Philip Reames [Mon, 5 Aug 2019 22:34:59 +0000 (22:34 +0000)]
Add a note to the release not about a potentially breaking optimization
This has come up twice already (once in pr42763 and once in the commit thread), so give warning of a new way in which UB can result in unexpected program behavior.
llvm-svn: 367941
George Burgess IV [Mon, 5 Aug 2019 22:15:40 +0000 (22:15 +0000)]
[Sema] Add -Wpointer-compare
This patch adds a warning that diagnoses comparisons of pointers to
'\0'. This is often indicative of a bug (e.g. the user might've
forgotten to dereference the pointer).
Patch by Elaina Guan!
Differential Revision: https://reviews.llvm.org/D65595
llvm-svn: 367940
Hansang Bae [Mon, 5 Aug 2019 21:46:13 +0000 (21:46 +0000)]
[OpenMP] Fix broken build due to new OMPT tests
New OMPT tests with teams construct should be disabled for GCC as it
emits code with a GOMP entry not supported in the LLVM runtime.
Differential Revision: https://reviews.llvm.org/D65757
llvm-svn: 367939
Peter Collingbourne [Mon, 5 Aug 2019 21:46:10 +0000 (21:46 +0000)]
hwasan: Untag global variable addresses in tests.
Once we start instrumenting globals, all addresses including those of string literals
that we pass to the operating system will start being tagged. Since we can't rely
on the operating system to be able to cope with these addresses, we need to untag
them before passing them to the operating system. This change introduces a macro
that does so and uses it everywhere it is needed.
Differential Revision: https://reviews.llvm.org/D65768
llvm-svn: 367938
Peter Collingbourne [Mon, 5 Aug 2019 21:44:45 +0000 (21:44 +0000)]
Add "REQUIRES: x86-registered-target" to test.
llvm-svn: 367937
Raphael Isemann [Mon, 5 Aug 2019 21:43:53 +0000 (21:43 +0000)]
[lldb][NFC] Document and refactor ClangPersistentVariables::RemovePersistentVariable
llvm-svn: 367936
Keno Fischer [Mon, 5 Aug 2019 21:36:09 +0000 (21:36 +0000)]
[WebAssembly] Fix conflict between ret legalization and sjlj
Summary:
When the WebAssembly backend encounters a return type that doesn't
fit within i32, SelectionDAG performs sret demotion, adding an
additional argument to the start of the function that contains
a pointer to an sret buffer to use instead. However, this conflicts
with the emscripten sjlj lowering pass. There we translate calls like:
```
call {i32, i32} @foo()
```
into (in pseudo-llvm)
```
%addr = @foo
call {i32, i32} @__invoke_{i32,i32}(%addr)
```
i.e. we perform an indirect call through an extra function.
However, the sret transform now transforms this into
the equivalent of
```
%addr = @foo
%sret = alloca {i32, i32}
call {i32, i32} @__invoke_{i32,i32}(%sret, %addr)
```
(while simultaneously translation the implementation of @foo as well).
Unfortunately, this doesn't work out. The __invoke_ ABI expected
the function address to be the first argument, causing crashes.
There is several possible ways to fix this:
1. Implementing the sret rewrite at the IR level as well and performing
it as part of lowering to __invoke
2. Fixing the wasm backend to recognize that __invoke has a special ABI
3. A change to the binaryen/emscripten ABI to recognize this situation
This revision implements the middle option, teaching the backend to
treat __invoke_ functions specially in sret lowering. This is achieved
by
1) Introducing a new CallingConv ID for invoke functions
2) When this CallingConv ID is seen in the backend and the first argument
is marked as sret (a function pointer would never be marked as sret),
swapping the first two arguments.
Reviewed By: tlively, aheejin
Differential Revision: https://reviews.llvm.org/D65463
llvm-svn: 367935
Johannes Doerfert [Mon, 5 Aug 2019 21:35:02 +0000 (21:35 +0000)]
[Attributor][Fix] Do not remove instructions during manifestation
When we remove instructions cached references could still be live. This
patch avoids removing invoke instructions that are replaced by calls and
instead keeps them around but in a dead block.
llvm-svn: 367933
Daniel Sanders [Mon, 5 Aug 2019 21:34:45 +0000 (21:34 +0000)]
Revert Register/MCRegister: Add conversion operators to avoid use of implicit convert to unsigned. NFC
MSVC finds ambiguity where clang doesn't and it looks like it's not going to be an easy fix
Reverting while I figure out how to fix it
This reverts r367916 (git commit
aa15ec3c231717826e3c262b5ef9813d2fb5cadb)
This reverts r367920 (git commit
5d14efe279b5db9f4746ff834ab5c70e249d3871)
llvm-svn: 367932
Johannes Doerfert [Mon, 5 Aug 2019 21:34:45 +0000 (21:34 +0000)]
[Attributor][Fix] Keep invokes if handlers catch asynchronous exceptions
Similar to other places where we transform invokes to calls we need to
be careful if the handler (=personality) can catch asynchronous
exceptions as they are not modeled as part of nounwind.
This is tested with D59978.
llvm-svn: 367931
Louis Dionne [Mon, 5 Aug 2019 21:26:50 +0000 (21:26 +0000)]
[libc++] Accept any non-zero return for .fail.cpp tests
llvm-svn: 367930
Eric Christopher [Mon, 5 Aug 2019 21:25:59 +0000 (21:25 +0000)]
BMI2 support is indicated in bit eight of EBX, not nine.
See Intel SDM, Vol 2A, Table 3-8:
https://www.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-vol-2a-manual.pdf#page=296
Differential Revision: https://reviews.llvm.org/D65766
llvm-svn: 367929
Louis Dionne [Mon, 5 Aug 2019 21:20:42 +0000 (21:20 +0000)]
[NFC][pstl] Remove stray semi-colon
llvm-svn: 367928
Peter Collingbourne [Mon, 5 Aug 2019 20:59:25 +0000 (20:59 +0000)]
llvm-symbolizer: Untag addresses in object files by default.
