Eli Friedman [Mon, 26 Jul 2021 18:59:40 +0000 (11:59 -0700)]
Fix clang regression test after
5c486ce0
Michael Liao [Sun, 18 Jul 2021 17:45:52 +0000 (13:45 -0400)]
[amdgpu] Add 64-bit PC support when expanding unconditional branches.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D106445
Craig Topper [Mon, 26 Jul 2021 18:46:31 +0000 (11:46 -0700)]
[TypePromotion] Remove redundant if. NFC
The same condition was checked in the previous if. Maybe this was
a bad merge resolution?
Raul Tambre [Wed, 21 Jul 2021 17:38:24 +0000 (20:38 +0300)]
[compiler-rt][CMake][arm64] Use a custom target for symlinking LSE sources
On Apple platforms the builtins may be built for both arm64 and arm64e.
With Makefile generators separate targets are built using Make sub-invocations.
This causes a race when creating the symlink which may sometimes fail.
Work around this by using a custom target that the builtin targets depend on.
This causes any sub-invocations to depend on the symlinks having been created before.
Mailing list thread: https://lists.llvm.org/pipermail/llvm-dev/2021-July/151822.html
Reviewed By: thakis, steven_wu
Differential Revision: https://reviews.llvm.org/D106305
Nemanja Ivanovic [Mon, 26 Jul 2021 17:35:19 +0000 (12:35 -0500)]
[PowerPC] Implement partial vector ld/st builtins for XL compatibility
XL provides functions __vec_ldrmb/__vec_strmb for loading/storing a
sequence of 1 to 16 bytes in big endian order, right justified in the
vector register (regardless of target endianness).
This is equivalent to vec_xl_len_r/vec_xst_len_r which are only
available on Power9.
This patch simply uses the Power9 functions when compiled for Power9,
but provides a more general implementation for Power8.
Differential revision: https://reviews.llvm.org/D106757
Derek Schuff [Fri, 23 Jul 2021 23:57:37 +0000 (16:57 -0700)]
[lld][WebAssembly] Do not remove name section with --strip-debug
Leave the name section in the output when using the --strip-debug
flag. This treats it more like ELF symbol tables, as the name
section has similar uses at runtime (e.g. wasm engines understand
it and it can be used for symbolization at runtime).
Fixes https://github.com/emscripten-core/emscripten/issues/14623
Differential Revision: https://reviews.llvm.org/D106728
Kevin P. Neal [Mon, 26 Jul 2021 17:57:12 +0000 (13:57 -0400)]
[FPEnv][InstSimplify] Enable more folds for constrained fadd
Precommit tests, try 2. My tree is up-to-date as of this morning so this
should go better than my first try.
Omar Emara [Mon, 26 Jul 2021 18:04:21 +0000 (11:04 -0700)]
[LLDB][GUI] Resolve paths in file/directory fields
This patch resolves the paths in the file/directory fields before
performing checks. Those checks are applied on the file system if
m_need_to_exist is true, so remote files can set this to false to avoid
performing host-side file system checks. Additionally, methods to get
a resolved and a direct file specs were added to be used by client code.
Reviewed By: clayborg
Differential Revision: https://reviews.llvm.org/D106553
Amara Emerson [Sun, 25 Jul 2021 14:32:44 +0000 (07:32 -0700)]
[AArch4][GlobalISel] Post-legalize combine s64 = G_MERGE s32, 0 -> G_ZEXT.
These are generated as a byproduce of legalization.
Differential Revision: https://reviews.llvm.org/D106768
Eli Friedman [Mon, 26 Jul 2021 17:51:00 +0000 (10:51 -0700)]
[LLVM IR] Allow volatile stores to trap.
Proposed alternative to D105338.
This is ugly, but short-term I think it's the best way forward: first,
let's formalize the hacks into a coherent model. Then we can consider
extensions of that model (we could have different flavors of volatile
with different rules).
Differential Revision: https://reviews.llvm.org/D106309
Amara Emerson [Sun, 25 Jul 2021 07:47:03 +0000 (00:47 -0700)]
[AArch64][GlobalISel] Enable some select combines after legalization.
The legalizer generates selects for some operations, which can have constant
condition values, resulting in lots of dead code if it's not folded away.
Differential Revision: https://reviews.llvm.org/D106762
Amara Emerson [Sun, 25 Jul 2021 06:44:29 +0000 (23:44 -0700)]
[GlobalISel] Add combine for merge(unmerge) and use AArch64 postlegal-combiner.
Differential Revision: https://reviews.llvm.org/D106761
Michael Jones [Fri, 23 Jul 2021 22:15:03 +0000 (22:15 +0000)]
[libc] add scudo wrappers to llvm libc
The previous patch included the implementations for the scudo allocator,
but not the wrappers. This change fixes that.
Reviewed By: sivachandra
Differential Revision: https://reviews.llvm.org/D106718
Heejin Ahn [Thu, 22 Jul 2021 05:29:31 +0000 (22:29 -0700)]
[WebAssembly] Improve pseudocode in LowerEmscriptenEHSjLj
Both `__THREW__` and `__threwValue` are global variables, and we have
been distinguishing the global variable `__THREW__` and the loaded value
`%__THREW__.val` in comments but not doing it for `__threwValue`. Made
the pseudocode comments consistent for both variables.
Reviewed By: dschuff
Differential Revision: https://reviews.llvm.org/D106524
Rob Suderman [Thu, 22 Jul 2021 22:12:43 +0000 (15:12 -0700)]
[mlir][tosa] Disable tosa shape verification between operands/results
Tosa shape verification prevent shape propagation when coming from a dialect
of known shape. Relax this constraint to allow ingestion / shape propagation
from these other dialects.
