Fangrui Song [Tue, 31 Dec 2019 23:00:59 +0000 (15:00 -0800)]
[ELF][RISCV][test] Test absolute/PC-relative/branch relocations to undefined weak symbols
Hideto Ueno [Wed, 1 Jan 2020 06:25:19 +0000 (15:25 +0900)]
[Attributor] AAValueConstantRange: Value range analysis using constant range
This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point.
One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument).
The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty).
Currently, AAValueConstantRange is created when AAValueSimplify cannot
simplify the value.
Supported
- BinaryOperator(add, sub, ...)
- CmpInst(icmp eq, ...)
- !range metadata
`AAValueConstantRange` is not intended to extend to polyhedral range value analysis.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D71620
Craig Topper [Wed, 1 Jan 2020 05:40:58 +0000 (21:40 -0800)]
[X86] Fix typo in getCMovOpcode.
The 64-bit HasMemoryOperand line was using CMOV32rm instead of
CMOV64rm. Not sure how to test this. We have no test coverage
that passes true for HasMemoryOperand.
Craig Topper [Wed, 1 Jan 2020 04:29:41 +0000 (20:29 -0800)]
[X86] Add X87 FCMOV support to X86FlagsCopyLowering.
Fixes PR44396
Matt Arsenault [Mon, 30 Dec 2019 19:46:24 +0000 (14:46 -0500)]
DAG: Stop trying to fold FP -(x-y) -> y-x in getNode with nsz
This was increasing the number of instructions when fsub was legalized
on AMDGPU with no signed zeros enabled. This fold should be guarded by
hasOneUse, and I don't think getNode should be doing that. The same
fold is already done as a regular combine through isNegatibleForFree.
This does require duplicating, even though isNegatibleForFree does
this combine already (and properly checks hasOneUse) to avoid one PPC
regression. In the regression, the outer fneg has nsz but the fsub
operand does not. isNegatibleForFree only sees the operand, and
doesn't see it's used from a nsz context. A nsz parameter needs to be
added and threaded through isNegatibleForFree to avoid this.
Fangrui Song [Tue, 31 Dec 2019 23:06:31 +0000 (15:06 -0800)]
[ELF][RISCV] Improve error message for unknown relocations
Like rLLD354040.
Craig Topper [Tue, 31 Dec 2019 23:57:17 +0000 (15:57 -0800)]
[X86] Constant fold KSHIFT of an all zeros vector to just an all zeros vector.
Craig Topper [Tue, 31 Dec 2019 23:06:47 +0000 (15:06 -0800)]
[X86][InstCombine] Add constant folding and simplification support for pdep and pext
The instructions use a mask to either pack disjoint bits together(pext) or spread bits to disjoint locations(pdep). If the mask is all 0s then no bits are extracted or deposited. If the mask is all ones, then the source value is written to the result since no compression or expansion happens. Otherwise if both the source and mask are constant we can walk the bits in the source/mask and calculate the result.
There other crazier things we could do like computeKnownBits or turning pext into shift/and if only a single contiguous range of bits is extracted.
Fixes PR44389
Differential Revision: https://reviews.llvm.org/D71952
Craig Topper [Tue, 31 Dec 2019 23:05:21 +0000 (15:05 -0800)]
[X86] Use carry flag from add for (seteq (add X, -1), -1).
If we just subtracted 1 and are checking if the result is -1. We can use the carry flag from the ADD instead of an explicit CMP. I'm using the same checks for the add users as EmitTest.
Fixes one case from PR44412
Differential Revision: https://reviews.llvm.org/D72019
Craig Topper [Tue, 31 Dec 2019 22:04:09 +0000 (14:04 -0800)]
[LegalizeVectorOps][AArch64] Stop asking for v4f16 fp_round and fp_extend to be promoted.
These operations are needed as building blocks for promoting so they
can't be promoted themselves.
This appeared to work because the fp_extend query type for operation
actions is the result type, not the input type so it never triggered
in the legalizer.
For fp_round, the vector op legalizer just ended up creating a
nop fp_extend that was elided by getNode, followed by a nop
fp_round that was also elided by getNode. This was followed by
a final fp_round from v4f32 back to vf416 which was CSEd to the
original node. Then legalize vector ops just believed that node
legalized to itself. LegalizeDAG took another crack at promoting
it, but didn't have a handler so just skipped it with a debug
message saying it wasn't promoted.
This patch just removes the operation actions to avoid this
non-sense. Found while trying to refactor LegalizeVectorOps to
handle multiple result nodes better.
Matt Arsenault [Tue, 31 Dec 2019 18:51:52 +0000 (13:51 -0500)]
AMDGPU: Precommit test showing extra instructions are introduced
Martin Storsjö [Tue, 31 Dec 2019 19:38:43 +0000 (21:38 +0200)]
Revert "[Diagnostic] Add ftabstop to -Wmisleading-indentation"
This reverts commit
b47b35ff51b355a446483777155290541ab64fae.
This caused failed asserts (Assertion `FIDAndOffset.second >
ColNo && "Column number smaller than file offset?"' failed.)
on a source file with a single line containing
"int main (void) { for( int i = 0; i < 9; i++ ); return 0; }".
