wanglian [Mon, 19 Sep 2022 02:28:30 +0000 (10:28 +0800)]
[CUDA][NFC] Rename 'addDeviceDepences' to 'addDeviceDependences' in DeviceActionBuilder.
Reviewed By: tra
Differential Revision: https://reviews.llvm.org/D134007
Chuanqi Xu [Mon, 19 Sep 2022 02:35:00 +0000 (10:35 +0800)]
[NFC] Move the position of CodeGen/module-initializer*.cpp
Previsouly the module-initializer*.cpp lives in the CodeGen dir instead
of CodeGenCXX dir, which is not consistency with other tests since
modules are features for C++.
LiaoChunyu [Mon, 19 Sep 2022 02:22:31 +0000 (10:22 +0800)]
[RISCV][NFC]Remove outdated comment from targetShrinkDemandedConstant
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D134154
Kazu Hirata [Mon, 19 Sep 2022 01:41:02 +0000 (18:41 -0700)]
Use std::make_unsigned_t (NFC)
Emilia Dreamer [Sun, 18 Sep 2022 23:49:58 +0000 (02:49 +0300)]
[clang-format] Disallow requires clauses to become function declarations
There already exists logic to disallow requires *expressions* to be
treated as function declarations, but this expands it to include
requires *clauses*, when they happen to also be parenthesized.
Previously, in the following case:
```
template <typename T>
requires(Foo<T>)
T foo();
```
The line with the requires clause was actually being considered as the
line with the function declaration due to the parentheses, and the
*real* function declaration on the next line became a trailing
annotation
(Together with https://reviews.llvm.org/D134049) Fixes https://github.com/llvm/llvm-project/issues/56213
Reviewed By: HazardyKnusperkeks, owenpan
Differential Revision: https://reviews.llvm.org/D134052
Emilia Dreamer [Sun, 18 Sep 2022 23:47:42 +0000 (02:47 +0300)]
[clang-format] Disallow trailing return arrows to be operators
In the following construction:
`template <typename T> requires Foo<T> || Bar<T> auto func() -> int;`
The `->` of the trailing return type was actually considered as an
operator as part of the binary operation in the requires clause, with
the precedence level of `PrecedenceArrowAndPeriod`, leading to fake
parens being inserted in strange locations, that would never be closed.
Fixes one part of https://github.com/llvm/llvm-project/issues/56213
(the rest will probably be in a separate patch)
Reviewed By: HazardyKnusperkeks, owenpan
Differential Revision: https://reviews.llvm.org/D134049
Lang Hames [Mon, 19 Sep 2022 00:47:12 +0000 (17:47 -0700)]
[ORC][ORC-RT] Make WrapperFunctionCall::Create support void functions.
Serialized calls to void-wrapper-functions should have zero bytes of argument
data, but accessing ArgData[0] may (and will, in the case of SmallVector) fail
if the argument data buffer is empty.
This commit fixes the issue by adding a check for empty argument buffers.
Kazu Hirata [Mon, 19 Sep 2022 00:50:17 +0000 (17:50 -0700)]
[X86] Fix the LEA optimization pass
The LEA optimization pass visits each basic block of a given machine
function. In each basic block, for each pair of LEAs that differ only
in their displacement fields, we replace all uses of the second LEA
with the first LEA while adjusting the displacement.
Now, without this patch, after all the replacements are made, the
following assert triggers:
assert(MRI->use_empty(LastVReg) &&
"The LEA's def register must have no uses");
The replacement loop uses:
for (MachineOperand &MO :
llvm::make_early_inc_range(MRI->use_operands(LastVReg))) {
which is equivalent to:
for (auto UI = MRI->use_begin(LastVReg), UE = MRI->use_end();
UI != UE;) {
MachineOperand &MO = *UI++; // <-- Look!
That is, immediately after the post increment, make_early_inc_range
already has the iterator for the next iteration in its mind.
The problem is that in one iteration of the loop, we could replace two
uses in a debug instruction like:
DBG_VALUE_LIST !"r", !DIExpression(DW_OP_LLVM_arg, 0), %0:gr64, %0:gr64, ...
So, the iterator for the next iteration becomes invalid. We end up
traversing a garbage use list from that point on. In turn, we don't
get to visit remaining uses.
The patch fixes the problem by switching to a "draining" while loop:
while (!MRI->use_empty(LastVReg)) {
MachineOperand &MO = *MRI->use_begin(LastVReg);
MachineInstr &MI = *MO.getParent();
The credit goes to Simon Pilgrim for reducing the test case.
Fixes https://github.com/llvm/llvm-project/issues/57673
Differential Revision: https://reviews.llvm.org/D133631
Kazu Hirata [Mon, 19 Sep 2022 00:46:53 +0000 (17:46 -0700)]
[mlir] Use empty (NFC)
Yaxun (Sam) Liu [Sat, 17 Sep 2022 21:57:35 +0000 (17:57 -0400)]
[SimplifyCFG] accumulate bonus insts cost
SimplifyCFG folds
bool foo() {
if (cond1) return false;
if (cond2) return false;
return true;
}
as
bool foo() {
if (cond1 | cond2) return false
return true;
}
'cond2' is called 'bonus insts' in branch folding since they introduce overhead
since the original CFG could do early exit but the folded CFG always executes
them. SimplifyCFG calculates the costs of 'bonus insts' of a folding a BB into
its predecessor BB which shares the destination. If it is below bonus-inst-threshold,
SimplifyCFG will fold that BB into its predecessor and cond2 will always be executed.
