Simon Pilgrim [Thu, 30 Apr 2020 17:45:30 +0000 (18:45 +0100)]
[X86][SSE] Canonicalize UNARYSHUFFLE(XOR(X,-1) -> XOR(UNARYSHUFFLE(X),-1)
This pushes the NOT pattern up the DAG to help expose it for further combines (AND->ANDN in particular).
The PSHUFD/MOVDDUP 'splat' cases are the only ones I've seen in the wild so far, we can further generalize if/when we need to.
Sam McCall [Wed, 29 Apr 2020 23:25:18 +0000 (01:25 +0200)]
[clangd] Render doc-comment code spans with `backticks` in plaintext mode
Reviewers: kadircet
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79142
Louis Dionne [Thu, 30 Apr 2020 18:09:39 +0000 (14:09 -0400)]
[libc++] Make sure we create the working directory for running config check commands
The internal Lit shell requires the current working directory to exist.
This didn't show up locally because the directories were already created
by previous runs of the tests.
River Riddle [Thu, 30 Apr 2020 17:56:33 +0000 (10:56 -0700)]
[mlir][OpInterfaceGen] Emit the utility Trait decl as a class and not using directive
This fixes the build on MSVC where it is unable to handle template-template using directives properly.
Jonas Devlieghere [Thu, 30 Apr 2020 18:04:51 +0000 (11:04 -0700)]
[lldb/CMake] Set the PYTHON_RPATH for the unit tests
The API and Python script interpreter unit tests also link against
Python and therefore need to set the RPATH when applicable.
Scott Linder [Thu, 30 Apr 2020 18:00:09 +0000 (14:00 -0400)]
[AMDGPU] Update DWARF proposal encodings
Update the tentative encodings to avoid a conflict with a GNU extension.
Differential Revision: https://reviews.llvm.org/D70523
Sam McCall [Wed, 29 Apr 2020 18:10:50 +0000 (20:10 +0200)]
[clangd] Move inserted include from detail -> documentation.
Summary: Many clients try to display all the detail inline, with poor results.
See https://github.com/clangd/clangd/issues/284
Reviewers: hokein
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79106
Jonas Devlieghere [Thu, 30 Apr 2020 17:54:03 +0000 (10:54 -0700)]
[lldb/CMake] Don't make PYTHON_RPATH a required variable.
Jonas Devlieghere [Thu, 30 Apr 2020 17:51:40 +0000 (10:51 -0700)]
[debuginfo-tests] Update Python variable in lit.site.cfg.py
Nicolas Vasilache [Thu, 30 Apr 2020 13:44:55 +0000 (09:44 -0400)]
[mlir][Linalg] Add support to lower named ops to loops.
This revision adds support to allow named ops to lower to loops.
Linalg.batch_matmul successfully lowers to loops and to LLVM.
In the process, this test also activates linalg to affine loops.
However padded convolutions to not lower to affine.load atm so this revision overrides the type of underlying load / store operation.
Differential Revision: https://reviews.llvm.org/D79135
Jonas Devlieghere [Thu, 30 Apr 2020 17:34:03 +0000 (10:34 -0700)]
[lldb] fix RPATH when linking against Python3.framework
The install name for the Python 3 framework in Xcode is relative to
the framework's location and not the dylib itself.
@rpath/Python3.framework/Versions/3.x/Python3
This means that we need to compute the path to the Python3.framework
and use that as the RPATH instead of the usual dylib's directory.
Raphael Isemann [Thu, 30 Apr 2020 17:28:23 +0000 (19:28 +0200)]
[lldb] Re-add deleted RUN line in module-ownership.mm
This was deleted in
681466f5e6412350a0b066791450e72325c2c074 by accident.
Louis Dionne [Thu, 30 Apr 2020 16:18:37 +0000 (12:18 -0400)]
[libc++] Use the internal Lit shell parser to run config tests
This makes the config tests portable across platforms.
Nikita Popov [Thu, 30 Apr 2020 16:49:01 +0000 (18:49 +0200)]
[Bitcode] Make attribute test more robust (NFC)
The ID of the { nobuiltin } attribute set increases every time
a new test is added to this file. Store it in a variable instead.
Sam McCall [Thu, 30 Apr 2020 08:49:32 +0000 (10:49 +0200)]
[clangd] Render code complete documentation as plaintext/markdown.
Summary:
Structure is parsed from the raw comment using the existing heuristics used
for hover.
Reviewers: kadircet
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79157
Louis Dionne [Thu, 30 Apr 2020 16:55:01 +0000 (12:55 -0400)]
[libc++] Do not allow customizing the arch in the trunk and back-deployment scripts
We should eventually be able to pass an arbitrary target triple, however
for the time being, just being able to pass 64/32 is not really that
useful.
Lucy Fox [Thu, 30 Apr 2020 16:47:19 +0000 (09:47 -0700)]
[MLIR] Modify Partial op conversion mode to optionally track all non-legalizable operations.
There are three op conversion modes: Partial, Full, and Analysis. This change modifies the Partial mode to optionally take a set of non-legalizable ops. If this parameter is specified, all ops that are not legalizable (i.e. would cause full conversion to fail) are tracked throughout the partial legalization.
Differential Revision: https://reviews.llvm.org/D78788
Nikita Popov [Thu, 30 Apr 2020 16:45:04 +0000 (18:45 +0200)]
[BitcodeReader] Simplify raw attribute handling (NFC)
Every new attribute we add from now on will not be supported in the
raw format, because we ran out of space. Don't bother listing each
affected attribute twice.
Arthur Eubanks [Thu, 30 Apr 2020 01:50:17 +0000 (18:50 -0700)]
[NFC] Rename *ByValOrInalloca* to *PassPointeeByValue*
Summary: In preparation for preallocated.
