Blame - docs/LangRef.html - platform/external/llvm

blob: 33fc7d4ff6cbc23b1596a8109623810510ae6741 [file] [log] [blame]

Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1	<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
				2	"http://www.w3.org/TR/html4/strict.dtd">
				3	<html>
				4	<head>
				5	<title>LLVM Assembly Language Reference Manual</title>
				6	<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
				7	<meta name="author" content="Chris Lattner">
				8	<meta name="description"
				9	content="LLVM Assembly Language Reference Manual.">
				10	<link rel="stylesheet" href="llvm.css" type="text/css">
				11	</head>
				12
				13	<body>
				14
				15	<div class="doc_title"> LLVM Language Reference Manual </div>
				16	<ol>
				17	<li><a href="#abstract">Abstract</a></li>
				18	<li><a href="#introduction">Introduction</a></li>
				19	<li><a href="#identifiers">Identifiers</a></li>
				20	<li><a href="#highlevel">High Level Structure</a>
				21	<ol>
				22	<li><a href="#modulestructure">Module Structure</a></li>
Bill Wendling	41a0785	2009-07-20 01:03:30 +0000	[diff] [blame]	23	<li><a href="#linkage">Linkage Types</a>
				24	<ol>
Bill Wendling	e275324	2009-07-20 02:41:50 +0000	[diff] [blame]	25	<li><a href="#linkage_private">'<tt>private</tt>' Linkage</a></li>
				26	<li><a href="#linkage_linker_private">'<tt>linker_private</tt>' Linkage</a></li>
				27	<li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
				28	<li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
				29	<li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
				30	<li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
				31	<li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
				32	<li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
				33	<li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
				34	<li><a href="#linkage_linkonce">'<tt>linkonce_odr</tt>' Linkage</a></li>
				35	<li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
				36	<li><a href="#linkage_external">'<tt>externally visible</tt>' Linkage</a></li>
				37	<li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
				38	<li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
Bill Wendling	41a0785	2009-07-20 01:03:30 +0000	[diff] [blame]	39	</ol>
				40	</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	41	<li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattner	5b6dc6e	2009-01-11 20:53:49 +0000	[diff] [blame]	42	<li><a href="#namedtypes">Named Types</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	43	<li><a href="#globalvars">Global Variables</a></li>
				44	<li><a href="#functionstructure">Functions</a></li>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	45	<li><a href="#aliasstructure">Aliases</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	46	<li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel	008cd3e	2008-09-26 23:51:19 +0000	[diff] [blame]	47	<li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen	13fe5e3	2007-12-10 03:18:06 +0000	[diff] [blame]	48	<li><a href="#gc">Garbage Collector Names</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	49	<li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
				50	<li><a href="#datalayout">Data Layout</a></li>
Dan Gohman	27b4701	2009-07-27 18:07:55 +0000	[diff] [blame]	51	<li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	52	</ol>
				53	</li>
				54	<li><a href="#typesystem">Type System</a>
				55	<ol>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	56	<li><a href="#t_classifications">Type Classifications</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	57	<li><a href="#t_primitive">Primitive Types</a>
				58	<ol>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	59	<li><a href="#t_floating">Floating Point Types</a></li>
				60	<li><a href="#t_void">Void Type</a></li>
				61	<li><a href="#t_label">Label Type</a></li>
Nick Lewycky	29aaef8	2009-05-30 05:06:04 +0000	[diff] [blame]	62	<li><a href="#t_metadata">Metadata Type</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	63	</ol>
				64	</li>
				65	<li><a href="#t_derived">Derived Types</a>
				66	<ol>
Chris Lattner	251ab81	2007-12-18 06:18:21 +0000	[diff] [blame]	67	<li><a href="#t_integer">Integer Type</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	68	<li><a href="#t_array">Array Type</a></li>
				69	<li><a href="#t_function">Function Type</a></li>
				70	<li><a href="#t_pointer">Pointer Type</a></li>
				71	<li><a href="#t_struct">Structure Type</a></li>
				72	<li><a href="#t_pstruct">Packed Structure Type</a></li>
				73	<li><a href="#t_vector">Vector Type</a></li>
				74	<li><a href="#t_opaque">Opaque Type</a></li>
				75	</ol>
				76	</li>
Chris Lattner	515195a	2009-02-02 07:32:36 +0000	[diff] [blame]	77	<li><a href="#t_uprefs">Type Up-references</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	78	</ol>
				79	</li>
				80	<li><a href="#constants">Constants</a>
				81	<ol>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	82	<li><a href="#simpleconstants">Simple Constants</a></li>
Chris Lattner	9706385	2009-02-28 18:32:25 +0000	[diff] [blame]	83	<li><a href="#complexconstants">Complex Constants</a></li>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	84	<li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
				85	<li><a href="#undefvalues">Undefined Values</a></li>
				86	<li><a href="#constantexprs">Constant Expressions</a></li>
Nick Lewycky	4dcf810	2009-04-04 07:22:01 +0000	[diff] [blame]	87	<li><a href="#metadata">Embedded Metadata</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	88	</ol>
				89	</li>
				90	<li><a href="#othervalues">Other Values</a>
				91	<ol>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	92	<li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	93	</ol>
				94	</li>
Chris Lattner	75c24e0	2009-07-20 05:55:19 +0000	[diff] [blame]	95	<li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
				96	<ol>
				97	<li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
Chris Lattner	1e0e0d1	2009-07-20 06:14:25 +0000	[diff] [blame]	98	<li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
				99	Global Variable</a></li>
Chris Lattner	75c24e0	2009-07-20 05:55:19 +0000	[diff] [blame]	100	<li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
				101	Global Variable</a></li>
				102	<li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
				103	Global Variable</a></li>
				104	</ol>
				105	</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	106	<li><a href="#instref">Instruction Reference</a>
				107	<ol>
				108	<li><a href="#terminators">Terminator Instructions</a>
				109	<ol>
				110	<li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
				111	<li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
				112	<li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
				113	<li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
				114	<li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
				115	<li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
				116	</ol>
				117	</li>
				118	<li><a href="#binaryops">Binary Operations</a>
				119	<ol>
				120	<li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	121	<li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	122	<li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	123	<li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	124	<li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	125	<li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	126	<li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
				127	<li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
				128	<li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
				129	<li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
				130	<li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
				131	<li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
				132	</ol>
				133	</li>
				134	<li><a href="#bitwiseops">Bitwise Binary Operations</a>
				135	<ol>
				136	<li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
				137	<li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
				138	<li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
				139	<li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
				140	<li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
				141	<li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
				142	</ol>
				143	</li>
				144	<li><a href="#vectorops">Vector Operations</a>
				145	<ol>
				146	<li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
				147	<li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
				148	<li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
				149	</ol>
				150	</li>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	151	<li><a href="#aggregateops">Aggregate Operations</a>
				152	<ol>
				153	<li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
				154	<li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
				155	</ol>
				156	</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	157	<li><a href="#memoryops">Memory Access and Addressing Operations</a>
				158	<ol>
				159	<li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
				160	<li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
				161	<li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
				162	<li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
				163	<li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
				164	<li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
				165	</ol>
				166	</li>
				167	<li><a href="#convertops">Conversion Operations</a>
				168	<ol>
				169	<li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
				170	<li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
				171	<li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
				172	<li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
				173	<li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
				174	<li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
				175	<li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
				176	<li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
				177	<li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
				178	<li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
				179	<li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
				180	<li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
				181	</ol>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	182	</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	183	<li><a href="#otherops">Other Operations</a>
				184	<ol>
				185	<li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
				186	<li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
				187	<li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
				188	<li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
				189	<li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
				190	<li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
				191	</ol>
				192	</li>
				193	</ol>
				194	</li>
				195	<li><a href="#intrinsics">Intrinsic Functions</a>
				196	<ol>
				197	<li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
				198	<ol>
				199	<li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
				200	<li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
				201	<li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
				202	</ol>
				203	</li>
				204	<li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
				205	<ol>
				206	<li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
				207	<li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
				208	<li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
				209	</ol>
				210	</li>
				211	<li><a href="#int_codegen">Code Generator Intrinsics</a>
				212	<ol>
				213	<li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
				214	<li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
				215	<li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
				216	<li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
				217	<li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
				218	<li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
				219	<li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
				220	</ol>
				221	</li>
				222	<li><a href="#int_libc">Standard C Library Intrinsics</a>
				223	<ol>
				224	<li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
				225	<li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
				226	<li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
				227	<li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
				228	<li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	229	<li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
				230	<li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
				231	<li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	232	</ol>
				233	</li>
				234	<li><a href="#int_manip">Bit Manipulation Intrinsics</a>
				235	<ol>
				236	<li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
				237	<li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
				238	<li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
				239	<li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	240	</ol>
				241	</li>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	242	<li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
				243	<ol>
Bill Wendling	3e1258b	2009-02-08 04:04:40 +0000	[diff] [blame]	244	<li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
				245	<li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
				246	<li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
				247	<li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
				248	<li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendling	bda98b6	2009-02-08 23:00:09 +0000	[diff] [blame]	249	<li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	250	</ol>
				251	</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	252	<li><a href="#int_debugger">Debugger intrinsics</a></li>
				253	<li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands	7407a9f	2007-09-11 14:10:23 +0000	[diff] [blame]	254	<li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	255	<ol>
				256	<li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	257	</ol>
				258	</li>
Bill Wendling	9127adb	2008-11-18 22:10:53 +0000	[diff] [blame]	259	<li><a href="#int_atomics">Atomic intrinsics</a>
				260	<ol>
				261	<li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
				262	<li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
				263	<li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
				264	<li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
				265	<li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
				266	<li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
				267	<li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
				268	<li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
				269	<li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
				270	<li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
				271	<li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
				272	<li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
				273	<li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
				274	</ol>
				275	</li>
Reid Spencer	b043f67	2007-07-20 19:59:11 +0000	[diff] [blame]	276	<li><a href="#int_general">General intrinsics</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	277	<ol>
Reid Spencer	b043f67	2007-07-20 19:59:11 +0000	[diff] [blame]	278	<li><a href="#int_var_annotation">
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	279	'<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattner	b306a9e	2007-09-21 22:59:12 +0000	[diff] [blame]	280	<li><a href="#int_annotation">
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	281	'<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikov	e6e764f	2008-01-15 22:31:34 +0000	[diff] [blame]	282	<li><a href="#int_trap">
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	283	'<tt>llvm.trap</tt>' Intrinsic</a></li>
				284	<li><a href="#int_stackprotector">
				285	'<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Tanya Lattner	b306a9e	2007-09-21 22:59:12 +0000	[diff] [blame]	286	</ol>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	287	</li>
				288	</ol>
				289	</li>
				290	</ol>
				291
				292	<div class="doc_author">
				293	<p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
				294	and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
				295	</div>
				296
				297	<!-- *********************************************************************** -->
				298	<div class="doc_section"> <a name="abstract">Abstract </a></div>
				299	<!-- *********************************************************************** -->
				300
				301	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	302
				303	<p>This document is a reference manual for the LLVM assembly language. LLVM is
				304	a Static Single Assignment (SSA) based representation that provides type
				305	safety, low-level operations, flexibility, and the capability of representing
				306	'all' high-level languages cleanly. It is the common code representation
				307	used throughout all phases of the LLVM compilation strategy.</p>
				308
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	309	</div>
				310
				311	<!-- *********************************************************************** -->
				312	<div class="doc_section"> <a name="introduction">Introduction</a> </div>
				313	<!-- *********************************************************************** -->
				314
				315	<div class="doc_text">
				316
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	317	<p>The LLVM code representation is designed to be used in three different forms:
				318	as an in-memory compiler IR, as an on-disk bitcode representation (suitable
				319	for fast loading by a Just-In-Time compiler), and as a human readable
				320	assembly language representation. This allows LLVM to provide a powerful
				321	intermediate representation for efficient compiler transformations and
				322	analysis, while providing a natural means to debug and visualize the
				323	transformations. The three different forms of LLVM are all equivalent. This
				324	document describes the human readable representation and notation.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	325
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	326	<p>The LLVM representation aims to be light-weight and low-level while being
				327	expressive, typed, and extensible at the same time. It aims to be a
				328	"universal IR" of sorts, by being at a low enough level that high-level ideas
				329	may be cleanly mapped to it (similar to how microprocessors are "universal
				330	IR's", allowing many source languages to be mapped to them). By providing
				331	type information, LLVM can be used as the target of optimizations: for
				332	example, through pointer analysis, it can be proven that a C automatic
				333	variable is never accessed outside of the current function... allowing it to
				334	be promoted to a simple SSA value instead of a memory location.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	335
				336	</div>
				337
				338	<!-- _______________________________________________________________________ -->
				339	<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
				340
				341	<div class="doc_text">
				342
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	343	<p>It is important to note that this document describes 'well formed' LLVM
				344	assembly language. There is a difference between what the parser accepts and
				345	what is considered 'well formed'. For example, the following instruction is
				346	syntactically okay, but not well formed:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	347
				348	<div class="doc_code">
				349	<pre>
				350	%x = <a href="#i_add">add</a> i32 1, %x
				351	</pre>
				352	</div>
				353
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	354	<p>...because the definition of <tt>%x</tt> does not dominate all of its
				355	uses. The LLVM infrastructure provides a verification pass that may be used
				356	to verify that an LLVM module is well formed. This pass is automatically run
				357	by the parser after parsing input assembly and by the optimizer before it
				358	outputs bitcode. The violations pointed out by the verifier pass indicate
				359	bugs in transformation passes or input to the parser.</p>
				360
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	361	</div>
				362
Chris Lattner	a83fdc0	2007-10-03 17:34:29 +0000	[diff] [blame]	363	<!-- Describe the typesetting conventions here. -->
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	364
				365	<!-- *********************************************************************** -->
				366	<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
				367	<!-- *********************************************************************** -->
				368
				369	<div class="doc_text">
				370
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	371	<p>LLVM identifiers come in two basic types: global and local. Global
				372	identifiers (functions, global variables) begin with the <tt>'@'</tt>
				373	character. Local identifiers (register names, types) begin with
				374	the <tt>'%'</tt> character. Additionally, there are three different formats
				375	for identifiers, for different purposes:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	376
				377	<ol>
Reid Spencer	c8245b0	2007-08-07 14:34:28 +0000	[diff] [blame]	378	<li>Named values are represented as a string of characters with their prefix.
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	379	For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
				380	<tt>%a.really.long.identifier</tt>. The actual regular expression used is
				381	'<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
				382	other characters in their names can be surrounded with quotes. Special
				383	characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
				384	ASCII code for the character in hexadecimal. In this way, any character
				385	can be used in a name value, even quotes themselves.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	386
Reid Spencer	c8245b0	2007-08-07 14:34:28 +0000	[diff] [blame]	387	<li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	388	prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	389
				390	<li>Constants, which are described in a <a href="#constants">section about
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	391	constants</a>, below.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	392	</ol>
				393
Reid Spencer	c8245b0	2007-08-07 14:34:28 +0000	[diff] [blame]	394	<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	395	don't need to worry about name clashes with reserved words, and the set of
				396	reserved words may be expanded in the future without penalty. Additionally,
				397	unnamed identifiers allow a compiler to quickly come up with a temporary
				398	variable without having to avoid symbol table conflicts.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	399
				400	<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	401	languages. There are keywords for different opcodes
				402	('<tt><a href="#i_add">add</a></tt>',
				403	'<tt><a href="#i_bitcast">bitcast</a></tt>',
				404	'<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
				405	('<tt><a href="#t_void">void</a></tt>',
				406	'<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
				407	reserved words cannot conflict with variable names, because none of them
				408	start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	409
				410	<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	411	'<tt>%X</tt>' by 8:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	412
				413	<p>The easy way:</p>
				414
				415	<div class="doc_code">
				416	<pre>
				417	%result = <a href="#i_mul">mul</a> i32 %X, 8
				418	</pre>
				419	</div>
				420
				421	<p>After strength reduction:</p>
				422
				423	<div class="doc_code">
				424	<pre>
				425	%result = <a href="#i_shl">shl</a> i32 %X, i8 3
				426	</pre>
				427	</div>
				428
				429	<p>And the hard way:</p>
				430
				431	<div class="doc_code">
				432	<pre>
				433	<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
				434	<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
				435	%result = <a href="#i_add">add</a> i32 %1, %1
				436	</pre>
				437	</div>
				438
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	439	<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
				440	lexical features of LLVM:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	441
				442	<ol>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	443	<li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	444	line.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	445
				446	<li>Unnamed temporaries are created when the result of a computation is not
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	447	assigned to a named value.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	448
				449	<li>Unnamed temporaries are numbered sequentially</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	450	</ol>
				451
				452	<p>...and it also shows a convention that we follow in this document. When
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	453	demonstrating instructions, we will follow an instruction with a comment that
				454	defines the type and name of value produced. Comments are shown in italic
				455	text.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	456
				457	</div>
				458
				459	<!-- *********************************************************************** -->
				460	<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
				461	<!-- *********************************************************************** -->
				462
				463	<!-- ======================================================================= -->
				464	<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
				465	</div>
				466
				467	<div class="doc_text">
				468
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	469	<p>LLVM programs are composed of "Module"s, each of which is a translation unit
				470	of the input programs. Each module consists of functions, global variables,
				471	and symbol table entries. Modules may be combined together with the LLVM
				472	linker, which merges function (and global variable) definitions, resolves
				473	forward declarations, and merges symbol table entries. Here is an example of
				474	the "hello world" module:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	475
				476	<div class="doc_code">
				477	<pre><i>; Declare the string constant as a global constant...</i>
				478	<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
				479	href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
				480
				481	<i>; External declaration of the puts function</i>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	482	<a href="#functionstructure">declare</a> i32 @puts(i8 ) <i>; i32(i8 )* </i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	483
				484	<i>; Definition of main function</i>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	485	define i32 @main() { <i>; i32()* </i>
Dan Gohman	0185238	2009-01-04 23:44:43 +0000	[diff] [blame]	486	<i>; Convert [13 x i8]* to i8 *...</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	487	%cast210 = <a
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	488	href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	489
				490	<i>; Call puts function to write out the string to stdout...</i>
				491	<a
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	492	href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	493	<a
				494	href="#i_ret">ret</a> i32 0<br>}<br>
				495	</pre>
				496	</div>
				497
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	498	<p>This example is made up of a <a href="#globalvars">global variable</a> named
				499	"<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function, and
				500	a <a href="#functionstructure">function definition</a> for
				501	"<tt>main</tt>".</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	502
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	503	<p>In general, a module is made up of a list of global values, where both
				504	functions and global variables are global values. Global values are
				505	represented by a pointer to a memory location (in this case, a pointer to an
				506	array of char, and a pointer to a function), and have one of the
				507	following <a href="#linkage">linkage types</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	508
				509	</div>
				510
				511	<!-- ======================================================================= -->
				512	<div class="doc_subsection">
				513	<a name="linkage">Linkage Types</a>
				514	</div>
				515
				516	<div class="doc_text">
				517
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	518	<p>All Global Variables and Functions have one of the following types of
				519	linkage:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	520
				521	<dl>
Rafael Espindola	a168fc9	2009-01-15 20:18:42 +0000	[diff] [blame]	522	<dt><tt><b><a name="linkage_private">private</a></b></tt>: </dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	523	<dd>Global values with private linkage are only directly accessible by objects
				524	in the current module. In particular, linking code into a module with an
				525	private global value may cause the private to be renamed as necessary to
				526	avoid collisions. Because the symbol is private to the module, all
				527	references can be updated. This doesn't show up in any symbol table in the
				528	object file.</dd>
Rafael Espindola	a168fc9	2009-01-15 20:18:42 +0000	[diff] [blame]	529
Bill Wendling	41a0785	2009-07-20 01:03:30 +0000	[diff] [blame]	530	<dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt>: </dt>
Bill Wendling	41a0785	2009-07-20 01:03:30 +0000	[diff] [blame]	531	<dd>Similar to private, but the symbol is passed through the assembler and
				532	removed by the linker after evaluation.</dd>
				533
Dale Johannesen	96e7e09	2008-05-23 23:13:41 +0000	[diff] [blame]	534	<dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	535	<dd>Similar to private, but the value shows as a local symbol
				536	(<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
				537	corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	538
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	539	<dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt>: </dt>
Chris Lattner	6843344	2009-04-13 05:44:34 +0000	[diff] [blame]	540	<dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	541	into the object file corresponding to the LLVM module. They exist to
				542	allow inlining and other optimizations to take place given knowledge of
				543	the definition of the global, which is known to be somewhere outside the
				544	module. Globals with <tt>available_externally</tt> linkage are allowed to
				545	be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
				546	This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner	6843344	2009-04-13 05:44:34 +0000	[diff] [blame]	547
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	548	<dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	549	<dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	550	the same name when linkage occurs. This is typically used to implement
				551	inline functions, templates, or other code which must be generated in each
				552	translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
				553	allowed to be discarded.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	554
Dale Johannesen	96e7e09	2008-05-23 23:13:41 +0000	[diff] [blame]	555	<dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	556	<dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
				557	linkage, except that unreferenced <tt>common</tt> globals may not be
				558	discarded. This is used for globals that may be emitted in multiple
				559	translation units, but that are not guaranteed to be emitted into every
				560	translation unit that uses them. One example of this is tentative
				561	definitions in C, such as "<tt>int X;</tt>" at global scope.</dd>
Dale Johannesen	96e7e09	2008-05-23 23:13:41 +0000	[diff] [blame]	562
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	563	<dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
Dale Johannesen	96e7e09	2008-05-23 23:13:41 +0000	[diff] [blame]	564	<dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	565	that some targets may choose to emit different assembly sequences for them
				566	for target-dependent reasons. This is used for globals that are declared
				567	"weak" in C source code.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	568
				569	<dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	570	<dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	571	pointer to array type. When two global variables with appending linkage
				572	are linked together, the two global arrays are appended together. This is
				573	the LLVM, typesafe, equivalent of having the system linker append together
				574	"sections" with identical names when .o files are linked.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	575
				576	<dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	577	<dd>The semantics of this linkage follow the ELF object file model: the symbol
				578	is weak until linked, if not linked, the symbol becomes null instead of
				579	being an undefined reference.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	580
Duncan Sands	19d161f	2009-03-07 15:45:40 +0000	[diff] [blame]	581	<dt><tt><b><a name="linkage_linkonce">linkonce_odr</a></b></tt>: </dt>
Duncan Sands	19d161f	2009-03-07 15:45:40 +0000	[diff] [blame]	582	<dt><tt><b><a name="linkage_weak">weak_odr</a></b></tt>: </dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	583	<dd>Some languages allow differing globals to be merged, such as two functions
				584	with different semantics. Other languages, such as <tt>C++</tt>, ensure
				585	that only equivalent globals are ever merged (the "one definition rule" -
				586	"ODR"). Such languages can use the <tt>linkonce_odr</tt>
				587	and <tt>weak_odr</tt> linkage types to indicate that the global will only
				588	be merged with equivalent globals. These linkage types are otherwise the
				589	same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands	19d161f	2009-03-07 15:45:40 +0000	[diff] [blame]	590
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	591	<dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	592	<dd>If none of the above identifiers are used, the global is externally
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	593	visible, meaning that it participates in linkage and can be used to
				594	resolve external symbol references.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	595	</dl>
				596
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	597	<p>The next two types of linkage are targeted for Microsoft Windows platform
				598	only. They are designed to support importing (exporting) symbols from (to)
				599	DLLs (Dynamic Link Libraries).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	600
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	601	<dl>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	602	<dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	603	<dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	604	or variable via a global pointer to a pointer that is set up by the DLL
				605	exporting the symbol. On Microsoft Windows targets, the pointer name is
				606	formed by combining <code>__imp_</code> and the function or variable
				607	name.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	608
				609	<dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	610	<dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	611	pointer to a pointer in a DLL, so that it can be referenced with the
				612	<tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
				613	name is formed by combining <code>__imp_</code> and the function or
				614	variable name.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	615	</dl>
				616
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	617	<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
				618	another module defined a "<tt>.LC0</tt>" variable and was linked with this
				619	one, one of the two would be renamed, preventing a collision. Since
				620	"<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
				621	declarations), they are accessible outside of the current module.</p>
				622
				623	<p>It is illegal for a function <i>declaration</i> to have any linkage type
				624	other than "externally visible", <tt>dllimport</tt>
				625	or <tt>extern_weak</tt>.</p>
				626
Duncan Sands	19d161f	2009-03-07 15:45:40 +0000	[diff] [blame]	627	<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	628	or <tt>weak_odr</tt> linkages.</p>
				629
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	630	</div>
				631
				632	<!-- ======================================================================= -->
				633	<div class="doc_subsection">
				634	<a name="callingconv">Calling Conventions</a>
				635	</div>
				636
				637	<div class="doc_text">
				638
				639	<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	640	and <a href="#i_invoke">invokes</a> can all have an optional calling
				641	convention specified for the call. The calling convention of any pair of
				642	dynamic caller/callee must match, or the behavior of the program is
				643	undefined. The following calling conventions are supported by LLVM, and more
				644	may be added in the future:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	645
				646	<dl>
				647	<dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	648	<dd>This calling convention (the default if no other calling convention is
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	649	specified) matches the target C calling conventions. This calling
				650	convention supports varargs function calls and tolerates some mismatch in
				651	the declared prototype and implemented declaration of the function (as
				652	does normal C).</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	653
				654	<dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	655	<dd>This calling convention attempts to make calls as fast as possible
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	656	(e.g. by passing things in registers). This calling convention allows the
				657	target to use whatever tricks it wants to produce fast code for the
				658	target, without having to conform to an externally specified ABI
				659	(Application Binary Interface). Implementations of this convention should
				660	allow arbitrary <a href="CodeGenerator.html#tailcallopt">tail call
				661	optimization</a> to be supported. This calling convention does not
				662	support varargs and requires the prototype of all callees to exactly match
				663	the prototype of the function definition.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	664
				665	<dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	666	<dd>This calling convention attempts to make code in the caller as efficient
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	667	as possible under the assumption that the call is not commonly executed.
				668	As such, these calls often preserve all registers so that the call does
				669	not break any live ranges in the caller side. This calling convention
				670	does not support varargs and requires the prototype of all callees to
				671	exactly match the prototype of the function definition.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	672
				673	<dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	674	<dd>Any calling convention may be specified by number, allowing
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	675	target-specific calling conventions to be used. Target specific calling
				676	conventions start at 64.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	677	</dl>
				678
				679	<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	680	support Pascal conventions or any other well-known target-independent
				681	convention.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	682
				683	</div>
				684
				685	<!-- ======================================================================= -->
				686	<div class="doc_subsection">
				687	<a name="visibility">Visibility Styles</a>
				688	</div>
				689
				690	<div class="doc_text">
				691
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	692	<p>All Global Variables and Functions have one of the following visibility
				693	styles:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	694
				695	<dl>
				696	<dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattner	9645148	2008-08-05 18:29:16 +0000	[diff] [blame]	697	<dd>On targets that use the ELF object file format, default visibility means
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	698	that the declaration is visible to other modules and, in shared libraries,
				699	means that the declared entity may be overridden. On Darwin, default
				700	visibility means that the declaration is visible to other modules. Default
				701	visibility corresponds to "external linkage" in the language.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	702
				703	<dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	704	<dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	705	object if they are in the same shared object. Usually, hidden visibility
				706	indicates that the symbol will not be placed into the dynamic symbol
				707	table, so no other module (executable or shared library) can reference it
				708	directly.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	709
				710	<dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	711	<dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	712	the dynamic symbol table, but that references within the defining module
				713	will bind to the local symbol. That is, the symbol cannot be overridden by
				714	another module.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	715	</dl>
				716
				717	</div>
				718
				719	<!-- ======================================================================= -->
				720	<div class="doc_subsection">
Chris Lattner	5b6dc6e	2009-01-11 20:53:49 +0000	[diff] [blame]	721	<a name="namedtypes">Named Types</a>
				722	</div>
				723
				724	<div class="doc_text">
				725
				726	<p>LLVM IR allows you to specify name aliases for certain types. This can make
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	727	it easier to read the IR and make the IR more condensed (particularly when
				728	recursive types are involved). An example of a name specification is:</p>
Chris Lattner	5b6dc6e	2009-01-11 20:53:49 +0000	[diff] [blame]	729
				730	<div class="doc_code">
				731	<pre>
				732	%mytype = type { %mytype*, i32 }
				733	</pre>
				734	</div>
				735
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	736	<p>You may give a name to any <a href="#typesystem">type</a> except
				737	"<a href="t_void">void</a>". Type name aliases may be used anywhere a type
				738	is expected with the syntax "%mytype".</p>
Chris Lattner	5b6dc6e	2009-01-11 20:53:49 +0000	[diff] [blame]	739
				740	<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	741	and that you can therefore specify multiple names for the same type. This
				742	often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
				743	uses structural typing, the name is not part of the type. When printing out
				744	LLVM IR, the printer will pick <em>one name</em> to render all types of a
				745	particular shape. This means that if you have code where two different
				746	source types end up having the same LLVM type, that the dumper will sometimes
				747	print the "wrong" or unexpected type. This is an important design point and
				748	isn't going to change.</p>
Chris Lattner	5b6dc6e	2009-01-11 20:53:49 +0000	[diff] [blame]	749
				750	</div>
				751
Chris Lattner	5b6dc6e	2009-01-11 20:53:49 +0000	[diff] [blame]	752	<!-- ======================================================================= -->
				753	<div class="doc_subsection">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	754	<a name="globalvars">Global Variables</a>
				755	</div>
				756
				757	<div class="doc_text">
				758
				759	<p>Global variables define regions of memory allocated at compilation time
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	760	instead of run-time. Global variables may optionally be initialized, may
				761	have an explicit section to be placed in, and may have an optional explicit
				762	alignment specified. A variable may be defined as "thread_local", which
				763	means that it will not be shared by threads (each thread will have a
				764	separated copy of the variable). A variable may be defined as a global
				765	"constant," which indicates that the contents of the variable
				766	will <b>never</b> be modified (enabling better optimization, allowing the
				767	global data to be placed in the read-only section of an executable, etc).
