Nikita Popov [Mon, 10 Aug 2020 20:06:48 +0000 (22:06 +0200)]
[InstSimplify] Forbid undef folds in expandBinOp
This is the replacement for D84250 based on D84792. As we recursively
fold with the same value twice, we need to disable undef folds,
to prevent an undef from being folded to two different values.
Reverting rG00f3579aea6e3d4a4b7464c3db47294f71cef9e4 and using the
test case from https://reviews.llvm.org/D83360#2145793, it no longer
performs the incorrect fold.
Differential Revision: https://reviews.llvm.org/D85684
David Goldman [Thu, 9 Jul 2020 18:29:15 +0000 (14:29 -0400)]
[clangd][ObjC] Improve xrefs for protocols and classes
Summary:
Previously clangd would jump to forward declarations for protocols
and classes instead of their definition/implementation.
Reviewers: sammccall
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D83501
Yitzhak Mandelbaum [Tue, 11 Aug 2020 16:35:36 +0000 (16:35 +0000)]
[libTooling] Cleanup and reorder `RewriteRule.h`.
This patch lifts `RootID` out of the `RewriteRule` class so that constructs
(e.g. inline functions) can that refer to the root id don't need to depend on
the `RewriteRule` class.
With this dependency, the patch is able to collect all `ASTEdit` helper function
declarations together with the class declaration, before the introduction of the
`RewriteRule` class. In the process, we also adjust some of the comments.
This patch is essentially a NFC.
Reviewed By: gribozavr2
Differential Revision: https://reviews.llvm.org/D85733
Eric Christopher [Tue, 11 Aug 2020 16:30:51 +0000 (09:30 -0700)]
Fold Opcode into assert uses to fix an unused variable warning without asserts.
Xing GUO [Tue, 11 Aug 2020 14:46:46 +0000 (22:46 +0800)]
[DWARFYAML] Teach yaml2obj emit the correct line table program.
The following issues are addressed in this patch.
1. The operands of DW_LNE_set_discriminator should be an ULEB128 number
rather than an address.
2. Test the emitted opcodes.
Reviewed By: jhenderson
Differential Revision: https://reviews.llvm.org/D85717
Simon Pilgrim [Tue, 11 Aug 2020 15:42:37 +0000 (16:42 +0100)]
[X86][SSE] combineShuffleWithHorizOp - avoid unnecessary subtraction. NFCI.
We can safely replace ((M - NumElts) % NumEltsPerLane) with (M % NumEltsPerLane) as the modulo result will be the same.
Jonas Devlieghere [Tue, 11 Aug 2020 15:43:44 +0000 (08:43 -0700)]
[lldb/CMake] Separate CMake code for Lua and Python (NFC)
Separate the CMake logic for Lua and Python to clearly distinguish
between code specific to either scripting language and the code shared
by both.
What this patch does is:
- Move Python specific code into the bindings/python subdirectory.
- Move the Lua specific code into the bindings/lua subdirectory.
- Add the _python suffix to Python specific functions/targets.
- Fix a dependency issue that would check the binding instead of
whether the scripting language is enabled.
Note that this patch also changes where the bindings are generated,
which might affect downstream projects that check them in.
Differential revision: https://reviews.llvm.org/D85708
Matt Arsenault [Thu, 23 Jul 2020 03:47:35 +0000 (23:47 -0400)]
AMDGPU/GlobalISel: Manually select llvm.amdgcn.writelane
Fixup the special case constant bus handling pre-gfx10.
Tim Keith [Tue, 11 Aug 2020 15:37:23 +0000 (08:37 -0700)]
[flang] Allow compiler directives in more places
Allow compiler directives in the implicit-part and before USE statements
in the specification-part.
Differential Revision: https://reviews.llvm.org/D85693
Whitney Tsang [Tue, 11 Aug 2020 15:33:10 +0000 (15:33 +0000)]
[NFC][LoopUnrollAndJam] Use BasicBlock::replacePhiUsesWith instead of
static function updatePHIBlocks.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D85673
Jay Foad [Tue, 11 Aug 2020 09:50:58 +0000 (10:50 +0100)]
[GlobalISel] Add G_ABS
This is equivalent to the new llvm.abs intrinsic added by D84125 with
is_int_min_poison=0.