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79152
Aaron Smith [Thu, 30 Apr 2020 16:38:19 +0000 (09:38 -0700)]
[Windows SEH] Fix abnormal-exits in _try
Summary:
Per Windows SEH Spec, except _leave, all other early exits of a _try (goto/return/continue/break) are considered abnormal exits. In those cases, the first parameter passes to its _finally funclet should be TRUE to indicate an abnormal-termination.
One way to implement abnormal exits in _try is to invoke Windows runtime _local_unwind() (MSVC approach) that will invoke _dtor funclet where abnormal-termination flag is always TRUE when calling _finally. Obviously this approach is less optimal and is complicated to implement in Clang.
Clang today has a NormalCleanupDestSlot mechanism to dispatch multiple exits at the end of _try. Since _leave (or try-end fall-through) is always Indexed with 0 in that NormalCleanupDestSlot, this fix takes the advantage of that mechanism and just passes NormalCleanupDest ID as 1st Arg to _finally.
Reviewers: rnk, eli.friedman, JosephTremoulet, asmith, efriedma
Reviewed By: efriedma
Subscribers: efriedma, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77936
Fangrui Song [Sat, 18 Apr 2020 06:08:11 +0000 (23:08 -0700)]
[ELF][PPC64] Suppress toc-indirect to toc-relative relaxation if R_PPC64_TOC16_LO is seen
The current implementation assumes that R_PPC64_TOC16_HA is always followed
by R_PPC64_TOC16_LO_DS. This can break with R_PPC64_TOC16_LO:
// Load the address of the TOC entry, instead of the value stored at that address
addis 3, 2, .LC0@tloc@ha # R_PPC64_TOC16_HA
addi 3, 3, .LC0@tloc@l # R_PPC64_TOC16_LO
blr
which is used by boringssl's util/fipstools/delocate/delocate.go
https://github.com/google/boringssl/blob/master/crypto/fipsmodule/FIPS.md has some documentation.
In short, this tool converts an assembly file to avoid any potential relocations.
The distance to an input .toc is not a constant after linking, so it cannot use an `addis;ld` pair.
Instead, it jumps to a stub which loads the TOC entry address with `addis;addi`.
This patch checks the presence of R_PPC64_TOC16_LO and suppresses
toc-indirect to toc-relative relaxation if R_PPC64_TOC16_LO is seen.
This approach is conservative and loses some relaxation opportunities but is easy to implement.
addis 3, 2, .LC0@toc@ha # no relaxation
addi 3, 3, .LC0@toc@l # no relaxation
li 9, 0
addis 4, 2, .LC0@toc@ha # can relax but suppressed
ld 4, .LC0@toc@l(4) # can relax but suppressed
Also note that interleaved R_PPC64_TOC16_HA and R_PPC64_TOC16_LO_DS is
possible and this patch accounts for that.
addis 3, 2, .LC1@toc@ha # can relax
addis 4, 2, .LC2@toc@ha # can relax
ld 3, .LC1@toc@l(3) # can relax
ld 4, .LC2@toc@l(4) # can relax
Reviewed By: #powerpc, sfertile
Differential Revision: https://reviews.llvm.org/D78431
Hans Wennborg [Thu, 30 Apr 2020 13:03:22 +0000 (15:03 +0200)]
lit googletest.py: Don't raise StopIteration in generator
The intention here seems to be to end the generator function, but with
modern Python, raising StopIteration causes a runtime error
(https://www.python.org/dev/peps/pep-0479/).
Differential revision: https://reviews.llvm.org/D79169
Aaron Smith [Thu, 30 Apr 2020 16:11:54 +0000 (09:11 -0700)]
[clang-format] Fix Microsoft style for enums
Summary:
Before this change enums were formatted incorrectly for the Microsoft style.
[C++ Example]
enum {
one,
two
} three, four;
[Incorrectly Formatted]
enum
{
one,
two
} three,
four;
[Correct Format with Patch]
enum
{
one,
two
} three, four;
Reviewers: jbcoe, MyDeveloperDay, rnk
Reviewed By: MyDeveloperDay
Subscribers: cfe-commits
Tags: #clang, #clang-format
Differential Revision: https://reviews.llvm.org/D78982
Louis Dionne [Mon, 27 Apr 2020 16:49:01 +0000 (12:49 -0400)]
[libc++] Make the verify-support feature implicit
Tests that require support for Clang-verify are already marked as such
explicitly by their extension, which is .verify.cpp. Requiring the use
of an explicit Lit feature is, after thought, not really helpful.
This is a change in design: we have been bitten in the past by tests not
being enabled when we thought they were. However, the issue was mostly
with file extensions being ignored. The fix for that is not to blindly
require explicit features all the time, but instead to report all files
that are in the suite but that don't match any known test format. This
can be implemented in a follow-up patch.
Fangrui Song [Thu, 30 Apr 2020 04:56:51 +0000 (21:56 -0700)]
[obj2yaml][test] Move tests to binary format specific subdirectories
Similar to D70264.
Create COFF/, ELF/, MachO/, Minidump/, XCOFF/, and move tests there.
Reviewed By: grimar, jhenderson, labath
Differential Revision: https://reviews.llvm.org/D79154
Kirill Naumov [Thu, 30 Apr 2020 15:09:51 +0000 (15:09 +0000)]
[InlineCost] Addressing a very strict assert check in CostAnnotationWriter::emitInstructionAnnot
The assert checks that every instruction must be annotated by this point while it is not
necessary. If the inlining process was interrupted because the threshold was reached, the rest
of the instructions would not be annotated which triggers the assert.
The added test shows the situation in which it can happen.
This is a recommit as the original commit fail due to the absence of REQUIRES: assert in the test.
