Jonas Devlieghere [Wed, 3 Mar 2021 00:24:37 +0000 (16:24 -0800)]
[lldb] Add more logging to __lldb_apple_objc_v2_get_dynamic_class_info
Victor Huang [Wed, 3 Mar 2021 00:16:14 +0000 (18:16 -0600)]
[AIX][TLS] Generate TLS variables in assembly files
This patch allows generating TLS variables in assembly files on AIX.
Initialized and external uninitialized variables are generated with the
.csect pseudo-op and local uninitialized variables are generated with
the .comm/.lcomm pseudo-ops. The patch also adds a check to
explicitly say that TLS is not yet supported on AIX.
Reviewed by: daltenty, jasonliu, lei, nemanjai, sfertile
Originally patched by: bsaleil
Commandeered by: NeHuang
Differential Revision: https://reviews.llvm.org/D96184
Petr Hosek [Fri, 26 Feb 2021 20:52:53 +0000 (12:52 -0800)]
[runtimes] Use standalone build only for compiler-rt
compiler-rt needs to use standalone build because of the assumptions
made by its build, but other runtimes can use non-standalone build.
Differential Revision: https://reviews.llvm.org/D97575
zoecarver [Wed, 3 Mar 2021 00:17:22 +0000 (16:17 -0800)]
[libc++] Add bind_front function (P0356R5).
Implementes [[ http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p0356r5.html | P0356R5 ]]. Adds `bind_front` to `functional`.
Reviewed By: ldionne, #libc, Quuxplusone
Differential Revision: https://reviews.llvm.org/D60368
Jonas Devlieghere [Wed, 3 Mar 2021 00:11:34 +0000 (16:11 -0800)]
[lldb] Extend Python convenience variable table with equivalent APIs
Add a column to the table of convenience variables with the equivalent
API to get to the current debugger, target, process, etc.
We often get asked to make convenience variables available outside of
the interactive interpreter. After explaining why that's not possible, a
common complaint is that it's hard to find out how to get to these
variables in a non-interactive context, for example how to get to the
current frame when given a thread. This patch aims to alleviate that by
including the APIs to navigate between these instances in the table.
Differential revision: https://reviews.llvm.org/D97778
Neal (nealsid) [Tue, 2 Mar 2021 03:50:27 +0000 (19:50 -0800)]
Migrate to llvm::unique_function instead of static member functions for callbacks
A few cleanups suggested in another patch review's comments:
1. Use llvm:unique_function for storing & invoking callbacks from
Editline to IOHandler
2. Change return type of one of the callback setters from bool to void,
since it's return value was never used
3. Moved the callback setters inline & made them nonstatic, since that's
more consistent with other setter definitions
4. Removed the baton parameter since we no longer need it anymore
Differential revision: https://reviews.llvm.org/D50299
Arthur Eubanks [Tue, 2 Mar 2021 22:49:46 +0000 (14:49 -0800)]
[opt] Error if -debug-pass is specified alongside the new PM
Reviewed By: ychen
Differential Revision: https://reviews.llvm.org/D97810
Andrei Elovikov [Tue, 2 Mar 2021 23:08:19 +0000 (15:08 -0800)]
[NFCI][VPlan] Modify Recipes' print methods to honor Indent parameter
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D97787
Fangrui Song [Tue, 2 Mar 2021 23:30:44 +0000 (15:30 -0800)]
[test] Fix CodeGen/VE/Scalar tests
Aart Bik [Tue, 2 Mar 2021 20:17:13 +0000 (12:17 -0800)]
[mlir][sparse] do not ignore ordering for "dense" tensor linked with sparse type
Reviewed By: bixia
Differential Revision: https://reviews.llvm.org/D97795
Peter Steinfeld [Tue, 2 Mar 2021 15:53:10 +0000 (07:53 -0800)]
[flang] Detect circularly defined interfaces of procedures
It's possible to define a procedure whose interface depends on a procedure
which has an interface that depends on the original procedure. Such a circular
definition was causing the compiler to fall into an infinite loop when
resolving the name of the second procedure. It's also possible to create
circular dependency chains of more than two procedures.
I fixed this by adding the function HasCycle() to the class DeclarationVisitor
and calling it from DeclareProcEntity() to detect procedures with such
circularly defined interfaces. I marked the associated symbols of such
procedures by calling SetError() on them. When processing subsequent
procedures, I called HasError() before attempting to analyze their interfaces.
Unfortunately, this did not work.
With help from Tim, we determined that the SymbolSet used to track the
erroneous symbols was instantiated using a "<" operator which was defined using
the location of the name of the procedure. But the location of the procedure
name was being changed by a call to ReplaceName() between the times that the
calls to SetError() and HasError() were made. This caused HasError() to
incorrectly report that a symbol was not in the set of erroneous symbols.
I fixed this by changing SymbolSet to be an unordered set that uses the
contents of the name of the symbol as the basis for its hash function. This
works because the contents of the name of the symbol is preserved by
ReplaceName() even though its location changes.
I also fixed the error message used when reporting recursively defined
dummy procedure arguments by removing extra apostrophes and sorting the
list of symbols.
I also added tests that will crash the compiler without this change.
Note that the "<" operator is used in other contexts, for example, in the map
of characterized procedures, maps of items in equivalence sets, maps of
structure constructor values, ... All of these situations happen after name
resolution has been completed and all calls to ReplaceName() have already
happened and thus are not subject to the problem I ran into when ReplaceName()
was called when processing procedure entities.
Note also that the implementation of the "<" operator uses the relative
location in the cooked character stream as the basis of its implementation.
This is potentially problematic when symbols from diffent compilation units
(for example symbols originating in .mod files) are put into the same map since
their names will appear in two different source streams which may not be
allocated in the same relative positions in memory. But I was unable to create
a test that caused a problem. Using a direct comparison of the content of the
name of the symbol in the "<" operator has problems. Symbols in enclosing or
parallel scopes can have the same name. Also using the location of the symbol
in the cooked character stream has the advantage that it preserves the the
order of the symbols in a structure constructor constant, which makes matching
the values with the symbols relatively easy.
This patch supersedes D97749.
Differential Revision: https://reviews.llvm.org/D97774
Nico Weber [Tue, 2 Mar 2021 23:10:25 +0000 (18:10 -0500)]
hack to unbreak check-llvm on win after https://reviews.llvm.org/D97335
fix attempt http://reviews.llvm.org/rGbbdb4c8c9bcef0e didn't work
The problem is that the test tries to look up
llvm_orc_registerJITLoaderGDBWrapper from the llvm-jitlink.exe
executable, but the symbol wasn't exported. Just manually export it
for now. There's a FIXME with a suggestion for a real fix.
