Nick Desaulniers [Mon, 24 May 2021 19:06:49 +0000 (12:06 -0700)]
fix up test from D102742
In D102742, I mistakenly put the split file designator above a bunch of
CHECK lines, which unintentionally removed the CHECKs from actually
being verified.
This can be verified by observing:
<build dir>/test/CodeGen/X86/Output/stack-protector-3.ll.tmp/main.ll
George [Mon, 24 May 2021 18:52:41 +0000 (11:52 -0700)]
Surface clone APIs in CAPI
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D102987
LLVM GN Syncbot [Mon, 24 May 2021 18:48:17 +0000 (18:48 +0000)]
[gn build] Port
b510e4cf1b96
Craig Topper [Mon, 24 May 2021 17:25:27 +0000 (10:25 -0700)]
[RISCV] Add a vsetvli insert pass that can be extended to be aware of incoming VL/VTYPE from other basic blocks.
This is a replacement for D101938 for inserting vsetvli
instructions where needed. This new version changes how
we track the information in such a way that we can extend
it to be aware of VL/VTYPE changes in other blocks. Given
how much it changes the previous patch, I've decided to
abandon the previous patch and post this from scratch.
For now the pass consists of a single phase that assumes
the incoming state from other basic blocks is unknown. A
follow up patch will extend this with a phase to collect
information about how VL/VTYPE change in each block and
a second phase to propagate this information to the entire
function. This will be used by a third phase to do the
vsetvli insertion.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D102737
LLVM GN Syncbot [Mon, 24 May 2021 18:36:50 +0000 (18:36 +0000)]
[gn build] Port
a64ebb863727
Heejin Ahn [Sun, 23 May 2021 09:09:17 +0000 (02:09 -0700)]
[WebAssembly] Add NullifyDebugValueLists pass
`WebAssemblyDebugValueManager` does not currently handle
`DBG_VALUE_LIST`, which is a recent addition to LLVM. We tried to
nullify them within the constructor of `WebAssemblyDebugValueManager` in
D102589, but it made the class error-prone to use because it deletes
instructions within the constructor and thus invalidates existing
iterators within the BB, so the user of the class should take special
care not to use invalidated iterators. This actually caused a bug in
ExplicitLocals pass.
Instead of trying to fix ExplicitLocals pass to make the iterator usage
correct, which is possible but error-prone, this adds
NullifyDebugValueLists pass that nullifies all `DBG_VALUE_LIST`
instructions before we run WebAssembly specific passes in the backend.
We can remove this pass after we implement handlers for
`DBG_VALUE_LIST`s in `WebAssemblyDebugValueManager` and elsewhere.
Fixes https://github.com/emscripten-core/emscripten/issues/14255.
Reviewed By: dschuff
Differential Revision: https://reviews.llvm.org/D102999
George Balatsouras [Fri, 21 May 2021 17:56:45 +0000 (10:56 -0700)]
[dfsan] Add function that prints origin stack trace to buffer
Reviewed By: stephan.yichao.zhao
Differential Revision: https://reviews.llvm.org/D102451
Artem Belevich [Fri, 21 May 2021 17:53:28 +0000 (10:53 -0700)]
[CUDA] Work around compatibility issue with libstdc++ 11.1.0
libstdc++ redeclares __failed_assertion multiple times and that results in the
function declared with conflicting set of attributes when we include <complex>
with __host__ __device__ attributes force-applied to all functions.
In order to work around the issue, we rename __failed_assertion within the
region with forced attributes.
See https://bugs.llvm.org/show_bug.cgi?id=50383 for the details.
Differential Revision: https://reviews.llvm.org/D102936
Stella Laurenzo [Mon, 24 May 2021 16:41:38 +0000 (16:41 +0000)]
Enable MLIR Python bindings for TOSA.
Differential Revision: https://reviews.llvm.org/D103035
Raphael Isemann [Mon, 24 May 2021 17:16:40 +0000 (19:16 +0200)]
[lldb] Add missing mutex guards to TargetList::CreateTarget
TestMultipleTargets is randomly failing on the bots. The reason for that is that
the test is calling `SBDebugger::CreateTarget` from multiple threads.
`TargetList::CreateTarget` is curiously missing the guard that all of its other
member functions have, so all the threads in the test end up changing the
internal TargetList state at the same time and end up corrupting it.
Reviewed By: vsk, JDevlieghere
Differential Revision: https://reviews.llvm.org/D103020
serge-sans-paille [Mon, 24 May 2021 17:43:40 +0000 (19:43 +0200)]
Revert "[NFC] remove explicit default value for strboolattr attribute in tests"
This reverts commit
bda6e5bee04c75b1f1332b4fd1ac4e8ef6c3c247.
See https://lab.llvm.org/buildbot/#/builders/109/builds/15424 for instance
serge-sans-paille [Sun, 23 May 2021 11:19:23 +0000 (13:19 +0200)]
[NFC] remove explicit default value for strboolattr attribute in tests
Since
d6de1e1a71406c75a4ea4d5a2fe84289f07ea3a1, no attributes is quivalent to
setting attribute to false.
