Chad Rosier [Fri, 1 Apr 2016 17:34:38 +0000 (17:34 +0000)]
[AArch64] Fix a typo. NFC.
llvm-svn: 265160
George Rimar [Fri, 1 Apr 2016 17:30:52 +0000 (17:30 +0000)]
[ELF] - Split Writer::assignAddresses(): extract code for initializing dummies sections
Extracts code for initializing dummies sections
to avoid possible duplication in following patches.
Differential review: http://reviews.llvm.org/D18691
llvm-svn: 265159
David Majnemer [Fri, 1 Apr 2016 17:28:17 +0000 (17:28 +0000)]
[InstCombine] Don't sink an instr after a catchswitch
A catchswitch is a terminator, instructions cannot be inserted after it.
llvm-svn: 265158
David Majnemer [Fri, 1 Apr 2016 17:28:15 +0000 (17:28 +0000)]
[SLPVectorizer] Don't insert an extractelement before a catchswitch
A catchswitch cannot be preceded by another instruction in the same
basic block (other than a PHI node).
Instead, insert the extract element right after the materialization of
the vectorized value. This isn't optimal but is a reasonable compromise
given the constraints of WinEH.
This fixes PR27163.
llvm-svn: 265157
Rui Ueyama [Fri, 1 Apr 2016 17:24:19 +0000 (17:24 +0000)]
Make error handling consistent.
Some functions in Writer reports error using HasError, and some reports
their return values. This patch makes them to consistently use HasError.
llvm-svn: 265156
Rui Ueyama [Fri, 1 Apr 2016 17:17:14 +0000 (17:17 +0000)]
Remove dead code.
Thanks to George Rimor for pointing it out.
llvm-svn: 265155
Rui Ueyama [Fri, 1 Apr 2016 17:11:42 +0000 (17:11 +0000)]
Don't do extra work if -r is given.
fixAbsoluteSymbols fixes linker-created symbol addresses. Since we don't
create such symbols for relocatable output, we don't need to call this
function.
llvm-svn: 265154
Reid Kleckner [Fri, 1 Apr 2016 17:09:12 +0000 (17:09 +0000)]
Fix warnings uncovered by building with clang-cl
Move ifdefs to avoid unused static helpers. Move alignment attribute so
that it is respected in GCC and MSVC.
llvm-svn: 265153
Reid Kleckner [Fri, 1 Apr 2016 17:09:08 +0000 (17:09 +0000)]
Fix a load of -Wunknown-pragma warnings from clang-cl
llvm-svn: 265152
Rui Ueyama [Fri, 1 Apr 2016 17:07:17 +0000 (17:07 +0000)]
Tidy up address and file offset assignments in the Writer.
assignAddressesRelocatable function did not set addresses to sections
despite its name. What it actually did is to set file offsets to sections.
assignAddresses function assigned addresses and file offsets to sections.
So there was a confusion what they were doing, and they had duplicate code.
This patch separates file offset assignments from address assignments.
A new function, assignFileOffsets assign file offsets. assignAddresses
do not care about file offsets anymore.
llvm-svn: 265151
Rafael Espindola [Fri, 1 Apr 2016 17:00:36 +0000 (17:00 +0000)]
Bring r264761 back with an extra fix.
The extra fix is to note that it still requires copy relocations.
Original message:
Change how we handle R_MIPS_LO16.
Mips aligns PT_LOAD to 16 bits (0x10000). That means that the lower 16
bits are always the same, so we can, effectively, say that the
relocation is relative.
P.S.: Suggestions for a better name for the predicate are welcome :-)
llvm-svn: 265150
Rong Xu [Fri, 1 Apr 2016 16:43:30 +0000 (16:43 +0000)]
[PGO] Refactor PGOFuncName meta data code to be used in clang
Refactor the code that gets and creates PGOFuncName meta data so that it can be
used in clang's value profile annotation.
Differential Revision: http://reviews.llvm.org/D18623
llvm-svn: 265149
Sanjay Patel [Fri, 1 Apr 2016 16:27:14 +0000 (16:27 +0000)]
[x86] avoid intermediate splat for non-zero memsets (PR27100)
Follow-up to D18566 - where we noticed that an intermediate splat was being
generated for memsets of non-zero chars.
That was because we told getMemsetStores() to use a 32-bit vector element type,
and it happily obliged by producing that constant using an integer multiply.
The tests that were added in the last patch are now equivalent for AVX1 and AVX2
(no splats, just a vector load), but we have PR27141 to track that splat difference.
In the new tests, the splat via shuffling looks ok to me, but there might be some
room for improvement depending on uarch there.
Note that the SSE1/2 paths are not changed in this patch. That can be a follow-up.
This patch should resolve PR27100.
Differential Revision: http://reviews.llvm.org/D18676
llvm-svn: 265148
Rafael Espindola [Fri, 1 Apr 2016 16:21:09 +0000 (16:21 +0000)]
Don't apply a recolation that the dynamic linker will rewrite.
