Amara Emerson [Mon, 4 Oct 2021 05:26:26 +0000 (22:26 -0700)]
[GlobalISel] Combine G_UMULH x, (1 << c)) -> x >> (bitwidth - c)
In order to not generate an unnecessary G_CTLZ, I extended the constant folder
in the CSEMIRBuilder to handle G_CTLZ. I also added some extra handing of
vector constants too. It seems we don't have any support for doing constant
folding of vector constants, so the tests show some other useless G_SUB
instructions too.
Differential Revision: https://reviews.llvm.org/D111036
Tobias Gysi [Fri, 8 Oct 2021 06:39:22 +0000 (06:39 +0000)]
[mlir][linalg] Add loop interchange to CodegenStrategy.
Add a loop interchange pass and integrate it with CodegenStrategy.
This patch depends on https://reviews.llvm.org/D110728 and https://reviews.llvm.org/D110746.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D110748
Jay Foad [Fri, 8 Oct 2021 06:32:44 +0000 (07:32 +0100)]
[TargetPassConfig] Remove an obsolete FIXME comment
The "coloring with register" functionality it refers to was removed ten
years ago in svn r144481 "Remove the -color-ss-with-regs option".
Tobias Gysi [Fri, 8 Oct 2021 06:06:12 +0000 (06:06 +0000)]
[mlir][linalg] Add generalization to CodegenStrategy.
Add a generalization pass and integrate it with CodegenStrategy.
This patch depends on https://reviews.llvm.org/D110728.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D110746
Simon Moll [Fri, 8 Oct 2021 06:07:51 +0000 (08:07 +0200)]
[VE][Test] Make Scalar/va_arg test generic
Make match patterns more permissive to be invariant to register
allocation choices.
Reviewed By: kaz7
Differential Revision: https://reviews.llvm.org/D111312
Vassil Vassilev [Tue, 5 Oct 2021 14:31:30 +0000 (14:31 +0000)]
Reland "[clang-repl] Re-implement clang-interpreter as a test case."
Original commit message: "
Original commit message: "
Original commit message:"
The current infrastructure in lib/Interpreter has a tool, clang-repl, very
similar to clang-interpreter which also allows incremental compilation.
This patch moves clang-interpreter as a test case and drops it as conditionally
built example as we already have clang-repl in place.
Differential revision: https://reviews.llvm.org/D107049
"
This patch also ignores ppc due to missing weak symbol for __gxx_personality_v0
which may be a feature request for the jit infrastructure. Also, adds a missing
build system dependency to the orc jit.
"
Additionally, this patch defines a custom exception type and thus avoids the
requirement to include header <exception>, making it easier to deploy across
systems without standard location of the c++ headers.
"
This patch also works around PR49692 and finds a way to use llvm::consumeError
in rtti mode.
Differential revision: https://reviews.llvm.org/D107049
Jimmy Zhongduo Lin [Fri, 8 Oct 2021 05:58:42 +0000 (11:28 +0530)]
[NFC][ScalarEvolution] Fix typo in getRangeForAffineAR and getRangeViaFactoring
Matthias Springer [Fri, 8 Oct 2021 05:46:45 +0000 (14:46 +0900)]
[mlir][linalg][bufferize][NFC] API change of aliasesNonWritableBuffer
The function now takes a Value instead of an OpOperand.
Differential Revision: https://reviews.llvm.org/D111378
Matthias Springer [Fri, 8 Oct 2021 05:36:43 +0000 (14:36 +0900)]
[mlir][linalg][bufferize] Fix/add missing case to getAliasingOpOperand
Differential Revision: https://reviews.llvm.org/D111377
Matthias Springer [Fri, 8 Oct 2021 03:14:05 +0000 (12:14 +0900)]
[mlir][linalg][bufferize] Add bufferRelation to op interface
Currently supported are: BufferRelation::None, BufferRelation::Equivalent.
Differential Revision: https://reviews.llvm.org/D111376
Mehdi Amini [Fri, 8 Oct 2021 05:18:44 +0000 (05:18 +0000)]
evert "[mlir] Limit Python dependency to Development.Module when possible."
This reverts commit
7aebdfc4fcc4a2667f5d5c30ffbdc146f9b014d2.
The build is broken with errors like:
GPUPasses.cpp:(.text.pybind11_object_init[pybind11_object_init]+0x118): undefined reference to `PyExc_TypeError'
Mike Urbach [Fri, 8 Oct 2021 04:08:15 +0000 (22:08 -0600)]
[mlir] Limit Python dependency to Development.Module when possible.
After CMake 3.18, we are able to limit the scope of the
find_package(Python3 ...) search to just Development.Module. Searching
for Development will fail in manylinux builds, and isn't necessary
since we are not embedding the Python interpreter. For more information, see:
https://pybind11.readthedocs.io/en/stable/compiling.html#findpython-mode
Reviewed By: stellaraccident
Differential Revision: https://reviews.llvm.org/D111383
Mehdi Amini [Fri, 8 Oct 2021 01:21:24 +0000 (01:21 +0000)]
Fix test-rsqrt.mlir to accept AMD's approximation of rsqrt as well
These kind of function can behave differently on these X86 chips, there
isn't really "one true answer" so we'll accept both.
Also remove spurious passes and use mattr="avx" to match the instruction
used here.