Any addresses that we pass to llvm-symbolizer are going to be untagged,
while any HWASAN instrumented globals are going to be tagged in the
symbol table. Therefore we need to untag the addresses before using them.
Differential Revision: https://reviews.llvm.org/D65769
llvm-svn: 367926
David Carlier [Mon, 5 Aug 2019 20:54:36 +0000 (20:54 +0000)]
[Sanitizer] Linux refactor shadow huge page mode handling
Disabling Transparent huge page mode refactored in one function.
Reviewers: vitalybuka
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D65771
llvm-svn: 367925
Raphael Isemann [Mon, 5 Aug 2019 20:31:47 +0000 (20:31 +0000)]
[lldb][NFC] Refactor ClangUserExpression::UpdateLanguageForExpr
The UpdateLanguageForExpr should only update the language, but over
time it started to do also do different things related to the generation
of the expression source code. This patch refactors all the source code
generation part into its own function.
llvm-svn: 367922
Lang Hames [Mon, 5 Aug 2019 20:30:35 +0000 (20:30 +0000)]
[ORC] Work around broken GCC/libstdc++ by adding an explicit conversion.
This should fix the bots that have been failing due to r367712.
llvm-svn: 367921
Daniel Sanders [Mon, 5 Aug 2019 20:03:43 +0000 (20:03 +0000)]
Fix MSVC error after r367916
It seems that MSVC sees ambiguity between the operator==()'s where clang
doesn't
llvm-svn: 367920
Amara Emerson [Mon, 5 Aug 2019 20:02:52 +0000 (20:02 +0000)]
[AArch64][GlobalISel] Inline tiny memcpy et al at -O0.
FastISel already does this since the initial arm64 port was upstreamed, so
it seems there are no issues with doing this at -O0 for very small memcpys.
Gives a 0.2% geomean code size improvement on CTMark.
Differential Revision: https://reviews.llvm.org/D65758
llvm-svn: 367919
JF Bastien [Mon, 5 Aug 2019 19:59:07 +0000 (19:59 +0000)]
[docs] don't use :option: for Wall Wextra
The bots are sad that they're not documented.
llvm-svn: 367918
Max Moroz [Mon, 5 Aug 2019 19:55:52 +0000 (19:55 +0000)]
[compiler-rt] Move FDP to include/fuzzer/FuzzedDataProvider.h for easier use.
Summary:
FuzzedDataProvider is a helper class for writing fuzz targets that fuzz
multple inputs simultaneously. The header is supposed to be used for fuzzing
engine agnostic fuzz targets (i.e. the same target can be used with libFuzzer,
AFL, honggfuzz, and other engines). The common thing though is that fuzz targets
are typically compiled with clang, as it provides all sanitizers as well as
different coverage instrumentation modes. Therefore, making this FDP class a
part of the compiler-rt installation package would make it easier to develop
and distribute fuzz targets across different projects, build systems, etc.
Some context also available in https://github.com/google/oss-fuzz/pull/2547.
This CL does not delete the header from `lib/fuzzer/utils` directory in order to
provide the downstream users some time for a smooth migration to the new
header location.
Reviewers: kcc, morehouse
Reviewed By: morehouse
Subscribers: lebedev.ri, kubamracek, dberris, mgorny, delcypher, #sanitizers, llvm-commits
Tags: #llvm, #sanitizers
Differential Revision: https://reviews.llvm.org/D65661
llvm-svn: 367917
Daniel Sanders [Mon, 5 Aug 2019 19:50:25 +0000 (19:50 +0000)]
Register/MCRegister: Add conversion operators to avoid use of implicit convert to unsigned. NFC
Summary:
This has no functional effect but makes it more obvious which parts of the
compiler do not use Register/MCRegister when you mark the implicit conversion
deprecated.
Implicit conversions for comparisons accounted for ~20% (~3k of ~13k) of
the implicit conversions when I first measured it. I haven't maintained
those numbers as other patches have landed though so it may be out of date.
Reviewers: arsenm
Reviewed By: arsenm
Subscribers: wdng, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D65678
llvm-svn: 367916
JF Bastien [Mon, 5 Aug 2019 19:45:23 +0000 (19:45 +0000)]
[docs] don't use :option: for C++ compat
The bots are sad that they're not documented.
llvm-svn: 367914
Nico Weber [Mon, 5 Aug 2019 19:35:59 +0000 (19:35 +0000)]
Fix another uninit read found by msan after r367829
llvm-svn: 367912
Nico Weber [Mon, 5 Aug 2019 19:25:35 +0000 (19:25 +0000)]
compiler-rt: Remove .cc from all lit config files
All cc files have been renamed to cpp now.
llvm-svn: 367911
Michael Kruse [Mon, 5 Aug 2019 19:12:10 +0000 (19:12 +0000)]
[libomptarget] Harmonize emitting CUDA errors and general debug messages.
Ensures that CUDA fail reasons (such as "No CUDA-capable device detected")
are printed together with libomptarget's debug message
(e.g. "Error when setting CUDA context"). Previously, the former was
printed only in CMAKE_BUILD_TYPE=Debug builds while the latter was
enabled by LIBOMPTARGET_ENABLE_DEBUG.
With this change, also only call cuGetErrorString when the error will be
printed.
Suggested-by: Ye Luo <xw111luoye@gmail.com>
Differential Revision: https://reviews.llvm.org/D65687
llvm-svn: 367910
Dmitri Gribenko [Mon, 5 Aug 2019 19:07:09 +0000 (19:07 +0000)]
Revert "Try to fix failing AMDGPU disasm test, both Lin/Win agree this is 0 not 0x0"
This reverts commit r367907, it broke the test.
llvm-svn: 367909
Anusha Basana [Mon, 5 Aug 2019 19:06:55 +0000 (19:06 +0000)]
[llvm-lipo] Implement -segalign
Sets section alignments of the specified architecture slices to the
alignment values.