This matches the behavior of both bfd and gold. Looks like we just got
here for mips because of a bad ordering of an if else chain.
llvm-svn: 265147
Anastasia Stulova [Fri, 1 Apr 2016 16:05:09 +0000 (16:05 +0000)]
[OpenCL] Moved nosvm attribute handling in Sema to other OpenCL attrs
llvm-svn: 265146
Benjamin Kramer [Fri, 1 Apr 2016 15:51:51 +0000 (15:51 +0000)]
[ADT] Make StringMap's tombstone aligned.
This avoids undefined behavior when casting pointers to it. Also make
sure that we don't cast to a derived StringMapEntry before checking for
tombstone, as that may have different alignment requirements.
llvm-svn: 265145
Vedant Kumar [Fri, 1 Apr 2016 15:45:33 +0000 (15:45 +0000)]
[PGOProfile] Rename a test to make it more reusable, NFC
llvm-svn: 265144
Rafael Espindola [Fri, 1 Apr 2016 14:36:09 +0000 (14:36 +0000)]
Pass a const pointer to getImplicitAddend. NFC.
llvm-svn: 265143
Rafael Espindola [Fri, 1 Apr 2016 14:14:48 +0000 (14:14 +0000)]
Refactor duplicated code.
We had almost identical code to handle creating a plt entry in two
places.
llvm-svn: 265142
Valery Pykhtin [Fri, 1 Apr 2016 13:13:12 +0000 (13:13 +0000)]
[AMDGPU] fix MADAK/MADMK instructions operand namings to match encoding fields.
$vsrc1 -> $src1, $k -> $imm
Differential Revision: http://reviews.llvm.org/D18659
llvm-svn: 265141
Pavel Labath [Fri, 1 Apr 2016 12:59:37 +0000 (12:59 +0000)]
Fix clean rule for a makefile
The test was failing on windows because the clean rule (which is executed even if the test is
skipped) returned an error there.
llvm-svn: 265140
Rafael Espindola [Fri, 1 Apr 2016 12:54:27 +0000 (12:54 +0000)]
Avoid creating duplicated relocations.
llvm-svn: 265139
Andrea Di Biagio [Fri, 1 Apr 2016 12:25:44 +0000 (12:25 +0000)]
[x86] Remove redundant call to setTargetDAGCombine for BUILD_VECTOR node type.
Since revision 235394, we no longer perform target specific combines on
build_vector nodes. No functional change intended.
llvm-svn: 265138
George Rimar [Fri, 1 Apr 2016 12:24:28 +0000 (12:24 +0000)]
Fix linux buildbot after r265136
http://lab.llvm.org:8011/builders/lld-x86_64-darwin13/builds/20138
Fix: Move method above class that uses it.
Works fine with MSVS.
llvm-svn: 265137
George Rimar [Fri, 1 Apr 2016 11:59:32 +0000 (11:59 +0000)]
Make dummySectionsNum() to be not a member of writer. NFC.
llvm-svn: 265136
Simon Pilgrim [Fri, 1 Apr 2016 11:57:51 +0000 (11:57 +0000)]
[X86][AVX512] Regenerated intrinsics tests
llvm-svn: 265135
Sagar Thakur [Fri, 1 Apr 2016 11:55:33 +0000 (11:55 +0000)]
[MIPS][LLVM-MC] Fix JR encoding for MIPSR6 ISA
Summary: The assembler was picking the wrong JR variant because the pre-R6 one was still enabled at R6.
Author: nitesh.jain
Reviewers: vkalintiris, dsanders
Subscribers: dsanders, llvm-commits, mohit.bhakkad, sagar, bhushan, jaydeep
Differential: D18387
llvm-svn: 265134
Benjamin Kramer [Fri, 1 Apr 2016 11:49:59 +0000 (11:49 +0000)]
[ThinLTO] Fix uninitialized flags.
Found by msan. Patch by Adrian Kuegel!
llvm-svn: 265133
Johannes Doerfert [Fri, 1 Apr 2016 11:25:47 +0000 (11:25 +0000)]
[FIX] Adjust the insert point for non-affine region PHIs
If a non-affine region PHI is generated we should not move the insert
point prior to the synthezised value in the same block as we might
split that block at the insert point later on. Only if the incoming
value should be placed in a different block we should change the
insertion point.
llvm-svn: 265132
George Rimar [Fri, 1 Apr 2016 11:04:47 +0000 (11:04 +0000)]
Split Writer::assignAddresses(): extract common code to fixFileOff(). NFC.
llvm-svn: 265131
George Rimar [Fri, 1 Apr 2016 10:49:14 +0000 (10:49 +0000)]
Split Writer::assignAddresses(): extract assignPhdrs(). NFC.
llvm-svn: 265130
George Rimar [Fri, 1 Apr 2016 10:23:32 +0000 (10:23 +0000)]
[ELF] - Move calculation of _end to fixAbsoluteSymbols()
That is consistent with other symbols: _edata, _etext
and can help to avoid duplicate code.
