OSDN Git Service
Dan Liew [Thu, 26 May 2016 21:54:25 +0000 (21:54 +0000)]
[LibFuzzer] Add missing #include<string>
This partially fixes the compilation of the LibFuzzer unit test
on OSX using AppleClang.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270926
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Zolotukhin [Thu, 26 May 2016 21:42:51 +0000 (21:42 +0000)]
[LoopUnrollAnalyzer] Fix a crash in analyzeLoopUnrollCost.
Condition might be simplified to a Constant, but it doesn't have to be
ConstantInt, so we should dyn_cast, instead of cast.
This fixes PR27886.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270924
91177308-0d34-0410-b5e6-
96231b3b80d8
Adrian Prantl [Thu, 26 May 2016 21:42:47 +0000 (21:42 +0000)]
PR26055: Speed up LiveDebugValues::transferDebugValue()
This patch builds upon r270776 and speeds up
LiveDebugValues::transferDebugValue() by adding an index that maps each
DebugVariable to its open VarLoc.
The transferDebugValue() function needs to close all open ranges for a
given DebugVariable. Iterating over the set bits of OpenRanges is
prohibitively slow in practice. I experimented with using the sorted map
of VarLocs in the UniqueVector to iterate only over the range of VarLocs
with a given DebugVariable, but the binary search turned out to be even
more expensive than just iterating over the set bits in OpenRanges.
Instead, this patch exploits the fact that there can only be one open
location for each DebugVariable and redundantly stores this location in a
DenseMap.
This patch brings the time spent in the LiveDebugValues pass down to an
almost neglectiable amount.
http://llvm.org/bugs/show_bug.cgi?id=26055
http://reviews.llvm.org/D20636
rdar://problem/
24091200
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270923
91177308-0d34-0410-b5e6-
96231b3b80d8
Kostya Serebryany [Thu, 26 May 2016 21:32:30 +0000 (21:32 +0000)]
[libFuzzer] refactor: hide CurrentUnitData inside an interface function. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270922
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Bieneman [Thu, 26 May 2016 21:31:56 +0000 (21:31 +0000)]
[Docs][WritingAnLLVMBackend] Makefiles are deprecated
Summary: * docs/WritingAnLLVMBackend.rst: Makefiles are no longer used. The users should use CMakeLists.txt. In order to add the target, the TARGETS_TO_BUILD is replaced with LLVM_ALL_TARGETS.
Reviewers: gribozavr, void, beanz
Subscribers: llvm-commits
Patch By: Visoiu Mistrih Francis (thegameg)
Differential Revision: http://reviews.llvm.org/D20700
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270921
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Bieneman [Thu, 26 May 2016 21:29:39 +0000 (21:29 +0000)]
[obj2yaml][yaml2obj] Support for MachO lazy bindings
This adds support for YAML round tripping dyld info lazy bindings. The storage and format of these is the same as regular bind opcodes, they are just interpreted differently by dyld, and can have DONE opcodes in the middle of the opcode lists.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270920
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 21:27:52 +0000 (21:27 +0000)]
[Kaleidoscope][BuildingAJIT] Remove leftover debugging output statements.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270919
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 21:21:18 +0000 (21:21 +0000)]
[Kaleidoscope][BuildingAJIT] Fix strong emphasis markup by removing whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270918
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 21:17:06 +0000 (21:17 +0000)]
[Kaleidoscope][BuildingAJIT] Add docs for Chapter 3 of the Building A JIT
tutorial.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270917
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 21:15:58 +0000 (21:15 +0000)]
[Error] Make ECError only constructible via errorCodeToError.
This enforces idiomatic usage of ECError removing the option to construct them
using make_error.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270916
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjoy Das [Thu, 26 May 2016 21:12:44 +0000 (21:12 +0000)]
Add myself as the code owner for SCEV and IndVarSimplify
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270915
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 21:00:29 +0000 (21:00 +0000)]
[Kaleidoscope][BuildingAJIT] Add explicit dependence on TransformUtils to Chapter3.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270914
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Liew [Thu, 26 May 2016 20:55:09 +0000 (20:55 +0000)]
[LibFuzzer] Emit error if LLVM_USE_SANITIZER is not correctly set.
Previously CMake would successfully configure and compile (with warnings
about ``-fsanitize-coverage=...`` being unused) but the tests LibFuzzer
tests would fail.
Differential Revision: http://reviews.llvm.org/D20662
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270913
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Liew [Thu, 26 May 2016 20:55:05 +0000 (20:55 +0000)]
[LibFuzzer] Allow LibFuzzer to be built in modes other than RELEASE.
Previously the flags were only being set correctly when the
build type was "Release". Now the build should work properly
for all the supported build types. When building libFuzzer
the optimization level respects whatever is used for the
rest of LLVM but for the LibFuzzer tests we force -O0.