Alignment values are hexadecimal values that are powers of 2.
Differential Revision: https://reviews.llvm.org/D65420
llvm-svn: 367908
Reid Kleckner [Mon, 5 Aug 2019 18:46:26 +0000 (18:46 +0000)]
Try to fix failing AMDGPU disasm test, both Lin/Win agree this is 0 not 0x0
llvm-svn: 367907
Jan Korous [Mon, 5 Aug 2019 18:44:07 +0000 (18:44 +0000)]
[DirectoryWatcher][linux] Fix build for older kernels
Apparently kernel support for IN_EXCL_UNLINK in inotify_add_watch() doesn't imply it's defined in sys/inotify.h.
https://bugs.llvm.org/show_bug.cgi?id=42824
llvm-svn: 367906
Michael Kruse [Mon, 5 Aug 2019 18:43:21 +0000 (18:43 +0000)]
[OpenMP 5.0] Codegen support for user-defined mappers.
This patch implements the code generation for OpenMP 5.0 declare mapper
(user-defined mapper) constructs. For each declare mapper, a mapper
function is generated. These mapper functions will be called by the
runtime and/or other mapper functions to achieve user defined mapping.
The design slides can be found at
https://github.com/lingda-li/public-sharing/blob/master/mapper_runtime_design.pptx
Re-commit after revert in r367773 because r367755 changed the LLVM-IR
output such that a CHECK line failed.
Patch by Lingda Li <lildmh@gmail.com>
Differential Revision: https://reviews.llvm.org/D59474
llvm-svn: 367905
Dmitri Gribenko [Mon, 5 Aug 2019 18:36:43 +0000 (18:36 +0000)]
Revert "[AMDGPU] Use S_DENORM_MODE for gfx10"
This reverts commit r367882. It broke the test
MC/Disassembler/AMDGPU/gfx10_dasm_all.txt.
llvm-svn: 367904
Louis Dionne [Mon, 5 Aug 2019 18:29:14 +0000 (18:29 +0000)]
[libc++] Take 2: Integrate the PSTL into libc++
Summary:
This commit allows specifying LIBCXX_ENABLE_PARALLEL_ALGORITHMS when
configuring libc++ in CMake. When that option is enabled, libc++ will
assume that the PSTL can be found somewhere on the CMake module path,
and it will provide the C++17 parallel algorithms based on the PSTL
(that is assumed to be available).
The commit also adds support for running the PSTL tests as part of
the libc++ test suite.
The first attempt to commit this failed because it exposed a bug in the
tests for modules. Now that this has been fixed, it should be safe to
commit this.
Reviewers: EricWF
Subscribers: mgorny, christof, jkorous, dexonsmith, libcxx-commits, mclow.lists, EricWF
Tags: #libc
Differential Revision: https://reviews.llvm.org/D60480
llvm-svn: 367903
Craig Topper [Mon, 5 Aug 2019 18:25:36 +0000 (18:25 +0000)]
[X86] Enable -x86-experimental-vector-widening-legalization by default.
This patch changes our defualt legalization behavior for 16, 32, and
64 bit vectors with i8/i16/i32/i64 scalar types from promotion to
widening. For example, v8i8 will now be widened to v16i8 instead of
promoted to v8i16. This keeps the elements widths the same and pads
with undef elements. We believe this is a better legalization strategy.
But it carries some issues due to the fragmented vector ISA. For
example, i8 shifts and multiplies get widened and then later have
to be promoted/split into vXi16 vectors.
This has the potential to cause regressions so we wanted to get
it in early in the 10.0 cycle so we have plenty of time to
address them.
Next steps will be to merge tests that explicitly test the command
line option. And then we can remove the option and its associated
code.
llvm-svn: 367901
Philip Reames [Mon, 5 Aug 2019 18:25:08 +0000 (18:25 +0000)]
Robustify update_test_checks.py to non-autogened tests, and add a mode to skip non-autogenerated ones
Intended use case is:
./utils/update_test_checks.py test/Transform/PassDir/* --update-only
(i.e. rapidly be able to see changes in autogened filed, before handing non-autogened tests individually)
Differential Revision: https://reviews.llvm.org/D65610
llvm-svn: 367900
David Blaikie [Mon, 5 Aug 2019 18:12:50 +0000 (18:12 +0000)]
Fix build when both gtest death tests and LLVM_NODISCARD are available.
(matching r367495)
llvm-svn: 367899
Evandro Menezes [Mon, 5 Aug 2019 18:09:14 +0000 (18:09 +0000)]
[AArch64] Expand bcmp() for small block lengths
Patch D56593 by @courbet results in calls to `bcmp()` in some cases, should
the target support the it. Unless `TTI::MemCmpExpansionOptions()`
is overridden by the target.
In a proprietary benchmark we see a performance drop of about 12% on PNG
compression before this patch, though it passes all tests.
This patch mirrors X86 for AArch64 and initializes
`TTI::MemCmpExpansionOptions()` to then expand calls to `bcmp()` when
appropriate. No tuning of the parameters was performed, but, at this point,
it's enough to recover the performance drop above.
This problem also exists on ARM. Once a consensus is reached for AArch64, we
can work to fix ARM as well.
