Commit Graph

378638 Commits

Author SHA1 Message Date
Joseph Huber fda4853998 [OpenMP] Fix seg fault in libomptarget when using Info with multiple threads
Summary:
One option for the LIBOMPTARGET_INFO environment variable is to print the current status of the device's data mappings. These are a shared resource among threads so this needs to be protected when using multiple streams.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D95786
2021-02-01 11:21:57 -05:00
Sam McCall 8712df7a62 [clangd] references: decls of overrides of x are refs to x, not decls
This requires a second index query for refs to overrides, as the refs
call doesn't tell you which ref points at which symbol.

Differential Revision: https://reviews.llvm.org/D95451
2021-02-01 17:19:19 +01:00
Simon Pilgrim 657e769688 Revert rGce587529ad8b5 - "[APFloat] multiplySignificand - pass IEEEFloat as const reference. NFCI."
Breaks on some buildbots
2021-02-01 16:15:23 +00:00
Sander de Smalen bf294953e7 NFC: Migrate SimplifyCFG to work on InstructionCost
This patch migrates cost values and arithmetic to work on InstructionCost.
When the interfaces to TargetTransformInfo are changed, any InstructionCost
state will propagate naturally.

See this patch for the introduction of the type: https://reviews.llvm.org/D91174
See this thread for context: http://lists.llvm.org/pipermail/llvm-dev/2020-November/146408.html

Reviewed By: spatel

Differential Revision: https://reviews.llvm.org/D95351
2021-02-01 16:14:05 +00:00
Sander de Smalen 880b64aa22 [SimplifyCFG] NFC: Rename static methods to clang-tidy standards.
This patch is a precursor to D95351, which changes the signature
of these methods.
2021-02-01 16:14:05 +00:00
David Green 5805521207 [ARM] Simplify VMOVRRD from extracts of buildvectors
Under the softfp calling convention, we are often left with
VMOVRRD(extract(bitcast(build_vector(a, b, c, d)))) for the return value
of the function. These can be simplified to a,b or c,d directly,
depending on the value of the extract.

Big endian is a little different because the bitcast switches the lanes
around, meaning we end up with b,a or d,c.

Differential Revision: https://reviews.llvm.org/D94989
2021-02-01 16:09:25 +00:00
Sam McCall ff4832dbff [clangd] Respect ReferencesParams.context.includeDeclarations
Unfortunately this treats overrides declarations as declarations, not as
references. I don't plan to land this until I have a fix for that issue.

Differential Revision: https://reviews.llvm.org/D95450
2021-02-01 17:07:02 +01:00
J-Y You 267b573b55 [TableGen] Fix anonymous record self-reference in foreach and multiclass
If we instantiate self-referenced anonymous records in foreach and
multiclass, the NAME value will point to incorrect record. It's because
anonymous name is resolved too early.

This patch adds AnonymousNameInit to represent an anonymous record name.
When instantiating an anonymous record, it will update the referred name.

Differential Revision: https://reviews.llvm.org/D95309
2021-02-01 10:59:07 -05:00
Simon Pilgrim ce587529ad [APFloat] multiplySignificand - pass IEEEFloat as const reference. NFCI.
Avoids unnecessary IEEEFloat copies.
2021-02-01 15:41:50 +00:00
Ruslan Arutyunyan c448ea948c [libc++] Fix for the Bug 41784
Add deleted volatile copy-assignment operator in the most derived atomic
to fix the Bug 41784. The root cause: there is an `operator=(T) volatile`
that has better match than the deleted copy-assignment operator of the base
class when `this` is `volatile`. The compiler sees that right operand of
the assignment operator can be converted to `T` and chooses that path
without taking into account the deleted copy-assignment operator of the
base class.

The current behavior on libstdc++ is different from what we have in libc++.
On the same test compilation fails with libstdc++. Proof: https://godbolt.org/z/nebPYd
(everything is the same except the -stdlib option).

I choose the way with explicit definition of copy-assignment for atomic
in the most derived class. But probably we can fix that by moving
`operator=(T)` overloads to the base class from both specializations.
At first glance, it shouldn't break anything.

Differential Revision: https://reviews.llvm.org/D90968
2021-02-01 10:14:22 -05:00
Haojian Wu 78c22fbce9 [clangd] Don't rely on builtin headers for document-link.test.
This test seems to be failing at HEAD.