This is a preliminary commit for https://reviews.llvm.org/D99080
Craig Topper [Mon, 24 May 2021 17:19:09 +0000 (10:19 -0700)]
[X86] Call insertDAGNode on trunc/zext created in tryShiftAmountMod.
This puts the new nodes in the proper place in the topologically
sorted list of nodes.
Fixes PR50431, which was introduced recently in D101944.
LLVM GN Syncbot [Mon, 24 May 2021 17:18:43 +0000 (17:18 +0000)]
[gn build] Port
095e91c9737b
Vitaly Buka [Sun, 23 May 2021 22:49:43 +0000 (15:49 -0700)]
[NFC][scudo] Small test cleanup
Fixing issues raised on D102979 review.
Reviewed By: cryptoad
Differential Revision: https://reviews.llvm.org/D102994
Jon Roelofs [Mon, 24 May 2021 16:49:32 +0000 (09:49 -0700)]
[Remarks] Add analysis remarks for memset/memcpy/memmove lengths
Re-landing now that the crasher this patch previously uncovered has been fixed
in: https://reviews.llvm.org/D102935
Differential revision: https://reviews.llvm.org/D102452
Roman Lebedev [Mon, 24 May 2021 17:09:04 +0000 (20:09 +0300)]
[X86][Costmodel] getMaskedMemoryOpCost(): don't scalarize non-power-of-two vectors with legal element type
This follows in steps of similar `getMemoryOpCost()` changes, D100099/D100684.
Intel SDM, `VPMASKMOV — Conditional SIMD Integer Packed Loads and Stores`:
```
Faults occur only due to mask-bit required memory accesses that caused the faults. Faults will not occur due to
referencing any memory location if the corresponding mask bit for that memory location is 0. For example, no
faults will be detected if the mask bits are all zero.
```
I.e., if mask is all-zeros, any address is fine.
Masked load/store's prime use-case is e.g. tail masking the loop remainder,
where for the last iteration, only first some few elements of a vector exist.
So much similarly, i don't see why must we scalarize non-power-of-two vectors,
iff the element type is something we can masked- store/load.
We simply need to legalize it, widen the mask, and be done with it.
And we even already count the cost of widening the mask.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D102990
luxufan [Mon, 24 May 2021 16:51:04 +0000 (09:51 -0700)]
[RISCV] Optimize getVLENFactoredAmount function.
If the local variable `NumOfVReg` isPowerOf2_32(NumOfVReg - 1) or isPowerOf2_32(NumOfVReg + 1), the ADDI and MUL instructions can be replaced with SLLI and ADD(or SUB) instructions.
Based on original patch by StephenFan.
Reviewed By: frasercrmck, StephenFan
Differential Revision: https://reviews.llvm.org/D100577
Markus Böck [Mon, 24 May 2021 16:40:39 +0000 (18:40 +0200)]
[mlir][doc] Fix links and references in top level docs directory
This is the fourth and final patch in a series of patches fixing markdown links and references inside the mlir documentation. This patch combined with the other three should fix almost every broken link on mlir.llvm.org as far as I can tell.
This patch in particular addresses all Markdown files in the top level docs directory.
Differential Revision: https://reviews.llvm.org/D103032
Jon Roelofs [Mon, 24 May 2021 16:19:31 +0000 (09:19 -0700)]
[Remarks] Look through inttoptr/ptrtoint for -ftrivial-auto-var-init remarks.
The crasher is a related problem that @aemerson found broke speck2k6/403.gcc
when I landed https://reviews.llvm.org/D102452. It has been reduced & modified
to reproduce without that patch.
Differential revision: https://reviews.llvm.org/D102935
Adrian Prantl [Mon, 24 May 2021 16:06:00 +0000 (09:06 -0700)]
CoroSplit: Replace ad-hoc implementation of reachability with API from CFG.h
The current ad-hoc implementation used to determine whether a basic
block is unreachable doesn't work correctly in the general case (for
example it won't detect successors of unreachable blocks as
unreachable). This patch replaces it with the correct API that uses a
DominatorTree to answer the question correctly and quickly.
rdar://
77181156
Differential Revision: https://reviews.llvm.org/D102963
Steven Wu [Mon, 24 May 2021 16:13:34 +0000 (09:13 -0700)]
[llvm] Revert align attr test in test/Bitcode/attribute-3.3.ll
Revert testcase changed in D87304 now the upgrader can correctly handle
the align attribute in upgrader.
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D102880
Suraj Sudhir [Mon, 24 May 2021 15:47:24 +0000 (15:47 +0000)]
[mlir][tosa] Align tensor rank specifications with current spec
Deconstrains several TOSA operators to align with the current TOSA spec, including all the elementwise ops.
Note: some more ops are under consideration for further cleanup; they will follow once the spec has been updated.
Reviewed By: stellaraccident
Differential Revision: https://reviews.llvm.org/D102958
Kostya Kortchinsky [Wed, 19 May 2021 16:10:30 +0000 (09:10 -0700)]
[scudo] Separate Fuchsia & Default SizeClassMap
The Fuchsia allocator config was using the default size class map.
