Adrian Prantl [Fri, 9 Mar 2018 00:45:04 +0000 (00:45 +0000)]
LowerDbgDeclare: ignore dbg.declares for allocas with volatile access
There is no point in lowering a dbg.declare describing an alloca that
has volatile loads or stores as users, since the alloca cannot be
elided. Lowering the dbg.declare will result in larger debug info that
may also have worse coverage than just describing the alloca.
rdar://problem/
34496278
llvm-svn: 327092
Kuba Mracek [Fri, 9 Mar 2018 00:41:29 +0000 (00:41 +0000)]
Mark the -overlap tests are "UNSUPPORTED: android". Android test harness doesn't know how to handle suppression files.
llvm-svn: 327091
Douglas Yung [Fri, 9 Mar 2018 00:38:51 +0000 (00:38 +0000)]
[DOXYGEN] Fix doxygen and content issues in mmintrin.h
- Fix instruction mappings/listings for various intrinsics
This patch was made by Craig Flores
Differential Revision: https://reviews.llvm.org/D41517
llvm-svn: 327090
Vedant Kumar [Fri, 9 Mar 2018 00:34:43 +0000 (00:34 +0000)]
[test] Skip a test which sporadically fails in its dsym variant
There is a mailing list discussion re: r325927 about why this test fails
in the dsym variant. I've marked it skipped for now, until the issue is
resolved.
llvm-svn: 327089
Jason Molenda [Fri, 9 Mar 2018 00:32:56 +0000 (00:32 +0000)]
Remove unneeded per-arch sdk specifications from debugserver xcode project file.
llvm-svn: 327088
Jason Molenda [Fri, 9 Mar 2018 00:28:51 +0000 (00:28 +0000)]
Remove the explicit dependency on libpmenergy and libpmsample.
llvm-svn: 327087
Eric Christopher [Fri, 9 Mar 2018 00:23:35 +0000 (00:23 +0000)]
Fix header comment on SHA1 code.
llvm-svn: 327086
Jason Molenda [Fri, 9 Mar 2018 00:23:29 +0000 (00:23 +0000)]
Fixed two more sdk inconsistencies with the pmenergy stuff.
llvm-svn: 327085
Jason Molenda [Fri, 9 Mar 2018 00:17:22 +0000 (00:17 +0000)]
Three little cleanups in the debugserver xcode project file.
1. Link against libpmenergy and pmsample unconditionally. It is available on
macOS 10.10 ("Yosemite") and newer. We're already linking against libcompression
unconditionally which is only available on macOS 10.11 & newer.
2. Change a few "sdk=macosx.internal"'s to sdk=macosx.
3. Clean up a few places where libcompression was being enabled inconsistently.
Note: the -DLLDB_ENERGY define is only set when building against the macosx.internal
SDK; it includes a header file that is not public. We link against the dylibs
unconditionally for simplicity.
llvm-svn: 327084
Paul Robinson [Fri, 9 Mar 2018 00:11:54 +0000 (00:11 +0000)]
Revert "[DWARF] Fix mixing assembler -g with DWARF .file directives."
This reverts commit
d6d9ac1ab5039ba1fe0f63c36eac2bdd9f0a79c9.
aka r327073
llvm-svn: 327083
Saleem Abdulrasool [Fri, 9 Mar 2018 00:06:10 +0000 (00:06 +0000)]
utils: add a helper class to lit for captured substitutions
On Windows, if the substitution contains a back reference, it would
removed due to the replacement of the escape character in lit. Create a
helper class to avoid this which will simply ignore the replacement and
mark the substitution as having capture groups being referenced.
llvm-svn: 327082
Kuba Mracek [Fri, 9 Mar 2018 00:03:09 +0000 (00:03 +0000)]
Mark strcat-overlap.cc as "UNSUPPORTED: win32"
llvm-svn: 327081
Kuba Mracek [Thu, 8 Mar 2018 23:53:29 +0000 (23:53 +0000)]
Move lto-constmerge-odr.cc to Posix (it's failing on Windows).
llvm-svn: 327080
Sameer AbuAsal [Thu, 8 Mar 2018 23:41:40 +0000 (23:41 +0000)]
Propagate flags to SDValue in SplitVecOp_VECREDUCE
This patch is a fix for PR36642.
While legalizing long vector types, make sure the smaller types get the
flags of the wider type.
link: https://bugs.llvm.org/show_bug.cgi?id=36642
Change-Id: I0c2829639f094c862c10a6b51b342d4c2563e1fa
llvm-svn: 327079
Adrian Prantl [Thu, 8 Mar 2018 23:11:46 +0000 (23:11 +0000)]
Add a debug info testcase for the trvial_abi attribute.
llvm-svn: 327078
Kevin Enderby [Thu, 8 Mar 2018 23:10:38 +0000 (23:10 +0000)]
For llvm-objdump and Mach-O files, update the printing of some thread states
from core files. I tested this against the couple of core files that were
getting errors about unknown thread flavors and it now produce the same output as
the Xcode otool-classic(1) tool. Since the core files are huge I didn’t include
them as test cases.
rdar://
38216356
llvm-svn: 327077
Jan Vesely [Thu, 8 Mar 2018 23:01:01 +0000 (23:01 +0000)]
amdgcn/fmax: fcanonicalize operands
v_max instruction needs canonicalized operands.