Reviewed By: mtrofin
Differential Revision: https://reviews.llvm.org/D79107
Simon Pilgrim [Thu, 30 Apr 2020 15:23:53 +0000 (16:23 +0100)]
[X86] Extend combine-bitselect tests
Add AVX512VL tests and v2i64/v8i64 mask broadcast tests that match the existing v4i64 versions
Simon Pilgrim [Thu, 30 Apr 2020 14:10:25 +0000 (15:10 +0100)]
[X86] combineANDXORWithAllOnesIntoANDNP - add BROADCAST handling
Fold BROADCAST(NOT(Y)) -> BROADCAST(Y) as part of finding a NOT inversion.
Jann Horn [Thu, 30 Apr 2020 09:05:01 +0000 (11:05 +0200)]
[AddressSanitizer] Instrument byval call arguments
Summary:
In the LLVM IR, "call" instructions read memory for each byval operand.
For example:
```
$ cat blah.c
struct foo { void *a, *b, *c; };
struct bar { struct foo foo; };
void func1(const struct foo);
void func2(struct bar *bar) { func1(bar->foo); }
$ [...]/bin/clang -S -flto -c blah.c -O2 ; cat blah.s
[...]
define dso_local void @func2(%struct.bar* %bar) local_unnamed_addr #0 {
entry:
%foo = getelementptr inbounds %struct.bar, %struct.bar* %bar, i64 0, i32 0
tail call void @func1(%struct.foo* byval(%struct.foo) align 8 %foo) #2
ret void
}
[...]
$ [...]/bin/clang -S -c blah.c -O2 ; cat blah.s
[...]
func2: # @func2
[...]
subq $24, %rsp
[...]
movq 16(%rdi), %rax
movq %rax, 16(%rsp)
movups (%rdi), %xmm0
movups %xmm0, (%rsp)
callq func1
addq $24, %rsp
[...]
retq
```
Let ASAN instrument these hidden memory accesses.
This is patch 4/4 of a patch series:
https://reviews.llvm.org/D77616 [PATCH 1/4] [AddressSanitizer] Refactor ClDebug{Min,Max} handling
https://reviews.llvm.org/D77617 [PATCH 2/4] [AddressSanitizer] Split out memory intrinsic handling
https://reviews.llvm.org/D77618 [PATCH 3/4] [AddressSanitizer] Refactor: Permit >1 interesting operands per instruction
https://reviews.llvm.org/D77619 [PATCH 4/4] [AddressSanitizer] Instrument byval call arguments
Reviewers: kcc, glider
Reviewed By: glider
Subscribers: hiraditya, dexonsmith, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77619
Jann Horn [Thu, 30 Apr 2020 09:05:00 +0000 (11:05 +0200)]
[AddressSanitizer] Refactor: Permit >1 interesting operands per instruction
Summary:
Refactor getInterestingMemoryOperands() so that information about the
pointer operand is returned through an array of structures instead of
passing each piece of information separately by-value.
This is in preparation for returning information about multiple pointer
operands from a single instruction.
A side effect is that, instead of repeatedly generating the same
information through isInterestingMemoryAccess(), it is now simply collected
once and then passed around; that's probably more efficient.
HWAddressSanitizer has a bunch of copypasted code from AddressSanitizer,
so these changes have to be duplicated.
This is patch 3/4 of a patch series:
https://reviews.llvm.org/D77616 [PATCH 1/4] [AddressSanitizer] Refactor ClDebug{Min,Max} handling
https://reviews.llvm.org/D77617 [PATCH 2/4] [AddressSanitizer] Split out memory intrinsic handling
https://reviews.llvm.org/D77618 [PATCH 3/4] [AddressSanitizer] Refactor: Permit >1 interesting operands per instruction
https://reviews.llvm.org/D77619 [PATCH 4/4] [AddressSanitizer] Instrument byval call arguments
[glider: renamed llvm::InterestingMemoryOperand::Type to OpType to fix
GCC compilation]
Reviewers: kcc, glider
Reviewed By: glider
Subscribers: hiraditya, jfb, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77618
Jann Horn [Thu, 30 Apr 2020 09:04:55 +0000 (11:04 +0200)]
[AddressSanitizer] Split out memory intrinsic handling
Summary:
In both AddressSanitizer and HWAddressSanitizer, we first collect
instructions whose operands should be instrumented and memory intrinsics,
then instrument them. Both during collection and when inserting
instrumentation, they are handled separately.
Collect them separately and instrument them separately. This is a bit
more straightforward, and prepares for collecting operands instead of
instructions in a future patch.
This is patch 2/4 of a patch series:
https://reviews.llvm.org/D77616 [PATCH 1/4] [AddressSanitizer] Refactor ClDebug{Min,Max} handling
https://reviews.llvm.org/D77617 [PATCH 2/4] [AddressSanitizer] Split out memory intrinsic handling
https://reviews.llvm.org/D77618 [PATCH 3/4] [AddressSanitizer] Refactor: Permit >1 interesting operands per instruction
https://reviews.llvm.org/D77619 [PATCH 4/4] [AddressSanitizer] Instrument byval call arguments
Reviewers: kcc, glider
Reviewed By: glider
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77617
Jann Horn [Thu, 30 Apr 2020 09:04:51 +0000 (11:04 +0200)]
[AddressSanitizer] Refactor ClDebug{Min,Max} handling
Summary:
A following commit will split the loop over ToInstrument into two.
To avoid having to duplicate the condition for suppressing instrumentation
sites based on ClDebug{Min,Max}, refactor it out into a new function.
While we're at it, we can also avoid the indirection through
NumInstrumented for setting FunctionModified.
