Qiu Chaofan [Tue, 12 May 2020 06:29:40 +0000 (14:29 +0800)]
[PowerPC] Exploit VSX neg, abs and nabs for f32
xsnegdp, xsabsdp and xsnabsdp can be used to operate on f32 operand.
This patch adds the missing patterns since we prefer VSX instructions
when available.
Reviewed By: steven.zhang
Differential Revision: https://reviews.llvm.org/D75344
Sam Parker [Wed, 13 May 2020 06:22:42 +0000 (07:22 +0100)]
[CostModel] Modify BasicTTI getCastInstrCost
Fix the assumption that all bitcasts of the same type sizes are free.
We now only assume that bitcasts between ints and ptrs of the same
size are free. This allows TTImpl to just call the concrete
implementation of getCastInstrCost.
Differential Revision: https://reviews.llvm.org/D78918
MaheshRavishankar [Wed, 13 May 2020 06:04:33 +0000 (23:04 -0700)]
[mlir][StandardToLLVM] Add SinOp to LLVM dialect and lowering of std.sin to this op.
Differential Revision: https://reviews.llvm.org/D79505
Qiu Chaofan [Wed, 13 May 2020 06:03:38 +0000 (14:03 +0800)]
[PowerPC] Respect SDNodeFlags in lowering SELECT_CC
Legalizer should respect both command-line options or SDNode-level
fast-math flags.
Also, this patch propagates other flags during custom simplifying.
Reviewed By: steven.zhang
Differential Revision: https://reviews.llvm.org/D79074
MaheshRavishankar [Wed, 13 May 2020 05:50:44 +0000 (22:50 -0700)]
[mlir][Linalg] Add folders and canonicalizers for
linalg.reshape/linalg.tensor_reshape operations.
Differential Revision: https://reviews.llvm.org/D79765
MaheshRavishankar [Wed, 13 May 2020 05:50:35 +0000 (22:50 -0700)]
[mlir][Linalg] Allow reshapes to collapse to a zero-rank tensor.
This is only valid if the source tensors (result tensor) is static
shaped with all unit-extents when the reshape is collapsing
(expanding) dimensions.
Differential Revision: https://reviews.llvm.org/D79764
Kang Zhang [Wed, 13 May 2020 05:59:52 +0000 (05:59 +0000)]
[PowerPC] Use add instead of addReg in ppc-early-ret pass
Summary:
The ppc-early-ret pass use the addReg() to add operand to the new
instruction, it can't reserve the flag of old operand. This has caused
machine verfications failed.
This patch use add() to instead of addReg().
Reviewed By: steven.zhang
Differential Revision: https://reviews.llvm.org/D77997
Stephen Neuendorffer [Wed, 13 May 2020 03:43:50 +0000 (20:43 -0700)]
[cmake] Update creation of object library dependencies for LINK_LIBS PUBLIC
We need to avoid declaring dependencies on strings which are valid
LINK_LIBS and not valid targets. Previously, we used if(TARGET) to
check this condition. However, if(TARGET) checks whether a target has
been created (in the cmake subdirectory traversal order) and not
whether it *will* be created. This results in annoying directory
ordering problems.
This patch changes the check to more explicitly eliminate problematic
libraries (namely -lpthread) using a REGEX.
Differential Revision: https://reviews.llvm.org/D79837
KAWASHIMA Takahiro [Thu, 7 May 2020 07:40:06 +0000 (16:40 +0900)]
[gcov] Fix simultaneous .gcda creation/lock
Fixes PR45673
The commit
9180c14fe4d (D76206) resolved only a part of the problem
of concurrent .gcda file creation. It ensured that only one process
creates the file but did not ensure that the process locks the
file first. If not, the process which created the file may clobber
the contents written by a process which locked the file first.
This is the cause of PR45673.
This commit prevents the clobbering by revising the assumption
that a process which creates the file locks the file first.
Regardless of file creation, a process which locked the file first
uses fwrite (new_file==1) and other processes use mmap (new_file==0).
I also tried to keep the creation/first-lock process same by using
mkstemp/link/unlink but the code gets long. This commit is more
simple.
Note: You may be confused with other changes which try to resolve
concurrent file access. My understanding is (may not be correct):
D76206: Resolve race of .gcda file creation (but not lock)
This one: Resolve race of .gcda file creation and lock
D54599: Same as D76206 but abandoned?
D70910: Resolve race of multi-threaded counter flushing
D74953: Resolve counter sharing between parent/children processes
D78477: Revision of D74953
Differential Revision: https://reviews.llvm.org/D79556
KAWASHIMA Takahiro [Thu, 7 May 2020 04:30:25 +0000 (13:30 +0900)]
[LoopReroll] Fix rerolling loop with use outside the loop
Fixes PR41696
The loop-reroll pass generates an invalid IR (or its assertion
fails in debug build) if values of the base instruction and
other root instructions (terms used in the loop-reroll pass)
are used outside the loop block. See IRs written in PR41696
as examples.
The current implementation of the loop-reroll pass can reroll
only loops that don't have values that are used outside the
loop, except reduced values (the last values of reduction chains).
This is described in the comment of the `LoopReroll::reroll`
function.
https://github.com/llvm/llvm-project/blob/llvmorg-10.0.0/llvm/lib/Transforms/Scalar/LoopRerollPass.cpp#L1600
This is checked in the `LoopReroll::DAGRootTracker::validate`
function.
https://github.com/llvm/llvm-project/blob/llvmorg-10.0.0/llvm/lib/Transforms/Scalar/LoopRerollPass.cpp#L1393
However, the base instruction and other root instructions skip
this check in the validation loop.
https://github.com/llvm/llvm-project/blob/llvmorg-10.0.0/llvm/lib/Transforms/Scalar/LoopRerollPass.cpp#L1229
Moving the check in front of the skip is the logically simplest
fix. However, inserting the check in an earlier stage is better
in terms of compilation time of unrerollable loops. This fix
inserts the check for the base instruction into the function
to validate possible base/root instructions. Check for other
root instructions is unnecessary because they don't match any
base instructions if they have uses outside the loop.