Passes CTS on carrizo
Reviewer: Aaron Watry <awatry@gmail.com>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 327076
Jan Vesely [Thu, 8 Mar 2018 23:00:58 +0000 (23:00 +0000)]
amdgcn/fmin: fcanonicalize operands
v_min instruction needs canonicalized operands.
Passes CTS on carrizo
Reviewer: Aaron Watry <awatry@gmail.com>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 327075
Eugene Zelenko [Thu, 8 Mar 2018 22:45:13 +0000 (22:45 +0000)]
[StaticAnalyzer] Fix some Clang-tidy modernize and Include What You Use warnings; other minor fixes (NFC).
llvm-svn: 327074
Paul Robinson [Thu, 8 Mar 2018 22:39:47 +0000 (22:39 +0000)]
[DWARF] Fix mixing assembler -g with DWARF .file directives.
We were effectively overriding an explicit '.file' directive with info
for the assembler source. That shouldn't happen.
Fixes PR36636.
Differential Revision: https://reviews.llvm.org/D44265
llvm-svn: 327073
Matt Davis [Thu, 8 Mar 2018 22:22:26 +0000 (22:22 +0000)]
[DebugInfo] Add DW_AT_byte_size to vectors
Summary:
This patch adds the DW_AT_byte_size dwarf attribute to vectors.
This fixes PR21924
LLVM will round a vector up to the next alignable address, which can result in
the vector's representation in the object file being larger than what the
debugger will calculate via NumberOfElements * ElementSize. In such a case calling sizeof(MyVec) in the source will result in a different value than what a debugger might present. This situation can occur because LLVM permits non-power of two 'vector_size' attributes.
Reviewers: echristo, dexonsmith, aprantl
Reviewed By: aprantl
Subscribers: probinson, aprantl, llvm-commits, JDevlieghere
Tags: #debug-info
Differential Revision: https://reviews.llvm.org/D44048
llvm-svn: 327072
Sanjay Patel [Thu, 8 Mar 2018 22:05:27 +0000 (22:05 +0000)]
[Reassociate] fix test to be independent of FP undef
llvm-svn: 327071
Benjamin Kramer [Thu, 8 Mar 2018 21:54:30 +0000 (21:54 +0000)]
[Support] Pacify -Wsign-compare in unit test.
llvm-svn: 327070
Craig Topper [Thu, 8 Mar 2018 21:53:36 +0000 (21:53 +0000)]
[TargetLowering] Remove redundant if condition in SimplifySetcc. NFC
We were checking the condition code a second time when we were already in a block with this same condition code check.
llvm-svn: 327069
Dan Liew [Thu, 8 Mar 2018 21:50:22 +0000 (21:50 +0000)]
[asan] Fix bug where suppression of overlapping accesses was ignored on
`strcpy()`, `strncpy()`, `strcat()`, and `strncat()`.
rdar://problem/
35576899
Differential Revision: https://reviews.llvm.org/D43702
llvm-svn: 327068
Benjamin Kramer [Thu, 8 Mar 2018 21:31:10 +0000 (21:31 +0000)]
[DebugInfo] Move RangeListEntries instead of copying.
This is needed for correctness as RangeListEntry is not copy-assignable,
which std::vector might rely on.
llvm-svn: 327067
Sanjay Patel [Thu, 8 Mar 2018 21:30:56 +0000 (21:30 +0000)]
[AMDGPU] fix test to survive more FP undef constant folding
llvm-svn: 327066
Philip Reames [Thu, 8 Mar 2018 21:25:30 +0000 (21:25 +0000)]
[NFC] Factor out a helper function for checking if a block has a potential early implicit exit.
llvm-svn: 327065
Vedant Kumar [Thu, 8 Mar 2018 21:15:26 +0000 (21:15 +0000)]
Low-hanging fruit optimization in string::__move_assign().
shrink_to_fit() ends up doing a lot work to get information that we
already know since we just called clear(). This change seems concise
enough to be worth the couple extra lines and my benchmarks show that it
is indeed a pretty decent win. It looks like the same thing is going on
twice in __copy_assign_alloc(), but I didn't want to go overboard since
this is my first contribution to llvm/libc++.
Patch by Timothy VanSlyke!
Differential Revision: https://reviews.llvm.org/D41976
llvm-svn: 327064
Zachary Turner [Thu, 8 Mar 2018 21:07:30 +0000 (21:07 +0000)]
Fix compilation failure with MSVC.
llvm-svn: 327063
Kuba Mracek [Thu, 8 Mar 2018 21:02:52 +0000 (21:02 +0000)]
[asan] Fix a false positive ODR violation due to LTO ConstantMerge pass [compiler-rt part, take 3]
This fixes a false positive ODR violation that is reported by ASan when using LTO. In cases, where two constant globals have the same value, LTO will merge them, which breaks ASan's ODR detection. See the included testcase for an example.
