blob: 10b35b56f9d3b073c64eaa5dbb98ecd3ff6da2ba [file] [log] [blame]
Fangrui Songee4e2e72018-01-30 00:40:05 +00001import re
Fangrui Song4f0f4262018-02-10 05:01:33 +00002import sys
Fangrui Songee4e2e72018-01-30 00:40:05 +00003
4from . import common
5
Fangrui Song4f0f4262018-02-10 05:01:33 +00006if sys.version_info[0] > 2:
7 class string:
8 expandtabs = str.expandtabs
9else:
10 import string
11
Fangrui Songee4e2e72018-01-30 00:40:05 +000012# RegEx: this is where the magic happens.
13
Fangrui Song4f0f4262018-02-10 05:01:33 +000014##### Assembly parser
15
Fangrui Songee4e2e72018-01-30 00:40:05 +000016ASM_FUNCTION_X86_RE = re.compile(
17 r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n[^:]*?'
18 r'(?P<body>^##?[ \t]+[^:]+:.*?)\s*'
19 r'^\s*(?:[^:\n]+?:\s*\n\s*\.size|\.cfi_endproc|\.globl|\.comm|\.(?:sub)?section|#+ -- End function)',
20 flags=(re.M | re.S))
21
22ASM_FUNCTION_ARM_RE = re.compile(
23 r'^(?P<func>[0-9a-zA-Z_]+):\n' # f: (name of function)
24 r'\s+\.fnstart\n' # .fnstart
25 r'(?P<body>.*?)\n' # (body of the function)
26 r'.Lfunc_end[0-9]+:', # .Lfunc_end0: or # -- End function
27 flags=(re.M | re.S))
28
29ASM_FUNCTION_AARCH64_RE = re.compile(
30 r'^_?(?P<func>[^:]+):[ \t]*\/\/[ \t]*@(?P=func)\n'
Sanjay Patel21d9c702018-04-20 17:16:23 +000031 r'(?:[ \t]+.cfi_startproc\n)?' # drop optional cfi noise
Fangrui Songee4e2e72018-01-30 00:40:05 +000032 r'(?P<body>.*?)\n'
33 # This list is incomplete
34 r'.Lfunc_end[0-9]+:\n',
35 flags=(re.M | re.S))
36
37ASM_FUNCTION_MIPS_RE = re.compile(
38 r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n[^:]*?' # f: (name of func)
39 r'(?:^[ \t]+\.(frame|f?mask|set).*?\n)+' # Mips+LLVM standard asm prologue
40 r'(?P<body>.*?)\n' # (body of the function)
41 r'(?:^[ \t]+\.(set|end).*?\n)+' # Mips+LLVM standard asm epilogue
42 r'(\$|\.L)func_end[0-9]+:\n', # $func_end0: (mips32 - O32) or
43 # .Lfunc_end0: (mips64 - NewABI)
44 flags=(re.M | re.S))
45
46ASM_FUNCTION_PPC_RE = re.compile(
47 r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n'
48 r'\.Lfunc_begin[0-9]+:\n'
49 r'(?:[ \t]+.cfi_startproc\n)?'
50 r'(?:\.Lfunc_[gl]ep[0-9]+:\n(?:[ \t]+.*?\n)*)*'
51 r'(?P<body>.*?)\n'
52 # This list is incomplete
53 r'(?:^[ \t]*(?:\.long[ \t]+[^\n]+|\.quad[ \t]+[^\n]+)\n)*'
54 r'.Lfunc_end[0-9]+:\n',
55 flags=(re.M | re.S))
56
57ASM_FUNCTION_RISCV_RE = re.compile(
58 r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n[^:]*?'