Michael Liao [Wed, 25 Dec 2019 05:47:18 +0000 (00:47 -0500)]
[amdgpu] Fix scoreboard updating on `s_waitcnt_vscnt`.
Summary: - Other counters are accidentally cleared.
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D71866
Alexey Bataev [Tue, 31 Dec 2019 17:41:57 +0000 (12:41 -0500)]
[OPENMP]Emit artificial threprivate vars as threadlocal, if possible.
It may improve performance for declare reduction constructs.
Jacques Pienaar [Tue, 31 Dec 2019 18:56:24 +0000 (10:56 -0800)]
[mlir][docs] Remove redundant path prefix
./ is not needed.
Craig Topper [Tue, 31 Dec 2019 18:57:58 +0000 (10:57 -0800)]
[X86] Add test case for opposite branch condition for PR44412. NFC
Craig Topper [Tue, 31 Dec 2019 18:33:30 +0000 (10:33 -0800)]
[CodeGen] Emit conj/conjf/confjl libcalls as fneg instructions if possible.
We already recognize the __builtin versions of these, might as well
recognize the libcall version.
Differential Revision: https://reviews.llvm.org/D72028
Jacques Pienaar [Tue, 31 Dec 2019 17:52:19 +0000 (09:52 -0800)]
[mlir] Make code blocks more consistent
Use the same form specification for the same type of code.
Sanjay Patel [Tue, 31 Dec 2019 17:29:53 +0000 (12:29 -0500)]
[InstCombine] fold zext of masked bit set/clear
This does not solve PR17101, but it is one of the
underlying diffs noted here:
https://bugs.llvm.org/show_bug.cgi?id=17101#c8
We could ease the one-use checks for the 'clear'
(no 'not' op) half of the transform, but I do not
know if that asymmetry would make things better
or worse.
Proofs:
https://rise4fun.com/Alive/uVB
Name: masked bit set
%sh1 = shl i32 1, %y
%and = and i32 %sh1, %x
%cmp = icmp ne i32 %and, 0
%r = zext i1 %cmp to i32
=>
%s = lshr i32 %x, %y
%r = and i32 %s, 1
Name: masked bit clear
%sh1 = shl i32 1, %y
%and = and i32 %sh1, %x
%cmp = icmp eq i32 %and, 0
%r = zext i1 %cmp to i32
=>
%xn = xor i32 %x, -1
%s = lshr i32 %xn, %y
%r = and i32 %s, 1
Sanjay Patel [Tue, 31 Dec 2019 17:00:07 +0000 (12:00 -0500)]
[InstCombine] add/adjust tests for masked bit; NFC
Stephen Kelly [Sun, 29 Dec 2019 19:21:57 +0000 (19:21 +0000)]
Implement additional traverse() overloads
Summary:
These overloads make it possible to wrap unless(), anyOf(), has() etc
with the traverse matcher.
Reviewers: aaron.ballman
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D71977
Stephen Kelly [Sun, 29 Dec 2019 19:19:55 +0000 (19:19 +0000)]
Unnest struct in Matcher implementation
This allows implementation of the traverse() matcher to surround
matchers like unless().
Stephen Kelly [Sun, 29 Dec 2019 19:19:35 +0000 (19:19 +0000)]
Match code following lambdas when ignoring invisible nodes
Reviewers: aaron.ballman
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D71976
Johannes Doerfert [Tue, 31 Dec 2019 16:55:07 +0000 (10:55 -0600)]
[Attributor][Fix] Avoid leaking memory after D68765
Nikita Popov [Tue, 31 Dec 2019 16:42:38 +0000 (17:42 +0100)]
Revert "[InstCombine] Fix infinite loop due to bitcast <-> phi transforms"
This reverts commit
27a0795943fee0f30b995fe5165428afc2dfd402.
Seems to break test-suite.
Jinsong Ji [Tue, 31 Dec 2019 16:24:35 +0000 (16:24 +0000)]
[PowerPC][NFC] Fix clang-tidy warning
Reported by
https://results.llvm-merge-guard.org/amd64_debian_testing_clang8-726/clang-tidy.txt
/mnt/disks/ssd0/agent/workspace/amd64_debian_testing_clang8/llvm/lib/Target/PowerPC/PPCISelLowering.cpp:11672:10:
warning: invalid case style for variable 'isEQ'
[readability-identifier-naming]
bool isEQ = (MI.getOpcode() == PPC::ANDI_rec_1_EQ_BIT ||
^~~~
IsEq
/mnt/disks/ssd0/agent/workspace/amd64_debian_testing_clang8/llvm/lib/Target/PowerPC/PPCISelLowering.cpp:11679:14:
warning: invalid case style for variable 'dl'
[readability-identifier-naming]
DebugLoc dl = MI.getDebugLoc();
^~
Dl
Sven van Haastregt [Tue, 31 Dec 2019 15:30:02 +0000 (15:30 +0000)]
[OpenCL] Remove redundant foreach in OpenCLBuiltins.td; NFC
Remove various `foreach` declarations where the iterator is used only
once. This makes the .td file more compact.