Differential Revision: https://reviews.llvm.org/D106610
Fangrui Song [Mon, 26 Jul 2021 17:00:05 +0000 (10:00 -0700)]
[test] Fix PayloadString: in lldb tests
Simon Pilgrim [Mon, 26 Jul 2021 16:57:09 +0000 (17:57 +0100)]
[X86][AVX] Add PR50053 test case
Florian Hahn [Mon, 26 Jul 2021 16:38:10 +0000 (17:38 +0100)]
[LAA] Remove RuntimeCheckingPtrGroup::RtCheck member (NFC).
This patch removes RtCheck from RuntimeCheckingPtrGroup to make it
possible to construct RuntimeCheckingPtrGroup objects without a
RuntimePointerChecking object. This should make it easier to
re-use the code to generate runtime checks, e.g. in D102834.
RtCheck was only used to access the pointer info for a given index.
Instead, the start and end expressions can be passed directly.
For code-gen, we also need to know the address space to use. This can
also be explicitly passed at construction.
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D105481
Stephen Tozer [Mon, 26 Jul 2021 08:59:20 +0000 (09:59 +0100)]
[DebugInfo] Correctly update debug users of SSA values in tail duplication
During tail duplication, SSA values may be updated and have their uses
replaced with a virtual register, and any debug instructions that use
that value are deleted. This patch fixes the implementation of the debug
instruction deletion to work correctly for debug instructions that use
the SSA value multiple times, by batching deletions so that we don't
attempt to delete the same instruction twice.
Differential Revision: https://reviews.llvm.org/D106557
Qiu Chaofan [Mon, 26 Jul 2021 16:20:56 +0000 (00:20 +0800)]
[PowerPC] Change altivec indexed load/store builtins argument type
This patch changes the index argument of lvxl?/lve[bhw]x and
stvxl?/stve[bhw]x builtins from int to long. Because on 64-bit
subtargets, an extra extsw will always been generated, which is
incorrect.
Reviewed By: nemanjai
Differential Revision: https://reviews.llvm.org/D106530
Louis Dionne [Fri, 23 Jul 2021 20:46:31 +0000 (16:46 -0400)]
[libc++] Add range_size_t
Differential Revision: https://reviews.llvm.org/D106708
Simon Pilgrim [Mon, 26 Jul 2021 16:15:15 +0000 (17:15 +0100)]
[Analysis] Fix getOrderedReductionCost to call target's getArithmeticInstrCost implementation
The getOrderedReductionCost implementation introduced in D105432 calls the CRTP base version getArithmeticInstrCost instead of the redirecting to the target version.
Differential Revision: https://reviews.llvm.org/D106795
Sander de Smalen [Mon, 26 Jul 2021 15:04:25 +0000 (16:04 +0100)]
[LV] Remove assert that VF cannot be scalable in setCostBasedWideningDecision.
Scalarization for scalable vectors is not (yet) supported, so the
LV discards a VF when scalarization is chosen as the widening
decision. It should therefore not assert that the VF is not scalable
when it computes the decision to scalarize.
The code can get here when both the interleave-cost, gather/scatter cost
and scalarization-cost are all illegal. This may e.g. happen for SVE
when the VF=1, to avoid generating `<vscale x 1 x eltty>` types that
the code-generator cannot yet handle.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D106656
Christopher Di Bella [Thu, 22 Jul 2021 20:30:43 +0000 (20:30 +0000)]
[clang][pp] adds '#pragma include_instead'
`#pragma clang include_instead(<header>)` is a pragma that can be used
by system headers (and only system headers) to indicate to a tool that
the file containing said pragma is an implementation-detail header and
should not be directly included by user code.
The library alternative is very messy code that can be seen in the first
diff of D106124, and we'd rather avoid that with something more
universal.
This patch takes the first step by warning a user when they include a
detail header in their code, and suggests alternative headers that the
user should include instead. Future work will involve adding a fixit to
automate the process, as well as cleaning up modules diagnostics to not
suggest said detail headers. Other tools, such as clangd can also take
advantage of this pragma to add the correct user headers.
Differential Revision: https://reviews.llvm.org/D106394
Nikita Popov [Mon, 26 Jul 2021 16:05:47 +0000 (18:05 +0200)]
[MergeICmps] Collect block instructions once (NFC)
Collect the relevant instructions for a given BCECmpBlock once
on construction, rather than repeating this logic in multiple
places.
Fangrui Song [Mon, 26 Jul 2021 16:05:18 +0000 (09:05 -0700)]
[llvm-objcopy] Drop GRP_COMDAT if the group signature is localized
See [GRP_COMDAT group with STB_LOCAL signature](https://groups.google.com/g/generic-abi/c/2X6mR-s2zoc)
objcopy PR: https://sourceware.org/bugzilla/show_bug.cgi?id=27931
GRP_COMDAT deduplication is purely based on the signature symbol name in
ld.lld/GNU ld/gold. The local/global status is not part of the equation.
If the signature symbol is localized by --localize-hidden or
--keep-global-symbol, the intention is likely to make the group fully
localized. Drop GRP_COMDAT to suppress deduplication.
Reviewed By: jhenderson
Differential Revision: https://reviews.llvm.org/D106782
Fangrui Song [Mon, 26 Jul 2021 16:04:51 +0000 (09:04 -0700)]
[yaml2obj][MachO] Rename PayloadString to Content
The new name is conciser and matches yaml2obj ELF & DWARF.