Differential Revision: https://reviews.llvm.org/D85718
Gousemoodhin Nadaf [Tue, 11 Aug 2020 14:48:24 +0000 (16:48 +0200)]
[clang] Do not crash for unsupported fixed point to floating point conversion
- Fixed point to floating point conversion is unimplemented.
- If one of the operands has a floating type and the other operand has a fixed-point type, the function
handleFloatConversion() is called because one of the operands has a floating type, but we do not handle fixed
point type in this function (Implementation of fixed point to floating point conversion is missing), due to this
compiler crashes. In order to avoid compiler crash, when one of the operands has a floating type and the other
operand has a fixed-point type, return NULL.
- FIXME: Implementation of fixed point to floating point conversion.
- I am going to resolve FIXME in followup patches.
- Add the test case.
Reviewed By: ebevhan
Differential Revision: https://reviews.llvm.org/D81904
Bevin Hansson [Tue, 7 Jul 2020 17:21:29 +0000 (19:21 +0200)]
[Sema] Iteratively strip sugar when removing address spaces.
ASTContext::removeAddrSpaceQualType does not properly deal
with sugar. QualTypes derive their ASes from the AS on the
canonical type, not the type itself. However,
removeAddrSpaceQualType only strips the outermost qualifiers,
which means that it can fail to remove addrspace qualifiers
if there is sugar in the way.
Change the function to desugar types until the address space
really no longer exists on the corresponding QualType. This
should guarantee the removal of the address space.
This fixes the erroneous behavior in D62574.
Reviewed By: rjmccall, svenvh
Differential Revision: https://reviews.llvm.org/D83325
Sanjay Patel [Tue, 11 Aug 2020 12:08:46 +0000 (08:08 -0400)]
[InstSimplify] fold min/max with matching min/max operands
I think this is the last remaining translation of an existing
instcombine transform for the corresponding cmp+sel idiom.
This interpretation is more general though - we can remove
mismatched signed/unsigned combinations in addition to the
more obvious cases.
min/max(X, Y) must produce X or Y as the result, so this is
just another clause in the existing transform that was already
matching a min/max of min/max.
Sanjay Patel [Mon, 10 Aug 2020 16:05:12 +0000 (12:05 -0400)]
[InstSimplify] add tests for min/max intrinsics with common operands; NFC
There are 4*4*4 = 64 variations. We currently handle some, but not all,
of the alternative patterns with cmp+sel in instcombine.
Sanjay Patel [Mon, 10 Aug 2020 12:53:28 +0000 (08:53 -0400)]
[InstCombine] add tests for pointer casts with insertelement; NFC
Alex Zinenko [Tue, 11 Aug 2020 15:17:27 +0000 (17:17 +0200)]
[mlir] Add verification to LLVM dialect types
Now that LLVM dialect types are implemented directly in the dialect, we can use
MLIR hooks for verifying type construction invariants. Implement the verifiers
and use them in the parser.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D85663
Simon Pilgrim [Tue, 11 Aug 2020 13:55:11 +0000 (14:55 +0100)]
[X86][SSE] Add HADD/SUB support to combineHorizOpWithShuffle
Handles some HOP(SHUFFLE,SHUFFLE) patterns and sets us up to improve some of the cases mentioned in PR41813.
Matt Arsenault [Sun, 24 May 2020 18:31:06 +0000 (14:31 -0400)]
AMDGPU/GlobalISel: Prepare for more custom load lowerings
Slight restructuring of the code to avoid formatting changes when more
cases are handled here.
Matt Arsenault [Sat, 8 Aug 2020 13:33:23 +0000 (09:33 -0400)]
AMDGPU/GlobalISel: Fix test bugs and add a few more cases
The wrong alignment or addrspace was used. Also add various cases to
stress a future patch.