Sanjay Patel [Tue, 31 Dec 2019 15:25:41 +0000 (10:25 -0500)]
[AArch64] add test for fsub+fneg; NFC
D72015 proposes to restrict the current behavior.
Sanjay Patel [Mon, 30 Dec 2019 19:56:57 +0000 (14:56 -0500)]
[InstCombine] add tests for masked bit set/clear; NFC
Nikita Popov [Sat, 7 Dec 2019 11:21:22 +0000 (12:21 +0100)]
[InstCombine] Fix infinite loop due to bitcast <-> phi transforms
Fix for https://bugs.llvm.org/show_bug.cgi?id=44245.
The optimizeBitCastFromPhi() and FoldPHIArgOpIntoPHI() end up
fighting against each other, because optimizeBitCastFromPhi()
assumes that bitcasts of loads will get folded. This doesn't happen
here, because a dangling phi node prevents the one-use fold in
https://github.com/llvm/llvm-project/blob/master/llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp#L620-L628 from triggering.
This patch fixes the issue by adding manually removing the old phis.
Differential Revision: https://reviews.llvm.org/D71164
Nicolas Vasilache [Mon, 30 Dec 2019 20:41:12 +0000 (15:41 -0500)]
[mlir][Linalg] Delete unused LinalgLibraryOps.td
Summary: This has been previously renamed to LinalgStructuredOps.td
Reviewers: ftynse
Subscribers: mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, llvm-commits, ftynse
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D72013
Miloš Stojanović [Tue, 31 Dec 2019 13:14:41 +0000 (14:14 +0100)]
[llvm-exegesis] Check counters before running
Check if the appropriate counters for the specified mode are defined on
the target. This is checked before any other work is done.
Differential Revision: https://reviews.llvm.org/D71927
Sam Parker [Tue, 31 Dec 2019 11:31:06 +0000 (11:31 +0000)]
[ARM][TypePromotion] Re-enable by default
Re-enable the pass after it was reverted and the bug fixed.
Michael Platings [Mon, 30 Dec 2019 08:21:42 +0000 (08:21 +0000)]
Fix external-names.c test when separator is \\
This fixes the following failure:
C:\[...]\llvm\tools\clang\test\VFS\external-names.c:34:26: error: CHECK-DEBUG-EXTERNAL: expected string not found in input
// CHECK-DEBUG-EXTERNAL: ![[Num]] = !DIFile(filename: "{{[^"]*}}Inputs{{.}}external-names.h"
^
[...]
<stdin>:42:54: note: possible intended match here
!10 = !DIFile(filename: "C:/[...]\\llvm\\tools\\clang\\test\\VFS\\Inputs\\external-names.h", directory: "")
Differential Revision: https://reviews.llvm.org/D71991
Connor Abbott [Tue, 31 Dec 2019 11:11:35 +0000 (12:11 +0100)]
[InstCombine] Don't rewrite phi-of-bitcast when the phi has other users
Judging by the existing comments, this was the intention, but the
transform never actually checked if the existing phi's would be removed.
See https://bugs.llvm.org/show_bug.cgi?id=44242 for an example where
this causes much worse code generation on AMDGPU.
Differential Revision: https://reviews.llvm.org/D71209
Connor Abbott [Tue, 31 Dec 2019 11:10:06 +0000 (12:10 +0100)]
[InstCombine] Add tests for PR44242
Differential Revision: https://reviews.llvm.org/D71260
Sylvestre Ledru [Tue, 31 Dec 2019 10:56:17 +0000 (11:56 +0100)]
clang-tidy doc: modernize-make-unique has an autofix
Ilya Biryukov [Tue, 31 Dec 2019 09:21:52 +0000 (10:21 +0100)]
[Attributor] Suppress unused warnings when assertions are disabled. NFC
Johannes Doerfert [Fri, 11 Oct 2019 01:23:17 +0000 (20:23 -0500)]
[Utils] Deal with occasionally deleted functions
When functions exist for some but not all run lines we need to be
careful when selecting the prefix. So far, a common prefix was
potentially chosen as there was never a "conflict" that would have
caused otherwise. With this patch we avoid common prefixes if they
are used by run lines that do not emit the function.
Reviewed By: lebedev.ri
Differential Revision: https://reviews.llvm.org/D68850
Johannes Doerfert [Thu, 10 Oct 2019 06:39:16 +0000 (01:39 -0500)]
[Attributor] Function signature rewrite infrastructure
As part of the Attributor manifest we want to change the signature of
functions. This patch introduces a fairly generic interface to do so.
As a first, very simple, use case, we remove unused arguments. A second
use case, pointer privatization, will be committed with this patch as
well.
A lot of the code and ideas are taken from argument promotion and we
run all argument promotion tests through this framework as well.
Reviewed By: uenoku
Differential Revision: https://reviews.llvm.org/D68765
Craig Topper [Tue, 31 Dec 2019 07:41:35 +0000 (23:41 -0800)]
[X86] Slightly improve our attempted error recovery for 64-bit -mno-sse2 in LowerCallResult to use FP1 if there are two return values.