Authors:
- Evandro Menezes (@evandro) <e.menezes@samsung.com>
- Brian Rzycki (@brzycki) <b.rzycki@samsung.com>
Differential revision: https://reviews.llvm.org/D64805
llvm-svn: 367898
Roman Lebedev [Mon, 5 Aug 2019 18:01:22 +0000 (18:01 +0000)]
[InstCombine][NFC] Tests for non-canonical clamp-like pattern
As discussed in https://reviews.llvm.org/D65148#1607019
The canonical fold is: https://rise4fun.com/Alive/FKe
llvm-svn: 367897
Chris Bieneman [Mon, 5 Aug 2019 18:00:55 +0000 (18:00 +0000)]
NFC. Documenting Native tablegen dependency
Adding documentation explaining why this dependency is required and should not be removed again.
llvm-svn: 367896
Chris Bieneman [Mon, 5 Aug 2019 17:50:08 +0000 (17:50 +0000)]
NATIVE tablegen needs to depend on target tablegen
This dependency was removed in r357486, which has lead to a stream of difficult to diagnose bugs.
Without this dependency, when building with `LLVM_OPTIMIZED_TABLEGEN=On` the native tablegen executible may not be rebuilt at all, and often won't get rebuilt before targets that use the tablegen headers. In the best case this results in a build-time failure, in the worst case it results in runtime failures.
llvm-svn: 367895
Pablo Barrio [Mon, 5 Aug 2019 17:38:58 +0000 (17:38 +0000)]
[AArch64] Set preferred function alignment to 16 bytes on Neoverse N1
Summary:
The Arm Neoverse N1 Software Optimization Guide [1], Section "4.8 Branch
instruction alignment" states:
"Consider aligning subroutine entry points and branch targets to 32B
boundaries, within the bounds of the code-density requirements of the
program."
This patch sets the preferred function alignment on Neoverse N1 to 2^4=16B.
This was already the case in some of the latest Cortex-A CPUs. Benchmarking
in previous Cortex-A CPUs suggested that 16B alignment is already better
than the default. See commit
d04ee305.
The reason we don't set it to 32B right now (as the optimisation guide
suggests) is that this will impact code size and perhaps the instruction
cache performance. Therefore we need benchmark numbers first.
I have also added testing for A75 and A76 that we were missing.
[1] https://developer.arm.com/docs/swog309707/latest
Reviewers: fhahn, greened, samparker, dmgreen
Reviewed By: dmgreen
Subscribers: dmgreen, javed.absar, kristof.beyls, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D65654
llvm-svn: 367894
Jan Vesely [Mon, 5 Aug 2019 17:23:23 +0000 (17:23 +0000)]
Revert "Rename F_{None,Text,Append} to OF_{None,Text,Append}. NFC"
This reverts commit
58c814614d2ac69bcf79b09543505fac80ada4e6.
Fixes build breakage using LLVM<7.
llvm-svn: 367893
Johan Vikstrom [Mon, 5 Aug 2019 17:14:46 +0000 (17:14 +0000)]
[AST] Fix buildbot failure because of raw string inside macro from 367839.
llvm-svn: 367892
Sanjay Patel [Mon, 5 Aug 2019 16:59:58 +0000 (16:59 +0000)]
[InstCombine] combine mul+shl separated by zext
This appears to slightly help patterns similar to what's
shown in PR42874:
https://bugs.llvm.org/show_bug.cgi?id=42874
...but not in the way requested.
That fix will require some later IR and/or backend pass to
decompose multiply/shifts into something more optimal per
target. Those transforms already exist in some basic forms,
but probably need enhancing to catch more cases.
https://rise4fun.com/Alive/Qzv2
llvm-svn: 367891
Nico Weber [Mon, 5 Aug 2019 16:55:04 +0000 (16:55 +0000)]
gn build: run "gn format"
llvm-svn: 367890
JF Bastien [Mon, 5 Aug 2019 16:53:45 +0000 (16:53 +0000)]
[docs] document -Weveything more betterer
Reviewers: aaron.ballman
Subscribers: jkorous, dexonsmith, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D65706
llvm-svn: 367889
Nico Weber [Mon, 5 Aug 2019 16:50:56 +0000 (16:50 +0000)]
Follow-up for r367863 and r367656
llvm-svn: 367888
Nico Weber [Mon, 5 Aug 2019 16:48:12 +0000 (16:48 +0000)]
compiler-rt: Rename cc files below test/asan to cpp
See r367803 and similar other changes.
llvm-svn: 367887
David Green [Mon, 5 Aug 2019 16:27:36 +0000 (16:27 +0000)]
[AST] Fix RecursiveASTVisitorTest multiline string literal. NFC
Some compiler, notably older gccs (< 8) can have trouble with multiline raw
string literals inside macros. This just moves the code outsize the macro, to
attempt to appease the bots.
llvm-svn: 367885
Jordan Rupprecht [Mon, 5 Aug 2019 16:26:48 +0000 (16:26 +0000)]
[llvm-readobj][test] Add llvm-readobj style test cases for r367878
llvm-svn: 367884
Sanjay Patel [Mon, 5 Aug 2019 16:17:07 +0000 (16:17 +0000)]
[InstCombine] add tests for shl+mul; NFC
llvm-svn: 367883
Austin Kerbow [Mon, 5 Aug 2019 16:09:49 +0000 (16:09 +0000)]
[AMDGPU] Use S_DENORM_MODE for gfx10
Summary: During fdiv32 lowering use S_DENORM_MODE to select denorm mode in gfx10.
Reviewers: arsenm, rampitec
Reviewed By: arsenm, rampitec
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D65620
llvm-svn: 367882
Tom Stellard [Mon, 5 Aug 2019 16:08:44 +0000 (16:08 +0000)]
AMDGPU/LoadStoreOptimizer: Set the correct offset whem merging MMOs
Summary:
This is a follow up to r367237. MachineFunction::getMachineMemOperand()
adds the offset parameter to the existing offset instead of resetting it.
So we need to reset the offset to the correct value after calling this
function.