Reviewed By: #lld-macho, thakis
Differential Revision: https://reviews.llvm.org/D106759
Nikita Popov [Mon, 26 Jul 2021 15:29:07 +0000 (17:29 +0200)]
[MergeICmps] Try to fix MSVC build failure
Apparently this fails to line up the types -- try to sidestep the
issue entirely by writing the code in a more reasonable way: Walk
over the operands and perform a set lookup, rather than walking
over the set and performing an operand scan.
Kazu Hirata [Mon, 26 Jul 2021 15:29:32 +0000 (08:29 -0700)]
[AsmParser] Remove MDRef (NFC)
The last use was removed on Jan 12, 2015 in commit
ab617d597708fcf3c4b829bf595e9d990ca66c07.
Paul Walker [Thu, 22 Jul 2021 16:16:20 +0000 (17:16 +0100)]
[SVE] Use reg+reg addressing mode for immediate offsets.
For reg+imm SVE addressing mode imm is implictly scaled by VL,
making them impractical for truely immediate offsets. However, if
the offset can be unscaled based on the storage element type we
can use the reg+reg SVE addressing mode and thus either reduce the
number of generate add instructions or replace them with a mov
instruction that can be hoisted from the hot code path.
Differential Revision: https://reviews.llvm.org/D106744
Louis Dionne [Mon, 26 Jul 2021 15:22:34 +0000 (11:22 -0400)]
[libc++][NFC] Change a few instances of > > to >> in C++20 code
Sanjay Patel [Mon, 26 Jul 2021 14:52:43 +0000 (10:52 -0400)]
[SimplifyLibCalls] avoid crash on pointer math
We could try harder to screen out libcalls by
function signature (and that would be a much larger
change than for sprintf alone), but that might make
the transition to type-less pointers more difficult.
https://llvm.org/PR51200
Sanjay Patel [Mon, 26 Jul 2021 14:28:43 +0000 (10:28 -0400)]
[SimplifyLibCalls] reduce code duplication; NFC
Nikita Popov [Mon, 26 Jul 2021 12:40:12 +0000 (14:40 +0200)]
[MergeICmps] Separate out BCECmp and use Optional (NFC)
Separate out the BCECmp part from BCECmpBlock, which just stores
the comparison atoms without the branch instruction. At the same
time switch the code to return Optional<> rather than objects in
invalid state and partially constructed objects.
Sander de Smalen [Mon, 26 Jul 2021 14:22:01 +0000 (15:22 +0100)]
[LV] Don't assume isScalarAfterVectorization if one of the uses needs widening.
This fixes an issue that was found in D105199, where a GEP instruction
is used both as the address of a store, as well as the value of a store.
For the former, the value is scalar after vectorization, but the latter
(as value) requires widening.
Other code in that function seems to prevent similar cases from happening,
but it seems this case was missed.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D106164
Bradley Smith [Thu, 8 Jul 2021 16:12:54 +0000 (16:12 +0000)]
[AArch64][SVE] Break false dependencies for inactive lanes of unary operations
Differential Revision: https://reviews.llvm.org/D105889
Shilei Tian [Mon, 26 Jul 2021 15:00:32 +0000 (11:00 -0400)]
[Clang][OpenMP] Remove the mandatory flush for capture for OpenMP 5.1
In OpenMP 5.1:
> If the `write` or `update` clause is specifieded, the atomic operation is not an atomic conditional update for which the comparison fails, and the effective memory ordering is `release`, `acq_rel`, or `seq_cst`, the strong flush on entry to the atomic operation is also a release flush. If the `read` or `update` clause is specified and the effective memory ordering is `acquire`, `acq_rel`, or `seq_cst` then the strong flush on exit from the atomic operation is also an acquire flush.
In OpenMP 5.0:
> If the `write`, `update`, or **`capture`** clause is specified and the `release`, `acq_rel`, or `seq_cst` clause is specified then the strong flush on entry to the atomic operation is also a release flush. If the `read` or `capture` clause is specified and the `acquire`, `acq_rel`, or `seq_cst` clause is specified then the strong flush on exit from the atomic operation is also an acquire flush.
From my understanding, in OpenMP 5.1, `capture` is removed from the requirement for flush, therefore we don't have to enforce it.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D100768
Ulrich Weigand [Tue, 20 Jul 2021 16:14:38 +0000 (18:14 +0200)]
[SystemZ] Add support for new cpu architecture - arch14
This patch adds support for the next-generation arch14
CPU architecture to the SystemZ backend.
This includes:
- Basic support for the new processor and its features.
- Detection of arch14 as host processor.
- Assembler/disassembler support for new instructions.
- New LLVM intrinsics for certain new instructions.
- Support for low-level builtins mapped to new LLVM intrinsics.
- New high-level intrinsics in vecintrin.h.
- Indicate support by defining __VEC__ == 10304.
Note: No currently available Z system supports the arch14
architecture. Once new systems become available, the
official system name will be added as supported -march name.
Florian Hahn [Mon, 26 Jul 2021 14:21:16 +0000 (15:21 +0100)]
Recommit "[VPlan] Add recipe for first-order rec phis, make splicing explicit."
This reverts the revert commit
b1777b04dc4b1a9fee0e7effa7e177892ab32ef0.
The patch originally got reverted due to a crash:
https://bugs.chromium.org/p/chromium/issues/detail?id=1232798#c2
The underlying issue was that we were not using the stored values from
the modified memory recipes, but the out-of-date values directly from
the IR (accessed via the VPlan). This should be fixed in
d995d6376. A
reduced version of the reproducer has been added in
93664503be6b.
Mark de Wever [Thu, 22 Jul 2021 09:17:53 +0000 (11:17 +0200)]
[libc++][ci] Detect not committed generated files.