This is patch 1/4 of a patch series:
https://reviews.llvm.org/D77616 [PATCH 1/4] [AddressSanitizer] Refactor ClDebug{Min,Max} handling
https://reviews.llvm.org/D77617 [PATCH 2/4] [AddressSanitizer] Split out memory intrinsic handling
https://reviews.llvm.org/D77618 [PATCH 3/4] [AddressSanitizer] Refactor: Permit >1 interesting operands per instruction
https://reviews.llvm.org/D77619 [PATCH 4/4] [AddressSanitizer] Instrument byval call arguments
Reviewers: kcc, glider
Reviewed By: glider
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77616
Benjamin Kramer [Thu, 30 Apr 2020 15:00:18 +0000 (17:00 +0200)]
Sebastian Neubauer [Mon, 27 Apr 2020 09:40:31 +0000 (11:40 +0200)]
[CMake] Fix cross-compiling with LLVM as CMake subproject
When embedding LLVM as a CMake subproject, using cross-compiling does
not work at the moment. This also affects -DLLVM_OPTIMIZED_TABLEGEN=1,
which uses the same CMake infrastructure.
This patch replaces global CMake variables with the current version,
which allows cross-compilation to work in a subproject.
CMAKE_BINARY_DIR -> CMAKE_CURRENT_BINARY_DIR
CMAKE_SOURCE_DIR -> CMAKE_CURRENT_SOURCE_DIR
CMAKE_PROJECT_NAME -> PROJECT_NAME
Differential Revision: https://reviews.llvm.org/D78913
Kadir Cetinkaya [Thu, 30 Apr 2020 14:40:56 +0000 (16:40 +0200)]
[clangd] Get rid of move semantics to unbreak windows build bots
Raphael Isemann [Thu, 30 Apr 2020 14:33:54 +0000 (16:33 +0200)]
Include SmallVector.h in IPO.h to fix modules build [NFC]
This file currently doesn't compile under LLVM_ENABLE_MODULES as SmallVector
is used in this header but is never forward declared or included in any way.
Let's include SmallVector.h instead and get rid of the SmallVectorImpl fwd
declaration which is now no longer necessary.
Jay Foad [Thu, 30 Apr 2020 14:05:40 +0000 (15:05 +0100)]
[AMDGPU] Use int and unsigned instead of other 32-bit integer types. NFC.
Alexander Potapenko [Thu, 30 Apr 2020 14:15:43 +0000 (16:15 +0200)]
Revert an accidental commit of four AddressSanitizer refactor CLs
I couldn't make arc land the changes properly, for some reason they all got
squashed. Reverting them now to land cleanly.
Summary: This reverts commit
cfb5f89b62f885a7a50b1dd49a0cb5532fc75176.
Reviewers: kcc, thejh
Subscribers:
diggerlin [Thu, 30 Apr 2020 13:53:41 +0000 (09:53 -0400)]
[AIX] emit .extern and .weak directive linkage
SUMMARY:
emit .extern and .weak directive linkage
Reviewers: hubert.reinterpretcast, Jason Liu
Subscribers: wuzish, nemanjai, hiraditya
Differential Revision: https://reviews.llvm.org/D76932
Simon Pilgrim [Thu, 30 Apr 2020 13:45:06 +0000 (14:45 +0100)]
[X86][SSE] Add bitselect tests where the mask is a broadcasted scalar
Shows issue that the IsNot() test can't see through shuffles/broadcasts
Kadir Cetinkaya [Thu, 30 Apr 2020 13:37:07 +0000 (15:37 +0200)]
[clangd] Second attempt at fixing VS2019 build bots
Jann Horn [Thu, 30 Apr 2020 09:05:01 +0000 (11:05 +0200)]
[AddressSanitizer] Refactor ClDebug{Min,Max} handling
Summary:
A following commit will split the loop over ToInstrument into two.
To avoid having to duplicate the condition for suppressing instrumentation
sites based on ClDebug{Min,Max}, refactor it out into a new function.
While we're at it, we can also avoid the indirection through
NumInstrumented for setting FunctionModified.
This is patch 1/4 of a patch series:
https://reviews.llvm.org/D77616 [PATCH 1/4] [AddressSanitizer] Refactor ClDebug{Min,Max} handling
https://reviews.llvm.org/D77617 [PATCH 2/4] [AddressSanitizer] Split out memory intrinsic handling
https://reviews.llvm.org/D77618 [PATCH 3/4] [AddressSanitizer] Refactor: Permit >1 interesting operands per instruction
https://reviews.llvm.org/D77619 [PATCH 4/4] [AddressSanitizer] Instrument byval call arguments
Reviewers: kcc, glider
Reviewed By: glider
Subscribers: jfb, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77616
Lei Zhang [Mon, 27 Apr 2020 16:31:29 +0000 (12:31 -0400)]
[mlir] Use memory effect to detecting allocation
This commit marks AllocLikeOp as MemAlloc in StandardOps.
Also in Linalg dependency analysis use memory effect to detect
allocation. This allows the dependency analysis to be more
general and recognize other allocation-like operations.
Differential Revision: https://reviews.llvm.org/D78705
Sanjay Patel [Thu, 30 Apr 2020 13:11:29 +0000 (09:11 -0400)]
[InstCombine] add tests for bitcast+inselt; NFC
Sanjay Patel [Thu, 30 Apr 2020 12:38:36 +0000 (08:38 -0400)]
[InstCombine] update auto-generated test checks; NFC
jasonliu [Thu, 30 Apr 2020 02:50:57 +0000 (02:50 +0000)]
[NFC][clang] Replace raw new/delete with unique_ptr to store ABIInfo in TargetCodeGenInfo
Use unique_ptr to manage the lifetime of ABIInfo member inside TargetCodeGenInfo.
Reviewed By: hubert.reinterpretcast
Differential Revision: https://reviews.llvm.org/D79033
Kadir Cetinkaya [Thu, 30 Apr 2020 11:57:42 +0000 (13:57 +0200)]
[clangd] Fix VS2019 build bots too
Simon Pilgrim [Thu, 30 Apr 2020 11:53:05 +0000 (12:53 +0100)]
[DAGCombine] Move the remaining X86 funnel shift patterns to DAGCombine
X86 matches several 'shift+xor' funnel shift patterns:
fold (or (srl (srl x1, 1), (xor y, 31)), (shl x0, y)) -> (fshl x0, x1, y)
fold (or (shl (shl x0, 1), (xor y, 31)), (srl x1, y)) -> (fshr x0, x1, y)
fold (or (shl (add x0, x0), (xor y, 31)), (srl x1, y)) -> (fshr x0, x1, y)
These patterns are also what we end up with the proposed expansion changes in D77301.