When SimplifyCFG calculates the cost of 'bonus insts', it only consider 'bonus' insts
in the current BB to be considered for folding. This causes issue for unrolled loops
which share destinations, e.g.
bool foo(int *a) {
for (int i = 0; i < 32; i++)
if (a[i] > 0) return false;
return true;
}
After unrolling, it becomes
bool foo(int *a) {
if(a[0]>0) return false
if(a[1]>0) return false;
//...
if(a[31]>0) return false;
return true;
}
SimplifyCFG will merge each BB with its predecessor BB,
and ends up with 32 'bonus insts' which are always executed, which
is much slower than the original CFG.
The root cause is that SimplifyCFG does not consider the
accumulated cost of 'bonus insts' which are folded from
different BB's.
This patch fixes that by introducing a ValueMap to track
costs of 'bonus insts' coming from different BB's into
the same BB, and cuts off if the accumulated cost
exceeds a threshold.
Reviewed by: Artem Belevich, Florian Hahn, Nikita Popov, Matt Arsenault
Differential Revision: https://reviews.llvm.org/D132408
Carl Ritson [Sun, 18 Sep 2022 23:23:43 +0000 (08:23 +0900)]
[AMDGPU] Fix isSGPRReg for special registers
Special registers, e.g. MODE, do not have register classes so
will cause null pointer exception if passed to isSGPRReg.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D134025
Kazu Hirata [Sun, 18 Sep 2022 23:23:19 +0000 (16:23 -0700)]
[BOLT] Remove duplicate types (NFC)
This patch, a follow-up for
588628de3e954d8029b20841aceb612e328ae104,
removes duplicate types like T and PointerT in favor of reference and
pointer, respectively.
Xiang Li [Mon, 12 Sep 2022 22:58:32 +0000 (15:58 -0700)]
[HLSL] [clang] Add vector version of abs for HLSL
Add vector version of abs as
```
__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
int2 abs (int2 );
__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
int3 abs (int3 );
```
To make this work.
Allowed custom type checking builtins to be recelareable.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D133737
Kazu Hirata [Sun, 18 Sep 2022 21:35:09 +0000 (14:35 -0700)]
[ModuleInliner] Capitalize a variable name (NFC)
Kazu Hirata [Sun, 18 Sep 2022 21:27:06 +0000 (14:27 -0700)]
[ModuleInliner] Remove unused using declarations (NFC)
Kazu Hirata [Sun, 18 Sep 2022 21:09:21 +0000 (14:09 -0700)]
[ModuleInliner] Move getInlineCostWrapper to an anonymous namespace (NFC)
This patch moves getInlineCostWrapper to an anonymous namespace.
While I am at it, I'm moving the function closer to the beginning of
the file so that I can use it elsewhere in the file without a forward
declaration.
Shafik Yaghmour [Sun, 18 Sep 2022 18:54:32 +0000 (11:54 -0700)]
[Clang] Fix compat diagnostic to detect a nontype template parameter has a placeholder type using getContainedAutoType()
Based on the changes introduced by
15361a21e01026e74cb17011b702c7d1c881ae94 it
looks like C++17 compatibility diagnostic should have been checking
getContainedAutoType().
This fixes: https://github.com/llvm/llvm-project/issues/57369
https://github.com/llvm/llvm-project/issues/57643
https://github.com/llvm/llvm-project/issues/57793
Differential Revision: https://reviews.llvm.org/D132990
Sanjay Patel [Sun, 18 Sep 2022 18:21:21 +0000 (14:21 -0400)]
[InstCombine] remove multi-use add demanded constant fold
This was originally part of D133788. There are no visible
regressions. All of the diffs show a large unsigned constant
becoming a small negative constant. This should be better
for analysis (and slightly less compile-time) and codegen.
Kazu Hirata [Sun, 18 Sep 2022 18:21:16 +0000 (11:21 -0700)]
[llvm] Use x.empty() instead of llvm::empty(x) (NFC)
I'm planning to deprecate and eventually remove llvm::empty.
Note that no use of llvm::empty requires the ability of llvm::empty to
determine the emptiness from begin/end only.
owenca [Sat, 17 Sep 2022 04:31:21 +0000 (21:31 -0700)]
[clang-format] Skip token annotation in passes that don't need it
Differential Revision: https://reviews.llvm.org/D134103
Kazu Hirata [Sun, 18 Sep 2022 18:10:11 +0000 (11:10 -0700)]
[flang] Use x.empty() instead of llvm::empty(x) (NFC)
I'm planning to deprecate and eventually remove llvm::empty.
Note that no use of llvm::empty requires the ability of llvm::empty to
determine the emptiness from begin/end only.
Kazu Hirata [Sun, 18 Sep 2022 18:07:50 +0000 (11:07 -0700)]
[clang] Use x.empty() instead of llvm::empty(x) (NFC)
I'm planning to deprecate and eventually remove llvm::empty.
Note that no use of llvm::empty requires the ability of llvm::empty to
determine the emptiness from begin/end only.
Kazu Hirata [Sun, 18 Sep 2022 18:01:56 +0000 (11:01 -0700)]
[BOLT] Use x.empty() instead of llvm::empty(x) (NFC)
I'm planning to deprecate and eventually remove llvm::empty.
Note that no use of llvm::empty requires the ability of llvm::empty to
determine the emptiness from begin/end only.