				768	Note that variables that need runtime initialization cannot be marked
				769	"constant" as there is a store to the variable.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	770
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	771	<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
				772	constant, even if the final definition of the global is not. This capability
				773	can be used to enable slightly better optimization of the program, but
				774	requires the language definition to guarantee that optimizations based on the
				775	'constantness' are valid for the translation units that do not include the
				776	definition.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	777
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	778	<p>As SSA values, global variables define pointer values that are in scope
				779	(i.e. they dominate) all basic blocks in the program. Global variables
				780	always define a pointer to their "content" type because they describe a
				781	region of memory, and all memory objects in LLVM are accessed through
				782	pointers.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	783
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	784	<p>A global variable may be declared to reside in a target-specific numbered
				785	address space. For targets that support them, address spaces may affect how
				786	optimizations are performed and/or what target instructions are used to
				787	access the variable. The default address space is zero. The address space
				788	qualifier must precede any other attributes.</p>
Christopher Lamb	dd0049d	2007-12-11 09:31:00 +0000	[diff] [blame]	789
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	790	<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	791	supports it, it will emit globals to the section specified.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	792
				793	<p>An explicit alignment may be specified for a global. If not present, or if
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	794	the alignment is set to zero, the alignment of the global is set by the
				795	target to whatever it feels convenient. If an explicit alignment is
				796	specified, the global is forced to have at least that much alignment. All
				797	alignments must be a power of 2.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	798
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	799	<p>For example, the following defines a global in a numbered address space with
				800	an initializer, section, and alignment:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	801
				802	<div class="doc_code">
				803	<pre>
Dan Gohman	21ef02c	2009-01-11 00:40:00 +0000	[diff] [blame]	804	@G = addrspace(5) constant float 1.0, section "foo", align 4
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	805	</pre>
				806	</div>
				807
				808	</div>
				809
				810
				811	<!-- ======================================================================= -->
				812	<div class="doc_subsection">
				813	<a name="functionstructure">Functions</a>
				814	</div>
				815
				816	<div class="doc_text">
				817
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	818	<p>LLVM function definitions consist of the "<tt>define</tt>" keyord, an
				819	optional <a href="#linkage">linkage type</a>, an optional
				820	<a href="#visibility">visibility style</a>, an optional
				821	<a href="#callingconv">calling convention</a>, a return type, an optional
				822	<a href="#paramattrs">parameter attribute</a> for the return type, a function
				823	name, a (possibly empty) argument list (each with optional
				824	<a href="#paramattrs">parameter attributes</a>), optional
				825	<a href="#fnattrs">function attributes</a>, an optional section, an optional
				826	alignment, an optional <a href="#gc">garbage collector name</a>, an opening
				827	curly brace, a list of basic blocks, and a closing curly brace.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	828
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	829	<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
				830	optional <a href="#linkage">linkage type</a>, an optional
				831	<a href="#visibility">visibility style</a>, an optional
				832	<a href="#callingconv">calling convention</a>, a return type, an optional
				833	<a href="#paramattrs">parameter attribute</a> for the return type, a function
				834	name, a possibly empty list of arguments, an optional alignment, and an
				835	optional <a href="#gc">garbage collector name</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	836
Chris Lattner	9645148	2008-08-05 18:29:16 +0000	[diff] [blame]	837	<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	838	(Control Flow Graph) for the function. Each basic block may optionally start
				839	with a label (giving the basic block a symbol table entry), contains a list
				840	of instructions, and ends with a <a href="#terminators">terminator</a>
				841	instruction (such as a branch or function return).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	842
				843	<p>The first basic block in a function is special in two ways: it is immediately
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	844	executed on entrance to the function, and it is not allowed to have
				845	predecessor basic blocks (i.e. there can not be any branches to the entry
				846	block of a function). Because the block can have no predecessors, it also
				847	cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	848
				849	<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	850	supports it, it will emit functions to the section specified.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	851
				852	<p>An explicit alignment may be specified for a function. If not present, or if
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	853	the alignment is set to zero, the alignment of the function is set by the
				854	target to whatever it feels convenient. If an explicit alignment is
				855	specified, the function is forced to have at least that much alignment. All
				856	alignments must be a power of 2.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	857
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	858	<h5>Syntax:</h5>
Devang Patel	d0bfcc7	2008-10-07 17:48:33 +0000	[diff] [blame]	859	<div class="doc_code">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	860	<pre>
Chris Lattner	1e5c5cd0	2008-10-13 16:55:18 +0000	[diff] [blame]	861	define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	862	[<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
				863	<ResultType> @<FunctionName> ([argument list])
				864	[<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
				865	[<a href="#gc">gc</a>] { ... }
				866	</pre>
Devang Patel	d0bfcc7	2008-10-07 17:48:33 +0000	[diff] [blame]	867	</div>
				868
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	869	</div>
				870
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	871	<!-- ======================================================================= -->
				872	<div class="doc_subsection">
				873	<a name="aliasstructure">Aliases</a>
				874	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	875
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	876	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	877
				878	<p>Aliases act as "second name" for the aliasee value (which can be either
				879	function, global variable, another alias or bitcast of global value). Aliases
				880	may have an optional <a href="#linkage">linkage type</a>, and an
				881	optional <a href="#visibility">visibility style</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	882
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	883	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	884	<div class="doc_code">
				885	<pre>
Duncan Sands	d7bfabf	2008-09-12 20:48:21 +0000	[diff] [blame]	886	@<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	887	</pre>
				888	</div>
				889
				890	</div>
				891
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	892	<!-- ======================================================================= -->
				893	<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	894
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	895	<div class="doc_text">
				896
				897	<p>The return type and each parameter of a function type may have a set of
				898	<i>parameter attributes</i> associated with them. Parameter attributes are
				899	used to communicate additional information about the result or parameters of
				900	a function. Parameter attributes are considered to be part of the function,
				901	not of the function type, so functions with different parameter attributes
				902	can have the same function type.</p>
				903
				904	<p>Parameter attributes are simple keywords that follow the type specified. If
				905	multiple parameter attributes are needed, they are space separated. For
				906	example:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	907
				908	<div class="doc_code">
				909	<pre>
Nick Lewycky	3022a74	2009-02-15 23:06:14 +0000	[diff] [blame]	910	declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattner	f33b845	2008-10-04 18:33:34 +0000	[diff] [blame]	911	declare i32 @atoi(i8 zeroext)
				912	declare signext i8 @returns_signed_char()
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	913	</pre>
				914	</div>
				915
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	916	<p>Note that any attributes for the function result (<tt>nounwind</tt>,
				917	<tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	918
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	919	<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner	275e6be	2008-01-11 06:20:47 +0000	[diff] [blame]	920
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	921	<dl>
				922	<dt><tt>zeroext</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	923	<dd>This indicates to the code generator that the parameter or return value
				924	should be zero-extended to a 32-bit value by the caller (for a parameter)
				925	or the callee (for a return value).</dd>
Chris Lattner	275e6be	2008-01-11 06:20:47 +0000	[diff] [blame]	926
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	927	<dt><tt>signext</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	928	<dd>This indicates to the code generator that the parameter or return value
				929	should be sign-extended to a 32-bit value by the caller (for a parameter)
				930	or the callee (for a return value).</dd>
Chris Lattner	275e6be	2008-01-11 06:20:47 +0000	[diff] [blame]	931
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	932	<dt><tt>inreg</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	933	<dd>This indicates that this parameter or return value should be treated in a
				934	special target-dependent fashion during while emitting code for a function
				935	call or return (usually, by putting it in a register as opposed to memory,
				936	though some targets use it to distinguish between two different kinds of
				937	registers). Use of this attribute is target-specific.</dd>
Chris Lattner	275e6be	2008-01-11 06:20:47 +0000	[diff] [blame]	938
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	939	<dt><tt><a name="byval">byval</a></tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	940	<dd>This indicates that the pointer parameter should really be passed by value
				941	to the function. The attribute implies that a hidden copy of the pointee
				942	is made between the caller and the callee, so the callee is unable to
				943	modify the value in the callee. This attribute is only valid on LLVM
				944	pointer arguments. It is generally used to pass structs and arrays by
				945	value, but is also valid on pointers to scalars. The copy is considered
				946	to belong to the caller not the callee (for example,
				947	<tt><a href="#readonly">readonly</a></tt> functions should not write to
				948	<tt>byval</tt> parameters). This is not a valid attribute for return
				949	values. The byval attribute also supports specifying an alignment with
				950	the align attribute. This has a target-specific effect on the code
				951	generator that usually indicates a desired alignment for the synthesized
				952	stack slot.</dd>
				953
				954	<dt><tt>sret</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	955	<dd>This indicates that the pointer parameter specifies the address of a
				956	structure that is the return value of the function in the source program.
				957	This pointer must be guaranteed by the caller to be valid: loads and
				958	stores to the structure may be assumed by the callee to not to trap. This
				959	may only be applied to the first parameter. This is not a valid attribute
				960	for return values. </dd>
				961
				962	<dt><tt>noalias</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	963	<dd>This indicates that the pointer does not alias any global or any other
				964	parameter. The caller is responsible for ensuring that this is the
				965	case. On a function return value, <tt>noalias</tt> additionally indicates
				966	that the pointer does not alias any other pointers visible to the
				967	caller. For further details, please see the discussion of the NoAlias
				968	response in
				969	<a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
				970	analysis</a>.</dd>
				971
				972	<dt><tt>nocapture</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	973	<dd>This indicates that the callee does not make any copies of the pointer
				974	that outlive the callee itself. This is not a valid attribute for return
				975	values.</dd>
				976
				977	<dt><tt>nest</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	978	<dd>This indicates that the pointer parameter can be excised using the
				979	<a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
				980	attribute for return values.</dd>
				981	</dl>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	982
				983	</div>
				984
				985	<!-- ======================================================================= -->
				986	<div class="doc_subsection">
Gordon Henriksen	13fe5e3	2007-12-10 03:18:06 +0000	[diff] [blame]	987	<a name="gc">Garbage Collector Names</a>
				988	</div>
				989
				990	<div class="doc_text">
Gordon Henriksen	13fe5e3	2007-12-10 03:18:06 +0000	[diff] [blame]	991
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	992	<p>Each function may specify a garbage collector name, which is simply a
				993	string:</p>
				994
				995	<div class="doc_code">
				996	<pre>
				997	define void @f() gc "name" { ...
				998	</pre>
				999	</div>
Gordon Henriksen	13fe5e3	2007-12-10 03:18:06 +0000	[diff] [blame]	1000
				1001	<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1002	collector which will cause the compiler to alter its output in order to
				1003	support the named garbage collection algorithm.</p>
				1004
Gordon Henriksen	13fe5e3	2007-12-10 03:18:06 +0000	[diff] [blame]	1005	</div>
				1006
				1007	<!-- ======================================================================= -->
				1008	<div class="doc_subsection">
Devang Patel	008cd3e	2008-09-26 23:51:19 +0000	[diff] [blame]	1009	<a name="fnattrs">Function Attributes</a>
Devang Patel	d468f1c	2008-09-04 23:05:13 +0000	[diff] [blame]	1010	</div>
				1011
				1012	<div class="doc_text">
Devang Patel	008cd3e	2008-09-26 23:51:19 +0000	[diff] [blame]	1013
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1014	<p>Function attributes are set to communicate additional information about a
				1015	function. Function attributes are considered to be part of the function, not
				1016	of the function type, so functions with different parameter attributes can
				1017	have the same function type.</p>
Devang Patel	008cd3e	2008-09-26 23:51:19 +0000	[diff] [blame]	1018
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1019	<p>Function attributes are simple keywords that follow the type specified. If
				1020	multiple attributes are needed, they are space separated. For example:</p>
Devang Patel	d468f1c	2008-09-04 23:05:13 +0000	[diff] [blame]	1021
				1022	<div class="doc_code">
Bill Wendling	74d3eac	2008-09-07 10:26:33 +0000	[diff] [blame]	1023	<pre>
Devang Patel	008cd3e	2008-09-26 23:51:19 +0000	[diff] [blame]	1024	define void @f() noinline { ... }
				1025	define void @f() alwaysinline { ... }
				1026	define void @f() alwaysinline optsize { ... }
				1027	define void @f() optsize
Bill Wendling	74d3eac	2008-09-07 10:26:33 +0000	[diff] [blame]	1028	</pre>
Devang Patel	d468f1c	2008-09-04 23:05:13 +0000	[diff] [blame]	1029	</div>
				1030
Bill Wendling	74d3eac	2008-09-07 10:26:33 +0000	[diff] [blame]	1031	<dl>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1032	<dt><tt>alwaysinline</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1033	<dd>This attribute indicates that the inliner should attempt to inline this
				1034	function into callers whenever possible, ignoring any active inlining size
				1035	threshold for this caller.</dd>
Bill Wendling	74d3eac	2008-09-07 10:26:33 +0000	[diff] [blame]	1036
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1037	<dt><tt>noinline</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1038	<dd>This attribute indicates that the inliner should never inline this
				1039	function in any situation. This attribute may not be used together with
				1040	the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel	008cd3e	2008-09-26 23:51:19 +0000	[diff] [blame]	1041
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1042	<dt><tt>optsize</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1043	<dd>This attribute suggests that optimization passes and code generator passes
				1044	make choices that keep the code size of this function low, and otherwise
				1045	do optimizations specifically to reduce code size.</dd>
Devang Patel	008cd3e	2008-09-26 23:51:19 +0000	[diff] [blame]	1046
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1047	<dt><tt>noreturn</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1048	<dd>This function attribute indicates that the function never returns
				1049	normally. This produces undefined behavior at runtime if the function
				1050	ever does dynamically return.</dd>
Bill Wendling	dfaabba	2008-11-13 01:02:51 +0000	[diff] [blame]	1051
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1052	<dt><tt>nounwind</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1053	<dd>This function attribute indicates that the function never returns with an
				1054	unwind or exceptional control flow. If the function does unwind, its
				1055	runtime behavior is undefined.</dd>
Bill Wendling	be9ec3f	2008-11-26 19:07:40 +0000	[diff] [blame]	1056
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1057	<dt><tt>readnone</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1058	<dd>This attribute indicates that the function computes its result (or decides
				1059	to unwind an exception) based strictly on its arguments, without
				1060	dereferencing any pointer arguments or otherwise accessing any mutable
				1061	state (e.g. memory, control registers, etc) visible to caller functions.
				1062	It does not write through any pointer arguments
				1063	(including <tt><a href="#byval">byval</a></tt> arguments) and never
				1064	changes any state visible to callers. This means that it cannot unwind
				1065	exceptions by calling the <tt>C++</tt> exception throwing methods, but
				1066	could use the <tt>unwind</tt> instruction.</dd>
Devang Patel	a2f9f41	2009-06-12 19:45:19 +0000	[diff] [blame]	1067
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1068	<dt><tt><a name="readonly">readonly</a></tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1069	<dd>This attribute indicates that the function does not write through any
				1070	pointer arguments (including <tt><a href="#byval">byval</a></tt>
				1071	arguments) or otherwise modify any state (e.g. memory, control registers,
				1072	etc) visible to caller functions. It may dereference pointer arguments
				1073	and read state that may be set in the caller. A readonly function always
				1074	returns the same value (or unwinds an exception identically) when called
				1075	with the same set of arguments and global state. It cannot unwind an
				1076	exception by calling the <tt>C++</tt> exception throwing methods, but may
				1077	use the <tt>unwind</tt> instruction.</dd>
Anton Korobeynikov	edd7d11	2009-07-17 18:07:26 +0000	[diff] [blame]	1078
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1079	<dt><tt><a name="ssp">ssp</a></tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1080	<dd>This attribute indicates that the function should emit a stack smashing
				1081	protector. It is in the form of a "canary"—a random value placed on
				1082	the stack before the local variables that's checked upon return from the
				1083	function to see if it has been overwritten. A heuristic is used to
				1084	determine if a function needs stack protectors or not.<br>
				1085	<br>
				1086	If a function that has an <tt>ssp</tt> attribute is inlined into a
				1087	function that doesn't have an <tt>ssp</tt> attribute, then the resulting
				1088	function will have an <tt>ssp</tt> attribute.</dd>
				1089
				1090	<dt><tt>sspreq</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1091	<dd>This attribute indicates that the function should <em>always</em> emit a
				1092	stack smashing protector. This overrides
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	1093	the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
				1094	<br>
				1095	If a function that has an <tt>sspreq</tt> attribute is inlined into a
				1096	function that doesn't have an <tt>sspreq</tt> attribute or which has
				1097	an <tt>ssp</tt> attribute, then the resulting function will have
				1098	an <tt>sspreq</tt> attribute.</dd>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1099
				1100	<dt><tt>noredzone</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1101	<dd>This attribute indicates that the code generator should not use a red
				1102	zone, even if the target-specific ABI normally permits it.</dd>
				1103
				1104	<dt><tt>noimplicitfloat</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1105	<dd>This attributes disables implicit floating point instructions.</dd>
				1106
				1107	<dt><tt>naked</tt></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1108	<dd>This attribute disables prologue / epilogue emission for the function.
				1109	This can have very system-specific consequences.</dd>
Bill Wendling	74d3eac	2008-09-07 10:26:33 +0000	[diff] [blame]	1110	</dl>
				1111
Devang Patel	d468f1c	2008-09-04 23:05:13 +0000	[diff] [blame]	1112	</div>
				1113
				1114	<!-- ======================================================================= -->
				1115	<div class="doc_subsection">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1116	<a name="moduleasm">Module-Level Inline Assembly</a>
				1117	</div>
				1118
				1119	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1120
				1121	<p>Modules may contain "module-level inline asm" blocks, which corresponds to
				1122	the GCC "file scope inline asm" blocks. These blocks are internally
				1123	concatenated by LLVM and treated as a single unit, but may be separated in
				1124	the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1125
				1126	<div class="doc_code">
				1127	<pre>
				1128	module asm "inline asm code goes here"
				1129	module asm "more can go here"
				1130	</pre>
				1131	</div>
				1132
				1133	<p>The strings can contain any character by escaping non-printable characters.
				1134	The escape sequence used is simply "\xx" where "xx" is the two digit hex code
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1135	for the number.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1136
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1137	<p>The inline asm code is simply printed to the machine code .s file when
				1138	assembly code is generated.</p>
				1139
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1140	</div>
				1141
				1142	<!-- ======================================================================= -->
				1143	<div class="doc_subsection">
				1144	<a name="datalayout">Data Layout</a>
				1145	</div>
				1146
				1147	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1148
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1149	<p>A module may specify a target specific data layout string that specifies how
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1150	data is to be laid out in memory. The syntax for the data layout is
				1151	simply:</p>
				1152
				1153	<div class="doc_code">
				1154	<pre>
				1155	target datalayout = "<i>layout specification</i>"
				1156	</pre>
				1157	</div>
				1158
				1159	<p>The <i>layout specification</i> consists of a list of specifications
				1160	separated by the minus sign character ('-'). Each specification starts with
				1161	a letter and may include other information after the letter to define some
				1162	aspect of the data layout. The specifications accepted are as follows:</p>
				1163
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1164	<dl>
				1165	<dt><tt>E</tt></dt>
				1166	<dd>Specifies that the target lays out data in big-endian form. That is, the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1167	bits with the most significance have the lowest address location.</dd>
				1168
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1169	<dt><tt>e</tt></dt>
Chris Lattner	9645148	2008-08-05 18:29:16 +0000	[diff] [blame]	1170	<dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1171	the bits with the least significance have the lowest address
				1172	location.</dd>
				1173
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1174	<dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
				1175	<dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1176	<i>preferred</i> alignments. All sizes are in bits. Specifying
				1177	the <i>pref</i> alignment is optional. If omitted, the
				1178	preceding <tt>:</tt> should be omitted too.</dd>
				1179
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1180	<dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
				1181	<dd>This specifies the alignment for an integer type of a given bit
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1182	<i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
				1183
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1184	<dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
				1185	<dd>This specifies the alignment for a vector type of a given bit
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1186	<i>size</i>.</dd>
				1187
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1188	<dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
				1189	<dd>This specifies the alignment for a floating point type of a given bit
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1190	<i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
				1191	(double).</dd>
				1192
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1193	<dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
				1194	<dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1195	<i>size</i>.</dd>
				1196
Daniel Dunbar	d88a97b	2009-06-08 22:17:53 +0000	[diff] [blame]	1197	<dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
				1198	<dd>This specifies the alignment for a stack object of a given bit
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1199	<i>size</i>.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1200	</dl>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1201
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1202	<p>When constructing the data layout for a given target, LLVM starts with a
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1203	default set of specifications which are then (possibly) overriden by the
				1204	specifications in the <tt>datalayout</tt> keyword. The default specifications
				1205	are given in this list:</p>
				1206
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1207	<ul>
				1208	<li><tt>E</tt> - big endian</li>
				1209	<li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
				1210	<li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
				1211	<li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
				1212	<li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
				1213	<li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner	9645148	2008-08-05 18:29:16 +0000	[diff] [blame]	1214	<li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1215	alignment of 64-bits</li>
				1216	<li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
				1217	<li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
				1218	<li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
				1219	<li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
				1220	<li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbar	d88a97b	2009-06-08 22:17:53 +0000	[diff] [blame]	1221	<li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1222	</ul>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1223
				1224	<p>When LLVM is determining the alignment for a given type, it uses the
				1225	following rules:</p>
				1226
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1227	<ol>
				1228	<li>If the type sought is an exact match for one of the specifications, that
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1229	specification is used.</li>
				1230
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1231	<li>If no match is found, and the type sought is an integer type, then the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1232	smallest integer type that is larger than the bitwidth of the sought type
				1233	is used. If none of the specifications are larger than the bitwidth then
				1234	the the largest integer type is used. For example, given the default
				1235	specifications above, the i7 type will use the alignment of i8 (next
				1236	largest) while both i65 and i256 will use the alignment of i64 (largest
				1237	specified).</li>
				1238
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1239	<li>If no match is found, and the type sought is a vector type, then the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1240	largest vector type that is smaller than the sought vector type will be
				1241	used as a fall back. This happens because <128 x double> can be
				1242	implemented in terms of 64 <2 x double>, for example.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1243	</ol>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1244
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1245	</div>
				1246
Dan Gohman	27b4701	2009-07-27 18:07:55 +0000	[diff] [blame]	1247	<!-- ======================================================================= -->
				1248	<div class="doc_subsection">
				1249	<a name="pointeraliasing">Pointer Aliasing Rules</a>
				1250	</div>
				1251
				1252	<div class="doc_text">
				1253
Andreas Bolka	11fbf43	2009-07-29 00:02:05 +0000	[diff] [blame]	1254	<p>Any memory access must be done through a pointer value associated
Andreas Bolka	23bece4	2009-07-27 20:37:10 +0000	[diff] [blame]	1255	with an address range of the memory access, otherwise the behavior
Dan Gohman	27b4701	2009-07-27 18:07:55 +0000	[diff] [blame]	1256	is undefined. Pointer values are associated with address ranges
				1257	according to the following rules:</p>
				1258
				1259	<ul>
Andreas Bolka	11fbf43	2009-07-29 00:02:05 +0000	[diff] [blame]	1260	<li>A pointer value formed from a
				1261	<tt><a href="#i_getelementptr">getelementptr</a></tt> instruction
				1262	is associated with the addresses associated with the first operand
				1263	of the <tt>getelementptr</tt>.</li>
				1264	<li>An address of a global variable is associated with the address
Dan Gohman	27b4701	2009-07-27 18:07:55 +0000	[diff] [blame]	1265	range of the variable's storage.</li>
				1266	<li>The result value of an allocation instruction is associated with
				1267	the address range of the allocated storage.</li>
				1268	<li>A null pointer in the default address-space is associated with
Andreas Bolka	11fbf43	2009-07-29 00:02:05 +0000	[diff] [blame]	1269	no address.</li>
				1270	<li>A pointer value formed by an
				1271	<tt><a href="#i_inttoptr">inttoptr</a></tt> is associated with all
				1272	address ranges of all pointer values that contribute (directly or
				1273	indirectly) to the computation of the pointer's value.</li>
				1274	<li>The result value of a
				1275	<tt><a href="#i_bitcast">bitcast</a></tt> is associated with all
Dan Gohman	27b4701	2009-07-27 18:07:55 +0000	[diff] [blame]	1276	addresses associated with the operand of the <tt>bitcast</tt>.</li>
				1277	<li>An integer constant other than zero or a pointer value returned
				1278	from a function not defined within LLVM may be associated with address
				1279	ranges allocated through mechanisms other than those provided by
Andreas Bolka	11fbf43	2009-07-29 00:02:05 +0000	[diff] [blame]	1280	LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman	27b4701	2009-07-27 18:07:55 +0000	[diff] [blame]	1281	allocated by mechanisms provided by LLVM.</li>
				1282	</ul>
				1283
				1284	<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka	11fbf43	2009-07-29 00:02:05 +0000	[diff] [blame]	1285	<tt><a href="#i_load">load</a></tt> merely indicates the size and
				1286	alignment of the memory from which to load, as well as the
				1287	interpretation of the value. The first operand of a
				1288	<tt><a href="#i_store">store</a></tt> similarly only indicates the size
				1289	and alignment of the store.</p>
Dan Gohman	27b4701	2009-07-27 18:07:55 +0000	[diff] [blame]	1290
				1291	<p>Consequently, type-based alias analysis, aka TBAA, aka
				1292	<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
				1293	LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
				1294	additional information which specialized optimization passes may use
				1295	to implement type-based alias analysis.</p>
				1296
				1297	</div>
				1298
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1299	<!-- *********************************************************************** -->
				1300	<div class="doc_section"> <a name="typesystem">Type System</a> </div>
				1301	<!-- *********************************************************************** -->
				1302
				1303	<div class="doc_text">
				1304
				1305	<p>The LLVM type system is one of the most important features of the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1306	intermediate representation. Being typed enables a number of optimizations
				1307	to be performed on the intermediate representation directly, without having
				1308	to do extra analyses on the side before the transformation. A strong type
				1309	system makes it easier to read the generated code and enables novel analyses
				1310	and transformations that are not feasible to perform on normal three address
				1311	code representations.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1312
				1313	</div>
				1314
				1315	<!-- ======================================================================= -->
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1316	<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1317	Classifications</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1318
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1319	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1320
				1321	<p>The types fall into a few useful classifications:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1322
				1323	<table border="1" cellspacing="0" cellpadding="4">
				1324	<tbody>
				1325	<tr><th>Classification</th><th>Types</th></tr>
				1326	<tr>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1327	<td><a href="#t_integer">integer</a></td>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1328	<td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
				1329	</tr>
				1330	<tr>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1331	<td><a href="#t_floating">floating point</a></td>
				1332	<td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1333	</tr>
				1334	<tr>
				1335	<td><a name="t_firstclass">first class</a></td>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1336	<td><a href="#t_integer">integer</a>,
				1337	<a href="#t_floating">floating point</a>,
				1338	<a href="#t_pointer">pointer</a>,
Dan Gohman	f6237db	2008-06-18 18:42:13 +0000	[diff] [blame]	1339	<a href="#t_vector">vector</a>,
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	1340	<a href="#t_struct">structure</a>,
				1341	<a href="#t_array">array</a>,
Nick Lewycky	29aaef8	2009-05-30 05:06:04 +0000	[diff] [blame]	1342	<a href="#t_label">label</a>,
				1343	<a href="#t_metadata">metadata</a>.
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1344	</td>
				1345	</tr>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1346	<tr>
				1347	<td><a href="#t_primitive">primitive</a></td>
				1348	<td><a href="#t_label">label</a>,
				1349	<a href="#t_void">void</a>,
Nick Lewycky	29aaef8	2009-05-30 05:06:04 +0000	[diff] [blame]	1350	<a href="#t_floating">floating point</a>,
				1351	<a href="#t_metadata">metadata</a>.</td>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1352	</tr>
				1353	<tr>
				1354	<td><a href="#t_derived">derived</a></td>
				1355	<td><a href="#t_integer">integer</a>,
				1356	<a href="#t_array">array</a>,
				1357	<a href="#t_function">function</a>,
				1358	<a href="#t_pointer">pointer</a>,
				1359	<a href="#t_struct">structure</a>,
				1360	<a href="#t_pstruct">packed structure</a>,
				1361	<a href="#t_vector">vector</a>,
				1362	<a href="#t_opaque">opaque</a>.
Dan Gohman	032ba85	2008-10-14 16:32:04 +0000	[diff] [blame]	1363	</td>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1364	</tr>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1365	</tbody>
				1366	</table>
				1367
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1368	<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
				1369	important. Values of these types are the only ones which can be produced by
				1370	instructions, passed as arguments, or used as operands to instructions.</p>
				1371
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1372	</div>
				1373
				1374	<!-- ======================================================================= -->
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1375	<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner	8643761	2008-01-04 04:34:14 +0000	[diff] [blame]	1376
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1377	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1378
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1379	<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1380	system.</p>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1381
Chris Lattner	8643761	2008-01-04 04:34:14 +0000	[diff] [blame]	1382	</div>
				1383
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1384	<!-- _______________________________________________________________________ -->
				1385	<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
				1386
				1387	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1388
				1389	<table>
				1390	<tbody>
				1391	<tr><th>Type</th><th>Description</th></tr>
				1392	<tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
				1393	<tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
				1394	<tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
				1395	<tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
				1396	<tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
				1397	</tbody>
				1398	</table>
				1399
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1400	</div>
				1401
				1402	<!-- _______________________________________________________________________ -->
				1403	<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
				1404
				1405	<div class="doc_text">
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	1406
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1407	<h5>Overview:</h5>
				1408	<p>The void type does not represent any value and has no size.</p>
				1409
				1410	<h5>Syntax:</h5>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1411	<pre>
				1412	void
				1413	</pre>
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	1414
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1415	</div>
				1416
				1417	<!-- _______________________________________________________________________ -->
				1418	<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
				1419
				1420	<div class="doc_text">
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	1421
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1422	<h5>Overview:</h5>
				1423	<p>The label type represents code labels.</p>
				1424
				1425	<h5>Syntax:</h5>
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1426	<pre>
				1427	label
				1428	</pre>
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	1429
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1430	</div>
				1431
Nick Lewycky	29aaef8	2009-05-30 05:06:04 +0000	[diff] [blame]	1432	<!-- _______________________________________________________________________ -->
				1433	<div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
				1434
				1435	<div class="doc_text">
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	1436
Nick Lewycky	29aaef8	2009-05-30 05:06:04 +0000	[diff] [blame]	1437	<h5>Overview:</h5>
				1438	<p>The metadata type represents embedded metadata. The only derived type that
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1439	may contain metadata is <tt>metadata*</tt> or a function type that returns or
				1440	takes metadata typed parameters, but not pointer to metadata types.</p>
Nick Lewycky	29aaef8	2009-05-30 05:06:04 +0000	[diff] [blame]	1441
				1442	<h5>Syntax:</h5>
Nick Lewycky	29aaef8	2009-05-30 05:06:04 +0000	[diff] [blame]	1443	<pre>
				1444	metadata
				1445	</pre>
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	1446
Nick Lewycky	29aaef8	2009-05-30 05:06:04 +0000	[diff] [blame]	1447	</div>
				1448
Chris Lattner	488772f	2008-01-04 04:32:38 +0000	[diff] [blame]	1449
				1450	<!-- ======================================================================= -->
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1451	<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
				1452
				1453	<div class="doc_text">
				1454
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1455	<p>The real power in LLVM comes from the derived types in the system. This is
				1456	what allows a programmer to represent arrays, functions, pointers, and other
				1457	useful types. Note that these derived types may be recursive: For example,
				1458	it is possible to have a two dimensional array.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1459
				1460	</div>
				1461
				1462	<!-- _______________________________________________________________________ -->
				1463	<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
				1464
				1465	<div class="doc_text">
				1466
				1467	<h5>Overview:</h5>
				1468	<p>The integer type is a very simple derived type that simply specifies an
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1469	arbitrary bit width for the integer type desired. Any bit width from 1 bit to
				1470	2^23-1 (about 8 million) can be specified.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1471
				1472	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1473	<pre>
				1474	iN
				1475	</pre>
				1476
				1477	<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1478	value.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1479
				1480	<h5>Examples:</h5>
				1481	<table class="layout">
Nick Lewycky	39382d6	2009-05-24 02:46:06 +0000	[diff] [blame]	1482	<tr class="layout">
				1483	<td class="left"><tt>i1</tt></td>
				1484	<td class="left">a single-bit integer.</td>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1485	</tr>
Nick Lewycky	39382d6	2009-05-24 02:46:06 +0000	[diff] [blame]	1486	<tr class="layout">
				1487	<td class="left"><tt>i32</tt></td>
				1488	<td class="left">a 32-bit integer.</td>
				1489	</tr>
				1490	<tr class="layout">
				1491	<td class="left"><tt>i1942652</tt></td>
				1492	<td class="left">a really big integer of over 1 million bits.</td>
				1493	</tr>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1494	</table>
djg	e93155c	2009-01-24 15:58:40 +0000	[diff] [blame]	1495
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1496	<p>Note that the code generator does not yet support large integer types to be
				1497	used as function return types. The specific limit on how large a return type
				1498	the code generator can currently handle is target-dependent; currently it's
				1499	often 64 bits for 32-bit targets and 128 bits for 64-bit targets.</p>
djg	e93155c	2009-01-24 15:58:40 +0000	[diff] [blame]	1500
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1501	</div>
				1502
				1503	<!-- _______________________________________________________________________ -->
				1504	<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
				1505
				1506	<div class="doc_text">
				1507
				1508	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1509	<p>The array type is a very simple derived type that arranges elements
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1510	sequentially in memory. The array type requires a size (number of elements)
				1511	and an underlying data type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1512
				1513	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1514	<pre>
				1515	[<# elements> x <elementtype>]
				1516	</pre>
				1517
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1518	<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
				1519	be any type with a size.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1520
				1521	<h5>Examples:</h5>
				1522	<table class="layout">
				1523	<tr class="layout">
Chris Lattner	7311d22	2007-12-19 05:04:11 +0000	[diff] [blame]	1524	<td class="left"><tt>[40 x i32]</tt></td>
				1525	<td class="left">Array of 40 32-bit integer values.</td>
				1526	</tr>
				1527	<tr class="layout">
				1528	<td class="left"><tt>[41 x i32]</tt></td>
				1529	<td class="left">Array of 41 32-bit integer values.</td>
				1530	</tr>
				1531	<tr class="layout">
				1532	<td class="left"><tt>[4 x i8]</tt></td>
				1533	<td class="left">Array of 4 8-bit integer values.</td>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1534	</tr>
				1535	</table>
				1536	<p>Here are some examples of multidimensional arrays:</p>
				1537	<table class="layout">
				1538	<tr class="layout">
Chris Lattner	7311d22	2007-12-19 05:04:11 +0000	[diff] [blame]	1539	<td class="left"><tt>[3 x [4 x i32]]</tt></td>
				1540	<td class="left">3x4 array of 32-bit integer values.</td>
				1541	</tr>
				1542	<tr class="layout">
				1543	<td class="left"><tt>[12 x [10 x float]]</tt></td>
				1544	<td class="left">12x10 array of single precision floating point values.</td>
				1545	</tr>
				1546	<tr class="layout">
				1547	<td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
				1548	<td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1549	</tr>
				1550	</table>
				1551
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1552	<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
				1553	length array. Normally, accesses past the end of an array are undefined in
				1554	LLVM (e.g. it is illegal to access the 5th element of a 3 element array). As
				1555	a special case, however, zero length arrays are recognized to be variable
				1556	length. This allows implementation of 'pascal style arrays' with the LLVM
				1557	type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1558
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1559	<p>Note that the code generator does not yet support large aggregate types to be
				1560	used as function return types. The specific limit on how large an aggregate
				1561	return type the code generator can currently handle is target-dependent, and
				1562	also dependent on the aggregate element types.</p>
djg	e93155c	2009-01-24 15:58:40 +0000	[diff] [blame]	1563
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1564	</div>
				1565
				1566	<!-- _______________________________________________________________________ -->
				1567	<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1568
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1569	<div class="doc_text">
Chris Lattner	43030e7	2008-04-23 04:59:35 +0000	[diff] [blame]	1570
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1571	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1572	<p>The function type can be thought of as a function signature. It consists of
				1573	a return type and a list of formal parameter types. The return type of a
				1574	function type is a scalar type, a void type, or a struct type. If the return
				1575	type is a struct type then all struct elements must be of first class types,
				1576	and the struct must have at least one element.</p>
Devang Patel	a3cc537	2008-03-10 20:49:15 +0000	[diff] [blame]	1577
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1578	<h5>Syntax:</h5>
Chris Lattner	43030e7	2008-04-23 04:59:35 +0000	[diff] [blame]	1579	<pre>
				1580	<returntype list> (<parameter list>)
				1581	</pre>
				1582
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1583	<p>...where '<tt><parameter list></tt>' is a comma-separated list of type
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1584	specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
				1585	which indicates that the function takes a variable number of arguments.