Differential Revision: http://reviews.llvm.org/D20558
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270912
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Bieneman [Thu, 26 May 2016 20:50:05 +0000 (20:50 +0000)]
[obj2yaml][yaml2obj] Support for MachO weak bindings
This adds support for YAML round tripping dyld info weak bindings. The storage and format of these is the same as regular bind opcodes, they are just interpreted differently by dyld.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270911
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 20:35:39 +0000 (20:35 +0000)]
[Kaleidoscope][BuildingAJIT] Add code for Chapter 3.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270909
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 20:33:37 +0000 (20:33 +0000)]
[Orc] Don't create empty globals modules in the CompileOnDemandLayer.
Global variables and aliases are emitted eagerly, but there may not be any in
the incoming module. In that case, we can save some memory and compile time by
not building, emitting and tracking an empty globals module.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270908
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Thu, 26 May 2016 20:31:00 +0000 (20:31 +0000)]
coff: fix weak alias to local.
We were creating a weak external that tried to reference a static symbol. That
would always fail to link with link.exe.
We now create an external symbol in the same position as the local and refer
to that. This works with link.exe and matches what gas does.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270906
91177308-0d34-0410-b5e6-
96231b3b80d8
Kostya Serebryany [Thu, 26 May 2016 20:25:49 +0000 (20:25 +0000)]
[libFuzzer] when there is a leak in the existing corpus report the reproducer properly
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270905
91177308-0d34-0410-b5e6-
96231b3b80d8
Richard Smith [Thu, 26 May 2016 20:21:55 +0000 (20:21 +0000)]
Don't use recursion to print out the PrettyStackTrace after a crash. If the
crash was due to a stack overflow, chances are good that this would also cause
a stack overflow.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270903
91177308-0d34-0410-b5e6-
96231b3b80d8
Richard Smith [Thu, 26 May 2016 20:13:50 +0000 (20:13 +0000)]
Add new .def file to module map.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270902
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Bieneman [Thu, 26 May 2016 20:06:14 +0000 (20:06 +0000)]
[obj2yaml][yaml2obj] Support for MachO bind opcodes
This adds support for YAML round tripping dyld info bind opcodes. Bind opcodes can have signed or unsigned LEB128 data, and they can have symbols associated with them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270901
91177308-0d34-0410-b5e6-
96231b3b80d8
Kostya Serebryany [Thu, 26 May 2016 20:03:02 +0000 (20:03 +0000)]
[libFuzzer] reimplement the way we do -only_ascii to allow more 'const' in function declarations. Add a test for -only_ascii. NFC intended
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270900
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 19:44:33 +0000 (19:44 +0000)]
[Kaleidoscope][BuildingAJIT] Fix a bug in the symbol resolver in Chapter2.
Symbol resolution should be done on the top layer of the stack unless there's a
good reason to do otherwise. In this case it would have worked because
OptimizeLayer::addModuleSet eagerly passes all modules down to the
CompileLayer, meaning that searches in CompileLayer will find the definitions.
In later chapters where the top layer's addModuleSet isn't a pass-through, this
would break.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270899
91177308-0d34-0410-b5e6-
96231b3b80d8
Krzysztof Parzyszek [Thu, 26 May 2016 19:44:28 +0000 (19:44 +0000)]
[Hexagon] Enable the post-RA scheduler
The aggressive anti-dependency breaker can rename the restored callee-
saved registers. To prevent this, mark these registers are live on all
paths to the return/tail-call instructions, and add implicit use operands
for them to these instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270898
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 26 May 2016 19:41:33 +0000 (19:41 +0000)]
[AArch64] Generate rev16/rev32 from bswap + srl when upper bits are known zero.
Canonicalize (srl (bswap i32 x), 16) to (rotr (bswap i32 x), 16), if the high
16-bits of x are zero. Similarly, canonicalize (srl (bswap i64 x), 32) to
(rotr (bswap i64 x), 32), if the high 32-bits of x are zero.
test_rev_w_srl16: test_rev_w_srl16:
and w8, w0, #0xffff and w8, w0, #0xffff
rev w8, w8 ---> rev16 w0, w8
lsr w0, w8, #16
test_rev_x_srl32: test_rev_x_srl32:
rev x8, x8 ---> rev32 x0, x8
lsr x0, x8, #32
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270896
91177308-0d34-0410-b5e6-
96231b3b80d8
Changpeng Fang [Thu, 26 May 2016 19:35:29 +0000 (19:35 +0000)]
AMDGPU/SI: Enable load-store-opt by default.
Summary: Enable load-store-opt by default, and update LIT tests.
Reviewers: arsenm
Differential Revision: http://reviews.llvm.org/D20694
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270894
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Kuperstein [Thu, 26 May 2016 19:30:49 +0000 (19:30 +0000)]
[BasicAA] Extend inbound GEP negative offset logic to GlobalVariables
r270777 improved the precision of alloca vs. inbounbds GEP alias queries: if
we have (a) an inbounds GEP and (b) a pointer based on an alloca, and the
beginning of the object the GEP points to would have a negative offset with
respect to the alloca, then the GEP can not alias pointer (b).
This makes the same logic fire when (b) is based on a GlobalVariable instead
of an alloca.