Fangrui Song [Mon, 10 Aug 2020 15:57:19 +0000 (08:57 -0700)]
[ELF] Move the outSecOff addend from relocAlloc/relocNonAlloc/... to InputSectionBase::relocate
For an InputSection, the `buf` argument of `InputSectionBase::relocate` points
to the content of the containing OutputSection, instead of the content of the
InputSection itself, so `outSecOff` needs to be added in its callees. This is
counter-intuitive and leads to many `- outSecOff` and `+ outSecOff`.
This patch makes `InputSection::writeTo` call `InputSectionBase::relocate` with
`outSecOff` added. relocAlloc/relocNonAlloc/relocateNonAllocForRelocatable can
thus be simplified now.
Updated test:
* non-abs-reloc.s: A minor offset bug is fixed for a diagnostic in `relocateNonAlloc`
Reviewed By: grimar
Differential Revision: https://reviews.llvm.org/D85618
Fangrui Song [Mon, 10 Aug 2020 15:52:16 +0000 (08:52 -0700)]
[ELF][test] Enhance x86-64-split-stack-prologue-adjust-success.s & non-abs-reloc.s
Catch problems when outSecOff is not zero and outSecOff is added twice or not added.
Fangrui Song [Mon, 10 Aug 2020 22:02:10 +0000 (15:02 -0700)]
[split-file] Fix sys::fs::remove() on Solaris after D83834
where stdio.h ::remove() may set errno to EEXIST instead of ENOTEMPTY.
POSIX.1-2017 allows EEXIST for unlink() (which is called by remove()):
> [EEXIST] or [ENOTEMPTY]
> The flag parameter has the AT_REMOVEDIR bit set and the path argument names a directory that is not an empty directory, or there are hard links to the directory other than dot or a single entry in dot-dot.
Reviewed By: ro
Differential Revision: https://reviews.llvm.org/D85690
Bruno Ricci [Tue, 11 Aug 2020 14:54:15 +0000 (15:54 +0100)]
[clang] Add a matcher for template template parameters.
There are already matchers for type template parameters and non-type template
parameters, but somehow no matcher exists for template template parameters
and I need it to write unit tests.
Differential Revision: https://reviews.llvm.org/D85536
Reviewed By: aaron.ballman
David Stenberg [Tue, 11 Aug 2020 11:48:30 +0000 (13:48 +0200)]
[DebugInfo] Allow GNU macro extension to be emitted
Allow the GNU .debug_macro extension to be emitted for DWARF versions
earlier than 5. The extension is basically what became DWARF 5's format,
except that a DW_AT_GNU_macros attribute is emitted, and some entries
like the strx entries are missing. In this patch I emit GNU's indirect
entries, which are the same as DWARF 5's strp entries.
This patch adds the extension behind a hidden LLVM flag,
-use-gnu-debug-macro. I would later want to enable it by default when
tuning for GDB and targeting DWARF versions earlier than 5.
The size of a Clang 8.0 binary built with RelWithDebInfo and the flags
"-gdwarf-4 -fdebug-macro" reduces from 1533 MB to 1349 MB with
.debug_macro (compared to 1296 MB without -fdebug-macro).
Reviewed By: SouraVX, dblaikie
Differential Revision: https://reviews.llvm.org/D82975
David Stenberg [Tue, 11 Aug 2020 11:46:59 +0000 (13:46 +0200)]
[DebugInfo] Simplify DwarfDebug::emitMacro
Broken out from a review comment on D82975. This is an NFC expect for
that the Macinfo macro string is now emitted using a single emitBytes()
invocation, so it can be done using a single string directive.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D83557
Benjamin Kramer [Tue, 11 Aug 2020 14:56:45 +0000 (16:56 +0200)]
[GlobalISel] Remove unused variable. NFC.
Xing GUO [Tue, 11 Aug 2020 14:41:14 +0000 (22:41 +0800)]
[DWARFYAML] Let the address size of line tables inferred from the object file.
Currently, the line table uses the first compilation unit's address size
as its address size. It's not the right behavior. The address size should be
inferred from the target machine.