This CL gives Fuchsia its own size class map and changes a couple of
things in the default one:
- make `SizeDelta` configurable in `Config` for a fixed size class map
as it currently is for a table size class map;
- switch `SizeDelta` to 0 for the default config, it allows for size
classes that allow for power of 2s, and overall better wrt pages
filling;
- increase the max number of caches pointers to 14 in the default,
this makes the transfer batch 64/128 bytes on 32/64-bit platforms,
which is cache-line friendly (previous size was 48/96 bytes).
The Fuchsia size class map remains untouched for now, this doesn't
impact Android which uses the table size class map.
Differential Revision: https://reviews.llvm.org/D102783
Nikita Popov [Mon, 24 May 2021 15:28:38 +0000 (17:28 +0200)]
[CVP] Add additional test for phi common val transform (NFC)
Nikita Popov [Mon, 24 May 2021 13:16:06 +0000 (15:16 +0200)]
[LoopUnroll] Add additional trip multiple test (NFC)
This uses a trip multiple on a (unique) non-latch exit.
Nikita Popov [Mon, 24 May 2021 10:20:47 +0000 (12:20 +0200)]
[LoopUnroll] Regenerate test checks (NFC)
Ivan Murashko [Mon, 24 May 2021 15:21:44 +0000 (11:21 -0400)]
Remark was added to clang tooling Diagnostic
The diff adds Remark to Diagnostic::Level for clang tooling. That makes
Remark diagnostic level ready to use in clang-tidy checks: the
clang-diagnostic-module-import becomes visible as a part of the change.
Simon Pilgrim [Mon, 24 May 2021 14:53:31 +0000 (15:53 +0100)]
[CostModel][X86] Add missing SSE41 v2iX sext/zext costs
Also fix existing v4i8->v4i16 sext cost to match the equivalents
Mark de Wever [Mon, 24 May 2021 14:44:22 +0000 (16:44 +0200)]
[libc++][doc] Update format paper status.
- Fixes paper number P1862 -> P1868. (The title was correct.)
- Marks P1868 as in progress.
- Marks P1892 as in progress.
- Marks LWG-3327 as nothing to do, since the wording change doesn't
impact the code. (Also updated on the general C++20 status page.)
thomasraoux [Mon, 24 May 2021 14:36:29 +0000 (07:36 -0700)]
[NVPTX] Fix lowering of frem for negative values
to match fmod frem result must have the dividend sign. Previous implementation
had the wrong sign when passing negative numbers. For ex: frem(-16, 7) was
returning 5 instead of -2. We should just a ftrunc instead of floor when
lowering to get the right behavior.
Differential Revision: https://reviews.llvm.org/D102528
Simon Pilgrim [Mon, 24 May 2021 14:40:42 +0000 (15:40 +0100)]
[CostModel][X86] Regenerate sse-itoi.ll test checks
Sanjay Patel [Mon, 24 May 2021 14:15:08 +0000 (10:15 -0400)]
[ConstProp] propagate poison from vector reduction element(s) to result
This follows from the underlying logic for binops and min/max.
Although it does not appear that we handle this for min/max
intrinsics currently.
https://alive2.llvm.org/ce/z/Kq9Xnh
Sanjay Patel [Mon, 24 May 2021 13:55:36 +0000 (09:55 -0400)]
[ConstProp] add tests for vector reductions with poison elements; NFC
Florian Hahn [Mon, 24 May 2021 13:14:08 +0000 (14:14 +0100)]
[VPlan] Add first VPlan version of sinkScalarOperands.
This patch adds a first VPlan-based implementation of sinking of scalar
operands.
The current version traverse a VPlan once and processes all operands of
a predicated REPLICATE recipe. If one of those operands can be sunk,
it is moved to the block containing the predicated REPLICATE recipe.
Continue with processing the operands of the sunk recipe.
The initial version does not re-process candidates after other recipes
have been sunk. It also cannot partially sink induction increments at
the moment. The VPlan only contains WIDEN-INDUCTION recipes and if the
induction is used for example in a GEP, only the first lane is used and
in the lowered IR the adds for the other lanes can be sunk into the
predicated blocks.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D100258
Raphael Isemann [Mon, 24 May 2021 14:24:45 +0000 (16:24 +0200)]
[lldb] Readd deleted variable in the sample test
In D102771 wanted to make `test_var` global to demonstrate the a no-launch test,
but the old variable is still needed for another test. This just creates the
global var with a different name to demonstrate the no-launch functionality.
Raphael Isemann [Mon, 24 May 2021 14:01:48 +0000 (16:01 +0200)]
[lldb] Introduce createTestTarget for creating a valid target in API tests
At the moment nearly every test calls something similar to
`self.dbg.CreateTarget(self.getBuildArtifact("a.out"))` and them sometimes
checks if the created target is actually valid with something like
`self.assertTrue(target.IsValid(), "some useless text")`.