Differential Revision: https://reviews.llvm.org/D111373
Arthur Eubanks [Fri, 8 Oct 2021 03:53:43 +0000 (20:53 -0700)]
Revert "[sanitizer] Support Intel CET"
This reverts commit
fdf4c035225de52f596899931b1f6100e5e3e928.
Breaks macOS bots, e.g. https://crbug.com/1257863.
Still figuring out if this is actually supported on macOS. Other places
that include <cet.h> only do so on Linux.
Craig Topper [Thu, 7 Oct 2021 19:02:35 +0000 (12:02 -0700)]
[RISCV] Remove experimental-b extension that includes all Zb* extensions
At this point it looks like a B extension will never exist. Instead
Zba, Zbb, Zbc, and Zbs are individual extensions being ratified
together as a package. Unknown at this time when or if the other
Zb* extensions will be ratified.
This patch removes references to the B extension. I've updated and
split tests accordingly.
This has been split from D110669 to make review a little easier.
Differential Revision: https://reviews.llvm.org/D111338
Chen Zheng [Fri, 8 Oct 2021 03:44:12 +0000 (03:44 +0000)]
use right separator for windows.
Max Kazantsev [Fri, 8 Oct 2021 02:09:22 +0000 (09:09 +0700)]
[LoopPeel] Peel loops with deoptimizing exits
Added support for peeling loops with "deoptimizing" exits -
such exits that it or any of its children (or any of their
children, etc) either has a @llvm.experimental.deoptimize call
prior to the terminating return instruction of this basic block
or is terminated with unreachable. All blocks in the the
sequence must have a single successor, maybe except for the last
one.
Previously we only checked the exit block for being deoptimizing.
Now we check if the last reachable block from the exit is deoptimizing.
Patch by Dmitry Makogon!
Differential Revision: https://reviews.llvm.org/D110922
Reviewed By: mkazantsev
Chen Zheng [Fri, 8 Oct 2021 02:33:24 +0000 (02:33 +0000)]
[XCOFF] support DWARF for 32-bit XCOFF for object output
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D97184
Joseph Huber [Thu, 7 Oct 2021 19:43:56 +0000 (15:43 -0400)]
[OpenMP] Introduce new flags to assert thread and team usage in the runtime
This patch adds two flags to be supported for the new runtime. The flags
are `-fopenmp-assume-threads-oversubscription` and
-fopenmp-assume-teams-oversubscription`. These add global values that
can be checked by the work sharing runtime functions to make better
judgements about how to distribute work between the threads.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D111348
Shilei Tian [Fri, 8 Oct 2021 02:15:23 +0000 (22:15 -0400)]
[OpenMP][DeviceRTL] Add the support for printf in a freestanding way
For NVPTX, `printf` can be used just with a function declaration. For AMDGCN, an
function definition is added, but it simply returns.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D109728
Itay Bookstein [Fri, 8 Oct 2021 01:55:55 +0000 (09:55 +0800)]
[SelectionDAG] Fix shift libcall ABI mismatch in shift-amount argument
The shift libcalls have a shift amount parameter of MVT::i32, but
sometimes ExpandIntRes_Shift may be called with a node whose
second operand is a type that is larger than that. This leads to
an ABI mismatch, and for example causes a spurious zeroing of
a register in RV32 for 64-bit shifts. Note that at present regular
shift intstructions already have their shift amount operand adapted
at SelectionDAGBuilder::visitShift time, and funnelled shifts bypass that.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D110508
Wang, Pengfei [Fri, 8 Oct 2021 01:05:55 +0000 (09:05 +0800)]
[X86] Optimize fdiv with reciprocal instructions for half type
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D110557
Ben Shi [Tue, 5 Oct 2021 08:37:44 +0000 (08:37 +0000)]
[RISCV][test] Add more tests of (add (mul r, c0), c1)
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D111140
Johannes Doerfert [Fri, 8 Oct 2021 00:57:22 +0000 (20:57 -0400)]
[OpenMP][FIX] Data race in the SPMD execution of the new runtime
We need to synchronize the threads *before* we destroy the RAII objects
that hold the old values and not after to avoid threads executing the
parallel region but seeing an inconsistent state.
Reviewed By: tianshilei1992
Differential Revision: https://reviews.llvm.org/D111369
Vitaly Buka [Wed, 6 Oct 2021 20:10:35 +0000 (13:10 -0700)]
[msan] Print both shadow and user address
before:
00 00 00 00 ff ff ff ff 00 00 00 00 00 00 00 00
Shadow map of [0x211000000005, 0x21100000012e), 297 bytes:
now:
0x2f60d213ac10[0x7f60d213ac10] 00 00 00 00 ff ff ff ff 00 00 00 00 00 00 00 00
Shadow map [0x211000000005, 0x21100000012e) of [0x711000000005, 0x711000000135), 297 bytes:
Differential Revision: https://reviews.llvm.org/D111261
MaheshRavishankar [Thu, 7 Oct 2021 23:21:32 +0000 (16:21 -0700)]
[mlir][Tensor] Add ReifyRankedShapedTypeOpInterface to tensor.extract_slice.
Differential Revision: https://reviews.llvm.org/D111263
Volodymyr Sapsai [Wed, 22 Sep 2021 18:20:52 +0000 (11:20 -0700)]
[modules] Fix IRGen assertion on accessing ObjC ivar inside a method.