Reviewed By: jhenderson
Differential Revision: https://reviews.llvm.org/D85707
Valentin Clement [Tue, 11 Aug 2020 14:43:56 +0000 (10:43 -0400)]
[flang][directives] Use TableGen information for clause classes in parse-tree
This patch takes advantage of the directive information and tablegen generation
to replace the clauses class parse tree and in the dump parse tree sections.
Reviewed By: sscalpone
Differential Revision: https://reviews.llvm.org/D85549
Matt Arsenault [Tue, 16 Jun 2020 01:35:15 +0000 (21:35 -0400)]
GlobalISel: Implement bitcast action for G_INSERT_VECTOR_ELT
This mirrors the support for the equivalent extracts. This also
creates a huge mess that would be greatly improved if we had any bit
operation combines.
Dinar Temirbulatov [Tue, 11 Aug 2020 14:34:47 +0000 (16:34 +0200)]
[NFC] Guard the cost report block of debug outputs with NDEBUG and
switch to SmallString, this is part of D57779.
Raphael Isemann [Tue, 11 Aug 2020 12:42:26 +0000 (14:42 +0200)]
[clang] Don't make ObjCIvarDecl visible twice when adding them to an implicit ObjCInterfaceDecl
`addDecl` is making the ivar visible in its primary context. The primary context
of the ivar here is in a 'fragile' ABI the ObjCInterfaceDecl and in a
'non-fragile' ABI the current ObjCImplementationDecl. The additional call to
`makeDeclVisibleInContext` to make the ivar visible in the ObjCInterfaceDecl is
only necessary in the 'non-fragile' case (as in the 'fragile' case the Decl
becomes automatically visible in the ObjCInterfaceDecl with the `addDecl` call
as thats its primary context). See `Sema::ActOnIvar` for where the ivar is put
into a different context depending on the ABI.
To put this into an example:
```
lang=c++
@implementation SomeClass
{
id ivar1;
}
@end
fragile case:
implicit ObjCInterfaceDecl 'SomeClass'
`- ivar1 (in primary context and will be automatically made visible)
ObjCImplementationDecl 'SomeClass'
non-fragile case:
implicit ObjCInterfaceDecl 'SomeClass'
`-<<<ivar1 not visible here and needs to be manually marked as visible.>>>
ObjCImplementationDecl 'SomeClass'
`- ivar1 (in its primary context and will be automatically made visible here)
```
Making a Decl visible multiple times in the same context is inefficient and
potentially can lead to crashes. See D84827 for more info and what this is
breaking.
Reviewed By: aprantl
Differential Revision: https://reviews.llvm.org/D84829
Raphael Isemann [Tue, 11 Aug 2020 12:37:53 +0000 (14:37 +0200)]
[clang] Don't make synthesized accessor stub functions visible twice
`addDecl` is making the Decl visible, so there is no need to make it explicitly
visible again. Making it visible twice will also make the lookup storage less
efficient and potentially lead to crashes, see D84827 for that.
Reviewed By: aprantl
Differential Revision: https://reviews.llvm.org/D84828
Matt Arsenault [Sun, 2 Aug 2020 21:23:52 +0000 (17:23 -0400)]
TableGen/GlobalISel: Hack the operand order for atomic_store
ISD::ATOMIC_STORE arbitrarily has the operands in the opposite order
from regular ISD::STORE, which always introduced an annoying
duplication of patterns to handle both cases. Since in GlobalISel
there's just the one G_STORE, we need to swap the operands to
correctly emit the type check for the pointer operand.
Some work started in
20aafa31569b5157e792daa8860d71dd0df8a53a to
migrate SelectionDAG to use ISD::STORE for atomics, but that work
seems to have stalled. Since this is the pretty much the last
operation which matters which isn't supported for AMDGPU, use this
compatibility hack to unblock declaring it functionally complete.
Not sure what's going on with the pending_phis AArch64 test. It seems
it didn't always use atomics, and I'm not sure what it was originally
testing matters anymore.