Beside being really verbose the error messages generated by this pattern are
always just indicating that the target failed to be created but now why.
This patch introduces a helper function `createTestTarget` to our Test class
that creates the target with the much more verbose `CreateTarget` overload that
gives us back an SBError (with a fancy error). If the target couldn't be created
the function prints out the SBError that LLDB returned and asserts for us. It
also defaults to the "a.out" build artifact path that nearly all tests are using
to avoid to hardcode "a.out" in every test.
I converted a bunch of tests to the new function but I'll do the rest of the
test suite as follow ups.
Reviewed By: JDevlieghere
Differential Revision: https://reviews.llvm.org/D102771
Raphael Isemann [Mon, 24 May 2021 13:01:15 +0000 (15:01 +0200)]
[lldb] Reland "Fix UB in half2float" to fix the ubsan bot.
This relands part of the UB fix in
4b074b49be206306330076b9fa40632ef1960823.
The original commit also added some additional tests that uncovered some
other issues (see D102845). I landed all the passing tests in
48780527dd6820698f3537f5ebf76499030ee349 and this patch is now just fixing
the UB in half2float. See D102846 for a proposed rewrite of the function.
Original commit message:
The added DumpDataExtractorTest uncovered that this is lshifting a negative
integer which upsets ubsan and breaks the sanitizer bot. This patch just
changes the variable we shift to be unsigned.
Anastasia Stulova [Mon, 24 May 2021 13:18:56 +0000 (14:18 +0100)]
[OpenCL][Docs] Minor update to OpenCL 3.0
Simon Pilgrim [Mon, 24 May 2021 12:59:15 +0000 (13:59 +0100)]
[CostModel][X86] Improve accuracy of vector non-uniform shift costs on XOP/AVX2 targets
By llvm-mca analysis, Haswell/Broadwell has a non-uniform vector shift recip-throughput cost of the AVX2 targets at 2 for both 128 and 256-bit vectors - XOP capable targets have better 128-bit vector shifts so improve the fallback in those cases.
Florian Hahn [Mon, 24 May 2021 10:44:13 +0000 (11:44 +0100)]
[VectorCombine] Fix load extract scalarization tests with assumes.
The input IR for @load_extract_idx_var_i64_known_valid_by_assume
and @load_extract_idx_var_i64_not_known_valid_by_assume_after_load
has been swapped.
This patch fixes the test so that @load_extract_idx_var_i64_known_valid_by_assume
has the assume before the load and the other test has it after.
Florian Hahn [Fri, 14 May 2021 22:18:04 +0000 (23:18 +0100)]
[VPlan] Add mayReadOrWriteMemory & friends.
This patch adds initial implementation of mayReadOrWriteMemory,
mayReadFromMemory and mayWriteToMemory to VPRecipeBase.
Used by D100258.
Anastasia Stulova [Mon, 24 May 2021 12:03:32 +0000 (13:03 +0100)]
[OpenCL] Fix test by adding SPIR triple
Bradley Smith [Tue, 11 May 2021 15:39:36 +0000 (16:39 +0100)]
[AArch64][SVE] Add fixed length codegen for FP_ROUND/FP_EXTEND
Depends on D102498
Differential Revision: https://reviews.llvm.org/D102607
Bradley Smith [Fri, 14 May 2021 11:22:52 +0000 (12:22 +0100)]
[AArch64][SVE] Improve codegen for fixed length vector concat
Differential Revision: https://reviews.llvm.org/D102498
Anastasia Stulova [Mon, 24 May 2021 11:38:02 +0000 (12:38 +0100)]
[OpenCL] Add clang extension for bit-fields.
Allow use of bit-fields as a clang extension
in OpenCL. The extension can be enabled using
pragma directives.
This fixes PR45339!
Differential Revision: https://reviews.llvm.org/D101843
David Green [Mon, 24 May 2021 11:22:15 +0000 (12:22 +0100)]
[ARM] Allow findLoopPreheader to return headers with multiple loop successors
The findLoopPreheader function will currently not find a preheader if it
branches to multiple different loop headers. This patch adds an option
to relax that, allowing ARMLowOverheadLoops to process more loops
successfully. This helps with WhileLoopStart setup instructions that can
branch/fallthrough to the low overhead loop and to branch to a separate
loop from the same preheader (but I don't believe it is possible for
both loops to be low overhead loops).
Differential Revision: https://reviews.llvm.org/D102747
Florian Hahn [Mon, 24 May 2021 09:11:38 +0000 (10:11 +0100)]
Recommit "[VectorCombine] Scalarize vector load/extract."
This reverts commit
94d54155e2f38b56171811757044a3e6f643c14b.
This fixes a sanitizer failure by moving scalarizeLoadExtract(I)
before foldSingleElementStore(I), which may remove instructions.
David Green [Mon, 24 May 2021 10:26:45 +0000 (11:26 +0100)]
[ARM] Ensure WLS preheader blocks have branches during memcpy lowering
This makes sure that the blocks created for lowering memcpy to loops end
up with branches, even if they fall through to the successor. Otherwise
IfCvt is getting confused with unanalyzable branches and creating
invalid block layouts.