Reviewers: arsenm
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D65557
llvm-svn: 367881
Sanjay Patel [Mon, 5 Aug 2019 16:04:07 +0000 (16:04 +0000)]
[InstCombine] add extra use constraint for shl-zext fold
As the test shows, we can end up with more instructions than
we started with if we don't include the extra-use check.
llvm-svn: 367880
Matt Arsenault [Mon, 5 Aug 2019 15:59:07 +0000 (15:59 +0000)]
AMDGPU: Correct behavior of f16 buffer loads
Don't assume format loads for f16. Also fixes support for targets
without i16.
llvm-svn: 367879
Jordan Rupprecht [Mon, 5 Aug 2019 15:43:20 +0000 (15:43 +0000)]
[llvm-readelf] Fix core note descriptions
Summary:
Core files have different descriptions for note values. llvm-readelf currently prints the generic note type, which is wrong when using it to read a core file.
To verify the constants/strings, see:
Values: https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=blob;f=include/elf/common.h;h=
75c4fb7e9d7c0f780d635ac305f579546b7b071b;hb=HEAD#l571
Strings: https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=blob;f=binutils/readelf.c;h=
c31a5c1266b7bb62a485895b01b49e1f832ade35;hb=HEAD#l16881
Note: this does not handle printing the note data for NT_FILE, it just fixes the descriptions.
Reviewers: MaskRay
Reviewed By: MaskRay
Subscribers: labath, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D65608
llvm-svn: 367878
Fangrui Song [Mon, 5 Aug 2019 15:34:01 +0000 (15:34 +0000)]
[ELF][test] Reorganize some tls-*.s tests
Some tls-*.s tests do not test generic TLS behavior but rather are x86 specific.
Rename them to i386-*.s or x86-64-*.s
Delete tls-static.s: covered by tls-opt.s
Delete tls-opt-no-plt.s: add --implicit-check-not=.plt to x86-64-tls-gdie.s to cover it
Rename tls-dynamic-i686.s to i386-tls-dynamic.s
Rename tls-i686.s to i386-tls-le.s
Rename tls-opt-i686.s to i386-tls-opt.s
Rename tls-opt-iele-i686-nopic.s to i386-tls-opt-iele-nopic.s
Rename tls-dynamic.s to x86-64-tls-dynamic.s . IE should be split off in the future.
Rename tls-error.s to x86-64-reloc-tpoff32-error.s
Rename tls-opt-gdie.s to x86-64-tls-gdie.s
Rename tls-opt-x86_64-noplt.s to x86-64-tls-opt-noplt.s
Rename tls-opt-local.s => x86-64-tls-ie-opt-local.s . It can be merged with x86-64-tls-ie-local.s in the future.
llvm-svn: 367877
Sanjay Patel [Mon, 5 Aug 2019 15:25:07 +0000 (15:25 +0000)]
[InstCombine] add test for shl-zext with extra use; NFC
llvm-svn: 367876
Nico Weber [Mon, 5 Aug 2019 15:23:10 +0000 (15:23 +0000)]
Undo what looks like an unintentional change in r367829
The MSan bot was (rightfully) complaining that NumASTLoaded was
unitialized, so put the initialization removed in r367829 back in.
While here, remove two needless semicolons added in that change.
llvm-svn: 367875
Nico Weber [Mon, 5 Aug 2019 15:10:15 +0000 (15:10 +0000)]
Try to fix OOB tests more on Windows after r367642
See PR42868 for more details.
The affected list of tests is:
Failing Tests (8):
AddressSanitizer-Unit :: ./Asan-x86_64-calls-Test.exe/AddressSanitizer.LargeOOBRightTest
AddressSanitizer-Unit :: ./Asan-x86_64-calls-Test.exe/AddressSanitizer.OOBRightTest
AddressSanitizer-Unit :: ./Asan-x86_64-calls-Test.exe/AddressSanitizer.OOB_char
AddressSanitizer-Unit :: ./Asan-x86_64-calls-Test.exe/AddressSanitizer.OOB_int
AddressSanitizer-Unit :: ./Asan-x86_64-inline-Test.exe/AddressSanitizer.LargeOOBRightTest
AddressSanitizer-Unit :: ./Asan-x86_64-inline-Test.exe/AddressSanitizer.OOBRightTest
AddressSanitizer-Unit :: ./Asan-x86_64-inline-Test.exe/AddressSanitizer.OOB_char
AddressSanitizer-Unit :: ./Asan-x86_64-inline-Test.exe/AddressSanitizer.OOB_int
llvm-svn: 367874
Nico Weber [Mon, 5 Aug 2019 15:04:20 +0000 (15:04 +0000)]
Try to fix OOB tests on at least Windows after r367642
gtest's built-in regex engine doesn't support (). Looks like it's not
needed, just remove it.
See PR42868 for more details.
llvm-svn: 367873
Matt Arsenault [Mon, 5 Aug 2019 14:57:59 +0000 (14:57 +0000)]
AMDGPU: Correct behavior of f16/i16 non-format store intrinsics
This was switching to use a format store for a non-format store for
f16 types. Also fixes i16/f16 stores on targets without legal f16.
The corresponding loads also need to be fixed.
llvm-svn: 367872
Matt Arsenault [Mon, 5 Aug 2019 14:40:26 +0000 (14:40 +0000)]
AMDGPU/GlobalISel: Alternative mappings for constants
Without context we assume SGPR. Allowing VGPR constants theoretically
helps avoid a copy. This seems to not actually work now, and the
choice isn't based on the use bank.
llvm-svn: 367871
Matt Arsenault [Mon, 5 Aug 2019 14:40:23 +0000 (14:40 +0000)]
AMDGPU/GlobalISel: Don't reject shader types
I'm not sure what complications these present, but the current
argument lowering is pretty much directly copied from the DAG
lowering, so I assume these work as they should.