				1586	Variable argument functions can access their arguments with
				1587	the <a href="#int_varargs">variable argument handling intrinsic</a>
				1588	functions. '<tt><returntype list></tt>' is a comma-separated list of
				1589	<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner	43030e7	2008-04-23 04:59:35 +0000	[diff] [blame]	1590
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1591	<h5>Examples:</h5>
				1592	<table class="layout">
				1593	<tr class="layout">
				1594	<td class="left"><tt>i32 (i32)</tt></td>
				1595	<td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
				1596	</td>
				1597	</tr><tr class="layout">
Reid Spencer	f234bed	2007-07-19 23:13:04 +0000	[diff] [blame]	1598	<td class="left"><tt>float (i16 signext, i32 )
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1599	</tt></td>
				1600	<td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
				1601	an <tt>i16</tt> that should be sign extended and a
				1602	<a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
				1603	<tt>float</tt>.
				1604	</td>
				1605	</tr><tr class="layout">
				1606	<td class="left"><tt>i32 (i8*, ...)</tt></td>
				1607	<td class="left">A vararg function that takes at least one
				1608	<a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
				1609	which returns an integer. This is the signature for <tt>printf</tt> in
				1610	LLVM.
				1611	</td>
Devang Patel	d4ba41d	2008-03-24 05:35:41 +0000	[diff] [blame]	1612	</tr><tr class="layout">
				1613	<td class="left"><tt>{i32, i32} (i32)</tt></td>
Misha Brukman	afc88b0	2008-11-27 06:41:20 +0000	[diff] [blame]	1614	<td class="left">A function taking an <tt>i32</tt>, returning two
				1615	<tt>i32</tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Patel	d4ba41d	2008-03-24 05:35:41 +0000	[diff] [blame]	1616	</td>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1617	</tr>
				1618	</table>
				1619
				1620	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1621
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1622	<!-- _______________________________________________________________________ -->
				1623	<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1624
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1625	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1626
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1627	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1628	<p>The structure type is used to represent a collection of data members together
				1629	in memory. The packing of the field types is defined to match the ABI of the
				1630	underlying processor. The elements of a structure may be any type that has a
				1631	size.</p>
				1632
				1633	<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
				1634	'<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
				1635	the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
				1636
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1637	<h5>Syntax:</h5>
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	1638	<pre>
				1639	{ <type list> }
				1640	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1641
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1642	<h5>Examples:</h5>
				1643	<table class="layout">
				1644	<tr class="layout">
				1645	<td class="left"><tt>{ i32, i32, i32 }</tt></td>
				1646	<td class="left">A triple of three <tt>i32</tt> values</td>
				1647	</tr><tr class="layout">
				1648	<td class="left"><tt>{ float, i32 (i32) * }</tt></td>
				1649	<td class="left">A pair, where the first element is a <tt>float</tt> and the
				1650	second element is a <a href="#t_pointer">pointer</a> to a
				1651	<a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
				1652	an <tt>i32</tt>.</td>
				1653	</tr>
				1654	</table>
djg	e93155c	2009-01-24 15:58:40 +0000	[diff] [blame]	1655
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1656	<p>Note that the code generator does not yet support large aggregate types to be
				1657	used as function return types. The specific limit on how large an aggregate
				1658	return type the code generator can currently handle is target-dependent, and
				1659	also dependent on the aggregate element types.</p>
djg	e93155c	2009-01-24 15:58:40 +0000	[diff] [blame]	1660
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1661	</div>
				1662
				1663	<!-- _______________________________________________________________________ -->
				1664	<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
				1665	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1666
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1667	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1668
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1669	<h5>Overview:</h5>
				1670	<p>The packed structure type is used to represent a collection of data members
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1671	together in memory. There is no padding between fields. Further, the
				1672	alignment of a packed structure is 1 byte. The elements of a packed
				1673	structure may be any type that has a size.</p>
				1674
				1675	<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
				1676	'<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
				1677	the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
				1678
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1679	<h5>Syntax:</h5>
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	1680	<pre>
				1681	< { <type list> } >
				1682	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1683
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1684	<h5>Examples:</h5>
				1685	<table class="layout">
				1686	<tr class="layout">
				1687	<td class="left"><tt>< { i32, i32, i32 } ></tt></td>
				1688	<td class="left">A triple of three <tt>i32</tt> values</td>
				1689	</tr><tr class="layout">
Bill Wendling	74d3eac	2008-09-07 10:26:33 +0000	[diff] [blame]	1690	<td class="left">
				1691	<tt>< { float, i32 (i32)* } ></tt></td>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1692	<td class="left">A pair, where the first element is a <tt>float</tt> and the
				1693	second element is a <a href="#t_pointer">pointer</a> to a
				1694	<a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
				1695	an <tt>i32</tt>.</td>
				1696	</tr>
				1697	</table>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1698
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1699	</div>
				1700
				1701	<!-- _______________________________________________________________________ -->
				1702	<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Chris Lattner	96edbd3	2009-02-08 19:53:29 +0000	[diff] [blame]	1703
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1704	<div class="doc_text">
				1705
				1706	<h5>Overview:</h5>
				1707	<p>As in many languages, the pointer type represents a pointer or reference to
				1708	another object, which must live in memory. Pointer types may have an optional
				1709	address space attribute defining the target-specific numbered address space
				1710	where the pointed-to object resides. The default address space is zero.</p>
				1711
				1712	<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
				1713	permit pointers to labels (<tt>label</tt>). Use <tt>i8</tt> instead.</p>
Chris Lattner	96edbd3	2009-02-08 19:53:29 +0000	[diff] [blame]	1714
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1715	<h5>Syntax:</h5>
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	1716	<pre>
				1717	<type> *
				1718	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1719
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1720	<h5>Examples:</h5>
				1721	<table class="layout">
				1722	<tr class="layout">
Dan Gohman	0185238	2009-01-04 23:44:43 +0000	[diff] [blame]	1723	<td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner	7311d22	2007-12-19 05:04:11 +0000	[diff] [blame]	1724	<td class="left">A <a href="#t_pointer">pointer</a> to <a
				1725	href="#t_array">array</a> of four <tt>i32</tt> values.</td>
				1726	</tr>
				1727	<tr class="layout">
				1728	<td class="left"><tt>i32 (i32 ) </tt></td>
				1729	<td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1730	href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner	7311d22	2007-12-19 05:04:11 +0000	[diff] [blame]	1731	<tt>i32</tt>.</td>
				1732	</tr>
				1733	<tr class="layout">
				1734	<td class="left"><tt>i32 addrspace(5)*</tt></td>
				1735	<td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
				1736	that resides in address space #5.</td>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1737	</tr>
				1738	</table>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1739
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1740	</div>
				1741
				1742	<!-- _______________________________________________________________________ -->
				1743	<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1744
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1745	<div class="doc_text">
				1746
				1747	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1748	<p>A vector type is a simple derived type that represents a vector of elements.
				1749	Vector types are used when multiple primitive data are operated in parallel
				1750	using a single instruction (SIMD). A vector type requires a size (number of
				1751	elements) and an underlying primitive data type. Vectors must have a power
				1752	of two length (1, 2, 4, 8, 16 ...). Vector types are considered
				1753	<a href="#t_firstclass">first class</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1754
				1755	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1756	<pre>
				1757	< <# elements> x <elementtype> >
				1758	</pre>
				1759
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1760	<p>The number of elements is a constant integer value; elementtype may be any
				1761	integer or floating point type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1762
				1763	<h5>Examples:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1764	<table class="layout">
				1765	<tr class="layout">
Chris Lattner	7311d22	2007-12-19 05:04:11 +0000	[diff] [blame]	1766	<td class="left"><tt><4 x i32></tt></td>
				1767	<td class="left">Vector of 4 32-bit integer values.</td>
				1768	</tr>
				1769	<tr class="layout">
				1770	<td class="left"><tt><8 x float></tt></td>
				1771	<td class="left">Vector of 8 32-bit floating-point values.</td>
				1772	</tr>
				1773	<tr class="layout">
				1774	<td class="left"><tt><2 x i64></tt></td>
				1775	<td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1776	</tr>
				1777	</table>
djg	e93155c	2009-01-24 15:58:40 +0000	[diff] [blame]	1778
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1779	<p>Note that the code generator does not yet support large vector types to be
				1780	used as function return types. The specific limit on how large a vector
				1781	return type codegen can currently handle is target-dependent; currently it's
				1782	often a few times longer than a hardware vector register.</p>
djg	e93155c	2009-01-24 15:58:40 +0000	[diff] [blame]	1783
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1784	</div>
				1785
				1786	<!-- _______________________________________________________________________ -->
				1787	<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
				1788	<div class="doc_text">
				1789
				1790	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1791	<p>Opaque types are used to represent unknown types in the system. This
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1792	corresponds (for example) to the C notion of a forward declared structure
				1793	type. In LLVM, opaque types can eventually be resolved to any type (not just
				1794	a structure type).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1795
				1796	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1797	<pre>
				1798	opaque
				1799	</pre>
				1800
				1801	<h5>Examples:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1802	<table class="layout">
				1803	<tr class="layout">
Chris Lattner	7311d22	2007-12-19 05:04:11 +0000	[diff] [blame]	1804	<td class="left"><tt>opaque</tt></td>
				1805	<td class="left">An opaque type.</td>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1806	</tr>
				1807	</table>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1808
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1809	</div>
				1810
Chris Lattner	515195a	2009-02-02 07:32:36 +0000	[diff] [blame]	1811	<!-- ======================================================================= -->
				1812	<div class="doc_subsection">
				1813	<a name="t_uprefs">Type Up-references</a>
				1814	</div>
				1815
				1816	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1817
Chris Lattner	515195a	2009-02-02 07:32:36 +0000	[diff] [blame]	1818	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1819	<p>An "up reference" allows you to refer to a lexically enclosing type without
				1820	requiring it to have a name. For instance, a structure declaration may
				1821	contain a pointer to any of the types it is lexically a member of. Example
				1822	of up references (with their equivalent as named type declarations)
				1823	include:</p>
Chris Lattner	515195a	2009-02-02 07:32:36 +0000	[diff] [blame]	1824
				1825	<pre>
Chris Lattner	5ad632d	2009-02-09 10:00:56 +0000	[diff] [blame]	1826	{ \2 * } %x = type { %x* }
Chris Lattner	515195a	2009-02-02 07:32:36 +0000	[diff] [blame]	1827	{ \2 }* %y = type { %y }*
				1828	\1* %z = type %z*
				1829	</pre>
				1830
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1831	<p>An up reference is needed by the asmprinter for printing out cyclic types
				1832	when there is no declared name for a type in the cycle. Because the
				1833	asmprinter does not want to print out an infinite type string, it needs a
				1834	syntax to handle recursive types that have no names (all names are optional
				1835	in llvm IR).</p>
Chris Lattner	515195a	2009-02-02 07:32:36 +0000	[diff] [blame]	1836
				1837	<h5>Syntax:</h5>
				1838	<pre>
				1839	\<level>
				1840	</pre>
				1841
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1842	<p>The level is the count of the lexical type that is being referred to.</p>
Chris Lattner	515195a	2009-02-02 07:32:36 +0000	[diff] [blame]	1843
				1844	<h5>Examples:</h5>
Chris Lattner	515195a	2009-02-02 07:32:36 +0000	[diff] [blame]	1845	<table class="layout">
				1846	<tr class="layout">
				1847	<td class="left"><tt>\1*</tt></td>
				1848	<td class="left">Self-referential pointer.</td>
				1849	</tr>
				1850	<tr class="layout">
				1851	<td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
				1852	<td class="left">Recursive structure where the upref refers to the out-most
				1853	structure.</td>
				1854	</tr>
				1855	</table>
Chris Lattner	515195a	2009-02-02 07:32:36 +0000	[diff] [blame]	1856
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1857	</div>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1858
				1859	<!-- *********************************************************************** -->
				1860	<div class="doc_section"> <a name="constants">Constants</a> </div>
				1861	<!-- *********************************************************************** -->
				1862
				1863	<div class="doc_text">
				1864
				1865	<p>LLVM has several different basic types of constants. This section describes
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1866	them all and their syntax.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1867
				1868	</div>
				1869
				1870	<!-- ======================================================================= -->
				1871	<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
				1872
				1873	<div class="doc_text">
				1874
				1875	<dl>
				1876	<dt><b>Boolean constants</b></dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1877	<dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1878	constants of the <tt><a href="#t_primitive">i1</a></tt> type.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1879
				1880	<dt><b>Integer constants</b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1881	<dd>Standard integers (such as '4') are constants of
				1882	the <a href="#t_integer">integer</a> type. Negative numbers may be used
				1883	with integer types.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1884
				1885	<dt><b>Floating point constants</b></dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1886	<dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1887	exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
				1888	notation (see below). The assembler requires the exact decimal value of a
				1889	floating-point constant. For example, the assembler accepts 1.25 but
				1890	rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
				1891	constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1892
				1893	<dt><b>Null pointer constants</b></dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1894	<dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1895	and must be of <a href="#t_pointer">pointer type</a>.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1896	</dl>
				1897
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1898	<p>The one non-intuitive notation for constants is the hexadecimal form of
				1899	floating point constants. For example, the form '<tt>double
				1900	0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
				1901	'<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
				1902	constants are required (and the only time that they are generated by the
				1903	disassembler) is when a floating point constant must be emitted but it cannot
				1904	be represented as a decimal floating point number in a reasonable number of
				1905	digits. For example, NaN's, infinities, and other special values are
				1906	represented in their IEEE hexadecimal format so that assembly and disassembly
				1907	do not cause any bits to change in the constants.</p>
				1908
Dale Johannesen	f82a52f	2009-02-11 22:14:51 +0000	[diff] [blame]	1909	<p>When using the hexadecimal form, constants of types float and double are
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1910	represented using the 16-digit form shown above (which matches the IEEE754
				1911	representation for double); float values must, however, be exactly
				1912	representable as IEE754 single precision. Hexadecimal format is always used
				1913	for long double, and there are three forms of long double. The 80-bit format
				1914	used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
				1915	The 128-bit format used by PowerPC (two adjacent doubles) is represented
				1916	by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
				1917	is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
				1918	currently supported target uses this format. Long doubles will only work if
				1919	they match the long double format on your target. All hexadecimal formats
				1920	are big-endian (sign bit at the left).</p>
				1921
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1922	</div>
				1923
				1924	<!-- ======================================================================= -->
Chris Lattner	9706385	2009-02-28 18:32:25 +0000	[diff] [blame]	1925	<div class="doc_subsection">
Bill Wendling	1a2630a	2009-07-20 02:32:41 +0000	[diff] [blame]	1926	<a name="aggregateconstants"></a> <!-- old anchor -->
				1927	<a name="complexconstants">Complex Constants</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1928	</div>
				1929
				1930	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1931
Chris Lattner	9706385	2009-02-28 18:32:25 +0000	[diff] [blame]	1932	<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1933	constants and smaller complex constants.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1934
				1935	<dl>
				1936	<dt><b>Structure constants</b></dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1937	<dd>Structure constants are represented with notation similar to structure
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1938	type definitions (a comma separated list of elements, surrounded by braces
				1939	(<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
				1940	where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
				1941	Structure constants must have <a href="#t_struct">structure type</a>, and
				1942	the number and types of elements must match those specified by the
				1943	type.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1944
				1945	<dt><b>Array constants</b></dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1946	<dd>Array constants are represented with notation similar to array type
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1947	definitions (a comma separated list of elements, surrounded by square
				1948	brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
				1949	]</tt>". Array constants must have <a href="#t_array">array type</a>, and
				1950	the number and types of elements must match those specified by the
				1951	type.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1952
				1953	<dt><b>Vector constants</b></dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1954	<dd>Vector constants are represented with notation similar to vector type
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1955	definitions (a comma separated list of elements, surrounded by
				1956	less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32
				1957	42, i32 11, i32 74, i32 100 ></tt>". Vector constants must
				1958	have <a href="#t_vector">vector type</a>, and the number and types of
				1959	elements must match those specified by the type.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1960
				1961	<dt><b>Zero initialization</b></dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1962	<dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1963	value to zero of <em>any</em> type, including scalar and aggregate types.
				1964	This is often used to avoid having to print large zero initializers
				1965	(e.g. for large arrays) and is always exactly equivalent to using explicit
				1966	zero initializers.</dd>
Nick Lewycky	4dcf810	2009-04-04 07:22:01 +0000	[diff] [blame]	1967
				1968	<dt><b>Metadata node</b></dt>
Nick Lewycky	f122c7e	2009-05-30 16:08:30 +0000	[diff] [blame]	1969	<dd>A metadata node is a structure-like constant with
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1970	<a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
				1971	i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
				1972	be interpreted as part of the instruction stream, metadata is a place to
				1973	attach additional information such as debug info.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1974	</dl>
				1975
				1976	</div>
				1977
				1978	<!-- ======================================================================= -->
				1979	<div class="doc_subsection">
				1980	<a name="globalconstants">Global Variable and Function Addresses</a>
				1981	</div>
				1982
				1983	<div class="doc_text">
				1984
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	1985	<p>The addresses of <a href="#globalvars">global variables</a>
				1986	and <a href="#functionstructure">functions</a> are always implicitly valid
				1987	(link-time) constants. These constants are explicitly referenced when
				1988	the <a href="#identifiers">identifier for the global</a> is used and always
				1989	have <a href="#t_pointer">pointer</a> type. For example, the following is a
				1990	legal LLVM file:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1991
				1992	<div class="doc_code">
				1993	<pre>
				1994	@X = global i32 17
				1995	@Y = global i32 42
				1996	@Z = global [2 x i32] [ i32 @X, i32* @Y ]
				1997	</pre>
				1998	</div>
				1999
				2000	</div>
				2001
				2002	<!-- ======================================================================= -->
				2003	<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
				2004	<div class="doc_text">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2005
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2006	<p>The string '<tt>undef</tt>' is recognized as a type-less constant that has no
				2007	specific value. Undefined values may be of any type and be used anywhere a
				2008	constant is permitted.</p>
				2009
				2010	<p>Undefined values indicate to the compiler that the program is well defined no
				2011	matter what value is used, giving the compiler more freedom to optimize.</p>
				2012
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2013	</div>
				2014
				2015	<!-- ======================================================================= -->
				2016	<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
				2017	</div>
				2018
				2019	<div class="doc_text">
				2020
				2021	<p>Constant expressions are used to allow expressions involving other constants
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2022	to be used as constants. Constant expressions may be of
				2023	any <a href="#t_firstclass">first class</a> type and may involve any LLVM
				2024	operation that does not have side effects (e.g. load and call are not
				2025	supported). The following is the syntax for constant expressions:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2026
				2027	<dl>
				2028	<dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2029	<dd>Truncate a constant to another type. The bit size of CST must be larger
				2030	than the bit size of TYPE. Both types must be integers.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2031
				2032	<dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2033	<dd>Zero extend a constant to another type. The bit size of CST must be
				2034	smaller or equal to the bit size of TYPE. Both types must be
				2035	integers.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2036
				2037	<dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2038	<dd>Sign extend a constant to another type. The bit size of CST must be
				2039	smaller or equal to the bit size of TYPE. Both types must be
				2040	integers.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2041
				2042	<dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2043	<dd>Truncate a floating point constant to another floating point type. The
				2044	size of CST must be larger than the size of TYPE. Both types must be
				2045	floating point.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2046
				2047	<dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2048	<dd>Floating point extend a constant to another type. The size of CST must be
				2049	smaller or equal to the size of TYPE. Both types must be floating
				2050	point.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2051
Reid Spencer	e6adee8	2007-07-31 14:40:14 +0000	[diff] [blame]	2052	<dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2053	<dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2054	constant. TYPE must be a scalar or vector integer type. CST must be of
				2055	scalar or vector floating point type. Both CST and TYPE must be scalars,
				2056	or vectors of the same number of elements. If the value won't fit in the
				2057	integer type, the results are undefined.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2058
				2059	<dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
				2060	<dd>Convert a floating point constant to the corresponding signed integer
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2061	constant. TYPE must be a scalar or vector integer type. CST must be of
				2062	scalar or vector floating point type. Both CST and TYPE must be scalars,
				2063	or vectors of the same number of elements. If the value won't fit in the
				2064	integer type, the results are undefined.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2065
				2066	<dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
				2067	<dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2068	constant. TYPE must be a scalar or vector floating point type. CST must be
				2069	of scalar or vector integer type. Both CST and TYPE must be scalars, or
				2070	vectors of the same number of elements. If the value won't fit in the
				2071	floating point type, the results are undefined.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2072
				2073	<dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
				2074	<dd>Convert a signed integer constant to the corresponding floating point
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2075	constant. TYPE must be a scalar or vector floating point type. CST must be
				2076	of scalar or vector integer type. Both CST and TYPE must be scalars, or
				2077	vectors of the same number of elements. If the value won't fit in the
				2078	floating point type, the results are undefined.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2079
				2080	<dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
				2081	<dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2082	<tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
				2083	type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
				2084	make it fit in <tt>TYPE</tt>.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2085
				2086	<dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2087	<dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
				2088	type. CST must be of integer type. The CST value is zero extended,
				2089	truncated, or unchanged to make it fit in a pointer size. This one is
				2090	<i>really</i> dangerous!</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2091
				2092	<dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
Chris Lattner	557bc5d	2009-02-28 18:27:03 +0000	[diff] [blame]	2093	<dd>Convert a constant, CST, to another TYPE. The constraints of the operands
				2094	are the same as those for the <a href="#i_bitcast">bitcast
				2095	instruction</a>.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2096
				2097	<dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
Dan Gohman	106b2ae	2009-07-27 21:53:46 +0000	[diff] [blame]	2098	<dt><b><tt>getelementptr inbounds ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2099	<dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2100	constants. As with the <a href="#i_getelementptr">getelementptr</a>
				2101	instruction, the index list may have zero or more indexes, which are
				2102	required to make sense for the type of "CSTPTR".</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2103
				2104	<dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2105	<dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2106
				2107	<dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
				2108	<dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
				2109
				2110	<dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
				2111	<dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
				2112
				2113	<dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2114	<dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
				2115	constants.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2116
				2117	<dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2118	<dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
				2119	constants.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2120
				2121	<dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2122	<dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
				2123	constants.</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2124
				2125	<dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2126	<dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
				2127	be any of the <a href="#binaryops">binary</a>
				2128	or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
				2129	on operands are the same as those for the corresponding instruction
				2130	(e.g. no bitwise operations on floating point values are allowed).</dd>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2131	</dl>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2132
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2133	</div>
				2134
Nick Lewycky	4dcf810	2009-04-04 07:22:01 +0000	[diff] [blame]	2135	<!-- ======================================================================= -->
				2136	<div class="doc_subsection"><a name="metadata">Embedded Metadata</a>
				2137	</div>
				2138
				2139	<div class="doc_text">
				2140
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2141	<p>Embedded metadata provides a way to attach arbitrary data to the instruction
				2142	stream without affecting the behaviour of the program. There are two
				2143	metadata primitives, strings and nodes. All metadata has the
				2144	<tt>metadata</tt> type and is identified in syntax by a preceding exclamation
				2145	point ('<tt>!</tt>').</p>
Nick Lewycky	4dcf810	2009-04-04 07:22:01 +0000	[diff] [blame]	2146
				2147	<p>A metadata string is a string surrounded by double quotes. It can contain
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2148	any character by escaping non-printable characters with "\xx" where "xx" is
				2149	the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
Nick Lewycky	4dcf810	2009-04-04 07:22:01 +0000	[diff] [blame]	2150
				2151	<p>Metadata nodes are represented with notation similar to structure constants
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2152	(a comma separated list of elements, surrounded by braces and preceeded by an
				2153	exclamation point). For example: "<tt>!{ metadata !"test\00", i32
				2154	10}</tt>".</p>
Nick Lewycky	4dcf810	2009-04-04 07:22:01 +0000	[diff] [blame]	2155
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2156	<p>A metadata node will attempt to track changes to the values it holds. In the
				2157	event that a value is deleted, it will be replaced with a typeless
				2158	"<tt>null</tt>", such as "<tt>metadata !{null, i32 10}</tt>".</p>
Nick Lewycky	117f438	2009-05-10 20:57:05 +0000	[diff] [blame]	2159
Nick Lewycky	4dcf810	2009-04-04 07:22:01 +0000	[diff] [blame]	2160	<p>Optimizations may rely on metadata to provide additional information about
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2161	the program that isn't available in the instructions, or that isn't easily
				2162	computable. Similarly, the code generator may expect a certain metadata
				2163	format to be used to express debugging information.</p>
				2164
Nick Lewycky	4dcf810	2009-04-04 07:22:01 +0000	[diff] [blame]	2165	</div>
				2166
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2167	<!-- *********************************************************************** -->
				2168	<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
				2169	<!-- *********************************************************************** -->
				2170
				2171	<!-- ======================================================================= -->
				2172	<div class="doc_subsection">
				2173	<a name="inlineasm">Inline Assembler Expressions</a>
				2174	</div>
				2175
				2176	<div class="doc_text">
				2177
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2178	<p>LLVM supports inline assembler expressions (as opposed
				2179	to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
				2180	a special value. This value represents the inline assembler as a string
				2181	(containing the instructions to emit), a list of operand constraints (stored
				2182	as a string), and a flag that indicates whether or not the inline asm
				2183	expression has side effects. An example inline assembler expression is:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2184
				2185	<div class="doc_code">
				2186	<pre>
				2187	i32 (i32) asm "bswap $0", "=r,r"
				2188	</pre>
				2189	</div>
				2190
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2191	<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
				2192	a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
				2193	have:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2194
				2195	<div class="doc_code">
				2196	<pre>
				2197	%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
				2198	</pre>
				2199	</div>
				2200
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2201	<p>Inline asms with side effects not visible in the constraint list must be
				2202	marked as having side effects. This is done through the use of the
				2203	'<tt>sideeffect</tt>' keyword, like so:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2204
				2205	<div class="doc_code">
				2206	<pre>
				2207	call void asm sideeffect "eieio", ""()
				2208	</pre>
				2209	</div>
				2210
				2211	<p>TODO: The format of the asm and constraints string still need to be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2212	documented here. Constraints on what can be done (e.g. duplication, moving,
				2213	etc need to be documented). This is probably best done by reference to
				2214	another document that covers inline asm from a holistic perspective.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2215
				2216	</div>
				2217
Chris Lattner	75c24e0	2009-07-20 05:55:19 +0000	[diff] [blame]	2218
				2219	<!-- *********************************************************************** -->
				2220	<div class="doc_section">
				2221	<a name="intrinsic_globals">Intrinsic Global Variables</a>
				2222	</div>
				2223	<!-- *********************************************************************** -->
				2224
				2225	<p>LLVM has a number of "magic" global variables that contain data that affect
				2226	code generation or other IR semantics. These are documented here. All globals
Chris Lattner	1e0e0d1	2009-07-20 06:14:25 +0000	[diff] [blame]	2227	of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
				2228	section and all globals that start with "<tt>llvm.</tt>" are reserved for use
				2229	by LLVM.</p>
Chris Lattner	75c24e0	2009-07-20 05:55:19 +0000	[diff] [blame]	2230
				2231	<!-- ======================================================================= -->
				2232	<div class="doc_subsection">
				2233	<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
				2234	</div>
				2235
				2236	<div class="doc_text">
				2237
				2238	<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
				2239	href="#linkage_appending">appending linkage</a>. This array contains a list of
				2240	pointers to global variables and functions which may optionally have a pointer
				2241	cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
				2242
				2243	<pre>
				2244	@X = global i8 4
				2245	@Y = global i32 123
				2246
				2247	@llvm.used = appending global [2 x i8*] [
				2248	i8* @X,
				2249	i8* bitcast (i32* @Y to i8*)
				2250	], section "llvm.metadata"
				2251	</pre>
				2252
				2253	<p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
				2254	compiler, assembler, and linker are required to treat the symbol as if there is
				2255	a reference to the global that it cannot see. For example, if a variable has
				2256	internal linkage and no references other than that from the <tt>@llvm.used</tt>
				2257	list, it cannot be deleted. This is commonly used to represent references from
				2258	inline asms and other things the compiler cannot "see", and corresponds to
				2259	"attribute((used))" in GNU C.</p>
				2260
				2261	<p>On some targets, the code generator must emit a directive to the assembler or
				2262	object file to prevent the assembler and linker from molesting the symbol.</p>
				2263
				2264	</div>
				2265
				2266	<!-- ======================================================================= -->
				2267	<div class="doc_subsection">
Chris Lattner	1e0e0d1	2009-07-20 06:14:25 +0000	[diff] [blame]	2268	<a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
				2269	</div>
				2270
				2271	<div class="doc_text">
				2272
				2273	<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
				2274	<tt>@llvm.used</tt> directive, except that it only prevents the compiler from
				2275	touching the symbol. On targets that support it, this allows an intelligent
				2276	linker to optimize references to the symbol without being impeded as it would be
				2277	by <tt>@llvm.used</tt>.</p>
				2278
				2279	<p>This is a rare construct that should only be used in rare circumstances, and
				2280	should not be exposed to source languages.</p>
				2281
				2282	</div>
				2283
				2284	<!-- ======================================================================= -->
				2285	<div class="doc_subsection">
Chris Lattner	75c24e0	2009-07-20 05:55:19 +0000	[diff] [blame]	2286	<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
				2287	</div>
				2288
				2289	<div class="doc_text">
				2290
				2291	<p>TODO: Describe this.</p>
				2292
				2293	</div>
				2294
				2295	<!-- ======================================================================= -->
				2296	<div class="doc_subsection">
				2297	<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
				2298	</div>
				2299
				2300	<div class="doc_text">
				2301
				2302	<p>TODO: Describe this.</p>
				2303
				2304	</div>
				2305
				2306
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2307	<!-- *********************************************************************** -->
				2308	<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
				2309	<!-- *********************************************************************** -->
				2310
				2311	<div class="doc_text">
				2312
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2313	<p>The LLVM instruction set consists of several different classifications of
				2314	instructions: <a href="#terminators">terminator
				2315	instructions</a>, <a href="#binaryops">binary instructions</a>,
				2316	<a href="#bitwiseops">bitwise binary instructions</a>,
				2317	<a href="#memoryops">memory instructions</a>, and
				2318	<a href="#otherops">other instructions</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2319
				2320	</div>
				2321
				2322	<!-- ======================================================================= -->
				2323	<div class="doc_subsection"> <a name="terminators">Terminator
				2324	Instructions</a> </div>
				2325
				2326	<div class="doc_text">
				2327
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2328	<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
				2329	in a program ends with a "Terminator" instruction, which indicates which
				2330	block should be executed after the current block is finished. These
				2331	terminator instructions typically yield a '<tt>void</tt>' value: they produce
				2332	control flow, not values (the one exception being the
				2333	'<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
				2334
				2335	<p>There are six different terminator instructions: the
				2336	'<a href="#i_ret"><tt>ret</tt></a>' instruction, the
				2337	'<a href="#i_br"><tt>br</tt></a>' instruction, the
				2338	'<a href="#i_switch"><tt>switch</tt></a>' instruction, the
				2339	'<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
				2340	'<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
				2341	'<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2342
				2343	</div>
				2344
				2345	<!-- _______________________________________________________________________ -->
				2346	<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
				2347	Instruction</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2348
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2349	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2350
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2351	<h5>Syntax:</h5>
Dan Gohman	3e70003	2008-10-04 19:00:07 +0000	[diff] [blame]	2352	<pre>
				2353	ret <type> <value> <i>; Return a value from a non-void function</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2354	ret void <i>; Return from void function</i>
				2355	</pre>
Chris Lattner	43030e7	2008-04-23 04:59:35 +0000	[diff] [blame]	2356
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2357	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2358	<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
				2359	a value) from a function back to the caller.</p>
				2360
				2361	<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
				2362	value and then causes control flow, and one that just causes control flow to
				2363	occur.</p>
Chris Lattner	43030e7	2008-04-23 04:59:35 +0000	[diff] [blame]	2364
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2365	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2366	<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
				2367	return value. The type of the return value must be a
				2368	'<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohman	3e70003	2008-10-04 19:00:07 +0000	[diff] [blame]	2369
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2370	<p>A function is not <a href="#wellformed">well formed</a> if it it has a
				2371	non-void return type and contains a '<tt>ret</tt>' instruction with no return
				2372	value or a return value with a type that does not match its type, or if it
				2373	has a void return type and contains a '<tt>ret</tt>' instruction with a
				2374	return value.</p>
Chris Lattner	43030e7	2008-04-23 04:59:35 +0000	[diff] [blame]	2375
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2376	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2377	<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
				2378	the calling function's context. If the caller is a
				2379	"<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
				2380	instruction after the call. If the caller was an
				2381	"<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
				2382	the beginning of the "normal" destination block. If the instruction returns
				2383	a value, that value shall set the call or invoke instruction's return
				2384	value.</p>
Chris Lattner	43030e7	2008-04-23 04:59:35 +0000	[diff] [blame]	2385
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2386	<h5>Example:</h5>
Chris Lattner	43030e7	2008-04-23 04:59:35 +0000	[diff] [blame]	2387	<pre>
				2388	ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2389	ret void <i>; Return from a void function</i>
Bill Wendling	d163e2d	2009-02-28 22:12:54 +0000	[diff] [blame]	2390	ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2391	</pre>
Dan Gohman	6096719	2009-01-12 23:12:39 +0000	[diff] [blame]	2392
djg	e93155c	2009-01-24 15:58:40 +0000	[diff] [blame]	2393	<p>Note that the code generator does not yet fully support large
				2394	return values. The specific sizes that are currently supported are
				2395	dependent on the target. For integers, on 32-bit targets the limit
				2396	is often 64 bits, and on 64-bit targets the limit is often 128 bits.