59 r'(?P<body>^##?[ \t]+[^:]+:.*?)\s*'
60 r'.Lfunc_end[0-9]+:\n',
61 flags=(re.M | re.S))
62
Daniel Cederman1c8fb182018-04-20 07:59:13 +000063ASM_FUNCTION_SPARC_RE = re.compile(
64 r'^_?(?P<func>[^:]+):[ \t]*!+[ \t]*@(?P=func)\n'
65 r'(?P<body>.*?)\s*'
66 r'.Lfunc_end[0-9]+:\n',
67 flags=(re.M | re.S))
68
Fangrui Songee4e2e72018-01-30 00:40:05 +000069ASM_FUNCTION_SYSTEMZ_RE = re.compile(
70 r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n'
71 r'[ \t]+.cfi_startproc\n'
72 r'(?P<body>.*?)\n'
73 r'.Lfunc_end[0-9]+:\n',
74 flags=(re.M | re.S))
75
76
77SCRUB_LOOP_COMMENT_RE = re.compile(
78 r'# =>This Inner Loop Header:.*|# in Loop:.*', flags=re.M)
79
80SCRUB_X86_SHUFFLES_RE = (
81 re.compile(
82 r'^(\s*\w+) [^#\n]+#+ ((?:[xyz]mm\d+|mem)( \{%k\d+\}( \{z\})?)? = .*)$',
83 flags=re.M))
Chandler Carruth6646bec2018-04-03 09:57:05 +000084SCRUB_X86_SPILL_RELOAD_RE = (
85 re.compile(
86 r'-?\d+\(%([er])[sb]p\)(.*(?:Spill|Reload))$',
87 flags=re.M))
Fangrui Songee4e2e72018-01-30 00:40:05 +000088SCRUB_X86_SP_RE = re.compile(r'\d+\(%(esp|rsp)\)')
89SCRUB_X86_RIP_RE = re.compile(r'[.\w]+\(%rip\)')
90SCRUB_X86_LCP_RE = re.compile(r'\.LCPI[0-9]+_[0-9]+')
91SCRUB_X86_RET_RE = re.compile(r'ret[l|q]')
92
93def scrub_asm_x86(asm, args):
94 # Scrub runs of whitespace out of the assembly, but leave the leading
95 # whitespace in place.
96 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
97 # Expand the tabs used for indentation.
98 asm = string.expandtabs(asm, 2)
99 # Detect shuffle asm comments and hide the operands in favor of the comments.
100 asm = SCRUB_X86_SHUFFLES_RE.sub(r'\1 {{.*#+}} \2', asm)
Chandler Carruth6646bec2018-04-03 09:57:05 +0000101 # Detect stack spills and reloads and hide their exact offset and whether
102 # they used the stack pointer or frame pointer.
Chandler Carruthff2f4fc2018-04-03 10:28:56 +0000103 asm = SCRUB_X86_SPILL_RELOAD_RE.sub(r'{{[-0-9]+}}(%\1{{[sb]}}p)\2', asm)
Fangrui Songee4e2e72018-01-30 00:40:05 +0000104 # Generically match the stack offset of a memory operand.
105 asm = SCRUB_X86_SP_RE.sub(r'{{[0-9]+}}(%\1)', asm)
106 # Generically match a RIP-relative memory operand.
107 asm = SCRUB_X86_RIP_RE.sub(r'{{.*}}(%rip)', asm)
108 # Generically match a LCP symbol.
109 asm = SCRUB_X86_LCP_RE.sub(r'{{\.LCPI.*}}', asm)
Simon Pilgrimee769442018-06-01 13:37:01 +0000110 if getattr(args, 'extra_scrub', False):
Fangrui Songee4e2e72018-01-30 00:40:05 +0000111 # Avoid generating different checks for 32- and 64-bit because of 'retl' vs 'retq'.
112 asm = SCRUB_X86_RET_RE.sub(r'ret{{[l|q]}}', asm)
113 # Strip kill operands inserted into the asm.
114 asm = common.SCRUB_KILL_COMMENT_RE.sub('', asm)
115 # Strip trailing whitespace.
116 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
117 return asm
118
119def scrub_asm_arm_eabi(asm, args):
120 # Scrub runs of whitespace out of the assembly, but leave the leading
121 # whitespace in place.
122 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
123 # Expand the tabs used for indentation.
124 asm = string.expandtabs(asm, 2)
125 # Strip kill operands inserted into the asm.
126 asm = common.SCRUB_KILL_COMMENT_RE.sub('', asm)
127 # Strip trailing whitespace.
128 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
129 return asm
130
131def scrub_asm_powerpc64(asm, args):
132 # Scrub runs of whitespace out of the assembly, but leave the leading
133 # whitespace in place.
134 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
135 # Expand the tabs used for indentation.
136 asm = string.expandtabs(asm, 2)
137 # Stripe unimportant comments
138 asm = SCRUB_LOOP_COMMENT_RE.sub(r'', asm)
139 # Strip trailing whitespace.