Differential revision: http://reviews.llvm.org/D18655
llvm-svn: 265129
Andrey Turetskiy [Fri, 1 Apr 2016 10:16:15 +0000 (10:16 +0000)]
[X86] Introduce Lakemont CPU.
Add a new Intel MCU CPU Lakemont, which doesn't support X87.
Differential Revision: http://reviews.llvm.org/D18650
llvm-svn: 265128
Alexey Bataev [Fri, 1 Apr 2016 10:12:06 +0000 (10:12 +0000)]
[OPENMP] Fixed documentation category for 'declare simd' attribute, NFC.
llvm-svn: 265127
Benjamin Kramer [Fri, 1 Apr 2016 10:04:07 +0000 (10:04 +0000)]
[Lexer] Let the compiler infer string lengths. No functionality change intended.
llvm-svn: 265126
Benjamin Kramer [Fri, 1 Apr 2016 09:58:45 +0000 (09:58 +0000)]
[Lexer] Don't read out of bounds if a conflict marker is at the end of a file
This can happen as we look for '<<<<' while scanning tokens but then expect
'<<<<\n' to tell apart perforce from diff3 conflict markers. Just harden
the pointer arithmetic.
Found by libfuzzer + asan!
llvm-svn: 265125
Pavel Labath [Fri, 1 Apr 2016 09:57:30 +0000 (09:57 +0000)]
Change a recently added assert into lldbassert
llvm-svn: 265124
Dmitry Polukhin [Fri, 1 Apr 2016 09:52:30 +0000 (09:52 +0000)]
[OPENMP] Avoid useless recursive calls in getDSA if it is called in a loop, NFC
llvm-svn: 265123
James Molloy [Fri, 1 Apr 2016 09:40:47 +0000 (09:40 +0000)]
Fix for pr24346: arm asm label calculation error in sub
Some ARM instructions encode 32-bit immediates as a 8-bit integer (0-255)
and a 4-bit rotation (0-30, even) in its least significant 12 bits. The
original fixup, FK_Data_4, patches the instruction by the value bit-to-bit,
regardless of the encoding. For example, assuming the label L1 and L2 are
0x0 and 0x104 respectively, the following instruction:
add r0, r0, #(L2 - L1) ; expects 0x104, i.e., 260
would be assembled to the following, which adds 1 to r0, instead of 260:
e2800104 add r0, r0, #4, 2 ; equivalently 1
The new fixup kind fixup_arm_mod_imm takes care of the encoding:
e2800f41 add r0, r0, #260
Patch by Ting-Yuan Huang!
llvm-svn: 265122
Alexey Bataev [Fri, 1 Apr 2016 09:23:34 +0000 (09:23 +0000)]
[OPENMP 4.5] Allow data members as loop counters in loop-based
directives.
OpenMP 4.5 allows privatization of non-static data members in non-static
member functions. Patch allows to use and implicit privatization of data
members used as counters in loop-based directives.
llvm-svn: 265121
Oliver Stannard [Fri, 1 Apr 2016 09:14:50 +0000 (09:14 +0000)]
[AArch64] Better errors for out-of-range fixups
When a fixup that can be resolved by the assembler is out of range, we should
report an error in the source, rather than crashing.
Differential Revision: http://reviews.llvm.org/D18402
llvm-svn: 265120
Alexey Bataev [Fri, 1 Apr 2016 08:43:42 +0000 (08:43 +0000)]
Revert "[OPENMP] Allow skip expression after comma in clauses with lists."
This reverts commit http://reviews.llvm.org/rL265003. After some
thoughts decided to emit errors here.
llvm-svn: 265119
Mehdi Amini [Fri, 1 Apr 2016 08:22:59 +0000 (08:22 +0000)]
ThinLTO: move ObjCARCContractPass in the CodeGen pipeline
This is to be coherent with Full LTO.
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265118
Haojian Wu [Fri, 1 Apr 2016 07:57:30 +0000 (07:57 +0000)]
[clang-tidy] Don't delete unused parameter in class override method in anonymous namespace.
Summary: Fixes PR26740.
Reviewers: alexfh
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D17926
llvm-svn: 265117
Jeroen Ketema [Fri, 1 Apr 2016 07:56:17 +0000 (07:56 +0000)]
[OCaml] Use LLVMCreateMessage with constant strings when calling llvm_raise
The llvm_string_of_message function, called by llvm_raise, calls
LLVMDisposeMessage, which expects the message to be dynamically
allocated; it fails freeing the message otherwise. So always
dynamically allocate with LLVMCreateMessage.
Differential Revision: http://reviews.llvm.org/D18675
llvm-svn: 265116
Jeroen Ketema [Fri, 1 Apr 2016 07:54:24 +0000 (07:54 +0000)]
[OCaml] Reinstate data_layout
Expose LLVMCreateTargetMachineData as data_layout.