No tests because I'm lazy and things are getting pretty close to the
point where the existing calling-conventions.ll can be shared with
SelectionDAG.
llvm-svn: 367870
Fangrui Song [Mon, 5 Aug 2019 14:31:39 +0000 (14:31 +0000)]
[ELF] Consistently prioritize non-* wildcards overs "*" in version scripts
We prioritize non-* wildcards overs VER_NDX_LOCAL/VER_NDX_GLOBAL "*".
This patch generalizes the rule to "*" of other versions and thus fixes PR40176.
I don't feel strongly about this GNU linkers' behavior but the
generalization simplifies code.
Delete `config->defaultSymbolVersion` which was used to special case
VER_NDX_LOCAL/VER_NDX_GLOBAL "*".
In `SymbolTable::scanVersionScript`, custom versions are handled the same
way as VER_NDX_LOCAL/VER_NDX_GLOBAL. So merge
`config->versionScript{Locals,Globals}` into `config->versionDefinitions`.
Overall this seems to simplify the code.
In `SymbolTable::assign{Exact,Wildcard}Versions`,
`sym->verdefIndex == config->defaultSymbolVersion` is changed to
`verdefIndex == UINT32_C(-1)`.
This allows us to give duplicate assignment diagnostics for
`{ global: foo; };` `V1 { global: foo; };`
In test/linkerscript/version-script.s:
vs_index of an undefined symbol changes from 0 to 1. This doesn't matter (arguably 1 is better because the binding is STB_GLOBAL) because vs_index of an undefined symbol is ignored.
Reviewed By: ruiu
Differential Revision: https://reviews.llvm.org/D65716
llvm-svn: 367869
Nico Weber [Mon, 5 Aug 2019 14:22:21 +0000 (14:22 +0000)]
gn build: Merge r367864
llvm-svn: 367868
Nilanjana Basu [Mon, 5 Aug 2019 14:16:58 +0000 (14:16 +0000)]
Changing representation of .cv_def_range directives in Codeview debug info assembly format for better readability
llvm-svn: 367867
Rainer Orth [Mon, 5 Aug 2019 14:06:16 +0000 (14:06 +0000)]
[Driver] Properly use values-X[ca].o, values-xpg[46].o on Solaris
Builtins-*-sunos :: compiler_rt_logbf_test.c currently FAILs on Solaris, both SPARC and
x86, 32 and 64-bit.
It turned out that this is due to different behaviour of logb depending on the C
standard compiled for, as documented on logb(3M):
RETURN VALUES
Upon successful completion, these functions return the exponent of x.
If x is subnormal:
o For SUSv3-conforming applications compiled with the c99 com-
piler driver (see standards(7)), the exponent of x as if x
were normalized is returned.
o Otherwise, if compiled with the cc compiler driver, -1022,
-126, and -16382 are returned for logb(), logbf(), and
logbl(), respectively.
Studio c99 and gcc control this by linking with the appropriate version of values-xpg[46].o, but clang uses neither of those.
The following patch fixes this by following what gcc does, as corrected some time ago in
Fix use of Solaris values-Xc.o (PR target/40411)
https://gcc.gnu.org/ml/gcc-patches/2018-01/msg02350.html and
https://gcc.gnu.org/ml/gcc-patches/2018-01/msg02384.html.
Tested on x86_64-pc-solaris2.11, sparcv9-sun-solaris2.11, and x86_64-pc-linux-gnu.
Differential Revision: https://reviews.llvm.org/D64793
llvm-svn: 367866
Rainer Orth [Mon, 5 Aug 2019 14:00:43 +0000 (14:00 +0000)]
[lldb][clang] Reflect LangStandard.h move to clang/Basic
D65562 <https://reviews.llvm.org/D65562> moves LangStandard.h from clang/Frontend to clang/Basic. This patch
adjusts the single file in lldb that uses it to match.
Tested on x86_64-pc-linux-gnu.
Differential Revision: https://reviews.llvm.org/D65717
llvm-svn: 367865
Rainer Orth [Mon, 5 Aug 2019 13:59:26 +0000 (13:59 +0000)]
Move LangStandard*, InputKind::Language to Basic
This patch is a prerequisite for using LangStandard from Driver in
https://reviews.llvm.org/D64793.
It moves LangStandard* and InputKind::Language to Basic. It is mostly
mechanical, with only a few changes of note:
- enum Language has been changed into enum class Language : uint8_t to
avoid a clash between OpenCL in enum Language and OpenCL in enum
LangFeatures and not to increase the size of class InputKind.
- Now that getLangStandardForName, which is currently unused, also checks
both canonical and alias names, I've introduced a helper getLangKind
which factors out a code pattern already used 3 times.
The patch has been tested on x86_64-pc-solaris2.11, sparcv9-sun-solaris2.11,
and x86_64-pc-linux-gnu.
There's a companion patch for lldb which uses LangStandard.h
(https://reviews.llvm.org/D65717).
While polly includes isl which in turn uses InputKind::C, that part of the
code isn't even built inside the llvm tree. I've posted a patch to allow
for both InputKind::C and Language::C upstream
(https://groups.google.com/forum/#!topic/isl-development/6oEvNWOSQFE).
Differential Revision: https://reviews.llvm.org/D65562
llvm-svn: 367864
Nico Weber [Mon, 5 Aug 2019 13:57:03 +0000 (13:57 +0000)]
compiler-rt: Rename cc files below test/sanitizer_common to cpp
See r367803 and similar other changes.
llvm-svn: 367863
Hubert Tong [Mon, 5 Aug 2019 13:55:41 +0000 (13:55 +0000)]
[yaml2obj][tests] Fix overly restrictive od output check
Summary:
rL364517 introduced further instances of `od` output checking of the
kind previously corrected by rL363829. This patch corrects the issue by
suppressing output of the input offset. The check remains sufficiently
sensitive to test for the intended value of the specific byte since the
relevant byte value is the only output we are expecting from `od`.