The extra branches should be removed as the tail predicated loop is
finalized in almost all cases.
David Green [Mon, 24 May 2021 10:01:58 +0000 (11:01 +0100)]
[ARM] Fix inline memcpy trip count sequence
The trip count for a memcpy/memset will be n/16 rounded up to the
nearest integer. So (n+15)>>4. The old code was including a BIC too, to
clear one of the bits, which does not seem correct. This remove the
extra BIC.
Note that ideally this would never actually be generated, as in the
creation of a tail predicated loop we will DCE that setup code, letting
the WLSTP perform the trip count calculation. So this doesn't usually
come up in testing (and apparently the ARMLowOverheadLoops pass does not
do any sort of validation on the tripcount). Only if the generation of
the WLTP fails will it use the incorrect BIC instructions.
Differential Revision: https://reviews.llvm.org/D102629
Uday Bondhugula [Mon, 24 May 2021 03:20:17 +0000 (08:50 +0530)]
[MLIR] Drop old cmake var names
Drop old cmake variable names that were kept around so that zorg
buildbot could be migrated, which has now happened (D102977). D102976
had fixed the inconsistent names.
Differential Revision: https://reviews.llvm.org/D102997
Fraser Cormack [Fri, 21 May 2021 12:00:19 +0000 (13:00 +0100)]
[RISCV] Prevent store combining from infinitely looping
RVV code generation does not successfully custom-lower BUILD_VECTOR in all
cases. When it resorts to default expansion it may, on occasion, be expanded to
scalar stores through the stack. Unfortunately these stores may then be picked
up by the post-legalization DAGCombiner which merges them again. The merged
store uses a BUILD_VECTOR which is then expanded, and so on.
This patch addresses the issue by overriding the `mergeStoresAfterLegalization`
hook. A lack of granularity in this method (being passed the scalar type) means
we opt out in almost all cases when RVV fixed-length vector support is enabled.
The only exception to this rule are mask vectors, which are always either
custom-lowered or are expanded to a load from a constant pool.
Reviewed By: HsiangKai
Differential Revision: https://reviews.llvm.org/D102913
James Henderson [Tue, 18 May 2021 09:57:32 +0000 (10:57 +0100)]
[debuginfo-tests] Stop using installed LLDB and remove redundancy
The removed code just replicated what use_llvm_tool does, plus looked
for an installed LLDB on the PATH to use. In a monorepo world, it seems
likely that if people want to run the tests that require LLDB, they
should enable and build LLDB itself. If users really want to use the
installed LLDB executable, they can specify the path to the executable
as an environment variable "LLDB".
See the discussion in https://reviews.llvm.org/D95339#2638619 for
more details.
Reviewed by: jmorse, aprantl
Differential Revision: https://reviews.llvm.org/D102680
Roman Lebedev [Mon, 24 May 2021 09:13:05 +0000 (12:13 +0300)]
[NFCI][LoopIdiom] 'left-shift until bittest': assert that BaseX is loop-invariant
Given that BaseX is an incoming value when coming from the preheader,
it *should* be loop-invariant, but let's just document this assumption.
Roman Lebedev [Mon, 24 May 2021 09:08:30 +0000 (12:08 +0300)]
[LoopIdiom] 'logical right shift until zero': the value must be loop-invariant
As per the reproducer provided by Mikael Holmén in post-commit review.
Thorsten Schütt [Mon, 24 May 2021 09:11:52 +0000 (11:11 +0200)]
flang: include limits
Florian Hahn [Mon, 24 May 2021 09:11:00 +0000 (10:11 +0100)]
Revert "[VectorCombine] Scalarize vector load/extract."
This reverts commit
86497785d540e59eaca24bed4219ddec183cbc9b.
One of the tests causes an ASAN failure.
https://lab.llvm.org/buildbot/#/builders/5/builds/7927/steps/12/logs/stdio
Simon Pilgrim [Sun, 23 May 2021 21:50:45 +0000 (22:50 +0100)]
[CostModel][X86] Improve accuracy of vXi64 MUL costs on AVX2/AVX512 targets
By llvm-mca analysis, Haswell/Broadwell has the worst v4i64 recip-throughput cost of the AVX2 targets at 6 (vs the currently used cost of 8). Similarly SkylakeServer (our only AVX512 target model) implements PMULLQ with an average cost of 1.5 (rounded up to 2.0), and the PMULUDQ-sequence (without AVX512DQ) as a cost of 6.
Pushpinder Singh [Fri, 21 May 2021 08:03:26 +0000 (08:03 +0000)]
[AMDGPU][Libomptarget] Remove global KernelNameMap
KernelNameMap contains entries like "key.kd" => key which clearly
could be replaced by simple logic of removing suffix from the key.
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D102691
Chen Zheng [Mon, 24 May 2021 08:32:23 +0000 (04:32 -0400)]
[Debug-Info]update section name to match AIX behaviour; nfc
Florian Hahn [Mon, 24 May 2021 08:19:40 +0000 (09:19 +0100)]
[VectorCombine] Scalarize vector load/extract.