As r263530 did for go. From that commit: "LLVMGetTargetDataLayout was
removed from the C API, and then TargetMachine.TargetData was removed.
Later, LLVMCreateTargetMachineData was added to the C API"
Differential Revision: http://reviews.llvm.org/D18677
llvm-svn: 265115
Tobias Grosser [Fri, 1 Apr 2016 07:15:19 +0000 (07:15 +0000)]
ScoPDetection: Obtain a known free diagnostic ID
... instead of hardcoding something that has been free at some point. This fixes
a crash triggered by r265084, where the diagnostic IDs have been shifted in a
way that resulted our hardcode ID to not be assigned any implementation. Our ID
was likely already wrong earlier on, but this time we really crashed nicely.
llvm-svn: 265114
Mehdi Amini [Fri, 1 Apr 2016 06:47:02 +0000 (06:47 +0000)]
Add a libLTO API to stop/restart ThinLTO between optimizations and CodeGen
This allows the linker to instruct ThinLTO to perform only the
optimization part or only the codegen part of the process.
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265113
Chuang-Yu Cheng [Fri, 1 Apr 2016 06:44:32 +0000 (06:44 +0000)]
[PPC64] Bug fix: when enabling sibling-call-opt and shrink-wrapping, the tail call branch instruction might disappear
Bug Pattern:
# BB#0: # %entry
cmpldi 3, 0
beq- 0, .LBB0_2
# BB#1: # %exit
lwz 4, 0(3)
#TC_RETURNd8 LVComputationKind 0
.LBB0_2: # %cond.false
mflr 0
std 0, 16(1)
stdu 1, -96(1)
.Ltmp0:
.cfi_def_cfa_offset 96
.Ltmp1:
.cfi_offset lr, 16
bl __assert_fail
nop
The branch instruction for tail call return is not generated, because the
shrink-wrapping pass choosing a new Restore Point: %cond.false, so %exit
block is not sent to emitEpilogue, that's why the branch is not generated.
Thanks Kit's opinions!
Reviewers: nemanjai hfinkel tjablin kbarton
http://reviews.llvm.org/D17606
llvm-svn: 265112
Mehdi Amini [Fri, 1 Apr 2016 05:33:11 +0000 (05:33 +0000)]
Add a module Hash in the bitcode and the combined index, implementing a kind of "build-id"
This is intended to be used for ThinLTO incremental build.
Differential Revision: http://reviews.llvm.org/D18213
This is a recommit of r265095 after fixing the Windows issues.
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265111
Mehdi Amini [Fri, 1 Apr 2016 05:19:14 +0000 (05:19 +0000)]
Fix MSVC warning "comparison of integers of different signs" (NFC)
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265110
Mehdi Amini [Fri, 1 Apr 2016 05:12:24 +0000 (05:12 +0000)]
Fix S390 big endian detection
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265109
Mehdi Amini [Fri, 1 Apr 2016 05:12:18 +0000 (05:12 +0000)]
Const correctness in raw_sha1_ostream (NFC)
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265108
Mehdi Amini [Fri, 1 Apr 2016 04:30:16 +0000 (04:30 +0000)]
Add support for computing SHA1 in LLVM
Provide a class to generate a SHA1 from a sequence of bytes, and
a convenience raw_ostream adaptor.
This will be used to provide a "build-id" by hashing the Module
block when writing bitcode. ThinLTO will use this information for
incremental build.
Reapply r265094 which was reverted in r265102 because it broke
MSVC bots (constexpr is not supported).
http://reviews.llvm.org/D16325
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265107
Sean Silva [Fri, 1 Apr 2016 03:54:03 +0000 (03:54 +0000)]
Improve CHECK-NOT robustness of dllexport tests
This changes some dllexport tests, to verify that some symbols that
should not be exported are not, in a way that improves the robustness
of CHECK-SAME interaction with CHECK-NOT.
We plan to enable dllimport/dllexport support for the PS4, and these
changes are for points we noticed in our internal testing.
Patch by Warren Ristow!
llvm-svn: 265106
Michael Kuperstein [Fri, 1 Apr 2016 03:45:08 +0000 (03:45 +0000)]
Use range-based for loops. NFC.
llvm-svn: 265105
Mike Spertus [Fri, 1 Apr 2016 03:31:43 +0000 (03:31 +0000)]
Visual Studio native visualizers for clang::TemplateSpecializationType
Displays a template specialization as, say, A<int, double>. Does not
yet handle UncommonTemplateNameStorage, QualifiedTemplateName, or
DependentTemplateName, but still more than worthwhile
llvm-svn: 265104
Devin Coughlin [Fri, 1 Apr 2016 03:24:13 +0000 (03:24 +0000)]
[analyzer] Prefer accessor method in extension over category in CallEvent.
