- Oct 11, 2013
-
-
Justin Holewinski authored
Make AsmPrinter::emitImplicitDef a virtual method so targets can emit custom comments for implicit defs For NVPTX, this fixes a crash where the emitImplicitDef implementation was expecting physical registers, while NVPTX uses virtual registers (with a couple of exceptions). Now, the implicit def comment will be emitted as a true PTX register name. Other targets can use this to customize the output of implicit def comments. Fixes PR17519 llvm-svn: 192444
-
Sergey Matveev authored
llvm-svn: 192443
-
Sergey Matveev authored
Add a test. llvm-svn: 192442
-
Alexey Samsonov authored
llvm-svn: 192441
-
Amara Emerson authored
llvm-svn: 192440
-
Robert Lytton authored
llvm-svn: 192439
-
Daniel Sanders authored
[mips][msa] Added support for matching maddv.[bhwd], and msubv.[bhwd] from normal IR (i.e. not intrinsics) llvm-svn: 192438
-
Robert Lytton authored
llvm-svn: 192437
-
Robert Lytton authored
Incorrect handling of 'double' and 'long long int' llvm-svn: 192436
-
Daniel Sanders authored
llvm-svn: 192435
-
Robert Lytton authored
llvm-svn: 192434
-
Robert Lytton authored
llvm-svn: 192433
-
Robert Lytton authored
When a ConstantExpr which uses a thread local is part of a PHI node instruction, the insruction that replaces the ConstantExpr must be inserted in the predecessor block, in front of the terminator instruction. If the predecessor block has multiple successors, the edge is first split. llvm-svn: 192432
-
Robert Lytton authored
llvm-svn: 192431
-
Daniel Sanders authored
llvm-svn: 192430
-
Daniel Sanders authored
[mips][msa] Added support for matching ffint_[us].[wd], and ftrunc_[us].[wd] from normal IR (i.e. not intrinsics) llvm-svn: 192429
-
Alexey Samsonov authored
[Sanitizer] Simplify StackTrace::FastUnwindStack interface and fix a bug with one-frame stack traces llvm-svn: 192428
-
Rui Ueyama authored
llvm-svn: 192427
-
Tobias Grosser authored
llvm-svn: 192426
-
Craig Topper authored
llvm-svn: 192425
-
Rui Ueyama authored
llvm-svn: 192424
-
Rui Ueyama authored
allowRemainingUndefines() is already checked in Resolver::resolve(), so we don't need to check it again after returning from it. It's actually not only superfluous but buggy because a failure of resolve() does not always mean that there is a remaining undefines. llvm-svn: 192423
-
Rui Ueyama authored
llvm-svn: 192422
-
NAKAMURA Takumi authored
llvm-svn: 192421
-
Jason Molenda authored
llvm-svn: 192420
-
Douglas Gregor authored
Fixes <rdar://problem/14468891>. llvm-svn: 192419
-
Rui Ueyama authored
This eliminates _elements's size check from the loop. llvm-svn: 192418
-
Nick Lewycky authored
Mark it as requiring the x86 target. llvm-svn: 192417
-
Rui Ueyama authored
llvm-svn: 192416
-
Rui Ueyama authored
This patch also handles errors other than no_more_files error. They were silently ignored. llvm-svn: 192415
-
Shankar Easwaran authored
llvm-svn: 192414
-
Nick Lewycky authored
-f{function,data}-sections had no tests at all, add some, and verify that the -fno variants work as well. llvm-svn: 192413
-
Nick Lewycky authored
llvm-svn: 192412
-
Kevin Qin authored
vget_lane, vset_lane, vcopy_lane, vcreate, vdup_n, vdup_lane, vmov_n. llvm-svn: 192411
-
Kevin Qin authored
llvm-svn: 192410
-
Shankar Easwaran authored
This also reverts the LayoutPass to use std::sort as all files have an ordinal now. llvm-svn: 192409
-
Ed Maste authored
MIPS's .dyanamic section is read-only. Instead of using DT_DEBUG for the pointer to dyld information it uses a separate tag DT_MIPS_RLD_MAP which points to storage in the read-write .rld_map section, which in turn points to the dyld information. Review: http://llvm-reviews.chandlerc.com/D1890 llvm-svn: 192408
-
Fariborz Jahanian authored
don't leave a blank line behind replacing the typedef decl. // rdar://15200949 llvm-svn: 192407
-
Matt Arsenault authored
llvm-svn: 192406
-
Hans Wennborg authored
This was just broken. llvm-svn: 192405
-