Differential Revision: http://reviews.llvm.org/D20652
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270893
91177308-0d34-0410-b5e6-
96231b3b80d8
David Majnemer [Thu, 26 May 2016 19:24:24 +0000 (19:24 +0000)]
[MemCpyOpt] Don't perform callslot optimization across may-throw calls
An exception could prevent a store from occurring but MemCpyOpt's
callslot optimization would fire anyway, causing the store to occur.
This fixes PR27849.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270892
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Thu, 26 May 2016 18:48:23 +0000 (18:48 +0000)]
coff: fix the section of weak symbols.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270889
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Kuperstein [Thu, 26 May 2016 18:43:57 +0000 (18:43 +0000)]
[BBVectorize] Don't vectorize selects with a scalar condition and vector operands.
This fixes PR27879.
Differential Revision: http://reviews.llvm.org/D20659
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270888
91177308-0d34-0410-b5e6-
96231b3b80d8
Krzysztof Parzyszek [Thu, 26 May 2016 18:35:45 +0000 (18:35 +0000)]
Add test/CodeGen/MIR/Hexagon/lit.local.cfg
Require that Hexagon is a registered target.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270887
91177308-0d34-0410-b5e6-
96231b3b80d8
Krzysztof Parzyszek [Thu, 26 May 2016 18:22:53 +0000 (18:22 +0000)]
Do not rename registers that do not start an independent live range
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270885
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Thu, 26 May 2016 18:04:53 +0000 (18:04 +0000)]
coff: fix the value of weak definitions.
It looks like this doesn't get a lot of use.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270883
91177308-0d34-0410-b5e6-
96231b3b80d8
Pete Cooper [Thu, 26 May 2016 17:40:07 +0000 (17:40 +0000)]
Don't allocate in APInt::slt. NFC.
APInt::slt was copying the LHS and RHS in to temporaries then making
them unsigned so that it could use an unsigned comparision. It did
this even on the paths which were trivial to give results for, such
as the sign bit of the LHS being set while RHS was not set.
This changes the logic to return out immediately in the trivial cases,
and use an unsigned comparison in the remaining cases. But this time,
just use the unsigned comparison directly without creating any temporaries.
This works because, for example:
true = (-2 slt -1) = (0xFE ult 0xFF)
Also added some tests explicitly for slt with APInt's larger than 64-bits
so that this new code is tested.
Using the memory for 'opt -O2 verify-uselistorder.lto.opt.bc -o opt.bc'
(see r236629 for details), this reduces the number of allocations from
26.8M to 23.9M.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270881
91177308-0d34-0410-b5e6-
96231b3b80d8
David Majnemer [Thu, 26 May 2016 17:36:22 +0000 (17:36 +0000)]
[CaptureTracking] Volatile operations capture their memory location
The memory location that corresponds to a volatile operation is very
special. They are observed by the machine in ways which we cannot
reason about.
Differential Revision: http://reviews.llvm.org/D20555
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270879
91177308-0d34-0410-b5e6-
96231b3b80d8
Artem Belevich [Thu, 26 May 2016 17:29:20 +0000 (17:29 +0000)]
Init member structs in constructor.
Fixes build error on windows where MSVC does not
support list initialization inside member initializer list.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270877
91177308-0d34-0410-b5e6-
96231b3b80d8
Xinliang David Li [Thu, 26 May 2016 17:27:22 +0000 (17:27 +0000)]
Revert 270865 -- unexplained bot failure on linux/ppcle
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270876
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 17:20:35 +0000 (17:20 +0000)]
[Orc] Merge some common code for creating CompileCallbackManagers and
IndirectStubsManagers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270874
91177308-0d34-0410-b5e6-
96231b3b80d8
Artem Belevich [Thu, 26 May 2016 17:02:56 +0000 (17:02 +0000)]
[NVPTX] Added NVVMIntrRange pass
NVVMIntrRange adds !range metadata to calls of NVVM intrinsics
that return values within known limited range.
This allows LLVM to generate optimal code for indexing arrays
based on tid/ctaid which is a frequently used pattern in CUDA code.
Differential Revision: http://reviews.llvm.org/D20644
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270872
91177308-0d34-0410-b5e6-
96231b3b80d8
Artem Tamazov [Thu, 26 May 2016 17:00:33 +0000 (17:00 +0000)]
[AMDGPU][llvm-mc] s_getreg/setreg* - hwreg - factor out strings/literals etc.
Hwreg(...) syntax implementation unified with sendmsg(...).
Common strings moved to Utils
MathExtras.h functionality utilized.
Added missing build dependency in Disassembler.