When have ObjCInterfaceDecl with the same name in 2 different modules,
hitting the assertion
> Assertion failed: (Index < RL->getFieldCount() && "Ivar is not inside record layout!"),
> function lookupFieldBitOffset, file llvm-project/clang/lib/AST/RecordLayoutBuilder.cpp, line 3434.
on accessing an ivar inside a method. The assertion happens because
ivar belongs to one module while its containing interface belongs to
another module and then we fail to find the ivar inside the containing
interface. We already keep a single ObjCInterfaceDecl definition in
redecleration chain and in this case containing interface was correct.
The issue is with ObjCIvarDecl. IVar decl for IRGen is taken from
ObjCIvarRefExpr that is created in `Sema::BuildIvarRefExpr` using ivar
decl returned from `Sema::LookupIvarInObjCMethod`. And ivar lookup
returns a wrong decl because basically we take the first ObjCIvarDecl
found in `ASTReader::FindExternalVisibleDeclsByName` (called by
`DeclContext::lookup`). And in `ASTReader.Lookups` lookup table for a
wrong module comes first because `ASTReader::finishPendingActions`
processes `PendingUpdateRecords` in reverse order and the first
encountered ObjCIvarDecl will end up the last in `ASTReader.Lookups`.
Fix by merging ObjCIvarDecl from different modules correctly and by
using a canonical one in IRGen.
rdar://
82854574
Differential Revision: https://reviews.llvm.org/D110280
Petr Hosek [Thu, 7 Oct 2021 23:25:15 +0000 (16:25 -0700)]
[CMake] Include llvm-libtool-darwin in Fuchsia toolchain
We want to use this tool in our build.
Differential Revision: https://reviews.llvm.org/D111366
Amy Zhuang [Thu, 7 Oct 2021 22:45:59 +0000 (15:45 -0700)]
[mlir] Fix a bug in Affine LICM.
Currently Affine LICM checks iterOperands and does not hoist out any
instruction containing iterOperands. We should check iterArgs instead.
Reviewed By: bondhugula
Differential Revision: https://reviews.llvm.org/D111090
Jonas Devlieghere [Thu, 7 Oct 2021 22:12:45 +0000 (15:12 -0700)]
[lldb] Parse and display reporting errors from JSON crashlogs
JSON crashlogs have an optional field named reportNotes that contains
any potential errors encountered by the crash reporter when generating
the crashlog. Parse and display them in LLDB.
Differential revision: https://reviews.llvm.org/D111339
Jonas Devlieghere [Thu, 7 Oct 2021 18:35:56 +0000 (11:35 -0700)]
[lldb] Support missing threadState in JSON crashlogs
Gracefully deal with JSON crashlogs that don't have thread state
available and print an error saying as much: "No thread state (register
information) available".
rdar://
83955858
Differential revision: https://reviews.llvm.org/D111341
Roman Lebedev [Thu, 7 Oct 2021 22:24:32 +0000 (01:24 +0300)]
Workaround broken FileCheck default yet another time
Keith Smiley [Thu, 7 Oct 2021 22:07:33 +0000 (15:07 -0700)]
[lld][test] Remove /usr/local/lib test requirement
This field only exists if the directory exists on the machine running
the test. It likely exists for most Intel macOS users because of
homebrew, but doesn't exist on some of the CI machines. This
unfortunately makes this test a bit less strict.
Differential Revision: https://reviews.llvm.org/D111361
Roman Lebedev [Thu, 7 Oct 2021 22:15:29 +0000 (01:15 +0300)]
[NFC][VectorCombine] Add baseline test coverage for GEP scalarization
Qiongsi Wu [Thu, 7 Oct 2021 22:07:30 +0000 (18:07 -0400)]
[NFC] Including <string> in llvm-cxxdump/Error.cpp
A [[ https://reviews.llvm.org/rGf6fa95b77f33c3690e4201e505cb8dce1433abd9 | recent commit ]] removed `<string>` from `ErrorHandling.h`. The removal caused `<string>` to be no longer included for `llvm/tools/llvm-cxxdump/Error.cpp` which uses the string type.
This patch adds `<string>` to `llvm/tools/llvm-cxxdump/Error.cpp`.
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D111354
Arthur Eubanks [Wed, 6 Oct 2021 23:51:38 +0000 (16:51 -0700)]
Don't print uselistorder in --print-changed
Using uselistorders is fairly niche, it shouldn't be on by default and mostly just clutters the output.
Reviewed By: jamieschmeiser
Differential Revision: https://reviews.llvm.org/D111282
Louis Dionne [Mon, 4 Oct 2021 19:01:27 +0000 (15:01 -0400)]
[libc++] Remove the CI job for Apple/System/Noexceptions
When we recently started using DYLD_LIBRARY_PATH to run the test suite
on the Apple/System configuration of the library, the -fno-exceptions
variant started failing.
It started failing because under that configuration, libc++abi.dylib
doesn't provide support for exceptions. For example, it doesn't provide
some symbols such as ___gxx_personality_v0. Now, the problem is that
when the test suite is run with DYLD_LIBRARY_PATH, /usr/lib/libobjc.dylib
uses the just-built libc++abi.dylib, which doesn't support exceptions,
and we end up with an unresolved reference to ___gxx_personality_v0.
Previously, using -Wl,-rpath,path/to/lib, we would be loading both
/usr/lib/libc++abi.dylib and <just-built>/lib/libc++abi.dylib.
/usr/lib/libobjc.dylib would use the system libc++abi.dylib, which
contains support for exceptions, and the tests would be using the
just-built one, which doesn't.