Differential Revision: https://reviews.llvm.org/D79549
Muhammad Omair Javaid [Wed, 13 May 2020 03:51:47 +0000 (08:51 +0500)]
[LLDB] Fix typo in xfail decorator assert.test
Fix a typo in earlier xfailed assert.test replace // with #.
Muhammad Omair Javaid [Wed, 13 May 2020 03:03:28 +0000 (08:03 +0500)]
[LLDB] Mark some xfails for arm-linux
This patch marks following tests as xfail for arm-linux target.
lldb/test/API/functionalities/load_using_paths/TestLoadUsingPaths.py
lldb/test/API/python_api/thread/TestThreadAPI.py
lldb/test/Shell/Recognizer/assert.test
Bugs have been filed for all of them for the corresponding failing
component.
aartbik [Wed, 13 May 2020 00:07:29 +0000 (17:07 -0700)]
[mlir] [VectorOps] Implement vector.constant_mask lowering to LLVM IR
Summary:
Makes this operation runnable on CPU by generating MLIR instructions
that are eventually folded into an LLVM IR constant for the mask.
Reviewers: nicolasvasilache, ftynse, reidtatge, bkramer, andydavis1
Reviewed By: nicolasvasilache, ftynse, andydavis1
Subscribers: mehdi_amini, rriddle, jpienaar, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, liufengdb, stephenneuendorffer, Joonsoo, grosul1, frgossen, Kayjukh, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79815
Muhammad Omair Javaid [Wed, 13 May 2020 02:15:10 +0000 (07:15 +0500)]
[LLDB] Fix minidebuginfo-set-and-hit-breakpoint.test for arm 32-bit
This patch fixes minidebuginfo-set-and-hit-breakpoint.test for arm-linux
targets. 32-bit elf executables use .rel.dyn and 64-bit uses .rela.dyn for
relocation entries for dynamic symbols.
Johannes Doerfert [Wed, 13 May 2020 01:51:21 +0000 (20:51 -0500)]
[Attributor][FIX] Stabilize the state of AAReturnedValues each update
For AAReturnedValues we treated new and existing information differently
in the updateImpl. Only the latter was properly analyzed and
categorized. The former was thought to be analyzed in the subsequent
update. Since the Attributor does not support "self-updates" we need to
make sure the state is "stable" after each updateImpl invocation. That
is, if the surrounding information does not change, the state is valid.
Now we make sure all return values have been handled and properly
categorized each iteration. We might not update again if we have not
requested a non-fix attribute so we cannot "wait" for the next update to
analyze a new return value.
Bug reported by @sdmitriev.
zoecarver [Wed, 13 May 2020 01:56:18 +0000 (18:56 -0700)]
[libcxx] Constrain function assignment operator (2574).
This patch fixes LWG issue 2574.
Differential Review: https://reviews.llvm.org/D62928
Zequan Wu [Wed, 13 May 2020 01:46:24 +0000 (18:46 -0700)]
test commit
Juneyoung Lee [Tue, 12 May 2020 15:37:38 +0000 (00:37 +0900)]
[ValueTracking] Fix crash in isGuaranteedNotToBeUndefOrPoison when V is in an unreachable block
Summary:
This fixes PR45885 by fixing isGuaranteedNotToBeUndefOrPoison so it does not look into dominating
branch conditions of V when V is an instruction in an unreachable block.
Reviewers: spatel, nikic, lebedev.ri
Reviewed By: nikic
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79790
Muhammad Omair Javaid [Wed, 13 May 2020 00:12:48 +0000 (05:12 +0500)]
Fix error in TestNumThreads.py when frame.GetFunctionName returns none
Summary:
This patch fixes an error happening in TestNumThreads.py when it encounters frame.GetFunctionName none for address only locations in stripped libc.
This error was showing up on arm-linux docker container running lldb buildbot.
Reviewers: labath
Reviewed By: labath
Subscribers: kristof.beyls, lldb-commits
Differential Revision: https://reviews.llvm.org/D79777
Nicolas Vasilache [Tue, 12 May 2020 21:17:34 +0000 (17:17 -0400)]
[mlir] Revisit std.subview handling of static information.
The main objective of this revision is to change the way static information is represented, propagated and canonicalized in the SubViewOp.
In the current implementation the issue is that canonicalization may strictly lose information because static offsets are combined in irrecoverable ways into the result type, in order to fit the strided memref representation.
The core semantics of the op do not change but the parser and printer do: the op always requires `rank` offsets, sizes and strides. These quantities can now be either SSA values or static integer attributes.
The result type is automatically deduced from the static information and more powerful canonicalizations (as powerful as the representation with sentinel `?` values allows). Previously static information was inferred on a best-effort basis from looking at the source and destination type.
Relevant tests are rewritten to use the idiomatic `offset: x, strides : [...]`-form. Bugs are corrected along the way that were not trivially visible in flattened strided memref form.
Lowering to LLVM is updated, simplified and now supports all cases.
A mixed static-dynamic mode test that wouldn't previously lower is added.
It is an open question, and a longer discussion, whether a better result type representation would be a nicer alternative. For now, the subview op carries the required semantic.
Differential Revision: https://reviews.llvm.org/D79662
Zequan Wu [Tue, 12 May 2020 21:07:50 +0000 (14:07 -0700)]
Add nomerge function attribute to supress tail merge optimization in simplifyCFG
We want to add a way to avoid merging identical calls so as to keep the
separate debug-information for those calls. There is also an asan
usecase where having this attribute would be beneficial to avoid
alternative work-arounds.
