- eb2d1f1 Removing a bunch of dead returns/breaks after llvm_unreachables. by David Blaikie · 13 years ago
- b219cfc Switch assert(0/false) llvm_unreachable. by David Blaikie · 13 years ago
- bcfd1f5 Extend the ASTContext constructor to delay the initialization of by Douglas Gregor · 13 years ago
- cec52f0 Slight optimization enabled by the previous assert: by John McCall · 13 years ago
- 8affed5 Assert that a call argument is a gl-value iff the parameter is a reference type. by John McCall · 13 years ago
- 410ffb2 Track whether an AggValueSlot is potentially aliased, and do not by John McCall · 13 years ago
- 041087c 'pure' and 'const' functions should also be marked nounwind. Migrate by Eric Christopher · 13 years ago
- 194f06a Handle "homogeneous aggregates" as required by the ARM AAPCS-VFP ABI. by Bob Wilson · 13 years ago
- 2d3ba4f Move ArrayRef to LLVM.h and eliminate now-redundant qualifiers, patch by Jon Mulder! by Chris Lattner · 13 years ago
- 5f9e272 remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports by Chris Lattner · 13 years ago
- b13eab9 fix rdar://9780211 - Clang crashes with an assertion failure building WKView.mm from WebKit by Chris Lattner · 13 years ago
- 2acc6e3 de-constify llvm::Type, patch by David Blaikie! by Chris Lattner · 13 years ago
- 4c7d9f1 Convert CallInst and InvokeInst APIs to use ArrayRef. by Jay Foad · 13 years ago
- d26c071 protect some calls to ConvertType when a function info is under construction by Chris Lattner · 13 years ago
- 71305cc Enhance the IR type lowering code to be much smarter about recursively lowering by Chris Lattner · 13 years ago
- 6af13f3 PR10337 reminds me that calls return values, lets handle them just by Chris Lattner · 13 years ago
- 811bf36 insert a bitcast in the 'expand' case of argument passing when needed. This by Chris Lattner · 13 years ago
- 21ca1fd fix PR10335 by watching out for IR type compatibility in call argument lists. by Chris Lattner · 13 years ago
- 7085544 add some aggressive assertions that call arguments match up to callee by Chris Lattner · 13 years ago
- be5f332 how about we initialize RecursionState. by Chris Lattner · 13 years ago
- f742eb0 Rename CGT::VerifyFuncTypeComplete to isFuncTypeConvertible since by Chris Lattner · 13 years ago
- 9cbe4f0 clang side to match the LLVM IR type system rewrite patch. by Chris Lattner · 13 years ago
- 545d996 LValue carries a type now, so simplify the main EmitLoad/Store APIs by John McCall · 13 years ago
- f85e193 Automatic Reference Counting. by John McCall · 13 years ago
- 97cb5a4 Per Chris's suggestion, simplify code using llvm::getOrEnforceKnownAlignment. by Eli Friedman · 13 years ago
- 70cbd2a Fix a regression from r132957 involving complex integers. (Fixes failures on gcc-testsuite bot.) by Eli Friedman · 13 years ago
- ea5e4da The LLVM IR representation of byval arguments has a rather strange property: if the alignment of an argument to a call is less than the specified byval alignment for that argument, there is no way to specify the alignment of the implied copy. Therefore, we must ensure that the alignment of the argument is at least the byval alignment. To do this, we have to mess with the alignment of relevant alloca's in some cases, and insert a copy that conceptually shouldn't be necessary in some cases. by Eli Friedman · 13 years ago
- 55d4848 Skip extra copy from aggregate where it isn't necessary; rdar://problem/8139919 . This shouldn't make much of a difference at -O3, but should substantially reduce the number of generated memcpy's at -O0. by Eli Friedman · 13 years ago
- 575a1c9 Introduce Type::isSignedIntegerOrEnumerationType() and by Douglas Gregor · 13 years ago
- badea57 For calls returning first-class aggregates, store by element instead of creating aggregate stores in common cases. This is more friendly to fast-isel. by Eli Friedman · 13 years ago
- 42e0611 More killing of std::vector. by John McCall · 13 years ago