Disentangling that led me to believe that we shouldn't try to test this
configuration where libc++/libc++abi are built as system libraries, but
where they don't support exceptions, since that just doesn't make any
sense. Doing so is like trying to build libc++/libc++abi and test it as
a system library after performing an ABI break -- of course nothing is
going to work.
For that reason, I am removing this configuration. Note that we could
still test the library on macOS without exceptions if we wanted, only
we wouldn't be building it as a system library. This patch doesn't add
that because we already have a -fno-exceptions CI job on Linux.
Differential Revision: https://reviews.llvm.org/D111349
Louis Dionne [Thu, 7 Oct 2021 18:54:14 +0000 (14:54 -0400)]
[libc++] Add a from-scratch testing config for GCC
Differential Revision: https://reviews.llvm.org/D111329
Leonard Chan [Thu, 7 Oct 2021 21:19:29 +0000 (14:19 -0700)]
Reland "[clang][Fuchsia] Re-enable compiler-rt tests in runtimes build"
This reverts commit
a625fd26cea579853bfe6c00f8fd8e6e88388630.
Round 3: The scudo test was addressed in
6727832c324c1fb43946275d24e2931fde94bc0d.
wlei [Thu, 16 Sep 2021 07:31:57 +0000 (00:31 -0700)]
[llvm-profgen] Ignore branch count against outline function
For some transformations like hot-cold split or coro split, it can outline its part of function ranges. Since sample loader is the early stage of backend and no split happens at that time, compiler can't recognize those function, so in llvm-profgen we should attribute the sample to the original function. This is already done for the body range samples since we use the symbols from dwarf which is created before the split.
But for branch samples, the call from master function to its outlined function is actually not a call to the original function, we shouldn't add head/callsie samples for it. So instead of dwarf symbol, we use the symbols from symbol table and ignore those functions with special suffixes(like `.cold` ,`.resume`) for accumulating the callsite/head samples.
Reviewed By: hoy, wenlei
Differential Revision: https://reviews.llvm.org/D110864
Kostya Kortchinsky [Thu, 7 Oct 2021 19:29:39 +0000 (12:29 -0700)]
[scudo] Reduce the scope of AllocAfterFork
`ScudoWrappersCppTest.AllocAfterFork` was failing obscurely sometimes.
Someone pointed us to Linux's `vm.max_map_count` that can be
significantly lower on some machines than others. It turned out that
on a machine with that setting set to 65530, some `ENOMEM` errors
would occur with `mmap` & `mprotect` during that specific test.
Reducing the number of times we fork, and the maximum size allocated
during that test makes it pass on those machines.
Differential Revision: https://reviews.llvm.org/D111342
Adrian Prantl [Thu, 7 Oct 2021 19:58:24 +0000 (12:58 -0700)]
Do not emit prologue_end for line 0 locs if there is a non-zero loc present
This change fixes a bug where the compiler generates a prologue_end
for line 0 locs. That is because line 0 is not associated with any
source location, so there should not be a prolgoue_end at a location
that doesn't correspond to a source location.
There were some LLVM tests that were explicitly checking for line 0
prologue_end's as well since I believe that to be incorrect, I had to
change those tests as well.
Patch by Shubham Rastogi!
Differential Revision: https://reviews.llvm.org/D110740
Adrian Prantl [Wed, 6 Oct 2021 22:04:36 +0000 (15:04 -0700)]
Recognize the Swift compiler in DW_AT_producer
This patch adds support for Swift compiler producer strings to DWARFUnit.
Differential Revision: https://reviews.llvm.org/D111278
Vitaly Buka [Thu, 7 Oct 2021 17:43:49 +0000 (10:43 -0700)]
[NFC][sanitizer] Annotate a few branches in StackDepot
Vitaly Buka [Wed, 6 Oct 2021 21:19:27 +0000 (14:19 -0700)]
[sanitizer] Remove traces from the header
This will simplify removing id proposed by @dvyukov on D111183
Also now we have more flexiliby for traces compressio they
are not interleaving with uncompressable headers.
Depends on D111256.
Differential Revision: https://reviews.llvm.org/D111274
Vitaly Buka [Wed, 6 Oct 2021 19:37:21 +0000 (12:37 -0700)]
[NFC][sanitizer] Remove global PersistentAllocator
This way is easier to track memory usage and do other
incremental refactorings.
Differential Revision: https://reviews.llvm.org/D111256
Vitaly Buka [Thu, 7 Oct 2021 17:13:00 +0000 (10:13 -0700)]
[sanitizer] Uninline slow path of PersistentAllocator::alloc
peter klausler [Wed, 6 Oct 2021 23:29:00 +0000 (16:29 -0700)]
[flang] Error checking for IBCLR/IBSET and ISHFT/SHIFT[ALR]
Bit positions for the intrinsics IBCLR and IBSET and shift counts
for the intrinsics ISHFT/SHIFTA/SHIFTL/SHIFTR should be validated
when folding.
Differential Revision: https://reviews.llvm.org/D111327
Philip Reames [Wed, 6 Oct 2021 23:44:24 +0000 (16:44 -0700)]
[scev] Put comments on the right fields [nfc]
Jay Foad [Thu, 7 Oct 2021 14:11:31 +0000 (15:11 +0100)]
[AMDGPU] Preserve MachineDominatorTree in SILowerControlFlow
Updating the MachineDominatorTree is easy since SILowerControlFlow only
splits and removes basic blocks. This should save a bit of compile time
because previously we would recompute the dominator tree from scratch
after this pass.