This patch adds a new combine that tries to scalarize chains of
`extractelement (load %ptr), %idx` to `load (gep %ptr, %idx)`. This is
profitable when extracting only a few elements out of a large vector.
At the moment, `store (extractelement (load %ptr), %idx), %ptr`
operations on large vectors result in huge code in the backend.
This can easily be triggered by using the matrix extension, e.g.
https://clang.godbolt.org/z/qsccPdPf4
This should complement D98240.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D100273
Tomasz Kamiński [Mon, 24 May 2021 08:16:52 +0000 (10:16 +0200)]
[analyzer] Correctly propagate ConstructionContextLayer thru ParenExpr
Previously, information about `ConstructionContextLayer` was not
propagated thru causing the expression like:
Var c = (createVar());
To produce unrelated temporary for the `createVar()` result and conjure
a new symbol for the value of `c` in C++17 mode.
Reviewed By: steakhal
Patch By: tomasz-kaminski-sonarsource!
Differential Revision: https://reviews.llvm.org/D102835
Johannes Doerfert [Mon, 10 May 2021 01:02:18 +0000 (20:02 -0500)]
[Attributor] Introduce a helper do deal with constant type mismatches
If we simplify values we sometimes end up with type mismatches. If the
value is a constant we can often cast it though to still allow
propagation. The logic is now put into a helper and it replaces some
ad hoc things we did before.
This also introduces the AA namespace for abstract attribute related
functions and types.
Johannes Doerfert [Sat, 8 May 2021 04:05:40 +0000 (23:05 -0500)]
[Attributor] Teach AAIsDead about undef values
Not only if the branch or switch condition is dead but also if it is
assumed `undef` we can delay AAIsDead exploration.
Johannes Doerfert [Thu, 6 May 2021 15:21:16 +0000 (10:21 -0500)]
[Attributor] Deal with address spaces gracefully
When we do value propagation we need to cast address spaces properly.
Johannes Doerfert [Sun, 16 May 2021 01:19:11 +0000 (20:19 -0500)]
[Attributor] Be more careful to not disturb the CG outside the SCC
We have seen various problems when the call graph was not updated or
the updated did not succeed because it involved functions outside the
SCC. This patch adds assertions and checks to avoid accidentally
changing something outside the SCC that would impact the call graph.
It also prevents us from reanalyzing functions outside the current
SCC which could cause problems on its own. Note that the transformations
we do might cause the CG to be "more precise" but the original one would
always be a super set of the most precise one. Since the call graph is
by nature an approximation, it is good enough to have a super set of all
call edges.
John Demme [Mon, 24 May 2021 03:37:55 +0000 (20:37 -0700)]
[MLIR] [Python] Add Operation.parent
Attribute to get the parent operation of an operation.
Reviewed By: stellaraccident
Differential Revision: https://reviews.llvm.org/D102981
Alexander Shaposhnikov [Mon, 24 May 2021 03:35:55 +0000 (20:35 -0700)]
[lld][MachO] Fix code formatting
Apply clang-format -style=llvm to InputFile.cpp. NFC.
Test plan: make check-all
Uday Bondhugula [Sat, 22 May 2021 14:25:38 +0000 (19:55 +0530)]
[MLIR] Make MLIR cmake variable names consistent
Fix inconsistent MLIR CMake variable names. Consistently name them as
MLIR_ENABLE_<feature>.
Eg: MLIR_CUDA_RUNNER_ENABLED -> MLIR_ENABLE_CUDA_RUNNER
MLIR follows (or has mostly followed) the convention of naming
cmake enabling variables in the from MLIR_ENABLE_... etc. Using a
convention here is easy and also important for convenience. A counter
pattern was started with variables named MLIR_..._ENABLED. This led to a
sequence of related counter patterns: MLIR_CUDA_RUNNER_ENABLED,
MLIR_ROCM_RUNNER_ENABLED, etc.. From a naming standpoint, the imperative
form is more meaningful. Additional discussion at:
https://llvm.discourse.group/t/mlir-cmake-enable-variable-naming-convention/3520
Switch all inconsistent ones to the ENABLE form. Keep the couple of old
mappings needed until buildbot config is migrated.
Differential Revision: https://reviews.llvm.org/D102976
Haruki Imai [Mon, 24 May 2021 03:04:45 +0000 (08:34 +0530)]
[mlir] Normalize dynamic memrefs with a map of tiled-layout.
Steps for normalizing dynamic memrefs for tiled layout map
1. Check if original map is tiled layout. Only tiled layout is supported.
2. Create normalized memrefType. Dimensions that include dynamic dimensions
in the map output will be dynamic dimensions.
3. Create new maps to calculate each dimension size of new memref.
In tiled layout, the dimension size can be calculated by replacing
"floordiv <tile size>" with "ceildiv <tile size>" and
"mod <tile size>" with "<tile size>".