The Generated output CI job only tests for modified files. This job
should also fail the generated output contains new files.
It would be possible to test modified and untracked files in one
execution of `git ls-files`. However the diff is stored as an artifact
so the execution of `git diff` would still be required.
Discussion: Would it be better to do `git ls-files -om` and remove the
excution of
`! grep -q '^--- a' ${BUILD_DIR}/generated_output.patch || false` ?
(Obviously then the name `generated_output.untracked` should change to
something like `generated_output.status`)
Reviewed By: #libc, ldionne
Differential Revision: https://reviews.llvm.org/D106534
Nikita Popov [Wed, 21 Jul 2021 19:23:38 +0000 (21:23 +0200)]
[IR] Consider non-willreturn as side effect (PR50511)
This adjusts mayHaveSideEffect() to return true for !willReturn()
instructions. Just like other side-effects, non-willreturn calls
(aka "divergence") cannot be removed and cannot be reordered relative
to other side effects. This fixes a number of bugs where
non-willreturn calls are either incorrectly dropped or moved. In
particular, it also fixes the last open problem in
https://bugs.llvm.org/show_bug.cgi?id=50511.
I performed a cursory review of all current mayHaveSideEffect()
uses, which convinced me that these are indeed the desired default
semantics. Places that do not want to consider non-willreturn as a
sideeffect generally do not want mayHaveSideEffect() semantics at
all. I identified two such cases, which are addressed by D106591
and D106742. Finally, there is a use in SCEV for which we don't
really have an appropriate API right now -- what it wants is
basically "would this be considered forward progress". I've just
spelled out the previous semantics there.
Differential Revision: https://reviews.llvm.org/D106749
Benjamin Kramer [Mon, 26 Jul 2021 14:32:22 +0000 (16:32 +0200)]
Simplify away some SmallVector copies. NFCI.
The lifetime of the initializer list is the full expression, so we can
skip storing it in a temporary vector.
Gabor Marton [Tue, 20 Jul 2021 14:46:15 +0000 (16:46 +0200)]
[Analyzer][solver][NFC] print constraints deterministically (ordered by their string representation)
This change is an extension to D103967 where I added dump methods for
(dis)equality classes of the State. There, the (dis)equality classes and their
contents are dumped in an ordered fashion, they are ordered based on their
string representation. This is very useful once we start to use FileCheck to
test the State dump in certain tests.
Differential Revision: https://reviews.llvm.org/D106642
Jeremy Morse [Mon, 26 Jul 2021 14:00:08 +0000 (15:00 +0100)]
[InstrRef][AArch64][1/4] Accept constant physreg variable locations
Late in SelectionDAG we join up instruction numbers with their defining
instructions, if it couldn't be done during the main part of SelectionDAG.
One exception is function arguments, where we have to point a DBG_PHI
instruction at the incoming live register, as they don't have a defining
instruction. This patch adds another exception, for constant physregs, like
aarch64 has.
It may seem wasteful to use two instructions where we could use a single
DBG_VALUE, however the whole point of instruction referencing is to
decouple the identification of values from the specification of where
variable location ranges start.
(Part of my aarch64 work to ease adoption of instruction referencing, as
in the meta comment on D104520)
Differential Revision: https://reviews.llvm.org/D104520
Florian Hahn [Mon, 26 Jul 2021 13:47:00 +0000 (14:47 +0100)]
[LV] Add test to store a first-order rec via interleave group.
This is a reduced version of the reproducer from
https://bugs.chromium.org/p/chromium/issues/detail?id=1232798#c2
Alexey Bataev [Mon, 26 Jul 2021 12:43:53 +0000 (05:43 -0700)]
[SLP]Fix costs calculations.
Need to fix several cost-related problems. The final type may be defined
incorrectly because of to early definition (we may end up with the wider
type), the CommonCost should not be redefined in ExtractElements
cost related calculations and the shuffle of the final insertelements
vectors should be calculated as a cost of single vector permutations
+ costs of two vector permutations for other n-1 incoming vectors.
Differential Revision: https://reviews.llvm.org/D106578
Anastasia Stulova [Mon, 26 Jul 2021 13:59:47 +0000 (14:59 +0100)]
[OpenCL] Change default standard version to CL1.2
Set default version for OpenCL C to 1.2. This means that the
absence of any standard flag will be equivalent to passing
'-cl-std=CL1.2'.
Note that this patch also fixes incorrect version check for
the pointer to pointer kernel arguments diagnostic and
atomic test.
Differential Revision: https://reviews.llvm.org/D106504
gbreynoo [Mon, 26 Jul 2021 13:49:53 +0000 (14:49 +0100)]
[llvm-readobj] Display multiple function names for stack size entries
The current implementation of displaying .stack_size information
presumes that each entry represents a single function but this is not
always the case. For example with the use of ICF multiple functions can
be represented with the same code, meaning that the address found in a
.stack_size entry corresponds to multiple function symbols.
This change allows multiple function names to be displayed when
appropriate.
Differential Revision: https://reviews.llvm.org/D105884
Jay Foad [Wed, 21 Jul 2021 14:53:59 +0000 (15:53 +0100)]
[AMDGPU][GISel] Fix MMO for raw/struct buffer access with non-constant offset
Codegen for the raw/struct buffer access intrinsics would update the
offset in the MMO to reflect the combined offset, if it was known to be
constant. If the combined offset was not known to be constant, or if
there was an index, it would set the offset in the MMO to 0. This is
unsafe because it makes it look like the access does not alias with
another access with a fixed non-zero offset.
Fix these cases by setting the pointer in the MMO to null, to reflect
the fact that we do not have any known IR value pointer + constant
offset for the access.