If the return value is a struct of 2 doubles we need two return
registers.
If SSE2 is disabled we can't return in XMM registers like the ABI says.
After logging an error we attempt to recover by using FP0 instead
of an XMM register. But if the return needs two registers, we may have
already used FP0. So if the register we were supposed to copy to is
XMM1, copy to FP1 in the recovery instead.
This seems to fix the assertion/crash in PR44413.
Johannes Doerfert [Tue, 31 Dec 2019 08:11:30 +0000 (02:11 -0600)]
[Utils][Fix] Minor test result change
Shengchen Kan [Tue, 31 Dec 2019 08:08:05 +0000 (16:08 +0800)]
[NFC] Style cleanup
1. make function Is16BitMemOperand static
2. Use Doxygen features in comment
3. Rename functions to make them start with a lower case letter
Johannes Doerfert [Fri, 1 Nov 2019 17:51:26 +0000 (12:51 -0500)]
[Utils] Reuse argument variable names in the body
If we have `int foo(int a) { return a; }` and we run with --function-signature
enabled, we want a single variable declaration for `a` which is reused
later.
Reviewed By: lebedev.ri
Differential Revision: https://reviews.llvm.org/D69722
Johannes Doerfert [Fri, 11 Oct 2019 01:32:04 +0000 (20:32 -0500)]
[Utils] Allow update_test_checks to scrub attribute annotations
Attribute annotations on calls, e.g., #0, are not useful on their own.
This patch adds a flag to update_test_checks.py to scrub them.
Reviewed By: lebedev.ri
Differential Revision: https://reviews.llvm.org/D68851
Johannes Doerfert [Tue, 31 Dec 2019 07:27:50 +0000 (01:27 -0600)]
[Attributor] Propagate known align from arguments to call sites arguments
Since the information is known we can simply use it at the call site.
This is especially useful for callbacks but also helps regular calls.
The test changes are mechanical.
Johannes Doerfert [Thu, 10 Oct 2019 06:19:57 +0000 (01:19 -0500)]
[Attributor] Use abstract call sites to determine associated arguments
This is the second step after D67871 to make use of abstract call sites.
In this patch the argument we associate with a abstract call site
argument can be the one in the callback callee instead of the one in the
callback broker.
Caveat: We cannot allow no-alias arguments for problematic callbacks:
As described in [1], adding no-alias (or restrict) to arguments could
break synchronization as the synchronization effect, e.g., a barrier,
does not "alias" with the pointer anymore. This disables no-alias
annotation for potentially problematic arguments until we implement the
fix described in [1].
Reviewed By: uenoku
Differential Revision: https://reviews.llvm.org/D68008
[1] Compiler Optimizations for OpenMP, J. Doerfert and H. Finkel,
International Workshop on OpenMP 2018,
http://compilers.cs.uni-saarland.de/people/doerfert/par_opt18.pdf
Johannes Doerfert [Tue, 31 Dec 2019 06:57:00 +0000 (00:57 -0600)]
[Attributor] Annotate the memory behavior of call site arguments
Especially for callbacks, annotating the call site arguments is
important. Doing so exposed a too strong dependence of AAMemoryBehavior
on AANoCapture since we handle the case of potentially captured pointers
explicitly.
The changes to the tests are all mechanical.
Shengchen Kan [Tue, 31 Dec 2019 07:10:08 +0000 (15:10 +0800)]
[NFC] Make X86MCCodeEmitter::isPCRel32Branch static
David Blaikie [Tue, 31 Dec 2019 06:32:08 +0000 (22:32 -0800)]
Revert "DebugInfo: Fix rangesBaseAddress DICompileUnit bitcode serialization/deserialization"
Seeing some curious CFI failures internally - which makes little sense
to me, as I don't think anyone is using this flag (even us,
internally)... so sounds like a bug in my code somewhere (possibly a
latent one that propagating this flag exposed, not sure). Reverting
while I investigate.
This reverts commit
c51b45e32ef7f35c11891f60871aa9c2c04cd991.
Shengchen Kan [Tue, 31 Dec 2019 06:23:07 +0000 (14:23 +0800)]
[NFC] Style cleanup
1. Remove function is64BitMode() and use STI.hasFeature(X86::Mode16Bit) directly
2. Use Doxygen features in comment
3. Rename functions to make them start with a lower case letter
4. Format the code with clang-format
River Riddle [Tue, 31 Dec 2019 04:49:47 +0000 (20:49 -0800)]
[mlir] Refactor operation results to use a single use list for all results of the operation.
Summary: A new class is added, IRMultiObjectWithUseList, that allows for representing an IR use list that holds multiple sub values(used in this case for OpResults). This class provides all of the same functionality as the base IRObjectWithUseList, but for specific sub-values. This saves a word per operation result and is a necessary step in optimizing the layout of operation results. For now the use list is placed on the operation itself, so zero-result operations grow by a word. When the work for optimizing layout is finished, this can be moved back to being a trailing object based on memory/runtime benchmarking.