4. Create AffineApplyOp to apply the new maps. The output of AffineApplyOp is
dynamicSizes for new AllocOp.
5. Add the new dynamic sizes in new AllocOp.
This patch also set MemRefsNormalizable trant in CastOp and DimOp since
they used with dynamic memrefs.
Reviewed By: bondhugula
Differential Revision: https://reviews.llvm.org/D97655
Johannes Doerfert [Sat, 8 May 2021 04:45:07 +0000 (23:45 -0500)]
[Attributor][FIX] Account for undef in the constant value lattice
The constant value lattice looks like this
```
<None>
|
<undef>
/ | \
... <0> ...
\ | /
<unknown>
```
We did not account for the undef and assumed a value meant we could not
change anymore. Now we actually check if we have the same value as
before, which will signal CHANGED to the users when we go from undef to
a specific constant.
This fixes, among other things, the bug exposed by @ipccp4 in
`value-simplify.ll`.
Johannes Doerfert [Thu, 6 May 2021 21:51:37 +0000 (16:51 -0500)]
[Attributor][FIX] Ensure we replace undef if we see the first "real" value
The state of AAPotentialValues tracks if undef is contained. It should
fold undef into the first non-undef value. However we missed a case
before. There was also a shadowing definition of two variables that
caused trouble. The test exposes both problems.
Johannes Doerfert [Fri, 21 May 2021 17:43:15 +0000 (12:43 -0500)]
[Attributor][NFC] Precommit test case with branch on undef
This test exposes a bug in the module pass as it simplifies ipccp4 to
unreachable, which is unfortunately wrong.
Johannes Doerfert [Sun, 16 May 2021 01:16:39 +0000 (20:16 -0500)]
[Attributor][NFC] Add helpful debug outputs
Johannes Doerfert [Thu, 6 May 2021 16:27:06 +0000 (11:27 -0500)]
[Attributor][NFC] Clang format the Attributor source files
Johannes Doerfert [Sun, 16 May 2021 01:32:59 +0000 (20:32 -0500)]
[Attributor][NFC] Rerun update_test_checks script on Attributor tests
Chen Zheng [Thu, 20 May 2021 09:52:34 +0000 (05:52 -0400)]
[Debug-Info] handle DW_TAG_rvalue_reference_type at strict DWARF.
When -gstrict-dwarf is specified, generate DW_TAG_rvalue_reference_type
at DWARF 4 or above
Reviewed By: dblaikie, aprantl
Differential Revision: https://reviews.llvm.org/D100630
Fady Ghanim [Sun, 23 May 2021 23:13:09 +0000 (19:13 -0400)]
[NFC] Removing leftover debug code
Removing a missed value::dump() used to debug during development of
OMPBuilder atomic.
Fangrui Song [Sun, 23 May 2021 22:20:56 +0000 (15:20 -0700)]
[AArch64] Delete unneeded fixup_aarch64_ldr_pcrel_imm19 VK_GOT special case
An AArch64 VK_GOT fixup must have a symbol. MCAssembler::evaluateFixup considers
such a fixup not resolved. The code path cannot trigger.
Fady Ghanim [Thu, 6 May 2021 21:23:28 +0000 (17:23 -0400)]
[OpenMP][OMPIRBuilder]Adding support for `omp atomic`
This patch adds support for generating `omp atomic` for all different
atomic clauses
Vitaly Buka [Sun, 23 May 2021 21:12:49 +0000 (14:12 -0700)]
[NFC][scudo] Enforce header size alignment
As-is it should not change struct size, but it will
help to keep correct size if more fields added.
Philipp Krones [Sun, 23 May 2021 21:15:23 +0000 (14:15 -0700)]
[MC] Refactor MCObjectFileInfo initialization and allow targets to create MCObjectFileInfo
This makes it possible for targets to define their own MCObjectFileInfo.
This MCObjectFileInfo is then used to determine things like section alignment.
This is a follow up to D101462 and prepares for the RISCV backend defining the
text section alignment depending on the enabled extensions.
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D101921
Joerg Sonnenberger [Sun, 23 May 2021 20:55:45 +0000 (22:55 +0200)]
[libc++] use more early returns for consistency
Reviewed By: #libc, ldionne
Differential Revision: https://reviews.llvm.org/D96983
Nikita Popov [Sun, 23 May 2021 21:08:32 +0000 (23:08 +0200)]
[LoopUnroll] Add test for partial unrolling again non-latch exit (NFC)
This test case would get miscompiled by the current version of
D102982, because unrolling does not respect the PreserveCondBr
flag for partial unrolling.
Chris Lattner [Sun, 23 May 2021 21:08:31 +0000 (14:08 -0700)]
[IR] Add a Location to BlockArgument
This adds the ability to specify a location when creating BlockArguments.
Notably Value::getLoc() will return this correctly, which makes diagnostics
more precise (e.g. the example in test-legalize-type-conversion.mlir).
This is currently optional to avoid breaking any existing code - if
absent, the BlockArgument defaults to using the location of its enclosing
operation (preserving existing behavior).