Reviewed By: kadircet, sammccall

Differential Revision: https://reviews.llvm.org/D95670
2021-02-01 15:36:34 +01:00
LLVM GN Syncbot 313a36130f [gn build] Port b63cd4db91 2021-02-01 14:24:45 +00:00
Sam McCall b63cd4db91 [clangd] Rename: merge index/AST refs path-insensitively where needed
If you have c:\foo open, and C:\foo indexed (case difference) then these
need to be considered the same file. Otherwise we emit edits to both,
and editors do... something that isn't pretty.

Maybe more centralized normalization is called for, but it's not trivial
to do this while also being case-preserving. see
https://github.com/clangd/clangd/issues/108

Fixes https://github.com/clangd/clangd/issues/665

Differential Revision: https://reviews.llvm.org/D95759
2021-02-01 15:15:21 +01:00
Kerry McLaughlin 9b4fcfaa9e [SVE][CodeGen] Remove performMaskedGatherScatterCombine
The AArch64 DAG combine added by D90945 & D91433 extends the index
of a scalable masked gather or scatter to i32 if necessary.

This patch removes the combine and instead adds shouldExtendGSIndex, which
is used by visitMaskedGather/Scatter in SelectionDAGBuilder to query whether
the index should be extended before calling getMaskedGather/Scatter.

Reviewed By: david-arm

Differential Revision: https://reviews.llvm.org/D94525
2021-02-01 14:10:00 +00:00
Florian Hahn f1e8136115
[SCEV] Bail out if URem operand cannot be zero-extended.
In some cases, LHS is larger than the target expression type. Bail out
in that case for now, to avoid crashing
2021-02-01 13:50:54 +00:00
Jeroen Dobbelaere 2556b413a7 Revert "[Verifier] enable llvm.experimental.noalias.scope.decl dominance check."
the 'clang-with-lto-ubuntu' buildbot triggers the assertion.

This reverts commit b43c395e60.
2021-02-01 14:38:33 +01:00
Florian Hahn 972212d29c
[ConstraintElimination] Add tests for signed predicates.
Add test coverage for conditions with signed predicates.
2021-02-01 13:23:05 +00:00
Med Ismail Bennani ee562e2315 [lldb/test] Skip `SBTarget::IsLoaded` test on windows (NFC)
This patch skips the test for the SBTarget::IsLoaded method on windows
since the logic is different.

Differential Revision: https://reviews.llvm.org/D95686

Signed-off-by: Med Ismail Bennani <medismail.bennani@gmail.com>
2021-02-01 14:09:50 +01:00
Tim Northover c2b322fc19 GlobalISel: check type size before getZExtValue()ing it.
Otherwise getZExtValue() asserts.
2021-02-01 12:43:33 +00:00
Cullen Rhodes 8cda227432 [LV] Fix crash when computing max VF too early
D90687 introduced a crash:

  llvm::LoopVectorizationCostModel::computeMaxVF(llvm::ElementCount, unsigned int):
    Assertion `WideningDecisions.empty() && Uniforms.empty() && Scalars.empty() &&
    "No decisions should have been taken at this point"' failed.

when compiling the following C code:

  typedef struct {
  char a;
  } b;

  b *c;
  int d, e;

  int f() {
    int g = 0;
    for (; d; d++) {
      e = 0;
      for (; e < c[d].a; e++)
        g++;
    }
    return g;
  }

with:

  clang -Os -target hexagon -mhvx -fvectorize -mv67 testcase.c -S -o -

This occurred since prior to D90687 computeFeasibleMaxVF would only be
called in computeMaxVF when a scalar epilogue was allowed, but now it's
always called. This causes the assert above since computeFeasibleMaxVF
collects all viable VFs larger than the default MaxVF, and for each VF
calculates the register usage which results in analysis being done the
assert above guards against. This can occur in computeFeasibleMaxVF if
TTI.shouldMaximizeVectorBandwidth and this target hook is implemented in
the hexagon backend to always return true.

Reported by @iajbar.