Here is the link to the feature request:
https://bugs.llvm.org/show_bug.cgi?id=42783.
`nomerge` is different from `noline`. `noinline` prevents function from
inlining at callsites, but `nomerge` prevents multiple identical calls
from being merged into one.
This patch adds `nomerge` to disable the optimization in IR level. A
followup patch will be needed to let backend understands `nomerge` and
avoid tail merge at backend.
Reviewed By: asbirlea, rnk
Differential Revision: https://reviews.llvm.org/D78659
Nico Weber [Tue, 12 May 2020 18:02:13 +0000 (14:02 -0400)]
[lld-macho] Ignore -platform_version and -syslibroot flags.
clang passes these flags; this makes it easier to try `clang -v`
output with `ld -flavor darwinnew`.
Differential Revision: https://reviews.llvm.org/D79797
Siva Chandra Reddy [Tue, 12 May 2020 23:11:15 +0000 (16:11 -0700)]
[libc][Obvious] Fix deps of few threads targets.
A missing dep has been added, and a few redundent deps have been
removed.
Casey Carter [Tue, 12 May 2020 23:11:22 +0000 (16:11 -0700)]
[libc++][test] Properly mark libc++-only XFAILs
These tests PASS on libstdc++ and MSVC.
Stanislav Mekhanoshin [Tue, 12 May 2020 21:18:53 +0000 (14:18 -0700)]
[AMDGPU] Make v4i64/v4f64/v8i64/v8f64 legal
We can produce such vectors in the Promote Alloca pass,
but we are unable to use movrel to operate it and lower
via scratch. Making it legal makes SI_INDIRECT patterns
work.
There is more work to do in subsequent changes:
1. We initialize m0 twice to access each dword. It shall
be possible to only do it once and increment base register
number instead.
2. We also need v16i64/v16f64 but these first need to be
added to tablegen.
Differential Revision: https://reviews.llvm.org/D79808
Jonas Devlieghere [Tue, 12 May 2020 20:18:35 +0000 (13:18 -0700)]
[lldb/Reproducers] Also record directories FileSystem::Collect.
Now that the FileCollector knows how to deal with directories we no
longer have to ignore them in the FileSystem class.
Sean Silva [Tue, 12 May 2020 20:40:00 +0000 (13:40 -0700)]
Revert of Revert of [mlir][shape] Tidy up shape.shape_of
Summary:
- Mark it NoSideEffect
- Add custom parser/printer
This reverts the temporary revert in
https://reviews.llvm.org/rG84a9c725742d26df04808a3c7349dbd98684c6cb
That was a false alarm. A downstream test actually needed to be updated.
Jan Korous [Tue, 12 May 2020 21:42:22 +0000 (14:42 -0700)]
[YAMLVFSWriter] Fix for delimiters
Differential Revision: https://reviews.llvm.org/D79809
Sanjay Patel [Tue, 12 May 2020 21:08:24 +0000 (17:08 -0400)]
[x86][CGP] enable target hook to sink funnel shift intrinsic's splatted shift amount
SDAG suffers when it can't see that a funnel operand is a splat value
(due to single-basic-block visibility), so invert the normal loop
hoisting rules to move a splat op closer to its use.
This would be part 1 of an enhancement similar to D63233.
This is needed to re-fix PR37426:
https://bugs.llvm.org/show_bug.cgi?id=37426
...because we got better at canonicalizing IR to funnel shift intrinsics.
The existing CGP code for shift opcodes is likely overstepping what it was
intended to do, so that will be fixed in a follow-up.
Differential Revision: https://reviews.llvm.org/D79718
Davide Italiano [Tue, 12 May 2020 22:37:44 +0000 (15:37 -0700)]
[GIsel] Update a comment and make it more precise.
This only covers ANYEXT/ZEXT. SEXT is covered in another test
I just checked in.
Alex Zinenko [Tue, 12 May 2020 22:30:54 +0000 (00:30 +0200)]
[mlir] Move Conversion/StandardToStandard to Dialect/StandardOps/Transforms/FuncConversions
Conversion/ folders were originally intended to store patterns for
DialectA->DialectB conversions that depend on both dialects and do not
conceptually belong to either of the dialects. As such, DialectA->DialectA
conversion does not make sense under Conversion/ and should rather live with
the dialect it operates on.
Differential Revision: https://reviews.llvm.org/D79569
Davide Italiano [Tue, 12 May 2020 22:31:34 +0000 (15:31 -0700)]
[GlobalISel] Assign the correct location when combining G_SEXT.
<rdar://problem/
62991635>
Alexey Lapshin [Tue, 12 May 2020 22:20:33 +0000 (01:20 +0300)]
Justin Hibbits [Sun, 19 Apr 2020 04:09:30 +0000 (23:09 -0500)]
PowerPC: Treat llvm.fma.f* intrinsic as using CTR with SPE
Summary:
The SPE doesn't have a 'fma' instruction, so the intrinsic becomes a
libcall. It really should become an expansion to two instructions, but
for some reason the compiler doesn't think that's as optimal as a
branch. Since this lowering is done after CTR is allocated for loops,
tell the optimizer that CTR may be used in this case. This prevents a
"Invalid PPC CTR loop!" assertion in the case that a fma() function call
is used in a C/C++ file, and clang converts it into an intrinsic.
Reviewed By: shchenz
Differential Revision: https://reviews.llvm.org/D78668
Alexey Lapshin [Tue, 12 May 2020 22:10:32 +0000 (01:10 +0300)]
Wei Mi [Tue, 12 May 2020 21:48:22 +0000 (14:48 -0700)]
[SampleFDO] Rename llvm-profdata flag -partial-profile to -gen-partial-profile.