D106284 did this for SelectionDAG. This is the corresponding fix for
GlobalISel.
Differential Revision: https://reviews.llvm.org/D106451
Jay Foad [Wed, 21 Jul 2021 10:29:04 +0000 (11:29 +0100)]
[AMDGPU] Pre-commit global-isel test case for D106451
This test case shows the scheduler wrongly reordering two buffer
accesses that might alias.
Jay Foad [Mon, 19 Jul 2021 15:29:39 +0000 (16:29 +0100)]
[AMDGPU] Fix MMO for raw/struct buffer access with non-constant offset
Codegen for the raw/struct buffer access intrinsics would update the
offset in the MMO to reflect the combined offset, if it was known to be
constant. If the combined offset was not known to be constant, or if
there was an index, it would set the offset in the MMO to 0. This is
unsafe because it makes it look like the access does not alias with
another access with a fixed non-zero offset.
Fix these cases by setting the pointer in the MMO to null, to reflect
the fact that we do not have any known IR value pointer + constant
offset for the access.
Differential Revision: https://reviews.llvm.org/D106284
David Green [Mon, 26 Jul 2021 13:26:38 +0000 (14:26 +0100)]
[ARM] Ensure correct regclass in distributing postinc
The register class required for some MVE loads/stores is more
constrained than the register we use when creating postinc. Make sure we
constrain the register class to keep the code correct.
Tim Northover [Mon, 26 Jul 2021 12:53:34 +0000 (13:53 +0100)]
AArch64: support i128 (& larger) returns in GlobalISel
Nikita Popov [Wed, 21 Jul 2021 20:34:28 +0000 (22:34 +0200)]
[SimplifyCFG] Improve store speculation check
isSafeToSpeculateStore() looks for a preceding store to the same
location to make sure that introducing a new store of the same
value is safe. It currently bails on intervening mayHaveSideEffect()
instructions. However, I believe just checking mayWriteToMemory()
is sufficient there -- we just need to make sure that we know which
value was stored, we don't care if we can unwind in the meantime.
While looking into this, I started having some doubts about the
correctness of the transform with regard to thread safety. While
we don't try to hoist non-simple stores, I believe we also need
to make sure that the preceding store is simple as well. Otherwise
we could introduce a spurious non-atomic write after an atomic write
-- under our memory model this would result in a subsequent undef
atomic read, even if the second write stores the same value as the
first.
Example: https://alive2.llvm.org/ce/z/q_3YAL
Differential Revision: https://reviews.llvm.org/D106742
Kerry McLaughlin [Mon, 26 Jul 2021 09:55:15 +0000 (10:55 +0100)]
[SVE] Fix casts to <FixedVectorType> in truncateToMinimalBitwidths
Fixes more casts to `<FixedVectorType>` for the cases where the
instruction is a Insert/ExtractElementInst.
For fixed-width, this part of truncateToMinimalBitWidths is tested by
AArch64/type-shrinkage-insertelt.ll. I attempted to write a test case for this part
of truncateToMinimalBitWidths which uses scalable vectors, but was unable to add
one. The tests in type-shrinkage-insertelt.ll rely on scalarization to create extract
element instructions for instance, which is not possible for scalable vectors.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D106163
Alexey Bataev [Mon, 26 Jul 2021 12:42:11 +0000 (05:42 -0700)]
Revert "[SLP]Fix costs calculations."
This reverts commit
a053afed49897aa34e08287f91c5255efa4e5131 to fix
buildbots.
Caroline Concatto [Mon, 26 Jul 2021 12:08:33 +0000 (13:08 +0100)]
[AArch65][SVE] Remove vector_splice from AddedComplexity pattern
The pattern for vector_splice with Index equal or bigger than
zero was misplaced in the AddedComplexity = 1 pattern in the AArch64
tablegen file. This patch fixes it by removing vector_splice pattern
from inside AddedComplexity = 1.
Tres Popp [Mon, 26 Jul 2021 12:15:37 +0000 (14:15 +0200)]
[mlir] split type conversion to two lines for GCC's sake
Alexey Bataev [Thu, 22 Jul 2021 17:48:36 +0000 (10:48 -0700)]
[SLP]Fix costs calculations.
Need to fix several cost-related problems. The final type may be defined
incorrectly because of to early definition (we may end up with the wider
type), the CommonCost should not be redefined in ExtractElements
cost related calculations and the shuffle of the final insertelements
vectors should be calculated as a cost of single vector permutations
+ costs of two vector permutations for other n-1 incoming vectors.
Differential Revision: https://reviews.llvm.org/D106578
Paul Walker [Sat, 24 Jul 2021 15:34:20 +0000 (16:34 +0100)]
[NFC] Change VFShape so it contains an ElementCount rather than seperate VF and IsScalable properties.
Differential Revision: https://reviews.llvm.org/D106750
Philipp Krones [Wed, 14 Jul 2021 11:21:40 +0000 (12:21 +0100)]
[Inliner] Make the CallPenalty configurable
Tests with multiple benchmarks, like Embench [1], showed that the
CallPenalty magic number has the most influence on inlining decisions
when optimizing for size.
On the other hand, there was no good default value for this parameter.
Some benchmarks profited strongly from a reduced call penalty. On
example is the picojpeg benchmark compiled for RISC-V, which got 6%
smaller with a CallPenalty of 10 instead of 12. Other benchmarks
increased in size, like matmult.
This commit makes the compromise of turning the magic number constant of
CallPenalty into a configurable value. This introduces the flag
`--inline-call-penalty`. With that flag users can fine tune the inliner
to their needs.