Kazu Hirata [Sun, 18 Sep 2022 17:53:49 +0000 (10:53 -0700)]
[mlir] Use x.empty() instead of llvm::empty(x) (NFC)
I'm planning to deprecate and eventually remove llvm::empty.
Note that no use of llvm::empty requires the ability of llvm::empty to
determine the emptiness from begin/end only.
Kazu Hirata [Sun, 18 Sep 2022 17:25:08 +0000 (10:25 -0700)]
Use std::decay_t (NFC)
Kazu Hirata [Sun, 18 Sep 2022 17:25:06 +0000 (10:25 -0700)]
Use std::conditional_t (NFC)
Kazu Hirata [Sun, 18 Sep 2022 17:25:04 +0000 (10:25 -0700)]
[ADT] Use std::common_type_t (NFC)
Marc Auberer [Sun, 18 Sep 2022 17:16:12 +0000 (13:16 -0400)]
[InstCombine] Fix bug when folding x + (x | -x) to x & (x - 1)
Addresses concern: https://reviews.llvm.org/rG09cdddea0c4d284c2c22f5dfade40a60850c5ea7
There was a copy/paste mistake in the code. Updated code and test ref.
Differential Revision: https://reviews.llvm.org/D134135
Sanjay Patel [Sun, 18 Sep 2022 16:40:53 +0000 (12:40 -0400)]
[InstCombine] fold full-shift of sdiv to icmp+extend
This is a disguised sign-bit test with offset:
(X / +DivC) >> (Width - 1) --> ext (X <= -DivC)
(X / -DivC) >> (Width - 1) --> ext (X >= +DivC)
https://alive2.llvm.org/ce/z/cO8JO4
We don't match/test poison in the sdiv constant because
that would be immediate undefined behavior.
Sanjay Patel [Fri, 16 Sep 2022 20:34:36 +0000 (16:34 -0400)]
[InstCombine] add tests for full-right-shift of sdiv; NFC
Aaron Puchert [Sun, 18 Sep 2022 17:11:54 +0000 (19:11 +0200)]
Make sure libLLVM users link with libatomic if needed
64-bit atomics are used in llvm/ADT/Statistic.h, which means that users
of libLLVM.so might also have to link with libatomic. To avoid having
to special-case the library here, we simply add all `LLVM_SYSTEM_LIBS`
as public link dependencies to libLLVM.
This fixes a build failure on PowerPC 32-bit.
Reviewed By: beanz
Differential Revision: https://reviews.llvm.org/D132799
Kazu Hirata [Sun, 18 Sep 2022 15:49:44 +0000 (08:49 -0700)]
[ModuleInliner] Remove InlineOrder::front (NFC)
InlineOrder::front is a remnant from the era when we had a nested
"while" loops in the module inliner, with the inner one grouping the
call sites with the same caller.
Now that we have a simple "while" loop draining the priority queue, we
can just use InlineOrder::pop.
Differential Revision: https://reviews.llvm.org/D134121
Phoebe Wang [Sun, 18 Sep 2022 13:55:26 +0000 (21:55 +0800)]
[Clang][NFC] update predicate and reduce redundant check
Simon Pilgrim [Sun, 18 Sep 2022 13:35:04 +0000 (14:35 +0100)]
[DAG] MatchRotate - reuse existing LHSShiftArg/RHSShiftArg variables. NFC.
Jun Zhang [Sat, 17 Sep 2022 09:41:46 +0000 (17:41 +0800)]
[Docs] Add a link that refers to C++ standard modules in Clang modules doc
Currently there're two pages that both talk about "Modules" in clang, but
they're different. The one that describes C++ standard modules explicitly
spells out the difference but the other one which targeting Clang modules
doesn't.
This patch adds a link that refers to the C++ standard modules
one in Clang modules doc, as you usually got the later page when
googling. I believe this will make newcomers less confused.
Signed-off-by: Jun Zhang <jun@junz.org>
Differential Revision: https://reviews.llvm.org/D134105
Phoebe Wang [Sun, 18 Sep 2022 10:00:36 +0000 (18:00 +0800)]
[Clang][NFC] update obsolete check predicate
Nikolas Klauser [Sat, 17 Sep 2022 12:30:29 +0000 (14:30 +0200)]
[libc++] Avoid including <tuple> in compressed_pair.h
compressed_pair is widely used in the library, but most of the uses don't use the tuple parts. To avoid including <tuple> everywhere, use the forward declaration instead in compressed_pair.h
Reviewed By: ldionne, #libc
Spies: libcxx-commits
Differential Revision: https://reviews.llvm.org/D133331
Chuanqi Xu [Sun, 18 Sep 2022 08:43:27 +0000 (16:43 +0800)]
[Modules] Don't judge if we're compiling a module unit by LangOpts::CurrentModule.empty()
Closing https://github.com/llvm/llvm-project/issues/57778.
Previously it judge if we're compiling a module unit by
LangOpts::CurrentModule.empty(). But it is not true since we can specify
the module name by `-fmodule-name` option for arbitrary module unit.
Then this patch adjuest the judgement properly.
Benjamin Kramer [Sun, 18 Sep 2022 07:15:32 +0000 (09:15 +0200)]
Silence unused variable warning in release builds. NFC
Shivam Gupta [Sun, 18 Sep 2022 06:59:40 +0000 (12:29 +0530)]
[NFC] Typo fix in ARCMTActions.h
Kazu Hirata [Sun, 18 Sep 2022 06:10:23 +0000 (23:10 -0700)]
[Transforms] Merge function attributes within InlineFunction (NFC)
In the past, we've had a bug resulting in a compiler crash after
forgetting to merge function attributes (D105729).