In ObjCMethodCall:getRuntimeDefinition(), if the method is an accessor in a
category, and it doesn't have a self declaration, first try to find the method
in a class extension. This works around a bug in Sema where multiple accessors
are synthesized for properties in class extensions that are redeclared in a
category. The implicit parameters are not filled in for the method on the
category, which causes a crash when trying to synthesize a getter for the
property in BodyFarm. The Sema bug is tracked as rdar://problem/
25481164.
rdar://problem/
25056531
llvm-svn: 265103
Mehdi Amini [Fri, 1 Apr 2016 03:03:21 +0000 (03:03 +0000)]
Revert "Add support for computing SHA1 in LLVM"
This reverts commit r265096, r265095, and r265094.
Windows build is broken, and the validation does not pass.
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265102
Sanjoy Das [Fri, 1 Apr 2016 02:51:30 +0000 (02:51 +0000)]
Don't insert stackrestore on deoptimizing returns
They're not necessary (since the stack pointer is trivially restored on
return), and the way LLVM inserts the stackrestore calls breaks the
IR (we get a stackrestore between the deoptimize call and the return).
llvm-svn: 265101
Sanjoy Das [Fri, 1 Apr 2016 02:51:26 +0000 (02:51 +0000)]
Don't insert lifetime end markers on deoptimizing returns
They're not necessary (since the lifetime of the alloca is trivially
over due to the return), and the way LLVM inserts the lifetime.end
markers breaks the IR (we get a lifetime end marker between the
deoptimize call and the return).
llvm-svn: 265100
Sanjoy Das [Fri, 1 Apr 2016 02:51:21 +0000 (02:51 +0000)]
Don't use an i64 return type with webkit_jscc
Re-enable an assertion enabled by Justin Lebar in rL265092. rL265092
was breaking test/CodeGen/X86/deopt-intrinsic.ll because webkit_jscc
does not like non-i64 return types. Change the test case to not do
that.
llvm-svn: 265099
Matthias Braun [Fri, 1 Apr 2016 02:49:17 +0000 (02:49 +0000)]
AArch64ISelLowering: Remove unused variables/arguments; NFC
llvm-svn: 265098
Chuang-Yu Cheng [Fri, 1 Apr 2016 02:05:29 +0000 (02:05 +0000)]
Fix Sub-register Rewriting in Aggressive Anti-Dependence Breaker
Previously, HandleLastUse would delete RegRef information for sub-registers
if they were dead even if their corresponding super-register were still live.
If the super-register were later renamed, then the definitions of the
sub-register would not be updated appropriately. This patch alters the
behavior so that RegInfo information for sub-registers is only deleted when
the sub-register and super-register are both dead.
This resolves PR26775. This is the mirror image of Hal's r227311 commit.
Author: Tom Jablin (tjablin)
Reviewers: kbarton uweigand nemanjai hfinkel
http://reviews.llvm.org/D18448
llvm-svn: 265097
Mehdi Amini [Fri, 1 Apr 2016 01:37:52 +0000 (01:37 +0000)]
Add missing test for the "Module hash in bitcode" added in r265095
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265096
Mehdi Amini [Fri, 1 Apr 2016 01:30:06 +0000 (01:30 +0000)]
Add a module Hash in the bitcode and the combined index, implementing a kind of "build-id"
This is intended to be used for ThinLTO incremental build.
Differential Revision: http://reviews.llvm.org/D18213
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265095
Mehdi Amini [Fri, 1 Apr 2016 01:29:54 +0000 (01:29 +0000)]
Add support for computing SHA1 in LLVM
Provide a class to generate a SHA1 from a sequence of bytes, and
a convenience raw_ostream adaptor.
This will be used to provide a "build-id" by hashing the Module
block when writing bitcode. ThinLTO will use this information for
incremental build.
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265094
Justin Lebar [Fri, 1 Apr 2016 01:23:23 +0000 (01:23 +0000)]
Revert "Protect some assertions with NDEBUG rather than DEBUG()."
This reverts r265092, because it breaks CodeGen/X86/deopt-intrinsic.ll.
llvm-svn: 265093
Justin Lebar [Fri, 1 Apr 2016 01:09:12 +0000 (01:09 +0000)]
Protect some assertions with NDEBUG rather than DEBUG().
DEBUG() only runs if you pass -debug, but these assertions are generally
useful.
llvm-svn: 265092
Justin Lebar [Fri, 1 Apr 2016 01:09:10 +0000 (01:09 +0000)]
[NVPTX] Add a truncate DAG node to some calls.
Summary:
Previously, we were running afoul of the assertion
EVT(CLI.Ins[i].VT) == InVals[i].getValueType() && "LowerCall emitted a value with the wrong type!"
in SelectionDAGBuilder.cpp when running the NVPTX/i8-param.ll test.
This is because our backend (for some reason) treats small return values
as i32, but it wasn't ever truncating the i32 back down to the expected
width in the DAG.