Differential Revision: https://reviews.llvm.org/D43959
llvm-svn: 327062
Kuba Mracek [Thu, 8 Mar 2018 21:02:18 +0000 (21:02 +0000)]
[asan] Fix a false positive ODR violation due to LTO ConstantMerge pass [llvm part, take 3]
This fixes a false positive ODR violation that is reported by ASan when using LTO. In cases, where two constant globals have the same value, LTO will merge them, which breaks ASan's ODR detection.
Differential Revision: https://reviews.llvm.org/D43959
llvm-svn: 327061
Zachary Turner [Thu, 8 Mar 2018 20:57:37 +0000 (20:57 +0000)]
Fix signed-unsigned comparison warning.
llvm-svn: 327060
Wolfgang Pieb [Thu, 8 Mar 2018 20:52:35 +0000 (20:52 +0000)]
[DWARF v5] Support for verbose dumping of .debug_rnglist entries
Adding verbose dumping to the recent implementation of dumping of v5 range list entries.
We're capturing the entries as is as they come in during extraction, including their file offset,
so we can dump them in more detail.
The offset table entries which are table-relative are shown as is (as in non-verbose mode)
and with the actual file offset they map to.
Reviewers: dblaikie, aprantl, jdevlieghere, jhenderson
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D43366
llvm-svn: 327059
Sanjay Patel [Thu, 8 Mar 2018 20:42:49 +0000 (20:42 +0000)]
[ConstantFold] fp_binop undef, undef --> undef
These are uncontroversial and independent of a proposed LangRef edits (D44216).
I tried to fix tests that would fold away:
rL327004
rL327028
rL327030
rL327034
I'm not sure if the Reassociate tests are meaningless yet, but they probably will be
as we add more folds, so if anyone has suggestions or wants to fix those, please do.
Differential Revision: https://reviews.llvm.org/D44258
llvm-svn: 327058
Zachary Turner [Thu, 8 Mar 2018 20:34:47 +0000 (20:34 +0000)]
[Support] Add WriteThroughMemoryBuffer.
This is like MemoryBuffer (read-only) and WritableMemoryBuffer
(writable private), but where the underlying file can be modified
after writing. This is useful when you want to open a file, make
some targeted edits, and then write it back out.
Differential Revision: https://reviews.llvm.org/D44230
llvm-svn: 327057
Andrea Di Biagio [Thu, 8 Mar 2018 20:21:55 +0000 (20:21 +0000)]
[llvm-mca] Fix handling of zero-latency instructions.
This patch fixes a problem found when testing zero latency instructions on
target AArch64 -mcpu=exynos-m3 / -mcpu=exynos-m1.
On Exynos-m3/m1, direct branches are zero-latency instructions that don't consume
any processor resources. The DispatchUnit marks zero-latency instructions as
"executed", so that no scheduling is required. The event of instruction
executed is then notified to all the listeners, and the reorder buffer (managed
by the RetireControlUnit) is updated. In particular, the entry associated to the
zero-latency instruction in the reorder buffer is marked as executed.
Before this patch, the DispatchUnit forgot to assign a retire control unit token
(RCUToken) to the zero-latency instruction. As a consequence, the RCUToken was
used uninitialized. This was causing a crash in the RetireControlUnit logic.
Fixes PR36650.
llvm-svn: 327056
Kuba Mracek [Thu, 8 Mar 2018 20:13:39 +0000 (20:13 +0000)]
Revert r327053.
llvm-svn: 327055
Simon Pilgrim [Thu, 8 Mar 2018 20:07:06 +0000 (20:07 +0000)]
[X86][AVX] Pull out variable permute creation from LowerBUILD_VECTORAsVariablePermute. NFCI.
This will make it easier to handle more complex cases than basic scaling or index masks.
llvm-svn: 327054
Kuba Mracek [Thu, 8 Mar 2018 20:05:45 +0000 (20:05 +0000)]
[asan] Fix a false positive ODR violation due to LTO ConstantMerge pass [llvm part, take 2]
This fixes a false positive ODR violation that is reported by ASan when using LTO. In cases, where two constant globals have the same value, LTO will merge them, which breaks ASan's ODR detection.
Differential Revision: https://reviews.llvm.org/D43959
llvm-svn: 327053
Vedant Kumar [Thu, 8 Mar 2018 19:46:39 +0000 (19:46 +0000)]
[test] Skip a test when using an out-of-tree debugserver
The test "test_fp_special_purpose_register_read" in TestRegisters.py
fails on Darwin machines configured to use an out-of-tree debugserver.
The error message is: 'register read ftag' returns expected result, got
'ftag = 0x80'. This indicates that the debugserver in use is too old.