The internal flag -partial-profile in llvm conflicts with the flag with
the same name in llvm-profdata. The conflict happens in builds with
LLVM_LINK_LLVM_DYLIB enabled. In this case the tools are linked with libLLVM
and we end up with two definitions for the same cl::opt.
The patch renames llvm-profdata flag -partial-profile to -gen-partial-profile.
Jonas Devlieghere [Tue, 12 May 2020 21:55:17 +0000 (14:55 -0700)]
[VirtualFileSystem] Add unit test that showcases another YAMLVFSWriter bug
This scenario generates another broken YAML mapping as illustrated below.
{
'type': 'directory',
'name': "c",
'contents': [
,
{
'type': 'directory',
'name': "d",
'contents': [
,
{
'type': 'directory',
'name': "e",
'contents': [
{
'type': 'file',
'name': "f",
'external-contents': "//root/a/c/d/e/f"
} {
'type': 'file',
'name': "g",
'external-contents': "//root/a/c/d/e/g"
}
]
}
]
}
]
},
Jonas Devlieghere [Tue, 12 May 2020 21:46:23 +0000 (14:46 -0700)]
[VirtualFileSystem] Add unit test that showcases YAMLVFSWriter bug
This scenario generates a broken YAML mapping as illustrated below.
{
'type': 'directory',
'name': "c",
'contents': [
{
'type': 'file',
'name': "d",
'external-contents': "//root/a/c/d"
} {
'type': 'file',
'name': "e",
'external-contents': "//root/a/c/e"
} {
'type': 'file',
'name': "f",
'external-contents': "//root/a/c/f"
}
]
},
Alexey Lapshin [Wed, 12 Feb 2020 17:47:39 +0000 (20:47 +0300)]
[X86][ISelLowering] refactor Varargs handling in X86ISelLowering.cpp
Summary:
This patch refactors handling of VarArgs in
X86TargetLowering::LowerFormalArguments.
That refactoring was requested while reviewing
D69372. Code related to varargs handling is removed
from X86TargetLowering::LowerFormalArguments and
is divided into smaller routines.
Reviewed By: aeubanks
Differential Revision: https://reviews.llvm.org/D74794
Fangrui Song [Thu, 7 May 2020 21:00:24 +0000 (14:00 -0700)]
[TargetLoweringObjectFileImpl] Produce .text.hot. instead of .text.hot for -fno-unique-section-names
GNU ld's internal linker script uses (https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=
add44f8d5c5c05e08b11e033127a744d61c26aee)
.text :
{
*(.text.unlikely .text.*_unlikely .text.unlikely.*)
*(.text.exit .text.exit.*)
*(.text.startup .text.startup.*)
*(.text.hot .text.hot.*)
*(SORT(.text.sorted.*))
*(.text .stub .text.* .gnu.linkonce.t.*)
/* .gnu.warning sections are handled specially by elf.em. */
*(.gnu.warning)
}
Because `*(.text.exit .text.exit.*)` is ordered before `*(.text .text.*)`, in a -ffunction-sections build, the C library function `exit` will be placed before other functions.
gold's `-z keep-text-section-prefix` has the same problem.
In lld, `-z keep-text-section-prefix` recognizes `.text.{exit,hot,startup,unlikely,unknown}.*`, but not `.text.{exit,hot,startup,unlikely,unknown}`, to avoid the strange placement problem.
In -fno-function-sections or -fno-unique-section-names mode, a function whose `function_section_prefix` is set to `.exit"`
will go to the output section `.text` instead of `.text.exit` when linked by lld.
To address the problem, append a dot to become `.text.exit.`
Reviewed By: grimar
Differential Revision: https://reviews.llvm.org/D79600
Louis Dionne [Tue, 12 May 2020 20:57:53 +0000 (16:57 -0400)]
[libc++abi] Adjust XFAIL on macOS for bug that was fixed in recent OSes
Sergey Dmitriev [Tue, 12 May 2020 20:31:03 +0000 (13:31 -0700)]
[Attributor] Fixup block addresses after rewriting function signature
Reviewers: jdoerfert, sstefan1, uenoku
Reviewed By: jdoerfert
Subscribers: hiraditya, uenoku, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79801
Richard Smith [Tue, 12 May 2020 20:14:32 +0000 (13:14 -0700)]
PR45589: Properly decompose overloaded `&&` and `||` operators in
constraint expressions.
We create overloaded `&&` and `||` operators to hold the possible
unqualified lookup results (if any) when the operands are dependent. We
could avoid building these in some cases (we will never use the stored
lookup results, and it would be better to not store them or perform the
lookups), but in the general case we will probably still need to handle
overloaded operators even with that optimization.
Stephen Neuendorffer [Tue, 12 May 2020 20:37:31 +0000 (13:37 -0700)]
[MLIR] Fix dependencies for Analysis libraries
cmake does not truly support dependencies on automatically generated files
which are not in the same directory as the targets which depend on them.
It works with ninja, but doesn't work with make
This patch adds an explicit dependence so that all dialects are built
before the analysis libraries.
Differential Revision: https://reviews.llvm.org/D79805
Artem Dergachev [Tue, 12 May 2020 19:43:32 +0000 (22:43 +0300)]
[analyzer] SATestBuild.py: Fix support for #NOPREFIX.
Regressed in ec2d93c.