Tim Keith [Tue, 11 Aug 2020 14:20:13 +0000 (07:20 -0700)]
[flang] Disable -Wmaybe-uninitialized for GCC
According to llvm/cmake/config-ix.cmake, gcc's potentially uninitialized
use analysis has lots of false positives. We are encountering one in
flang/lib/Lower/CharacterExpr.cpp.
That warning is disabled for gcc in in-tree builds; this does the same
thing for out-of-tree builds.
Differential Revision: https://reviews.llvm.org/D85694
Sam Parker [Tue, 11 Aug 2020 14:10:49 +0000 (15:10 +0100)]
[NFC][ARM][SimplifyCFG] Add some tests.
Add some tests around thresholds and minsize.
Kadir Cetinkaya [Sat, 8 Aug 2020 23:02:23 +0000 (01:02 +0200)]
[clang][HeaderInsert] Do not treat defines with values as header guards
This was resulting in inserting headers at bogus locations, see
https://github.com/ycm-core/YouCompleteMe/issues/3736 for an example.
Differential Revision: https://reviews.llvm.org/D85590
Kadir Cetinkaya [Tue, 11 Aug 2020 13:03:32 +0000 (15:03 +0200)]
[clangd] Disable ExtractFunction for C
This tweak uses constructs like auto and refs, which are not available
in C.
Differential Revision: https://reviews.llvm.org/D85727
Kadir Cetinkaya [Tue, 11 Aug 2020 11:32:00 +0000 (13:32 +0200)]
[clangd] Unify macro matching in code completion for AST and Index based macros
fixes https://github.com/clangd/clangd/issues/489
Differential Revision: https://reviews.llvm.org/D85721
clementval [Tue, 11 Aug 2020 13:54:04 +0000 (09:54 -0400)]
Revert "[flang][directives] Use TableGen information for clause classes in parse-tree"
This reverts commit
bf93edc4758b512e84b70ec6335d15b0cf386a18.
Buildbot failure
Georgii Rymar [Wed, 5 Aug 2020 13:41:30 +0000 (16:41 +0300)]
[llvm-readobj/elf] - Remove `unwrapOrError` calls from DumpStyle<ELFT>::printRelocationsHelper.
It removes all `unwrapOrError` calls except the first one, which
is is bit different and can be removed separately.
Differential revision: https://reviews.llvm.org/D85303
Rainer Orth [Tue, 11 Aug 2020 13:46:38 +0000 (15:46 +0200)]
[compiler-rt][builtins] Un-xfail two tests on sparcv9
Two tests currently `XPASS` on sparcv9:
Unexpectedly Passed Tests (2):
Builtins-sparcv9-sunos :: compiler_rt_logbl_test.c
Builtins-sparcv9-sunos :: divtc3_test.c
The following patch fixes this.
Tested on `sparcv9-sun-solaris2.11`.
Differential Revision: https://reviews.llvm.org/D85119
Georgii Rymar [Tue, 11 Aug 2020 13:44:50 +0000 (16:44 +0300)]
[llvm-readobj][test] - Rename gnu-phdrs.test to program-headers.test, NFCI.
program-headers.test was merged to gnu-phdrs.test and removed in D85526.
Now, gnu-phdrs.test contains tests for both GNU and LLVM style.
It was decided that this renaming should be done separately for a better
commit history and diffing purposes.
Valentin Clement [Mon, 10 Aug 2020 19:44:38 +0000 (15:44 -0400)]
[flang][directives] Use TableGen information for clause classes in parse-tree
This patch takes advantage of the directive information and tablegen generation
to replace the clauses class parse tree and in the dump parse tree sections.
Reviewed By: sscalpone
Differential Revision: https://reviews.llvm.org/D85549
Georgii Rymar [Fri, 7 Aug 2020 13:42:33 +0000 (16:42 +0300)]
[llvm-readobj] - Refine program headers testing (`-l`, `--program-headers` and `--segment`).
We have `gnu-phdrs.test` that tests
`-l`, `--program-headers` and `--segment` options for `llvm-readelf`.
We also have:
1) `program-headers.test` that tests `--program-headers` and `-l`
for `llvm-readobj`. It doesn't test `--segment` and also uses
lots of precompiled objects. It is very incomplete in compare
with the `gnu-phdrs.test`.