This patch teaches InlineFunction to merge function attributes. This
way, we minimize the "time" when the IR is valid, but the function
attributes are not.
Differential Revision: https://reviews.llvm.org/D134117
Kazu Hirata [Sun, 18 Sep 2022 03:59:54 +0000 (20:59 -0700)]
[Analysis] Introduce isSoleCallToLocalFunction (NFC)
We check to see if a given CallBase is a sole call to a local function
at multiple places in InlineCost.cpp. This patch factors out the
common code.
Differential Revision: https://reviews.llvm.org/D134114
Yaxun (Sam) Liu [Tue, 30 Aug 2022 03:03:29 +0000 (23:03 -0400)]
[SimplifyCFG] add a test for branch folding multiple BB
Reviewed by: Florian Hahn
Differential Revision: https://reviews.llvm.org/D132910
Kai Nacke [Sat, 3 Sep 2022 16:04:55 +0000 (16:04 +0000)]
[GISel] Fix match tree emitter.
The following changes are necessasy to get the generated tree
matcher to compile:
- In CodeExpansions::declare(), the assert() prevents connecting
two instructions. E.g. the match code
(match (MUL $t, $s1, $s2),
(SUB $d, $t, $s3)),
results in two declarations of $t, one for the def and one for
the use. Removing the assertion allows this construct.
If $t is later used, it is one of the operands, which should be
perfectly fine.
- The code emitted in GIMatchTreeVRegDefPartitioner::generatePartitionSelectorCode()
is not compilable:
- The value of NewInstrID should be emitted, not the name
- Both calls involving getOperand() end with one parenthesis too many
- Swaps generated condition for the partition code in the latter function
It also changes the rules i2p_to_p2i, fabs_fabs_fold, and fneg_fneg_fold
to use the tree matcher for a linear match. These rules are tested by:
CodeGen/AArch64/GlobalISel/combine-fabs.mir
CodeGen/AArch64/GlobalISel/combine-fneg.mir
CodeGen/AArch64/GlobalISel/combine-ptrtoint.mir
CodeGen/AMDGPU/GlobalISel/combine-add-nullptr.mir
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D133257
Kazu Hirata [Sat, 17 Sep 2022 21:16:32 +0000 (14:16 -0700)]
[ModuleInliner] Set Changed earlier (NFC)
It makes more sense to set Changed to true immediately after a
successful inlining.
isuckatcs [Sat, 10 Sep 2022 17:04:28 +0000 (19:04 +0200)]
[analyzer] Cleanup some artifacts from non-POD array evaluation
Most of the state traits used for non-POD array evaluation were
only cleaned up if the ctors/dtors were inlined, since the cleanup
happened in ExprEngine::processCallExit(). This patch makes sure
they are removed even if said functions are not inlined.
Differential Revision: https://reviews.llvm.org/D133643
Kazu Hirata [Sat, 17 Sep 2022 20:36:16 +0000 (13:36 -0700)]
[mlir] Don't include SetVector.h (NFC)
Kazu Hirata [Sat, 17 Sep 2022 20:36:15 +0000 (13:36 -0700)]
[lld] Don't include SetVector.h (NFC)
Kazu Hirata [Sat, 17 Sep 2022 20:36:13 +0000 (13:36 -0700)]
[clang] Don't include SetVector.h (NFC)
LLVM GN Syncbot [Sat, 17 Sep 2022 19:55:35 +0000 (19:55 +0000)]
[gn build] Port
e5e3dccd0741
Aiden Grossman [Sat, 17 Sep 2022 19:27:18 +0000 (19:27 +0000)]
[mlgo] Add in-development instruction based features for regalloc advisor
This patch adds in instruction based features to the regalloc advisor
gated behind a flag so a user can decide at runtime whether or not they
want to enable the feature. The features are only enabled when LLVM is
compiled in MLGO develpment mode (LLVM_HAVE_TF_API) is set to true.
To extract the instruction features, I'm taking a list of segments from
each LiveInterval and noting the start and end SlotIndices. This list is then
sorted based on the start SlotIndex and I iterate through each SlotIndex
to grab instructions, making sure to check for overlaps. This results in
a vector of opcodes and binary mapping matrix that maps live ranges to the
opcodes of the instructions within that LR.
Reviewed By: mtrofin
Differential Revision: https://reviews.llvm.org/D131930
Kazu Hirata [Sat, 17 Sep 2022 19:36:43 +0000 (12:36 -0700)]
[llvm] Don't including SetVector.h (NFC)
llvm/lib/ProfileData/RawMemProfReader.cpp uses SetVector without
including SetVector.h, so this patch adds an appropriate #include
there.
Fangrui Song [Sat, 17 Sep 2022 19:35:17 +0000 (12:35 -0700)]
[Support] Rename llvm::compression::{zlib,zstd}::uncompress to more appropriate decompress
This improves consistency with other places (e.g. llvm::compression::decompress,
llvm::object::Decompressor::decompress, llvm-objcopy).
Note: when zstd::uncompress was added, we noticed that the API `ZSTD_decompress`
is fine while the zlib API `uncompress` is a misnomer.
Kazu Hirata [Sat, 17 Sep 2022 19:17:51 +0000 (12:17 -0700)]
[ModuleInliner] Don't include SetVector.h (NFC)
We don't use SetVector in the module inliner.