Reviewed By: jpienaar
Differential Revision: https://reviews.llvm.org/D71955
Craig Topper [Tue, 31 Dec 2019 03:07:36 +0000 (19:07 -0800)]
[TargetLowering][AMDGPU] Make scalarizeVectorLoad return a pair of SDValues instead of creating a MERGE_VALUES node. NFCI
This allows us to clean up some places that were peeking through
the MERGE_VALUES node after the call. By returning the SDValues
directly, we can clean that up.
Unfortunately, there are several call sites in AMDGPU that wanted
the MERGE_VALUES and now need to create their own.
Craig Topper [Tue, 31 Dec 2019 02:01:59 +0000 (18:01 -0800)]
[SelectionDAG] Fix copy/paste mistake in comment. NFC
I think this was copied from scalarizeVectorLoad where that is
what happens.
jasonliu [Tue, 31 Dec 2019 03:29:50 +0000 (03:29 +0000)]
[NFC] Add comments in unit test aix-xcoff-toc.ll to clarify the intent
Address David's post review comment in https://reviews.llvm.org/D71667.
Add comments to clarify what we are testing in that file.
Craig Topper [Mon, 30 Dec 2019 22:40:56 +0000 (14:40 -0800)]
[X86] Add test case for PR44412. NFC
Craig Topper [Mon, 30 Dec 2019 21:25:23 +0000 (13:25 -0800)]
[CodeGen] Use IRBuilder::CreateFNeg for __builtin_conj
This replaces the fsub -0.0 idiom with an fneg instruction. We didn't see to have a test that showed the current codegen. Just some tests for constant folding and a test that was only checking the declare lines for libcalls. The latter just checked that we did not have a declare for @conj when using __builtin_conj.
Differential Revision: https://reviews.llvm.org/D72012
Craig Topper [Mon, 30 Dec 2019 21:24:08 +0000 (13:24 -0800)]
[CodeGen] Use CreateFNeg in buildFMulAdd
We have an fneg instruction now and should use it instead of the fsub -0.0 idiom. Looks like we had no test that showed that we handled the negation cases here so I've added new tests.
Differential Revision: https://reviews.llvm.org/D72010
Eric Astor [Mon, 30 Dec 2019 21:11:28 +0000 (16:11 -0500)]
Remove a redundant `default:` on an exhaustive switch(enum).
Johannes Doerfert [Mon, 30 Dec 2019 20:58:28 +0000 (14:58 -0600)]
[OpenMP][FIX] Generalize a test check line
The new check line is compatible with the clang code generation check
line as it allows a 64 and 32 bit value.
I hope this makes the llvm-clang-win-x-armv7l buildbot happy.
Jon Chesterfield [Mon, 30 Dec 2019 20:49:56 +0000 (20:49 +0000)]
[libomptarget][nfc] Change unintentional target_impl prefix to kmpc_impl
Jinsong Ji [Mon, 30 Dec 2019 20:21:46 +0000 (20:21 +0000)]
[PowerPC][docs] Update Embedded PowerPC docs in Compiler Writers Info page
Summary:
Embedded PowerPC are still actively supported, especially SPE...
So update some important references here:
* adding EREF
* adding SPE/VLE ref
Delete deprecated ones into "Other documents..".
Reviewers: #powerpc, jhibbits, hfinkel
Reviewed By: #powerpc, jhibbits
Subscribers: wuzish, merge_guards_bot, nemanjai, shchenz, steven.zhang, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D72008
Johannes Doerfert [Thu, 26 Dec 2019 17:23:38 +0000 (11:23 -0600)]
[OpenMP] Use the OpenMPIRBuilder for `omp parallel`
This allows to use the OpenMPIRBuilder for parallel regions. Code was
extracted from D61953 and adapted to work with the new version (D70109).
All but one feature should be supported. An update of this patch will
provide test coverage and privatization other than shared.
Reviewed By: fghanim
Differential Revision: https://reviews.llvm.org/D70290
Johannes Doerfert [Fri, 27 Dec 2019 21:53:37 +0000 (15:53 -0600)]
[OpenMP] Use the OpenMPIRBuilder for `omp cancel`
An `omp cancel parallel` needs to be emitted by the OpenMPIRBuilder if
the `parallel` was emitted by the OpenMPIRBuilder. This patch makes
this possible. The cancel logic is shared with the cancel barriers.
Testing is done via unit tests and the clang cancel_codegen.cpp file
once D70290 lands.
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D71948
Eric Astor [Mon, 30 Dec 2019 19:33:56 +0000 (14:33 -0500)]
[X86][AsmParser] re-introduce 'offset' operator
Summary:
Amend MS offset operator implementation, to more closely fit with its MS counterpart:
1. InlineAsm: evaluate non-local source entities to their (address) location
2. Provide a mean with which one may acquire the address of an assembly label via MS syntax, rather than yielding a memory reference (i.e. "offset asm_label" and "$asm_label" should be synonymous
3. address PR32530
Based on http://llvm.org/D37461
Fix broken test where the break appears unrelated.
- Set up appropriate memory-input rewrites for variable references.
- Intel-dialect assembly printing now correctly handles addresses by adding "offset".
- Pass offsets as immediate operands (using "r" constraint for offsets of locals).