Differential Revision: http://reviews.llvm.org/D20381
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270871
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Bieneman [Thu, 26 May 2016 16:32:40 +0000 (16:32 +0000)]
[CMake] Restrict libxar linkage to just llvm-objdump
At some point we're going to need libObject to have this dependency, but as it is now this is causing too many headaches. This commit will reduce the linkage to just llvm-objdump where it is strictly needed, and we'll cross the libObject bridge later when we need it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270866
91177308-0d34-0410-b5e6-
96231b3b80d8
Xinliang David Li [Thu, 26 May 2016 16:28:01 +0000 (16:28 +0000)]
Use new interface in Triple /NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270865
91177308-0d34-0410-b5e6-
96231b3b80d8
Adrian Prantl [Thu, 26 May 2016 16:06:04 +0000 (16:06 +0000)]
Undo a suboptimal clang-format decision. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270861
91177308-0d34-0410-b5e6-
96231b3b80d8
Ismail Donmez [Thu, 26 May 2016 15:52:23 +0000 (15:52 +0000)]
Since some time clang itself figures out the default for ms-compatibility-version and uses it. Trying to figure it out during build is redundant and also will not work when the environment variable VSINSTALLDIR is not defined (which is not defined if you don't install whole Visual Studio but use Visual C++ Build Tools package).
Tested by bootstrapping clang with clang-cl.
Differential Revision: http://reviews.llvm.org/D20672
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270860
91177308-0d34-0410-b5e6-
96231b3b80d8
Artem Tamazov [Thu, 26 May 2016 15:52:16 +0000 (15:52 +0000)]
Fix build warning introduced in r270552 "[AMDGPU][llvm-mc] Disassembler: support for TTMP/TBA/TMA registers."
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270859
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Thu, 26 May 2016 15:40:36 +0000 (15:40 +0000)]
[X86][SSE] When lowering a 256-bit shuffle as PMOVZX, reduce the input vector to the lower 128-bit subvector.
Most often as not this is what it started out as, the extraction is zero-cost on AVX and the PMOVZX/PMOVSX folding logic is based around 128-bit loads.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270858
91177308-0d34-0410-b5e6-
96231b3b80d8
Krzysztof Parzyszek [Thu, 26 May 2016 15:38:50 +0000 (15:38 +0000)]
[Hexagon] Select the aggressive anti-dependency breaker
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270857
91177308-0d34-0410-b5e6-
96231b3b80d8
Diana Picus [Thu, 26 May 2016 15:24:55 +0000 (15:24 +0000)]
[AMDGPU] Remove exit-on-error flag from test (PR27762)
Similar to r269948, but for argument lowering.
Fixes PR27762
Differential Revision: http://reviews.llvm.org/D20430
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270856
91177308-0d34-0410-b5e6-
96231b3b80d8
Diana Picus [Thu, 26 May 2016 15:23:50 +0000 (15:23 +0000)]
[BPF] Remove exit-on-error flag in test (PR27767)
The exit-on-error flag is needed to avoid an assert where
llvm::SelectionDAGISel::LowerArguments doesn't create enough arguments. Fill up
with zeroes to reach the right number of args.
Fixes PR27767.
Differential Revision: http://reviews.llvm.org/D20571
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270855
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 26 May 2016 14:58:51 +0000 (14:58 +0000)]
[InstCombine] Catch more bswap cases missed due to zext and truncs.
Fixes PR27824.
Differential Revision: http://reviews.llvm.org/D20591.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270853
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Thu, 26 May 2016 14:45:30 +0000 (14:45 +0000)]
[X86][SSE] Added load_zext_16i8_to_8i32 test
Odd issue with input vector not being folded into pmovzx on AVX2+ targets
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270852
91177308-0d34-0410-b5e6-
96231b3b80d8
Teresa Johnson [Thu, 26 May 2016 14:16:52 +0000 (14:16 +0000)]
[ThinLTO] Resolve LinkOnceAny
Summary:
Ensure we keep prevailing copy of LinkOnceAny by converting it to
WeakAny.
Rename odr_resolution test to the now more appropriate weak_resolution
(weak in the linker sense includes linkonce).
Reviewers: joker.eph
Subscribers: llvm-commits, joker.eph
Differential Revision: http://reviews.llvm.org/D20634
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270850
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 26 May 2016 13:27:56 +0000 (13:27 +0000)]
[AArch64] Generate a BFI/BFXIL from 'or (and X, MaskImm), OrImm'.
If and only if the value being inserted sets only known zero bits.
This combine transforms things like
and w8, w0, #0xfffffff0
movz w9, #5
orr w0, w8, w9
into
movz w8, #5
bfxil w0, w8, #0, #4
The combine is tuned to make sure we always reduce the number of instructions.
We avoid churning code for what is expected to be performance neutral changes
(e.g., converted AND+OR to OR+BFI).
Differential Revision: http://reviews.llvm.org/D20387
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270846
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Thu, 26 May 2016 12:42:55 +0000 (12:42 +0000)]
Use shouldAssumeDSOLocal on AArch64.
This reduces code duplication and now AArch64 also handles PIE.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270844
91177308-0d34-0410-b5e6-
96231b3b80d8
Igor Breger [Thu, 26 May 2016 12:42:25 +0000 (12:42 +0000)]
[AVX512] Fix intrinsic cmp{sd|ss} lowering.