Kazu Hirata [Sat, 17 Sep 2022 19:05:35 +0000 (12:05 -0700)]
[ModuleInliner] Remove unnecessary #includes (NFC)
While I am at it, this patch removes an unnecessary forward
declaration.
Kazu Hirata [Sat, 17 Sep 2022 18:41:28 +0000 (11:41 -0700)]
[ModuleInliner] Move UseInlinePriority to InlineOrder.cpp (NFC)
UseInlinePriority specifies the priority function. This patch
simplifies the code by moving UseInlinePriority closer to the actual
consumer -- the switch statement inside getInlineOrder.
Differential Revision: https://reviews.llvm.org/D134100
Sander de Smalen [Sat, 17 Sep 2022 16:23:32 +0000 (16:23 +0000)]
[AArch64][SME] Add intrinsics for enabling/disabling ZA.
This adds the intrinsics:
* void @llvm.aarch64.sme.za.enable() -> smstart za
* void @llvm.aarch64.sme.za.disable() -> smstop za
Reviewed By: aemerson
Differential Revision: https://reviews.llvm.org/D133894
Sander de Smalen [Sat, 17 Sep 2022 15:38:50 +0000 (15:38 +0000)]
[AArch64][SME] Disable tail-call optimization when streaming mode change or lazy-save may be required.
When a streaming mode change is (or may be) required for a call, it will
need to restore the original mode after the call, which prevents the use of
tail-call optimization. The same holds true for a call that requires the lazy-save
mechanism to be set up before the call, and possibly restored after.
More details about the SME attributes and design can be found
in D131562.
Reviewed By: aemerson
Differential Revision: https://reviews.llvm.org/D131579
Nikolas Klauser [Fri, 26 Aug 2022 16:35:25 +0000 (18:35 +0200)]
[libc++][NFC] Inline the string constructors
This removes a lot of boilerplate code.
Reviewed By: ldionne, #libc
Spies: EricWF, libcxx-commits
Differential Revision: https://reviews.llvm.org/D128081
Evgeny Shulgin [Wed, 14 Sep 2022 18:08:34 +0000 (18:08 +0000)]
[Clang] Support label at end of compound statement
Implements paper P2324R2
https://wg21.link/p2324r2
https://github.com/cplusplus/papers/issues/1006
Reviewed By: cor3ntin
Differential Revision: https://reviews.llvm.org/D133887
Dmitry Polukhin [Thu, 15 Sep 2022 14:39:10 +0000 (07:39 -0700)]
[clang][C++20] Fix clang/clangd assert/crash after compilation errors
After compilation errors, expression a transformation result may not be usable.
It triggers an assert in RemoveNestedImmediateInvocation and SIGSEGV in case of
builds without asserts. This issue significantly affects clangd because source
may not be valid during typing. Tests cases that I attached was reduce from huge
C++ translation unit.
Test Plan: check-clang
Differential Revision: https://reviews.llvm.org/D133948
Florian Hahn [Sat, 17 Sep 2022 13:47:38 +0000 (14:47 +0100)]
[ConstraintElimination] Fix crash when combining results.
f213128b292d didn't account for the possibility that the result of
decompose may be empty. Fix that by explicitly checking. Use a newly
introduced helper to also reduce some duplication.
Thanks @bjope for finding the issue!
Shivam Gupta [Sat, 17 Sep 2022 13:39:26 +0000 (19:09 +0530)]
[NFC] Fix a comment in InitializePasses.h
Aaron Ballman [Sat, 17 Sep 2022 12:43:17 +0000 (08:43 -0400)]
Add code examples to the potentially breaking changes
It may help users to better understand the change by showing them a
contrived code example which demonstrates the difference in behavior.
Nikolas Klauser [Fri, 2 Sep 2022 11:38:01 +0000 (13:38 +0200)]
[libc++] Add test to ensure that type trait aliases in dependent return types can be mangled
Reviewed By: ldionne, #libc
Spies: libcxx-commits, jeroen.dobbelaere
Differential Revision: https://reviews.llvm.org/D133196
Aaron Ballman [Sat, 17 Sep 2022 12:21:33 +0000 (08:21 -0400)]
Fix release note formatting and style; NFC
Uses double backticks where appropriate, changes some instances of
GH12345 to be Issue 12345, etc.
Aaron Ballman [Sat, 17 Sep 2022 12:06:16 +0000 (08:06 -0400)]
Fix this test to be more robust
The test is failing because it lacks a target triple, so the number of
diagnostics differs between Windows and Linux targets.
This should correct the issue found by:
https://lab.llvm.org/buildbot/#/builders/109/builds/46804
Aaron Ballman [Sat, 17 Sep 2022 11:55:10 +0000 (07:55 -0400)]
Correctly diagnose use of long long literals w/o a suffix
We would diagnose use of `long long` as an extension in C89 and C++98
modes when the user spelled the type `long long` or used the `LL`
literal suffix, but failed to diagnose when the literal had no suffix
but required a `long long` to represent the value.
Christian Sigg [Fri, 16 Sep 2022 20:27:04 +0000 (22:27 +0200)]
[Bazel] Allow lit_test() macro to be used from other repos.
Wrap implicit dependencies in Label() so that they refer to @llvm-project, see https://bazel.build/rules/lib/Label#Label.
This change allows lit_test() to be used from other bazel repositories.