Another reason for doing this is that SILowerControlFlow preserves
LiveIntervals which transitively requires MachineDominatorTree. I think
that means that SILowerControlFlow is obliged to preserve
MachineDominatorTree too as explained here:
https://lists.llvm.org/pipermail/llvm-dev/2020-November/146923.html
although it does not seem to have caused any problems in practice yet.
Differential Revision: https://reviews.llvm.org/D111313
Jay Foad [Thu, 7 Oct 2021 20:24:50 +0000 (21:24 +0100)]
[TargetPassConfig] Enable machine verification after miscellaneous passes
In a couple of places machine verification was disabled for no apparent
reason, probably just because an "addPass(..., false)" line was cut and
pasted from elsewhere.
After this patch the only remaining place where machine verification is
disabled in the generic TargetPassConfig code, is after addPreEmitPass.
Nikita Popov [Thu, 7 Oct 2021 20:06:53 +0000 (22:06 +0200)]
[BasicAA] Use base of decomposed GEP in recursive queries (NFC)
DecompGEP.Base and UnderlyingV are currently always the same.
However, logically DecompGEP.Base is the right value to use here,
because the decomposed offset is relative to that base.
Mark Schimmel [Thu, 7 Oct 2021 19:59:39 +0000 (12:59 -0700)]
[ARC] ARCRegisterInfo cleanup prior to adding core register pairs (ARC32) and 64-bit core registers (ARC64)
Differential Revision: https://reviews.llvm.org/D11108
Keith Smiley [Wed, 6 Oct 2021 20:29:08 +0000 (13:29 -0700)]
[clang] Fix darwin REQUIRES test annotation (NFC)
Some subprojects like compiler-rt define the `darwin` feature in their
lit config, but clang does not do that, so we need to use the global
`system-darwin` here instead.
Differential Revision: https://reviews.llvm.org/D111267
Nikita Popov [Thu, 7 Oct 2021 18:38:37 +0000 (20:38 +0200)]
[LoopFlatten] Mark loop analyses as preserved
LoopFlatten does preserve loop analyses (DT, LI and SCEV), but
currently doesn't mark them as preserved in the NewPM (they are
marked as preserved in the LegacyPM). I think this doesn't really
have an effect in the end because the loop pass adaptor will just
assume they're preserved anyway, but let's be explicit about this
for the sake of clarity.
Differential Revision: https://reviews.llvm.org/D111328
Geoffrey Martin-Noble [Thu, 7 Oct 2021 19:45:53 +0000 (12:45 -0700)]
[Bazel] Update config for
3b01cf9286
Updates the Bazel config for changes from
https://github.com/llvm/llvm-project/commit/
3b01cf9286
by adding configuration for the new OpenMPOpsInterfaces tablegn target.
Differential Revision: https://reviews.llvm.org/D111347
Louis Dionne [Tue, 28 Sep 2021 19:54:41 +0000 (15:54 -0400)]
[runtimes] Add tests for vendor-specific properties
Vendors take libc++ and ship it in various ways. Some vendors might
ship it differently from what upstream LLVM does, i.e. the install
location might be different, some ABI properties might differ, etc.
In the past few years, I've come across several instances where
having a place to test some of these properties would have been
incredibly useful. I also just got bitten by the lack of tests
of that kind, so I'm adding some now.
The tests added by this commit for Apple platforms have numerous
TODOs that capture discrepancies between the upstream LLVM CMake
and the slightly-modified build we perform internally to produce
Apple's system libc++. In the future, the goal would be to upstream
all those differences so that it's possible to build a faithful
Apple system libc++ with the upstream LLVM sources only.
But this isn't only useful for Apple - this lays out the path for
any vendor being able to add their own checks (either upstream or
downstream) to libc++.
This is a re-application of
9892d1644f, which was reverted in
138dc27186be
because it broke the build. The issue was that we didn't apply the required
changes to libunwind and our CI didn't notice it because we were not
running the libunwind tests. This has been fixed now, and we're running
the libunwind tests in CI now too.
Differential Revision: https://reviews.llvm.org/D110736
Keith Smiley [Wed, 6 Oct 2021 20:41:14 +0000 (13:41 -0700)]
[lld][test] Fix darwin REQUIRES (NFC)
Some subprojects like compiler-rt define the `darwin` feature in their
lit config, but lld does not do that, so we need to use the global
system-darwin here instead. This test seems to have drifted from the
actual behavior so I also had to add `/usr/local/lib` here to make it
pass.
Differential Revision: https://reviews.llvm.org/D111268
Jamie Schmeiser [Thu, 7 Oct 2021 19:23:48 +0000 (15:23 -0400)]
Revert "Reland A new option -print-on-crash that prints the IR as it was upon entering the last pass when there is a crash."
This reverts commit
13d1592716a65444314f501109ec9ca344ef1f87.
Jon Chesterfield [Thu, 7 Oct 2021 19:17:02 +0000 (20:17 +0100)]
[libomptarget] Reapply
2bc4d48a78b which was accidentally reverted
Sanjay Patel [Thu, 7 Oct 2021 18:56:29 +0000 (14:56 -0400)]
[InstCombine] ease use check for fold of bitcasted extractelt to trunc
This helps with examples like:
https://llvm.org/PR52057
...but we need at least one more fold to fix that case.