Reviewed By: fhahn

Differential Revision: https://reviews.llvm.org/D94869
2021-02-01 12:14:59 +00:00
Sander de Smalen 3b8a1d581e NFC: Migrate SpeculativeExecution to work on InstructionCost
This patch migrates cost values and arithmetic to work on InstructionCost.
When the interfaces to TargetTransformInfo are changed, any InstructionCost
state will propagate naturally.

See this patch for the introduction of the type: https://reviews.llvm.org/D91174
See this thread for context: http://lists.llvm.org/pipermail/llvm-dev/2020-November/146408.html

Reviewed By: david-arm

Differential Revision: https://reviews.llvm.org/D95356
2021-02-01 12:13:23 +00:00
Andy Yankovsky 754ab803b8 [lldb] Use current execution context in SBDebugger
Use `GetSelectedExecutionContext()` instead of
`GetCommandInterpreter().GetExecutionContext()` in
`SBDebugger::GetInternalVariableValue/SBDebugger::SetInternalVariable`. The
execution context in the command interpreter might be empty, if no commands has
been executed yet (it is updated only when handling commands or completions --
e.g.
https://github.com/llvm/llvm-project/blob/main/lldb/source/Interpreter/CommandInterpreter.cpp#L1855).

Reviewed By: teemperor

Differential Revision: https://reviews.llvm.org/D95761
2021-02-01 13:12:42 +01:00
Jan Svoboda eefa8a9ff8 Revert "[clang][cli] Port OpenMP-related LangOpts to marshalling system"
This reverts commit 9ad94c12

It turns out that to correctly generate command line flags for LangOptions::OpenMP and LangOptions::OpenMPSimd, we need the flexibility of C++.
2021-02-01 12:50:48 +01:00
Martin Storsjö 11e74e512d [lldb] Remove a stray semicolon, fixing GCC warnings. NFC. 2021-02-01 13:45:07 +02:00
Dmitry Preobrazhensky 99b5631649 [AMDGPU][MC] Corrected error position for invalid operands
Generic parser may report an incorrect error position when an offending operand is followed by a comma.
See bug 48884 for details: https://bugs.llvm.org/show_bug.cgi?id=48884.

Differential Revision: https://reviews.llvm.org/D95674
2021-02-01 14:31:08 +03:00
Med Ismail Bennani b8923c0022 [lldb/API] Expose Module::IsLoadedInTarget() to SB API (NFC)
This patch adds an `SBTarget::IsLoaded(const SBModule&) const` endpoint
to lldb's Scripting Bridge API. As the name suggests, it will allow the
user to know if the module is loaded in a specific target.

rdar://37957625

Differential Review: https://reviews.llvm.org/D95686

Signed-off-by: Med Ismail Bennani <medismail.bennani@gmail.com>
2021-02-01 12:16:33 +01:00
xgupta 94fac81fcc [Branch-Rename] Fix some links
According to the [[ https://foundation.llvm.org/docs/branch-rename/ | status of branch rename ]], the master branch of the LLVM repository is removed on 28 Jan 2021.

Reviewed By: mehdi_amini

Differential Revision: https://reviews.llvm.org/D95766
2021-02-01 16:43:21 +05:30
David Green ad12e6ee95 [ARM] Turn sext_inreg(VGetLaneu) into VGetLaneu
This adds a DAG combine for converting sext_inreg of VGetLaneu into
VGetLanes, providing the types match correctly.

Differential Revision: https://reviews.llvm.org/D95073
2021-02-01 11:10:35 +00:00
Jeroen Dobbelaere b43c395e60 [Verifier] enable llvm.experimental.noalias.scope.decl dominance check.
Now that Loop Peeling has been fixed (80cdd30eb9),
enable the dominance check by default.

This reverts commit 3b5d36ece2.
2021-02-01 11:53:01 +01:00
Simon Pilgrim 5211af4818 [X86][AVX] combineExtractWithShuffle - combine extracts from 256/512-bit vector shuffles.
We can only legally extract from the lowest 128-bit subvector, so extract the correct subvector to allow us to handle 256/512-bit vector element extracts.
2021-02-01 10:31:43 +00:00
David Green 6ab792b68d [ARM] Simplify extract of VMOVDRR
Under SoftFP calling conventions, we can be left with
extract(bitcast(BUILD_VECTOR(VMOVDRR(a, b), ..))) patterns that can
simplify to a or b, depending on the extract lane.