The CallPenalty constant was also used for loops. This commit replaces
the CallPenalty constant with a new LoopPenalty constant that is now
used instead.
This is a slimmed down version of https://reviews.llvm.org/D30899
[1]: https://github.com/embench/embench-iot
Differential Revision: https://reviews.llvm.org/D105976
Florian Hahn [Mon, 26 Jul 2021 10:38:39 +0000 (11:38 +0100)]
[VPlan] Use stored value from recipes for interleave groups.
Instead of getting the VPValue for the stored IR values through the
current plan, use the stored value of the recipes directly.
This way, the correct VPValues are used if the store recipes have been
modified in the VPlan and the IR value is not correct any longer. This
can happen, e.g. due to D105008.
Dylan Fleming [Mon, 26 Jul 2021 10:26:28 +0000 (11:26 +0100)]
[SVE] Add support for folding for select + masked loads
Add folds to instcombine to support the removal of select instruction when the masked_load is guaranteed to zero the same lanes, i.e. select(mask, mload(,,mask,0), 0) -> mload(,,mask,0).
Patch originally authored by @paulwalker-arm
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D106376
Caroline Concatto [Mon, 19 Jul 2021 10:14:20 +0000 (11:14 +0100)]
[SVE][AArch64] Improve code generation for vector_splice for Imm > 0
This patch implements vector_splice in tablegen for all cases when the
Immediate is positive and lower than the known minimum value of
a scalable vector.
Vector_splice can be implemented using SVE instruction EXT.
For instance :
@llvm.experimental.vector.splice(Vector_1, Vector_2, Imm)
@llvm.experimental.vector.splice(<A,B,C,D>, <E,F,G,H>, 1) ==> <B, C, D, E>
EXT Vector_1, Vector_2, Imm // Vector_1 = B, C, D + Vector_2 = E
Depends on D105633
Differential Revision: https://reviews.llvm.org/D106273
David Sherwood [Mon, 26 Jul 2021 10:38:49 +0000 (11:38 +0100)]
Caroline Concatto [Tue, 6 Jul 2021 08:19:16 +0000 (09:19 +0100)]
[AArch64][SVE] Improve code generation for vector_splice for Imm == -1
This patch implements vector_splice in tablegen for:
a) when the immediate is equal to -1 (Imm==1) and uses:
INSR + LASTB
For instance :
@llvm.experimental.vector.splice(Vector_1, Vector_2, -1)
@llvm.experimental.vector.splice(<A,B,C,D>, <E,F,G,H>, 1) ==> <D, E, F, G>
LAST RegLast, Vector_1 // RegLast = D
INSR Res, (Vector_1 >> 1), RegLast // Res = D + E, F, G
Differential Revision: https://reviews.llvm.org/D105633
Simon Pilgrim [Mon, 26 Jul 2021 10:11:56 +0000 (11:11 +0100)]
[X86][AVX] Prefer vinsertf128 to vperm2f128 on AVX1 targets
Splatting the lower xmm with vinsertf128 is at least as quick as vperm2f128, and a lot faster on some AMD targets.
First step towards PR50053
Simon Pilgrim [Mon, 26 Jul 2021 10:03:31 +0000 (11:03 +0100)]
[X86][SSE] Don't scrub address math from interleaved shuffle tests
Sam McCall [Mon, 26 Jul 2021 08:42:03 +0000 (10:42 +0200)]
Revert "[clangd] Avoid range-loop init-list lifetime subtleties."
This reverts commit
253b8145dedbe8d10792f44b4af7f52dbecd527f.
This doesn't actually fix anything - I should stop guessing.
See https://github.com/clangd/clangd/issues/800 for update
Cullen Rhodes [Mon, 26 Jul 2021 09:23:59 +0000 (09:23 +0000)]
[AArch64][AsmParser] NFC: Parser.getTok().getLoc() -> getLoc()
Reviewed By: tmatheson
Differential Revision: https://reviews.llvm.org/D106635
David Sherwood [Wed, 7 Jul 2021 12:18:20 +0000 (13:18 +0100)]
[Analysis] Add simple cost model for strict (in-order) reductions
I have added a new FastMathFlags parameter to getArithmeticReductionCost
to indicate what type of reduction we are performing:
1. Tree-wise. This is the typical fast-math reduction that involves
continually splitting a vector up into halves and adding each
half together until we get a scalar result. This is the default
behaviour for integers, whereas for floating point we only do this
if reassociation is allowed.
2. Ordered. This now allows us to estimate the cost of performing
a strict vector reduction by treating it as a series of scalar
operations in lane order. This is the case when FP reassociation
is not permitted. For scalable vectors this is more difficult
because at compile time we do not know how many lanes there are,
and so we use the worst case maximum vscale value.
I have also fixed getTypeBasedIntrinsicInstrCost to pass in the
FastMathFlags, which meant fixing up some X86 tests where we always
assumed the vector.reduce.fadd/mul intrinsics were 'fast'.
New tests have been added here:
Analysis/CostModel/AArch64/reduce-fadd.ll
Analysis/CostModel/AArch64/sve-intrinsics.ll
Transforms/LoopVectorize/AArch64/strict-fadd-cost.ll
Transforms/LoopVectorize/AArch64/sve-strict-fadd-cost.ll
Differential Revision: https://reviews.llvm.org/D105432
Fraser Cormack [Thu, 22 Jul 2021 17:03:40 +0000 (18:03 +0100)]
[SelectionDAG] Support scalable-vector splats in yet more cases
This patch extends support for (scalable-vector) splats in the
DAGCombiner via the `ISD::matchBinaryPredicate` function, which enable a
variety of simple combines of constants.