This commit introduces a decorator which can be used to skip tests which
rely on having a just-built debugserver. This resolves the issue:
$ ./bin/llvm-dotest -p TestRegisters.py -v
1 out of 617 test suites processed - TestRegisters.py
Test Methods: 7
Success: 6
Skip: 1
...
llvm-svn: 327052
Vlad Tsyrklevich [Thu, 8 Mar 2018 19:46:19 +0000 (19:46 +0000)]
Specify that test from r327041 requires asserts
llvm-svn: 327051
Zachary Turner [Thu, 8 Mar 2018 19:45:20 +0000 (19:45 +0000)]
Fix detection of COFF executable files.
One overload of this function would try to identify a file
by opening it and using the first 32 bytes to identify the magic
of the file. This didn't work properly when more than 32 bytes
is actually needed for magic detection to succeed. So now we
have this overload read in the entire file.
Differential Revision: https://reviews.llvm.org/D44225
llvm-svn: 327050
Zachary Turner [Thu, 8 Mar 2018 19:33:47 +0000 (19:33 +0000)]
Resubmit "Write a hash of the executable into the PE timestamp fields."
This fixes the broken tests that were causing failures. The tests
before were verifying that the time stamp was 0, but now that we
are actually writing a timestamp, I just removed the match against
the timestamp value.
llvm-svn: 327049
Matt Davis [Thu, 8 Mar 2018 19:31:37 +0000 (19:31 +0000)]
[DebugInfo] Add verifier for DICompositeType vector
Summary:
This patch adds verification logic for DICompositeType vectors, ensuring that they only have one element, and that element is of type subrange.
This patch complements https://reviews.llvm.org/D44048
Reviewers: aprantl
Reviewed By: aprantl
Subscribers: JDevlieghere, llvm-commits
Tags: #debug-info
Differential Revision: https://reviews.llvm.org/D44262
llvm-svn: 327048
Vlad Tsyrklevich [Thu, 8 Mar 2018 19:20:08 +0000 (19:20 +0000)]
Fix test failure introduced in r327041
The "Assertion: `...' failed" error message format is not identical
across platforms.
llvm-svn: 327047
Alina Sbirlea [Thu, 8 Mar 2018 19:15:00 +0000 (19:15 +0000)]
[MemorySSA] Split PtrIntPair as this fails on win/arm.
Summary: Split PtrIntPair into Instruction and OptionalAlias<Result>. The latter needs 3 bits, which appear unavailable on certain archs.
Subscribers: sanjoy, jlebar, Prazek, llvm-commits, kristof.beyls
Differential Revision: https://reviews.llvm.org/D44268
llvm-svn: 327046
Jan Vesely [Thu, 8 Mar 2018 18:58:07 +0000 (18:58 +0000)]
amdgcn,popcount: Workaround broken llvm.ctpop intrinsic on some GCN ASICs
This is only really needed for VI+ ASICs. However, llvm would cast the value to
i32 for older asics anyway. The proper fix is in LLVM-7 (r326535).
Fixes CTS popcount on carrizo.
Reviewer: Aaron Watry <awatry@gmail.com>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 327044
Jan Vesely [Thu, 8 Mar 2018 18:58:05 +0000 (18:58 +0000)]
integer/gentype: Add __CLC_VECSIZE macro
Reviewer: Aaron Watry <awatry@gmail.com>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 327043
Jan Vesely [Thu, 8 Mar 2018 18:58:00 +0000 (18:58 +0000)]
popcount: Provide function implementation rather than intrinsic redirect
amdgcn will need to override this
Reviewer: Aaron Watry <awatry@gmail.com>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 327042
Vlad Tsyrklevich [Thu, 8 Mar 2018 18:48:03 +0000 (18:48 +0000)]
[ThinLTO] Keep available_externally symbols live
Summary:
This change fixes PR36483. The bug was originally introduced by a change
that marked non-prevailing symbols dead. This broke LowerTypeTests
handling of available_externally functions, which are non-prevailing.
LowerTypeTests uses liveness information to avoid emitting thunks for
unused functions.
Marking available_externally functions dead is incorrect, the functions
are used though the function definitions are not. This change keeps them
live, and lets the EliminateAvailableExternally/GlobalDCE passes remove
them later instead.
I've also enabled EliminateAvailableExternally for all optimization
levels, I believe it being disabled for O1 was an oversight.
Reviewers: pcc, tejohnson
Reviewed By: tejohnson
Subscribers: grimar, mehdi_amini, inglorion, eraman, llvm-commits
Differential Revision: https://reviews.llvm.org/D43690
llvm-svn: 327041
Gheorghe-Teodor Bercea [Thu, 8 Mar 2018 18:44:02 +0000 (18:44 +0000)]
[OpenMP][libomptarget] Fix union.
Summary: To make the two parts of the union have the same size, the size of vect needs to be increased by 16 bits.