Differential Revision: http://reviews.llvm.org/D20615
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270843
91177308-0d34-0410-b5e6-
96231b3b80d8
John Brawn [Thu, 26 May 2016 11:16:43 +0000 (11:16 +0000)]
Add auto-exporting of symbols from tools so that plugins work on Windows
The problem with plugins on Windows is that when building a plugin DLL it needs
to explicitly link against something (an exe or DLL) if it uses symbols from
that thing, and that thing must explicitly export those symbols. Also there's a
limit of 65535 symbols that can be exported. This means that currently plugins
only work on Windows when using BUILD_SHARED_LIBS, and that doesn't work with
MSVC.
This patch adds an LLVM_EXPORT_SYMBOLS_FOR_PLUGINS option, which when enabled
automatically exports from all LLVM tools the symbols that a plugin could want
to use so that a plugin can link against a tool directly. Plugins can specify
what tool they link against by using PLUGIN_TOOL argument to llvm_add_library.
The option can also be enabled on Linux, though there all it should do is
restrict the set of symbols that are exported as by default all symbols are
exported.
This option is currently OFF by default, as while I've verified that it works
with MSVC, linux gcc, and cygwin gcc, I haven't tried mingw gcc and I have no
idea what will happen on OSX. Also unfortunately we can't turn on
LLVM_ENABLE_PLUGINS when the option is ON as bugpoint-passes needs to be
loaded by both bugpoint.exe and opt.exe which is incompatible with this
approach. Also currently clang plugins don't work with this approach, which
will be fixed in future patches.
Differential Revision: http://reviews.llvm.org/D18826
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270839
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Thu, 26 May 2016 10:26:56 +0000 (10:26 +0000)]
[X86][F16C] Added F16C fast-isel tests to match clang/test/CodeGen/f16c-builtins.c
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270837
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Thu, 26 May 2016 10:07:05 +0000 (10:07 +0000)]
[X86][AVX2] Added gather fast-isel tests to match clang/test/CodeGen/avx2-builtins.c
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270835
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Dewhurst [Thu, 26 May 2016 07:28:31 +0000 (07:28 +0000)]
[Sparc] Extend the assembler printing support for Sparc back-end.
Allows display of floating-point registers and display of assembler meta-data output.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270829
91177308-0d34-0410-b5e6-
96231b3b80d8
David Majnemer [Thu, 26 May 2016 07:11:09 +0000 (07:11 +0000)]
[MergedLoadStoreMotion] Don't transform across may-throw calls
It is unsafe to hoist a load before a function call which may throw, the
throw might prevent a pointer dereference.
Likewise, it is unsafe to sink a store after a call which may throw.
The caller might be able to observe the difference.
This fixes PR27858.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270828
91177308-0d34-0410-b5e6-
96231b3b80d8
Adam Nemet [Thu, 26 May 2016 07:08:09 +0000 (07:08 +0000)]
[ConstantFold] NFC cleanup after previous change.
Merge two conditions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270827
91177308-0d34-0410-b5e6-
96231b3b80d8
Adam Nemet [Thu, 26 May 2016 07:08:05 +0000 (07:08 +0000)]
[ConstantFold] Fix incorrect index rewrites for GEPs
Summary:
If an index for a vector or array type is out-of-range GEP constant
folding tries to factor it into preceding dimensions. The code however
does not consider addressing of structure field padding which should not
qualify as out-of-range index.
As demonstrated by the testcase, this can occur if the indexing
performed on a vector type and the preceding index is an array type.
SROA generates GEPs for example involving padding bytes as it slices an
alloca.
My fix disables this folding if the element type is a vector type. I
believe that this is the only way we can end up with padding. (We have
no access to DataLayout so I am not sure if there is actual robust way
of actually checking the presence of padding.)
Reviewers: majnemer
Subscribers: llvm-commits, Gerolf
Differential Revision: http://reviews.llvm.org/D20663
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270826
91177308-0d34-0410-b5e6-
96231b3b80d8
David Majnemer [Thu, 26 May 2016 05:43:12 +0000 (05:43 +0000)]
[MergedLoadStoreMotion] Small cleanup
No functional change is intended.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270824
91177308-0d34-0410-b5e6-
96231b3b80d8
Peter Collingbourne [Thu, 26 May 2016 04:58:46 +0000 (04:58 +0000)]
MemorySSA: Revert r269678 and r268068; replace with special casing in MemorySSA.
It turns out that too many passes are relying on alias analysis results
for control dependencies. Until we fix that by introducing a more accurate
modelling of control dependencies, special case assume in MemorySSA instead.
Also introduce tests to ensure we don't regress the FunctionAttrs or LICM
passes.
Differential Revision: http://reviews.llvm.org/D20658
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270823
91177308-0d34-0410-b5e6-
96231b3b80d8
Steven Wu [Thu, 26 May 2016 04:35:35 +0000 (04:35 +0000)]
Revert "[CMake] LINK_LIBS need to be public for Darwin dylib targets"
This reverts r270723. This commit breaks greendragon.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270820
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Thu, 26 May 2016 04:28:45 +0000 (04:28 +0000)]
[X86] Add the AVX storeu intrinsics to InstCombine and LoopStrengthReduce in the same places that the SSE/SSE2 storeu intrinsics appear.