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D71436
Matt Arsenault [Sun, 8 Sep 2019 22:11:51 +0000 (18:11 -0400)]
AMDGPU/GlobalISel: Select mul24 intrinsics
Matt Arsenault [Mon, 30 Dec 2019 17:05:25 +0000 (12:05 -0500)]
TableGen: Fix assert on PatFrags with predicate code
This assumed a single pattern if there was a predicate. Relax this a
bit, and allow multiple patterns as long as they have the same class.
This was only broken for the DAG path. GlobalISel seems to have
handled this correctly already.
Craig Topper [Mon, 30 Dec 2019 18:50:04 +0000 (10:50 -0800)]
[X86] Add X86ISD::PCMPGT to SimplifyMultipleUseDemandedBitsForTargetNode.
If only the sign bit is demanded, and the LHS is all zeroes, then
we can bypass the PCMPGT.
Bryan Chan [Fri, 27 Dec 2019 22:26:24 +0000 (17:26 -0500)]
[test] do not parse ls output for file size; NFCI
Parsing `ls -l` output to obtain the size of a file is unreliable; the
exact output format is not specified, and some user or group names may
contain multiple words, causing `cut -f5 -d' '` to extract an incorrect
value. `wc -c`, on the other hand, is portable, and there are precendents
of its use in test cases.
Matt Arsenault [Sun, 29 Dec 2019 14:05:56 +0000 (09:05 -0500)]
AMDGPU/GlobalISel: Re-use MRI available in selector
Fangrui Song [Wed, 25 Dec 2019 02:12:15 +0000 (18:12 -0800)]
Ignore "no-frame-pointer-elim" and "no-frame-pointer-elim-non-leaf" in favor of "frame-pointer"
D56351 (included in LLVM 8.0.0) introduced "frame-pointer". All tests
which use "no-frame-pointer-elim" or "no-frame-pointer-elim-non-leaf"
have been migrated to use "frame-pointer".
Implement UpgradeFramePointerAttributes to upgrade the two obsoleted
function attributes for bitcode. Their semantics are ignored.
Differential Revision: https://reviews.llvm.org/D71863
Sanjay Patel [Mon, 30 Dec 2019 17:38:49 +0000 (12:38 -0500)]
[InstCombine] remove stale comment on test; NFC
Petar Avramovic [Mon, 30 Dec 2019 17:06:29 +0000 (18:06 +0100)]
[MIPS GlobalISel] Select bitreverse. Recommit
G_BITREVERSE is generated from llvm.bitreverse.<type> intrinsics,
clang genrates these intrinsics from __builtin_bitreverse32 and
__builtin_bitreverse64.
Add lower and narrowscalar for G_BITREVERSE.
Lower G_BITREVERSE on MIPS32.
Recommit notes:
Introduce temporary variables in order to make sure
instructions get inserted into MachineFunction in same order
regardless of compiler used to build llvm.
Differential Revision: https://reviews.llvm.org/D71363
Matt Arsenault [Sun, 8 Sep 2019 21:44:09 +0000 (17:44 -0400)]
AMDGPU/GlobalISel: Select llvm.amdgcn.fmad.ftz
Sanjay Patel [Mon, 30 Dec 2019 16:04:00 +0000 (11:04 -0500)]
[InstCombine] propagate sign argument through nested copysigns
This is another optimization suggested in PR44153:
https://bugs.llvm.org/show_bug.cgi?id=44153
Diogo Sampaio [Mon, 30 Dec 2019 15:43:32 +0000 (15:43 +0000)]
[ARM][Thumb][FIX] Add unwinding information to t4
Summary:
Add missing part of patch D71361. Now that the stack-frame
can be operated using a addw/subw instruction, they should
appear in the unwinding list.
Reviewers: dmgreen, efriedma
Reviewed By: dmgreen
Subscribers: kristof.beyls, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D72000
Matt Arsenault [Sun, 8 Sep 2019 19:39:00 +0000 (15:39 -0400)]
AMDGPU/GlobalISel: Add select test for fexp2
Matt Arsenault [Sat, 27 Jul 2019 21:47:08 +0000 (17:47 -0400)]
GlobalISel: moreElementsVector for FP min/max
Matt Arsenault [Tue, 24 Dec 2019 16:07:45 +0000 (11:07 -0500)]
AMDGPU: Improve llvm.round.f64 lowering for CI+
The path already used for f16/f32 works a lot better when v_trunc_f64
is available.
Matt Arsenault [Tue, 24 Dec 2019 16:24:30 +0000 (11:24 -0500)]
AMDGPU: Generate check lines
Matt Arsenault [Sat, 21 Dec 2019 20:34:30 +0000 (15:34 -0500)]
AMDGPU/GlobalISel: Account for G_PHI result bank
Sometimes the result bank of the phi is already assigned to something,
and should not be ignored. This is in preparation for additional
boolean phi handling changes.
Also refine the logic to fix some cases that were incorrectly deciding
to use SGPRs.