Filipp Zhinkin [Fri, 16 Sep 2022 18:01:09 +0000 (21:01 +0300)]
[ARM] Add more tests on instructions fusion with comparison with zero; NFC
Baseline tests for D131786
Daniel Bertalan [Fri, 16 Sep 2022 16:11:55 +0000 (18:11 +0200)]
[lld-macho] Simplify base address calculation for init offsets (NFC)
Xiang Li [Sat, 17 Sep 2022 07:11:44 +0000 (00:11 -0700)]
[NFC} update CodeGenHLSL tests to use cc1 instead of driver-mode
Alex Zinenko [Fri, 16 Sep 2022 14:16:27 +0000 (16:16 +0200)]
[mlir] use strided layouts in vector transfer on memrefs
One of the vector transformation patterns has been indiscriminately
converting layouts to affine maps. Leverage the strided form when
possible.
Reviewed By: nicolasvasilache, dcaballe
Differential Revision: https://reviews.llvm.org/D134047
Alex Zinenko [Fri, 16 Sep 2022 13:36:40 +0000 (15:36 +0200)]
[mlir] use strided layout in structured codegen-related tests
All relevant operations have been switched to primarily use the strided
layout, but still support the affine map layout. Update the relevant
tests to use the strided format instead for compatibility with how ops
now print by default.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D134045
Emmmer [Sun, 11 Sep 2022 08:36:12 +0000 (16:36 +0800)]
[LLDB][RISCV] Add RVM and RVA instruction support for EmulateInstructionRISCV
Add:
- RVM and RVA instructions sets.
- corresponding unittests.
Further work:
- implement RVC, RVF, RVD, and RVV extension.
Reviewed By: DavidSpickett
Differential Revision: https://reviews.llvm.org/D133670
Kazu Hirata [Sat, 17 Sep 2022 01:26:19 +0000 (18:26 -0700)]
Revert "[llvm] Remove llvm::is_trivially_{copy/move}_constructible (NFC)"
This reverts commit
01ffe31cbb54bfd8e38e71b3cf804a1d67ebf9c1.
A build breakage with GCC 7.3 has been reported:
https://reviews.llvm.org/D132311#3797053
FWIW, GCC 7.5 is OK according to Pavel Chupin. I also personally
tested GCC 8.4.0.
Vladislav Dzhidzhoev [Fri, 9 Sep 2022 23:15:39 +0000 (02:15 +0300)]
[GlobalISel][DebugInfo] Salvage trivially dead instructions
Use salvageDebugInfo for instructions erased as trivially dead in
GlobalISel.
It would be helpful to implement support of G_PTR_ADD and G_FRAME_INDEX
in salvageDebugInfo in future in order to preserve more variable
location.
Reviewed by: arsenm
Differential Revision: https://reviews.llvm.org/D133986
Maksim Panchenko [Fri, 16 Sep 2022 23:45:19 +0000 (16:45 -0700)]
[BOLT][NFC] Remove unreachable assertion
Reviewed By: ayermolo
Differential Revision: https://reviews.llvm.org/D134094
Aart Bik [Fri, 16 Sep 2022 22:22:48 +0000 (15:22 -0700)]
[mlir][sparse] add loop simplification to sparsification pass
Reviewed By: bixia
Differential Revision: https://reviews.llvm.org/D134090
Matheus Izvekov [Sat, 18 Jun 2022 02:21:59 +0000 (04:21 +0200)]
[clang] Fix AST representation of expanded template arguments.
Extend clang's SubstTemplateTypeParm to represent the pack substitution index.
Fixes PR56099.
Signed-off-by: Matheus Izvekov <mizvekov@gmail.com>
Differential Revision: https://reviews.llvm.org/D128113
Peiming Liu [Fri, 16 Sep 2022 18:21:59 +0000 (18:21 +0000)]
[mlir][sparse] Only try to compute a better iteraton graph when needed
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D134059
Michael Jones [Thu, 15 Sep 2022 18:24:47 +0000 (11:24 -0700)]
[libc][cmake] separate installing headers
Now libc headers can be installed separately from installing the rest of
the libc.
Reviewed By: sivachandra
Differential Revision: https://reviews.llvm.org/D133960
Kazu Hirata [Fri, 16 Sep 2022 22:36:40 +0000 (15:36 -0700)]
[Inliner] Retire DefaultInlineOrder (NFC)
DefaultInlineOrder was largely an exercise in generalizing the
traversal order of call sites within the inliner.
Now that the module inliner is starting to form its shape, there is no
point in sharing DefaultInlineOrder between the module inliner and the
CGSCC inliner. DefaultInlineOrder and all the other inline orders are
mutually exclusive in the following sense:
- The use of DefaultInlineOrder doesn't make sense in the module
inliner because there is no priority inherent in the order in which
call sites are added to the list of call sites -- SmallVector.
- The use of any other inline order doesn't make sense in the CGSCC
inliner because little prioritization can be done within one CGSCC.
This patch essentially reverts the addition of DefaultInlineOrder so
that the loop structure of Inliner.cpp looks like the state just
before we started working on the module inliner (circa June 2021).
At the same time, ww remove the choice of DefaultInlineOrder from
UseInlinePriority.
Differential Revision: https://reviews.llvm.org/D134080
Jacques Pienaar [Fri, 16 Sep 2022 22:08:55 +0000 (15:08 -0700)]
[mlir][emacs] Enable loading bytecode files as text
Use auto-compression-mode to read bytecode files in human readable
manner.