I don't really know how to test this. Just seemed like we should be consistent.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270819
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjoy Das [Thu, 26 May 2016 01:50:18 +0000 (01:50 +0000)]
[IRCE] Use C++11 style initializers; NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270815
91177308-0d34-0410-b5e6-
96231b3b80d8
Teresa Johnson [Thu, 26 May 2016 01:46:41 +0000 (01:46 +0000)]
[ThinLTO/gold] Handle bitcode archives
Summary:
Several changes were required for ThinLTO links involving bitcode
archive static libraries. With this patch clang/llvm bootstraps with
ThinLTO and gold.
The first is that the gold callbacks get_input_file and
release_input_file can normally be used to get file information for
each constituent bitcode file within an archive. However, these
interfaces lock the underlying file and can't be for each archive
constituent for ThinLTO backends where we get all the input files up
front and don't release any until after the backend threads complete.
However, it is sufficient to only get and release once per file, and
then each consituent bitcode file can be accessed via get_view. This
required saving some information to identify which file handle is the
"leader" for each claimed file sharing the same file descriptor, and
other information so that get_input_file isn't necessary later when
processing the backends.
Second, the module paths in the index need to distinguish between
different constituent bitcode files within the same archive file,
otherwise they will all end up with the same archive file path.
Do this by appending the offset within the archive for the start of the
bitcode file, returned by get_input_file when we claim each bitcode file,
and saving that along with the file handle.
Third, rather than have the function importer try to load a file based
on the module path identifier (which now contains a suffix to
distinguish different bitcode files within an archive), use a custom
module loader. This is the same approach taken in libLTO, and I am using
the support refactored into the new LTO.h header in r270509. The module
loader parses the bitcode files out of the memory buffers returned from
gold via the get_view callback and saved in a map. This also means that
we call the function importer directly, rather than add it to the pass
pipeline (which was in the plan to do already for other reasons).
Reviewers: pcc, joker.eph
Subscribers: llvm-commits, joker.eph
Differential Revision: http://reviews.llvm.org/D20559
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270814
91177308-0d34-0410-b5e6-
96231b3b80d8
Saleem Abdulrasool [Thu, 26 May 2016 01:45:12 +0000 (01:45 +0000)]
llvm-objdump: support dumping AUX records for weak externals
This is a support COFF feature. Ensure that we can display the weak externals
auxiliary symbol. It contains useful information (such as the default binding
and how to resolve the symbol).
This reapplies the previous patch with a modification which hopefully should fix
the endianness issues. The variadic call would promote the ulittle32_t to a
uint32_t which would lose the byte-swapping behaviour desired.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270813
91177308-0d34-0410-b5e6-
96231b3b80d8
Peter Collingbourne [Thu, 26 May 2016 01:19:17 +0000 (01:19 +0000)]
MemorySSA: Remove argument to createNewAccess function.
There is only one caller of MemorySSA::createNewAccess, and it passes true
as the IgnoreNonMemory argument. Remove that argument and fold its behavior
into createNewAccess.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270812
91177308-0d34-0410-b5e6-
96231b3b80d8
Peter Collingbourne [Thu, 26 May 2016 01:19:14 +0000 (01:19 +0000)]
MemorySSA: Fix example in header comment.
This fixes the example so that it matches the pass's behavior. I was a
little confused by the example until I tried running it and realized that
there was a mistake.
Differential Revision: http://reviews.llvm.org/D20657
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270811
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 00:38:04 +0000 (00:38 +0000)]
[Kaleidoscope][BuildingAJIT] Add a stub Chapter 2 doc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270809
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 26 May 2016 00:24:18 +0000 (00:24 +0000)]
[Kaleidoscope][BuildingAJIT] Make the optimizeModule method for Chapter2
private.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270807
91177308-0d34-0410-b5e6-
96231b3b80d8
David Blaikie [Thu, 26 May 2016 00:22:26 +0000 (00:22 +0000)]
PR11740: Disable assembly debug info when assembly already contains line directives
If there is already debug info in the assembly file, and user hope to
use -g option for compiling, we think we should not directly report an
error.
According to what GNU assembler did, it just reused the debug info in
the assembly file, and turned off the DEBUG_TYPE option so that there
will be no new debug info emitted by assembler. This fix is just as what
GNU assembler did.
The concern is the situation that there are two .text sections in the
assembly file, one with debug info and the other one without. Currently
with this fix, the assembler will no longer generate any debug info for
the second .text section. And this is what GNU assembler exactly did for
this situation. So I think this still make some sense.
Patch by Zhizhou Yang!