This patch moves these to DAGCombine's generic MatchFunnelPosNeg.
All existing X86 test cases still pass, and we just have a small codegen change in pr32282.ll.
Reviewed By: @spatel
Differential Revision: https://reviews.llvm.org/D78935
Simon Pilgrim [Thu, 30 Apr 2020 10:22:01 +0000 (11:22 +0100)]
[DAG] Add TODO comment regarding ADD(X,X) -> SHL(X,1) canonicalization
As discussed on D78935
Sanjay Patel [Wed, 29 Apr 2020 18:45:50 +0000 (14:45 -0400)]
[InstCombine] add tests for FP->int->FP->FP casting; NFC
Jay Foad [Thu, 30 Apr 2020 10:40:57 +0000 (11:40 +0100)]
Fix silly mistake in
31c09d03a1f [AMDGPU] Remove WaitcntBrackets::MixedPendingEvents[]. NFC.
David Spickett [Thu, 23 Apr 2020 13:40:57 +0000 (14:40 +0100)]
[globalopt] Don't emit DWARF fragments for members
of a struct that cover the whole struct
This can happen when the rest of the
members of are zero length. Following
the same pattern applied to the SROA
pass in:
d7f6f1636d53c3e2faf55cdf20fbb44a1a149df1
Fixes: https://bugs.llvm.org/show_bug.cgi?id=45335
Differential Revision: https://reviews.llvm.org/D78720
Sam Elliott [Thu, 30 Apr 2020 09:38:35 +0000 (10:38 +0100)]
[RISCV][NFC] Remove Duplicated F Extension Patterns
Cullen Rhodes [Tue, 28 Apr 2020 10:55:07 +0000 (10:55 +0000)]
[AArch64][SVE] Remove unused FP reduction intrinsic definitions
Summary: FP reductions no longer use these intrinsics since D78723.
Reviewers: efriedma, sdesmalen
Reviewed By: efriedma, sdesmalen
Differential Revision: https://reviews.llvm.org/D79010
Cullen Rhodes [Tue, 21 Apr 2020 10:44:46 +0000 (10:44 +0000)]
[AArch64][SVE] Custom lowering of floating-point reductions
Summary:
This patch implements custom floating-point reduction ISD nodes that
have vector results, which are used to lower the following intrinsics:
* llvm.aarch64.sve.fadda
* llvm.aarch64.sve.faddv
* llvm.aarch64.sve.fmaxv
* llvm.aarch64.sve.fmaxnmv
* llvm.aarch64.sve.fminv
* llvm.aarch64.sve.fminnmv
SVE reduction instructions keep their result within a vector register,
with all other bits set to zero.
Changes in this patch were implemented by Paul Walker and Sander de
Smalen.
Reviewers: sdesmalen, efriedma, rengolin
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D78723
David Sherwood [Fri, 24 Apr 2020 06:55:53 +0000 (07:55 +0100)]
[CodeGen] Add support for inserting elements into scalable vectors
Summary:
This patch tries to ensure that we do something sensible when
generating code for the ISD::INSERT_VECTOR_ELT DAG node when operating
on scalable vectors. Previously we always returned 'undef' when
inserting an element into an out-of-bounds lane index, whereas now
we only do this for fixed length vectors. For scalable vectors it
is assumed that the backend will do the right thing in the same way
that we have to deal with variable lane indices.
In this patch I have permitted a few basic combinations for scalable
vector types where it makes sense, but in general avoided most cases
for now as they currently require the use of BUILD_VECTOR nodes.
This patch includes tests for all scalable vector types when inserting
into lane 0, but I've only included one or two vector types for other
cases such as variable lane inserts.
Differential Revision: https://reviews.llvm.org/D78992
Kadir Cetinkaya [Thu, 30 Apr 2020 10:13:12 +0000 (12:13 +0200)]
[clangd][Hover] Get rid of unused private field in Paragraph
mydeveloperday [Thu, 30 Apr 2020 10:11:34 +0000 (11:11 +0100)]
[clang-format] Fix a bug causing BeforeLambdaBody to affect brace initialiser formatting
Summary: The condition added with the new setting checked whether the character was an l-brace, not specifically a lambda l-brace, when deciding whether it was possible to break before it or not. This caused the l-brace of some initialiser lists to break onto the next line with the first argument of the initialiser list when the setting was enabled.
Reviewed By: MyDeveloperDay, Wawha
Patch By: duncan-llvm
Subscribers: cfe-commits
Tags: #clang, #clang-format
Differential Revision: https://reviews.llvm.org/D79022
mydeveloperday [Thu, 30 Apr 2020 10:05:02 +0000 (11:05 +0100)]
[clang-format] Correct the AfterControlStatement configuration option output style
Summary:
Due to the order in which the enum cases were defined the old options which were retained for backwards compatibility were being preferred over the new options when printing with the --dump-config option.
Reviewers: MyDeveloperDay
Reviewed By: MyDeveloperDay
Patch By: duncan-llvm
Subscribers: cfe-commits
Tags: #clang, #clang-format
Differential Revision: https://reviews.llvm.org/D79020
mydeveloperday [Thu, 30 Apr 2020 10:02:42 +0000 (11:02 +0100)]
[clang-format] Fix lambda with ellipsis in return type
Summary:
BTW my actual code that hit this issue is like
```
[a, b = std::move(b)](auto &&... c) mutable -> std::invoke_result_t<decltype(b), decltype(c)...> { /* omitted */ }
```
where this explicit return type is required for SFINAE.