Kamlesh Kumar [Tue, 2 Mar 2021 23:02:47 +0000 (04:32 +0530)]
[libunwind] [risc-v] This patch is for fixing
immediate build failure when Cross Unwinding enabled.
Follow up patch will cleanup some Macros handling.
Differential Revision: https://reviews.llvm.org/D97762
Hansang Bae [Fri, 12 Feb 2021 23:36:02 +0000 (17:36 -0600)]
[OpenMP] Add allocator support for target memory
This is a preview of allocator support for target memory that depends on the
offload runtime API which allocates memory as described below.
llvm_omp_target_alloc_host(size_t size, int device_num);
-- Returns non-migratable memory owned by host.
-- Memory is accessible by host and device(s).
llvm_omp_target_alloc_shared(size_t size, int device_num);
-- Returns migratable memory owned by host and device.
-- Memory is accessible by host and device.
llvm_omp_target_alloc_device(size_t size, int device_num);
-- Returns memory owned by device.
-- Memory is only accessible by device.
New memory space and predefined allocator names are
-- llvm_omp_target_host_mem_space
-- llvm_omp_target_shared_mem_space
-- llvm_omp_target_device_mem_space
-- llvm_omp_target_host_mem_alloc
-- llvm_omp_target_shared_mem_alloc
-- llvm_omp_target_device_mem_alloc
Differential Revision: https://reviews.llvm.org/D96669
Christopher Di Bella [Tue, 2 Mar 2021 22:33:21 +0000 (22:33 +0000)]
[libcxx] adds common_reference to <type_traits>
Implements part of P0898R3 Standard Library Concepts
Reworks D74351 to use requires-clauses over SFINAE and so that it more
closely follows the wording.
Co-authored by: Michael Schellenberger Costa <mschellenbergercosta@googlemail.com>
(Michael did all the heavy lifting and I came in to polish it for
submission, since Michael is focussing on `std::format` now.)
Reviewed By: ldionne, #libc
Differential Revision: https://reviews.llvm.org/D96657
Matt Arsenault [Tue, 9 Feb 2021 17:09:20 +0000 (12:09 -0500)]
GlobalISel: Merge and cleanup more AMDGPU call lowering code
This merges more AMDGPU ABI lowering code into the generic call
lowering. Start cleaning up by factoring away more of the pack/unpack
logic into the buildCopy{To|From}Parts functions. These could use more
improvement, and the SelectionDAG versions are significantly more
complex, and we'll eventually have to emulate all of those cases too.
This is mostly NFC, but does result in some minor instruction
reordering. It also removes some of the limitations with mismatched
sizes the old code had. However, similarly to the merge on the input,
this is forcing gfx6/gfx7 to use the gfx8+ ABI (which is what we
actually want, but SelectionDAG is stuck using the weird emergent
ABI).
This also changes the load/store size for stack passed EVTs for
AArch64, which makes it consistent with the DAG behavior.
Adrian Prantl [Tue, 2 Mar 2021 19:03:38 +0000 (11:03 -0800)]
Promote scalars to load addresses when dereferencing them.
This is a follow-up to
188b0747c1664d962e94f00b5e3caac529fa1e26. This
is a very narrow fix to a more general problem. LLDB should be better
at distinguishing between implict and memory location descriptions.
rdar://
74902042
Nikita Popov [Tue, 2 Mar 2021 22:09:36 +0000 (23:09 +0100)]
[AST] Remove unused Loop member (NFC)
To fix some build bots after D89264.
Sam McCall [Mon, 1 Mar 2021 23:16:33 +0000 (00:16 +0100)]
[clangd] Move DraftStore from ClangdLSPServer into ClangdServer.
ClangdServer already gets notified of every change, so it makes sense for it to
be the source of truth.
This is a step towards having ClangdServer expose a FS that includes dirty
buffers: D94554
Related changes:
- version is now optional for ClangdServer, to preserve our existing fuzziness
in this area (missing version ==> autoincrement)
- ClangdServer::format{File,Range} are now more regular ClangdServer functions
that don't need the code passed in. While here, combine into one function.
- incremental content update logic is moved from DraftStore to
ClangdLSPServer, with most of the implementation in SourceCode.cpp.
DraftStore is now fairly trivial, and will probably ultimately be
*replaced* by the dirty FS stuff.
Differential Revision: https://reviews.llvm.org/D97738
Nathan James [Tue, 2 Mar 2021 21:57:16 +0000 (21:57 +0000)]
[cte][NFC] Remove all references to stdlib stream headers.
Inclusion of iostream is frobidden and using other stream classes from standard library is discouraged as per https://llvm.org/docs/CodingStandards.html#include-iostream-is-forbidden
Reviewed By: sammccall
Differential Revision: https://reviews.llvm.org/D97771
Markus Böck [Tue, 2 Mar 2021 21:44:51 +0000 (22:44 +0100)]
[PR48898][CMake] Support MinGW Toolchain tool sin llvm_ExternalProject_Add
Windows is in the unique position of having two drivers, clang-cl and normal GNU clang, depending on whether a GNU or MSVC target is used. The current implementation with the USE_TOOLCHAIN argument assumes that when CMAKE_SYSTEM_NAME is set to Windows that clang-cl should be used, which is the incorrect choice when targeting a GNU environment.
This patch solves this problem by adding an optional TARGET_TRIPLE argument to llvm_ExternalProject_Add, which sets the various CMAKE_<LANG>_COMPILER_TARGET variables. Additionally, if the triple is detected as an MSVC environment, clang-cl and similar MSVC specific tools will be used instead of the GNU tools.
Heejin Ahn [Sun, 28 Feb 2021 05:59:05 +0000 (21:59 -0800)]
[WebAssembly] Fix more ExceptionInfo grouping bugs
This fixes two bugs in `WebAssemblyExceptionInfo` grouping, created by
D97247. These two bugs are not easy to split into two different CLs,
because tests that fail for one also tend to fail for the other.
- In D97247, when fixing `ExceptionInfo` grouping by taking out
the unwind destination' exception from the unwind src's exception, we
just iterated the BBs in the function order, but this was incorrect;
this changes it to dominator tree preorder. Please refer to the
comments in the code for the reason and an example.