				2397	For aggregate types, the current limits are dependent on the element
				2398	types; for example targets are often limited to 2 total integer
				2399	elements and 2 total floating-point elements.</p>
Dan Gohman	6096719	2009-01-12 23:12:39 +0000	[diff] [blame]	2400
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2401	</div>
				2402	<!-- _______________________________________________________________________ -->
				2403	<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2404
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2405	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2406
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2407	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2408	<pre>
				2409	br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2410	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2411
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2412	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2413	<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
				2414	different basic block in the current function. There are two forms of this
				2415	instruction, corresponding to a conditional branch and an unconditional
				2416	branch.</p>
				2417
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2418	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2419	<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
				2420	'<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
				2421	of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
				2422	target.</p>
				2423
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2424	<h5>Semantics:</h5>
				2425	<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2426	argument is evaluated. If the value is <tt>true</tt>, control flows to the
				2427	'<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
				2428	control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
				2429
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2430	<h5>Example:</h5>
Bill Wendling	6ec4061	2009-07-20 02:39:26 +0000	[diff] [blame]	2431	<pre>
				2432	Test:
				2433	%cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
				2434	br i1 %cond, label %IfEqual, label %IfUnequal
				2435	IfEqual:
				2436	<a href="#i_ret">ret</a> i32 1
				2437	IfUnequal:
				2438	<a href="#i_ret">ret</a> i32 0
				2439	</pre>
				2440
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2441	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2442
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2443	<!-- _______________________________________________________________________ -->
				2444	<div class="doc_subsubsection">
				2445	<a name="i_switch">'<tt>switch</tt>' Instruction</a>
				2446	</div>
				2447
				2448	<div class="doc_text">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2449
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2450	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2451	<pre>
				2452	switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
				2453	</pre>
				2454
				2455	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2456	<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2457	several different places. It is a generalization of the '<tt>br</tt>'
				2458	instruction, allowing a branch to occur to one of many possible
				2459	destinations.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2460
				2461	<h5>Arguments:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2462	<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2463	comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
				2464	and an array of pairs of comparison value constants and '<tt>label</tt>'s.
				2465	The table is not allowed to contain duplicate constant entries.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2466
				2467	<h5>Semantics:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2468	<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2469	destinations. When the '<tt>switch</tt>' instruction is executed, this table
				2470	is searched for the given value. If the value is found, control flow is
				2471	transfered to the corresponding destination; otherwise, control flow is
				2472	transfered to the default destination.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2473
				2474	<h5>Implementation:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2475	<p>Depending on properties of the target machine and the particular
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2476	<tt>switch</tt> instruction, this instruction may be code generated in
				2477	different ways. For example, it could be generated as a series of chained
				2478	conditional branches or with a lookup table.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2479
				2480	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2481	<pre>
				2482	<i>; Emulate a conditional br instruction</i>
				2483	%Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman	0185238	2009-01-04 23:44:43 +0000	[diff] [blame]	2484	switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2485
				2486	<i>; Emulate an unconditional br instruction</i>
				2487	switch i32 0, label %dest [ ]
				2488
				2489	<i>; Implement a jump table:</i>
Dan Gohman	0185238	2009-01-04 23:44:43 +0000	[diff] [blame]	2490	switch i32 %val, label %otherwise [ i32 0, label %onzero
				2491	i32 1, label %onone
				2492	i32 2, label %ontwo ]
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2493	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2494
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2495	</div>
				2496
				2497	<!-- _______________________________________________________________________ -->
				2498	<div class="doc_subsubsection">
				2499	<a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
				2500	</div>
				2501
				2502	<div class="doc_text">
				2503
				2504	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2505	<pre>
Devang Patel	d0bfcc7	2008-10-07 17:48:33 +0000	[diff] [blame]	2506	<result> = invoke [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] <ptr to function ty> <function ptr val>(<function args>) [<a href="#fnattrs">fn attrs</a>]
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2507	to label <normal label> unwind label <exception label>
				2508	</pre>
				2509
				2510	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2511	<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2512	function, with the possibility of control flow transfer to either the
				2513	'<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
				2514	function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
				2515	control flow will return to the "normal" label. If the callee (or any
				2516	indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
				2517	instruction, control is interrupted and continued at the dynamically nearest
				2518	"exception" label.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2519
				2520	<h5>Arguments:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2521	<p>This instruction requires several arguments:</p>
				2522
				2523	<ol>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2524	<li>The optional "cconv" marker indicates which <a href="#callingconv">calling
				2525	convention</a> the call should use. If none is specified, the call
				2526	defaults to using C calling conventions.</li>
Devang Patel	ac2fc27	2008-10-06 18:50:38 +0000	[diff] [blame]	2527
				2528	<li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2529	return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
				2530	'<tt>inreg</tt>' attributes are valid here.</li>
Devang Patel	ac2fc27	2008-10-06 18:50:38 +0000	[diff] [blame]	2531
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2532	<li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2533	function value being invoked. In most cases, this is a direct function
				2534	invocation, but indirect <tt>invoke</tt>s are just as possible, branching
				2535	off an arbitrary pointer to function value.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2536
				2537	<li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2538	function to be invoked. </li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2539
				2540	<li>'<tt>function args</tt>': argument list whose types match the function
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2541	signature argument types. If the function signature indicates the
				2542	function accepts a variable number of arguments, the extra arguments can
				2543	be specified.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2544
				2545	<li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2546	executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2547
				2548	<li>'<tt>exception label</tt>': the label reached when a callee returns with
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2549	the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2550
Devang Patel	d0bfcc7	2008-10-07 17:48:33 +0000	[diff] [blame]	2551	<li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2552	'<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
				2553	'<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2554	</ol>
				2555
				2556	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2557	<p>This instruction is designed to operate as a standard
				2558	'<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
				2559	primary difference is that it establishes an association with a label, which
				2560	is used by the runtime library to unwind the stack.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2561
				2562	<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2563	cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
				2564	exception. Additionally, this is important for implementation of
				2565	'<tt>catch</tt>' clauses in high-level languages that support them.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2566
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2567	<p>For the purposes of the SSA form, the definition of the value returned by the
				2568	'<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
				2569	block to the "normal" label. If the callee unwinds then no return value is
				2570	available.</p>
Dan Gohman	140ba5d	2009-05-22 21:47:08 +0000	[diff] [blame]	2571
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2572	<h5>Example:</h5>
				2573	<pre>
Nick Lewycky	a1c11a1	2008-03-16 07:18:12 +0000	[diff] [blame]	2574	%retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2575	unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewycky	a1c11a1	2008-03-16 07:18:12 +0000	[diff] [blame]	2576	%retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2577	unwind label %TestCleanup <i>; {i32}:retval set</i>
				2578	</pre>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2579
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2580	</div>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2581
				2582	<!-- _______________________________________________________________________ -->
				2583
				2584	<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
				2585	Instruction</a> </div>
				2586
				2587	<div class="doc_text">
				2588
				2589	<h5>Syntax:</h5>
				2590	<pre>
				2591	unwind
				2592	</pre>
				2593
				2594	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2595	<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2596	at the first callee in the dynamic call stack which used
				2597	an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
				2598	This is primarily used to implement exception handling.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2599
				2600	<h5>Semantics:</h5>
Chris Lattner	8b094fc	2008-04-19 21:01:16 +0000	[diff] [blame]	2601	<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2602	immediately halt. The dynamic call stack is then searched for the
				2603	first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
				2604	Once found, execution continues at the "exceptional" destination block
				2605	specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
				2606	instruction in the dynamic call chain, undefined behavior results.</p>
				2607
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2608	</div>
				2609
				2610	<!-- _______________________________________________________________________ -->
				2611
				2612	<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
				2613	Instruction</a> </div>
				2614
				2615	<div class="doc_text">
				2616
				2617	<h5>Syntax:</h5>
				2618	<pre>
				2619	unreachable
				2620	</pre>
				2621
				2622	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2623	<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2624	instruction is used to inform the optimizer that a particular portion of the
				2625	code is not reachable. This can be used to indicate that the code after a
				2626	no-return function cannot be reached, and other facts.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2627
				2628	<h5>Semantics:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2629	<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2630
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2631	</div>
				2632
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2633	<!-- ======================================================================= -->
				2634	<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2635
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2636	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2637
				2638	<p>Binary operators are used to do most of the computation in a program. They
				2639	require two operands of the same type, execute an operation on them, and
				2640	produce a single value. The operands might represent multiple data, as is
				2641	the case with the <a href="#t_vector">vector</a> data type. The result value
				2642	has the same type as its operands.</p>
				2643
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2644	<p>There are several different binary operators:</p>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2645
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2646	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2647
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2648	<!-- _______________________________________________________________________ -->
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2649	<div class="doc_subsubsection">
				2650	<a name="i_add">'<tt>add</tt>' Instruction</a>
				2651	</div>
				2652
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2653	<div class="doc_text">
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2654
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2655	<h5>Syntax:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2656	<pre>
Dan Gohman	ed5fcb2	2009-07-20 22:41:19 +0000	[diff] [blame]	2657	<result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	46e9601	2009-07-22 22:44:56 +0000	[diff] [blame]	2658	<result> = nuw add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2659	<result> = nsw add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2660	<result> = nuw nsw add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2661	</pre>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2662
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2663	<h5>Overview:</h5>
				2664	<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2665
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2666	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2667	<p>The two arguments to the '<tt>add</tt>' instruction must
				2668	be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
				2669	integer values. Both arguments must have identical types.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2670
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2671	<h5>Semantics:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2672	<p>The value produced is the integer sum of the two operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2673
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2674	<p>If the sum has unsigned overflow, the result returned is the mathematical
				2675	result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2676
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2677	<p>Because LLVM integers use a two's complement representation, this instruction
				2678	is appropriate for both signed and unsigned integers.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2679
Dan Gohman	46e9601	2009-07-22 22:44:56 +0000	[diff] [blame]	2680	<p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
				2681	and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
				2682	<tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
				2683	is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
Dan Gohman	ed5fcb2	2009-07-20 22:41:19 +0000	[diff] [blame]	2684
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2685	<h5>Example:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2686	<pre>
				2687	<result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2688	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2689
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2690	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2691
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2692	<!-- _______________________________________________________________________ -->
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2693	<div class="doc_subsubsection">
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2694	<a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
				2695	</div>
				2696
				2697	<div class="doc_text">
				2698
				2699	<h5>Syntax:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2700	<pre>
				2701	<result> = fadd <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2702	</pre>
				2703
				2704	<h5>Overview:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2705	<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
				2706
				2707	<h5>Arguments:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2708	<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2709	<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
				2710	floating point values. Both arguments must have identical types.</p>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2711
				2712	<h5>Semantics:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2713	<p>The value produced is the floating point sum of the two operands.</p>
				2714
				2715	<h5>Example:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2716	<pre>
				2717	<result> = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
				2718	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2719
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2720	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2721
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2722	<!-- _______________________________________________________________________ -->
				2723	<div class="doc_subsubsection">
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2724	<a name="i_sub">'<tt>sub</tt>' Instruction</a>
				2725	</div>
				2726
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2727	<div class="doc_text">
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2728
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2729	<h5>Syntax:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2730	<pre>
Dan Gohman	46e9601	2009-07-22 22:44:56 +0000	[diff] [blame]	2731	<result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2732	<result> = nuw sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2733	<result> = nsw sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2734	<result> = nuw nsw sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2735	</pre>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2736
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2737	<h5>Overview:</h5>
				2738	<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2739	operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2740
				2741	<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2742	'<tt>neg</tt>' instruction present in most other intermediate
				2743	representations.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2744
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2745	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2746	<p>The two arguments to the '<tt>sub</tt>' instruction must
				2747	be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
				2748	integer values. Both arguments must have identical types.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2749
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2750	<h5>Semantics:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2751	<p>The value produced is the integer difference of the two operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2752
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2753	<p>If the difference has unsigned overflow, the result returned is the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2754	mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
				2755	result.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2756
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2757	<p>Because LLVM integers use a two's complement representation, this instruction
				2758	is appropriate for both signed and unsigned integers.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2759
Dan Gohman	46e9601	2009-07-22 22:44:56 +0000	[diff] [blame]	2760	<p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
				2761	and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
				2762	<tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
				2763	is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
Dan Gohman	ed5fcb2	2009-07-20 22:41:19 +0000	[diff] [blame]	2764
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2765	<h5>Example:</h5>
				2766	<pre>
				2767	<result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
				2768	<result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
				2769	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2770
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2771	</div>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2772
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2773	<!-- _______________________________________________________________________ -->
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2774	<div class="doc_subsubsection">
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2775	<a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
				2776	</div>
				2777
				2778	<div class="doc_text">
				2779
				2780	<h5>Syntax:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2781	<pre>
				2782	<result> = fsub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2783	</pre>
				2784
				2785	<h5>Overview:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2786	<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2787	operands.</p>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2788
				2789	<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2790	'<tt>fneg</tt>' instruction present in most other intermediate
				2791	representations.</p>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2792
				2793	<h5>Arguments:</h5>
Bill Wendling	1a2630a	2009-07-20 02:32:41 +0000	[diff] [blame]	2794	<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2795	<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
				2796	floating point values. Both arguments must have identical types.</p>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2797
				2798	<h5>Semantics:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2799	<p>The value produced is the floating point difference of the two operands.</p>
				2800
				2801	<h5>Example:</h5>
				2802	<pre>
				2803	<result> = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
				2804	<result> = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
				2805	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2806
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2807	</div>
				2808
				2809	<!-- _______________________________________________________________________ -->
				2810	<div class="doc_subsubsection">
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2811	<a name="i_mul">'<tt>mul</tt>' Instruction</a>
				2812	</div>
				2813
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2814	<div class="doc_text">
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2815
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2816	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2817	<pre>
Dan Gohman	46e9601	2009-07-22 22:44:56 +0000	[diff] [blame]	2818	<result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2819	<result> = nuw mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2820	<result> = nsw mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2821	<result> = nuw nsw mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2822	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2823
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2824	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2825	<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2826
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2827	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2828	<p>The two arguments to the '<tt>mul</tt>' instruction must
				2829	be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
				2830	integer values. Both arguments must have identical types.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2831
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2832	<h5>Semantics:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2833	<p>The value produced is the integer product of the two operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2834
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2835	<p>If the result of the multiplication has unsigned overflow, the result
				2836	returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
				2837	width of the result.</p>
				2838
				2839	<p>Because LLVM integers use a two's complement representation, and the result
				2840	is the same width as the operands, this instruction returns the correct
				2841	result for both signed and unsigned integers. If a full product
				2842	(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
				2843	be sign-extended or zero-extended as appropriate to the width of the full
				2844	product.</p>
				2845
Dan Gohman	46e9601	2009-07-22 22:44:56 +0000	[diff] [blame]	2846	<p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
				2847	and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
				2848	<tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
				2849	is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
Dan Gohman	ed5fcb2	2009-07-20 22:41:19 +0000	[diff] [blame]	2850
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2851	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2852	<pre>
				2853	<result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2854	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2855
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2856	</div>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2857
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2858	<!-- _______________________________________________________________________ -->
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2859	<div class="doc_subsubsection">
				2860	<a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
				2861	</div>
				2862
				2863	<div class="doc_text">
				2864
				2865	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2866	<pre>
				2867	<result> = fmul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2868	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2869
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2870	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2871	<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2872
				2873	<h5>Arguments:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2874	<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2875	<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
				2876	floating point values. Both arguments must have identical types.</p>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2877
				2878	<h5>Semantics:</h5>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2879	<p>The value produced is the floating point product of the two operands.</p>
				2880
				2881	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2882	<pre>
				2883	<result> = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2884	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2885
Dan Gohman	7ce405e	2009-06-04 22:49:04 +0000	[diff] [blame]	2886	</div>
				2887
				2888	<!-- _______________________________________________________________________ -->
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2889	<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
				2890	</a></div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2891
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2892	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2893
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2894	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2895	<pre>
				2896	<result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2897	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2898
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2899	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2900	<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2901
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2902	<h5>Arguments:</h5>
				2903	<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2904	<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
				2905	values. Both arguments must have identical types.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2906
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2907	<h5>Semantics:</h5>
Chris Lattner	9aba1e2	2008-01-28 00:36:27 +0000	[diff] [blame]	2908	<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2909
Chris Lattner	9aba1e2	2008-01-28 00:36:27 +0000	[diff] [blame]	2910	<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2911	operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
				2912
Chris Lattner	9aba1e2	2008-01-28 00:36:27 +0000	[diff] [blame]	2913	<p>Division by zero leads to undefined behavior.</p>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2914
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2915	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2916	<pre>
				2917	<result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2918	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2919
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2920	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2921
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2922	<!-- _______________________________________________________________________ -->
				2923	<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
				2924	</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2925
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2926	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2927
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2928	<h5>Syntax:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2929	<pre>
Dan Gohman	ed5fcb2	2009-07-20 22:41:19 +0000	[diff] [blame]	2930	<result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				2931	<result> = exact sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2932	</pre>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2933
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2934	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2935	<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2936
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2937	<h5>Arguments:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2938	<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2939	<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
				2940	values. Both arguments must have identical types.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2941
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2942	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2943	<p>The value produced is the signed integer quotient of the two operands rounded
				2944	towards zero.</p>
				2945
Chris Lattner	9aba1e2	2008-01-28 00:36:27 +0000	[diff] [blame]	2946	<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2947	operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
				2948
Chris Lattner	9aba1e2	2008-01-28 00:36:27 +0000	[diff] [blame]	2949	<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2950	undefined behavior; this is a rare case, but can occur, for example, by doing
				2951	a 32-bit division of -2147483648 by -1.</p>
				2952
Dan Gohman	67fa48e	2009-07-22 00:04:19 +0000	[diff] [blame]	2953	<p>If the <tt>exact</tt> keyword is present, the result value of the
				2954	<tt>sdiv</tt> is undefined if the result would be rounded or if overflow
				2955	would occur.</p>
Dan Gohman	ed5fcb2	2009-07-20 22:41:19 +0000	[diff] [blame]	2956
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2957	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2958	<pre>
				2959	<result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2960	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2961
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2962	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2963
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2964	<!-- _______________________________________________________________________ -->
				2965	<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
				2966	Instruction</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2967
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2968	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2969
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2970	<h5>Syntax:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2971	<pre>
Gabor Greif	d9068fe	2008-08-07 21:46:00 +0000	[diff] [blame]	2972	<result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2973	</pre>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2974
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2975	<h5>Overview:</h5>
				2976	<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2977
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2978	<h5>Arguments:</h5>
				2979	<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2980	<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
				2981	floating point values. Both arguments must have identical types.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2982
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2983	<h5>Semantics:</h5>
				2984	<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2985
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2986	<h5>Example:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2987	<pre>
				2988	<result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2989	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2990
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2991	</div>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	2992
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2993	<!-- _______________________________________________________________________ -->
				2994	<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
				2995	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2996
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2997	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	2998
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2999	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3000	<pre>
				3001	<result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3002	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3003
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3004	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3005	<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
				3006	division of its two arguments.</p>
				3007
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3008	<h5>Arguments:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3009	<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3010	<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
				3011	values. Both arguments must have identical types.</p>
				3012
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3013	<h5>Semantics:</h5>
				3014	<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3015	This instruction always performs an unsigned division to get the
				3016	remainder.</p>
				3017
Chris Lattner	9aba1e2	2008-01-28 00:36:27 +0000	[diff] [blame]	3018	<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3019	distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
				3020
Chris Lattner	9aba1e2	2008-01-28 00:36:27 +0000	[diff] [blame]	3021	<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3022
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3023	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3024	<pre>
				3025	<result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3026	</pre>
				3027
				3028	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3029
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3030	<!-- _______________________________________________________________________ -->
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3031	<div class="doc_subsubsection">
				3032	<a name="i_srem">'<tt>srem</tt>' Instruction</a>
				3033	</div>
				3034
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3035	<div class="doc_text">
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3036
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3037	<h5>Syntax:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3038	<pre>
Gabor Greif	d9068fe	2008-08-07 21:46:00 +0000	[diff] [blame]	3039	<result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3040	</pre>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3041
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3042	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3043	<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
				3044	division of its two operands. This instruction can also take
				3045	<a href="#t_vector">vector</a> versions of the values in which case the
				3046	elements must be integers.</p>
Chris Lattner	08497ce	2008-01-04 04:33:49 +0000	[diff] [blame]	3047
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3048	<h5>Arguments:</h5>
				3049	<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3050	<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
				3051	values. Both arguments must have identical types.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3052
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3053	<h5>Semantics:</h5>
				3054	<p>This instruction returns the <i>remainder</i> of a division (where the result
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3055	has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
				3056	operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
				3057	a value. For more information about the difference,
				3058	see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
				3059	Math Forum</a>. For a table of how this is implemented in various languages,
				3060	please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
				3061	Wikipedia: modulo operation</a>.</p>
				3062
Chris Lattner	9aba1e2	2008-01-28 00:36:27 +0000	[diff] [blame]	3063	<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3064	distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
				3065
Chris Lattner	9aba1e2	2008-01-28 00:36:27 +0000	[diff] [blame]	3066	<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3067	Overflow also leads to undefined behavior; this is a rare case, but can
				3068	occur, for example, by taking the remainder of a 32-bit division of
				3069	-2147483648 by -1. (The remainder doesn't actually overflow, but this rule
				3070	lets srem be implemented using instructions that return both the result of
				3071	the division and the remainder.)</p>
				3072
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3073	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3074	<pre>
				3075	<result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3076	</pre>
				3077
				3078	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3079
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3080	<!-- _______________________________________________________________________ -->
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3081	<div class="doc_subsubsection">
				3082	<a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
				3083
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3084	<div class="doc_text">
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3085
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3086	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3087	<pre>
				3088	<result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3089	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3090
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3091	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3092	<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
				3093	its two operands.</p>
				3094
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3095	<h5>Arguments:</h5>
				3096	<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3097	<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
				3098	floating point values. Both arguments must have identical types.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3099
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3100	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3101	<p>This instruction returns the <i>remainder</i> of a division. The remainder
				3102	has the same sign as the dividend.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3103
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3104	<h5>Example:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3105	<pre>
				3106	<result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3107	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3108
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3109	</div>
				3110
				3111	<!-- ======================================================================= -->
				3112	<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
				3113	Operations</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3114
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3115	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3116
				3117	<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
				3118	program. They are generally very efficient instructions and can commonly be
				3119	strength reduced from other instructions. They require two operands of the
				3120	same type, execute an operation on them, and produce a single value. The
				3121	resulting value is the same type as its operands.</p>
				3122
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3123	</div>
				3124
				3125	<!-- _______________________________________________________________________ -->
				3126	<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
				3127	Instruction</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3128
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3129	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3130
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3131	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3132	<pre>
				3133	<result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3134	</pre>
Chris Lattner	d939d9f	2007-10-03 21:01:14 +0000	[diff] [blame]	3135
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3136	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3137	<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
				3138	a specified number of bits.</p>
Chris Lattner	d939d9f	2007-10-03 21:01:14 +0000	[diff] [blame]	3139
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3140	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3141	<p>Both arguments to the '<tt>shl</tt>' instruction must be the
				3142	same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
				3143	integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattner	d939d9f	2007-10-03 21:01:14 +0000	[diff] [blame]	3144
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3145	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3146	<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
				3147	2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
				3148	is (statically or dynamically) negative or equal to or larger than the number
				3149	of bits in <tt>op1</tt>, the result is undefined. If the arguments are
				3150	vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
				3151	shift amount in <tt>op2</tt>.</p>
Chris Lattner	d939d9f	2007-10-03 21:01:14 +0000	[diff] [blame]	3152
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3153	<h5>Example:</h5>
				3154	<pre>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3155	<result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
				3156	<result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
				3157	<result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattner	d939d9f	2007-10-03 21:01:14 +0000	[diff] [blame]	3158	<result> = shl i32 1, 32 <i>; undefined</i>
Mon P Wang	9901e73	2008-12-09 05:46:39 +0000	[diff] [blame]	3159	<result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3160	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3161
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3162	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3163
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3164	<!-- _______________________________________________________________________ -->
				3165	<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
				3166	Instruction</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3167
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3168	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3169
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3170	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3171	<pre>
				3172	<result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3173	</pre>
				3174
				3175	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3176	<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
				3177	operand shifted to the right a specified number of bits with zero fill.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3178
				3179	<h5>Arguments:</h5>
				3180	<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3181	<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
				3182	type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3183
				3184	<h5>Semantics:</h5>
				3185	<p>This instruction always performs a logical shift right operation. The most
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3186	significant bits of the result will be filled with zero bits after the shift.