Nemanja Ivanovic [Mon, 30 Dec 2019 13:38:27 +0000 (07:38 -0600)]
[PowerPC] Legalize rounding nodes
VSX provides a full complement of rounding instructions yet we somehow ended up
with some of them legal and others not. This just legalizes all of the FP
rounding nodes and the FP -> int rounding nodes with unsafe math.
Differential revision: https://reviews.llvm.org/D69949
Dmitri Gribenko [Mon, 30 Dec 2019 13:28:56 +0000 (14:28 +0100)]
Revert "[MIPS GlobalISel] Select bitreverse"
This reverts commit
dbc136e0fe7e14c64dcb78e72321bb41af60afa4.
It broke buildbots:
http://lab.llvm.org:8011/builders/clang-x86_64-debian-fast/builds/21066
David Green [Mon, 30 Dec 2019 09:39:14 +0000 (09:39 +0000)]
[ARM] Sink splat to ICmp
This adds ICmp to the list of instructions that we sink a splat to in a
loop, allowing the register forms of instructions to be selected more
often. It does not add FCmp yet as the results look a little odd, trying
to keep the register in an float reg and having to move it back to a GPR.
Differential Revision: https://reviews.llvm.org/D70997
David Green [Mon, 30 Dec 2019 09:28:10 +0000 (09:28 +0000)]
[ARM] MVE sink ICmp test. NFC
Evgeniy Brevnov [Mon, 9 Dec 2019 08:11:30 +0000 (15:11 +0700)]
[LV][NFC] Keep dominator tree up to date during vectorization.
Evgeniy Brevnov [Thu, 5 Dec 2019 10:57:27 +0000 (17:57 +0700)]
[LV][NFC] Some refactoring and renaming to facilitate next change.
Diogo Sampaio [Mon, 30 Dec 2019 10:59:45 +0000 (10:59 +0000)]
[ARM][THUMB2] Allow emitting T3 types of add and sub
Summary:
This patch allows to emit thumb2 add and sub
instructions with 12 bit immediates in the
emitT2RegPlusImmediate function.
- Splitting parts of the D70680
Reviewers: eli.friedman, olista01, efriedma
Reviewed By: efriedma
Subscribers: efriedma, kristof.beyls, hiraditya, dmgreen, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D71361
Sven van Haastregt [Mon, 30 Dec 2019 10:47:58 +0000 (10:47 +0000)]
[OpenCL] Add mipmap builtin functions
Add the mipmap builtin functions from the OpenCL extension
specification.
Patch by Pierre Gondois and Sven van Haastregt.
Petar Avramovic [Mon, 30 Dec 2019 10:26:45 +0000 (11:26 +0100)]
[MIPS GlobalISel] Select bitreverse
G_BITREVERSE is generated from llvm.bitreverse.<type> intrinsics,
clang genrates these intrinsics from __builtin_bitreverse32 and
__builtin_bitreverse64.
Add lower and narrowscalar for G_BITREVERSE.
Lower G_BITREVERSE on MIPS32.
Differential Revision: https://reviews.llvm.org/D71363
Petar Avramovic [Mon, 30 Dec 2019 10:13:22 +0000 (11:13 +0100)]
[MIPS GlobalISel] Select bswap
G_BSWAP is generated from llvm.bswap.<type> intrinsics, clang genrates
these intrinsics from __builtin_bswap32 and __builtin_bswap64.
Add lower and narrowscalar for G_BSWAP.
Lower G_BSWAP on MIPS32, select G_BSWAP on MIPS32 revision 2 and later.
Differential Revision: https://reviews.llvm.org/D71362
Kai Luo [Mon, 30 Dec 2019 08:31:41 +0000 (16:31 +0800)]
[MCP] Add stats for backward copy propagation. NFC.
Peter Kokai [Mon, 30 Dec 2019 08:22:55 +0000 (00:22 -0800)]
[opt] Fix run-twice crash and detection problem
1. Execute `opt -run-twice a.ll` with in a terminal will crash.
https://bugs.llvm.org/show_bug.cgi?id=44382
2. `-run-twice` saves output into two buffers and compares them.
When outputing the result is disabled, that produces two empty string thus
they are going to be equal all the time resulting false-positive results.
The proposed solution is to generate the results even if the output will not be
emitted, as that is required for the comparision.
Differential Revision: https://reviews.llvm.org/D71967
Tyker [Sun, 29 Dec 2019 23:14:20 +0000 (00:14 +0100)]
[Diagnostic] Add ftabstop to -Wmisleading-indentation
Summary:
this allow much better support of codebases like the linux kernel that mix tabs and spaces.
-ftabstop=//Width// allow specifying how large tabs are considered to be.
Reviewers: xbolva00, aaron.ballman, rsmith
Reviewed By: aaron.ballman
Subscribers: jyknight, riccibruno, rsmith, nathanchance
Differential Revision: https://reviews.llvm.org/D71037
Qiu Chaofan [Mon, 30 Dec 2019 08:09:07 +0000 (16:09 +0800)]
[NFC] Add test for load-insert-store pattern
This patch adds necessary test cases for load-update-store pattern
which only updates single element of vector.