Reviewers: grokos, carlo.bertolli, caomhin, ABataev
Reviewed By: grokos, ABataev
Subscribers: fedor.sergeev, guansong, openmp-commits
Differential Revision: https://reviews.llvm.org/D44254
llvm-svn: 327040
Sanjay Patel [Thu, 8 Mar 2018 18:34:23 +0000 (18:34 +0000)]
[InstCombine] add min/max tests with not ops; NFC
These are based on:
https://bugs.llvm.org/show_bug.cgi?id=35875
It's not clear if/how instcombine can reduce these,
but we should have the tests here either way to
document current behavior.
llvm-svn: 327039
Kuba Mracek [Thu, 8 Mar 2018 18:20:03 +0000 (18:20 +0000)]
More revert of r327031
llvm-svn: 327038
Krzysztof Parzyszek [Thu, 8 Mar 2018 18:15:13 +0000 (18:15 +0000)]
[Hexagon] Ignore indexed loads when handling unaligned loads
llvm-svn: 327037
David Zarzycki [Thu, 8 Mar 2018 18:03:15 +0000 (18:03 +0000)]
[Tests] Remove empty test file that causes the test suite to fail
This empty file was "created" by r327033, which attempted to revert
r327029, which introduced the file.
llvm-svn: 327036
Alina Sbirlea [Thu, 8 Mar 2018 18:03:14 +0000 (18:03 +0000)]
Expose must/may alias info in MemorySSA.
Summary:
Building MemorySSA gathers alias information for Defs/Uses.
Store and expose this information when optimizing uses (when building MemorySSA),
and when optimizing defs or updating uses (getClobberingMemoryAccess).
Current patch does not propagate alias information through MemoryPhis.
Reviewers: gbiv, dberlin
Subscribers: Prazek, sanjoy, llvm-commits
Differential Revision: https://reviews.llvm.org/D38569
llvm-svn: 327035
Sanjay Patel [Thu, 8 Mar 2018 17:34:25 +0000 (17:34 +0000)]
[AMDGPU] fix test to survive the most basic undef constant folding
This will likely need to be changed again for anything more than:
fmul undef, undef -> undef
llvm-svn: 327034
Kuba Mracek [Thu, 8 Mar 2018 17:32:00 +0000 (17:32 +0000)]
Revert r327029
llvm-svn: 327033
Kuba Mracek [Thu, 8 Mar 2018 17:31:31 +0000 (17:31 +0000)]
Revert r327031
llvm-svn: 327032
Kuba Mracek [Thu, 8 Mar 2018 17:24:47 +0000 (17:24 +0000)]
[asan] Fix a false positive ODR violation due to LTO ConstantMerge pass [compiler-rt part]
This fixes a false positive ODR violation that is reported by ASan when using LTO. In cases, where two constant globals have the same value, LTO will merge them, which breaks ASan's ODR detection.
Differential Revision: https://reviews.llvm.org/D43959
llvm-svn: 327031
Sanjay Patel [Thu, 8 Mar 2018 17:24:30 +0000 (17:24 +0000)]
[x86] fix test to be independent of FP undef
llvm-svn: 327030
Kuba Mracek [Thu, 8 Mar 2018 17:24:06 +0000 (17:24 +0000)]
[asan] Fix a false positive ODR violation due to LTO ConstantMerge pass [llvm part]
This fixes a false positive ODR violation that is reported by ASan when using LTO. In cases, where two constant globals have the same value, LTO will merge them, which breaks ASan's ODR detection.
Differential Revision: https://reviews.llvm.org/D43959
llvm-svn: 327029
Sanjay Patel [Thu, 8 Mar 2018 17:13:57 +0000 (17:13 +0000)]
[StructurizeCFG] fix test to be independent of FP undef
llvm-svn: 327028
Andrea Di Biagio [Thu, 8 Mar 2018 17:02:28 +0000 (17:02 +0000)]
[llvm-mca] add override keyword to method ResourcePressureView::printView().
NFC.
llvm-svn: 327027
Sanjay Patel [Thu, 8 Mar 2018 16:56:49 +0000 (16:56 +0000)]
[x86] regenerate checks; NFC
This test will fail if we fix FP undef constant folding.
llvm-svn: 327026
Andrea Di Biagio [Thu, 8 Mar 2018 16:34:19 +0000 (16:34 +0000)]
[llvm-mca] HWEventListener is a class, not struct.
This should appease the buildbots.
llvm-svn: 327025
Tom Stellard [Thu, 8 Mar 2018 16:29:08 +0000 (16:29 +0000)]
merge-request.sh: Update 6.0 metabug for 6.0.1
llvm-svn: 327024
Haojian Wu [Thu, 8 Mar 2018 16:28:12 +0000 (16:28 +0000)]
[clangd] Early return for #include goto definition.
Summary: This would save cost of walking over the AST, NFC.