Jay Foad [Fri, 1 Oct 2021 16:31:21 +0000 (17:31 +0100)]
[TwoAddressInstruction] Enable machine verification after this pass
Differential Revision: https://reviews.llvm.org/D111007
Jay Foad [Sun, 3 Oct 2021 08:10:29 +0000 (09:10 +0100)]
[PHIElimination] Enable machine verification after this pass
Differential Revision: https://reviews.llvm.org/D111006
Jay Foad [Thu, 30 Sep 2021 14:35:43 +0000 (15:35 +0100)]
[PHIElimination] Account for INLINEASM_BR when inserting kills
When PHIElimination adds kills after lowering PHIs to COPYs it knows
that some instructions after the inserted COPY might use the same
SrcReg, but it was only looking at the terminator instructions at the
end of the block, not at other instructions like INLINEASM_BR that can
appear after the COPY insertion point.
Since we have already called findPHICopyInsertPoint, which knows about
INLINEASM_BR, we might as well reuse the insertion point that it
calculated when looking for instructions that might use SrcReg.
This fixes a machine verification failure if you force machine
verification to run after PHIElimination (currently it is disabled for
other reasons) when running
test/CodeGen/X86/callbr-asm-phi-placement.ll.
Differential Revision: https://reviews.llvm.org/D110834
Jay Foad [Fri, 1 Oct 2021 18:18:45 +0000 (19:18 +0100)]
[PHIElimination] Pre-commit a test case for D110834
Jamie Schmeiser [Thu, 7 Oct 2021 19:02:19 +0000 (15:02 -0400)]
Reland A new option -print-on-crash that prints the IR as it was upon entering the last pass when there is a crash.
Summary:
The IR is saved in its print form before each pass is started and a
signal handler is registered. If the compilation crashes, the signal
handler will print the saved IR to dbgs(). This option
can be modified using -print-module-scope to get the IR for the complete
module. Filtering options can be used to improve performance by limiting
which passes (or functions) save the IR. Note that this option only works
with the new pass manager.
Author: Jamie Schmeiser <schmeise@ca.ibm.com>
Reviewed By: aeubanks (Arthur Eubanks) yrouban (Yevgeny Rouban)
Differential Revision: https://reviews.llvm.org/D86657
Kiran Chandramohan [Thu, 7 Oct 2021 18:52:15 +0000 (20:52 +0200)]
[mlir][openmp] Add an interface for Outlineable OpenMP ops
Add an interface for outlineable OpenMP operations.
This patch was initially done in fir-dev and is now needed
for the upstreaming.
Reviewed By: schweitz
Differential Revision: https://reviews.llvm.org/D111310
Stella Laurenzo [Thu, 7 Oct 2021 18:47:05 +0000 (11:47 -0700)]
[mlir][python] Temporarily disable test for converting unsupported DenseElementsAttr types to a buffer.
* Need to investigate the proper solution to https://github.com/pybind/pybind11/issues/3336 or engineer something different.
* The attempt to produce an empty buffer_info as a workaround triggers asan/ubsan.
* Usage of this API does not arise naturally in practice yet, and it is more important to be asan/crash clean than have a solution right now.
* Switching back to raising an exception, even though that triggers terminate().
Jay Foad [Thu, 30 Sep 2021 13:18:52 +0000 (14:18 +0100)]
[X86] Special-case ADD of two identical registers in convertToThreeAddress
X86InstrInfo::convertToThreeAddress would convert this:
%1:gr32 = ADD32rr killed %0:gr32(tied-def 0), %0:gr32, implicit-def dead $eflags
to this:
undef %2.sub_32bit:gr64 = COPY killed %0:gr32
undef %3.sub_32bit:gr64_nosp = COPY %0:gr32
%1:gr32 = LEA64_32r killed %2:gr64, 1, killed %3:gr64_nosp, 0, $noreg
Note that in the ADD32rr, %0 was used twice and the first use had a kill
flag, which is what MachineInstr::addRegisterKilled does.
In the converted code, each use of %0 is copied to a new reg, and the
first COPY inherits the kill flag from the ADD32rr. This causes
machine verification to fail (if you force it to run after
TwoAddressInstructionPass) because the second COPY uses %0 after it is
killed. Note that machine verification is currently disabled after
TwoAddressInstructionPass but this is a step towards being able to
enable it.
Fix this by not inserting more than one COPY from the same source
register.
Differential Revision: https://reviews.llvm.org/D110829
Jay Foad [Tue, 5 Oct 2021 13:51:12 +0000 (14:51 +0100)]
[X86] Pre-commit a test case for D110829
Artem Belevich [Wed, 29 Sep 2021 22:02:36 +0000 (15:02 -0700)]
[CUDA] Make sure <string.h> is included with original __THROW defined.
Otherwise we may end up with an inconsistent redeclarations of the standard
library functions if _FORTIFY_SOURCE is in effect.
https://bugs.llvm.org/show_bug.cgi?id=47869
Differential Revision: https://reviews.llvm.org/D110781
Amara Emerson [Wed, 29 Sep 2021 06:41:11 +0000 (23:41 -0700)]
[GlobalISel] Port the udiv -> mul by constant combine.
This is a straight port from the equivalent DAG combine.
Differential Revision: https://reviews.llvm.org/D110890
Craig Topper [Thu, 7 Oct 2021 18:27:07 +0000 (11:27 -0700)]
[RISCV] Correct FileCheck prefixes in rv32zbc-intrinsic.ll and rv64zbc-intrinsic.ll. NFC
Zbc RUN lines should use ZBC instead of BC in their prefix.