- 6f08777 Revert r130717, which caused a regression (<rdar://problem/9402621>). by Douglas Gregor · 13 years ago
- b5a03b6 Skip extra copy from aggregate where it isn't necessary; rdar://problem/8139919 . This shouldn't make much of a difference at -O3, but should substantially reduce the number of generated memcpy's at -O0. by Eli Friedman · 13 years ago
- c6d0782 Switch CallArgList from an std::pair to a new CallArg struct (which will eventually gain more members). Working towards modifying call emission to avoid unnecessary copies. by Eli Friedman · 13 years ago
- 414d896 Implement ARM pcs attribute. Basically it's another way of calling convention selection (AAPCS or by Anton Korobeynikov · 13 years ago
- a49218e PR8369: make __attribute((regparm(0))) work correctly. Original patch by by Eli Friedman · 13 years ago
- 207f4d8 Add support for language-specific address spaces. On top of that, by Peter Collingbourne · 13 years ago
- 8026f6d Instead of storing an ASTContext* in FunctionProtoTypes with computed noexcept specifiers, unique FunctionProtoTypes with a ContextualFoldingSet, as suggested by John McCall. by Sebastian Redl · 13 years ago
- 60618fa Propagate the new exception information to FunctionProtoType. by Sebastian Redl · 13 years ago
- 413ebdb Use a slightly more semantic interface for emitting call arguments. by John McCall · 13 years ago
- c95a8fc lib/CodeGen/CGCall.cpp: Don't invoke multiple Builder.CreateBitCast() on Builder.CreateMemCpy. Or we would see sideeffect incompatibility among gcc and clang. by NAKAMURA Takumi · 13 years ago
- d26bc76 Use the "undergoes default argument promotion" bit on parameters to by John McCall · 13 years ago
- 093ac46 Encode argument numbering in debug info so that code generator can emit them in order. by Devang Patel · 13 years ago
- 9c6082f Revert "Add CC_Win64ThisCall and set it in the necessary places." by Tilmann Scheller · 13 years ago
- 41ba267 revert r126858. by Devang Patel · 13 years ago
- 88d117c Add CC_Win64ThisCall and set it in the necessary places. by Tilmann Scheller · 13 years ago
- 810b07c by Devang Patel · 13 years ago
- 3da5cb2 Revert r126794. by Devang Patel · 13 years ago
- 8a04585 Encode argument numbering in debug info so that code generator can emit them in order. by Devang Patel · 13 years ago
- c0ddef2 objc IRGen for Next runtime message API. by Fariborz Jahanian · 13 years ago
- 3475380 Simplify test to check an aggregate argument that has non trivial constructor or destructor. by Devang Patel · 13 years ago
- 7f6890e Fix family-friendly-o, tsk tsk. by Daniel Dunbar · 13 years ago
- 17d3fea Driver/Frontend: Wire up -mregparm=. by Daniel Dunbar · 13 years ago
- 570729e If an aggregate argument is passed indirectly because it has non trivial by Devang Patel · 13 years ago
- d16c2cf Reorganize CodeGen{Function,Module} to eliminate the unfortunate by John McCall · 13 years ago
- fe71008 Replace calls to getTypeSize() and getTypeAlign() with their 'InChars' by Ken Dyck · 14 years ago
- 9f0c7cc Simplify mem{cpy, move, set} creation with IRBuilder. by Benjamin Kramer · 14 years ago
- 9cac494 Fix Whitespace. by Michael J. Spencer · 14 years ago
- cf3b6f2 IRgen/ABI: Add support for realigning structures which are passed by indirect by Daniel Dunbar · 14 years ago
- 52fc314 Add symantic support for the Pascal calling convention via by Dawn Perchik · 14 years ago
- fc40028 Re-commit r112916 with an additional fix for the self-host failures. by John McCall · 14 years ago
- 2b8d60d Revert r112916, it's breaking selfhost pretty badly. by John McCall · 14 years ago
- 52968ab It's not safe to use the generic CXXMethodDecl overload of CGT::getFunctionInfo by John McCall · 14 years ago
- 4c40d98 Teach IR generation to return 'this' from constructors and destructors by John McCall · 14 years ago