Reviewers: ilya-biryukov
Subscribers: klimek, jkorous-apple, cfe-commits, ioeric
Differential Revision: https://reviews.llvm.org/D44251
llvm-svn: 327023
Sanjay Patel [Thu, 8 Mar 2018 16:25:37 +0000 (16:25 +0000)]
[StructurizeCFG] auto-generate full checks; NFC
Not sure what the intent of this test is, but this will change when we fix FP undef constant folding.
llvm-svn: 327022
Stefan Pintilie [Thu, 8 Mar 2018 16:24:33 +0000 (16:24 +0000)]
[Power9] Add more missing instructions to the Power 9 scheduler
With this patch we should be able to mark the Power 9 model as complete.
llvm-svn: 327021
Matt Arsenault [Thu, 8 Mar 2018 16:24:16 +0000 (16:24 +0000)]
AMDGPU/GlobalISel: Pass subtarget + TM to LegalizerInfo
These are the parameters x86 already uses.
llvm-svn: 327020
Eric Liu [Thu, 8 Mar 2018 16:14:11 +0000 (16:14 +0000)]
[clangd] Bump vscode extension version to 0.0.5
llvm-svn: 327019
Andrea Di Biagio [Thu, 8 Mar 2018 16:08:43 +0000 (16:08 +0000)]
[llvm-mca] Unify the API for the various views. NFCI
This allows the customization of the performance report.
Users can specify their own custom sequence of views.
Each view contributes a portion of the performance report generated by the
BackendPrinter.
Internally, class BackendPrinter keeps a sequence of views; views are printed
out in sequence when method 'printReport()' is called.
This patch addresses one of the two review comments from Clement in D43951.
llvm-svn: 327018
Pavel Labath [Thu, 8 Mar 2018 16:03:09 +0000 (16:03 +0000)]
Fix std unique pointer pretty-printer for new stl versions
Summary: The unique pointer layout was changed in libstdc++ 6.0.23.
Reviewers: labath, clayborg
Reviewed By: labath, clayborg
Subscribers: luporl, lbianc, lldb-commits
Differential Revision: https://reviews.llvm.org/D44015
Patch by Alexandre Yukio Yamashita <alexandre.yamashita@eldorado.org.br>.
llvm-svn: 327017
Pavel Labath [Thu, 8 Mar 2018 15:52:46 +0000 (15:52 +0000)]
Install lldb's SB headers (pr36630)
These were removed in r309021 in what looks like an accidentally
committed change. This brings them back.
I also rename the header component to lldb-headers (instead of
lldb_headers) to match the llvm style and add a special
install-lldb-headers target, which installs just the headers.
llvm-svn: 327016
Daniel Sanders [Thu, 8 Mar 2018 15:52:45 +0000 (15:52 +0000)]
Fix unused function warning in StatisticTest.cpp
llvm-svn: 327015
Sanjay Patel [Thu, 8 Mar 2018 15:46:38 +0000 (15:46 +0000)]
[InstCombine] regenerate checks; NFC
We may not need any of these tests after rL327012, but leaving
them here for now until that's confirmed.
llvm-svn: 327014
Pavel Labath [Thu, 8 Mar 2018 15:41:13 +0000 (15:41 +0000)]
[LLDB][PPC64] Fix single step and LldbGdbServer tests
Summary:
On PPC64, the tested functions were being entered through their local entry point, while the tests expected the program to stop at the function start address, that, for PPC64, corresponds to the global entry point.
To fix the issue, the test program was modified to call the functions to be tested through function pointers, which, on PPC64, force the calls through the global entry point, while not affecting the test on other platforms.
Reviewers: clayborg, labath
Reviewed By: labath
Subscribers: alexandreyy, lbianc
Differential Revision: https://reviews.llvm.org/D43768
Patch by Leandro Lupori <leandro.lupori@gmail.com>.
llvm-svn: 327013
Sanjay Patel [Thu, 8 Mar 2018 15:39:39 +0000 (15:39 +0000)]
[InstSimplify] add more tests for FP undef; NFC
llvm-svn: 327012
Pavel Labath [Thu, 8 Mar 2018 15:34:42 +0000 (15:34 +0000)]
DWARFVerifier: Basic verification of .debug_names
Summary:
This patch adds basic .debug_names verification capabilities to the
DWARF verifier. Right now, it checks that the headers and abbreviation
tables of the individual name indexes can be parsed correctly, it
verifies the buckets table and the cross-checks the CU lists for
consistency. I intend to add further checks in follow-up patches.
Reviewers: JDevlieghere, aprantl, probinson, dblaikie
Subscribers: vleschuk, echristo, clayborg, llvm-commits
Differential Revision: https://reviews.llvm.org/D44211
llvm-svn: 327011
Andrea Di Biagio [Thu, 8 Mar 2018 15:34:38 +0000 (15:34 +0000)]
[llvm-mca] Emit the 'Instruction Info' table before the resource pressure view.
In future, both the summary information and the 'instruction info' table should
be moved into a separate "Summary" view.
llvm-svn: 327010
Eric Liu [Thu, 8 Mar 2018 15:28:42 +0000 (15:28 +0000)]
[clangd:vscode] Resolve symlinks for file paths from clangd.