- After this subexception-taking-out fix, there still can be remaining
BBs we have to take out. When Exception B is taken out of Exception A
(because EHPad B is the unwind destination of EHPad A), there can
still be BBs within Exception A that are reachable from Exception B,
which also should be taken out. Please refer to the comments in the
code for more detailed explanation on why this can happen. To make
this possible, this splits `WebAssemblyException::addBlock` into two
parts: adding to a set and adding to a vector. We need to iterate on
BBs within a `WebAssemblyException` to fix this, so we add BBs to sets
first. But we add BBs to vectors later after we fix all incorrectness
because deleting BBs from vectors is expensive. I considered removing
the vector from `WebAssemblyException`, but it was not easy because
this class has to maintain a similar interface with `MachineLoop` to
be wrapped into a single interface `SortRegion`, which is used in
CFGSort.
Other misc. drive-by fixes:
- Make `WebAssemblyExceptionInfo` do not even run when wasm EH is not
used or the function doesn't have any EH pads, not to waste time
- Add `LLVM_DEBUG` lines for easy debugging
- Fix `preds` comments in cfg-stackify-eh.ll
- Fix `__cxa_throw`'s signature in cfg-stackify-eh.ll
Fixes https://github.com/emscripten-core/emscripten/issues/13554.
Reviewed By: dschuff, tlively
Differential Revision: https://reviews.llvm.org/D97677
Stephen Kelly [Mon, 28 Dec 2020 01:47:25 +0000 (01:47 +0000)]
[clang-tidy] Simplify unused RAII check
Fix handling of default construction where the constructor has a default arg.
Differential Revision: https://reviews.llvm.org/D97142
Nikita Popov [Sun, 11 Oct 2020 19:56:09 +0000 (21:56 +0200)]
[LICM] Make promotion faster
Even when MemorySSA-based LICM is used, an AST is still populated
for scalar promotion. As the AST has quadratic complexity, a lot
of time is spent in this step despite the existing access count
limit. This patch optimizes the identification of promotable stores.
The idea here is pretty simple: We're only interested in must-alias
mod sets of loop invariant pointers. As such, only populate the AST
with loop-invariant loads and stores (anything else is definitely
not promotable) and then discard any sets which alias with any of
the remaining, definitely non-promotable accesses.
If we promoted something, check whether this has made some other
accesses loop invariant and thus possible promotion candidates.
This is much faster in practice, because we need to perform AA
queries for O(NumPromotable^2 + NumPromotable*NumNonPromotable)
instead of O(NumTotal^2), and NumPromotable tends to be small.
Additionally, promotable accesses have loop invariant pointers,
for which AA is cheaper.
This has a signicant positive compile-time impact. We save ~1.8%
geomean on CTMark at O3, with 6% on lencod in particular and 25%
on individual files.
Conceptually, this change is NFC, but may not be so in practice,
because the AST is only an approximation, and can produce
different results depending on the order in which accesses are
added. However, there is at least no impact on the number of promotions
(licm.NumPromoted) in test-suite O3 configuration with this change.
Differential Revision: https://reviews.llvm.org/D89264
Yonghong Song [Tue, 2 Mar 2021 17:35:21 +0000 (09:35 -0800)]
BPF: Fix a bug in peephole TRUNC elimination optimization
Andrei Matei reported a llvm11 core dump for his bpf program
https://bugs.llvm.org/show_bug.cgi?id=48578
The core dump happens in LiveVariables analysis phase.
#4 0x00007fce54356bb0 __restore_rt
#5 0x00007fce4d51785e llvm::LiveVariables::HandleVirtRegUse(unsigned int,
llvm::MachineBasicBlock*, llvm::MachineInstr&)
#6 0x00007fce4d519abe llvm::LiveVariables::runOnInstr(llvm::MachineInstr&,
llvm::SmallVectorImpl<unsigned int>&)
#7 0x00007fce4d519ec6 llvm::LiveVariables::runOnBlock(llvm::MachineBasicBlock*, unsigned int)
#8 0x00007fce4d51a4bf llvm::LiveVariables::runOnMachineFunction(llvm::MachineFunction&)
The bug can be reproduced with llvm12 and latest trunk as well.
Futher analysis shows that there is a bug in BPF peephole
TRUNC elimination optimization, which tries to remove
unnecessary TRUNC operations (a <<= 32; a >>= 32).
Specifically, the compiler did wrong transformation for the
following patterns:
%1 = LDW ...
%2 = SLL_ri %1, 32
%3 = SRL_ri %2, 32
... %3 ...
%4 = SRA_ri %2, 32
... %4 ...
The current transformation did not check how many uses of %2
and did transformation like
%1 = LDW ...
... %1 ...
%4 = SRL_ri %2, 32
... %4 ...
and pseudo register %2 is used by not defined and
caused LiveVariables analysis core dump.
To fix the issue, when traversing back from SRL_ri to SLL_ri,
check to ensure SLL_ri has only one use. Otherwise, don't
do transformation.
Differential Revision: https://reviews.llvm.org/D97792
Thomas Preud'homme [Tue, 2 Mar 2021 17:04:34 +0000 (17:04 +0000)]
Add __builtin_isnan(__fp16) testcase
Reviewed By: rjmccall
Differential Revision: https://reviews.llvm.org/D97777
Greg McGary [Tue, 2 Mar 2021 09:20:22 +0000 (01:20 -0800)]
[lld-macho] Rework length check when opening input files
This reverts diff D97610 (commit
0223ab035c199e537a0040857ba147ced87fd533) and adds a one-line fix to verify that a `MemoryBufferRef` has sufficient length before reading a 4-byte magic number.
Differential Revision: https://reviews.llvm.org/D97757
Amara Emerson [Thu, 25 Feb 2021 06:45:25 +0000 (22:45 -0800)]
[AArch64][GlobalISel] Enable use of the optsize predicate in the selector.
To do this while supporting the existing functionality in SelectionDAG of using
PGO info, we add the ProfileSummaryInfo and LazyBlockFrequencyInfo analysis
dependencies to the instruction selector pass.
Then, use the predicate to generate constant pool loads for f32 materialization,
if we're targeting optsize/minsize.
Differential Revision: https://reviews.llvm.org/D97732
Hanhan Wang [Tue, 2 Mar 2021 20:48:12 +0000 (12:48 -0800)]
[mlir][linalg] Only generate one var for an attrUse.
Some variables are unused after D97383 landed. We should generate one symbol for one attrUse.