Stephen Neuendorffer [Wed, 6 May 2020 19:52:00 +0000 (12:52 -0700)]
[MLIR] Rename SideEffects.td -> SideEffectInterfaces.td
This normalize the name of the tablegen file with the name of the generated
files (SideEffectInterfaces.h.inc) and the other Interface tablegen files,
which all end in Interface(s).td
Differential Revision: https://reviews.llvm.org/D79517
Fred Riss [Fri, 8 May 2020 22:14:14 +0000 (15:14 -0700)]
[lldb/Driver] Support terminal resizing
Summary:
The comment in the Editine.h header made it sound like editline was
just unable to handle terminal resizing. We were not ever telling
editline that the terminal had changed size, which might explain why
it wasn't working.
This patch threads a `TerminalSizeChanged()` callback through the
IOHandler and invokes it from the SIGWINCH handler in the driver. Our
`Editline` class already had a `TerminalSizeChanged()` method which
was invoked only when editline was configured.
This patch also changes `Editline` to not apply the changes right away
in `TerminalSizeChanged()`, but instead defer that to the next
character read. During my testing, it happened once that the signal
was received while our `ConnectionFileDescriptor::Read` was allocating
memory. As `el_resize` seems to allocate memory too, this crashed.
Reviewers: labath, teemperor
Subscribers: lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D79654
David Blaikie [Tue, 12 May 2020 18:39:01 +0000 (11:39 -0700)]
Avoid binding pointers to "auto&" (by dereferencing the pointer that's non-null anyway)
Based on @djtodoro's
2552dc5317e0
zoecarver [Tue, 12 May 2020 01:42:50 +0000 (18:42 -0700)]
[libcxx] Re-commit: shared_ptr changes from library fundamentals (P0414R2).
Implements P0414R2:
* Adds support for array types in std::shared_ptr.
* Adds reinterpret_pointer_cast for shared_ptr.
Re-committing now that the leaking tests are fixed.
Differential Revision: https://reviews.llvm.org/D62259
Kamau Bridgeman [Tue, 12 May 2020 18:15:06 +0000 (13:15 -0500)]
[PowerPC] Fold redundant load immediates of zero and delete if possible
This patch folds redundant load immediates into a zero for instructions
which recognise this as the value zero and not the register. If the load
immediate is no longer in use it is then deleted.
This is already done in earlier passes but the ppc-mi-peephole allows for
a more general implementation.
Differential Revision: https://reviews.llvm.org/D69168
Jonas Devlieghere [Tue, 12 May 2020 17:28:27 +0000 (10:28 -0700)]
[Reproducers] Serialize process arguments in ProcessInfo
While debugging why TestProcessList.py failed during passive replay, I
remembered that we don't serialize the arguments for ProcessInfo. This
is necessary to make the test pass and to make platform process list -v
behave the same during capture and replay.
Differential revision: https://reviews.llvm.org/D79646
Jan Korous [Fri, 8 May 2020 23:29:17 +0000 (16:29 -0700)]
[FileCollector][NFC] Add comments
Differential Revision: https://reviews.llvm.org/D78961
Juneyoung Lee [Mon, 20 Apr 2020 15:35:45 +0000 (00:35 +0900)]
[ValueTracking] Let propagatesPoison support binops/unaryops/cast/etc.
Summary:
This patch makes propagatesPoison be more accurate by returning true on
more bin ops/unary ops/casts/etc.
The changed test in ScalarEvolution/nsw.ll was introduced by
https://github.com/llvm/llvm-project/commit/
a19edc4d15b0dae0210b90615775edd76f021008 .
IIUC, the goal of the tests is to show that iv.inc's SCEV expression still has
no-overflow flags even if the loop isn't in the wanted form.
It becomes more accurate with this patch, so think this is okay.
Reviewers: spatel, lebedev.ri, jdoerfert, reames, nikic, sanjoy
Reviewed By: spatel, nikic
Subscribers: regehr, nlopes, efriedma, fhahn, javed.absar, llvm-commits, hiraditya
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D78615
Craig Topper [Tue, 12 May 2020 17:10:42 +0000 (10:10 -0700)]
[X86] Remove the v16i8->v16i16 path for MULHS with AVX2.
We have a couple main strategies for legalizing MULH.
-If the vXi16 type is legal, extend to do the full i16 multiply
and then shift and truncate the results.
-Use unpcks to split each 128 bit lane into high and low halves.a
For signed we have an extra case to split a v32i8 to v16i8 and then
use the extending to v16i16 strategy.
This patch proposes to use the unpck strategy instead. Which is
what we already do for unsigned.
This seems to be 1 instruction shorter when the RHS is constant
like the idiv case. It's 1 instruction longer for the smulo case.
But we're trading cross lane shuffles for inlane shuffles and a
shift.
Differential Revision: https://reviews.llvm.org/D79652
Dimitry Andric [Tue, 12 May 2020 10:33:15 +0000 (12:33 +0200)]
[arm] Add big-endian version of pcrel fixups for adr instructions
Summary:
In
2e24219d3cbf, a number of ARM pcrel fixups were resolved at assembly
time, to solve PR44929. This only covered little-endian ARM however, so
add similar fixups for big-endian ARM. Also extend the test case to
cover big-endian ARM.
Reviewers: hans, psmith, MaskRay
Reviewed By: psmith, MaskRay
Subscribers: kristof.beyls, hiraditya, danielkiss, emaste, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79774
Austin Kerbow [Tue, 12 May 2020 03:30:10 +0000 (20:30 -0700)]
[AMDGPU] Add AGPRs to getRegClassForSizeOnBank
Differential Revision: https://reviews.llvm.org/D79761
Craig Topper [Tue, 12 May 2020 16:43:24 +0000 (09:43 -0700)]
[CodeGen] Use Align in MachineConstantPool.