Summary:
For features like go-to-definition, clangd can point clients to symlink paths
(e.g. in bazel execroot) which might not be desired if the symlink points to a
file in the workspace. Clangd might not be able to build the file, and users
might prefer to edit the file on the real path.
This change converts file paths from clangd to real path (e.g. resolving symlinks).
Long term, we might want to the symlink handling logic to clangd where clangd
can better decide whether symlinks should be resolved according to e.g. compile
commands.
Reviewers: sammccall
Subscribers: klimek, ilya-biryukov, jkorous-apple, cfe-commits
Differential Revision: https://reviews.llvm.org/D44158
llvm-svn: 327009
Alexey Bataev [Thu, 8 Mar 2018 15:24:08 +0000 (15:24 +0000)]
[OPENMP] Emit sizes/init ptrs etc. data for task reductions before
using.
We may emit the code in wrong order because of incorrect implementation
of the runtime functions for task reductions. Threadprivate storages may
be initialized after real initialization of the reduction items. Patch
fixes this problem.
llvm-svn: 327008
Michal Gorny [Thu, 8 Mar 2018 15:09:38 +0000 (15:09 +0000)]
[cmake] Append -Wl,-rpath-link conditionally to GNULD
Append -Wl,-rpath-link conditionally to whether GNU ld.bfd is used
rather than the Linux+!gold conditionals. Also move it out of 'else'
branch of *BSD handling. This fixes build failures with ld.bfd
on Gentoo/FreeBSD, and should cause no harm on other systems using
ld.bfd.
This patch improves the original logic by reusing results of linker
detection introduced in r307852.
Differential Revision: https://reviews.llvm.org/D43751
llvm-svn: 327007
George Rimar [Thu, 8 Mar 2018 15:06:58 +0000 (15:06 +0000)]
[ELF] - Fix crash relative to SHF_LINK_ORDER sections.
Our code assumes all input sections in an output SHF_LINK_ORDER
section has SHF_LINK_ORDER flag. We do not check that and that can cause a crash.
That happens because we call
std::stable_sort(Sections.begin(), Sections.end(), compareByFilePosition);,
where compareByFilePosition predicate does not expect to see
null when calls getLinkOrderDep.
The same might happen when sections refer to non-regular sections.
Test cases demonstrate the issues, patch fixes them.
Differential revision: https://reviews.llvm.org/D44193
llvm-svn: 327006
Marshall Clow [Thu, 8 Mar 2018 15:01:50 +0000 (15:01 +0000)]
Implement LWG#2518 - Non-member swap for propagate_const should call member swap
llvm-svn: 327005
Sanjay Patel [Thu, 8 Mar 2018 14:57:08 +0000 (14:57 +0000)]
[InstCombine, NewGVN] remove FP undef from tests
I'm trying to preserve the intent of these tests by using
non-undef operands; if we fix FP undef folding these tests
will not pass.
llvm-svn: 327004
George Rimar [Thu, 8 Mar 2018 14:54:38 +0000 (14:54 +0000)]
[ELF] - Support "INSERT AFTER" statement.
This implements INSERT AFTER in a following way:
During reading scripts it collects all insert statements.
After we done and read all files it inserts statements into script commands list.
With that:
* Rest of code does know nothing about INSERT.
* Approach is straightforward and have no visible limitations.
* It is also easy to support INSERT BEFORE (was seen in clang code once).
* Should work for PR35877 and similar cases.
Cons:
* It assumes we have "main" scripts that describes sections.
Differential revision: https://reviews.llvm.org/D43468
llvm-svn: 327003
David Zarzycki [Thu, 8 Mar 2018 14:43:24 +0000 (14:43 +0000)]
[CMake] Add missing test dependency
This makes 'ninja clean ; ninja check-all' work again.
llvm-svn: 327002
Hans Wennborg [Thu, 8 Mar 2018 14:27:28 +0000 (14:27 +0000)]
[COFF] Make the DOS stub a real DOS program
It only adds a few bytes and is nice for backward compatibility.
Differential Revision: https://reviews.llvm.org/D44018
llvm-svn: 327001
Ed Maste [Thu, 8 Mar 2018 13:52:04 +0000 (13:52 +0000)]
Use ellipsis ... to indicate omitted commands
In an example like "clang -fxray-instrument .." the .. could be confused
with a literal .. (parent directory), which is used in commands like
"cmake -GNinja .."
llvm-svn: 327000
Andrea Di Biagio [Thu, 8 Mar 2018 13:43:11 +0000 (13:43 +0000)]
Add llvm-mca.rst to the table of contents in docs/CommandGuide.
This should fix the documentation error reported by builder llvm-sphinx-docs
(build #16407) after r326998.
llvm-svn: 326999
Andrea Di Biagio [Thu, 8 Mar 2018 13:05:02 +0000 (13:05 +0000)]
[llvm-mca] LLVM Machine Code Analyzer.
llvm-mca is an LLVM based performance analysis tool that can be used to
statically measure the performance of code, and to help triage potential
problems with target scheduling models.
llvm-mca uses information which is already available in LLVM (e.g. scheduling
models) to statically measure the performance of machine code in a specific cpu.