				3187	If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
				3188	number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
				3189	vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
				3190	shift amount in <tt>op2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3191
				3192	<h5>Example:</h5>
				3193	<pre>
				3194	<result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
				3195	<result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
				3196	<result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
				3197	<result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattner	d939d9f	2007-10-03 21:01:14 +0000	[diff] [blame]	3198	<result> = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang	9901e73	2008-12-09 05:46:39 +0000	[diff] [blame]	3199	<result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3200	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3201
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3202	</div>
				3203
				3204	<!-- _______________________________________________________________________ -->
				3205	<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
				3206	Instruction</a> </div>
				3207	<div class="doc_text">
				3208
				3209	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3210	<pre>
				3211	<result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3212	</pre>
				3213
				3214	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3215	<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
				3216	operand shifted to the right a specified number of bits with sign
				3217	extension.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3218
				3219	<h5>Arguments:</h5>
				3220	<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3221	<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
				3222	type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3223
				3224	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3225	<p>This instruction always performs an arithmetic shift right operation, The
				3226	most significant bits of the result will be filled with the sign bit
				3227	of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
				3228	larger than the number of bits in <tt>op1</tt>, the result is undefined. If
				3229	the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
				3230	the corresponding shift amount in <tt>op2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3231
				3232	<h5>Example:</h5>
				3233	<pre>
				3234	<result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
				3235	<result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
				3236	<result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
				3237	<result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattner	d939d9f	2007-10-03 21:01:14 +0000	[diff] [blame]	3238	<result> = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang	9901e73	2008-12-09 05:46:39 +0000	[diff] [blame]	3239	<result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3240	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3241
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3242	</div>
				3243
				3244	<!-- _______________________________________________________________________ -->
				3245	<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
				3246	Instruction</a> </div>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3247
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3248	<div class="doc_text">
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3249
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3250	<h5>Syntax:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3251	<pre>
Gabor Greif	d9068fe	2008-08-07 21:46:00 +0000	[diff] [blame]	3252	<result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3253	</pre>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3254
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3255	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3256	<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
				3257	operands.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3258
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3259	<h5>Arguments:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3260	<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3261	<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
				3262	values. Both arguments must have identical types.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3263
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3264	<h5>Semantics:</h5>
				3265	<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3266
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3267	<table border="1" cellspacing="0" cellpadding="4">
				3268	<tbody>
				3269	<tr>
				3270	<td>In0</td>
				3271	<td>In1</td>
				3272	<td>Out</td>
				3273	</tr>
				3274	<tr>
				3275	<td>0</td>
				3276	<td>0</td>
				3277	<td>0</td>
				3278	</tr>
				3279	<tr>
				3280	<td>0</td>
				3281	<td>1</td>
				3282	<td>0</td>
				3283	</tr>
				3284	<tr>
				3285	<td>1</td>
				3286	<td>0</td>
				3287	<td>0</td>
				3288	</tr>
				3289	<tr>
				3290	<td>1</td>
				3291	<td>1</td>
				3292	<td>1</td>
				3293	</tr>
				3294	</tbody>
				3295	</table>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3296
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3297	<h5>Example:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3298	<pre>
				3299	<result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3300	<result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
				3301	<result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
				3302	</pre>
				3303	</div>
				3304	<!-- _______________________________________________________________________ -->
				3305	<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3306
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3307	<div class="doc_text">
				3308
				3309	<h5>Syntax:</h5>
				3310	<pre>
				3311	<result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
				3312	</pre>
				3313
				3314	<h5>Overview:</h5>
				3315	<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
				3316	two operands.</p>
				3317
				3318	<h5>Arguments:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3319	<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3320	<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
				3321	values. Both arguments must have identical types.</p>
				3322
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3323	<h5>Semantics:</h5>
				3324	<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3325
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3326	<table border="1" cellspacing="0" cellpadding="4">
				3327	<tbody>
				3328	<tr>
				3329	<td>In0</td>
				3330	<td>In1</td>
				3331	<td>Out</td>
				3332	</tr>
				3333	<tr>
				3334	<td>0</td>
				3335	<td>0</td>
				3336	<td>0</td>
				3337	</tr>
				3338	<tr>
				3339	<td>0</td>
				3340	<td>1</td>
				3341	<td>1</td>
				3342	</tr>
				3343	<tr>
				3344	<td>1</td>
				3345	<td>0</td>
				3346	<td>1</td>
				3347	</tr>
				3348	<tr>
				3349	<td>1</td>
				3350	<td>1</td>
				3351	<td>1</td>
				3352	</tr>
				3353	</tbody>
				3354	</table>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3355
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3356	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3357	<pre>
				3358	<result> = or i32 4, %var <i>; yields {i32}:result = 4 \| %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3359	<result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
				3360	<result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
				3361	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3362
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3363	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3364
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3365	<!-- _______________________________________________________________________ -->
				3366	<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
				3367	Instruction</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3368
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3369	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3370
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3371	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3372	<pre>
				3373	<result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3374	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3375
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3376	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3377	<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
				3378	its two operands. The <tt>xor</tt> is used to implement the "one's
				3379	complement" operation, which is the "~" operator in C.</p>
				3380
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3381	<h5>Arguments:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3382	<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3383	<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
				3384	values. Both arguments must have identical types.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	3385
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3386	<h5>Semantics:</h5>
				3387	<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3388
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3389	<table border="1" cellspacing="0" cellpadding="4">
				3390	<tbody>
				3391	<tr>
				3392	<td>In0</td>
				3393	<td>In1</td>
				3394	<td>Out</td>
				3395	</tr>
				3396	<tr>
				3397	<td>0</td>
				3398	<td>0</td>
				3399	<td>0</td>
				3400	</tr>
				3401	<tr>
				3402	<td>0</td>
				3403	<td>1</td>
				3404	<td>1</td>
				3405	</tr>
				3406	<tr>
				3407	<td>1</td>
				3408	<td>0</td>
				3409	<td>1</td>
				3410	</tr>
				3411	<tr>
				3412	<td>1</td>
				3413	<td>1</td>
				3414	<td>0</td>
				3415	</tr>
				3416	</tbody>
				3417	</table>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3418
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3419	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3420	<pre>
				3421	<result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3422	<result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
				3423	<result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
				3424	<result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
				3425	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3426
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3427	</div>
				3428
				3429	<!-- ======================================================================= -->
				3430	<div class="doc_subsection">
				3431	<a name="vectorops">Vector Operations</a>
				3432	</div>
				3433
				3434	<div class="doc_text">
				3435
				3436	<p>LLVM supports several instructions to represent vector operations in a
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3437	target-independent manner. These instructions cover the element-access and
				3438	vector-specific operations needed to process vectors effectively. While LLVM
				3439	does directly support these vector operations, many sophisticated algorithms
				3440	will want to use target-specific intrinsics to take full advantage of a
				3441	specific target.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3442
				3443	</div>
				3444
				3445	<!-- _______________________________________________________________________ -->
				3446	<div class="doc_subsubsection">
				3447	<a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
				3448	</div>
				3449
				3450	<div class="doc_text">
				3451
				3452	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3453	<pre>
				3454	<result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
				3455	</pre>
				3456
				3457	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3458	<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
				3459	from a vector at a specified index.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3460
				3461
				3462	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3463	<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
				3464	of <a href="#t_vector">vector</a> type. The second operand is an index
				3465	indicating the position from which to extract the element. The index may be
				3466	a variable.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3467
				3468	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3469	<p>The result is a scalar of the same type as the element type of
				3470	<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
				3471	<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
				3472	results are undefined.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3473
				3474	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3475	<pre>
				3476	%result = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
				3477	</pre>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3478
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3479	</div>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3480
				3481	<!-- _______________________________________________________________________ -->
				3482	<div class="doc_subsubsection">
				3483	<a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
				3484	</div>
				3485
				3486	<div class="doc_text">
				3487
				3488	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3489	<pre>
Dan Gohman	bcc3c50	2008-05-12 23:38:42 +0000	[diff] [blame]	3490	<result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3491	</pre>
				3492
				3493	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3494	<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
				3495	vector at a specified index.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3496
				3497	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3498	<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
				3499	of <a href="#t_vector">vector</a> type. The second operand is a scalar value
				3500	whose type must equal the element type of the first operand. The third
				3501	operand is an index indicating the position at which to insert the value.
				3502	The index may be a variable.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3503
				3504	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3505	<p>The result is a vector of the same type as <tt>val</tt>. Its element values
				3506	are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
				3507	value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
				3508	results are undefined.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3509
				3510	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3511	<pre>
				3512	%result = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
				3513	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3514
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3515	</div>
				3516
				3517	<!-- _______________________________________________________________________ -->
				3518	<div class="doc_subsubsection">
				3519	<a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
				3520	</div>
				3521
				3522	<div class="doc_text">
				3523
				3524	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3525	<pre>
Mon P Wang	bff5d9c	2008-11-10 04:46:22 +0000	[diff] [blame]	3526	<result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3527	</pre>
				3528
				3529	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3530	<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
				3531	from two input vectors, returning a vector with the same element type as the
				3532	input and length that is the same as the shuffle mask.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3533
				3534	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3535	<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
				3536	with types that match each other. The third argument is a shuffle mask whose
				3537	element type is always 'i32'. The result of the instruction is a vector
				3538	whose length is the same as the shuffle mask and whose element type is the
				3539	same as the element type of the first two operands.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3540
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3541	<p>The shuffle mask operand is required to be a constant vector with either
				3542	constant integer or undef values.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3543
				3544	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3545	<p>The elements of the two input vectors are numbered from left to right across
				3546	both of the vectors. The shuffle mask operand specifies, for each element of
				3547	the result vector, which element of the two input vectors the result element
				3548	gets. The element selector may be undef (meaning "don't care") and the
				3549	second operand may be undef if performing a shuffle from only one vector.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3550
				3551	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3552	<pre>
				3553	%result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
				3554	<4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
				3555	%result = shufflevector <4 x i32> %v1, <4 x i32> undef,
				3556	<4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
Mon P Wang	bff5d9c	2008-11-10 04:46:22 +0000	[diff] [blame]	3557	%result = shufflevector <8 x i32> %v1, <8 x i32> undef,
				3558	<4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
				3559	%result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
				3560	<8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 > <i>; yields <8 x i32></i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3561	</pre>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3562
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3563	</div>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3564
				3565	<!-- ======================================================================= -->
				3566	<div class="doc_subsection">
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3567	<a name="aggregateops">Aggregate Operations</a>
				3568	</div>
				3569
				3570	<div class="doc_text">
				3571
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3572	<p>LLVM supports several instructions for working with aggregate values.</p>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3573
				3574	</div>
				3575
				3576	<!-- _______________________________________________________________________ -->
				3577	<div class="doc_subsubsection">
				3578	<a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
				3579	</div>
				3580
				3581	<div class="doc_text">
				3582
				3583	<h5>Syntax:</h5>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3584	<pre>
				3585	<result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
				3586	</pre>
				3587
				3588	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3589	<p>The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
				3590	or array element from an aggregate value.</p>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3591
				3592	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3593	<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
				3594	of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
				3595	operands are constant indices to specify which value to extract in a similar
				3596	manner as indices in a
				3597	'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3598
				3599	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3600	<p>The result is the value at the position in the aggregate specified by the
				3601	index operands.</p>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3602
				3603	<h5>Example:</h5>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3604	<pre>
Dan Gohman	e5febe4	2008-05-31 00:58:22 +0000	[diff] [blame]	3605	%result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3606	</pre>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3607
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3608	</div>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3609
				3610	<!-- _______________________________________________________________________ -->
				3611	<div class="doc_subsubsection">
				3612	<a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
				3613	</div>
				3614
				3615	<div class="doc_text">
				3616
				3617	<h5>Syntax:</h5>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3618	<pre>
Dan Gohman	e5febe4	2008-05-31 00:58:22 +0000	[diff] [blame]	3619	<result> = insertvalue <aggregate type> <val>, <ty> <val>, <idx> <i>; yields <n x <ty>></i>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3620	</pre>
				3621
				3622	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3623	<p>The '<tt>insertvalue</tt>' instruction inserts a value into a struct field or
				3624	array element in an aggregate.</p>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3625
				3626
				3627	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3628	<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
				3629	of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
				3630	second operand is a first-class value to insert. The following operands are
				3631	constant indices indicating the position at which to insert the value in a
				3632	similar manner as indices in a
				3633	'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
				3634	value to insert must have the same type as the value identified by the
				3635	indices.</p>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3636
				3637	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3638	<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
				3639	that of <tt>val</tt> except that the value at the position specified by the
				3640	indices is that of <tt>elt</tt>.</p>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3641
				3642	<h5>Example:</h5>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3643	<pre>
Dan Gohman	b1aab4e	2008-06-23 15:26:37 +0000	[diff] [blame]	3644	%result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3645	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3646
Dan Gohman	74d6faf	2008-05-12 23:51:09 +0000	[diff] [blame]	3647	</div>
				3648
				3649
				3650	<!-- ======================================================================= -->
				3651	<div class="doc_subsection">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3652	<a name="memoryops">Memory Access and Addressing Operations</a>
				3653	</div>
				3654
				3655	<div class="doc_text">
				3656
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3657	<p>A key design point of an SSA-based representation is how it represents
				3658	memory. In LLVM, no memory locations are in SSA form, which makes things
				3659	very simple. This section describes how to read, write, allocate, and free
				3660	memory in LLVM.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3661
				3662	</div>
				3663
				3664	<!-- _______________________________________________________________________ -->
				3665	<div class="doc_subsubsection">
				3666	<a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
				3667	</div>
				3668
				3669	<div class="doc_text">
				3670
				3671	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3672	<pre>
				3673	<result> = malloc <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
				3674	</pre>
				3675
				3676	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3677	<p>The '<tt>malloc</tt>' instruction allocates memory from the system heap and
				3678	returns a pointer to it. The object is always allocated in the generic
				3679	address space (address space zero).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3680
				3681	<h5>Arguments:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3682	<p>The '<tt>malloc</tt>' instruction allocates
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3683	<tt>sizeof(<type>)*NumElements</tt> bytes of memory from the operating
				3684	system and returns a pointer of the appropriate type to the program. If
				3685	"NumElements" is specified, it is the number of elements allocated, otherwise
				3686	"NumElements" is defaulted to be one. If a constant alignment is specified,
				3687	the value result of the allocation is guaranteed to be aligned to at least
				3688	that boundary. If not specified, or if zero, the target can choose to align
				3689	the allocation on any convenient boundary compatible with the type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3690
				3691	<p>'<tt>type</tt>' must be a sized type.</p>
				3692
				3693	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3694	<p>Memory is allocated using the system "<tt>malloc</tt>" function, and a
				3695	pointer is returned. The result of a zero byte allocation is undefined. The
				3696	result is null if there is insufficient memory available.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3697
				3698	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3699	<pre>
Dan Gohman	f54f50a	2009-01-04 23:49:44 +0000	[diff] [blame]	3700	%array = malloc [4 x i8] <i>; yields {[%4 x i8]*}:array</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3701
				3702	%size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
				3703	%array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
				3704	%array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
				3705	%array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
				3706	%array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
				3707	</pre>
Dan Gohman	6096719	2009-01-12 23:12:39 +0000	[diff] [blame]	3708
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3709	<p>Note that the code generator does not yet respect the alignment value.</p>
Dan Gohman	6096719	2009-01-12 23:12:39 +0000	[diff] [blame]	3710
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3711	</div>
				3712
				3713	<!-- _______________________________________________________________________ -->
				3714	<div class="doc_subsubsection">
				3715	<a name="i_free">'<tt>free</tt>' Instruction</a>
				3716	</div>
				3717
				3718	<div class="doc_text">
				3719
				3720	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3721	<pre>
Dan Gohman	f54f50a	2009-01-04 23:49:44 +0000	[diff] [blame]	3722	free <type> <value> <i>; yields {void}</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3723	</pre>
				3724
				3725	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3726	<p>The '<tt>free</tt>' instruction returns memory back to the unused memory heap
				3727	to be reallocated in the future.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3728
				3729	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3730	<p>'<tt>value</tt>' shall be a pointer value that points to a value that was
				3731	allocated with the '<tt><a href="#i_malloc">malloc</a></tt>' instruction.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3732
				3733	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3734	<p>Access to the memory pointed to by the pointer is no longer defined after
				3735	this instruction executes. If the pointer is null, the operation is a
				3736	noop.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3737
				3738	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3739	<pre>
Dan Gohman	f54f50a	2009-01-04 23:49:44 +0000	[diff] [blame]	3740	%array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3741	free [4 x i8]* %array
				3742	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3743
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3744	</div>
				3745
				3746	<!-- _______________________________________________________________________ -->
				3747	<div class="doc_subsubsection">
				3748	<a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
				3749	</div>
				3750
				3751	<div class="doc_text">
				3752
				3753	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3754	<pre>
				3755	<result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
				3756	</pre>
				3757
				3758	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3759	<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3760	currently executing function, to be automatically released when this function
				3761	returns to its caller. The object is always allocated in the generic address
				3762	space (address space zero).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3763
				3764	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3765	<p>The '<tt>alloca</tt>' instruction
				3766	allocates <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the
				3767	runtime stack, returning a pointer of the appropriate type to the program.
				3768	If "NumElements" is specified, it is the number of elements allocated,
				3769	otherwise "NumElements" is defaulted to be one. If a constant alignment is
				3770	specified, the value result of the allocation is guaranteed to be aligned to
				3771	at least that boundary. If not specified, or if zero, the target can choose
				3772	to align the allocation on any convenient boundary compatible with the
				3773	type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3774
				3775	<p>'<tt>type</tt>' may be any sized type.</p>
				3776
				3777	<h5>Semantics:</h5>
Bill Wendling	2a45457	2009-05-08 20:49:29 +0000	[diff] [blame]	3778	<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3779	there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
				3780	memory is automatically released when the function returns. The
				3781	'<tt>alloca</tt>' instruction is commonly used to represent automatic
				3782	variables that must have an address available. When the function returns
				3783	(either with the <tt><a href="#i_ret">ret</a></tt>
				3784	or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
				3785	reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3786
				3787	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3788	<pre>
Dan Gohman	f54f50a	2009-01-04 23:49:44 +0000	[diff] [blame]	3789	%ptr = alloca i32 <i>; yields {i32*}:ptr</i>
				3790	%ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
				3791	%ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
				3792	%ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3793	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3794
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3795	</div>
				3796
				3797	<!-- _______________________________________________________________________ -->
				3798	<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
				3799	Instruction</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3800
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3801	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3802
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3803	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3804	<pre>
				3805	<result> = load <ty>* <pointer>[, align <alignment>]
				3806	<result> = volatile load <ty>* <pointer>[, align <alignment>]
				3807	</pre>
				3808
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3809	<h5>Overview:</h5>
				3810	<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3811
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3812	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3813	<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
				3814	from which to load. The pointer must point to
				3815	a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
				3816	marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
				3817	number or order of execution of this <tt>load</tt> with other
				3818	volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
				3819	instructions. </p>
				3820
				3821	<p>The optional constant "align" argument specifies the alignment of the
				3822	operation (that is, the alignment of the memory address). A value of 0 or an
				3823	omitted "align" argument means that the operation has the preferential
				3824	alignment for the target. It is the responsibility of the code emitter to
				3825	ensure that the alignment information is correct. Overestimating the
				3826	alignment results in an undefined behavior. Underestimating the alignment may
				3827	produce less efficient code. An alignment of 1 is always safe.</p>
				3828
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3829	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3830	<p>The location of memory pointed to is loaded. If the value being loaded is of
				3831	scalar type then the number of bytes read does not exceed the minimum number
				3832	of bytes needed to hold all bits of the type. For example, loading an
				3833	<tt>i24</tt> reads at most three bytes. When loading a value of a type like
				3834	<tt>i20</tt> with a size that is not an integral number of bytes, the result
				3835	is undefined if the value was not originally written using a store of the
				3836	same type.</p>
				3837
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3838	<h5>Examples:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3839	<pre>
				3840	%ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
				3841	<a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3842	%val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
				3843	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3844
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3845	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3846
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3847	<!-- _______________________________________________________________________ -->
				3848	<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
				3849	Instruction</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3850
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3851	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3852
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3853	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3854	<pre>
				3855	store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3856	volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
				3857	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3858
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3859	<h5>Overview:</h5>
				3860	<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3861
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3862	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3863	<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
				3864	and an address at which to store it. The type of the
				3865	'<tt><pointer></tt>' operand must be a pointer to
				3866	the <a href="#t_firstclass">first class</a> type of the
				3867	'<tt><value></tt>' operand. If the <tt>store</tt> is marked
				3868	as <tt>volatile</tt>, then the optimizer is not allowed to modify the number
				3869	or order of execution of this <tt>store</tt> with other
				3870	volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
				3871	instructions.</p>
				3872
				3873	<p>The optional constant "align" argument specifies the alignment of the
				3874	operation (that is, the alignment of the memory address). A value of 0 or an
				3875	omitted "align" argument means that the operation has the preferential
				3876	alignment for the target. It is the responsibility of the code emitter to
				3877	ensure that the alignment information is correct. Overestimating the
				3878	alignment results in an undefined behavior. Underestimating the alignment may
				3879	produce less efficient code. An alignment of 1 is always safe.</p>
				3880
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3881	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3882	<p>The contents of memory are updated to contain '<tt><value></tt>' at the
				3883	location specified by the '<tt><pointer></tt>' operand. If
				3884	'<tt><value></tt>' is of scalar type then the number of bytes written
				3885	does not exceed the minimum number of bytes needed to hold all bits of the
				3886	type. For example, storing an <tt>i24</tt> writes at most three bytes. When
				3887	writing a value of a type like <tt>i20</tt> with a size that is not an
				3888	integral number of bytes, it is unspecified what happens to the extra bits
				3889	that do not belong to the type, but they will typically be overwritten.</p>
				3890
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3891	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3892	<pre>
				3893	%ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling	63ffa14	2007-10-22 05:10:05 +0000	[diff] [blame]	3894	store i32 3, i32* %ptr <i>; yields {void}</i>
				3895	%val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3896	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3897
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3898	</div>
				3899
				3900	<!-- _______________________________________________________________________ -->
				3901	<div class="doc_subsubsection">
				3902	<a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
				3903	</div>
				3904
				3905	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3906
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3907	<h5>Syntax:</h5>
				3908	<pre>
Matthijs Kooijman	32a080e	2008-10-13 13:44:15 +0000	[diff] [blame]	3909	<result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
Dan Gohman	106b2ae	2009-07-27 21:53:46 +0000	[diff] [blame]	3910	<result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3911	</pre>
				3912
				3913	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3914	<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
				3915	subelement of an aggregate data structure. It performs address calculation
				3916	only and does not access memory.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3917
				3918	<h5>Arguments:</h5>
Matthijs Kooijman	32a080e	2008-10-13 13:44:15 +0000	[diff] [blame]	3919	<p>The first argument is always a pointer, and forms the basis of the
Chris Lattner	e92fc83	2009-07-29 06:44:13 +0000	[diff] [blame]	3920	calculation. The remaining arguments are indices that indicate which of the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3921	elements of the aggregate object are indexed. The interpretation of each
				3922	index is dependent on the type being indexed into. The first index always
				3923	indexes the pointer value given as the first argument, the second index
				3924	indexes a value of the type pointed to (not necessarily the value directly
				3925	pointed to, since the first index can be non-zero), etc. The first type
				3926	indexed into must be a pointer value, subsequent types can be arrays, vectors
				3927	and structs. Note that subsequent types being indexed into can never be
				3928	pointers, since that would require loading the pointer before continuing
				3929	calculation.</p>
Matthijs Kooijman	32a080e	2008-10-13 13:44:15 +0000	[diff] [blame]	3930
				3931	<p>The type of each index argument depends on the type it is indexing into.
Chris Lattner	e92fc83	2009-07-29 06:44:13 +0000	[diff] [blame]	3932	When indexing into a (optionally packed) structure, only <tt>i32</tt> integer
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3933	<b>constants</b> are allowed. When indexing into an array, pointer or
Chris Lattner	e92fc83	2009-07-29 06:44:13 +0000	[diff] [blame]	3934	vector, integers of any width are allowed, and they are not required to be
				3935	constant.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3936
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3937	<p>For example, let's consider a C code fragment and how it gets compiled to
				3938	LLVM:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3939
				3940	<div class="doc_code">
				3941	<pre>
				3942	struct RT {
				3943	char A;
				3944	int B[10][20];
				3945	char C;
				3946	};
				3947	struct ST {
				3948	int X;
				3949	double Y;
				3950	struct RT Z;
				3951	};
				3952
				3953	int foo(struct ST s) {
				3954	return &s[1].Z.B[5][13];
				3955	}
				3956	</pre>
				3957	</div>
				3958
				3959	<p>The LLVM code generated by the GCC frontend is:</p>
				3960
				3961	<div class="doc_code">
				3962	<pre>
Chris Lattner	5b6dc6e	2009-01-11 20:53:49 +0000	[diff] [blame]	3963	%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
				3964	%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3965
Dan Gohman	4736084	2009-07-25 02:23:48 +0000	[diff] [blame]	3966	define i32* @foo(%ST* %s) {
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3967	entry:
				3968	%reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
				3969	ret i32* %reg
				3970	}
				3971	</pre>
				3972	</div>
				3973
				3974	<h5>Semantics:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3975	<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3976	type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
				3977	}</tt>' type, a structure. The second index indexes into the third element
				3978	of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
				3979	i8 }</tt>' type, another structure. The third index indexes into the second
				3980	element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
				3981	array. The two dimensions of the array are subscripted into, yielding an
				3982	'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
				3983	pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3984
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	3985	<p>Note that it is perfectly legal to index partially through a structure,
				3986	returning a pointer to an inner element. Because of this, the LLVM code for
				3987	the given testcase is equivalent to:</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3988
				3989	<pre>
Dan Gohman	4736084	2009-07-25 02:23:48 +0000	[diff] [blame]	3990	define i32* @foo(%ST* %s) {
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3991	%t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
				3992	%t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
				3993	%t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
				3994	%t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
				3995	%t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
				3996	ret i32* %t5
				3997	}
				3998	</pre>
				3999
Dan Gohman	106b2ae	2009-07-27 21:53:46 +0000	[diff] [blame]	4000	<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohman	4cba156	2009-07-29 16:00:30 +0000	[diff] [blame^]	4001	<tt>getelementptr</tt> is undefined if the base pointer is not an
				4002	<i>in bounds</i> address of an allocated object, or if any of the addresses
				4003	formed by successive addition of the offsets implied by the indices to
				4004	the base address are not an <i>in bounds</i> address of that allocated
				4005	object.