Unclear to me whether this fixes any actual bugs -- in this test, at
least, the generated code is unchanged.
Reviewers: jingyue
Subscribers: llvm-commits, tra, jholewinski
Differential Revision: http://reviews.llvm.org/D17872
llvm-svn: 265091
Justin Lebar [Fri, 1 Apr 2016 01:09:07 +0000 (01:09 +0000)]
[NVPTX] Read __CUDA_FTZ from module flags in NVVMReflect.
Summary:
Previously the NVVMReflect pass would read its configuration from
command-line flags or a static configuration given to the pass at
instantiation time.
This doesn't quite work for clang's use-case. It needs to pass a value
for __CUDA_FTZ down on a per-module basis. We use a module flag for
this, so the NVVMReflect pass needs to be updated to read said flag.
Reviewers: tra, rnk
Subscribers: cfe-commits, jholewinski
Differential Revision: http://reviews.llvm.org/D18672
llvm-svn: 265090
Justin Lebar [Fri, 1 Apr 2016 01:09:05 +0000 (01:09 +0000)]
[NVPTX] Annotate some instructions as hasSideEffects = 0.
Summary:
Tablegen tries to infer this from the selection DAG patterns defined for
the instructions, but it can't always.
An instructive example is CLZr64. CLZr32 is correctly inferred to have
no side-effects, but the selection DAG pattern for CLZr64 is slightly
more complicated, and in particular the ctlz DAG node is not at the root
of the pattern. Thus tablegen can't infer that CLZr64 has no
side-effects.
Reviewers: jholewinski
Subscribers: jholewinski, tra, llvm-commits
Differential Revision: http://reviews.llvm.org/D17472
llvm-svn: 265089
Justin Lebar [Fri, 1 Apr 2016 01:09:03 +0000 (01:09 +0000)]
[ifcnv] Add brief comment explaining what ifcnv is.
llvm-svn: 265088
Mehdi Amini [Fri, 1 Apr 2016 00:52:05 +0000 (00:52 +0000)]
Fix Windows build (typo in disk_space() implementation)
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265087
Greg Clayton [Fri, 1 Apr 2016 00:41:29 +0000 (00:41 +0000)]
Fixed an issue that could cause debugserver to return two stop reply packets ($T packets) for one \x03 interrupt. The problem was that when a \x03 byte is sent to debugserver while the process is running, and up calling:
rnb_err_t
RNBRemote::HandlePacket_stop_process (const char *p)
{
if (!DNBProcessInterrupt(m_ctx.ProcessID()))
HandlePacket_last_signal (NULL);
return rnb_success;
}
In the call to DNBProcessInterrupt we did:
nub_bool_t
DNBProcessInterrupt(nub_process_t pid)
{
MachProcessSP procSP;
if (GetProcessSP (pid, procSP))
return procSP->Interrupt();
return false;
}
This would always return false. It would cause HandlePacket_stop_process to always call "HandlePacket_last_signal (NULL);" which would send an extra stop reply packet _if_ the process is stopped. On a machine with enough cores, it would call DNBProcessInterrupt(...) and then HandlePacket_last_signal(NULL) so quickly that it will never send out an extra stop reply packet. But if the machine is slow enough or doesn't have enough cores, it could cause the call to HandlePacket_last_signal() to actually succeed and send an extra stop reply packet. This would cause problems up in GDBRemoteCommunicationClient::SendContinuePacketAndWaitForResponse() where it would get the first stop reply packet and then possibly return or execute an async packet. If it returned, then the next packet that was sent will get the second stop reply as its response. If it executes an async packet, the async packet will get the wrong response.
To fix this I did the following:
1 - in debugserver, I fixed "bool MachProcess::Interrupt()" to return true if it sends the signal so we avoid sending the stop reply twice on slower machines
2 - Added a log line to RNBRemote::HandlePacket_stop_process() to say if we ever send an extra stop reply so we will see this in the darwin console output if this does happen
3 - Added response validators to StringExtractorGDBRemote so that we can verify some responses to some packets.
4 - Added validators to packets that often follow stop reply packets like the "m" packet for memory reads, JSON packets since "jThreadsInfo" is often sent immediately following a stop reply.
5 - Modified GDBRemoteCommunicationClient::SendPacketAndWaitForResponseNoLock() to validate responses. Any "StringExtractorGDBRemote &response" that contains a valid response verifier will verify the response and keep looking for correct responses up to 3 times. This will help us get back on track if we do get extra stop replies. If a StringExtractorGDBRemote does not have a response validator, it will accept any packet in response.