Performance is measured in terms of throughput as well as processor resource
consumption. The tool currently works for processors with an out-of-order
backend, for which there is a scheduling model available in LLVM.
The main goal of this tool is not just to predict the performance of the code
when run on the target, but also help with diagnosing potential performance
issues.
Given an assembly code sequence, llvm-mca estimates the IPC (instructions per
cycle), as well as hardware resources pressure. The analysis and reporting style
were mostly inspired by the IACA tool from Intel.
This patch is related to the RFC on llvm-dev visible at this link:
http://lists.llvm.org/pipermail/llvm-dev/2018-March/121490.html
Differential Revision: https://reviews.llvm.org/D43951
llvm-svn: 326998
Krasimir Georgiev [Thu, 8 Mar 2018 11:29:27 +0000 (11:29 +0000)]
[clang-format] Use NestedBlockIndent as a 0 column in formatted raw strings
Summary:
This makes the formatter of raw string literals use NestedBlockIndent for
determining the 0 column of the content inside. This makes the formatting use
less horizonal space and fixes a case where two newlines before and after the
raw string prefix were selected instead of a single newline after it:
Before:
```
aaaa = ffff(
R"pb(
key: value)pb");
```
After:
```
aaaa = ffff(R"pb(
key: value)pb");
```
Reviewers: djasper, sammccall
Reviewed By: sammccall
Subscribers: klimek, cfe-commits
Differential Revision: https://reviews.llvm.org/D44141
llvm-svn: 326996
James Henderson [Thu, 8 Mar 2018 10:53:34 +0000 (10:53 +0000)]
[DWARF] Don't attempt to parse line tables at invalid offsets
Whilst working on improvements to the error handling of the debug line
parsing code, I noticed that if an invalid offset were to be specified
in a call to getOrParseLineTable(), an entry in the LineTableMap would
still be created, even if the offset was not within the section range.
The immediate parsing attempt afterwards would fail (it would end up
getting a version of 0), and thereafter, any subsequent calls to
getOrParseLineTable or getLineTable would return the default-
constructed, invalid line table. In reality, we shouldn't even attempt
to parse this table, and we should always return a nullptr from these
two functions for this situation.
I have tested this via a unit test, which required some new framework
for unit testing debug line. My plan is to add quite a few more unit
tests for the new error reporting mechanism that will follow shortly,
hence the reason why the supporting code for the tests are written the
way they are - I intend to extend the DwarfGenerator class to support
generating debug line. At that point, I'll make sure that there are a
few positive test cases for this and the parsing code too.
Differential Revision: https://reviews.llvm.org/D44200
Reviewers: JDevlieghere, aprantl
llvm-svn: 326995
Jonas Devlieghere [Thu, 8 Mar 2018 10:39:12 +0000 (10:39 +0000)]
[dsymutil] Embed toolchain in dSYM bundle
Allow us to embed the (Xcode) toolchain in the dSYM bundle's property
list.
Differential revision: https://reviews.llvm.org/D44151
llvm-svn: 326994
Andrea Di Biagio [Thu, 8 Mar 2018 10:38:45 +0000 (10:38 +0000)]
[MCSchedule] Always generate processor resource names.
With this patch, the tablegen 'SubtargetEmitter' always generates processor
resource names.
The impact of this patch on the code size of other llvm tools is small. I have
observed an average increase of 0.03% in code size when doing a release build of
LLVM (on windows, using MSVC) with all the default backends.
This change is done in preparation for the upcoming llvm-mca patch.
llvm-svn: 326993
Craig Topper [Thu, 8 Mar 2018 08:02:52 +0000 (08:02 +0000)]
[X86] Change X86::PMULDQ/PMULUDQ opcodes to take vXi64 type as input instead of vXi32.
This instruction can be thought of as reading either the even elements of a vXi32 input or the lower half of each element of a vXi64 input. We currently use the vXi32 interpretation, but vXi64 matches better with its broadcast behavior in EVEX.
I'm looking at moving MULDQ/MULUDQ creation to a DAG combine so we can do it when AVX512DQ is enabled without having to go through Custom lowering. But in some of the test cases we failed to use a broadcast load due to the size difference. This should help with that.
I'm also wondering if we can model these instructions in native IR and remove the intrinsics and I think using a vXi64 type will work better with that.
llvm-svn: 326991
Stephan Bergmann [Thu, 8 Mar 2018 07:34:40 +0000 (07:34 +0000)]
Propagate DLLAttr to friend re-declarations of member functions
...that have already been constructed (e.g., in inner classes) while parsing the
class definition. They would otherwise lack any DLLAttr inherited from the
class, which are only set here (called from Sema::CheckCompletedClass) after the
class definition has been parsed completely.
Differential revision: https://reviews.llvm.org/D16632
llvm-svn: 326990