Reviewers: grimar, xingxue, daltenty, jasonliu, jhenderson, MaskRay
Reviewed By: grimar, MaskRay
Subscribers: llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D65680
llvm-svn: 367862
Nilanjana Basu [Mon, 5 Aug 2019 13:55:21 +0000 (13:55 +0000)]
Revert "Changing representation of .cv_def_range directives in Codeview debug info assembly format for better readability"
This reverts commit
a885afa9fa8cab3b34f1ddf3d21535f88b662881.
llvm-svn: 367861
George Rimar [Mon, 5 Aug 2019 13:54:35 +0000 (13:54 +0000)]
[yaml2obj] - Allow overriding sh_entsize for SHT_GNU_versym sections.
This allows to write a test case for one of untested errors
in llvm/Object/ELF.h.
I did it in this patch to demonstrate.
Differential revision: https://reviews.llvm.org/D65394
llvm-svn: 367860
Cullen Rhodes [Mon, 5 Aug 2019 13:44:10 +0000 (13:44 +0000)]
[AArch64] Implement initial SVE calling convention support
Summary:
This patch adds initial support for the SVE calling convention such that
SVE types can be passed as arguments and return values to/from a
subroutine.
The SVE AAPCS states [1]:
z0-z7 are used to pass scalable vector arguments to a subroutine,
and to return scalable vector results from a function. If a
subroutine takes arguments in scalable vector or predicate
registers, or if it is a function that returns results in such
registers, it must ensure that the entire contents of z8-z23 are
preserved across the call. In other cases it need only preserve the
low 64 bits of z8-z15, as described in ยง5.1.2.
p0-p3 are used to pass scalable predicate arguments to a subroutine
and to return scalable predicate results from a function. If a
subroutine takes arguments in scalable vector or predicate
registers, or if it is a function that returns results in these
registers, it must ensure that p4-p15 are preserved across the call.
In other cases it need not preserve any scalable predicate register
contents.
SVE predicate and data registers are passed indirectly (i.e. spilled to the
stack and pass the address) if they exceed the registers used for argument
passing defined by the PCS referenced above. Until SVE stack support is merged
we can't spill SVE registers to the stack, so currently an llvm_unreachable is
used where we will eventually handle this.
[1] https://static.docs.arm.com/100986/0000/100986_0000.pdf
Reviewed By: ostannard
Differential Revision: https://reviews.llvm.org/D65448
llvm-svn: 367859
Nico Weber [Mon, 5 Aug 2019 13:42:31 +0000 (13:42 +0000)]
compiler-rt: Rename remaining cc files in test/profile to cpp
See r367803 and similar other changes.
llvm-svn: 367858
Raphael Isemann [Mon, 5 Aug 2019 13:40:39 +0000 (13:40 +0000)]
[lldb][NFC] Fix documentation for ClangPersistentVariables::m_next_persistent_variable_id
llvm-svn: 367857
Nico Weber [Mon, 5 Aug 2019 13:27:37 +0000 (13:27 +0000)]
More follow-up to r367851
llvm-svn: 367856
Nico Weber [Mon, 5 Aug 2019 13:23:38 +0000 (13:23 +0000)]
compiler-rt: Rename last few cc files below test/ubsan to cpp
See r367803 and similar other changes.
llvm-svn: 367855
Nico Weber [Mon, 5 Aug 2019 13:19:28 +0000 (13:19 +0000)]
compiler-rt: Rename .cc file in test/dfsan to cpp
See r367849 et al.
llvm-svn: 367854
Andrea Di Biagio [Mon, 5 Aug 2019 13:18:37 +0000 (13:18 +0000)]
[MCA][doc] Add a section for the 'Bottleneck Analysis'.
Also clarify the meaning of 'Block RThroughput' and 'RThroughput'.
llvm-svn: 367853
George Rimar [Mon, 5 Aug 2019 13:16:06 +0000 (13:16 +0000)]
[obj2yaml] - Teach tool to dump SHT_NULL sections.
Recently an advanced support of SHT_NULL sections
was implemented in yaml2obj.
This patch adds a corresponding support to obj2yaml.
Differential revision: https://reviews.llvm.org/D65215
llvm-svn: 367852
Nico Weber [Mon, 5 Aug 2019 13:12:23 +0000 (13:12 +0000)]
compiler-rt: Rename cc files in test/hwasan/TestCases subdirectories as well
Should've been part of r367849.
llvm-svn: 367851
Nilanjana Basu [Mon, 5 Aug 2019 13:11:51 +0000 (13:11 +0000)]
Changing representation of .cv_def_range directives in Codeview debug info assembly format for better readability
llvm-svn: 367850
Nico Weber [Mon, 5 Aug 2019 13:10:50 +0000 (13:10 +0000)]
compiler-rt: Rename .cc file in test/hwasan to .cpp
Like r367463, but for test/hwasan.
llvm-svn: 367849
Hans Wennborg [Mon, 5 Aug 2019 13:04:12 +0000 (13:04 +0000)]
test-release.sh: Perform the sed substitution on both files (PR42739)
The comparison would otherwise fail if Phase2 occurrs naturally in the
object file. It would get replaced with Phase3 in the one .o, but not
in the other.
We were already running both files through sed to have them processed in
this same way; this is a logical extension of that.
llvm-svn: 367847
Hans Wennborg [Mon, 5 Aug 2019 13:04:07 +0000 (13:04 +0000)]
Write the RequiredLibraries for 'all' in LibraryDependencies.inc in a deterministic order (PR42739)
llvm-svn: 367846
Haojian Wu [Mon, 5 Aug 2019 12:48:09 +0000 (12:48 +0000)]
[clangd] Add a callback mechanism for handling responses from client.
Summary:
The callback will be invoked in clangd when we receive a reply from the client.