6 - In GDBRemoteCommunicationClient::SendPacketAndWaitForResponse we copy the response validator from the "response" argument over into m_async_response so that if we send the packet by interrupting the running process, we can validate the response we actually get in GDBRemoteCommunicationClient::SendContinuePacketAndWaitForResponse()
7 - Modified GDBRemoteCommunicationClient::SendContinuePacketAndWaitForResponse() to always check for an extra stop reply packet for 100ms when the process is interrupted. We were already doing this because we might interrupt a process with a \x03 packet, yet the process was in the process of stopping due to another reason. This race condition could cause an extra stop reply packet because the GDB remote protocol says if a \x03 packet is sent while the process is stopped, we should send a stop reply packet back. Now we always check for an extra stop reply packet when we manually interrupt a process.
The issue was showing up when our IDE would attempt to set a breakpoint while the process is running and this would happen:
--> \x03
<-- $T<stop reply 1>
--> z0,AAAAA,BB (set breakpoint)
<-- $T<stop reply 1> (incorrect extra stop reply packet)
--> c
<-- OK (response from z0 packet)
Now all packet traffic was off by one response. Since we now have a validator on the response for "z" packets, we do this:
--> \x03
<-- $T<stop reply 1>
--> z0,AAAAA,BB (set breakpoint)
<-- $T<stop reply 1> (Ignore this because this can't be the response to z0 packets)
<-- OK -- (we are back on track as this is a valid response to z0)
...
As time goes on we should add more packet validators.
<rdar://problem/
22859505>
llvm-svn: 265086
Davide Italiano [Fri, 1 Apr 2016 00:35:29 +0000 (00:35 +0000)]
[LTO] Inherit options from Codegen before initializing TargetMachine.
This fixes bootstrap of llvm-tblgen (with LTO) and PR27150.
Slightly longer explanation follows.
Emission of .init_array instead of .ctors is supported only on a
subset of the Target LLVM supports. Codegen needs to be conservative
and always emit .ctors unless instructed otherwise (based on target).
If the dynamic linker sees .init_array it completely ignores
what's inside .ctors and therefore some constructors are not called
(and this causes llvm-tblgen to crash on startup).
Teach LLD/LTO about the Codegen options so we end up always emitting
.init_array and avoid this issue.
In future, we might end up supporting mix of .ctors and .init_array
in different input files if this shows up as a real-world use case.
The way gold handles this case is mapping .ctors from input into
.init_array in output. There's also another caveat because
as far as I understand .ctors run in reverse order so when we do
the copy/mapping we need to reverse copy in the output if there's
more than one ctor. That's why I'd rather avoid this complicate logic
unless there's a real need.
An analogous reasoning holds for .dtors/.fini_array.
llvm-svn: 265085
Akira Hatanaka [Fri, 1 Apr 2016 00:34:39 +0000 (00:34 +0000)]
[LoopVectorize] Don't unconditionally print vectorization diagnostics
when compiling with LTO.
r244523 a new class DiagnosticInfoOptimizationRemarkAnalysisAliasing for
optimization analysis remarks related to pointer aliasing without
guarding it in isDiagnosticEnabled in LLVMContext.cpp. This caused the
diagnostic message to be printed unconditionally when compiling with
LTO.
This commit cleans up isDiagnosticEnabled and makes sure all the
vectorization optimization remarks are guarded.
rdar://problem/
25382153
llvm-svn: 265084
Justin Lebar [Fri, 1 Apr 2016 00:25:42 +0000 (00:25 +0000)]
[CUDA] Fix typo in __clang_cuda_runtime_wrapper.h.
We're #including the wrong file!
llvm-svn: 265083
Mehdi Amini [Fri, 1 Apr 2016 00:18:08 +0000 (00:18 +0000)]
Add disk_space() to llvm::fs
Summary: Adapted from Boost::filesystem.
(This is a reapply by reverting commit r265080 and fixing the WinAPI part)
Differential Revision: http://reviews.llvm.org/D18467
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265082
Adrian Prantl [Fri, 1 Apr 2016 00:16:49 +0000 (00:16 +0000)]
testcase gardening: update the emissionKind enum to the new syntax. (NFC)
llvm-svn: 265081
Mehdi Amini [Fri, 1 Apr 2016 00:13:31 +0000 (00:13 +0000)]
Revert "Add disk_space() to llvm::fs"
This reverts commit r265074 and r265068.
Breaks windows build
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265080
Adrian Prantl [Fri, 1 Apr 2016 00:00:27 +0000 (00:00 +0000)]
Fix a captialization error in r265077.
llvm-svn: 265079
Adrian Prantl [Thu, 31 Mar 2016 23:57:45 +0000 (23:57 +0000)]
Adapt to LLVM API change in r265077.
EmissionKind moved from DIBuilder to DICompileUnit.
<rdar://problem/
25427165>
llvm-svn: 265078
Adrian Prantl [Thu, 31 Mar 2016 23:56:58 +0000 (23:56 +0000)]
Move the DebugEmissionKind enum from DIBuilder into DICompileUnit.
This mostly cosmetic patch moves the DebugEmissionKind enum from DIBuilder
into DICompileUnit. DIBuilder is not the right place for this enum to live
in — a metadata consumer should not have to include DIBuilder.h.