Differential Revision: https://reviews.llvm.org/D71886
Hideto Ueno [Mon, 30 Dec 2019 08:08:48 +0000 (17:08 +0900)]
[Attributor] Use `changeUseAfterManifest` in AAValueSimplify manifest
Summary: This patch makes `AAValueSimplify` use `changeUsesAfterManifest` in `manifest`. This will invoke simple folding after the manifest.
Reviewers: jdoerfert, sstefan1
Reviewed By: jdoerfert
Subscribers: hiraditya, arphaman, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D71972
Fangrui Song [Wed, 18 Dec 2019 00:45:04 +0000 (16:45 -0800)]
[ELF][PPC64] Improve "call lacks nop" diagnostic and make it compatible with GCC<5.5 and GCC<6.4
GCC before r245813 (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79439)
did not emit nop after b/bl. This can happen with recursive calls.
r245813 was back ported to GCC 5.5 and GCC 6.4.
This is common, for example, libstdc++.a(locale.o) shipped with GCC 4.9
and many objects in netlib lapack can cause lld to error. gold allows
such calls to the same section. Our __plt_foo symbol's `section` field
is used for ThunkSection, so we can't implement a similar loosen rule
easily. But we can make use of its `file` field which is currently NULL.
Differential Revision: https://reviews.llvm.org/D71639
Fangrui Song [Tue, 17 Dec 2019 19:23:37 +0000 (11:23 -0800)]
[ELF][PPC32] Implement IPLT code sequence for non-preemptible IFUNC
Similar to D71509 (EM_PPC64), on EM_PPC, the IPLT code sequence should
be similar to a PLT call stub. Unlike EM_PPC64, EM_PPC -msecure-plt has
small/large PIC model differences.
* -fpic/-fpie: R_PPC_PLTREL24 r_addend=0. The call stub loads an address relative to `_GLOBAL_OFFSET_TABLE_`.
* -fPIC/-fPIE: R_PPC_PLTREL24 r_addend=0x8000. (A partial linked object
file may have an addend larger than 0x8000.) The call stub loads an address relative to .got2+0x8000.
Just assume large PIC model for now. This patch makes:
// clang -fuse-ld=lld -msecure-plt -fno-pie -no-pie a.c
// clang -fuse-ld=lld -msecure-plt -fPIE -pie a.c
#include <stdio.h>
static void impl(void) { puts("meow"); }
void thefunc(void) __attribute__((ifunc("resolver")));
void *resolver(void) { return &impl; }
int main(void) {
thefunc();
void (*theptr)(void) = &thefunc;
theptr();
}
work on Linux glibc. -fpie will crash because the compiler and the
linker do not agree on the value which r30 stores (_GLOBAL_OFFSET_TABLE_
vs .got2+0x8000).
Differential Revision: https://reviews.llvm.org/D71621
Fangrui Song [Sat, 14 Dec 2019 02:30:21 +0000 (18:30 -0800)]
[ELF][PPC64] Implement IPLT code sequence for non-preemptible IFUNC
Non-preemptible IFUNC are placed in in.iplt (.glink on EM_PPC64). If
there is a non-GOT non-PLT relocation, for pointer equality, we change
the type of the symbol from STT_IFUNC and STT_FUNC and bind it to the
.glink entry.
On EM_386, EM_X86_64, EM_ARM, and EM_AARCH64, the PLT code sequence
loads the address from its associated .got.plt slot. An IPLT also has an
associated .got.plt slot and can use the same code sequence.
On EM_PPC64, the PLT code sequence is actually a bl instruction in
.glink . It jumps to `__glink_PLTresolve` (the PLT header). and
`__glink_PLTresolve` computes the .plt slot (relocated by
R_PPC64_JUMP_SLOT).
An IPLT does not have an associated R_PPC64_JUMP_SLOT, so we cannot use
`bl` in .iplt . Instead, create a call stub which has a similar code
sequence as PPC64PltCallStub. We don't save the TOC pointer, so such
scenarios will not work: a function pointer to a non-preemptible ifunc,
which resolves to a function defined in another DSO. This is the
restriction described by https://sourceware.org/glibc/wiki/GNU_IFUNC
(though on many architectures it works in practice):
Requirement (a): Resolver must be defined in the same translation unit as the implementations.
If an ifunc is taken address but not called, technically we don't need
an entry for it, but we currently do that.
This patch makes
// clang -fuse-ld=lld -fno-pie -no-pie a.c
// clang -fuse-ld=lld -fPIE -pie a.c
#include <stdio.h>
static void impl(void) { puts("meow"); }
void thefunc(void) __attribute__((ifunc("resolver")));
void *resolver(void) { return &impl; }
int main(void) {
thefunc();
void (*theptr)(void) = &thefunc;
theptr();
}
work on Linux glibc and FreeBSD. Calling a function pointer pointing to
a Non-preemptible IFUNC never worked before.
Differential Revision: https://reviews.llvm.org/D71509
Fangrui Song [Mon, 30 Dec 2019 04:17:31 +0000 (20:17 -0800)]
[SelectionDAT] Simplify SelectionDAGBuilder::visitInlineAsm
Indirect C_Immediate or C_Other constraints have been excluded.
Also simplify an unneeded change to indirect 'X' by D60942.