Sanjay Patel [Tue, 12 May 2020 16:48:51 +0000 (12:48 -0400)]
[VectorCombine] add test to check for iterative improvements; NFC
Thomas Lively [Tue, 12 May 2020 16:39:01 +0000 (09:39 -0700)]
[WebAssembly] Implement pseudo-min/max SIMD instructions
Summary:
As proposed in https://github.com/WebAssembly/simd/pull/122. Since
these instructions are not yet merged to the SIMD spec proposal, this
patch makes them entirely opt-in by surfacing them only through LLVM
intrinsics and clang builtins. If these instructions are made
official, these intrinsics and builtins should be replaced with simple
instruction patterns.
Reviewers: aheejin
Subscribers: dschuff, sbc100, jgravelle-google, hiraditya, sunfish, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D79742
Fangrui Song [Tue, 12 May 2020 16:21:19 +0000 (09:21 -0700)]
[gcov][test] Fix clang test
Fangrui Song [Tue, 12 May 2020 06:20:33 +0000 (23:20 -0700)]
[gcov] Default coverage version to '408*' and delete CC1 option -coverage-exit-block-before-body
gcov 4.8 (r189778) moved the exit block from the last to the second.
The .gcda format is compatible with 4.7 but
* decoding libgcov 4.7 produced .gcda with gcov [4.7,8) can mistake the
exit block, emit bogus `%s:'%s' has arcs from exit block\n` warnings,
and print wrong `" returned %s` for branch statistics (-b).
* decoding libgcov 4.8 produced .gcda with gcov 4.7 has similar issues.
Also, rename "return block" to "exit block" because the latter is the
appropriate term.
Whitney Tsang [Tue, 12 May 2020 15:56:14 +0000 (15:56 +0000)]
[PassBuilder] Moved ProfileSummaryAnalysis in buildInlinerPipeline.
Summary:
As commented in the code, ProfileSummaryAnalysis is required for inliner
pass to query, so this patch moved
RequireAnalysisPass<ProfileSummaryAnalysis> in the recently created
buildInlinerPipeline.
Reviewer: mtrofin, davidxl, tejohnson, dblaikie, jdoerfert, sstefan1
Reviewed By: mtrofin, davidxl, jdoerfert
Subscribers: hiraditya, steven_wu, dexonsmith, wuzish, llvm-commits,
jsji
Tag: LLVM
Differential Revision: https://reviews.llvm.org/D79696
Jay Foad [Fri, 17 Apr 2020 15:17:26 +0000 (16:17 +0100)]
[GlobalISel][IRTranslator] Fix <1 x Ty> handling in ConstantExprs
Summary:
ConstantExprs involving operations on <1 x Ty> could translate into MIR
that failed to verify with:
*** Bad machine code: Reading virtual register without a def ***
The problem was that translate(const Constant &C, Register Reg) had
recursive calls that passed the same Reg in for the translation of a
subexpression, but without updating VMap for the subexpression first as
translate(const Constant &C, Register Reg) expects.
Fix this by using the same translateCopy helper function that we use for
translating Instructions. In some cases this causes extra G_COPY
MIR instructions to be generated.
Fixes https://bugs.llvm.org/show_bug.cgi?id=45576
Reviewers: arsenm, volkan, t.p.northover, aditya_nandakumar
Subscribers: jvesely, wdng, nhaehnle, rovka, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D78378
Jay Foad [Fri, 17 Apr 2020 13:36:01 +0000 (14:36 +0100)]
[GlobalISel][IRTranslator] New helper function translateCopy. NFC.
Reviewers: arsenm, volkan, t.p.northover, aditya_nandakumar
Subscribers: wdng, rovka, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D78377
Florian Hahn [Tue, 12 May 2020 15:37:01 +0000 (16:37 +0100)]
[Matrix] Check non-dependent elt type before creating DepSizedMatrix.
We should check non-dependent element types before creating a
DependentSizedMatrixType. Otherwise we do not generate an error message
for dependent-sized matrix types with invalid non-dependent element
types, if the template is never instantiated. See the make5 struct in
the tests.
It also moves the SEMA template tests to
clang/test/SemaTemplate/matrix-type.cpp and introduces a few more test
cases.
Michael Kruse [Tue, 12 May 2020 14:44:34 +0000 (09:44 -0500)]
[docs] Corrected inaccuracies in Common Problems section.
Changed the language in LLVM_USE_LINKER to more strongly recommend LLD
and to specify that the GNU gold linker is only useful if LLD is
unavailable in binary form and it is the first build of LLVM. Added that
LLD will help when used on ELF-based platforms.
Corrected information in CMAKE_BUILD_TYPE regarding the Release build
type and enabling assertions.
Added option LLVM_ENABLE_ASSERTIONS and mentioned enabling this option
with a Release build as an alternative to using a Debug build.
Specified that the LLVM_OPTIMIZED_TABLEGEN
option is only for Debug builds, that the LLVM_USE_SPLIT_DWARF option
is only available on ELF host platforms, and that setting
CLANG_ENABLE_STATIC_ANALYZER to OFF only slightly improves build time.
These changes address comments made in D75425.
Reviewed By: Meinersbur
Differential Revision: https://reviews.llvm.org/D77346
Jez Ng [Fri, 24 Apr 2020 03:16:49 +0000 (20:16 -0700)]
[lld-macho] Add support for creating and reading reexported dylibs
This unblocks the linking of real programs, since many core system
functions are only available as sub-libraries of libSystem.
Differential Revision: https://reviews.llvm.org/D79228
Jez Ng [Tue, 12 May 2020 14:29:15 +0000 (07:29 -0700)]
[lld-macho] Re-add dylink-lazy test
This reverts commit
eb81de2de4003e3045fdf743e093f77e37aee9bf; the
test commands just needed to be run under llvm-lit.
James Y Knight [Tue, 12 May 2020 04:22:54 +0000 (00:22 -0400)]
Add comment for SelectionDAGBuilder::SL field.