				4006	The <i>in bounds</i> addresses for an allocated object are all the addresses
				4007	that point into the object, plus the address one past the end.</p>
Dan Gohman	106b2ae	2009-07-27 21:53:46 +0000	[diff] [blame]	4008
				4009	<p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
				4010	the base address with silently-wrapping two's complement arithmetic, and
				4011	the result value of the <tt>getelementptr</tt> may be outside the object
				4012	pointed to by the base pointer. The result value may not necessarily be
				4013	used to access memory though, even if it happens to point into allocated
				4014	storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
				4015	section for more information.</p>
				4016
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4017	<p>The getelementptr instruction is often confusing. For some more insight into
				4018	how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4019
				4020	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4021	<pre>
				4022	<i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman	32a080e	2008-10-13 13:44:15 +0000	[diff] [blame]	4023	%aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
				4024	<i>; yields i8*:vptr</i>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	4025	%vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
Matthijs Kooijman	32a080e	2008-10-13 13:44:15 +0000	[diff] [blame]	4026	<i>; yields i8*:eptr</i>
				4027	%eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta	4f9a0dc	2009-04-25 07:27:44 +0000	[diff] [blame]	4028	<i>; yields i32*:iptr</i>
Sanjiv Gupta	1e46c58	2009-04-24 16:38:13 +0000	[diff] [blame]	4029	%iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4030	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4031
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4032	</div>
				4033
				4034	<!-- ======================================================================= -->
				4035	<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
				4036	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4037
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4038	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4039
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4040	<p>The instructions in this category are the conversion instructions (casting)
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4041	which all take a single operand and a type. They perform various bit
				4042	conversions on the operand.</p>
				4043
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4044	</div>
				4045
				4046	<!-- _______________________________________________________________________ -->
				4047	<div class="doc_subsubsection">
				4048	<a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
				4049	</div>
				4050	<div class="doc_text">
				4051
				4052	<h5>Syntax:</h5>
				4053	<pre>
				4054	<result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
				4055	</pre>
				4056
				4057	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4058	<p>The '<tt>trunc</tt>' instruction truncates its operand to the
				4059	type <tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4060
				4061	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4062	<p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
				4063	be an <a href="#t_integer">integer</a> type, and a type that specifies the
				4064	size and type of the result, which must be
				4065	an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
				4066	be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
				4067	allowed.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4068
				4069	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4070	<p>The '<tt>trunc</tt>' instruction truncates the high order bits
				4071	in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
				4072	source size must be larger than the destination size, <tt>trunc</tt> cannot
				4073	be a <i>no-op cast</i>. It will always truncate bits.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4074
				4075	<h5>Example:</h5>
				4076	<pre>
				4077	%X = trunc i32 257 to i8 <i>; yields i8:1</i>
				4078	%Y = trunc i32 123 to i1 <i>; yields i1:true</i>
				4079	%Y = trunc i32 122 to i1 <i>; yields i1:false</i>
				4080	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4081
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4082	</div>
				4083
				4084	<!-- _______________________________________________________________________ -->
				4085	<div class="doc_subsubsection">
				4086	<a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
				4087	</div>
				4088	<div class="doc_text">
				4089
				4090	<h5>Syntax:</h5>
				4091	<pre>
				4092	<result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
				4093	</pre>
				4094
				4095	<h5>Overview:</h5>
				4096	<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4097	<tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4098
				4099
				4100	<h5>Arguments:</h5>
				4101	<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4102	<a href="#t_integer">integer</a> type, and a type to cast it to, which must
				4103	also be of <a href="#t_integer">integer</a> type. The bit size of the
				4104	<tt>value</tt> must be smaller than the bit size of the destination type,
				4105	<tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4106
				4107	<h5>Semantics:</h5>
				4108	<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4109	bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4110
				4111	<p>When zero extending from i1, the result will always be either 0 or 1.</p>
				4112
				4113	<h5>Example:</h5>
				4114	<pre>
				4115	%X = zext i32 257 to i64 <i>; yields i64:257</i>
				4116	%Y = zext i1 true to i32 <i>; yields i32:1</i>
				4117	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4118
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4119	</div>
				4120
				4121	<!-- _______________________________________________________________________ -->
				4122	<div class="doc_subsubsection">
				4123	<a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
				4124	</div>
				4125	<div class="doc_text">
				4126
				4127	<h5>Syntax:</h5>
				4128	<pre>
				4129	<result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
				4130	</pre>
				4131
				4132	<h5>Overview:</h5>
				4133	<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
				4134
				4135	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4136	<p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
				4137	<a href="#t_integer">integer</a> type, and a type to cast it to, which must
				4138	also be of <a href="#t_integer">integer</a> type. The bit size of the
				4139	<tt>value</tt> must be smaller than the bit size of the destination type,
				4140	<tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4141
				4142	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4143	<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
				4144	bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
				4145	of the type <tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4146
				4147	<p>When sign extending from i1, the extension always results in -1 or 0.</p>
				4148
				4149	<h5>Example:</h5>
				4150	<pre>
				4151	%X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
				4152	%Y = sext i1 true to i32 <i>; yields i32:-1</i>
				4153	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4154
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4155	</div>
				4156
				4157	<!-- _______________________________________________________________________ -->
				4158	<div class="doc_subsubsection">
				4159	<a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
				4160	</div>
				4161
				4162	<div class="doc_text">
				4163
				4164	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4165	<pre>
				4166	<result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
				4167	</pre>
				4168
				4169	<h5>Overview:</h5>
				4170	<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4171	<tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4172
				4173	<h5>Arguments:</h5>
				4174	<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4175	point</a> value to cast and a <a href="#t_floating">floating point</a> type
				4176	to cast it to. The size of <tt>value</tt> must be larger than the size of
				4177	<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
				4178	<i>no-op cast</i>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4179
				4180	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4181	<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
				4182	<a href="#t_floating">floating point</a> type to a smaller
				4183	<a href="#t_floating">floating point</a> type. If the value cannot fit
				4184	within the destination type, <tt>ty2</tt>, then the results are
				4185	undefined.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4186
				4187	<h5>Example:</h5>
				4188	<pre>
				4189	%X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
				4190	%Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
				4191	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4192
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4193	</div>
				4194
				4195	<!-- _______________________________________________________________________ -->
				4196	<div class="doc_subsubsection">
				4197	<a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
				4198	</div>
				4199	<div class="doc_text">
				4200
				4201	<h5>Syntax:</h5>
				4202	<pre>
				4203	<result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
				4204	</pre>
				4205
				4206	<h5>Overview:</h5>
				4207	<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4208	floating point value.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4209
				4210	<h5>Arguments:</h5>
				4211	<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4212	<a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
				4213	a <a href="#t_floating">floating point</a> type to cast it to. The source
				4214	type must be smaller than the destination type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4215
				4216	<h5>Semantics:</h5>
				4217	<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4218	<a href="#t_floating">floating point</a> type to a larger
				4219	<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
				4220	used to make a <i>no-op cast</i> because it always changes bits. Use
				4221	<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4222
				4223	<h5>Example:</h5>
				4224	<pre>
				4225	%X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
				4226	%Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
				4227	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4228
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4229	</div>
				4230
				4231	<!-- _______________________________________________________________________ -->
				4232	<div class="doc_subsubsection">
				4233	<a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
				4234	</div>
				4235	<div class="doc_text">
				4236
				4237	<h5>Syntax:</h5>
				4238	<pre>
Reid Spencer	e6adee8	2007-07-31 14:40:14 +0000	[diff] [blame]	4239	<result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4240	</pre>
				4241
				4242	<h5>Overview:</h5>
Reid Spencer	e6adee8	2007-07-31 14:40:14 +0000	[diff] [blame]	4243	<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4244	unsigned integer equivalent of type <tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4245
				4246	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4247	<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
				4248	scalar or vector <a href="#t_floating">floating point</a> value, and a type
				4249	to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
				4250	type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
				4251	vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4252
				4253	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4254	<p>The '<tt>fptoui</tt>' instruction converts its
				4255	<a href="#t_floating">floating point</a> operand into the nearest (rounding
				4256	towards zero) unsigned integer value. If the value cannot fit
				4257	in <tt>ty2</tt>, the results are undefined.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4258
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4259	<h5>Example:</h5>
				4260	<pre>
Reid Spencer	e6adee8	2007-07-31 14:40:14 +0000	[diff] [blame]	4261	%X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner	681f1e8	2007-09-22 03:17:52 +0000	[diff] [blame]	4262	%Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencer	e6adee8	2007-07-31 14:40:14 +0000	[diff] [blame]	4263	%X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4264	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4265
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4266	</div>
				4267
				4268	<!-- _______________________________________________________________________ -->
				4269	<div class="doc_subsubsection">
				4270	<a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
				4271	</div>
				4272	<div class="doc_text">
				4273
				4274	<h5>Syntax:</h5>
				4275	<pre>
				4276	<result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
				4277	</pre>
				4278
				4279	<h5>Overview:</h5>
				4280	<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4281	<a href="#t_floating">floating point</a> <tt>value</tt> to
				4282	type <tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4283
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4284	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4285	<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
				4286	scalar or vector <a href="#t_floating">floating point</a> value, and a type
				4287	to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
				4288	type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
				4289	vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4290
				4291	<h5>Semantics:</h5>
				4292	<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4293	<a href="#t_floating">floating point</a> operand into the nearest (rounding
				4294	towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
				4295	the results are undefined.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4296
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4297	<h5>Example:</h5>
				4298	<pre>
				4299	%X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner	681f1e8	2007-09-22 03:17:52 +0000	[diff] [blame]	4300	%Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4301	%X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
				4302	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4303
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4304	</div>
				4305
				4306	<!-- _______________________________________________________________________ -->
				4307	<div class="doc_subsubsection">
				4308	<a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
				4309	</div>
				4310	<div class="doc_text">
				4311
				4312	<h5>Syntax:</h5>
				4313	<pre>
				4314	<result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
				4315	</pre>
				4316
				4317	<h5>Overview:</h5>
				4318	<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4319	integer and converts that value to the <tt>ty2</tt> type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4320
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4321	<h5>Arguments:</h5>
Nate Begeman	78246ca	2007-11-17 03:58:34 +0000	[diff] [blame]	4322	<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4323	scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
				4324	it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
				4325	type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
				4326	floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4327
				4328	<h5>Semantics:</h5>
				4329	<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4330	integer quantity and converts it to the corresponding floating point
				4331	value. If the value cannot fit in the floating point value, the results are
				4332	undefined.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4333
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4334	<h5>Example:</h5>
				4335	<pre>
				4336	%X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	4337	%Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4338	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4339
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4340	</div>
				4341
				4342	<!-- _______________________________________________________________________ -->
				4343	<div class="doc_subsubsection">
				4344	<a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
				4345	</div>
				4346	<div class="doc_text">
				4347
				4348	<h5>Syntax:</h5>
				4349	<pre>
				4350	<result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
				4351	</pre>
				4352
				4353	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4354	<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
				4355	and converts that value to the <tt>ty2</tt> type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4356
				4357	<h5>Arguments:</h5>
Nate Begeman	78246ca	2007-11-17 03:58:34 +0000	[diff] [blame]	4358	<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4359	scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
				4360	it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
				4361	type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
				4362	floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4363
				4364	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4365	<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
				4366	quantity and converts it to the corresponding floating point value. If the
				4367	value cannot fit in the floating point value, the results are undefined.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4368
				4369	<h5>Example:</h5>
				4370	<pre>
				4371	%X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	4372	%Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4373	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4374
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4375	</div>
				4376
				4377	<!-- _______________________________________________________________________ -->
				4378	<div class="doc_subsubsection">
				4379	<a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
				4380	</div>
				4381	<div class="doc_text">
				4382
				4383	<h5>Syntax:</h5>
				4384	<pre>
				4385	<result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
				4386	</pre>
				4387
				4388	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4389	<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
				4390	the integer type <tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4391
				4392	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4393	<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
				4394	must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
				4395	<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4396
				4397	<h5>Semantics:</h5>
				4398	<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4399	<tt>ty2</tt> by interpreting the pointer value as an integer and either
				4400	truncating or zero extending that value to the size of the integer type. If
				4401	<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
				4402	<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
				4403	are the same size, then nothing is done (<i>no-op cast</i>) other than a type
				4404	change.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4405
				4406	<h5>Example:</h5>
				4407	<pre>
				4408	%X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
				4409	%Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
				4410	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4411
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4412	</div>
				4413
				4414	<!-- _______________________________________________________________________ -->
				4415	<div class="doc_subsubsection">
				4416	<a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
				4417	</div>
				4418	<div class="doc_text">
				4419
				4420	<h5>Syntax:</h5>
				4421	<pre>
				4422	<result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
				4423	</pre>
				4424
				4425	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4426	<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
				4427	pointer type, <tt>ty2</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4428
				4429	<h5>Arguments:</h5>
				4430	<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4431	value to cast, and a type to cast it to, which must be a
				4432	<a href="#t_pointer">pointer</a> type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4433
				4434	<h5>Semantics:</h5>
				4435	<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4436	<tt>ty2</tt> by applying either a zero extension or a truncation depending on
				4437	the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
				4438	size of a pointer then a truncation is done. If <tt>value</tt> is smaller
				4439	than the size of a pointer then a zero extension is done. If they are the
				4440	same size, nothing is done (<i>no-op cast</i>).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4441
				4442	<h5>Example:</h5>
				4443	<pre>
				4444	%X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
				4445	%X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
				4446	%Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
				4447	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4448
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4449	</div>
				4450
				4451	<!-- _______________________________________________________________________ -->
				4452	<div class="doc_subsubsection">
				4453	<a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
				4454	</div>
				4455	<div class="doc_text">
				4456
				4457	<h5>Syntax:</h5>
				4458	<pre>
				4459	<result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
				4460	</pre>
				4461
				4462	<h5>Overview:</h5>
				4463	<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4464	<tt>ty2</tt> without changing any bits.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4465
				4466	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4467	<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
				4468	non-aggregate first class value, and a type to cast it to, which must also be
				4469	a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
				4470	of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
				4471	identical. If the source type is a pointer, the destination type must also be
				4472	a pointer. This instruction supports bitwise conversion of vectors to
				4473	integers and to vectors of other types (as long as they have the same
				4474	size).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4475
				4476	<h5>Semantics:</h5>
				4477	<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4478	<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
				4479	this conversion. The conversion is done as if the <tt>value</tt> had been
				4480	stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
				4481	be converted to other pointer types with this instruction. To convert
				4482	pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
				4483	<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4484
				4485	<h5>Example:</h5>
				4486	<pre>
				4487	%X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
				4488	%Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	4489	%Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4490	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4491
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4492	</div>
				4493
				4494	<!-- ======================================================================= -->
				4495	<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4496
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4497	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4498
				4499	<p>The instructions in this category are the "miscellaneous" instructions, which
				4500	defy better classification.</p>
				4501
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4502	</div>
				4503
				4504	<!-- _______________________________________________________________________ -->
				4505	<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
				4506	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4507
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4508	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4509
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4510	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4511	<pre>
				4512	<result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4513	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4514
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4515	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4516	<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
				4517	boolean values based on comparison of its two integer, integer vector, or
				4518	pointer operands.</p>
				4519
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4520	<h5>Arguments:</h5>
				4521	<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4522	the condition code indicating the kind of comparison to perform. It is not a
				4523	value, just a keyword. The possible condition code are:</p>
				4524
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4525	<ol>
				4526	<li><tt>eq</tt>: equal</li>
				4527	<li><tt>ne</tt>: not equal </li>
				4528	<li><tt>ugt</tt>: unsigned greater than</li>
				4529	<li><tt>uge</tt>: unsigned greater or equal</li>
				4530	<li><tt>ult</tt>: unsigned less than</li>
				4531	<li><tt>ule</tt>: unsigned less or equal</li>
				4532	<li><tt>sgt</tt>: signed greater than</li>
				4533	<li><tt>sge</tt>: signed greater or equal</li>
				4534	<li><tt>slt</tt>: signed less than</li>
				4535	<li><tt>sle</tt>: signed less or equal</li>
				4536	</ol>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4537
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4538	<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4539	<a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
				4540	typed. They must also be identical types.</p>
				4541
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4542	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4543	<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
				4544	condition code given as <tt>cond</tt>. The comparison performed always yields
				4545	either an <a href="#t_primitive"><tt>i1</tt></a> or vector of <tt>i1</tt>
				4546	result, as follows:</p>
				4547
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4548	<ol>
				4549	<li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4550	<tt>false</tt> otherwise. No sign interpretation is necessary or
				4551	performed.</li>
				4552
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4553	<li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4554	<tt>false</tt> otherwise. No sign interpretation is necessary or
				4555	performed.</li>
				4556
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4557	<li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4558	<tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
				4559
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4560	<li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4561	<tt>true</tt> if <tt>op1</tt> is greater than or equal
				4562	to <tt>op2</tt>.</li>
				4563
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4564	<li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4565	<tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
				4566
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4567	<li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4568	<tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
				4569
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4570	<li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4571	<tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
				4572
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4573	<li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4574	<tt>true</tt> if <tt>op1</tt> is greater than or equal
				4575	to <tt>op2</tt>.</li>
				4576
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4577	<li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4578	<tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
				4579
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4580	<li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4581	<tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4582	</ol>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4583
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4584	<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4585	values are compared as if they were integers.</p>
				4586
				4587	<p>If the operands are integer vectors, then they are compared element by
				4588	element. The result is an <tt>i1</tt> vector with the same number of elements
				4589	as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4590
				4591	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4592	<pre>
				4593	<result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4594	<result> = icmp ne float* %X, %X <i>; yields: result=false</i>
				4595	<result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
				4596	<result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
				4597	<result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
				4598	<result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
				4599	</pre>
Dan Gohman	a53eb38	2009-01-22 01:39:38 +0000	[diff] [blame]	4600
				4601	<p>Note that the code generator does not yet support vector types with
				4602	the <tt>icmp</tt> instruction.</p>
				4603
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4604	</div>
				4605
				4606	<!-- _______________________________________________________________________ -->
				4607	<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
				4608	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4609
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4610	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4611
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4612	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4613	<pre>
				4614	<result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4615	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4616
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4617	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4618	<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
				4619	values based on comparison of its operands.</p>
				4620
				4621	<p>If the operands are floating point scalars, then the result type is a boolean
				4622	(<a href="#t_primitive"><tt>i1</tt></a>).</p>
				4623
				4624	<p>If the operands are floating point vectors, then the result type is a vector
				4625	of boolean with the same number of elements as the operands being
				4626	compared.</p>
				4627
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4628	<h5>Arguments:</h5>
				4629	<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4630	the condition code indicating the kind of comparison to perform. It is not a
				4631	value, just a keyword. The possible condition code are:</p>
				4632
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4633	<ol>
				4634	<li><tt>false</tt>: no comparison, always returns false</li>
				4635	<li><tt>oeq</tt>: ordered and equal</li>
				4636	<li><tt>ogt</tt>: ordered and greater than </li>
				4637	<li><tt>oge</tt>: ordered and greater than or equal</li>
				4638	<li><tt>olt</tt>: ordered and less than </li>
				4639	<li><tt>ole</tt>: ordered and less than or equal</li>
				4640	<li><tt>one</tt>: ordered and not equal</li>
				4641	<li><tt>ord</tt>: ordered (no nans)</li>
				4642	<li><tt>ueq</tt>: unordered or equal</li>
				4643	<li><tt>ugt</tt>: unordered or greater than </li>
				4644	<li><tt>uge</tt>: unordered or greater than or equal</li>
				4645	<li><tt>ult</tt>: unordered or less than </li>
				4646	<li><tt>ule</tt>: unordered or less than or equal</li>
				4647	<li><tt>une</tt>: unordered or not equal</li>
				4648	<li><tt>uno</tt>: unordered (either nans)</li>
				4649	<li><tt>true</tt>: no comparison, always returns true</li>
				4650	</ol>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4651
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4652	<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4653	<i>unordered</i> means that either operand may be a QNAN.</p>
				4654
				4655	<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
				4656	a <a href="#t_floating">floating point</a> type or
				4657	a <a href="#t_vector">vector</a> of floating point type. They must have
				4658	identical types.</p>
				4659
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4660	<h5>Semantics:</h5>
Gabor Greif	d9068fe	2008-08-07 21:46:00 +0000	[diff] [blame]	4661	<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4662	according to the condition code given as <tt>cond</tt>. If the operands are
				4663	vectors, then the vectors are compared element by element. Each comparison
				4664	performed always yields an <a href="#t_primitive">i1</a> result, as
				4665	follows:</p>
				4666
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4667	<ol>
				4668	<li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4669
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4670	<li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4671	<tt>op1</tt> is equal to <tt>op2</tt>.</li>
				4672
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4673	<li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4674	<tt>op1</tt> is greather than <tt>op2</tt>.</li>
				4675
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4676	<li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4677	<tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
				4678
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4679	<li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4680	<tt>op1</tt> is less than <tt>op2</tt>.</li>
				4681
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4682	<li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4683	<tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
				4684
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4685	<li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4686	<tt>op1</tt> is not equal to <tt>op2</tt>.</li>
				4687
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4688	<li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4689
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4690	<li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4691	<tt>op1</tt> is equal to <tt>op2</tt>.</li>
				4692
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4693	<li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4694	<tt>op1</tt> is greater than <tt>op2</tt>.</li>
				4695
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4696	<li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4697	<tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
				4698
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4699	<li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4700	<tt>op1</tt> is less than <tt>op2</tt>.</li>
				4701
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4702	<li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4703	<tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
				4704
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4705	<li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4706	<tt>op1</tt> is not equal to <tt>op2</tt>.</li>
				4707
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4708	<li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4709
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4710	<li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
				4711	</ol>
				4712
				4713	<h5>Example:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4714	<pre>
				4715	<result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
Dan Gohman	b60ca3c	2008-09-09 01:02:47 +0000	[diff] [blame]	4716	<result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
				4717	<result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
				4718	<result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4719	</pre>
Dan Gohman	a53eb38	2009-01-22 01:39:38 +0000	[diff] [blame]	4720
				4721	<p>Note that the code generator does not yet support vector types with
				4722	the <tt>fcmp</tt> instruction.</p>
				4723
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4724	</div>
				4725
				4726	<!-- _______________________________________________________________________ -->
Nate Begeman	646fa48	2008-05-12 19:01:56 +0000	[diff] [blame]	4727	<div class="doc_subsubsection">
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	4728	<a name="i_phi">'<tt>phi</tt>' Instruction</a>
				4729	</div>
				4730
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4731	<div class="doc_text">
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	4732
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4733	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4734	<pre>
				4735	<result> = phi <ty> [ <val0>, <label0>], ...
				4736	</pre>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	4737
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4738	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4739	<p>The '<tt>phi</tt>' instruction is used to implement the φ node in the
				4740	SSA graph representing the function.</p>
				4741
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4742	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4743	<p>The type of the incoming values is specified with the first type field. After
				4744	this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
				4745	one pair for each predecessor basic block of the current block. Only values
				4746	of <a href="#t_firstclass">first class</a> type may be used as the value
				4747	arguments to the PHI node. Only labels may be used as the label
				4748	arguments.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	4749
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4750	<p>There must be no non-phi instructions between the start of a basic block and
				4751	the PHI instructions: i.e. PHI instructions must be first in a basic
				4752	block.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	4753
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4754	<p>For the purposes of the SSA form, the use of each incoming value is deemed to
				4755	occur on the edge from the corresponding predecessor block to the current
				4756	block (but after any definition of an '<tt>invoke</tt>' instruction's return
				4757	value on the same edge).</p>
Jay Foad	8e2fd2c	2009-06-03 10:20:10 +0000	[diff] [blame]	4758
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4759	<h5>Semantics:</h5>
				4760	<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4761	specified by the pair corresponding to the predecessor basic block that
				4762	executed just prior to the current block.</p>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	4763
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4764	<h5>Example:</h5>
Chris Lattner	6704c21	2008-05-20 20:48:21 +0000	[diff] [blame]	4765	<pre>
				4766	Loop: ; Infinite loop that counts from 0 on up...
				4767	%indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
				4768	%nextindvar = add i32 %indvar, 1
				4769	br label %Loop
				4770	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4771
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4772	</div>
				4773
				4774	<!-- _______________________________________________________________________ -->
				4775	<div class="doc_subsubsection">
				4776	<a name="i_select">'<tt>select</tt>' Instruction</a>
				4777	</div>
				4778
				4779	<div class="doc_text">
				4780
				4781	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4782	<pre>
Dan Gohman	b60ca3c	2008-09-09 01:02:47 +0000	[diff] [blame]	4783	<result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
				4784
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	4785	<i>selty</i> is either i1 or {<N x i1>}
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4786	</pre>
				4787
				4788	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4789	<p>The '<tt>select</tt>' instruction is used to choose one value based on a
				4790	condition, without branching.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4791
				4792
				4793	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4794	<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
				4795	values indicating the condition, and two values of the
				4796	same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
				4797	vectors and the condition is a scalar, then entire vectors are selected, not
				4798	individual elements.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4799
				4800	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4801	<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
				4802	first value argument; otherwise, it returns the second value argument.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4803
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4804	<p>If the condition is a vector of i1, then the value arguments must be vectors
				4805	of the same size, and the selection is done element by element.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4806
				4807	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4808	<pre>
				4809	%X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
				4810	</pre>
Dan Gohman	a53eb38	2009-01-22 01:39:38 +0000	[diff] [blame]	4811
				4812	<p>Note that the code generator does not yet support conditions
				4813	with vector type.</p>
				4814
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4815	</div>
				4816
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4817	<!-- _______________________________________________________________________ -->
				4818	<div class="doc_subsubsection">
				4819	<a name="i_call">'<tt>call</tt>' Instruction</a>
				4820	</div>
				4821
				4822	<div class="doc_text">
				4823
				4824	<h5>Syntax:</h5>
				4825	<pre>
Devang Patel	d0bfcc7	2008-10-07 17:48:33 +0000	[diff] [blame]	4826	<result> = [tail] call [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] <ty> [<fnty>*] <fnptrval>(<function args>) [<a href="#fnattrs">fn attrs</a>]
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4827	</pre>
				4828
				4829	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4830	<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
				4831
				4832	<h5>Arguments:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4833	<p>This instruction requires several arguments:</p>
				4834
				4835	<ol>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4836	<li>The optional "tail" marker indicates whether the callee function accesses
				4837	any allocas or varargs in the caller. If the "tail" marker is present,
				4838	the function call is eligible for tail call optimization. Note that calls
				4839	may be marked "tail" even if they do not occur before
				4840	a <a href="#i_ret"><tt>ret</tt></a> instruction.</li>
Devang Patel	ac2fc27	2008-10-06 18:50:38 +0000	[diff] [blame]	4841
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4842	<li>The optional "cconv" marker indicates which <a href="#callingconv">calling
				4843	convention</a> the call should use. If none is specified, the call
				4844	defaults to using C calling conventions.</li>
Devang Patel	ac2fc27	2008-10-06 18:50:38 +0000	[diff] [blame]	4845
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4846	<li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
				4847	return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
				4848	'<tt>inreg</tt>' attributes are valid here.</li>
				4849
				4850	<li>'<tt>ty</tt>': the type of the call instruction itself which is also the
				4851	type of the return value. Functions that return no value are marked
				4852	<tt><a href="#t_void">void</a></tt>.</li>
				4853
				4854	<li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
				4855	being invoked. The argument types must match the types implied by this
				4856	signature. This type can be omitted if the function is not varargs and if
				4857	the function type does not return a pointer to a function.</li>
				4858
				4859	<li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
				4860	be invoked. In most cases, this is a direct function invocation, but
				4861	indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
				4862	to function value.</li>
				4863
				4864	<li>'<tt>function args</tt>': argument list whose types match the function
				4865	signature argument types. All arguments must be of
				4866	<a href="#t_firstclass">first class</a> type. If the function signature
				4867	indicates the function accepts a variable number of arguments, the extra
				4868	arguments can be specified.</li>
				4869
				4870	<li>The optional <a href="#fnattrs">function attributes</a> list. Only
				4871	'<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
				4872	'<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4873	</ol>
				4874
				4875	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4876	<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
				4877	a specified function, with its incoming arguments bound to the specified
				4878	values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
				4879	function, control flow continues with the instruction after the function
				4880	call, and the return value of the function is bound to the result
				4881	argument.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4882
				4883	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4884	<pre>
Nick Lewycky	93082fc	2007-09-08 13:57:50 +0000	[diff] [blame]	4885	%retval = call i32 @test(i32 %argc)
Chris Lattner	5e893ef	2008-03-21 17:24:17 +0000	[diff] [blame]	4886	call i32 (i8 , ...) @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
				4887	%X = tail call i32 @foo() <i>; yields i32</i>
				4888	%Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
				4889	call void %foo(i8 97 signext)
Devang Patel	a3cc537	2008-03-10 20:49:15 +0000	[diff] [blame]	4890
				4891	%struct.A = type { i32, i8 }
Devang Patel	ac2fc27	2008-10-06 18:50:38 +0000	[diff] [blame]	4892	%r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman	3e70003	2008-10-04 19:00:07 +0000	[diff] [blame]	4893	%gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
				4894	%gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattner	ac454b3	2008-10-08 06:26:11 +0000	[diff] [blame]	4895	%Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman	2c4e05a	2008-10-07 10:03:45 +0000	[diff] [blame]	4896	%ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4897	</pre>
				4898
				4899	</div>
				4900
				4901	<!-- _______________________________________________________________________ -->
				4902	<div class="doc_subsubsection">
				4903	<a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
				4904	</div>
				4905
				4906	<div class="doc_text">
				4907
				4908	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4909	<pre>
				4910	<resultval> = va_arg <va_list*> <arglist>, <argty>
				4911	</pre>
				4912
				4913	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4914	<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4915	the "variable argument" area of a function call. It is used to implement the
				4916	<tt>va_arg</tt> macro in C.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4917
				4918	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4919	<p>This instruction takes a <tt>va_list*</tt> value and the type of the
				4920	argument. It returns a value of the specified argument type and increments
				4921	the <tt>va_list</tt> to point to the next argument. The actual type
				4922	of <tt>va_list</tt> is target specific.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4923
				4924	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4925	<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
				4926	from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
				4927	to the next argument. For more information, see the variable argument
				4928	handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4929
				4930	<p>It is legal for this instruction to be called in a function which does not
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4931	take a variable number of arguments, for example, the <tt>vfprintf</tt>
				4932	function.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4933
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4934	<p><tt>va_arg</tt> is an LLVM instruction instead of
				4935	an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
				4936	argument.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4937
				4938	<h5>Example:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4939	<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
				4940
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4941	<p>Note that the code generator does not yet fully support va_arg on many
				4942	targets. Also, it does not currently support va_arg with aggregate types on
				4943	any target.</p>
Dan Gohman	6096719	2009-01-12 23:12:39 +0000	[diff] [blame]	4944
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4945	</div>
				4946
				4947	<!-- *********************************************************************** -->
				4948	<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
				4949	<!-- *********************************************************************** -->
				4950
				4951	<div class="doc_text">
				4952
				4953	<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4954	well known names and semantics and are required to follow certain
				4955	restrictions. Overall, these intrinsics represent an extension mechanism for
				4956	the LLVM language that does not require changing all of the transformations
				4957	in LLVM when adding to the language (or the bitcode reader/writer, the
				4958	parser, etc...).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4959
				4960	<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4961	prefix is reserved in LLVM for intrinsic names; thus, function names may not
				4962	begin with this prefix. Intrinsic functions must always be external
				4963	functions: you cannot define the body of intrinsic functions. Intrinsic
				4964	functions may only be used in call or invoke instructions: it is illegal to
				4965	take the address of an intrinsic function. Additionally, because intrinsic
				4966	functions are part of the LLVM language, it is required if any are added that
				4967	they be documented here.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4968
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4969	<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
				4970	family of functions that perform the same operation but on different data
				4971	types. Because LLVM can represent over 8 million different integer types,
				4972	overloading is used commonly to allow an intrinsic function to operate on any
				4973	integer type. One or more of the argument types or the result type can be
				4974	overloaded to accept any integer type. Argument types may also be defined as
				4975	exactly matching a previous argument's type or the result type. This allows
				4976	an intrinsic function which accepts multiple arguments, but needs all of them
				4977	to be of the same type, to only be overloaded with respect to a single
				4978	argument or the result.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4979
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4980	<p>Overloaded intrinsics will have the names of its overloaded argument types
				4981	encoded into its function name, each preceded by a period. Only those types
				4982	which are overloaded result in a name suffix. Arguments whose type is matched
				4983	against another type do not. For example, the <tt>llvm.ctpop</tt> function
				4984	can take an integer of any width and returns an integer of exactly the same
				4985	integer width. This leads to a family of functions such as
				4986	<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
				4987	%val)</tt>. Only one type, the return type, is overloaded, and only one type
				4988	suffix is required. Because the argument's type is matched against the return
				4989	type, it does not require its own name suffix.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4990
				4991	<p>To learn how to add an intrinsic function, please see the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	4992	<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	4993
				4994	</div>
				4995
				4996	<!-- ======================================================================= -->
				4997	<div class="doc_subsection">
				4998	<a name="int_varargs">Variable Argument Handling Intrinsics</a>
				4999	</div>
				5000
				5001	<div class="doc_text">
				5002
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5003	<p>Variable argument support is defined in LLVM with
				5004	the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
				5005	intrinsic functions. These functions are related to the similarly named
				5006	macros defined in the <tt><stdarg.h></tt> header file.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5007
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5008	<p>All of these functions operate on arguments that use a target-specific value
				5009	type "<tt>va_list</tt>". The LLVM assembly language reference manual does
				5010	not define what this type is, so all transformations should be prepared to
				5011	handle these functions regardless of the type used.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5012
				5013	<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5014	instruction and the variable argument handling intrinsic functions are
				5015	used.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5016
				5017	<div class="doc_code">
				5018	<pre>
				5019	define i32 @test(i32 %X, ...) {
				5020	; Initialize variable argument processing
				5021	%ap = alloca i8*
				5022	%ap2 = bitcast i8** %ap to i8*
				5023	call void @llvm.va_start(i8* %ap2)
				5024
				5025	; Read a single integer argument
				5026	%tmp = va_arg i8** %ap, i32
				5027
				5028	; Demonstrate usage of llvm.va_copy and llvm.va_end
				5029	%aq = alloca i8*
				5030	%aq2 = bitcast i8** %aq to i8*
				5031	call void @llvm.va_copy(i8* %aq2, i8* %ap2)
				5032	call void @llvm.va_end(i8* %aq2)
				5033
				5034	; Stop processing of arguments.
				5035	call void @llvm.va_end(i8* %ap2)
				5036	ret i32 %tmp
				5037	}
				5038
				5039	declare void @llvm.va_start(i8*)
				5040	declare void @llvm.va_copy(i8, i8)
				5041	declare void @llvm.va_end(i8*)
				5042	</pre>
				5043	</div>
				5044
				5045	</div>
				5046
				5047	<!-- _______________________________________________________________________ -->
				5048	<div class="doc_subsubsection">
				5049	<a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
				5050	</div>
				5051
				5052
				5053	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5054
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5055	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5056	<pre>
				5057	declare void %llvm.va_start(i8* <arglist>)
				5058	</pre>
				5059
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5060	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5061	<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt>
				5062	for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5063
				5064	<h5>Arguments:</h5>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	5065	<p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5066
				5067	<h5>Semantics:</h5>
Dan Gohman	2672f3e	2008-10-14 16:51:45 +0000	[diff] [blame]	5068	<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5069	macro available in C. In a target-dependent way, it initializes
				5070	the <tt>va_list</tt> element to which the argument points, so that the next
				5071	call to <tt>va_arg</tt> will produce the first variable argument passed to
				5072	the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
				5073	need to know the last argument of the function as the compiler can figure
				5074	that out.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5075
				5076	</div>
				5077
				5078	<!-- _______________________________________________________________________ -->
				5079	<div class="doc_subsubsection">
				5080	<a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
				5081	</div>
				5082
				5083	<div class="doc_text">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5084
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5085	<h5>Syntax:</h5>
				5086	<pre>
				5087	declare void @llvm.va_end(i8* <arglist>)
				5088	</pre>
				5089
				5090	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5091	<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5092	which has been initialized previously
				5093	with <tt><a href="#int_va_start">llvm.va_start</a></tt>
				5094	or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5095
				5096	<h5>Arguments:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5097	<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
				5098
				5099	<h5>Semantics:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5100	<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5101	macro available in C. In a target-dependent way, it destroys
				5102	the <tt>va_list</tt> element to which the argument points. Calls
				5103	to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
				5104	and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
				5105	with calls to <tt>llvm.va_end</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5106
				5107	</div>
				5108
				5109	<!-- _______________________________________________________________________ -->
				5110	<div class="doc_subsubsection">
				5111	<a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
				5112	</div>
				5113
				5114	<div class="doc_text">
				5115
				5116	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5117	<pre>
				5118	declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
				5119	</pre>
				5120
				5121	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5122	<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5123	from the source argument list to the destination argument list.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5124
				5125	<h5>Arguments:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5126	<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5127	The second argument is a pointer to a <tt>va_list</tt> element to copy
				5128	from.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5129
				5130	<h5>Semantics:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5131	<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5132	macro available in C. In a target-dependent way, it copies the
				5133	source <tt>va_list</tt> element into the destination <tt>va_list</tt>
				5134	element. This intrinsic is necessary because
				5135	the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
				5136	arbitrarily complex and require, for example, memory allocation.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5137
				5138	</div>
				5139
				5140	<!-- ======================================================================= -->
				5141	<div class="doc_subsection">
				5142	<a name="int_gc">Accurate Garbage Collection Intrinsics</a>
				5143	</div>
				5144
				5145	<div class="doc_text">
				5146
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5147	<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner	9645148	2008-08-05 18:29:16 +0000	[diff] [blame]	5148	Collection</a> (GC) requires the implementation and generation of these
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5149	intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
				5150	roots on the stack</a>, as well as garbage collector implementations that
				5151	require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
				5152	barriers. Front-ends for type-safe garbage collected languages should generate
				5153	these intrinsics to make use of the LLVM garbage collectors. For more details,
				5154	see <a href="GarbageCollection.html">Accurate Garbage Collection with
				5155	LLVM</a>.</p>
Christopher Lamb	cfe0096	2007-12-17 01:00:21 +0000	[diff] [blame]	5156
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5157	<p>The garbage collection intrinsics only operate on objects in the generic
				5158	address space (address space zero).</p>
Christopher Lamb	cfe0096	2007-12-17 01:00:21 +0000	[diff] [blame]	5159
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5160	</div>
				5161
				5162	<!-- _______________________________________________________________________ -->
				5163	<div class="doc_subsubsection">
				5164	<a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
				5165	</div>
				5166
				5167	<div class="doc_text">
				5168
				5169	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5170	<pre>
Chris Lattner	38bd5dd	2007-09-21 17:30:40 +0000	[diff] [blame]	5171	declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5172	</pre>
				5173
				5174	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5175	<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5176	the code generator, and allows some metadata to be associated with it.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5177
				5178	<h5>Arguments:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5179	<p>The first argument specifies the address of a stack object that contains the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5180	root pointer. The second pointer (which must be either a constant or a
				5181	global value address) contains the meta-data to be associated with the
				5182	root.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5183
				5184	<h5>Semantics:</h5>
Chris Lattner	a7d94ba	2008-04-24 05:59:56 +0000	[diff] [blame]	5185	<p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5186	location. At compile-time, the code generator generates information to allow
				5187	the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
				5188	intrinsic may only be used in a function which <a href="#gc">specifies a GC
				5189	algorithm</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5190
				5191	</div>
				5192
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5193	<!-- _______________________________________________________________________ -->
				5194	<div class="doc_subsubsection">
				5195	<a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
				5196	</div>
				5197
				5198	<div class="doc_text">
				5199
				5200	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5201	<pre>
Chris Lattner	38bd5dd	2007-09-21 17:30:40 +0000	[diff] [blame]	5202	declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5203	</pre>
				5204
				5205	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5206	<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5207	locations, allowing garbage collector implementations that require read
				5208	barriers.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5209
				5210	<h5>Arguments:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5211	<p>The second argument is the address to read from, which should be an address
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5212	allocated from the garbage collector. The first object is a pointer to the
				5213	start of the referenced object, if needed by the language runtime (otherwise
				5214	null).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5215
				5216	<h5>Semantics:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5217	<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5218	instruction, but may be replaced with substantially more complex code by the
				5219	garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
				5220	may only be used in a function which <a href="#gc">specifies a GC
				5221	algorithm</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5222
				5223	</div>
				5224
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5225	<!-- _______________________________________________________________________ -->
				5226	<div class="doc_subsubsection">
				5227	<a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
				5228	</div>
				5229
				5230	<div class="doc_text">
				5231
				5232	<h5>Syntax:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5233	<pre>
Chris Lattner	38bd5dd	2007-09-21 17:30:40 +0000	[diff] [blame]	5234	declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5235	</pre>
				5236
				5237	<h5>Overview:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5238	<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5239	locations, allowing garbage collector implementations that require write
				5240	barriers (such as generational or reference counting collectors).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5241
				5242	<h5>Arguments:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5243	<p>The first argument is the reference to store, the second is the start of the
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5244	object to store it to, and the third is the address of the field of Obj to
				5245	store to. If the runtime does not require a pointer to the object, Obj may
				5246	be null.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5247
				5248	<h5>Semantics:</h5>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5249	<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5250	instruction, but may be replaced with substantially more complex code by the
				5251	garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
				5252	may only be used in a function which <a href="#gc">specifies a GC
				5253	algorithm</a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5254
				5255	</div>
				5256
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5257	<!-- ======================================================================= -->
				5258	<div class="doc_subsection">
				5259	<a name="int_codegen">Code Generator Intrinsics</a>
				5260	</div>
				5261
				5262	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5263
				5264	<p>These intrinsics are provided by LLVM to expose special features that may
				5265	only be implemented with code generator support.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5266
				5267	</div>
				5268
				5269	<!-- _______________________________________________________________________ -->
				5270	<div class="doc_subsubsection">
				5271	<a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
				5272	</div>
				5273
				5274	<div class="doc_text">
				5275
				5276	<h5>Syntax:</h5>
				5277	<pre>
				5278	declare i8 *@llvm.returnaddress(i32 <level>)
				5279	</pre>
				5280
				5281	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5282	<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
				5283	target-specific value indicating the return address of the current function
				5284	or one of its callers.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5285
				5286	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5287	<p>The argument to this intrinsic indicates which function to return the address
				5288	for. Zero indicates the calling function, one indicates its caller, etc.