Reviewed By: krasimir, MyDeveloperDay
Patch By: johnchen902
Subscribers: krasimir, dexonsmith, cfe-commits
Tags: #clang-format, #clang
Differential Revision: https://reviews.llvm.org/D78694
mydeveloperday [Thu, 30 Apr 2020 09:55:59 +0000 (10:55 +0100)]
[clang-format] [PR45357] Fix issue found with operator spacing
Summary:
This is a tentative fix for https://bugs.llvm.org/show_bug.cgi?id=45357
Spaces seem to be introduced between * and * due to changes brought in for {D69573}
Reviewers: sylvestre.ledru, mitchell-stellar, sammccall, Abpostelnicu, krasimir, jbcoe
Reviewed By: Abpostelnicu
Subscribers: tstellar, hans, Abpostelnicu, cfe-commits
Tags: #clang, #clang-format
Differential Revision: https://reviews.llvm.org/D78879
mydeveloperday [Thu, 30 Apr 2020 09:47:23 +0000 (10:47 +0100)]
[clang-format] NFC Correct clang-format headers file so documentation can be once again autogenerated
Summary:
`ClangFormatStyleOptions.rst` should ALWAYS be autogenerated from Format.h using `clang/docs/tools/dump_format_style.py` if not its liable to get removed leaving options undocumented.
This revision reworks the documentation for {D73354} {D73768} to ensure we can continue to regenerated
Fix other minor changes that ensure the documentation remains consistent (Format.h obviously got re clang-formatted after the rst had been regenerated previously)
Reviewed By: krasimir
Subscribers: cfe-commits
Tags: #clang, #clang-format
Differential Revision: https://reviews.llvm.org/D79095
James Henderson [Thu, 30 Apr 2020 09:40:11 +0000 (10:40 +0100)]
[docs][llvm-cxxfilt] Fix indentation in rst file
This makes it consistent throughout the options, although the end result
is unchanged.
serge-sans-paille [Thu, 30 Apr 2020 08:14:37 +0000 (10:14 +0200)]
Fix spurious warning in ExtensionDependencies.inc [nfc]
Tres Popp [Wed, 29 Apr 2020 13:45:50 +0000 (15:45 +0200)]
[MLIR] Add isa<> support for Dialects.
Summary:
The purpose of this is to aid in having code behave differently on
Operations based on their Dialect without caring about the specific
Op. Additionally this is consistent with most other types supporting
isa<> and dyn_cast<>.
A Dialect matches isa<> based only on its namespace and relies on each
namespace being unique.
Differential Revision: https://reviews.llvm.org/D79088
Evgeniy Brevnov [Wed, 29 Apr 2020 11:42:36 +0000 (18:42 +0700)]
[BPI][NFC] IRCE shoud qequest BPI through analysis manager.
Summary: There is no need to create BPI explicitly. It should be requested through AM in a normal way.
Reviewers: skatkov
Reviewed By: skatkov
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79080
Kadir Cetinkaya [Thu, 30 Apr 2020 08:24:39 +0000 (10:24 +0200)]
[clangd] Fix windows build bots without rvalue refs
Alexey Lapshin [Wed, 29 Apr 2020 17:00:31 +0000 (20:00 +0300)]
[Debuginfo][NFC] findRecursively: Replace std::vector by SmallVector
Summary:
Change std::vector to SmallVector to prevent re-allocations and to
have small pre-allocated storage.
Reviewers: clayborg, dblaikie
Differential Revision: https://reviews.llvm.org/D79123
Jay Foad [Tue, 28 Apr 2020 07:54:19 +0000 (08:54 +0100)]
[AMDGPU] Simplify loops in SIInsertWaitcnts::generateWaitcntInstBefore
The loops over use operands and def operands were mostly identical.
Combine them, and likewise for load memoperands and store memoperands.
NFC.
Jay Foad [Mon, 27 Apr 2020 19:14:52 +0000 (20:14 +0100)]
[AMDGPU] Remove Def argument from WaitcntBrackets::getRegInterval. NFC.
It's cleaner to check this in the callers instead.
Kadir Cetinkaya [Wed, 29 Apr 2020 09:21:14 +0000 (11:21 +0200)]
[clangd] Make use of URIs in FileShardedIndex
Summary:
This makes FileShardedIndex more robust and gets rid of the need for a
URIToFileCache, as it is done by the callers now and it is only once per file,
rather than once per symbol.
Reviewers: sammccall
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79079
Fangrui Song [Thu, 30 Apr 2020 06:17:14 +0000 (23:17 -0700)]
[MC] Move MCInstrAnalysis::evaluateBranch to X86MCInstrAnalysis::evaluateBranch
The generic implementation is actually specific to x86. It assumes the
offset is relative to the end of the instruction and the immediate is
not scaled (which is false on most RISC).
Evgeniy Brevnov [Wed, 29 Apr 2020 07:08:01 +0000 (14:08 +0700)]
[BPI] Incorrect probability reported in case of mulptiple edges.
Summary:
By design 'BranchProbabilityInfo:: getEdgeProbability(const BasicBlock *Src, const BasicBlock *Dst) const' should return sum of probabilities over all edges from Src to Dst. Current implementation is buggy and returns 1/num_of_successors if probabilities are not explicitly set.
Note current implementation of BPI printing has an issue as well and annotates each edge with sum of probabilities over all ages from one basic block to another. That's why 30% probability reported (instead of 10%) in the lit test. This is not urgent issue since only printing is affected.
Note also current implementation assumes that either all or none edges have probabilities set. This is not the only place which uses such assumption. At least we should assert that in verifier. In addition we can think on a more robust API of BPI which would prevent situations.
Reviewers: skatkov, yrouban, taewookoh
Reviewed By: skatkov
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79071
Evgeniy Brevnov [Tue, 28 Apr 2020 09:31:20 +0000 (16:31 +0700)]
[BPI][NFC] Reuse post dominantor tree from analysis manager when available
Summary: Currenlty BPI unconditionally creates post dominator tree each time. While this is not incorrect we can save compile time by reusing existing post dominator tree (when it's valid) provided by analysis manager.