Reviewed By: stellaraccident
Differential Revision: https://reviews.llvm.org/D97794
Stefan Gränitz [Tue, 2 Mar 2021 20:42:11 +0000 (21:42 +0100)]
[llvm-jitlink] Prevent missing symbols from JITLoaderGDB with MSVC mangling
The issue came up on builder clang-x64-windows-msvc after
5182a7901a5d83dfd15021d01e8a1899910130ec
Martin Storsjö [Sat, 17 Oct 2020 20:57:19 +0000 (20:57 +0000)]
[libcxx] [test] Pass some windows environment variables through to test processes
Normally, the run.py wrapper script runs the child processes in
a clean environment, with only the environment variables available
that are passed via the --env parameter.
However, the COMSPEC and TEMP variables are kind of necessary when
running some tests; COMSPEC is necessary for finding the interpreter
when executing commands via std::system().
Before
f1a96de1bc8db527b5eb820c36c17e275900ca2b, tests were executed
via an intermediate shell which implicitly readded the COMSPEC variable.
The TEMP variable allows temp files to be placed in a sensible
location; if unset, they're placed in the default temp fallback of
C:\Windows instead.
Differential Revision: https://reviews.llvm.org/D97452
Martin Storsjö [Fri, 18 Dec 2020 11:34:35 +0000 (13:34 +0200)]
[libcxx] Explicitly return the expected error code in create_directories if the parent isn't a directory
On windows, going ahead and actually trying to create the directory
doesn't return an error code that maps to
std::errc::not_a_directory in this case.
This fixes two cases of
TEST_CHECK(ErrorIs(ec, std::errc::not_a_directory))
in filesystems/fs.op.funcs/fs.op.create_directories/create_directories.pass.cpp
for windows (in testcases added in
59c72a70121567f7aee347e96b4ac8f3cfe9f4b2).
Differential Revision: https://reviews.llvm.org/D97090
Sanjay Patel [Tue, 2 Mar 2021 18:50:22 +0000 (13:50 -0500)]
[SDAG] allow partial undef vector constants with select->logic folds
This is an enhancement suggested in the original review/commit:
D97730 /
7fce3322a283
Sanjay Patel [Tue, 2 Mar 2021 18:07:16 +0000 (13:07 -0500)]
[AArch64] add select tests with partial vector undefs; NFC
Mehdi Amini [Tue, 2 Mar 2021 19:20:10 +0000 (19:20 +0000)]
Fix `debuginfo-tests/llvm-prettyprinters` build after MLIR API change in
e6260ad043d8 (NFC)
David Green [Tue, 2 Mar 2021 19:01:14 +0000 (19:01 +0000)]
[ARM] Use 0, not ZR during ISel for CSINC/INV/NEG
Instead of converting the 0 into a ZR reg during lowering, do that with
tablegen by matching the zero immediate. This when combined with other
optimizations is more likely to use ZR and helps keep the DAG more
easily optimizable. It should not otherwise effect code generation.
Jonas Paulsson [Thu, 25 Feb 2021 22:05:43 +0000 (16:05 -0600)]
[SystemZ] Assign the full space for promoted and split outgoing args.
When a large "irregular" (e.g. i96) integer call argument is converted to
indirect, 64-bit parts are stored to the stack. The full stack space
(e.g. i128) was not allocated prior to this patch, but rather just the exact
space of the original type. This caused neighboring values on the stack to be
overwritten.
Thanks to Josh Stone for reporting this.
Review: Ulrich Weigand
Fixes https://bugs.llvm.org/show_bug.cgi?id=49322
Differential Revision: https://reviews.llvm.org/D97514
Peter Steinfeld [Tue, 2 Mar 2021 15:53:10 +0000 (07:53 -0800)]
[flang] Detect circularly defined interfaces of procedures
It's possible to define a procedure whose interface depends on a procedure
which has an interface that depends on the original procedure. Such a circular
definition was causing the compiler to fall into an infinite loop when
resolving the name of the second procedure. It's also possible to create
circular dependency chains of more than two procedures.
I fixed this by adding the function HasCycle() to the class DeclarationVisitor
and calling it from DeclareProcEntity() to detect procedures with such
circularly defined interfaces. I marked the associated symbols of such
procedures by calling SetError() on them. When processing subsequent
procedures, I called HasError() before attempting to analyze their interfaces.
Unfortunately, this did not work.
With help from Tim, we determined that the SymbolSet used to track the
erroneous symbols was instantiated using a "<" operator which was defined using
the location of the name of the procedure. But the location of the procedure
name was being changed by a call to ReplaceName() between the times that the
calls to SetError() and HasError() were made. This caused HasError() to
incorrectly report that a symbol was not in the set of erroneous symbols.
I fixed this by changing SymbolSet to be an unordered set that uses the
contents of the name of the symbol as the basis for its hash function. This
works because the contents of the name of the symbol is preserved by
ReplaceName() even though its location changes.
I also fixed the error message used when reporting recursively defined
dummy procedure arguments by removing extra apostrophes and sorting the
list of symbols.
I also added tests that will crash the compiler without this change.
Note that the "<" operator is used in other contexts, for example, in the map
of characterized procedures, maps of items in equivalence sets, maps of
structure constructor values, ... All of these situations happen after name
resolution has been completed and all calls to ReplaceName() have already
happened and thus are not subject to the problem I ran into when ReplaceName()
was called when processing procedure entities.
Note also that the implementation of the "<" operator uses the relative
location in the cooked character stream as the basis of its implementation.
This is potentially problematic when symbols from diffent compilation units
(for example symbols originating in .mod files) are put into the same map since
their names will appear in two different source streams which may not be
allocated in the same relative positions in memory. But I was unable to create
a test that caused a problem. Using a direct comparison of the content of the
name of the symbol in the "<" operator has problems. Symbols in enclosing or
parallel scopes can have the same name. Also using the location of the symbol
in the cooked character stream has the advantage that it preserves the the
order of the symbols in a structure constructor constant, which makes matching
the values with the symbols relatively easy.
This patch supersedes D97749.
Differential Revision: https://reviews.llvm.org/D97774
Nico Weber [Tue, 2 Mar 2021 18:40:56 +0000 (13:40 -0500)]
[gn build] fix llvm-jitlink tests on linux after
ef2389235c5dec0
Joe Nash [Fri, 26 Feb 2021 21:31:56 +0000 (16:31 -0500)]
[AMDGPU] Make OMod explicit for V_CVT_{U,I}*
Make OMod explicit instead of implied by HasModifiers in the
operand list. Requires explicitly setting HasOMod=1 for
irregular OMod usage in instruction V_CVT_{U,I}*
Reviewed By: foad
Differential Revision: https://reviews.llvm.org/D97587
Change-Id: I230e1476f529e816eec60e242531f23a99e3839f
Martin Storsjö [Fri, 26 Feb 2021 07:58:50 +0000 (09:58 +0200)]
[libcxx] [test] Fix a test error with condvars with trivial destruction
If the destructor is trivial (_LIBCPP_HAS_TRIVIAL_CONDVAR_DESTRUCTION,
the constructor always is), the compiler warns about the
std::condition_variable being unused.