Haojian Wu [Mon, 11 May 2020 09:02:34 +0000 (11:02 +0200)]
[clangd] Add metrics for selection tree and recovery expressions.
Reviewers: sammccall
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79701
Carl Ritson [Tue, 12 May 2020 14:02:05 +0000 (23:02 +0900)]
[AMDGPU] Order pos exports before param exports
Summary:
Modify export clustering DAG mutation to move position exports
before other exports types.
Reviewers: foad, arsenm, rampitec, nhaehnle
Reviewed By: foad
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79670
Matt Arsenault [Fri, 27 Mar 2020 16:28:24 +0000 (12:28 -0400)]
HIP: Merge builtin library handling
Merge with the new --rocm-path handling used for OpenCL. This looks
for a usable set of device libraries upfront, rather than giving a
generic "no such file or directory error". If any of the required
bitcode libraries are missing, this will now produce a "cannot find
ROCm installation." error. This differs from the existing hip specific
flags by pointing to a rocm root install instead of a single directory
with bitcode files.
This tries to maintain compatibility with the existing the
--hip-device-lib and --hip-device-lib-path flags, as well as the
HIP_DEVICE_LIB_PATH environment variable, or at least the range of
uses with testcases. The existing range of uses and behavior doesn't
entirely make sense to me, so some of the untested edge cases change
behavior. Currently the two path forms seem to have the double purpose
of a search path for an arbitrary --hip-device-lib, and for finding
the stock set of libraries. Since the stock set of libraries This also
changes the behavior when multiple paths are specified, and only takes
the last one (and the environment variable only handles a single
path).
If --hip-device-lib is used, it now only treats --hip-device-lib-path
as the search path for it, and does not attempt to find the rocm
installation. If not, --hip-device-lib-path and the environment
variable are used as the directory to search instead of the rocm root
based path.
This should also automatically fix handling of the options to use
wave64.
Matt Arsenault [Fri, 10 Apr 2020 16:56:20 +0000 (12:56 -0400)]
AMDGPU: Search for new ROCm bitcode library structure
The current install situation is a mess, but I'm working on fixing
it. Search for the target layout instead of one of the N options that
exist today.
Reid Kleckner [Mon, 11 May 2020 20:16:25 +0000 (13:16 -0700)]
[LLD] Rename iDTable -> idTable, NFC
The variable renaming change did not handle this variable well.
Benjamin Kramer [Tue, 12 May 2020 13:26:59 +0000 (15:26 +0200)]
Fold single-use variables into assert
This avoids unused variable warnings in Release builds.
Kristof Beyls [Thu, 7 May 2020 15:32:44 +0000 (17:32 +0200)]
Add Linux SVE Ptrace macros.
Differential Revision: https://reviews.llvm.org/D79623
Sam McCall [Tue, 12 May 2020 13:18:50 +0000 (15:18 +0200)]
Revert "[mlir] Revisit std.subview handling of static information."
This reverts commit
80d133b24f77d1b9d351251315606441c971ef9b.
Per Stephan Herhut: The canonicalizer pattern that was added creates
forms of the subview op that cannot be lowered.
This is shown by failing Tensorflow XLA tests such as:
tensorflow/compiler/xla/service/mlir_gpu/tests:abs.hlo.test
Will provide more details offline, they rely on logs from private CI.
Melanie Blower [Fri, 8 May 2020 15:05:34 +0000 (08:05 -0700)]
[PATCH] #pragma float_control should be permitted in namespace scope.
Summary: Erroneous error diagnostic observed in VS2017 <numeric> header
Also correction to propagate usesFPIntrin from template func to instantiation.
Reviewers: rjmccall, erichkeane (no feedback received)
Differential Revision: https://reviews.llvm.org/D79631
Simon Pilgrim [Tue, 12 May 2020 12:43:38 +0000 (13:43 +0100)]
[X86] combineX86ShuffleChain - use narrowShuffleMaskElts scale == 1 builtin handling. NFC.
narrowShuffleMaskElts already has the fast-path for scale == 1, no need to reimplement it here.
Yaxun (Sam) Liu [Fri, 24 Apr 2020 20:41:24 +0000 (16:41 -0400)]
[CUDA][HIP] Workaround for resolving host device function against wrong-sided function
recommit
c77a4078e01033aa2206c31a579d217c8a07569b with fix
https://reviews.llvm.org/D77954 caused regressions due to diagnostics in implicit
host device functions.
For now, it seems the most feasible workaround is to treat implicit host device function and explicit host
device function differently. Basically in device compilation for implicit host device functions, keep the
old behavior, i.e. give host device candidates and wrong-sided candidates equal preference. For explicit
host device functions, favor host device candidates against wrong-sided candidates.
The rationale is that explicit host device functions are blessed by the user to be valid host device functions,
that is, they should not cause diagnostics in both host and device compilation. If diagnostics occur, user is
able to fix them. However, there is no guarantee that implicit host device function can be compiled in
device compilation, therefore we need to preserve its overloading resolution in device compilation.
Differential Revision: https://reviews.llvm.org/D79526
Sam Parker [Tue, 12 May 2020 12:05:09 +0000 (13:05 +0100)]
[NFC][AArch64] More casts tests...
Don't use truncs are users because sometimes they're free too.
Simon Pilgrim [Tue, 12 May 2020 11:31:07 +0000 (12:31 +0100)]
[X86][AVX] Use X86ISD::VPERM2X128 for blend-with-zero if optimizing for size
Last part of PR22984 - avoid the zero-register dependency if optimizing for size
Simon Pilgrim [Sun, 10 May 2020 20:50:13 +0000 (21:50 +0100)]
FuzzerCLI.h - reduce StringRef.h include to forward declaration. NFC.
Simon Pilgrim [Sun, 10 May 2020 19:33:51 +0000 (20:33 +0100)]
DebugCounter.h - remove unused includes. NFC.