- 79c3928 IRgen: Switch more MakeAddr() users to MakeAddrLValue; this time for calls which were previously not computing the qualifier list. In most cases, I don't think it matters, but I believe this is conservatively more correct / consistent. by Daniel Dunbar · 14 years ago
- 91a16fa IRgen: Change Emit{Load,Store}OfScalar to take a required Alignment argument and by Daniel Dunbar · 14 years ago
- 117e3f4 fix PR5179 and correctly fix PR5831 to not miscompile. by Chris Lattner · 14 years ago
- 1ed7267 fix a builder, why didn't clang++ catch this? by Chris Lattner · 14 years ago
- 800588f Kill off the 'coerce' ABI passing form. Now 'direct' and 'extend' always by Chris Lattner · 14 years ago
- ee5dcd0 dissolve some more complexity: make the x86-64 abi lowering code by Chris Lattner · 14 years ago
- a3c109b now that ABIInfo depends on CGT, it has trivial access to such by Chris Lattner · 14 years ago
- 2eb9cdd tidy up by Chris Lattner · 14 years ago
- deabde2 some cleanups and get alignments correct for various coerce cases. by Chris Lattner · 14 years ago
- f609462 Vectors are not integer types, so the type system should not classify by Douglas Gregor · 14 years ago
- d3f265d Fix regression caused by r108911. by Devang Patel · 14 years ago
- 4751a53 Use getDebugLoc and setDebugLoc instead of getDbgMetadata and setDbgMetadata, by Dan Gohman · 14 years ago
- dacf9dd CodeGen/ObjC/NeXT: Fix Obj-C message send to match llvm-gcc when choosing by Daniel Dunbar · 14 years ago
- 9c0c1f3 Mark calls to 'throw()' functions as nounwind, and mark the functions nounwind by John McCall · 14 years ago
- f1549f6 Validated by nightly-test runs on x86 and x86-64 darwin, including after by John McCall · 14 years ago
- 9282688 Generate fewer first class aggregate values for other by Chris Lattner · 14 years ago
- 121b3fa in the "coerce" case, the ABI handling code ends up making the by Chris Lattner · 14 years ago
- a9fa858 fix rdar://8147692 - yet another crash due to my abi work. by Chris Lattner · 14 years ago
- 21fcc8f IRgen: Fix debug info regression in r106970; when we eliminate the return value by Daniel Dunbar · 14 years ago
- bcaedae Reapply: by Chris Lattner · 14 years ago
- 7bbe03d Revert r107173, "fix PR7519: after thrashing around and remembering how all this stuff", it broke bootstrap. by Daniel Dunbar · 14 years ago
- a9e0515 Revert r107216, "fix PR7523, which was caused by the ABI code calling ConvertType instead", it is part of a boostrap breaking sequence. by Daniel Dunbar · 14 years ago
- f1d5d8a fix PR7523, which was caused by the ABI code calling ConvertType instead by Chris Lattner · 14 years ago
- bb52114 relax the CGFunctionInfo::CGFunctionInfo ctor to allow any sequence by Chris Lattner · 14 years ago
- 376fe5e fix PR7519: after thrashing around and remembering how all this stuff by Chris Lattner · 14 years ago
- ce93399 minor cleanups. by Chris Lattner · 14 years ago
- 8640cd6 Pass the LLVM IR version of argument types down into computeInfo. by Chris Lattner · 14 years ago
- 225e286 add IR names to coerced arguments. by Chris Lattner · 14 years ago
- 309c59f make the argument passing stuff in the FCA case smarter still, by by Chris Lattner · 14 years ago
- ce70016 Change CGCall to handle the "coerce" case where the coerce-to type by Chris Lattner · 14 years ago
- 6ae0069 make the trivial forms of CreateCoerced{Load|Store} trivial. by Chris Lattner · 14 years ago
- 77b89b8 finally get around to doing a significant cleanup to irgen: by Chris Lattner · 14 years ago
- 6d11cdb If coercing something from int or pointer type to int or pointer type by Chris Lattner · 14 years ago
- e7bb777 Same patch as the previous on the store side. Before we compiled this: by Chris Lattner · 14 years ago
- 08dd2a0 improve CreateCoercedLoad a bit to generate slightly less awful by Chris Lattner · 14 years ago
- 35b21b8 Change IR generation for return (in the simple case) to avoid doing silly by Chris Lattner · 14 years ago