Add a cast to void to silence the warning about the object being unused.
Differential Revision: https://reviews.llvm.org/D97540
Andrzej Warzynski [Tue, 2 Mar 2021 18:13:04 +0000 (18:13 +0000)]
[flang][f18] Add missing line in help text (nfc)
Louis Dionne [Tue, 2 Mar 2021 17:59:29 +0000 (12:59 -0500)]
[libc++] Add a utility script to run the Docker image used by builders
Several contributors have been asking me how to reproduce the CI
environment locally. This is the last step towards making that work
out-of-the-box. Basically, just run `libcxx/utils/ci/run-buildbot-container`
and you're good to go.
Differential Revision: https://reviews.llvm.org/D97782
Jez Ng [Tue, 2 Mar 2021 17:50:51 +0000 (12:50 -0500)]
[clang+lld] Pass -platform_version args to ld64.lld
Fix regression where we aren't passing `-platform_version` to new ld64.lld after {D95204}.
Most of the changes were originally in D95204, but I backed them out due to
test failures on builds which have `CLANG_DEFAULT_LINKER=lld`. The tests are
properly updated in this diff.
Reviewed By: #lld-macho, thakis
Differential Revision: https://reviews.llvm.org/D97741
Arnamoy Bhattacharyya [Tue, 2 Mar 2021 17:05:07 +0000 (17:05 +0000)]
[flang][driver] Fix -fdefault* family bug
This patch provides a fix for the `fdefault-*` family in f18
(Please consult `D96344` for details)
Differential Revision: https://reviews.llvm.org/D97724
Vy Nguyen [Thu, 25 Feb 2021 04:47:22 +0000 (23:47 -0500)]
[lld-macho] Change loadReexport to handle the case where a TAPI re-exports to reference documents nested within other TBD.
Currently, it was delibrately impleneted to not handle this case, but as it has turnt out, we need this feature.
The concrete use case is
`System/Library/Frameworks/Cocoa.framework/Versions/A/Cocoa` reexports
/System/Library/Frameworks/AppKit.framework/Versions/C/AppKit , which then rexports
/System/Library/PrivateFrameworks/UIFoundation.framework/Versions/A/UIFoundation
The current implemention uses a global currentTopLevelTapi, which is not reset until it finishes loading the whole tree.
This is a problem because if the top-level is set to Cocoa, then when we get to UIFoundation, it will try to find UIFoundation in the current top level, which is Cocoa and will not find it.
The right thing should be:
- When loading a library from a TBD file, re-exports need to be looked up in the auxiliary documents within the same TBD.
- When loading from an actual dylib, no additional TBD documents need to be examined.
- In no case does a re-export mentioned in one TBD file need to be looked up in a document in an auxiliary document from a different TBD file
Differential Revision: https://reviews.llvm.org/D97438
Krzysztof Parzyszek [Tue, 2 Mar 2021 14:16:26 +0000 (06:16 -0800)]
[TableGen] Add IntrNoMerge as intrinsic property
There is a function attribute 'nomerge' in addition to 'noduplicate'
and 'convergent'. Both 'noduplicate' and 'convergent' have corresponding
intrinsic properties. This patch adds an intrinsic property for the
'nomerge' attribute.
Differential Revision: https://reviews.llvm.org/D96364
Utkarsh Saxena [Tue, 2 Mar 2021 15:36:11 +0000 (16:36 +0100)]
Revert "Revert "[clangd] Use ML Code completion ranking as default.""
The ASAN failure was fixed by
https://github.com/llvm/llvm-project/commit/
bf935a034b345e905907c80030c67ef8f737d56a.
This reverts commit
7f086d74c347750c1da619058eb5b3e79c2fae14.
Fraser Cormack [Mon, 1 Mar 2021 15:31:56 +0000 (15:31 +0000)]
[RISCV] Support fixed-length INSERT_VECTOR_ELT
This patch enables support for lowering INSERT_VECTOR_ELT on
fixed-length vector types. The strategy follows that for scalable vector
types.
This patch also includes a quick fix to prevent the compiler infinitely
looping between lowering BUILD_VECTOR as VECTOR_SHUFFLE and back again.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D97698
Sam McCall [Fri, 26 Feb 2021 08:11:48 +0000 (09:11 +0100)]
[clangd] Include macro expansions in documentSymbol hierarchy
Browsing macro-generated symbols is confusing.
On the one hand, it seems very *useful* to be able to see the summary of
symbols that were generated.
On the other hand, some macros spew a lot of confusing symbols into the
namespace and when used repeatedly (ABSL_FLAG) can create a lot of spam
that's hard to navigate.
Design constraints:
- the macro expansion tree need not align with the AST, though it often
does in practice.
We address this by defining the nesting based on the *primary*
location of decls, rather than their ranges.
- DocumentSymbol.children[*].range should nest within DocumentSymbol.range
(This constraint is not in LSP "breadcrumbs" breaks without it)
We adjust macro ranges so they cover their "children", rather than
just the macro expansion
- LSP does not have a "macro expansion" symbolkind, nor does it allow a
symbol to have no kind. I've arbitrarily picked "null" as this is
unlikely to conflict with anything useful.
This patch makes all macros and children visible for simplicity+consistency,
though in some cases it may be better to elide the macro node.
We may consider adding heuristics for this in future (e.g. when it expands
to one decl only?) but it doesn't seem clear-cut to me.
Differential Revision: https://reviews.llvm.org/D97615
Alexey Bataev [Tue, 2 Mar 2021 16:23:50 +0000 (08:23 -0800)]
[Instcombine][NFC]Simplify logical reductions tests, NFC.
dfukalov [Mon, 21 Dec 2020 12:33:43 +0000 (15:33 +0300)]
[AA] Cache (optionally) estimated PartialAlias offsets.
For the cases of two clobbering loads and one loaded object is fully contained
in the second `BasicAAResult::aliasGEP` returns just `PartialAlias` that
is actually more common case of partial overlap, it doesn't say anything about
actual overlapping sizes.