This is a prerequisite of implementing a generic mechanism for chainable
refactorings (e.g. extract variable and rename), this would allow server to
trigger a new request to the LSP client after receiving a reply from the client.
Reviewers: sammccall
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D65387
llvm-svn: 367845
Nico Weber [Mon, 5 Aug 2019 12:44:53 +0000 (12:44 +0000)]
gn build: Merge r367839
llvm-svn: 367844
Raphael Isemann [Mon, 5 Aug 2019 12:37:54 +0000 (12:37 +0000)]
[lldb][NFC] Clang format GetNextPersistentVariableName signature
llvm-svn: 367843
Raphael Isemann [Mon, 5 Aug 2019 12:32:59 +0000 (12:32 +0000)]
[lldb] Move redundant persistent variable counter to ClangPersistentVariables
Currently Target::m_next_persistent_variable_index is counting up
for our persistent variables ($0, $1, ...) but we also have a
unused counter that is supposed to do this in
ClangPersistentVariables but that stays always at 0 (because
we currently increase the target counter when we should increase
that unused counter).
This patch removes the counter in Target and lets the documented
counter in ClangPersistentVariables do the variable counting.
Patch *should* be NFC, but it might unexpectedly bring LLDB to
new code paths that could contain exciting new bugs to fix.
llvm-svn: 367842
Nico Weber [Mon, 5 Aug 2019 12:28:08 +0000 (12:28 +0000)]
Remove another empty directory.
llvm-svn: 367841
Raphael Isemann [Mon, 5 Aug 2019 12:23:39 +0000 (12:23 +0000)]
[clang][NFC] Remove unused private variable 'CI' in CrossTranslationUnit.h
It seems because of the recent refactorings this variable has become unused
and now we get this warning in the build logs:
In file included from llvm/clang/lib/CrossTU/CrossTranslationUnit.cpp:12:
llvm/clang/include/clang/CrossTU/CrossTranslationUnit.h:200:21: warning: private field 'CI' is not used [-Wunused-private-field]
CompilerInstance &CI;
^
I'll remove them for now to get the builds back to green.
llvm-svn: 367840
Johan Vikstrom [Mon, 5 Aug 2019 12:20:43 +0000 (12:20 +0000)]
[AST] Fix RecursiveASTVisitor visiting implicit constructor initializers.
Summary: RecursiveASTVisitor was visiting implcit constructor initializers. This caused semantic highlighting in clangd to emit error logs. Fixes this by checking if the constructor is written or if the visitor should visit implicit decls.
Reviewers: hokein, ilya-biryukov
Subscribers: kadircet, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D65735
llvm-svn: 367839
Nico Weber [Mon, 5 Aug 2019 12:18:04 +0000 (12:18 +0000)]
Remove empty directory.
The directory was added in rL292747 which was reverted again in 292751 --
but the revert was apparently done via git and the empty dir stuck around in
svn.
llvm-svn: 367838
Martin Storsjo [Mon, 5 Aug 2019 11:57:06 +0000 (11:57 +0000)]
[MinGW] Add an lld specific option for requesting to delay load libraries
With GNU tools, delayload is handled completely differently. (One
creates a specific delayload import library using dlltool and then
links against it instead of the normal import library.)
Instead of requiring using -Xlink=-delayload:lib.dll, we can provide
an lld specific option for this.
Differential Revision: https://reviews.llvm.org/D65728
llvm-svn: 367837
Martin Storsjo [Mon, 5 Aug 2019 11:57:00 +0000 (11:57 +0000)]
[COFF] Omit automatically imported symbols from the symbol table
These symbols actually point to the symbol's IAT entry, which
obviously is different from the symbol itself (which is imported
from a different module and doesn't exist in the current one).
Omitting this symbol helps gdb inspect automatically imported
symbols, see https://sourceware.org/bugzilla/show_bug.cgi?id=24574
for discussion on the matter.
Surprisingly, those extra symbols don't seem to be an issue for
gdb when the sources have been built with clang, only with gcc.
The actual logic in gdb that this depends on still is unknown, but
omitting these symbols from the symbol table is the right thing to
do in any case.
Differential Revision: https://reviews.llvm.org/D65727
llvm-svn: 367836
Pavel Labath [Mon, 5 Aug 2019 11:29:01 +0000 (11:29 +0000)]
Fix PDB tests after r367820
The commit changed Module dumping code to call SymbolFile::Dump
directly, which meant that we were no longer showing the plugin name in
the output (as that was done in the SymbolVendor).
This adds the plugin name printing code to the SymbolFile dump method,
and tweak the assertions in the PDB tests to match it correctly.
llvm-svn: 367835
Sanjay Patel [Mon, 5 Aug 2019 11:27:07 +0000 (11:27 +0000)]
[DAGCombiner][x86] prevent infinite loop from truncate/extend transforms
The test case is based on the example from the post-commit thread for:
https://reviews.llvm.org/rGc9171bd0a955
This replaces the x86-specific simple-type check from:
rL367766
with a check in the DAGCombiner. Adding the check isn't
strictly necessary after the fix from:
rL367768
...but it seems likely that we're heading for trouble if
we are creating weird types in this transform.
I combined the earlier legality check into the initial
clause to simplify the code.
So we should only try the trunc/sext transform at the
earliest combine stage, but we limit the transform to
simple types anyway because the TLI hook is probably
too lax about what it considers a free truncate.
llvm-svn: 367834
George Rimar [Mon, 5 Aug 2019 11:19:28 +0000 (11:19 +0000)]
[llvm/Object] - Remove ELFFile<ELFT>::getSection(const StringRef SectionName). NFC.
This method is dead. It was introduced in D47989,
but now the logic from D63475 is used in llvm-readobj instead.
Also it has a problem: it returns the first matching section,
even if there are multiple sections with the same name.
Differential revision: https://reviews.llvm.org/D65393
llvm-svn: 367833