Reviewers: skatkov, taewookoh, yrouban
Reviewed By: skatkov
Subscribers: hiraditya, steven_wu, dexonsmith, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D78987
Julian Lettner [Tue, 14 Apr 2020 22:55:38 +0000 (15:55 -0700)]
[lit] Provide extension API for custom result categories
The lnt test suite defines custom result codes [1]. Support those via
an extension API instead of "by accident", which should offer the
advantage of properly handling them when we print test results.
[1] https://reviews.llvm.org/D77986
Reviewed By: Meinersbur
Differential Revision: https://reviews.llvm.org/D78164
Arthur Eubanks [Wed, 29 Apr 2020 20:50:19 +0000 (13:50 -0700)]
Make wrong preallocated arg count verifier error clearer
Reviewers: rnk
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79128
Nicolas Vasilache [Thu, 30 Apr 2020 01:12:10 +0000 (21:12 -0400)]
[mlir][Vector] Provide progressive lowering of masked n-D vector transfers
This revision allows masked vector transfers with m-D buffers and n-D vectors to
progressively lower to m-D buffer and 1-D vector transfers.
For a vector.transfer_read, assuming a `memref<(leading_dims) x (major_dims) x (minor_dims) x type>` and a `vector<(minor_dims) x type>` are involved in the transfer, this generates pseudo-IR resembling:
```
if (any_of(%ivs_major + %offsets, <, major_dims)) {
%v = vector_transfer_read(
{%offsets_leading, %ivs_major + %offsets_major, %offsets_minor},
%ivs_minor):
memref<(leading_dims) x (major_dims) x (minor_dims) x type>,
vector<(minor_dims) x type>;
} else {
%v = splat(vector<(minor_dims) x type>, %fill)
}
```
Differential Revision: https://reviews.llvm.org/D79062
Fangrui Song [Tue, 28 Apr 2020 02:30:09 +0000 (19:30 -0700)]
[ELF] Add --print-archive-stats=
gold has an option --print-symbol-counts= which prints:
// For each archive
archive $archive $members $fetched_members
// For each object file
symbols $object $defined_symbols $used_defined_symbols
In most cases, `$defined_symbols = $used_defined_symbols` unless weak
symbols are present. Strangely `$used_defined_symbols` includes symbols defined relative to --gc-sections discarded sections.
The `symbols` lines do not appear to be useful.
`archive` lines are useful: `$fetched_members=0` lines correspond to
unused archives. The information can be used to trim dependencies.
This patch implements --print-archive-stats= which prints the number of
members and the number of fetched members for each archive.
Reviewed By: grimar
Differential Revision: https://reviews.llvm.org/D78983
Fangrui Song [Thu, 30 Apr 2020 00:31:35 +0000 (17:31 -0700)]
[ELF] Add --rosegment to complement --no-rosegment
This option can cancel --no-rosegment and it just seems right to have
a corresponding positive option for a --no-* negative option.
Anecdotally, gold had --rosegment but did not have --no-rosegment.
I added --no-rosegment (https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=
9a6c68caa9543e09b064b7ac7c2b658f277bc19c) for binutils>=2.35
Dan Albert [Thu, 30 Apr 2020 00:11:01 +0000 (17:11 -0700)]
Reset another globalMemCounter.
Evgenii Stepanov [Wed, 29 Apr 2020 21:41:48 +0000 (14:41 -0700)]
[scudo] Initialize the allocator in setTrackAllocationStacks.
Summary:
If this is called before the malloc call in a thread (or in the whole
program), the lazy initialization of the allocation can overwrite
Options.
Reviewers: pcc, cryptoad
Subscribers: #sanitizers, llvm-commits
Tags: #sanitizers
Differential Revision: https://reviews.llvm.org/D79130
River Riddle [Wed, 29 Apr 2020 23:09:43 +0000 (16:09 -0700)]
[mlir] Simplify BranchOpInterface by using MutableOperandRange
This range allows for performing many different operations on successor operands, including erasing/adding/setting. This removes the need for the explicit canEraseSuccessorOperand and eraseSuccessorOperand methods.
Differential Revision: https://reviews.llvm.org/D79077
River Riddle [Wed, 29 Apr 2020 23:09:30 +0000 (16:09 -0700)]
[mlir][DeclareOpInterfaceMethods] Allow specifying a set of methods to force declaration generation for.
Currently a declaration won't be generated if the method has a default implementation. Meaning that operations that wan't to override the default have to explicitly declare the method in the extraClassDeclarations. This revision adds an optional list parameter to DeclareOpInterfaceMethods to allow for specifying a set of methods that should always have the declarations generated, even if there is a default.
Differential Revision: https://reviews.llvm.org/D79030
River Riddle [Wed, 29 Apr 2020 23:09:20 +0000 (16:09 -0700)]
[mlir] Move the operation equivalence out of CSE and into OperationSupport
This provides a general hash and comparison for checking if two operations are equivalent. This revision also optimizes the handling of result types to take advantage of how result types are stored on the operation.
Differential Revision: https://reviews.llvm.org/D79029
River Riddle [Wed, 29 Apr 2020 23:09:11 +0000 (16:09 -0700)]
[mlir] Add a new MutableOperandRange class for adding/remove operands
This class allows for mutating an operand range in-place, and provides vector like API for adding/erasing/setting. ODS now uses this class to generate mutable wrappers for named operands, with the name `MutableOperandRange <operand-name>Mutable()`
Differential Revision: https://reviews.llvm.org/D78892
Puyan Lotfi [Wed, 29 Apr 2020 23:43:17 +0000 (19:43 -0400)]
[test][MachineOutliner] REQUIRES: asserts
This new test checks some of the debug output to ensure what iteration
the outliner reached a fixed point. For now I am making it REQUIRES:
asserts so that it wont break any bots that have asserts disabled.