Akira Hatanaka [Thu, 7 Oct 2021 18:25:01 +0000 (11:25 -0700)]
Refactor code in ObjCARC.cpp. NFC
This is in preparation for another patch I'm planning to send later.
Fangrui Song [Thu, 7 Oct 2021 18:14:40 +0000 (11:14 -0700)]
[LangRef] Update ifunc syntax
Extracted from Itay Bookstein's D108872.
Kevin P. Neal [Wed, 6 Oct 2021 18:35:38 +0000 (14:35 -0400)]
[NFC] Rename functions to match our naming scheme.
In the review of D111085 it was pointed out that these functions don't
conform to the naming scheme in use in LLVM. With this commit we should
be good for all of FPEnv.h.
Jay Foad [Thu, 7 Oct 2021 09:38:38 +0000 (10:38 +0100)]
[MIRParser] Add support for IsInlineAsmBrIndirectTarget
Print this basic block flag as inlineasm-br-indirect-target and parse
it. This allows you to write MIR test cases for INLINEASM_BR. The test
case I added is one that I wanted to precommit anyway for D110834.
Differential Revision: https://reviews.llvm.org/D111291
Bjorn Pettersson [Wed, 6 Oct 2021 10:58:52 +0000 (12:58 +0200)]
[LoopRotate] Forget SCEV values in RewriteUsesOfClonedInstructions
This patch fixes problems reported in PR51981.
When rotating a loop it isn't enough to just forget SCEV for that
loop nest. When rotating we might clone some instructions from the
old header into the preheader, and insert new PHI nodes to merge
values together. There could be users of the original value that are
updated to use the PHI result. And those users were not necessarily
depending on a PHI node earlier, so they weren't cleaned up when just
forgetting all SCEV:s for the loop nest. So we need to explicitly
forget those values to avoid invalid cached SCEV expressions.
Reviewed By: fhahn, mkazantsev
Differential Revision: https://reviews.llvm.org/D110813
Bjorn Pettersson [Thu, 30 Sep 2021 10:33:31 +0000 (12:33 +0200)]
[test] Pre-commit test case for PR51981. NFC
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D110812
Luke Drummond [Thu, 7 Oct 2021 14:44:38 +0000 (15:44 +0100)]
Workaround build error for mingw-g++
mingw-g++ does not correctly support the full `std::errc` namespace as
worded in the standard[1]. As such, we cannot reliably use all names
therein. This patch changes the use of
`std::errc::state_not_recoverable`, to use portable error codes from the
`llvm::errc` equivalent.
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71444
Reviewed by v.g.vassilev
Differential Revision: https://reviews.llvm.org/D111315
Kazu Hirata [Thu, 7 Oct 2021 17:25:05 +0000 (10:25 -0700)]
[lldb] Fix a "missing field" warning
This patch fixes:
llvm-project/lldb/source/Plugins/ABI/PowerPC/ABISysV_ppc.cpp:204:6:
error: missing field 'invalidate_regs' initializer
[-Werror,-Wmissing-field-initializers]
Craig Topper [Thu, 7 Oct 2021 00:14:08 +0000 (17:14 -0700)]
[RISCV] Handle vector of pointer in getTgtMemIntrinsic for strided load/store.
getScalarSizeInBits() doesn't work if the scalar type is a pointer.
For that we need to go through DataLayout.
Corentin Jabot [Thu, 7 Oct 2021 17:07:15 +0000 (13:07 -0400)]
Add information about partially implemented features
Desccribe in cxx_status.html the missing parts of the partially
implemented proposals described in cxx_status.html.
Uses <details> blocks so the information appears collapsed
by default.
Paul Robinson [Thu, 30 Sep 2021 21:12:29 +0000 (14:12 -0700)]
[PS4][TargetLibraryInfo] Set TLI info correctly for PS4
Amy Kwan [Thu, 7 Oct 2021 15:55:50 +0000 (10:55 -0500)]
[NFC] Update return type of vec_popcnt to vector unsigned.
This patch updates the vec_popcnt builtins to return vector unsigned,
as defined by the Power Vector Intrinsics Programming Reference.
This patch is NFC and all existing tests pass.
Differential Revision: https://reviews.llvm.org/D110934
Sanjay Patel [Thu, 7 Oct 2021 15:53:16 +0000 (11:53 -0400)]
[InstSimplify] (x || y) && (x || !y) --> x
https://alive2.llvm.org/ce/z/4BE33w
This is the logical (select-form) equivalent of the bitwise logic fold:
e36d351d19b1
This is another part of solving the regression from:
https://llvm.org/PR52077
gbreynoo [Thu, 7 Oct 2021 16:09:52 +0000 (17:09 +0100)]
[llvm-readelf][docs] Add missing options and details to the help output and the command guide
This change is to keep the help text and command guide of llvm-readelf
in tandem.
- In the help text mention that --section-data, --section-relocations,
--section-symbols and --stack-sizes have no effect on GNU style
output; give the accepted values for --elf-output-style and update
the description of --gnu-hash-table to use the command guide
description.
- In the command guide add the missing options -a,
--dependant-libraries,--no-demangle, --wide and -W. Also update the
description of --symbols so it matches the help text.