Users of this function may now have to distinguish between
`BUILD_VECTOR` and `SPLAT_VECTOR` vector operands. The way of dealing
with this in-tree follows the approach added for
`ISD::matchUnaryPredicate` implemented in D94501.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D106575
Kadir Cetinkaya [Sat, 24 Jul 2021 17:44:15 +0000 (19:44 +0200)]
Revert "Revert D106562 "[clangd] Get rid of arg adjusters in CommandMangler""
This reverts commit
2aa0cf19e7fe17c9eb5eb2555e10184061b933f1.
Get rid of reference to the temporary.
Jon Chesterfield [Mon, 26 Jul 2021 08:54:50 +0000 (09:54 +0100)]
[libomptarget] Build amdgpu plugin without hsa
Default to building the amdgpu plugin to use dlopen when hsa is
not found instead of disabling it.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D106600
Jon Chesterfield [Mon, 26 Jul 2021 08:54:30 +0000 (09:54 +0100)]
[libomptarget][nfc] Squash unused variable warning
Suppress only current warning on openmp-clang-x86_64-linux-debian
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D106777
Vladislav Vinogradov [Wed, 7 Jul 2021 15:07:51 +0000 (18:07 +0300)]
[mlir] Fix RankedTensorType::walkImmediateSubElements method
Add 'enconding' attribute visitor.
Without it ASM printer doesn't use attribute aliases for 'enconding'.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D105554
Guillaume Chatelet [Mon, 26 Jul 2021 08:43:45 +0000 (08:43 +0000)]
[libc] fix LibcUnitTestMain when building with shared libraries
Lang Hames [Mon, 26 Jul 2021 07:49:05 +0000 (17:49 +1000)]
[ORC][ORC-RT] Add initial Objective-C and Swift support to MachOPlatform.
This allows ORC to execute code containing Objective-C and Swift classes and
methods (provided that the language runtime is loaded into the executor).
Marcel Koester [Fri, 23 Jul 2021 09:59:21 +0000 (11:59 +0200)]
[mlir] Added new RegionBranchTerminatorOpInterface and adapted uses of hasTrait<ReturnLike>.
This CL adds a new RegionBranchTerminatorOpInterface to query information about operands that can be
passed to successor regions. Similar to the BranchOpInterface, it allows to freely define the
involved operands. However, in contrast to the BranchOpInterface, it expects an additional region
number to distinguish between various use cases which might require different operands passed to
different regions.
Moreover, we added new utility functions (namely getMutableRegionBranchSuccessorOperands and
getRegionBranchSuccessorOperands) to query (mutable) operand ranges for operations equiped with the
ReturnLike trait and/or implementing the newly added interface. This simplifies reasoning about
terminators in the scope of the nested regions.
We also adjusted the SCF.ConditionOp to benefit from the newly added capabilities.
Differential Revision: https://reviews.llvm.org/D105018
Michael Kruse [Mon, 26 Jul 2021 02:39:08 +0000 (21:39 -0500)]
[Preprocessor] Implement -fminimize-whitespace.
This patch adds the -fminimize-whitespace with the following effects:
* If combined with -E, remove as much non-line-breaking whitespace as
possible.
* If combined with -E -P, removes as much whitespace as possible,
including line-breaks.
The motivation is to reduce the amount of insignificant changes in the
preprocessed output with source files where only whitespace has been
changed (add/remove comments, clang-format, etc.) which is in particular
useful with ccache.
A patch for ccache for using this flag has been proposed to ccache as well:
https://github.com/ccache/ccache/pull/815, which will use
-fnormalize-whitespace when clang-13 has been detected, and additionally
uses -P in "unify_mode". ccache already had a unify_mode in an older
version which was removed because of problems that using the
preprocessor itself does not have (such that the custom tokenizer did
not recognize C++11 raw strings).
This patch slightly reorganizes which part is responsible for adding
newlines that are required for semantics. It is now either
startNewLineIfNeeded() or MoveToLine() but never both; this avoids the
ShouldUpdateCurrentLine workaround and avoids redundant lines being
inserted in some cases. It also fixes a mandatory newline not inserted
after a _Pragma("...") that is expanded into a #pragma.
Reviewed By: aaron.ballman
Differential Revision: https://reviews.llvm.org/D104601
Yuanfang Chen [Mon, 26 Jul 2021 04:12:28 +0000 (21:12 -0700)]
[Object] make SourceMgr available to MCContext during inline asm symbols
collection
Fixes PR51210.
Esme-Yi [Mon, 26 Jul 2021 03:47:02 +0000 (03:47 +0000)]
[Debug-Info][llvm-dwarfdump] Don't use DW_FORM_data4/8
to encode the constants for DW_AT_data_member_location.
Summary: In DWARF v3, DW_FORM_data4/8 in
DW_AT_data_member_location are interpreted as location
list pointers. Interpreting constants as pointers is
not expected, so we use DW_FORM_udata to encode the
constants.
Reviewed By: probinson
Differential Revision: https://reviews.llvm.org/D105687
Mehdi Amini [Mon, 26 Jul 2021 03:08:26 +0000 (03:08 +0000)]
Revert "Build libSupport with -Werror=global-constructors (NFC)"
This reverts commit
579cc9ad2e2db6c3f1670b9f42c2cfe67bc5722c.
This breaks on Windows.
Mehdi Amini [Fri, 16 Jul 2021 03:32:59 +0000 (03:32 +0000)]
Build libSupport with -Werror=global-constructors (NFC)
Ensure that libSupport does not carry any static global initializer.
libSupport can be embedded in use cases where we don't want to load all
cl::opt unless we want to parse the command line.
ManagedStatic can be used to enable lazy-initialization of globals.