Differential Revision: https://reviews.llvm.org/D94990
2021-02-01 10:24:57 +00:00
Kazushi (Jam) Marukawa 4648098f97 [VE] Change inetger constants 32-bit friendly
Correct integer constants like `1UL << 63` to `UINT64_C(1) << 63` in
order to make them work on 32-bit machines.  Tested on both an i386
and x86_64 machines.

Reviewed By: mgorny

Differential Revision: https://reviews.llvm.org/D95724
2021-02-01 19:00:47 +09:00
Florian Hahn a9583a1923
[LoopUnswitch] Pacify compiler warnings.
Attempt to fix some compiler warnings on some bots after
b8c81fa5c7.
2021-02-01 09:13:06 +00:00
Florian Hahn b8c81fa5c7
[LoopUnswitch] Add shortcut if unswitched path is a no-op.
If we determine that the invariant path through the loop has no effects,
we can directly branch to the exit block, instead to unswitching first.

Besides avoiding some extra work (unswitching first, then deleting the
loop again) this allows to be more aggressive than regular unswitching
with respect to cost-modeling. This approach should always be be
desirable.

This is similar in spirit to D93734, just that it uses the previously
added checks for loop-unswitching.

I tried to add the required no-op checks from scratch, as we only check
a subset of the loop. There is potential to unify the checks with
LoopDeletion, at the cost of adding a predicate whether a block should
be considered.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D95468
2021-02-01 09:03:30 +00:00
Jeroen Dobbelaere 80cdd30eb9 [LoopPeel] Use llvm.experimental.noalias.scope.decl for duplicating noalias metadata as needed.
The reduction of a sanitizer build failure when enabling the dominance check (D95335) showed that loop peeling also needs to take care of scope duplication, just like loop unrolling (D92887).

Reviewed By: nikic

Differential Revision: https://reviews.llvm.org/D95544
2021-02-01 10:01:17 +01:00
Raphael Isemann 2939d2e1b4 [lldb][docs] Attempt to disable the generated GitHub button on the LLDB website
For unknown reasons the alabaster theme on the docs server is always generating
a GitHub link in the side bar. Beside the privacy problems of having an iframe
to some third-party service, we never configured any GitHub integration so
this button just links to the GitHub main site.

The button generation should be disabled by default, but as that's apparently
not true in the alabaster theme on the server, this patch tries working around
the issue by just explicitly turning off the GitHub integration.
2021-02-01 09:47:39 +01:00
Tres Popp 2790cbedd0 Revert "[mlir][Linalg] Replace SimplePad with PadTensor in hoist-padding"
This reverts commit d9b953d84b.

This commit resulted in build bot failures and the author is away from a
computer, so I am reverting on their behalf until they have a chance to
look into this.
2021-02-01 09:43:55 +01:00
Christian Sigg a4b7d52f3a [mlir] Fix missing null termination in cuLinkAddData argument.
Reviewed By: ftynse

Differential Revision: https://reviews.llvm.org/D95679
2021-02-01 09:32:50 +01:00
Hanhan Wang d9b953d84b [mlir][Linalg] Replace SimplePad with PadTensor in hoist-padding
This is the last revision to migrate using SimplePadOp to PadTensorOp, and the
SimplePadOp is removed in the patch. Update a bit in SliceAnalysis because the
PadTensorOp takes a region different from SimplePadOp. This is not covered by
LinalgOp because it is not a structured op.

Also, remove a duplicated comment from cpp file, which is already described in a
header file. And update the pseudo-mlir in the comment.

Reviewed By: nicolasvasilache

Differential Revision: https://reviews.llvm.org/D95671
2021-02-01 00:02:37 -08:00
Tobias Hieta c3c02d0d5a [OpenMP] Fix python3 compatibility in openmp's lit.cfg
Differential Revision: https://reviews.llvm.org/D95669
2021-02-01 08:20:26 +01:00
Craig Topper 77f077b8c3 [TableGen] Don't commute isel patterns if it would put an immAllOnesV or immAllZerosV on the left hand side.
This primarily occurs with isel patterns using vnot. This reduces
the number of variants in the isel tables.