AA users such as GVN and DSE have no functionality to estimate aliasing of GEPs
with non-constant offsets. The change stores estimated relative offsets so they
can be used further.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D93529
Thomas Preud'homme [Wed, 9 Dec 2020 22:18:32 +0000 (22:18 +0000)]
Stop traping on sNaN in __builtin_isinf
__builtin_isinf currently generates a floating-point compare operation
which triggers a trap when faced with a signaling NaN in StrictFP mode.
This commit uses integer operations instead to not generate any trap in
such a case.
Reviewed By: mibintc
Differential Revision: https://reviews.llvm.org/D97125
Jingu Kang [Tue, 2 Mar 2021 15:49:01 +0000 (15:49 +0000)]
Test Commmit
Jan Kratochvil [Tue, 2 Mar 2021 15:44:07 +0000 (16:44 +0100)]
[nfc] [lldb] DWARFDebugInfoEntry::Extract(): reindent
Jan Kratochvil [Tue, 2 Mar 2021 15:39:04 +0000 (16:39 +0100)]
[lldb] DWARFDebugInfoEntry::Extract(): Print an error for unsupported DW_FORM_*
Sam McCall [Mon, 22 Feb 2021 22:05:48 +0000 (23:05 +0100)]
[clangd] Show hex value of numeric constants
Don't show negative numbers
Don't show numbers <10 (hex is the same as decimal)
Show numeric enum values in hex too
Differential Revision: https://reviews.llvm.org/D97226
Nico Weber [Tue, 2 Mar 2021 15:31:40 +0000 (10:31 -0500)]
[gn build] (manually) port
99a6d003edbe
Nico Weber [Tue, 2 Mar 2021 15:26:34 +0000 (10:26 -0500)]
[gn build] Port
f47ff8cff1ed
Nico Weber [Tue, 2 Mar 2021 15:26:34 +0000 (10:26 -0500)]
[gn build] Port
ef2389235c5d
Utkarsh Saxena [Tue, 2 Mar 2021 14:23:36 +0000 (15:23 +0100)]
[clangd] Make categorical features 64 bit in DecisionForest Model.
CodeCompletionContext::Kind has 36 Kinds. The completion model used to
support categorical features of 32 cardinality.
Due to this clangd tests were failing asan tests due to overflow.
This patch makes the completion model support 64 cardinality of
categorical features by storing ENUM Features as uint64_t instead of
uint32_t.
Verified that this fixes the asan failures.
Latency: 6.7ms (old) VS 6.8ms (new) per 1000 predictions.
Differential Revision: https://reviews.llvm.org/D97770
Alexey Bataev [Thu, 11 Feb 2021 20:09:15 +0000 (12:09 -0800)]
[OPENMP50]Mapping of the subcomponents with the 'default' mappers.
If the mapped structure has data members, which have 'default' mappers,
need to map these members individually using their 'default' mappers.
Differential Revision: https://reviews.llvm.org/D92195
Tim Northover [Tue, 2 Mar 2021 14:38:58 +0000 (14:38 +0000)]
AArch64: report fp16 arithmetic is present for apple-a11 CPU.
AArch64.td got it right, but the target-parser dropped it, leading to missing
feature flags in Clang.
Simon Pilgrim [Tue, 2 Mar 2021 14:02:26 +0000 (14:02 +0000)]
[clang] DeltaTree::AddDelta - fix "initialization is never read" clang-tidy warning. NFCI.
Simon Pilgrim [Tue, 2 Mar 2021 13:42:03 +0000 (13:42 +0000)]
[DSE] eliminateDeadStoresMemorySSA - fix "initialization is never read" clang-tidy warning. NFCI.
Andrzej Warzynski [Tue, 2 Mar 2021 14:36:29 +0000 (14:36 +0000)]
Revert "[flang] Detect circularly defined interfaces of procedures"
This reverts commit
93c5e6bb49ca502d266700dd292e3873dfa51bb6.
This patch updates resolve102.f90 which is now failing in 6 out 8 of our
public buildbots:
* http://lab.llvm.org:8011/#/builders/21/builds/9625
* http://lab.llvm.org:8011/#/builders/134/builds/2395
* http://lab.llvm.org:8011/#/builders/79/builds/6298
* http://lab.llvm.org:8011/#/builders/66/builds/2084
* http://lab.llvm.org:8011/#/builders/135/builds/2485
* http://lab.llvm.org:8011/#/builders/32/builds/3551
Please see the following revisions for more context:
* https://reviews.llvm.org/D97201
* https://reviews.llvm.org/D97749
Sam McCall [Tue, 2 Mar 2021 14:53:55 +0000 (15:53 +0100)]
[clangd] findExplicitReferences impl filters nulls centrally. NFC
Alexey Bataev [Tue, 19 Jan 2021 20:10:39 +0000 (12:10 -0800)]
[SLP]Merge reorder and reuse shuffles.
It is possible to merge reuse and reorder shuffles and reduce the total
cost of the vectorization tree/number of final instructions.
Differential Revision: https://reviews.llvm.org/D94992
Stefan Gränitz [Tue, 2 Mar 2021 14:34:09 +0000 (15:34 +0100)]
[Orc] Fix MSVC error: conversion from 'initializer list' requires a narrowing
Frederik Gossen [Tue, 2 Mar 2021 14:29:08 +0000 (15:29 +0100)]
Split `ElementwiseMappable` trait into four more precise traits.
Some elementwise operations are not scalarizable, vectorizable, or tensorizable.
Split `ElementwiseMappable` trait into the following, more precise traits.
- `Elementwise`
- `Scalarizable`
- `Vectorizable`
- `Tensorizable`
This allows for reuse of `Elementwise` in dialects like HLO.
Differential Revision: https://reviews.llvm.org/D97674
Sanjay Patel [Tue, 2 Mar 2021 13:49:22 +0000 (08:49 -0500)]
[SDAG] allow vector types for select->logic folds
This prepares codegen for a change that will remove the identical
folds from IR because they are not poison-safe. See
D93065 / D97360
for details.
We already generically support scalar types, and there are various
target-specific transforms that overlap the vector folds. For example,
x86 recognizes the and patterns, but not or. We can end up with 1
extra instruction there, but I think that is still preferred over the
blendv alternative that loads a constant vector.
If this is not optimal, then it should be fixed with a later transform
(this change is not expected to result in any regressions because
InstCombine currently does the same thing).
Removing custom code and supporting undefs in constant-pattern-matching
can be follow-up changes.
Differential Revision: https://reviews.llvm.org/D97730
Jan Kratochvil [Tue, 2 Mar 2021 14:17:23 +0000 (15:17 +0100)]
[lldb] Support DWARF-5 DW_FORM_line_strp (used by GCC)
LLDB has been failing on binaries produced by new GCC 11.0 with -gdwarf-5.