				5289	The argument is <b>required</b> to be a constant integer value.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5290
				5291	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5292	<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
				5293	indicating the return address of the specified call frame, or zero if it
				5294	cannot be identified. The value returned by this intrinsic is likely to be
				5295	incorrect or 0 for arguments other than zero, so it should only be used for
				5296	debugging purposes.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5297
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5298	<p>Note that calling this intrinsic does not prevent function inlining or other
				5299	aggressive transformations, so the value returned may not be that of the
				5300	obvious source-language caller.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5301
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5302	</div>
				5303
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5304	<!-- _______________________________________________________________________ -->
				5305	<div class="doc_subsubsection">
				5306	<a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
				5307	</div>
				5308
				5309	<div class="doc_text">
				5310
				5311	<h5>Syntax:</h5>
				5312	<pre>
Chris Lattner	38bd5dd	2007-09-21 17:30:40 +0000	[diff] [blame]	5313	declare i8 *@llvm.frameaddress(i32 <level>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5314	</pre>
				5315
				5316	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5317	<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
				5318	target-specific frame pointer value for the specified stack frame.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5319
				5320	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5321	<p>The argument to this intrinsic indicates which function to return the frame
				5322	pointer for. Zero indicates the calling function, one indicates its caller,
				5323	etc. The argument is <b>required</b> to be a constant integer value.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5324
				5325	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5326	<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
				5327	indicating the frame address of the specified call frame, or zero if it
				5328	cannot be identified. The value returned by this intrinsic is likely to be
				5329	incorrect or 0 for arguments other than zero, so it should only be used for
				5330	debugging purposes.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5331
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5332	<p>Note that calling this intrinsic does not prevent function inlining or other
				5333	aggressive transformations, so the value returned may not be that of the
				5334	obvious source-language caller.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5335
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5336	</div>
				5337
				5338	<!-- _______________________________________________________________________ -->
				5339	<div class="doc_subsubsection">
				5340	<a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
				5341	</div>
				5342
				5343	<div class="doc_text">
				5344
				5345	<h5>Syntax:</h5>
				5346	<pre>
Chris Lattner	38bd5dd	2007-09-21 17:30:40 +0000	[diff] [blame]	5347	declare i8 *@llvm.stacksave()
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5348	</pre>
				5349
				5350	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5351	<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
				5352	of the function stack, for use
				5353	with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
				5354	useful for implementing language features like scoped automatic variable
				5355	sized arrays in C99.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5356
				5357	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5358	<p>This intrinsic returns a opaque pointer value that can be passed
				5359	to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
				5360	an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
				5361	from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
				5362	to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
				5363	In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
				5364	stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5365
				5366	</div>
				5367
				5368	<!-- _______________________________________________________________________ -->
				5369	<div class="doc_subsubsection">
				5370	<a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
				5371	</div>
				5372
				5373	<div class="doc_text">
				5374
				5375	<h5>Syntax:</h5>
				5376	<pre>
				5377	declare void @llvm.stackrestore(i8 * %ptr)
				5378	</pre>
				5379
				5380	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5381	<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
				5382	the function stack to the state it was in when the
				5383	corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
				5384	executed. This is useful for implementing language features like scoped
				5385	automatic variable sized arrays in C99.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5386
				5387	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5388	<p>See the description
				5389	for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5390
				5391	</div>
				5392
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5393	<!-- _______________________________________________________________________ -->
				5394	<div class="doc_subsubsection">
				5395	<a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
				5396	</div>
				5397
				5398	<div class="doc_text">
				5399
				5400	<h5>Syntax:</h5>
				5401	<pre>
Chris Lattner	38bd5dd	2007-09-21 17:30:40 +0000	[diff] [blame]	5402	declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5403	</pre>
				5404
				5405	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5406	<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
				5407	insert a prefetch instruction if supported; otherwise, it is a noop.
				5408	Prefetches have no effect on the behavior of the program but can change its
				5409	performance characteristics.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5410
				5411	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5412	<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
				5413	specifier determining if the fetch should be for a read (0) or write (1),
				5414	and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
				5415	locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
				5416	and <tt>locality</tt> arguments must be constant integers.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5417
				5418	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5419	<p>This intrinsic does not modify the behavior of the program. In particular,
				5420	prefetches cannot trap and do not produce a value. On targets that support
				5421	this intrinsic, the prefetch can provide hints to the processor cache for
				5422	better performance.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5423
				5424	</div>
				5425
				5426	<!-- _______________________________________________________________________ -->
				5427	<div class="doc_subsubsection">
				5428	<a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
				5429	</div>
				5430
				5431	<div class="doc_text">
				5432
				5433	<h5>Syntax:</h5>
				5434	<pre>
Chris Lattner	38bd5dd	2007-09-21 17:30:40 +0000	[diff] [blame]	5435	declare void @llvm.pcmarker(i32 <id>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5436	</pre>
				5437
				5438	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5439	<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
				5440	Counter (PC) in a region of code to simulators and other tools. The method
				5441	is target specific, but it is expected that the marker will use exported
				5442	symbols to transmit the PC of the marker. The marker makes no guarantees
				5443	that it will remain with any specific instruction after optimizations. It is
				5444	possible that the presence of a marker will inhibit optimizations. The
				5445	intended use is to be inserted after optimizations to allow correlations of
				5446	simulation runs.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5447
				5448	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5449	<p><tt>id</tt> is a numerical id identifying the marker.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5450
				5451	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5452	<p>This intrinsic does not modify the behavior of the program. Backends that do
				5453	not support this intrinisic may ignore it.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5454
				5455	</div>
				5456
				5457	<!-- _______________________________________________________________________ -->
				5458	<div class="doc_subsubsection">
				5459	<a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
				5460	</div>
				5461
				5462	<div class="doc_text">
				5463
				5464	<h5>Syntax:</h5>
				5465	<pre>
				5466	declare i64 @llvm.readcyclecounter( )
				5467	</pre>
				5468
				5469	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5470	<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
				5471	counter register (or similar low latency, high accuracy clocks) on those
				5472	targets that support it. On X86, it should map to RDTSC. On Alpha, it
				5473	should map to RPCC. As the backing counters overflow quickly (on the order
				5474	of 9 seconds on alpha), this should only be used for small timings.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5475
				5476	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5477	<p>When directly supported, reading the cycle counter should not modify any
				5478	memory. Implementations are allowed to either return a application specific
				5479	value or a system wide value. On backends without support, this is lowered
				5480	to a constant 0.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5481
				5482	</div>
				5483
				5484	<!-- ======================================================================= -->
				5485	<div class="doc_subsection">
				5486	<a name="int_libc">Standard C Library Intrinsics</a>
				5487	</div>
				5488
				5489	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5490
				5491	<p>LLVM provides intrinsics for a few important standard C library functions.
				5492	These intrinsics allow source-language front-ends to pass information about
				5493	the alignment of the pointer arguments to the code generator, providing
				5494	opportunity for more efficient code generation.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5495
				5496	</div>
				5497
				5498	<!-- _______________________________________________________________________ -->
				5499	<div class="doc_subsubsection">
				5500	<a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
				5501	</div>
				5502
				5503	<div class="doc_text">
				5504
				5505	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5506	<p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
				5507	integer bit width. Not all targets support all bit widths however.</p>
				5508
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5509	<pre>
Chris Lattner	82c2e43	2008-11-21 16:42:48 +0000	[diff] [blame]	5510	declare void @llvm.memcpy.i8(i8 * <dest>, i8 * <src>,
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5511	i8 <len>, i32 <align>)
Chris Lattner	82c2e43	2008-11-21 16:42:48 +0000	[diff] [blame]	5512	declare void @llvm.memcpy.i16(i8 * <dest>, i8 * <src>,
				5513	i16 <len>, i32 <align>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5514	declare void @llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
				5515	i32 <len>, i32 <align>)
				5516	declare void @llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
				5517	i64 <len>, i32 <align>)
				5518	</pre>
				5519
				5520	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5521	<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
				5522	source location to the destination location.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5523
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5524	<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
				5525	intrinsics do not return a value, and takes an extra alignment argument.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5526
				5527	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5528	<p>The first argument is a pointer to the destination, the second is a pointer
				5529	to the source. The third argument is an integer argument specifying the
				5530	number of bytes to copy, and the fourth argument is the alignment of the
				5531	source and destination locations.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5532
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5533	<p>If the call to this intrinisic has an alignment value that is not 0 or 1,
				5534	then the caller guarantees that both the source and destination pointers are
				5535	aligned to that boundary.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5536
				5537	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5538	<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
				5539	source location to the destination location, which are not allowed to
				5540	overlap. It copies "len" bytes of memory over. If the argument is known to
				5541	be aligned to some boundary, this can be specified as the fourth argument,
				5542	otherwise it should be set to 0 or 1.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5543
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5544	</div>
				5545
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5546	<!-- _______________________________________________________________________ -->
				5547	<div class="doc_subsubsection">
				5548	<a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
				5549	</div>
				5550
				5551	<div class="doc_text">
				5552
				5553	<h5>Syntax:</h5>
Chris Lattner	82c2e43	2008-11-21 16:42:48 +0000	[diff] [blame]	5554	<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5555	width. Not all targets support all bit widths however.</p>
				5556
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5557	<pre>
Chris Lattner	82c2e43	2008-11-21 16:42:48 +0000	[diff] [blame]	5558	declare void @llvm.memmove.i8(i8 * <dest>, i8 * <src>,
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5559	i8 <len>, i32 <align>)
Chris Lattner	82c2e43	2008-11-21 16:42:48 +0000	[diff] [blame]	5560	declare void @llvm.memmove.i16(i8 * <dest>, i8 * <src>,
				5561	i16 <len>, i32 <align>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5562	declare void @llvm.memmove.i32(i8 * <dest>, i8 * <src>,
				5563	i32 <len>, i32 <align>)
				5564	declare void @llvm.memmove.i64(i8 * <dest>, i8 * <src>,
				5565	i64 <len>, i32 <align>)
				5566	</pre>
				5567
				5568	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5569	<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
				5570	source location to the destination location. It is similar to the
				5571	'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
				5572	overlap.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5573
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5574	<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
				5575	intrinsics do not return a value, and takes an extra alignment argument.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5576
				5577	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5578	<p>The first argument is a pointer to the destination, the second is a pointer
				5579	to the source. The third argument is an integer argument specifying the
				5580	number of bytes to copy, and the fourth argument is the alignment of the
				5581	source and destination locations.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5582
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5583	<p>If the call to this intrinisic has an alignment value that is not 0 or 1,
				5584	then the caller guarantees that the source and destination pointers are
				5585	aligned to that boundary.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5586
				5587	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5588	<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
				5589	source location to the destination location, which may overlap. It copies
				5590	"len" bytes of memory over. If the argument is known to be aligned to some
				5591	boundary, this can be specified as the fourth argument, otherwise it should
				5592	be set to 0 or 1.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5593
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5594	</div>
				5595
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5596	<!-- _______________________________________________________________________ -->
				5597	<div class="doc_subsubsection">
				5598	<a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
				5599	</div>
				5600
				5601	<div class="doc_text">
				5602
				5603	<h5>Syntax:</h5>
Chris Lattner	82c2e43	2008-11-21 16:42:48 +0000	[diff] [blame]	5604	<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5605	width. Not all targets support all bit widths however.</p>
				5606
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5607	<pre>
Chris Lattner	82c2e43	2008-11-21 16:42:48 +0000	[diff] [blame]	5608	declare void @llvm.memset.i8(i8 * <dest>, i8 <val>,
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5609	i8 <len>, i32 <align>)
Chris Lattner	82c2e43	2008-11-21 16:42:48 +0000	[diff] [blame]	5610	declare void @llvm.memset.i16(i8 * <dest>, i8 <val>,
				5611	i16 <len>, i32 <align>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5612	declare void @llvm.memset.i32(i8 * <dest>, i8 <val>,
				5613	i32 <len>, i32 <align>)
				5614	declare void @llvm.memset.i64(i8 * <dest>, i8 <val>,
				5615	i64 <len>, i32 <align>)
				5616	</pre>
				5617
				5618	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5619	<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
				5620	particular byte value.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5621
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5622	<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
				5623	intrinsic does not return a value, and takes an extra alignment argument.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5624
				5625	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5626	<p>The first argument is a pointer to the destination to fill, the second is the
				5627	byte value to fill it with, the third argument is an integer argument
				5628	specifying the number of bytes to fill, and the fourth argument is the known
				5629	alignment of destination location.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5630
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5631	<p>If the call to this intrinisic has an alignment value that is not 0 or 1,
				5632	then the caller guarantees that the destination pointer is aligned to that
				5633	boundary.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5634
				5635	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5636	<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
				5637	at the destination location. If the argument is known to be aligned to some
				5638	boundary, this can be specified as the fourth argument, otherwise it should
				5639	be set to 0 or 1.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5640
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5641	</div>
				5642
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5643	<!-- _______________________________________________________________________ -->
				5644	<div class="doc_subsubsection">
				5645	<a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
				5646	</div>
				5647
				5648	<div class="doc_text">
				5649
				5650	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5651	<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
				5652	floating point or vector of floating point type. Not all targets support all
				5653	types however.</p>
				5654
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5655	<pre>
Dale Johannesen	f9adbb6	2007-10-02 17:47:38 +0000	[diff] [blame]	5656	declare float @llvm.sqrt.f32(float %Val)
				5657	declare double @llvm.sqrt.f64(double %Val)
				5658	declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
				5659	declare fp128 @llvm.sqrt.f128(fp128 %Val)
				5660	declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5661	</pre>
				5662
				5663	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5664	<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
				5665	returning the same value as the libm '<tt>sqrt</tt>' functions would.
				5666	Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
				5667	behavior for negative numbers other than -0.0 (which allows for better
				5668	optimization, because there is no need to worry about errno being
				5669	set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5670
				5671	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5672	<p>The argument and return value are floating point numbers of the same
				5673	type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5674
				5675	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5676	<p>This function returns the sqrt of the specified operand if it is a
				5677	nonnegative floating point number.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5678
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5679	</div>
				5680
				5681	<!-- _______________________________________________________________________ -->
				5682	<div class="doc_subsubsection">
				5683	<a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
				5684	</div>
				5685
				5686	<div class="doc_text">
				5687
				5688	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5689	<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
				5690	floating point or vector of floating point type. Not all targets support all
				5691	types however.</p>
				5692
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5693	<pre>
Dale Johannesen	f9adbb6	2007-10-02 17:47:38 +0000	[diff] [blame]	5694	declare float @llvm.powi.f32(float %Val, i32 %power)
				5695	declare double @llvm.powi.f64(double %Val, i32 %power)
				5696	declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
				5697	declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
				5698	declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5699	</pre>
				5700
				5701	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5702	<p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
				5703	specified (positive or negative) power. The order of evaluation of
				5704	multiplications is not defined. When a vector of floating point type is
				5705	used, the second argument remains a scalar integer value.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5706
				5707	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5708	<p>The second argument is an integer power, and the first is a value to raise to
				5709	that power.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5710
				5711	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5712	<p>This function returns the first value raised to the second power with an
				5713	unspecified sequence of rounding operations.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5714
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5715	</div>
				5716
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5717	<!-- _______________________________________________________________________ -->
				5718	<div class="doc_subsubsection">
				5719	<a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
				5720	</div>
				5721
				5722	<div class="doc_text">
				5723
				5724	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5725	<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
				5726	floating point or vector of floating point type. Not all targets support all
				5727	types however.</p>
				5728
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5729	<pre>
				5730	declare float @llvm.sin.f32(float %Val)
				5731	declare double @llvm.sin.f64(double %Val)
				5732	declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
				5733	declare fp128 @llvm.sin.f128(fp128 %Val)
				5734	declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
				5735	</pre>
				5736
				5737	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5738	<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5739
				5740	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5741	<p>The argument and return value are floating point numbers of the same
				5742	type.</p>
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5743
				5744	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5745	<p>This function returns the sine of the specified operand, returning the same
				5746	values as the libm <tt>sin</tt> functions would, and handles error conditions
				5747	in the same way.</p>
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5748
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5749	</div>
				5750
				5751	<!-- _______________________________________________________________________ -->
				5752	<div class="doc_subsubsection">
				5753	<a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
				5754	</div>
				5755
				5756	<div class="doc_text">
				5757
				5758	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5759	<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
				5760	floating point or vector of floating point type. Not all targets support all
				5761	types however.</p>
				5762
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5763	<pre>
				5764	declare float @llvm.cos.f32(float %Val)
				5765	declare double @llvm.cos.f64(double %Val)
				5766	declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
				5767	declare fp128 @llvm.cos.f128(fp128 %Val)
				5768	declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
				5769	</pre>
				5770
				5771	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5772	<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5773
				5774	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5775	<p>The argument and return value are floating point numbers of the same
				5776	type.</p>
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5777
				5778	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5779	<p>This function returns the cosine of the specified operand, returning the same
				5780	values as the libm <tt>cos</tt> functions would, and handles error conditions
				5781	in the same way.</p>
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5782
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5783	</div>
				5784
				5785	<!-- _______________________________________________________________________ -->
				5786	<div class="doc_subsubsection">
				5787	<a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
				5788	</div>
				5789
				5790	<div class="doc_text">
				5791
				5792	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5793	<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
				5794	floating point or vector of floating point type. Not all targets support all
				5795	types however.</p>
				5796
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5797	<pre>
				5798	declare float @llvm.pow.f32(float %Val, float %Power)
				5799	declare double @llvm.pow.f64(double %Val, double %Power)
				5800	declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
				5801	declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
				5802	declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
				5803	</pre>
				5804
				5805	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5806	<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
				5807	specified (positive or negative) power.</p>
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5808
				5809	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5810	<p>The second argument is a floating point power, and the first is a value to
				5811	raise to that power.</p>
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5812
				5813	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5814	<p>This function returns the first value raised to the second power, returning
				5815	the same values as the libm <tt>pow</tt> functions would, and handles error
				5816	conditions in the same way.</p>
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5817
Dan Gohman	361079c	2007-10-15 20:30:11 +0000	[diff] [blame]	5818	</div>
				5819
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5820	<!-- ======================================================================= -->
				5821	<div class="doc_subsection">
				5822	<a name="int_manip">Bit Manipulation Intrinsics</a>
				5823	</div>
				5824
				5825	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5826
				5827	<p>LLVM provides intrinsics for a few important bit manipulation operations.
				5828	These allow efficient code generation for some algorithms.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5829
				5830	</div>
				5831
				5832	<!-- _______________________________________________________________________ -->
				5833	<div class="doc_subsubsection">
				5834	<a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
				5835	</div>
				5836
				5837	<div class="doc_text">
				5838
				5839	<h5>Syntax:</h5>
				5840	<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5841	type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
				5842
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5843	<pre>
Chandler Carruth	a228e39	2007-08-04 01:51:18 +0000	[diff] [blame]	5844	declare i16 @llvm.bswap.i16(i16 <id>)
				5845	declare i32 @llvm.bswap.i32(i32 <id>)
				5846	declare i64 @llvm.bswap.i64(i64 <id>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5847	</pre>
				5848
				5849	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5850	<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
				5851	values with an even number of bytes (positive multiple of 16 bits). These
				5852	are useful for performing operations on data that is not in the target's
				5853	native byte order.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5854
				5855	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5856	<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
				5857	and low byte of the input i16 swapped. Similarly,
				5858	the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
				5859	bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
				5860	2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
				5861	The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
				5862	extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
				5863	more, respectively).</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5864
				5865	</div>
				5866
				5867	<!-- _______________________________________________________________________ -->
				5868	<div class="doc_subsubsection">
				5869	<a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
				5870	</div>
				5871
				5872	<div class="doc_text">
				5873
				5874	<h5>Syntax:</h5>
				5875	<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5876	width. Not all targets support all bit widths however.</p>
				5877
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5878	<pre>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	5879	declare i8 @llvm.ctpop.i8(i8 <src>)
Chandler Carruth	a228e39	2007-08-04 01:51:18 +0000	[diff] [blame]	5880	declare i16 @llvm.ctpop.i16(i16 <src>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5881	declare i32 @llvm.ctpop.i32(i32 <src>)
Chandler Carruth	a228e39	2007-08-04 01:51:18 +0000	[diff] [blame]	5882	declare i64 @llvm.ctpop.i64(i64 <src>)
				5883	declare i256 @llvm.ctpop.i256(i256 <src>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5884	</pre>
				5885
				5886	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5887	<p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
				5888	in a value.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5889
				5890	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5891	<p>The only argument is the value to be counted. The argument may be of any
				5892	integer type. The return type must match the argument type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5893
				5894	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5895	<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5896
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5897	</div>
				5898
				5899	<!-- _______________________________________________________________________ -->
				5900	<div class="doc_subsubsection">
				5901	<a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
				5902	</div>
				5903
				5904	<div class="doc_text">
				5905
				5906	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5907	<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
				5908	integer bit width. Not all targets support all bit widths however.</p>
				5909
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5910	<pre>
Chandler Carruth	a228e39	2007-08-04 01:51:18 +0000	[diff] [blame]	5911	declare i8 @llvm.ctlz.i8 (i8 <src>)
				5912	declare i16 @llvm.ctlz.i16(i16 <src>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5913	declare i32 @llvm.ctlz.i32(i32 <src>)
Chandler Carruth	a228e39	2007-08-04 01:51:18 +0000	[diff] [blame]	5914	declare i64 @llvm.ctlz.i64(i64 <src>)
				5915	declare i256 @llvm.ctlz.i256(i256 <src>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5916	</pre>
				5917
				5918	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5919	<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
				5920	leading zeros in a variable.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5921
				5922	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5923	<p>The only argument is the value to be counted. The argument may be of any
				5924	integer type. The return type must match the argument type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5925
				5926	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5927	<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
				5928	zeros in a variable. If the src == 0 then the result is the size in bits of
				5929	the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5930
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5931	</div>
				5932
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5933	<!-- _______________________________________________________________________ -->
				5934	<div class="doc_subsubsection">
				5935	<a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
				5936	</div>
				5937
				5938	<div class="doc_text">
				5939
				5940	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5941	<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
				5942	integer bit width. Not all targets support all bit widths however.</p>
				5943
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5944	<pre>
Chandler Carruth	a228e39	2007-08-04 01:51:18 +0000	[diff] [blame]	5945	declare i8 @llvm.cttz.i8 (i8 <src>)
				5946	declare i16 @llvm.cttz.i16(i16 <src>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5947	declare i32 @llvm.cttz.i32(i32 <src>)
Chandler Carruth	a228e39	2007-08-04 01:51:18 +0000	[diff] [blame]	5948	declare i64 @llvm.cttz.i64(i64 <src>)
				5949	declare i256 @llvm.cttz.i256(i256 <src>)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5950	</pre>
				5951
				5952	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5953	<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
				5954	trailing zeros.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5955
				5956	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5957	<p>The only argument is the value to be counted. The argument may be of any
				5958	integer type. The return type must match the argument type.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5959
				5960	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5961	<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
				5962	zeros in a variable. If the src == 0 then the result is the size in bits of
				5963	the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5964
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	5965	</div>
				5966
Bill Wendling	3e1258b	2009-02-08 04:04:40 +0000	[diff] [blame]	5967	<!-- ======================================================================= -->
				5968	<div class="doc_subsection">
				5969	<a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
				5970	</div>
				5971
				5972	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5973
				5974	<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendling	3e1258b	2009-02-08 04:04:40 +0000	[diff] [blame]	5975
				5976	</div>
				5977
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	5978	<!-- _______________________________________________________________________ -->
				5979	<div class="doc_subsubsection">
Bill Wendling	3e1258b	2009-02-08 04:04:40 +0000	[diff] [blame]	5980	<a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	5981	</div>
				5982
				5983	<div class="doc_text">
				5984
				5985	<h5>Syntax:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	5986	<p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5987	on any integer bit width.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	5988
				5989	<pre>
				5990	declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
				5991	declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
				5992	declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
				5993	</pre>
				5994
				5995	<h5>Overview:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	5996	<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	5997	a signed addition of the two arguments, and indicate whether an overflow
				5998	occurred during the signed summation.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	5999
				6000	<h5>Arguments:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6001	<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6002	be of integer types of any bit width, but they must have the same bit
				6003	width. The second element of the result structure must be of
				6004	type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
				6005	undergo signed addition.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6006
				6007	<h5>Semantics:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6008	<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6009	a signed addition of the two variables. They return a structure — the
				6010	first element of which is the signed summation, and the second element of
				6011	which is a bit specifying if the signed summation resulted in an
				6012	overflow.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6013
				6014	<h5>Examples:</h5>
				6015	<pre>
				6016	%res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
				6017	%sum = extractvalue {i32, i1} %res, 0
				6018	%obit = extractvalue {i32, i1} %res, 1
				6019	br i1 %obit, label %overflow, label %normal
				6020	</pre>
				6021
				6022	</div>
				6023
				6024	<!-- _______________________________________________________________________ -->
				6025	<div class="doc_subsubsection">
Bill Wendling	3e1258b	2009-02-08 04:04:40 +0000	[diff] [blame]	6026	<a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6027	</div>
				6028
				6029	<div class="doc_text">
				6030
				6031	<h5>Syntax:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6032	<p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6033	on any integer bit width.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6034
				6035	<pre>
				6036	declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
				6037	declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
				6038	declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
				6039	</pre>
				6040
				6041	<h5>Overview:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6042	<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6043	an unsigned addition of the two arguments, and indicate whether a carry
				6044	occurred during the unsigned summation.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6045
				6046	<h5>Arguments:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6047	<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6048	be of integer types of any bit width, but they must have the same bit
				6049	width. The second element of the result structure must be of
				6050	type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
				6051	undergo unsigned addition.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6052
				6053	<h5>Semantics:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6054	<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6055	an unsigned addition of the two arguments. They return a structure —
				6056	the first element of which is the sum, and the second element of which is a
				6057	bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6058
				6059	<h5>Examples:</h5>
				6060	<pre>
				6061	%res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
				6062	%sum = extractvalue {i32, i1} %res, 0
				6063	%obit = extractvalue {i32, i1} %res, 1
				6064	br i1 %obit, label %carry, label %normal
				6065	</pre>
				6066
				6067	</div>
				6068
				6069	<!-- _______________________________________________________________________ -->
				6070	<div class="doc_subsubsection">
Bill Wendling	3e1258b	2009-02-08 04:04:40 +0000	[diff] [blame]	6071	<a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6072	</div>
				6073
				6074	<div class="doc_text">
				6075
				6076	<h5>Syntax:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6077	<p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6078	on any integer bit width.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6079
				6080	<pre>
				6081	declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
				6082	declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
				6083	declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
				6084	</pre>
				6085
				6086	<h5>Overview:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6087	<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6088	a signed subtraction of the two arguments, and indicate whether an overflow
				6089	occurred during the signed subtraction.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6090
				6091	<h5>Arguments:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6092	<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6093	be of integer types of any bit width, but they must have the same bit
				6094	width. The second element of the result structure must be of
				6095	type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
				6096	undergo signed subtraction.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6097
				6098	<h5>Semantics:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6099	<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6100	a signed subtraction of the two arguments. They return a structure —
				6101	the first element of which is the subtraction, and the second element of
				6102	which is a bit specifying if the signed subtraction resulted in an
				6103	overflow.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6104
				6105	<h5>Examples:</h5>
				6106	<pre>
				6107	%res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
				6108	%sum = extractvalue {i32, i1} %res, 0
				6109	%obit = extractvalue {i32, i1} %res, 1
				6110	br i1 %obit, label %overflow, label %normal
				6111	</pre>
				6112
				6113	</div>
				6114
				6115	<!-- _______________________________________________________________________ -->
				6116	<div class="doc_subsubsection">
Bill Wendling	3e1258b	2009-02-08 04:04:40 +0000	[diff] [blame]	6117	<a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6118	</div>
				6119
				6120	<div class="doc_text">
				6121
				6122	<h5>Syntax:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6123	<p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6124	on any integer bit width.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6125
				6126	<pre>
				6127	declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
				6128	declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
				6129	declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
				6130	</pre>
				6131
				6132	<h5>Overview:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6133	<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6134	an unsigned subtraction of the two arguments, and indicate whether an
				6135	overflow occurred during the unsigned subtraction.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6136
				6137	<h5>Arguments:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6138	<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6139	be of integer types of any bit width, but they must have the same bit
				6140	width. The second element of the result structure must be of
				6141	type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
				6142	undergo unsigned subtraction.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6143
				6144	<h5>Semantics:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6145	<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6146	an unsigned subtraction of the two arguments. They return a structure —
				6147	the first element of which is the subtraction, and the second element of
				6148	which is a bit specifying if the unsigned subtraction resulted in an
				6149	overflow.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6150
				6151	<h5>Examples:</h5>
				6152	<pre>
				6153	%res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
				6154	%sum = extractvalue {i32, i1} %res, 0
				6155	%obit = extractvalue {i32, i1} %res, 1
				6156	br i1 %obit, label %overflow, label %normal
				6157	</pre>
				6158
				6159	</div>
				6160
				6161	<!-- _______________________________________________________________________ -->
				6162	<div class="doc_subsubsection">
Bill Wendling	3e1258b	2009-02-08 04:04:40 +0000	[diff] [blame]	6163	<a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6164	</div>
				6165
				6166	<div class="doc_text">
				6167
				6168	<h5>Syntax:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6169	<p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6170	on any integer bit width.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6171
				6172	<pre>
				6173	declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
				6174	declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
				6175	declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
				6176	</pre>
				6177
				6178	<h5>Overview:</h5>
				6179
				6180	<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6181	a signed multiplication of the two arguments, and indicate whether an
				6182	overflow occurred during the signed multiplication.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6183
				6184	<h5>Arguments:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6185	<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6186	be of integer types of any bit width, but they must have the same bit
				6187	width. The second element of the result structure must be of
				6188	type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
				6189	undergo signed multiplication.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6190
				6191	<h5>Semantics:</h5>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6192	<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6193	a signed multiplication of the two arguments. They return a structure —
				6194	the first element of which is the multiplication, and the second element of
				6195	which is a bit specifying if the signed multiplication resulted in an
				6196	overflow.</p>
Bill Wendling	3f8cebe	2009-02-08 01:40:31 +0000	[diff] [blame]	6197
				6198	<h5>Examples:</h5>
				6199	<pre>
				6200	%res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
				6201	%sum = extractvalue {i32, i1} %res, 0
				6202	%obit = extractvalue {i32, i1} %res, 1
				6203	br i1 %obit, label %overflow, label %normal
				6204	</pre>
				6205
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6206	</div>
				6207
Bill Wendling	bda98b6	2009-02-08 23:00:09 +0000	[diff] [blame]	6208	<!-- _______________________________________________________________________ -->
				6209	<div class="doc_subsubsection">
				6210	<a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
				6211	</div>
				6212
				6213	<div class="doc_text">
				6214
				6215	<h5>Syntax:</h5>
Bill Wendling	bda98b6	2009-02-08 23:00:09 +0000	[diff] [blame]	6216	<p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6217	on any integer bit width.</p>
Bill Wendling	bda98b6	2009-02-08 23:00:09 +0000	[diff] [blame]	6218
				6219	<pre>
				6220	declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
				6221	declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
				6222	declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
				6223	</pre>
				6224
				6225	<h5>Overview:</h5>
Bill Wendling	bda98b6	2009-02-08 23:00:09 +0000	[diff] [blame]	6226	<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6227	a unsigned multiplication of the two arguments, and indicate whether an
				6228	overflow occurred during the unsigned multiplication.</p>
Bill Wendling	bda98b6	2009-02-08 23:00:09 +0000	[diff] [blame]	6229
				6230	<h5>Arguments:</h5>
Bill Wendling	bda98b6	2009-02-08 23:00:09 +0000	[diff] [blame]	6231	<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6232	be of integer types of any bit width, but they must have the same bit
				6233	width. The second element of the result structure must be of
				6234	type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
				6235	undergo unsigned multiplication.</p>
Bill Wendling	bda98b6	2009-02-08 23:00:09 +0000	[diff] [blame]	6236
				6237	<h5>Semantics:</h5>
Bill Wendling	bda98b6	2009-02-08 23:00:09 +0000	[diff] [blame]	6238	<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6239	an unsigned multiplication of the two arguments. They return a structure
				6240	— the first element of which is the multiplication, and the second
				6241	element of which is a bit specifying if the unsigned multiplication resulted
				6242	in an overflow.</p>
Bill Wendling	bda98b6	2009-02-08 23:00:09 +0000	[diff] [blame]	6243
				6244	<h5>Examples:</h5>
				6245	<pre>
				6246	%res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
				6247	%sum = extractvalue {i32, i1} %res, 0
				6248	%obit = extractvalue {i32, i1} %res, 1
				6249	br i1 %obit, label %overflow, label %normal
				6250	</pre>
				6251
				6252	</div>
				6253
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6254	<!-- ======================================================================= -->
				6255	<div class="doc_subsection">
				6256	<a name="int_debugger">Debugger Intrinsics</a>
				6257	</div>
				6258
				6259	<div class="doc_text">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6260
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6261	<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
				6262	prefix), are described in
				6263	the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
				6264	Level Debugging</a> document.</p>
				6265
				6266	</div>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6267
				6268	<!-- ======================================================================= -->
				6269	<div class="doc_subsection">
				6270	<a name="int_eh">Exception Handling Intrinsics</a>
				6271	</div>
				6272
				6273	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6274
				6275	<p>The LLVM exception handling intrinsics (which all start with
				6276	<tt>llvm.eh.</tt> prefix), are described in
				6277	the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
				6278	Handling</a> document.</p>
				6279
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6280	</div>
				6281
				6282	<!-- ======================================================================= -->
				6283	<div class="doc_subsection">
Duncan Sands	7407a9f	2007-09-11 14:10:23 +0000	[diff] [blame]	6284	<a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6285	</div>
				6286
				6287	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6288
				6289	<p>This intrinsic makes it possible to excise one parameter, marked with
				6290	the <tt>nest</tt> attribute, from a function. The result is a callable
				6291	function pointer lacking the nest parameter - the caller does not need to
				6292	provide a value for it. Instead, the value to use is stored in advance in a
				6293	"trampoline", a block of memory usually allocated on the stack, which also
				6294	contains code to splice the nest value into the argument list. This is used
				6295	to implement the GCC nested function address extension.</p>
				6296
				6297	<p>For example, if the function is
				6298	<tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
				6299	pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
				6300	follows:</p>
				6301
				6302	<div class="doc_code">
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6303	<pre>
Duncan Sands	7407a9f	2007-09-11 14:10:23 +0000	[diff] [blame]	6304	%tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
				6305	%tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
				6306	%p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8), i8 %nval )
				6307	%fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6308	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6309	</div>
				6310
				6311	<p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
				6312	to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
				6313
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6314	</div>
				6315
				6316	<!-- _______________________________________________________________________ -->
				6317	<div class="doc_subsubsection">
				6318	<a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
				6319	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6320
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6321	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6322
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6323	<h5>Syntax:</h5>
				6324	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6325	declare i8* @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6326	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6327
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6328	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6329	<p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
				6330	function pointer suitable for executing it.</p>
				6331
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6332	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6333	<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
				6334	pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
				6335	sufficiently aligned block of memory; this memory is written to by the
				6336	intrinsic. Note that the size and the alignment are target-specific - LLVM
				6337	currently provides no portable way of determining them, so a front-end that
				6338	generates this intrinsic needs to have some target-specific knowledge.