The bulk of this change is plumbing location tracking through the parser
and printer to make sure it can round trip (in -mlir-print-debuginfo
mode). This is complete for generic operations, but requires manual
adoption for custom ops.
I added support for function-like ops to round trip their argument
locations - they print correctly, but when parsing the locations are
dropped on the floor. I intend to fix this, but it will require more
invasive plumbing through "function_like_impl" stuff so I think it
best to split it out to its own patch.
This is a reapply of the patch here: https://reviews.llvm.org/D102567
with an additional change: we now never defer block argument locations,
guaranteeing that we can round trip correctly.
This isn't required in all cases, but allows us to hill climb here and
works around unrelated bugs like https://bugs.llvm.org/show_bug.cgi?id=50451
Differential Revision: https://reviews.llvm.org/D102991
Fangrui Song [Sun, 23 May 2021 20:58:16 +0000 (13:58 -0700)]
[AArch64][MC] Remove unneeded "in .xxx directive" from diagnostics
The prevailing style does not add the message. The directive name is not useful
because the next line replicates the error line which includes the directive.
Joerg Sonnenberger [Sun, 23 May 2021 20:52:59 +0000 (22:52 +0200)]
[SPARC] recognize the "rd %pc, reg" special form
Differential Revision: https://reviews.llvm.org/D96312
Sander de Smalen [Sun, 23 May 2021 13:41:13 +0000 (14:41 +0100)]
NFC: cleaned up and renamed scalable-vf-analysis.ll -> scalable-vectorization.ll
* Removes unnecessary loop hints.
* Use RUN line with '-scalable-vectorization=preferred' instead of 'on'
for the maximize-bandwidth behaviour. This prepares the test for enabling
scalable vectorization; With a forced instruction-cost of 1, 'on' will
always favour fixed-width VF to be chosen, whereas with 'preferred'
we can check that the maximize-bandwidth option in combination with
scalable-vectorization=preferred actually picks a scalable VF.
* Renamed to scalable-vectorization.ll, because a follow-up patch will
test more than just analysis.
Roman Lebedev [Sun, 23 May 2021 18:42:30 +0000 (21:42 +0300)]
[NFC][X86][Costmodel] Add tests with with masked loads/stores w/non-power-of-two vectors
Fangrui Song [Sun, 23 May 2021 18:35:42 +0000 (11:35 -0700)]
[AArch64] Use \t in AsmStreamer to match the prevailing style
Markus Böck [Sun, 23 May 2021 18:00:44 +0000 (20:00 +0200)]
[mlir][doc] Fix links and indentation of mlir::ModuleOp description
All lines after the first are currently indented by one char further to the left than the first line. This leads to the first character of each sentence being cut from the resulting Markdown file after compilation. The text also contains 3 references to sections of other markdown files. One was missing the file, while the other two had outdated files, leading to 404 errors in the documentation.
Differential Revision: https://reviews.llvm.org/D102983
Simon Pilgrim [Sun, 23 May 2021 16:19:36 +0000 (17:19 +0100)]
Fix bugs URL for PR relocations
The PR works from llvm.org, not bugs.llvm.org
Simon Pilgrim [Sun, 23 May 2021 09:29:34 +0000 (10:29 +0100)]
[CostModel][X86] Align v2i64 MUL costs on SSE42+ targets with worst case
Based on worst case of sandybridge (which seems to match nehalem for this SSE sequence) (vs btver2 + bdver2) llvm-mca analysis
Nico Weber [Sun, 23 May 2021 14:01:06 +0000 (10:01 -0400)]
[gn build] (semi-manually) port
0bccdf82f705
Sanjay Patel [Sun, 23 May 2021 13:42:35 +0000 (09:42 -0400)]
[InstSimplify] add more tests for rem-mul-div; NFC
See D102864 for discussion.
maekawatoshiki [Sun, 23 May 2021 13:32:01 +0000 (22:32 +0900)]
[LoopUnrollAndJam] Change LoopUnrollAndJamPass to LoopNest pass
This patch changes LoopUnrollAndJamPass from FunctionPass to LoopNest pass.
The next patch will utilize LoopNest to effectively handle loop nests.
Reviewed By: Whitney
Differential Revision: https://reviews.llvm.org/D99149
Nikita Popov [Sun, 23 May 2021 08:49:11 +0000 (10:49 +0200)]
[LoopUnroll] Add test for unrollable non-latch multi-exit (NFC)
This test case requires unrolling against a non-latch exit in
a multiple-exit loop with exiting latch. It's not covered by
exiting heuristics or the extension in D102635.
David Green [Sun, 23 May 2021 07:52:13 +0000 (08:52 +0100)]
[ARM] Add extra debug messages for gather/scatter lowering. NFC
Vitaly Buka [Sun, 23 May 2021 05:55:53 +0000 (22:55 -0700)]
[NFC][scudo] Replace size_t with uptr
Vitaly Buka [Sun, 23 May 2021 05:30:03 +0000 (22:30 -0700)]
[NFC][scudo] Add releasePagesToOS test