We generally canonicalize build_vectors of constants to the RHS. I think
we might fail if there is a bitcast on the build_vector, but that
should be easy to fix if we can find a case. Usually the
bitcast is introduced by type legalization or lowering. It's
likely canonicalization would have already occured.
2021-01-31 21:18:21 -08:00
Serge Pavlov bf416d166b [FPEnv] Intrinsic for setting rounding mode
To set non-default rounding mode user usually calls function 'fesetround'
from standard C library. This way has some disadvantages.

* It creates unnecessary dependency on libc. On the other hand, setting
  rounding mode requires few instructions and could be made by compiler.
  Sometimes standard C library even is not available, like in the case of
  GPU or AI cores that execute small kernels.
* Compiler could generate more effective code if it knows that a particular
  call just sets rounding mode.

This change introduces new IR intrinsic, namely 'llvm.set.rounding', which
sets current rounding mode, similar to 'fesetround'. It however differs
from the latter, because it is a lower level facility:

* 'llvm.set.rounding' does not return any value, whereas 'fesetround'
  returns non-zero value in the case of failure. In glibc 'fesetround'
  reports failure if its argument is invalid or unsupported or if floating
  point operations are unavailable on the hardware. Compiler usually knows
  what core it generates code for and it can validate arguments in many
  cases.
* Rounding mode is specified in 'fesetround' using constants like
  'FE_TONEAREST', which are target dependent. It is inconvenient to work
  with such constants at IR level.

C standard provides a target-independent way to specify rounding mode, it
is used in FLT_ROUNDS, however it does not define standard way to set
rounding mode using this encoding.

This change implements only IR intrinsic. Lowering it to machine code is
target-specific and will be implemented latter. Mapping of 'fesetround'
to 'llvm.set.rounding' is also not implemented here.

Differential Revision: https://reviews.llvm.org/D74729
2021-02-01 11:28:14 +07:00
Craig Topper ce27b32715 [Mips] Cleanup isel patterns to use 'vnot' instead of (xor X, immAllOnesV). NFCI
A couple patterns used bitconvert on the immAllOnesV, but
the isel matching uses ISD::isBuildVectorAllOnes which
is able to look through bitcasts. So isel patterns don't need
to do it explicitly.
2021-01-31 20:01:05 -08:00
Craig Topper 94206f1f90 [PowerPC] Remove vnot_ppc and replace with the standard vnot.
immAllOnesV has special support for looking through bitcasts
automatically so isel patterns don't need to explicitly look
for the bitconvert.
2021-01-31 19:41:33 -08:00
Craig Topper ff46026897 [X86] Cleanup isel patterns to use 'vnot' instead of (xor X, immAllOnesV) to improve readability. NFC 2021-01-31 18:53:40 -08:00
Lang Hames d2bb07949b Revert "[JITLink] Add missing symbols for ELF ehframe testcase, re-enable ...."
This reverts commit 6e58539659.

This failed in http://lab.llvm.org:8011/#/builders/123/builds/2676. I guess
were're still missing some symbols, but unfortunately the specific error is
masked by a bug in python/lit that hides stderr. This test will have to remain
disabled on Windows until I can get help to debug it further.
2021-02-01 13:32:11 +11:00
Craig Topper 44cc5abbf9 [RISCV] Custom lower fshl/fshr with Zbt extension.
We need to add a mask to the shift amount for these operations
to use the FSR/FSL instructions. We were previously doing this
in isel patterns, but custom lowering will make the mask
visible to optimizations earlier.
2021-01-31 17:49:15 -08:00
Lang Hames 6e58539659 [JITLink] Add missing symbols for ELF ehframe testcase, re-enable on Windows.
This testcase was failing on windows due to missing definitions. This commit
adds definitions of the missing symbols (as absolute symbols) to eliminate the
errors.
2021-02-01 12:24:24 +11:00
Mike Rice e94a35a744 [OpenMP] Fix comment and assertion strings (NFC). 2021-01-31 17:17:33 -08:00
James Y Knight 20b1c1300c Fix test in "CFG: Create scope for non-compound range-for body."
The constant 4 is sometimes printed as "4L", or "4LL", in CFG dump
output, depending on platform; accept all variants.

Ammends commit 8f670d5b6d.
2021-01-31 19:56:26 -05:00