Differential Revision: http://reviews.llvm.org/D20002
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270806
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjoy Das [Thu, 26 May 2016 00:09:02 +0000 (00:09 +0000)]
[IRCE] Optimize conjunctions of range checks
After this change, we do the expected thing for cases like
```
Check0Passed = /* range check IRCE can optimize */
Check1Passed = /* range check IRCE can optimize */
if (!(Check0Passed && Check1Passed))
throw_Exception();
```
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270804
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjoy Das [Thu, 26 May 2016 00:08:24 +0000 (00:08 +0000)]
[IRCE] Refactor out a parseRangeCheckFromCond; NFC
This will later hold more general logic to parse conjunctions of range
checks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270802
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Wed, 25 May 2016 23:44:01 +0000 (23:44 +0000)]
[PM/PartiallyInlineLibCalls] Commit missing header.
This should have been committed with the previous commit, but
I forgot to `git add`. Sorry.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270800
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Wed, 25 May 2016 23:42:48 +0000 (23:42 +0000)]
[KaleidoscopeJIT][BuildingAJIT] Remove some copy-pasta from Chapter 1.
This text was accidentally left in when the original document was copied from
Chapter 7 of the Kaleidoscope language series.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270799
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Wed, 25 May 2016 23:38:53 +0000 (23:38 +0000)]
[PM] Port PartiallyInlineLibCalls to the new pass manager.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270798
91177308-0d34-0410-b5e6-
96231b3b80d8
David Blaikie [Wed, 25 May 2016 23:37:06 +0000 (23:37 +0000)]
llvm-dwp: Ensure uncompressed sections are not relocated during processing of later inputs
Richard Smith identified this in post commit review of r270466. The
string sections in particular (in the future, possibly all sections - so
I'm not going to bother pulling out just the string sections for the
extra lifetime handling right now) need to remain valid during
processing of all inputs so that elements of the DWPStringPool can be
looked up repeatedly without having to make in-memory copies of string
contents in the noncompressed case (more common in dwp+dwp merge steps
where the memory is a bigger problem because the files are larger).
Using the SmallVector (or any vector) a reallocation on push_back could
cause any of the nested SmallStrings in small mode to move in memory and
invalid pointers to their contents. Using a deque the SmallStrings will
never move around since no elements are removed from the container.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270797
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Wed, 25 May 2016 23:34:19 +0000 (23:34 +0000)]
[Kaleidoscope][BuildingAJIT] Shorten the name of the BuildingAJIT tutorial
series.
The original name was pretty long, and likely to look awkward as more
chapters get added.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270796
91177308-0d34-0410-b5e6-
96231b3b80d8
Reid Kleckner [Wed, 25 May 2016 23:29:08 +0000 (23:29 +0000)]
Revert "[MC] Support symbolic expressions in assembly directives"
This reverts commit r270786, it causes the directive_fill.s to fail.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270795
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Wed, 25 May 2016 23:25:23 +0000 (23:25 +0000)]
[Kaleidoscope][BuildingAJIT] Add code for the 2nd chapter of the BuildingAJIT
tutorial.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270794
91177308-0d34-0410-b5e6-
96231b3b80d8
Reid Kleckner [Wed, 25 May 2016 23:16:12 +0000 (23:16 +0000)]
[codeview] Use comdats for debug info describing comdat functions
Summary:
This allows the linker to discard unused symbol information for comdat
functions that were discarded during the link. Before this change,
searching for the name of an inline function in the debugger would
return multiple results, one per symbol subsection in the object file.
After this change, there is only one result, the result for the function
chosen by the linker.
Reviewers: zturner, majnemer
Subscribers: aaboud, amccarth, llvm-commits
Differential Revision: http://reviews.llvm.org/D20642
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270792
91177308-0d34-0410-b5e6-
96231b3b80d8
Manman Ren [Wed, 25 May 2016 23:14:48 +0000 (23:14 +0000)]
Objective-C Class Properties: Autoupgrade "Class Properties" module flag.
When we have "Image Info Version" module flag but don't have "Class Properties"
module flag, set "Class Properties" module flag to 0, so we can correctly emit
errors when one module has the flag set and another module does not.
rdar://
26469641
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270791
91177308-0d34-0410-b5e6-
96231b3b80d8
Justin Lebar [Wed, 25 May 2016 23:12:38 +0000 (23:12 +0000)]
[NVPTX] Don't (incorrectly) say that the NVVMReflect pass preserves all analyses.
Reviewers: tra
Subscribers: jholewinski, llvm-commits
Differential Revision: http://reviews.llvm.org/D20585
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270790
91177308-0d34-0410-b5e6-
96231b3b80d8
Justin Lebar [Wed, 25 May 2016 23:11:31 +0000 (23:11 +0000)]
[CUDA] Add section to docs about controlling fp optimizations.
Reviewers: rnk
Subscribers: llvm-commits, tra
Differential Revision: http://reviews.llvm.org/D20494
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270789
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Wed, 25 May 2016 22:56:58 +0000 (22:56 +0000)]
Remove unused header.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270788
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Wed, 25 May 2016 22:48:43 +0000 (22:48 +0000)]
[Kaleidoscope][BuildingAJIT] Delete trailing whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270787
91177308-0d34-0410-b5e6-
96231b3b80d8
Petr Hosek [Wed, 25 May 2016 22:47:51 +0000 (22:47 +0000)]
[MC] Support symbolic expressions in assembly directives
This matches the behavior of GNU assembler which supports symbolic
expressions in absolute expressions used in assembly directives.