Differential Revision: https://reviews.llvm.org/D97721
Stefan Gränitz [Fri, 12 Feb 2021 13:03:41 +0000 (13:03 +0000)]
[Orc] Fix remaining memory size of slab allocator
Stefan Gränitz [Tue, 2 Mar 2021 13:14:44 +0000 (14:14 +0100)]
[docs][JITLink] Fix a typo (NFC)
Stefan Gränitz [Tue, 2 Mar 2021 13:03:29 +0000 (14:03 +0100)]
[Orc] Extend lli debug support tests to JITLink
Stefan Gränitz [Tue, 23 Feb 2021 14:54:21 +0000 (14:54 +0000)]
[lli] Add JITLink in-process debug support
lli aims to provide both, RuntimeDyld and JITLink, as the dynamic linkers/loaders for it's JIT implementations. And they both offer debugging via the GDB JIT interface, which builds on the two well-known symbol names `__jit_debug_descriptor` and `__jit_debug_register_code`. As these symbols must be unique accross the linked executable, we can only define them in one of the libraries and make the other depend on it. OrcTargetProcess is a minimal stub for embedding a JIT client in remote executors. For the moment it seems reasonable to have the definition there and let ExecutionEngine depend on it, until we find a better solution.
This is the second commit for the reviewed patch.
Reviewed By: lhames
Differential Revision: https://reviews.llvm.org/D97339
Stefan Gränitz [Tue, 2 Mar 2021 11:37:48 +0000 (12:37 +0100)]
[Orc] Add JITLink debug support plugin for ELF x86-64
Add a new ObjectLinkingLayer plugin `DebugObjectManagerPlugin` and infrastructure to handle creation of `DebugObject`s as well as their registration in OrcTargetProcess. The current implementation only covers ELF on x86-64, but the infrastructure is not limited to that.
The journey starts with a new `LinkGraph` / `JITLinkContext` pair being created for a `MaterializationResponsibility` in ORC's `ObjectLinkingLayer`. It sends a `notifyMaterializing()` notification, which is forwarded to all registered plugins. The `DebugObjectManagerPlugin` aims to create a `DebugObject` form the provided target triple and object buffer. (Future implementations might create `DebugObject`s from a `LinkGraph` in other ways.) On success it will track it as the pending `DebugObject` for the `MaterializationResponsibility`.
This patch only implements the `ELFDebugObject` for `x86-64` targets. It follows the RuntimeDyld approach for debug object setup: it captures a copy of the input object, parses all section headers and prepares to patch their load-address fields with their final addresses in target memory. It instructs the plugin to report the section load-addresses once they are available. The plugin overrides `modifyPassConfig()` and installs a JITLink post-allocation pass to capture them.
Once JITLink emitted the finalized executable, the plugin emits and registers the `DebugObject`. For emission it requests a new `JITLinkMemoryManager::Allocation` with a single read-only segment, copies the object with patched section load-addresses over to working memory and triggers finalization to target memory. For registration, it notifies the `DebugObjectRegistrar` provided in the constructor and stores the previously pending`DebugObject` as registered for the corresponding MaterializationResponsibility.
The `DebugObjectRegistrar` registers the `DebugObject` with the target process. `llvm-jitlink` uses the `TPCDebugObjectRegistrar`, which calls `llvm_orc_registerJITLoaderGDBWrapper()` in the target process via `TargetProcessControl` to emit a `jit_code_entry` compatible with the GDB JIT interface [1]. So far the implementation only supports registration and no removal. It appears to me that it wouldn't raise any new design questions, so I left this as an addition for the near future.
[1] https://sourceware.org/gdb/current/onlinedocs/gdb/JIT-Interface.html
Reviewed By: lhames
Differential Revision: https://reviews.llvm.org/D97335
Stefan Gränitz [Tue, 23 Feb 2021 14:36:02 +0000 (14:36 +0000)]
[Orc] Rename local variable to avoid confusion with equally-named class member (NFC)
Stefan Gränitz [Thu, 18 Feb 2021 13:48:23 +0000 (13:48 +0000)]
[Orc] Fix a file header (NFC)
Stefan Gränitz [Thu, 18 Feb 2021 13:47:47 +0000 (13:47 +0000)]
[JITLink] LinkGraph::getName() can be const
Stefan Gränitz [Fri, 12 Feb 2021 13:58:57 +0000 (13:58 +0000)]
[JITLink] Remove some std::move(MemoryBufferRef) below createLinkGraphFromObject() (NFC)
Stefan Gränitz [Tue, 23 Feb 2021 14:33:11 +0000 (14:33 +0000)]
[llvm-jitlink] Remove duplicate type defintion (NFC)
Stefan Gränitz [Tue, 2 Mar 2021 09:41:52 +0000 (10:41 +0100)]
[lli] Add --jit-linker command line argument
The argument value determines the dynamic linker to use (`default`, `rtdyld` or `jitlink`). The JITLink implementation only supports in-process JITing for now. This is the first commit for the reviewed patch.
Reviewed By: lhames
Differential Revision: https://reviews.llvm.org/D97339
Utkarsh Saxena [Tue, 2 Mar 2021 13:57:26 +0000 (14:57 +0100)]
Revert "[clangd] Use ML Code completion ranking as default."
CodeCompletionContext::Kind has 36 Kinds. The completion model currently
only handles categorical features of 32 cardinality.
Changing the datatype to uint64_t will solve the problem.
This reverts commit
438b5bb05a429d697674088d400e4800c1108658.
Peyton, Jonathan L [Tue, 2 Mar 2021 13:44:15 +0000 (07:44 -0600)]
[OpenMP] Fix clang-cl build error regarding TSX intrinsics
Fix for https://bugs.llvm.org/show_bug.cgi?id=49339
The CMake check for the RTM intrinsics needs the -mrtm flag to be set
during the test. This way clang-cl correctly detects it has the
_xbegin() intrinsic. Otherwise, the CMake check fails.
Differential Revision: https://reviews.llvm.org/D97413
Ed Maste [Mon, 1 Mar 2021 18:17:30 +0000 (13:17 -0500)]
[Driver] Fix -gz=zlib options for linker also on FreeBSD
ccb4124a4172 fixed translating -gz=zlib to --compress-debug-sections for
linker invocation for several ToolChains, but omitted FreeBSD.