Esme-Yi [Mon, 26 Jul 2021 02:37:49 +0000 (02:37 +0000)]
[yaml2obj] Do not write the string table if there is no string entry.
Summary: yaml2obj shouldn't create the string table that isn't needed
- doing so wastes time and disk space.
Reviewed By: jhenderson
Differential Revision: https://reviews.llvm.org/D106420
Dave Airlie [Mon, 26 Jul 2021 01:04:25 +0000 (11:04 +1000)]
[OPENCL] opencl-c.h: add initial CL 3.0 conditionals for atomic operations.
This adds the optional wrappers around things, however this isn't sufficient yet for CL 3.0 without generic address space, I've got one more additional patch to add all those APIs, but this is an easier to review precursor.
Reviewed By: Anastasia
Differential Revision: https://reviews.llvm.org/D106111
Mehdi Amini [Mon, 26 Jul 2021 00:55:36 +0000 (00:55 +0000)]
Revert "Build libSupport with -Werror=global-constructors (NFC)"
This reverts commit
5eb2e9aa64b7be7cd8ed7f36de19c2c9bdf1977c.
This broke MacOS builds, needs to have a safer check guarding the flag
addition.
Mehdi Amini [Fri, 16 Jul 2021 03:32:59 +0000 (03:32 +0000)]
Build libSupport with -Werror=global-constructors (NFC)
Ensure that libSupport does not carry any static global initializer.
libSupport can be embedded in use cases where we don't want to load all
cl::opt unless we want to parse the command line.
ManagedStatic can be used to enable lazy-initialization of globals.
Mehdi Amini [Mon, 26 Jul 2021 00:20:24 +0000 (00:20 +0000)]
Remove the NotUnderValgrind caching flag
The motivation for this caching wasn't clear, remove it in an effort to
simplify the code and make libSupport free of global dynamic constructor.
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D106206
Roman Lebedev [Sun, 25 Jul 2021 23:57:19 +0000 (02:57 +0300)]
[SimplifyCFG] Fold branch to common dest: if branch is unpredictable, prefer to speculate
This is consistent with the two other usages of prof md in this pass.
Roman Lebedev [Sun, 25 Jul 2021 23:38:40 +0000 (02:38 +0300)]
[SimplifyCFG] Don't speculatively execute BB[s] if they are predictably not taken
Same as D106650, but for `FoldTwoEntryPHINode()`
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D106717
Roman Lebedev [Sun, 25 Jul 2021 23:30:31 +0000 (02:30 +0300)]
[SimplifyCFG] Don't speculatively execute BB if it's predictably not taken
If the branch isn't `unpredictable`, and it is predicted to *not* branch
to the block we are considering speculatively executing,
then it seems counter-productive to execute the code that is predicted not to be executed.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D106650
Roman Lebedev [Sun, 25 Jul 2021 23:30:12 +0000 (02:30 +0300)]
[NFC][SimplifyCFG] Add more negative tests for profmd-induced speculation avoidance
Fangrui Song [Sun, 25 Jul 2021 23:26:37 +0000 (16:26 -0700)]
[ELF] Support quoted symbols in symbol assignments
glibc/elf/tst-absolute-zero-lib.lds uses `"absolute" = 0;`
Nico Weber [Sun, 25 Jul 2021 22:37:49 +0000 (18:37 -0400)]
[lld/mac] Make comment style uniform in start-end.s test
Nico Weber [Fri, 23 Jul 2021 14:12:55 +0000 (10:12 -0400)]
[lld/mac] Add support for segment$start$ and segment$end$ symbols
These symbols are somewhat interesting in that they create non-existing
segments, which as far as I know is the only way to create segments
that don't contain any sections.
Final part of part of PR50760. Like D106629, but for segments instead
of sections. I'm not aware of anything that needs this in practice.
Differential Revision: https://reviews.llvm.org/D106767
Nico Weber [Sun, 25 Jul 2021 14:09:37 +0000 (10:09 -0400)]
[lld/mac] Move output segment rename logic into OutputSegment
Fixes the output segment name if both -rename_section and
-rename_segment are used and the post-section-rename segment
name is the same as the pre-segment-rename segment name to
match ld64's behavior.
The motivation is that segment$start$ can create section-less segments,
and this makes a corner case in the interaction between segment$start and
-rename_segment in the upcoming segment$start patch.
Differential Revision: https://reviews.llvm.org/D106766
Nico Weber [Sun, 25 Jul 2021 14:05:58 +0000 (10:05 -0400)]
[lld/mac] Reland: Add tests for the interaction between -rename_section and -rename_segment
No behavior change.
Differential Revision: https://reviews.llvm.org/D106765
Jon Chesterfield [Sun, 25 Jul 2021 22:14:42 +0000 (23:14 +0100)]
[libomptarget][amdgpu] More robust handling of failure to init HSA
If hsa_init fails, subsequent calls into hsa are not safe. Except for
hsa_init, but we don't retry on failure.
This patch:
- deletes a print that called into hsa to ask why it can't call into hsa
- drops a merge conflict block next to that print
- reliably initializes number of devices to zero
- skips the plugin destructor contents if the constructor failed to init hsa
Tested by making hsa_init return error, and by forcing the dynamic library
use which was then deleted from disk. Before this patch, both segv. After it,
friendly message about offloading being unavailable.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D106774
Nico Weber [Sun, 25 Jul 2021 22:11:36 +0000 (18:11 -0400)]
Revert "[lld/mac] Add tests for the interaction between -rename_section and -rename_segment"
This reverts commit
a6eb34624dcfa5a33caa0211f4a16710b22079c2.
The test fails, I screwed something up.