Differential Revision: https://reviews.llvm.org/D111240
Mark de Wever [Tue, 28 Sep 2021 17:15:18 +0000 (19:15 +0200)]
[libc++] Use addressof in assignment operator.
Replace `&__rhs` with `_VSTD::addressof(__rhs)` to guard against ADL hijacking
of `operator&` in `operator=`. Thanks to @CaseyCarter for bringing it to our
attention.
Similar issues with hijacking `operator&` still exist, they will be
addressed separately.
Reviewed By: #libc, Quuxplusone, ldionne
Differential Revision: https://reviews.llvm.org/D110852
David Spickett [Thu, 7 Oct 2021 16:08:08 +0000 (16:08 +0000)]
[lldb] Mark abort signal test unsupported on AArch64 Linux
This has started failing since we moved our bots to Focal.
For unknown reasons the abort_caller stack is missing when
we check from the handler breakpoint.
Mark unsupported while I investigate.
Mark Schimmel [Thu, 7 Oct 2021 16:02:19 +0000 (09:02 -0700)]
C] Add option to ARCOptAddrMode to disable the pass and diagnose errors
Fixed formatting issues reported by clang-format
Differential Revision: https://reviews.llvm.org/D111255
Stella Laurenzo [Thu, 7 Oct 2021 01:41:22 +0000 (18:41 -0700)]
[mlir] Extend C and Python API to support bulk loading of DenseElementsAttr.
* This already half existed in terms of reading the raw buffer backing a DenseElementsAttr.
* Documented the precise expectations of the buffer layout.
* Extended the Python API to support construction from bitcasted buffers, allowing construction of all primitive element types (even those that lack a compatible representation in Python).
* Specifically, the Python API can now load all integer types at all bit widths and all floating point types (f16, f32, f64, bf16).
Differential Revision: https://reviews.llvm.org/D111284
Chris Jackson [Thu, 7 Oct 2021 14:22:52 +0000 (14:22 +0000)]
[DebugInfo][LSR] Limit the size of SCEV translated to DIExpression
SCEV-based salvaging will use excessive resources if it encounters
very long SCEV expressions. This patch places a limit on the length of
SCEV expression that salvaging will attempt to translate.
Reviewed by: Orlando
Differential Revision: https://reviews.llvm.org/D110558
Mark de Wever [Thu, 7 Oct 2021 15:34:47 +0000 (17:34 +0200)]
[libcxx[ Run generate_private_header_tests.py
The script was recently updated to generate different output. This
breaks the CI due the patches which used the old version of the script.
Erik Desjardins [Thu, 7 Oct 2021 15:14:56 +0000 (08:14 -0700)]
[Inline] Introduce Constant::hasOneLiveUse, use it instead of hasOneUse in inline cost model (PR51667)
Otherwise, inlining costs may be pessimized by dead constants.
Fixes https://bugs.llvm.org/show_bug.cgi?id=51667.
Reviewed By: mtrofin, aeubanks
Differential Revision: https://reviews.llvm.org/D109294
gbreynoo [Thu, 7 Oct 2021 15:26:26 +0000 (16:26 +0100)]
[llvm-objdump][docs] Add details to the help output and command guide
This change is to add some missing details, clarifies some options and
brings the help text and command guide of objdump closer together.
- Added to the help that --all-headers also outputs symbols and
relocations to match the command guide.
- Added to the help that --debug-vars accepts an optional
ascii/unicode format to match the command guide.
- Changed the help descriptions for --disassemble,
--disassemble-all, --dwarf=<value>, --fault-map-section,
--line-numbers, --no-leading-addr and --source descriptions to
match the command guide.
- Added to the help that --start-address and --stop-address also
effect relocation entries and the symbol table output to match
the command guide.
- Added a note to the command guide that --unwind-info and -u
are not available for the elf format.
Differential Revision: https://reviews.llvm.org/D110633
Kazu Hirata [Thu, 7 Oct 2021 15:29:42 +0000 (08:29 -0700)]
[lldb, mlir] Migrate from getNumArgOperands and arg_operands (NFC)
Note that getNumArgOperands and arg_operands are considered legacy
names. See llvm/include/llvm/IR/InstrTypes.h for details.
Bradley Smith [Tue, 5 Oct 2021 11:05:20 +0000 (11:05 +0000)]
[AArch64][SVE] Improve VECTOR_SPLICE codegen for VL > 128-bit
Differential Revision: https://reviews.llvm.org/D111135
LLVM GN Syncbot [Thu, 7 Oct 2021 15:19:45 +0000 (15:19 +0000)]
[gn build] Port
7fb9f99f3bb6
LLVM GN Syncbot [Thu, 7 Oct 2021 15:19:44 +0000 (15:19 +0000)]
[gn build] Port
49e736d845d8
Mark de Wever [Mon, 14 Dec 2020 16:39:15 +0000 (17:39 +0100)]
[libc++][format] Adds bool formatter.
Implements the formatter for Boolean types.
[format.formatter.spec]/2.3
For each charT, for each cv-unqualified arithmetic type ArithmeticT other
than char, wchar_t, char8_t, char16_t, or char32_t, a specialization
```
template<> struct formatter<ArithmeticT, charT>;
```
This removes the stub implemented in D96664.
Implements parts of:
- P0645 Text Formatting
- P1652 Printf corner cases in std::format
Completes:
- P1868 width: clarifying units of width and precision in std::format
Reviewed By: #libc, ldionne
Differential Revision: https://reviews.llvm.org/D103670