Differential Revision: https://reviews.llvm.org/D133879
Alexey Bataev [Wed, 14 Sep 2022 19:28:31 +0000 (12:28 -0700)]
[SLP]Improve isUndefVector function by adding insertelement analysis.
Added the mask and the analysis of the buildvector sequence in the
isUndefVector function, improves codegen and cost estimation.
Metric: SLP.NumVectorInstructions
Program SLP.NumVectorInstructions
results results0 diff
test-suite :: External/SPEC/CFP2017rate/526.blender_r/526.blender_r.test 27362.00 27360.00 -0.0%
Metric: size..text
Program size..text
results results0 diff
test-suite :: External/SPEC/CFP2017rate/508.namd_r/508.namd_r.test 805299.00 806035.00 0.1%
526.blender_r - some extra code is vectorized.
508.namd_r - some extra code is optimized out.
Differential Revision: https://reviews.llvm.org/D133891
Siva Chandra Reddy [Fri, 16 Sep 2022 19:34:44 +0000 (19:34 +0000)]
[libc] Add implementation of POSIX "uname" function.
Reviewed By: lntue
Differential Revision: https://reviews.llvm.org/D134065
Siva Chandra Reddy [Fri, 16 Sep 2022 20:54:57 +0000 (20:54 +0000)]
[libc][Obvious] Fix typo in struct rlimit name - remove the "_t" suffix.
Chris Bieneman [Fri, 16 Sep 2022 19:42:40 +0000 (14:42 -0500)]
[HLSL] Enable availability attribute
Some HLSL functionality is gated on the target shader model version.
Enabling the use of availability markup allows us to diagnose
availability issues easily in the frontend.
Reviewed By: erichkeane
Differential Revision: https://reviews.llvm.org/D134067
Sotiris Apostolakis [Fri, 16 Sep 2022 20:34:37 +0000 (20:34 +0000)]
[SelectOpti] Restrict load sinking
This is a follow-up to D133777, which resolved a use-after-free case but
did not cover all possible memory bugs due to misplacement of loads.
In short, the overall problem was that sinked loads could be moved after
state-modifying instructions leading to memory bugs.
The solution is to restrict load sinking unless it is found to be sound.
i) Within a basic block (to-be-sinked load and select-user are in the same BB),
loads can be sinked only if there is no intervening state-modifying instruction.
This is a conservative approach to avoid resorting to alias analysis to detect
potential memory overlap.
ii) Across basic blocks, sinking of loads is avoided. This is because going over
multiple basic blocks looking for memory conflicts could be computationally
expensive and also unlikely to allow loads to sink. Further, experiments showed
that not sinking these loads has a slight positive performance effect.
Maybe for some of these loads, having some separation allows enough time
for the load to be executed in time for its user. This is not the case for
floating point operations that benefit more from sinking.
The solution in D133777 was essentially undone in this patch,
since the latter is a complete solution to the observed problem.
Overall, the performance impact of this patch is minimal.
Tested on two internal Google workloads with instrPGO.
Search application showed <0.05% perf difference,
while the database one showed a slight improvement,
but not statistically significant.
Reviewed By: davidxl
Differential Revision: https://reviews.llvm.org/D133999
Siva Chandra Reddy [Thu, 15 Sep 2022 22:22:59 +0000 (22:22 +0000)]
[libc] Add implementation of POSIX setrlimit and getrlimit functions.
Reviewed By: lntue
Differential Revision: https://reviews.llvm.org/D134016
Teresa Johnson [Fri, 16 Sep 2022 03:23:44 +0000 (20:23 -0700)]
[WPD/LTT] Lower type test feeding assumes via phi correctly
This fixes https://github.com/llvm/llvm-project/issues/57616.
Type test lowering in ThinLTO modules relies on having type id
summaries set up for the referenced types, which provide the type
test resolution. If there is no summary, the type tests are lowered
to false. At the very least, a default type id summary gives the
type tests a resolution of Unknown, which is handled correctly (ignored
by the first invocation of LTT, and lowered to true by the second).
WPD sets up the type id summaries (with a default type test resolution)
as it is processing the type tests, but only does this for the patterns
handled by WPD, which is a type test directly feeding an assume. In the
case of type tests feeding an assume via a phi, the type id summary was
not being set up, leading to the type tests being lowered to false
incorrectly.
Fix this by adding the default type id summary entries for all type ids
used on globals during index-only WPD.
This is not an issue for hybrid (split-lto-unit) LTO, as in that case
the type test resolution is determined and set up during LTT, since the
type definitions are in the regular LTO split module, and exported via
the summary to the ThinLTO split module.
Differential Revision: https://reviews.llvm.org/D134012
Lang Hames [Fri, 16 Sep 2022 18:23:08 +0000 (11:23 -0700)]
[ORC][ORC-RT][MachO] Reset __data and __common sections on library close.
If we want to be able to close and then re-open a library then we need to reset
the data section states when the library is closed. This commit updates
MachOPlatform and the ORC runtime to track __data and __common sections, and
reset the state in MachOPlatformRuntimeState::dlcloseDeinitialize.
This is only a first step to full support -- there are other data sections that
we're not capturing, and we'll probably want a more efficient representation
for the sections (rather than passing their string name over IPC), but this is
a reasonable first step.
This commit also contains a fix to MapperJITLinkMemoryManager that prevents it
from calling OnDeallocated twice in the case of an error.