Added explicit StringRef.h include as we need the full definition for several inline functions in DebugCounter.h.
Pierre-vh [Wed, 8 Apr 2020 10:55:09 +0000 (11:55 +0100)]
[Target][ARM] Replace outdated getARMVPTBlockMask function
getARMVPTBlockMask was an outdated function that only handled basic
block masks: T, TT, TTT and TTTT. This worked fine before the MVE
VPT Block Insertion Pass improvements as it was the only kind of
masks that it could generate, but now it can generate more complex
masks that uses E predicates, so it's dangerous to use that function
to calculate VPT/VPST block masks.
I replaced it with 2 different functions:
- expandPredBlockMask, in ARMBaseInfo. This adds an "E" or "T" at
the end of an existing PredBlockMask.
- recomputeVPTBlockMask, in Thumb2InstrInfo. This takes an iterator
to a VPT/VPST instruction and recomputes its block mask by looking
at the predicated instructions that follows it. This should be
used to recompute a block mask after removing/adding a predicated
instruction to the block.
The expandPredBlockMask function is pretty much imported from the MVE
VPT Blocks pass.
I had to change the ARMLowOverheadLoops and MVEVPTBlocks passes as well
so they could use these new functions.
Differential Revision: https://reviews.llvm.org/D78201
Pierre-vh [Thu, 2 Apr 2020 14:24:14 +0000 (15:24 +0100)]
[Target][ARM] Replace re-uses of old VPR values with VPNOTs
Differential Revision: https://reviews.llvm.org/D76847
David Zarzycki [Tue, 12 May 2020 10:44:37 +0000 (06:44 -0400)]
[libcxx testing] Remove ALLOW_RETRIES from sleep_for.pass.cpp
Operating systems are best effort by default, so we cannot assume that
sleep-like APIs return as soon as we'd like.
Even if a sleep-like API returns when we want it to, the potential for
preemption means that attempts to measure time are subject to delays.
Sander de Smalen [Tue, 12 May 2020 10:23:38 +0000 (11:23 +0100)]
[CodeGen][SVE] Add patterns for whole vector predicate select
Added patterns to implement `select i1 %p, <vty> %a, <vty> %b`
Reviewed By: efriedma
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79356
Jim Lin [Tue, 12 May 2020 10:49:17 +0000 (18:49 +0800)]
Revert "[RISCV] Make CanLowerReturn protected for downstream maintenance"
This reverts commit
d775841d7d6ee3e8bbf3a420590be9bb19433eaa.
Sam Parker [Tue, 12 May 2020 10:30:54 +0000 (11:30 +0100)]
[NFC][AArch64] More cast cost tests
Add truncating stores and casts with users.
Sander de Smalen [Tue, 12 May 2020 10:01:18 +0000 (11:01 +0100)]
[SveEmitter] Add builtins for svdup and svindex
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D79357
Petre-Ionut Tudor [Tue, 21 Apr 2020 13:11:13 +0000 (14:11 +0100)]
[ARM] Refactor lower to S[LR]I optimization
Summary:
The optimization has been refactored to fix certain bugs and
limitations. The condition for lowering to S[LR]I has been changed
to reflect the manual pseudocode description of SLI and SRI operation.
The optimization can now handle more cases of operand type and order.
Subscribers: kristof.beyls, hiraditya, danielkiss, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79233
Sam Parker [Thu, 7 May 2020 12:21:31 +0000 (13:21 +0100)]
[ARM][CostModel] Improve getCastInstrCost
- Specifically check for sext/zext users which have 'long' form NEON
instructions.
- Add more entries to the table for sext/zexts so that we can report
more accurately the number of vmovls required for NEON.
- Pass the instruction to the pass implementation.
Differential Revision: https://reviews.llvm.org/D79561
Sam Parker [Tue, 12 May 2020 07:54:18 +0000 (08:54 +0100)]
[AArch64][CostModel] getCastInstrCost
Pass the instruction to the base implementation.
Differential Revision: https://reviews.llvm.org/D79562
Manoel Roemmer [Tue, 12 May 2020 08:05:00 +0000 (10:05 +0200)]
[Openmp][VE] Libomptarget plugin for NEC SX-Aurora
This patch adds a libomptarget plugin for the NEC SX-Aurora TSUBASA Vector
Engine (VE target). The code is largely based on the existing generic-elf
plugin and uses the NEC VEO and VEOSINFO libraries for offloading.
Differential Revision: https://reviews.llvm.org/D76843
Haojian Wu [Tue, 12 May 2020 08:16:30 +0000 (10:16 +0200)]
get rid of the NDEBUG usage in RecoveryExpr, NFC.
use the llvm::all_of, per dblaikie's suggestion.
Sam Parker [Tue, 12 May 2020 07:47:28 +0000 (08:47 +0100)]
[NFC][AArch64] Update tests
Add cost model tests for extending loads.
Eric Christopher [Tue, 12 May 2020 07:44:05 +0000 (00:44 -0700)]
Fix typos encountered while working on pass pipeline for O1.
Djordje Todorovic [Tue, 12 May 2020 07:40:47 +0000 (09:40 +0200)]
Revert "[NFC][DwarfDebug] Prefer explicit to auto type deduction"
This wasn't proposed by the LLVM Style Guide.
Please see https://reviews.llvm.org/D79624.
This reverts commit rG2552dc5317e0.
Djordje Todorovic [Tue, 12 May 2020 07:04:57 +0000 (09:04 +0200)]
Revert "[NFC][DwarfDebug] Avoid default capturing when using lambdas"
Reverting this because we found it isn't that useful.
Please see https://reviews.llvm.org/D79616.
This reverts commit rG45e5a32a8bd3.