140 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
141 return asm
142
143def scrub_asm_mips(asm, args):
144 # Scrub runs of whitespace out of the assembly, but leave the leading
145 # whitespace in place.
146 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
147 # Expand the tabs used for indentation.
148 asm = string.expandtabs(asm, 2)
149 # Strip trailing whitespace.
150 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
151 return asm
152
153def scrub_asm_riscv(asm, args):
154 # Scrub runs of whitespace out of the assembly, but leave the leading
155 # whitespace in place.
156 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
157 # Expand the tabs used for indentation.
158 asm = string.expandtabs(asm, 2)
159 # Strip trailing whitespace.
160 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
161 return asm
162
Daniel Cederman1c8fb182018-04-20 07:59:13 +0000163def scrub_asm_sparc(asm, args):
164 # Scrub runs of whitespace out of the assembly, but leave the leading
165 # whitespace in place.
166 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
167 # Expand the tabs used for indentation.
168 asm = string.expandtabs(asm, 2)
169 # Strip trailing whitespace.
170 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
171 return asm
172
Fangrui Songee4e2e72018-01-30 00:40:05 +0000173def scrub_asm_systemz(asm, args):
174 # Scrub runs of whitespace out of the assembly, but leave the leading
175 # whitespace in place.
176 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
177 # Expand the tabs used for indentation.
178 asm = string.expandtabs(asm, 2)
179 # Strip trailing whitespace.
180 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
181 return asm
182
183
184def build_function_body_dictionary_for_triple(args, raw_tool_output, triple, prefixes, func_dict):
185 target_handlers = {
186 'x86_64': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
187 'i686': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
188 'x86': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
189 'i386': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
190 'aarch64': (scrub_asm_arm_eabi, ASM_FUNCTION_AARCH64_RE),
191 'arm-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
192 'thumb-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
193 'thumbv6': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
194 'thumbv6-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
195 'thumbv6t2': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
196 'thumbv6t2-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
197 'thumbv6m': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
198 'thumbv6m-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
199 'thumbv7': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
200 'thumbv7-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
201 'thumbv7m': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
202 'thumbv7m-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
203 'thumbv8-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
204 'thumbv8m.base': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
205 'thumbv8m.main': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
206 'armv6': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
207 'armv7': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
208 'armv7-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
209 'armeb-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
210 'armv7eb-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
211 'armv7eb': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
212 'mips': (scrub_asm_mips, ASM_FUNCTION_MIPS_RE),
213 'powerpc64': (scrub_asm_powerpc64, ASM_FUNCTION_PPC_RE),
214 'powerpc64le': (scrub_asm_powerpc64, ASM_FUNCTION_PPC_RE),
215 'riscv32': (scrub_asm_riscv, ASM_FUNCTION_RISCV_RE),
216 'riscv64': (scrub_asm_riscv, ASM_FUNCTION_RISCV_RE),
Daniel Cederman1c8fb182018-04-20 07:59:13 +0000217 'sparc': (scrub_asm_sparc, ASM_FUNCTION_SPARC_RE),
218 'sparcv9': (scrub_asm_sparc, ASM_FUNCTION_SPARC_RE),
Fangrui Songee4e2e72018-01-30 00:40:05 +0000219 's390x': (scrub_asm_systemz, ASM_FUNCTION_SYSTEMZ_RE),
220 }
221 handlers = None
222 for prefix, s in target_handlers.items():
223 if triple.startswith(prefix):
224 handlers = s
225 break
226 else:
227 raise KeyError('Triple %r is not supported' % (triple))
228
229 scrubber, function_re = handlers
230 common.build_function_body_dictionary(
231 function_re, scrubber, [args], raw_tool_output, prefixes,
232 func_dict, args.verbose)
Fangrui Song4f0f4262018-02-10 05:01:33 +0000233
234##### Generator of assembly CHECK lines
235
Simon Pilgrim702ec042018-04-05 09:50:58 +0000236def add_asm_checks(output_lines, comment_marker, prefix_list, func_dict, func_name):
237 # Label format is based on ASM string.
238 check_label_format = '{} %s-LABEL: %s:'.format(comment_marker)
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000239 common.add_checks(output_lines, comment_marker, prefix_list, func_dict, func_name, check_label_format, True, False)