Differential Revision: http://reviews.llvm.org/D20337
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270786
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Wed, 25 May 2016 22:44:06 +0000 (22:44 +0000)]
Don't repeat name in comment and git-clang-format.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270785
91177308-0d34-0410-b5e6-
96231b3b80d8
Adrian Prantl [Wed, 25 May 2016 22:37:29 +0000 (22:37 +0000)]
Work around an MSVC compiler issue in r270776.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270783
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Wed, 25 May 2016 22:33:25 +0000 (22:33 +0000)]
[Kaleidoscope][BuildingAJIT] Fix code-block indents.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270782
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Wed, 25 May 2016 22:29:34 +0000 (22:29 +0000)]
[LazyValueInfo] Simplify `return after else`. NFCI.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270779
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Wed, 25 May 2016 22:27:25 +0000 (22:27 +0000)]
[Kaleidoscope][BuildingAJIT] Add a description of the KaleidoscopeJIT addModule
method to Chapter1 of the BuildingAJIT tutorial.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270778
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Kuperstein [Wed, 25 May 2016 22:23:08 +0000 (22:23 +0000)]
[BasicAA] Improve precision of alloca vs. inbounds GEP alias queries
If a we have (a) a GEP and (b) a pointer based on an alloca, and the
beginning of the object the GEP points would have a negative offset with
repsect to the alloca, then the GEP can not alias pointer (b).
For example, consider code like:
struct { int f0, int f1, ...} foo;
...
foo alloca;
foo *random = bar(alloca);
int *f0 = &alloca.f0
int *f1 = &random->f1;
Which is lowered, approximately, to:
%alloca = alloca %struct.foo
%random = call %struct.foo* @random(%struct.foo* %alloca)
%f0 = getelementptr inbounds %struct, %struct.foo* %alloca, i32 0, i32 0
%f1 = getelementptr inbounds %struct, %struct.foo* %random, i32 0, i32 1
Assume %f1 and %f0 alias. Then %f1 would point into the object allocated
by %alloca. Since the %f1 GEP is inbounds, that means %random must also
point into the same object. But since %f0 points to the beginning of %alloca,
the highest %f1 can be is (%alloca + 3). This means %random can not be higher
than (%alloca - 1), and so is not inbounds, a contradiction.
Differential Revision: http://reviews.llvm.org/D20495
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270777
91177308-0d34-0410-b5e6-
96231b3b80d8
Adrian Prantl [Wed, 25 May 2016 22:21:12 +0000 (22:21 +0000)]
PR26055: Speed up LiveDebugValues by replacing lists with bitvectors.
This patch modifies the LiveDebugValues pass to use more efficient set
data structures as outlined in PR26055. Both VarLocSet and VarLocList are
now SparseBitVectors which allows us to perform much faster bitvector
arithmetic on them.
The speedup can be in the order of minutes especially on ASANified code.
The change is not NFC in the assembler output because the inserted
DBG_VALUEs are now sorted by variable and location.
Many thanks to Daniel Berlin for helping design the improved algorithm and
reviewing the patch.
https://llvm.org/bugs/show_bug.cgi?id=26055
http://reviews.llvm.org/D20178
rdar://problem/
24091200
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270776
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Wed, 25 May 2016 21:53:46 +0000 (21:53 +0000)]
[MBB] Early exit to reduce indentation, per coding guidelines. NFC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270773
91177308-0d34-0410-b5e6-
96231b3b80d8
Hal Finkel [Wed, 25 May 2016 21:42:37 +0000 (21:42 +0000)]
Look for a loop's starting location in the llvm.loop metadata
Getting accurate locations for loops is important, because those locations are
used by the frontend to generate optimization remarks. Currently, optimization
remarks for loops often appear on the wrong line, often the first line of the
loop body instead of the loop itself. This is confusing because that line might
itself be another loop, or might be somewhere else completely if the body was
inlined function call. This happens because of the way we find the loop's
starting location. First, we look for a preheader, and if we find one, and its
terminator has a debug location, then we use that. Otherwise, we look for a
location on an instruction in the loop header.
The fallback heuristic is not bad, but will almost always find the beginning of
the body, and not the loop statement itself. The preheader location search
often fails because there's often not a preheader, and even when there is a
preheader, depending on how it was formed, it sometimes carries the location of
some preceeding code.
I don't see any good theoretical way to fix this problem. On the other hand,
this seems like a straightforward solution: Put the debug location in the
loop's llvm.loop metadata. A companion Clang patch will cause Clang to insert
llvm.loop metadata with appropriate locations when generating debugging
information. With these changes, our loop remarks have much more accurate
locations.
Differential Revision: http://reviews.llvm.org/D19738
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270771
91177308-0d34-0410-b5e6-
96231b3b80d8