LLVM GN Syncbot [Wed, 29 Apr 2020 22:52:50 +0000 (22:52 +0000)]
[gn build] Port
9854edd817c
Jez Ng [Wed, 29 Apr 2020 22:42:45 +0000 (15:42 -0700)]
[lld-macho] Support X86_64_RELOC_BRANCH
Relatively straightforward diff, to set the stage for calling functions
in dylibs.
Differential Revision: https://reviews.llvm.org/D78269
Jez Ng [Wed, 29 Apr 2020 22:42:40 +0000 (15:42 -0700)]
[lld-macho] Have Symbol::getVA() return a non-relative virtual address
Currently, getVA() returns a virtual address with the assumption that
the ImageBase is zero. As I understand, this is what lld-ELF is doing.
However, under our current design, it seems like an awkward setup --
I'm finding that I have to add and subtract ImageBase in several places
to make things work out.
As such, I think it's simpler to have getVA() return a non-relative VA,
but I'm not sure if I'm missing something. Would love to hear more from
folks familiar with lld-ELF.
Differential Revision: https://reviews.llvm.org/D78168
Jez Ng [Wed, 29 Apr 2020 22:42:36 +0000 (15:42 -0700)]
[lld-macho] Support reading of universal binaries
Differential Revision: https://reviews.llvm.org/D77006
Jez Ng [Wed, 29 Apr 2020 22:42:32 +0000 (15:42 -0700)]
[lld-macho] Disable colors in errors when not printing to a pty
This makes for better tests (and is just the right thing to do)
Differential Revision: https://reviews.llvm.org/D79069
Jez Ng [Wed, 29 Apr 2020 22:42:19 +0000 (15:42 -0700)]
[lld-macho] Implement basic export trie
Build the trie by performing a three-way radix quicksort: We start by
sorting the strings by their first characters, then sort the strings
with the same first characters by their second characters, and so on
recursively. Each time the prefixes diverge, we add a node to the trie.
Thanks to @ruiu for the idea.
I used llvm-mc's radix quicksort implementation as a starting point. The
trie offset fixpoint code was taken from
MachONormalizedFileBinaryWriter.cpp.
Differential Revision: https://reviews.llvm.org/D76977
Mircea Trofin [Wed, 29 Apr 2020 16:01:38 +0000 (09:01 -0700)]
[llvm][NFC] Use CallBase explicitly instead of Instruction in FunctionComparator
Reviewers: dblaikie, craig.topper
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79098
Puyan Lotfi [Wed, 29 Apr 2020 07:33:47 +0000 (03:33 -0400)]
[NFCi] Iterative Outliner + clang-format refactoring.
Prior to D69446 I had done some NFC cleanup to make landing an iterative
outliner a cleaner more straight-forward patch. Since then, it seems that has
landed but I noticed some ways it could be cleaned up. Specifically:
1) doOutline was meant to be the re-runable function, but instead
runOnceOnModule was created that just calls doOutline.
2) In D69446 we discussed that the flag allowing the re-run of the
outliner should be a flag to tell how many additional times to run
the outliner again, not the total number of times. I don't think it
makes sense to introduce a flag, but print an error if the flag is
set to 0.
This is an NFCi, the i being that I get rid of the way that the
machine-outline-runs flag could be used to tell the outliner to not run
at all, and because I renamed the flag to '-machine-outliner-reruns'.
Differential Revision: https://reviews.llvm.org/D79070
Mircea Trofin [Wed, 29 Apr 2020 16:27:35 +0000 (09:27 -0700)]
[llvm][NFC] Inliner: rename call site variables.
Summary:
Renamed 'CS' to 'CB', and, in one case, to a more specific name to avoid
naming collision with outer scope (a maintainability/readability reason,
not correctness)
Also updated comments.
Reviewers: davidxl, dblaikie, jdoerfert
Subscribers: eraman, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79101
Tom Stellard [Wed, 29 Apr 2020 22:34:48 +0000 (15:34 -0700)]
libclc: Pass system libraries to the linker after llvm libraries
Summary:
The llvm libraries depend on the symbols in the system libaries, so
the system libraries need to be added after.
Reviewers: jvesely
Reviewed By: jvesely
Subscribers: mgorny, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D78535
River Riddle [Wed, 29 Apr 2020 22:08:25 +0000 (15:08 -0700)]
[mlir][Pass] Register a signal handler when generating crash reproducers.
The current implementation uses CrashRecoveryContext, but this only supports recovering in a certain number of cases. This revision adds a signal handler to support even more situations.
This revision was able to properly generate a reproducer for a segfault in the Inliner, that the current recovery couldn't.
Differential Revision: https://reviews.llvm.org/D78315
River Riddle [Wed, 29 Apr 2020 22:08:15 +0000 (15:08 -0700)]
[mlir][Pass] Add support for generating local crash reproducers
This revision adds a mode to the crash reproducer generator to attempt to generate a more local reproducer. This will attempt to generate a reproducer right before the offending pass that fails. This is useful for the majority of failures that are specific to a single pass, and situations where some passes in the pipeline are not registered with a specific tool.
Differential Revision: https://reviews.llvm.org/D78314
River Riddle [Wed, 29 Apr 2020 22:08:05 +0000 (15:08 -0700)]
[mlir][Pass][NFC] Merge OpToOpPassAdaptor and OpToOpPassAdaptorParallel
This moves the threading check to runOnOperation. This produces a much cleaner interface for the adaptor pass, and will allow for the ability to enable/disable threading in a much cleaner way in the future.
Differential Revision: https://reviews.llvm.org/D78313
Sam McCall [Sat, 4 Apr 2020 07:12:30 +0000 (09:12 +0200)]
[clangd] Parse `foo` in documentation comments and render as code.
Reviewers: kadircet
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77456