I also added a Verifier check that checks that the emission kind of a
DICompileUnit is actually legal.
http://reviews.llvm.org/D18612
<rdar://problem/
25427165>
llvm-svn: 265077
Hans Wennborg [Thu, 31 Mar 2016 23:55:16 +0000 (23:55 +0000)]
Follow-up to r265036: I got these iterators mixed up
llvm-svn: 265076
Mehdi Amini [Thu, 31 Mar 2016 23:27:37 +0000 (23:27 +0000)]
clang-format the large enum definitions in LLVMBitCodes.h
Suggestion from Duncan in a review thread.
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265075
Mehdi Amini [Thu, 31 Mar 2016 23:14:45 +0000 (23:14 +0000)]
Use const ref instead of value for Twine in the disk_space() API
Thanks Rui for noticing!
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265074
Peter Collingbourne [Thu, 31 Mar 2016 23:12:18 +0000 (23:12 +0000)]
ELF: Correctly handle --whole-archive for thin archives.
Differential Revision: http://reviews.llvm.org/D18669
llvm-svn: 265073
Eugene Zelenko [Thu, 31 Mar 2016 23:09:42 +0000 (23:09 +0000)]
Update release notes with list of checks added since 3.8.
Fix some checks documentation style.
Differential revision: http://reviews.llvm.org/D18582
llvm-svn: 265072
Matthias Braun [Thu, 31 Mar 2016 23:08:55 +0000 (23:08 +0000)]
lit: python3 compatibility fix
llvm-svn: 265070
Peter Collingbourne [Thu, 31 Mar 2016 23:07:50 +0000 (23:07 +0000)]
Create thin archive in GNU format to fix test on OS X.
llvm-svn: 265069
Mehdi Amini [Thu, 31 Mar 2016 23:05:26 +0000 (23:05 +0000)]
Add disk_space() to llvm::fs
Summary: Adapted from Boost::filesystem.
(This is a reapply by reverting commit r265062 and fixing the WinAPI part)
Differential Revision: http://reviews.llvm.org/D18467
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265068
Tim Shen [Thu, 31 Mar 2016 22:32:10 +0000 (22:32 +0000)]
Move asm-printer-topological-order.ll to PowerPC backend
llvm-svn: 265067
Peter Collingbourne [Thu, 31 Mar 2016 22:08:57 +0000 (22:08 +0000)]
Object: Add function for moving thin archive object buffer vector out of Archive.
Differential Revision: http://reviews.llvm.org/D18664
llvm-svn: 265066
Peter Collingbourne [Thu, 31 Mar 2016 22:08:31 +0000 (22:08 +0000)]
Object: Correctly read thin archives containing absolute paths.
Differential Revision: http://reviews.llvm.org/D18666
llvm-svn: 265065
Tim Shen [Thu, 31 Mar 2016 22:08:19 +0000 (22:08 +0000)]
[AsmPrinter] Print aliases in topological order
Print aliases in topological order, that is, for any alias a = b,
b must be printed before a. This is because on some targets (e.g. PowerPC)
linker expects aliases in such an order to generate correct TOC information.
GCC also prints aliases in topological order.
llvm-svn: 265064
Chandler Carruth [Thu, 31 Mar 2016 21:55:58 +0000 (21:55 +0000)]
Fix PR26940 where compiles times regressed massively.
Patch by Jonas Paulsson. Original description:
Bugfix in buildSchedGraph() to make -dag-maps-huge-region work properly
I found that the reduction of the maps did in fact never happen in this
test case. This was because *all* the stores / loads were made with
addresses from arguments and they thus became "unknown" stores / loads.
Fixed by removing continue statements and making sure that the test for
reduction always takes place.
Differential Revision: http://reviews.llvm.org/D18673
llvm-svn: 265063
Mehdi Amini [Thu, 31 Mar 2016 21:55:35 +0000 (21:55 +0000)]
Revert "Add disk_space() to llvm::fs"
Breaks windows bot.
This reverts commit r265050.
This reverts commit r265055.
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 265062
Evgeniy Stepanov [Thu, 31 Mar 2016 21:55:11 +0000 (21:55 +0000)]
Preserve blockaddress use edges in the module splitter.
"blockaddress" can not apply to an external function. All
blockaddress constant uses must belong to the same module as the
definition of the target function.
llvm-svn: 265061
David Majnemer [Thu, 31 Mar 2016 21:29:57 +0000 (21:29 +0000)]
[NVPTX] Infer __nvvm_reflect as nounwind, readnone
This patch simply mirrors the attributes we give to @llvm.nvvm.reflect
to the __nvvm_reflect libdevice call. This shaves about 30% of the code
in libdevice away because of CSE opportunities. It's also helps us
figure out that libdevice implementations of transcendental functions
don't have side-effects.
llvm-svn: 265060