Differential Revision: https://reviews.llvm.org/D97752
Simon Pilgrim [Tue, 2 Mar 2021 12:22:16 +0000 (12:22 +0000)]
[DAG] DAGCombiner::tryStoreMergeOfLoads - remove unused StartAddress variable. NFCI.
Noticed in "initialization is never read" clang-tidy warning - the only StartAddress set/used is inside the load combine loop.
Benjamin Kramer [Tue, 2 Mar 2021 12:41:50 +0000 (13:41 +0100)]
[MCParser] Bring back srcmanager diagnostics in AsmParser
AsmParser may have no LLVMContext attached to it, which means after
5de2d189e6ad466a1f0616195e8c524a4eb3cbc0 everything goes to stderr.
Restore the old behavior.
Christian Sigg [Tue, 2 Mar 2021 12:27:35 +0000 (13:27 +0100)]
[mlir] Use mlir::OpState::operator->() to get to Operation::getAttrs().
This is a preparation step to remove getAttrs() from OpState.
Muhammad Omair Javaid [Tue, 2 Mar 2021 12:20:46 +0000 (17:20 +0500)]
Support GDB remote g packet partial read
GDB remote protocol does not specify length of g packet for register read. It depends on remote to include all or exclude certain registers from g packet. In case a register or set of registers is not included as part of g packet then we should fall back to p packet for reading all registers excluded from g packet by remote. This patch adds support for above feature and adds a test-case for the same.
Reviewed By: labath
Differential Revision: https://reviews.llvm.org/D97498
Jan Svoboda [Tue, 2 Mar 2021 12:14:00 +0000 (13:14 +0100)]
[clang][cli] NFC: Use .str() instead of operator<<
Simon Pilgrim [Tue, 2 Mar 2021 12:06:10 +0000 (12:06 +0000)]
[AMDGPU] Fix "initialization is never read" clang-tidy warnings. NFCI.
Jan Svoboda [Tue, 2 Mar 2021 12:02:04 +0000 (13:02 +0100)]
[clang][cli] NFC: Use .str() instead of operator<<
Fraser Cormack [Mon, 1 Mar 2021 15:57:42 +0000 (15:57 +0000)]
[RISCV] Lower CONCAT_VECTORS to INSERT_SUBVECTOR nodes
The default expansion of CONCAT_VECTORS goes through the stack. This
patch avoids that penalty by custom-lowering CONCAT_VECTORS to a series
of INSERT_SUBVECTOR nodes. Futher optimizations are possible, but this
is a good start.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D97692
Peter Smith [Fri, 26 Feb 2021 13:14:21 +0000 (13:14 +0000)]
[LLD][ELF][ARM] Refactor inBranchRange to use addend for PC Bias
In AArch32 ARM, the PC reads two instructions ahead of the currently
executiing instruction. This evaluates to 8 in ARM state and 4 in
Thumb state. Branch instructions on AArch32 compensate for this by
subtracting the PC bias from the addend. For a branch to symbol this
will result in an addend of -8 in ARM state and -4 in Thumb state.
The existing ARM Target::inBranchRange function accounted for this
implict addend within the function meaning that if the addend were
to be taken into account by the caller then it would be double
counted. This complicates the interface for all Targets as callers
wanting to account for addends had to account for the ARM PC-bias.
In certain situations such as:
https://github.com/ClangBuiltLinux/linux/issues/1305
the PC-bias compensation code didn't match up. In particular
normalizeExistingThunk() didn't put the PC-bias back in as Arm
thunks did not store the addend.
The simplest fix for the problem is to add the PC bias in
normalizeExistingThunk when restoring the addend. However I think
it is worth refactoring the Arm inBranchRange implementation so
that fewer calls to getPCBias are needed for other Targets. I
wasn't able to remove getPCBias completely but hopefully the
Relocations.cpp code is simpler now.
In principle a test could be written to replicate the linux kernel
build failure but I wasn't able to reproduce with a small example
that I could build up from scratch.
Fixes https://github.com/ClangBuiltLinux/linux/issues/1305
Differential Revision: https://reviews.llvm.org/D97550
Jan Svoboda [Tue, 2 Mar 2021 10:53:40 +0000 (11:53 +0100)]
[clang][cli] NFC: Rename marshalling multiclass
The new name drops `String` from `MarshallingInfoStringInt`, which follows the naming convention of other marshalling multiclasses.
Florian Hahn [Tue, 2 Mar 2021 10:43:33 +0000 (10:43 +0000)]
[LV] Add test cases that require a larger number of RT checks.
Precommit tests cases for D75981.
KareemErgawy-TomTom [Tue, 2 Mar 2021 10:16:28 +0000 (11:16 +0100)]
[MLIR][LinAlg] Detensorize interal function control flow.
This patch continues detensorizing implementation by detensoring
internal control flow in functions.
In order to detensorize functions, all the non-entry block's arguments
are detensored and branches between such blocks are properly updated to
reflect the detensored types as well. Function entry block (signature)
is left intact.
This continues work towards handling github/google/iree#1159.
Reviewed By: silvas
Differential Revision: https://reviews.llvm.org/D97148
Jan Svoboda [Tue, 2 Mar 2021 10:32:03 +0000 (11:32 +0100)]
[clang][docs] Fix code blocks rendering
Some code blocks that render fine locally don't appear on the llvm.org website. Attempting to fix this by specifying the `text` type.
Benjamin Kramer [Tue, 2 Mar 2021 10:21:54 +0000 (11:21 +0100)]
Revert "[X86] Fold shuffle(not(x),undef) -> not(shuffle(x,undef))"
This reverts commit
925093d88ae74560a8e94cf66f95d60ea3ffa2d3.
Causes an infinite loop when compiling some shuffles:
$ cat bugpoint-reduced-simplified.ll
target triple = "x86_64-unknown-linux-gnu"
define void @foo() {
entry:
%0 = load i8, i8* undef, align 1
%broadcast.splatinsert = insertelement <16 x i8> poison, i8 %0, i32 0
%1 = icmp ne <16 x i8> %broadcast.splatinsert, zeroinitializer
%2 = shufflevector <16 x i1> %1, <16 x i1> undef, <16 x i32> zeroinitializer
%wide.load = load <16 x i8>, <16 x i8>* undef, align 1
%3 = icmp ne <16 x i8> %wide.load, zeroinitializer
%4 = and <16 x i1> %3, %2
%5 = zext <16 x i1> %4 to <16 x i8>
store <16 x i8> %5, <16 x i8>* undef, align 1
ret void
}
$ llc < bugpoint-reduced-simplified.ll
<timeout>