				6339	The <tt>func</tt> argument must hold a function bitcast to
				6340	an <tt>i8*</tt>.</p>
				6341
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6342	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6343	<p>The block of memory pointed to by <tt>tramp</tt> is filled with target
				6344	dependent code, turning it into a function. A pointer to this function is
				6345	returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
				6346	function pointer type</a> before being called. The new function's signature
				6347	is the same as that of <tt>func</tt> with any arguments marked with
				6348	the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
				6349	is allowed, and it must be of pointer type. Calling the new function is
				6350	equivalent to calling <tt>func</tt> with the same argument list, but
				6351	with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
				6352	calling <tt>llvm.init.trampoline</tt>, the memory pointed to
				6353	by <tt>tramp</tt> is modified, then the effect of any later call to the
				6354	returned function pointer is undefined.</p>
				6355
Duncan Sands	38947cd	2007-07-27 12:58:54 +0000	[diff] [blame]	6356	</div>
				6357
				6358	<!-- ======================================================================= -->
				6359	<div class="doc_subsection">
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6360	<a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
				6361	</div>
				6362
				6363	<div class="doc_text">
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6364
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6365	<p>These intrinsic functions expand the "universal IR" of LLVM to represent
				6366	hardware constructs for atomic operations and memory synchronization. This
				6367	provides an interface to the hardware, not an interface to the programmer. It
				6368	is aimed at a low enough level to allow any programming models or APIs
				6369	(Application Programming Interfaces) which need atomic behaviors to map
				6370	cleanly onto it. It is also modeled primarily on hardware behavior. Just as
				6371	hardware provides a "universal IR" for source languages, it also provides a
				6372	starting point for developing a "universal" atomic operation and
				6373	synchronization IR.</p>
				6374
				6375	<p>These do <em>not</em> form an API such as high-level threading libraries,
				6376	software transaction memory systems, atomic primitives, and intrinsic
				6377	functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
				6378	application libraries. The hardware interface provided by LLVM should allow
				6379	a clean implementation of all of these APIs and parallel programming models.
				6380	No one model or paradigm should be selected above others unless the hardware
				6381	itself ubiquitously does so.</p>
				6382
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6383	</div>
				6384
				6385	<!-- _______________________________________________________________________ -->
				6386	<div class="doc_subsubsection">
				6387	<a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
				6388	</div>
				6389	<div class="doc_text">
				6390	<h5>Syntax:</h5>
				6391	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6392	declare void @llvm.memory.barrier( i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss>, i1 <device> )
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6393	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6394
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6395	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6396	<p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
				6397	specific pairs of memory access types.</p>
				6398
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6399	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6400	<p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
				6401	The first four arguments enables a specific barrier as listed below. The
				6402	fith argument specifies that the barrier applies to io or device or uncached
				6403	memory.</p>
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6404
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6405	<ul>
				6406	<li><tt>ll</tt>: load-load barrier</li>
				6407	<li><tt>ls</tt>: load-store barrier</li>
				6408	<li><tt>sl</tt>: store-load barrier</li>
				6409	<li><tt>ss</tt>: store-store barrier</li>
				6410	<li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
				6411	</ul>
				6412
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6413	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6414	<p>This intrinsic causes the system to enforce some ordering constraints upon
				6415	the loads and stores of the program. This barrier does not
				6416	indicate <em>when</em> any events will occur, it only enforces
				6417	an <em>order</em> in which they occur. For any of the specified pairs of load
				6418	and store operations (f.ex. load-load, or store-load), all of the first
				6419	operations preceding the barrier will complete before any of the second
				6420	operations succeeding the barrier begin. Specifically the semantics for each
				6421	pairing is as follows:</p>
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6422
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6423	<ul>
				6424	<li><tt>ll</tt>: All loads before the barrier must complete before any load
				6425	after the barrier begins.</li>
				6426	<li><tt>ls</tt>: All loads before the barrier must complete before any
				6427	store after the barrier begins.</li>
				6428	<li><tt>ss</tt>: All stores before the barrier must complete before any
				6429	store after the barrier begins.</li>
				6430	<li><tt>sl</tt>: All stores before the barrier must complete before any
				6431	load after the barrier begins.</li>
				6432	</ul>
				6433
				6434	<p>These semantics are applied with a logical "and" behavior when more than one
				6435	is enabled in a single memory barrier intrinsic.</p>
				6436
				6437	<p>Backends may implement stronger barriers than those requested when they do
				6438	not support as fine grained a barrier as requested. Some architectures do
				6439	not need all types of barriers and on such architectures, these become
				6440	noops.</p>
				6441
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6442	<h5>Example:</h5>
				6443	<pre>
				6444	%ptr = malloc i32
				6445	store i32 4, %ptr
				6446
				6447	%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
				6448	call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
				6449	<i>; guarantee the above finishes</i>
				6450	store i32 8, %ptr <i>; before this begins</i>
				6451	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6452
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6453	</div>
				6454
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6455	<!-- _______________________________________________________________________ -->
				6456	<div class="doc_subsubsection">
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6457	<a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6458	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6459
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6460	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6461
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6462	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6463	<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
				6464	any integer bit width and for different address spaces. Not all targets
				6465	support all bit widths however.</p>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6466
				6467	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6468	declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* <ptr>, i8 <cmp>, i8 <val> )
				6469	declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* <ptr>, i16 <cmp>, i16 <val> )
				6470	declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* <ptr>, i32 <cmp>, i32 <val> )
				6471	declare i64 @llvm.atomic.cmp.swap.i64.p0i64( i64* <ptr>, i64 <cmp>, i64 <val> )
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6472	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6473
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6474	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6475	<p>This loads a value in memory and compares it to a given value. If they are
				6476	equal, it stores a new value into the memory.</p>
				6477
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6478	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6479	<p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
				6480	as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
				6481	same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
				6482	this integer type. While any bit width integer may be used, targets may only
				6483	lower representations they support in hardware.</p>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6484
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6485	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6486	<p>This entire intrinsic must be executed atomically. It first loads the value
				6487	in memory pointed to by <tt>ptr</tt> and compares it with the
				6488	value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
				6489	memory. The loaded value is yielded in all cases. This provides the
				6490	equivalent of an atomic compare-and-swap operation within the SSA
				6491	framework.</p>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6492
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6493	<h5>Examples:</h5>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6494	<pre>
				6495	%ptr = malloc i32
				6496	store i32 4, %ptr
				6497
				6498	%val1 = add i32 4, 4
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6499	%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6500	<i>; yields {i32}:result1 = 4</i>
				6501	%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
				6502	%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
				6503
				6504	%val2 = add i32 1, 1
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6505	%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6506	<i>; yields {i32}:result2 = 8</i>
				6507	%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
				6508
				6509	%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
				6510	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6511
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6512	</div>
				6513
				6514	<!-- _______________________________________________________________________ -->
				6515	<div class="doc_subsubsection">
				6516	<a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
				6517	</div>
				6518	<div class="doc_text">
				6519	<h5>Syntax:</h5>
				6520
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6521	<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
				6522	integer bit width. Not all targets support all bit widths however.</p>
				6523
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6524	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6525	declare i8 @llvm.atomic.swap.i8.p0i8( i8* <ptr>, i8 <val> )
				6526	declare i16 @llvm.atomic.swap.i16.p0i16( i16* <ptr>, i16 <val> )
				6527	declare i32 @llvm.atomic.swap.i32.p0i32( i32* <ptr>, i32 <val> )
				6528	declare i64 @llvm.atomic.swap.i64.p0i64( i64* <ptr>, i64 <val> )
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6529	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6530
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6531	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6532	<p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
				6533	the value from memory. It then stores the value in <tt>val</tt> in the memory
				6534	at <tt>ptr</tt>.</p>
				6535
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6536	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6537	<p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
				6538	the <tt>val</tt> argument and the result must be integers of the same bit
				6539	width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
				6540	integer type. The targets may only lower integer representations they
				6541	support.</p>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6542
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6543	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6544	<p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
				6545	stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
				6546	equivalent of an atomic swap operation within the SSA framework.</p>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6547
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6548	<h5>Examples:</h5>
				6549	<pre>
				6550	%ptr = malloc i32
				6551	store i32 4, %ptr
				6552
				6553	%val1 = add i32 4, 4
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6554	%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6555	<i>; yields {i32}:result1 = 4</i>
				6556	%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
				6557	%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
				6558
				6559	%val2 = add i32 1, 1
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6560	%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6561	<i>; yields {i32}:result2 = 8</i>
				6562
				6563	%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
				6564	%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
				6565	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6566
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6567	</div>
				6568
				6569	<!-- _______________________________________________________________________ -->
				6570	<div class="doc_subsubsection">
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6571	<a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6572
				6573	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6574
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6575	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6576
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6577	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6578	<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
				6579	any integer bit width. Not all targets support all bit widths however.</p>
				6580
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6581	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6582	declare i8 @llvm.atomic.load.add.i8..p0i8( i8* <ptr>, i8 <delta> )
				6583	declare i16 @llvm.atomic.load.add.i16..p0i16( i16* <ptr>, i16 <delta> )
				6584	declare i32 @llvm.atomic.load.add.i32..p0i32( i32* <ptr>, i32 <delta> )
				6585	declare i64 @llvm.atomic.load.add.i64..p0i64( i64* <ptr>, i64 <delta> )
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6586	</pre>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6587
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6588	<h5>Overview:</h5>
				6589	<p>This intrinsic adds <tt>delta</tt> to the value stored in memory
				6590	at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
				6591
				6592	<h5>Arguments:</h5>
				6593	<p>The intrinsic takes two arguments, the first a pointer to an integer value
				6594	and the second an integer value. The result is also an integer value. These
				6595	integer types can have any bit width, but they must all have the same bit
				6596	width. The targets may only lower integer representations they support.</p>
				6597
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6598	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6599	<p>This intrinsic does a series of operations atomically. It first loads the
				6600	value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
				6601	to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6602
				6603	<h5>Examples:</h5>
				6604	<pre>
				6605	%ptr = malloc i32
				6606	store i32 4, %ptr
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6607	%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6608	<i>; yields {i32}:result1 = 4</i>
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6609	%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6610	<i>; yields {i32}:result2 = 8</i>
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6611	%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6612	<i>; yields {i32}:result3 = 10</i>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6613	%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6614	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6615
Andrew Lenharth	e44f390	2008-02-21 06:45:13 +0000	[diff] [blame]	6616	</div>
				6617
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6618	<!-- _______________________________________________________________________ -->
				6619	<div class="doc_subsubsection">
				6620	<a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
				6621
				6622	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6623
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6624	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6625
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6626	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6627	<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
				6628	any integer bit width and for different address spaces. Not all targets
				6629	support all bit widths however.</p>
				6630
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6631	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6632	declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* <ptr>, i8 <delta> )
				6633	declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* <ptr>, i16 <delta> )
				6634	declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* <ptr>, i32 <delta> )
				6635	declare i64 @llvm.atomic.load.sub.i64.p0i32( i64* <ptr>, i64 <delta> )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6636	</pre>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6637
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6638	<h5>Overview:</h5>
				6639	<p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
				6640	<tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
				6641
				6642	<h5>Arguments:</h5>
				6643	<p>The intrinsic takes two arguments, the first a pointer to an integer value
				6644	and the second an integer value. The result is also an integer value. These
				6645	integer types can have any bit width, but they must all have the same bit
				6646	width. The targets may only lower integer representations they support.</p>
				6647
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6648	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6649	<p>This intrinsic does a series of operations atomically. It first loads the
				6650	value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
				6651	result to <tt>ptr</tt>. It yields the original value stored
				6652	at <tt>ptr</tt>.</p>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6653
				6654	<h5>Examples:</h5>
				6655	<pre>
				6656	%ptr = malloc i32
				6657	store i32 8, %ptr
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6658	%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6659	<i>; yields {i32}:result1 = 8</i>
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6660	%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6661	<i>; yields {i32}:result2 = 4</i>
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6662	%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6663	<i>; yields {i32}:result3 = 2</i>
				6664	%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
				6665	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6666
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6667	</div>
				6668
				6669	<!-- _______________________________________________________________________ -->
				6670	<div class="doc_subsubsection">
				6671	<a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
				6672	<a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
				6673	<a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
				6674	<a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6675	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6676
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6677	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6678
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6679	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6680	<p>These are overloaded intrinsics. You can
				6681	use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
				6682	<tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
				6683	bit width and for different address spaces. Not all targets support all bit
				6684	widths however.</p>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6685
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6686	<pre>
				6687	declare i8 @llvm.atomic.load.and.i8.p0i8( i8* <ptr>, i8 <delta> )
				6688	declare i16 @llvm.atomic.load.and.i16.p0i16( i16* <ptr>, i16 <delta> )
				6689	declare i32 @llvm.atomic.load.and.i32.p0i32( i32* <ptr>, i32 <delta> )
				6690	declare i64 @llvm.atomic.load.and.i64.p0i64( i64* <ptr>, i64 <delta> )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6691	</pre>
				6692
				6693	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6694	declare i8 @llvm.atomic.load.or.i8.p0i8( i8* <ptr>, i8 <delta> )
				6695	declare i16 @llvm.atomic.load.or.i16.p0i16( i16* <ptr>, i16 <delta> )
				6696	declare i32 @llvm.atomic.load.or.i32.p0i32( i32* <ptr>, i32 <delta> )
				6697	declare i64 @llvm.atomic.load.or.i64.p0i64( i64* <ptr>, i64 <delta> )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6698	</pre>
				6699
				6700	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6701	declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* <ptr>, i8 <delta> )
				6702	declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* <ptr>, i16 <delta> )
				6703	declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* <ptr>, i32 <delta> )
				6704	declare i64 @llvm.atomic.load.nand.i64.p0i32( i64* <ptr>, i64 <delta> )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6705	</pre>
				6706
				6707	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6708	declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* <ptr>, i8 <delta> )
				6709	declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* <ptr>, i16 <delta> )
				6710	declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* <ptr>, i32 <delta> )
				6711	declare i64 @llvm.atomic.load.xor.i64.p0i32( i64* <ptr>, i64 <delta> )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6712	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6713
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6714	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6715	<p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
				6716	the value stored in memory at <tt>ptr</tt>. It yields the original value
				6717	at <tt>ptr</tt>.</p>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6718
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6719	<h5>Arguments:</h5>
				6720	<p>These intrinsics take two arguments, the first a pointer to an integer value
				6721	and the second an integer value. The result is also an integer value. These
				6722	integer types can have any bit width, but they must all have the same bit
				6723	width. The targets may only lower integer representations they support.</p>
				6724
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6725	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6726	<p>These intrinsics does a series of operations atomically. They first load the
				6727	value stored at <tt>ptr</tt>. They then do the bitwise
				6728	operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
				6729	original value stored at <tt>ptr</tt>.</p>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6730
				6731	<h5>Examples:</h5>
				6732	<pre>
				6733	%ptr = malloc i32
				6734	store i32 0x0F0F, %ptr
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6735	%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6736	<i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6737	%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6738	<i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6739	%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6740	<i>; yields {i32}:result2 = 0xF0</i>
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6741	%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6742	<i>; yields {i32}:result3 = FF</i>
				6743	%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
				6744	</pre>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6745
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6746	</div>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6747
				6748	<!-- _______________________________________________________________________ -->
				6749	<div class="doc_subsubsection">
				6750	<a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
				6751	<a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
				6752	<a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
				6753	<a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6754	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6755
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6756	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6757
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6758	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6759	<p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
				6760	<tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
				6761	<tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
				6762	address spaces. Not all targets support all bit widths however.</p>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6763
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6764	<pre>
				6765	declare i8 @llvm.atomic.load.max.i8.p0i8( i8* <ptr>, i8 <delta> )
				6766	declare i16 @llvm.atomic.load.max.i16.p0i16( i16* <ptr>, i16 <delta> )
				6767	declare i32 @llvm.atomic.load.max.i32.p0i32( i32* <ptr>, i32 <delta> )
				6768	declare i64 @llvm.atomic.load.max.i64.p0i64( i64* <ptr>, i64 <delta> )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6769	</pre>
				6770
				6771	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6772	declare i8 @llvm.atomic.load.min.i8.p0i8( i8* <ptr>, i8 <delta> )
				6773	declare i16 @llvm.atomic.load.min.i16.p0i16( i16* <ptr>, i16 <delta> )
				6774	declare i32 @llvm.atomic.load.min.i32..p0i32( i32* <ptr>, i32 <delta> )
				6775	declare i64 @llvm.atomic.load.min.i64..p0i64( i64* <ptr>, i64 <delta> )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6776	</pre>
				6777
				6778	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6779	declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* <ptr>, i8 <delta> )
				6780	declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* <ptr>, i16 <delta> )
				6781	declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* <ptr>, i32 <delta> )
				6782	declare i64 @llvm.atomic.load.umax.i64.p0i64( i64* <ptr>, i64 <delta> )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6783	</pre>
				6784
				6785	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6786	declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* <ptr>, i8 <delta> )
				6787	declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* <ptr>, i16 <delta> )
				6788	declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* <ptr>, i32 <delta> )
				6789	declare i64 @llvm.atomic.load.umin.i64..p0i64( i64* <ptr>, i64 <delta> )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6790	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6791
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6792	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6793	<p>These intrinsics takes the signed or unsigned minimum or maximum of
				6794	<tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
				6795	original value at <tt>ptr</tt>.</p>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6796
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6797	<h5>Arguments:</h5>
				6798	<p>These intrinsics take two arguments, the first a pointer to an integer value
				6799	and the second an integer value. The result is also an integer value. These
				6800	integer types can have any bit width, but they must all have the same bit
				6801	width. The targets may only lower integer representations they support.</p>
				6802
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6803	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6804	<p>These intrinsics does a series of operations atomically. They first load the
				6805	value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
				6806	max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
				6807	yield the original value stored at <tt>ptr</tt>.</p>
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6808
				6809	<h5>Examples:</h5>
				6810	<pre>
				6811	%ptr = malloc i32
				6812	store i32 7, %ptr
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6813	%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6814	<i>; yields {i32}:result0 = 7</i>
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6815	%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6816	<i>; yields {i32}:result1 = -2</i>
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6817	%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6818	<i>; yields {i32}:result2 = 8</i>
Mon P Wang	ce3ac89	2008-07-30 04:36:53 +0000	[diff] [blame]	6819	%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6820	<i>; yields {i32}:result3 = 8</i>
				6821	%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
				6822	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6823
Mon P Wang	6bde9ec	2008-06-25 08:15:39 +0000	[diff] [blame]	6824	</div>
Andrew Lenharth	785610d	2008-02-16 01:24:58 +0000	[diff] [blame]	6825
				6826	<!-- ======================================================================= -->
				6827	<div class="doc_subsection">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6828	<a name="int_general">General Intrinsics</a>
				6829	</div>
				6830
				6831	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6832
				6833	<p>This class of intrinsics is designed to be generic and has no specific
				6834	purpose.</p>
				6835
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6836	</div>
				6837
				6838	<!-- _______________________________________________________________________ -->
				6839	<div class="doc_subsubsection">
				6840	<a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
				6841	</div>
				6842
				6843	<div class="doc_text">
				6844
				6845	<h5>Syntax:</h5>
				6846	<pre>
				6847	declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int> )
				6848	</pre>
				6849
				6850	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6851	<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6852
				6853	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6854	<p>The first argument is a pointer to a value, the second is a pointer to a
				6855	global string, the third is a pointer to a global string which is the source
				6856	file name, and the last argument is the line number.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6857
				6858	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6859	<p>This intrinsic allows annotation of local variables with arbitrary strings.
				6860	This can be useful for special purpose optimizations that want to look for
				6861	these annotations. These have no other defined use, they are ignored by code
				6862	generation and optimization.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6863
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6864	</div>
				6865
Tanya Lattner	b306a9e	2007-09-21 22:59:12 +0000	[diff] [blame]	6866	<!-- _______________________________________________________________________ -->
				6867	<div class="doc_subsubsection">
Tanya Lattner	c9869b1	2007-09-21 23:57:59 +0000	[diff] [blame]	6868	<a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattner	b306a9e	2007-09-21 22:59:12 +0000	[diff] [blame]	6869	</div>
				6870
				6871	<div class="doc_text">
				6872
				6873	<h5>Syntax:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6874	<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
				6875	any integer bit width.</p>
				6876
Tanya Lattner	b306a9e	2007-09-21 22:59:12 +0000	[diff] [blame]	6877	<pre>
Tanya Lattner	09161fe	2007-09-22 00:03:01 +0000	[diff] [blame]	6878	declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int> )
				6879	declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int> )
				6880	declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int> )
				6881	declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int> )
				6882	declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int> )
Tanya Lattner	b306a9e	2007-09-21 22:59:12 +0000	[diff] [blame]	6883	</pre>
				6884
				6885	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6886	<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattner	b306a9e	2007-09-21 22:59:12 +0000	[diff] [blame]	6887
				6888	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6889	<p>The first argument is an integer value (result of some expression), the
				6890	second is a pointer to a global string, the third is a pointer to a global
				6891	string which is the source file name, and the last argument is the line
				6892	number. It returns the value of the first argument.</p>
Tanya Lattner	b306a9e	2007-09-21 22:59:12 +0000	[diff] [blame]	6893
				6894	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6895	<p>This intrinsic allows annotations to be put on arbitrary expressions with
				6896	arbitrary strings. This can be useful for special purpose optimizations that
				6897	want to look for these annotations. These have no other defined use, they
				6898	are ignored by code generation and optimization.</p>
Tanya Lattner	b306a9e	2007-09-21 22:59:12 +0000	[diff] [blame]	6899
Tanya Lattner	b306a9e	2007-09-21 22:59:12 +0000	[diff] [blame]	6900	</div>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6901
Anton Korobeynikov	e6e764f	2008-01-15 22:31:34 +0000	[diff] [blame]	6902	<!-- _______________________________________________________________________ -->
				6903	<div class="doc_subsubsection">
				6904	<a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
				6905	</div>
				6906
				6907	<div class="doc_text">
				6908
				6909	<h5>Syntax:</h5>
				6910	<pre>
				6911	declare void @llvm.trap()
				6912	</pre>
				6913
				6914	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6915	<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikov	e6e764f	2008-01-15 22:31:34 +0000	[diff] [blame]	6916
				6917	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6918	<p>None.</p>
Anton Korobeynikov	e6e764f	2008-01-15 22:31:34 +0000	[diff] [blame]	6919
				6920	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6921	<p>This intrinsics is lowered to the target dependent trap instruction. If the
				6922	target does not have a trap instruction, this intrinsic will be lowered to
				6923	the call of the <tt>abort()</tt> function.</p>
Anton Korobeynikov	e6e764f	2008-01-15 22:31:34 +0000	[diff] [blame]	6924
Anton Korobeynikov	e6e764f	2008-01-15 22:31:34 +0000	[diff] [blame]	6925	</div>
				6926
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	6927	<!-- _______________________________________________________________________ -->
				6928	<div class="doc_subsubsection">
Misha Brukman	5dd7f4d	2008-11-22 23:55:29 +0000	[diff] [blame]	6929	<a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	6930	</div>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6931
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	6932	<div class="doc_text">
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6933
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	6934	<h5>Syntax:</h5>
				6935	<pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6936	declare void @llvm.stackprotector( i8* <guard>, i8** <slot> )
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	6937	</pre>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6938
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	6939	<h5>Overview:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6940	<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
				6941	stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
				6942	ensure that it is placed on the stack before local variables.</p>
				6943
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	6944	<h5>Arguments:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6945	<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
				6946	arguments. The first argument is the value loaded from the stack
				6947	guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
				6948	that has enough space to hold the value of the guard.</p>
				6949
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	6950	<h5>Semantics:</h5>
Bill Wendling	f85859d	2009-07-20 02:29:24 +0000	[diff] [blame]	6951	<p>This intrinsic causes the prologue/epilogue inserter to force the position of
				6952	the <tt>AllocaInst</tt> stack slot to be before local variables on the
				6953	stack. This is to ensure that if a local variable on the stack is
				6954	overwritten, it will destroy the value of the guard. When the function exits,
				6955	the guard on the stack is checked against the original guard. If they're
				6956	different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
				6957	function.</p>
				6958
Bill Wendling	e416459	2008-11-19 05:56:17 +0000	[diff] [blame]	6959	</div>
				6960
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6961	<!-- *********************************************************************** -->
				6962	<hr>
				6963	<address>
				6964	<a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman	947321d	2008-12-11 17:34:48 +0000	[diff] [blame]	6965	src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6966	<a href="http://validator.w3.org/check/referer"><img
Misha Brukman	947321d	2008-12-11 17:34:48 +0000	[diff] [blame]	6967	src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6968
				6969	<a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
				6970	<a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
				6971	Last modified: $Date$
				6972	</address>
Chris Lattner	08497ce	2008-01-04 04:33:49 +0000	[diff] [blame]	6973
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6974	</body>
				6975	</html>