Maksim Panchenko [Sat, 10 Sep 2022 01:06:13 +0000 (18:06 -0700)]
[BOLT] Change base class of ExecutableFileMemoryManager
When we derive EFMM from SectionMemoryManager, it brings into EFMM extra
functionality, such as the registry of exception handling sections,
page permission management, etc. Such functionality is of no use to
llvm-bolt and can even be detrimental (see
https://github.com/llvm/llvm-project/issues/56726).
Change the base class of ExecutableFileMemoryManager to MemoryManager,
avoid registering EH sections, and skip memory finalization.
Fixes #56726
Reviewed By: yota9
Differential Revision: https://reviews.llvm.org/D133994
Maksim Panchenko [Thu, 15 Sep 2022 20:31:52 +0000 (13:31 -0700)]
[BOLT] Fix empty function emission in non-relocation mode
In non-relocation mode, every function is emitted in its own section. If
a function is empty, RuntimeDyld will still allocate 1-byte section
for the function and initialize it with zero. As a result, we will
overwrite the first byte of the original function contents with zero.
Such scenario can happen when the input function had only NOP
instructions which BOLT removes by default. Even though such functions
likely cause undefined behavior, it's better to preserve their contents.
Reviewed By: yota9
Differential Revision: https://reviews.llvm.org/D133978
Jessica Paquette [Fri, 16 Sep 2022 20:32:42 +0000 (13:32 -0700)]
[GlobalISel] Combine select + fcmp to fminnum/fmaxnum/fminimum/fmaximum
This is a partial port of the code used by the SelectionDAGBuilder to
translate selects.
In particular, see matchSelectPattern in ValueTracking.cpp. This is a
GISel-equivalent of the portion which handles fminnum/fmaxnum/fminimum/fmaximum.
I tried to set it up so it'd be easy to add the non-FP cases. Those are simpler.
On the AArch64-end, it seems like the FP cases are more important for perf
right now, so I bit the bullet and went at the more complicated problem. :)
I elected to do this as a post-legalize combine rather than in the
IRTranslator because
Deciding which fmax/fmin to use can depend on legalization rules
Philosophically-speaking (TM), putting it in a combine just feels cleaner
Being able to enable/disable the combine is handy
Another option would be to use the ValueTracking code in the IRTranslator and
match what SelectionDAGBuilder::visitSelect does. I think that may be somewhat
annoying since we'd need to write lowerings back into the selects in the
legalizer. I'm not strongly opposed to the approach.
We'd also want to be careful with vector selects once that's implemented,
which explicitly check if a vector select is legal on the target. That'd
probably need a hook.
From what I can tell, doing this as a combine is probably a cleaner option
long-term.
Differential Revision: https://reviews.llvm.org/D116702
Pengxuan Zheng [Fri, 16 Sep 2022 00:52:46 +0000 (17:52 -0700)]
[MachineCSE] Add a threshold to avoid spending too much time in isProfitableToCSE
Currently, it can become extremely costly to compute MayIncreasePressure if the
size of CSUses turns out to be very large. In that case, it's no longer cost
effective to keep computing MayIncreasePressure. Therefore, to limit the amount
of time spent in isProfitableToCSE, we simply conservatively assume
MayIncreasePressure if the size of CSUses is too large. This can reduce overall
compile time by 30% for some benchmarks.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D134003
Craig Topper [Fri, 16 Sep 2022 19:59:13 +0000 (12:59 -0700)]
[VP][VE] Default VP_SREM/UREM to Expand and add generic expansion using VP_SDIV/UDIV+VP_MUL+VP_SUB.
I want to default all VP operations to Expand. These 2 were blocking
because VE doesn't support them and the tests were expecting them
to fail a specific way. Using Expand caused them to fail differently.
Seemed better to emulate them using operations that are supported.
@simoll mentioned on Discord that VE has some expansion downstream. Not
sure if its done like this or in the VE target.
Reviewed By: frasercrmck, efocht
Differential Revision: https://reviews.llvm.org/D133514
Xing Xue [Fri, 16 Sep 2022 20:08:40 +0000 (16:08 -0400)]
[libc++][lit][AIX] Enable test case last_write_time.pass.cpp for AIX
Summary:
This patch enables libc++ LIT test case last_write_time.pass.cpp for AIX. Because system call utimensat() of AIX which is used in the libc++ implementation of last_write_time() does not accept the times parameter with a negative tv_sec or tv_nsec field, testing of setting file time to before epoch time is excluded for AIX.
Reviewed by: ldionne, libc++
Differential Revision: https://reviews.llvm.org/D133124
Craig Topper [Fri, 16 Sep 2022 19:55:31 +0000 (12:55 -0700)]
[RISCV] Simplify some code in vector fp<->int handling. NFC
We changed the way container types are selected since this code
was written. We no longer need to use the largest type.
Aiden Grossman [Fri, 16 Sep 2022 19:35:50 +0000 (19:35 +0000)]
[Clang] Give error message for invalid profile path when compiling IR
Before this patch, when compiling an IR file (eg the .llvmbc section
from an object file compiled with -Xclang -fembed-bitcode=all) and
profile data was passed in using the -fprofile-instrument-use-path
flag, there would be no error printed (as the previous implementation
relied on the error getting caught again in the constructor of
CodeGenModule which isn't called when -x ir is set). This patch
moves the error checking directly to where the error is caught
originally rather than failing silently in setPGOUseInstrumentor and
waiting to catch it in CodeGenModule to print diagnostic information to
the user.
Regression test added.
Reviewed By: xur, mtrofin
Differential Revision: https://reviews.llvm.org/D132991