2) `pt-gnu-property.test` that contains a simple test for the
`PT_GNU_PROPERTY` segment. There is no reason to have it in
a separate file.
This patch:
1) Merges `program-headers.test` and `pt-gnu-property.test` to `gnu-phdrs.test`.
2) Removes 2 precompiled binaries used by `program-headers.test`
(other ones are still used by another tests).
Differential revision: https://reviews.llvm.org/D85526
Sjoerd Meijer [Tue, 11 Aug 2020 12:55:14 +0000 (13:55 +0100)]
[ARM][MVE] Added extra tail-predication runs for auto-correlation test case. NFC
Pavel Labath [Tue, 11 Aug 2020 13:20:59 +0000 (15:20 +0200)]
[lld] s/dyn_cast/isa
Fixes some unused variable warnings with gcc.
Pavel Labath [Tue, 28 Jul 2020 13:26:23 +0000 (15:26 +0200)]
[cmake] Make gtest macro definitions a part the library interface
These definitions are needed by any file which uses gtest. Previously we
were adding them in the add_unittest function, but over time we've
accumulated libraries (which don't go through add_unittest) building on
gtest and this has resulted in proliferation of the definitions.
Making this a part of the library interface enables them to be managed
centrally. This follows a patch for -Wno-suggest-override (D84554) which
took a similar approach.
Differential Revision: https://reviews.llvm.org/D84748
Simon Pilgrim [Tue, 11 Aug 2020 13:01:45 +0000 (14:01 +0100)]
[X86][SSE] Add tests for 256-bit HOP(SHUFFLE(X,Y),SHUFFLE(X,Y)) --> SHUFFLE(HOP(X,Y)) patterns
This should be addressed by combineHorizOpWithShuffle once we enable it for HADD/SUB
Simon Pilgrim [Tue, 11 Aug 2020 12:35:12 +0000 (13:35 +0100)]
[X86][SSE] Add tests for HOP(LOSUBVECTOR(SHUFFLE(X)),HISUBVECTOR(SHUFFLE(X))) patterns
This should be addressed by combineHorizOpWithShuffle once we enable it for HADD/SUB
Maximilian Fickert [Tue, 11 Aug 2020 12:54:46 +0000 (14:54 +0200)]
[clang-format] use spaces for alignment of binary/ternary expressions with UT_AlignWithSpaces
Use spaces to align binary and ternary expressions when using AlignOperands and UT_AlignWithSpaces.
This fixes an oversight in the new UT_AlignWithSpaces option (see D75034), which did not correctly identify the alignment of binary/ternary expressions.
Reviewed By: curdeius
Patch by: fickert
Differential Revision: https://reviews.llvm.org/D85600
Raphael Isemann [Tue, 11 Aug 2020 11:57:54 +0000 (13:57 +0200)]
[lldb] Add SubstTemplateTypeParm to RemoveWrappingTypes
Like the other type sugar removed by RemoveWrappingTypes, SubstTemplateTypeParm
is just pure sugar that should be ignored. If we don't ignore it (as we do now),
LLDB will fail to read values from record fields that have a
SubstTemplateTypeParm type.
Only way to produce such a type in LLDB is to either use the `import-std-module`
setting to get a template into the expression parser or just create your own
template directly in the expression parser which is what we do in the test.
Reviewed By: jarin
Differential Revision: https://reviews.llvm.org/D85132
Gongyu Deng [Tue, 11 Aug 2020 11:38:36 +0000 (13:38 +0200)]
[lldb] watchpoint ID common completion for commands `watchpoint delete/enable/disable/modify/ignore`
1. Added a common completion WatchPointIDs to complete with a list of the IDs of the current watchpoints;
2. Applied the completion to these commands: watchpoint delete/enable/disable/modify/ignore;
3. Added a correlated test case.
Reviewed By: teemperor
Differential Revision: https://reviews.llvm.org/D84104
Florian Hahn [Tue, 11 Aug 2020 12:05:04 +0000 (13:05 +0100)]
[SCEV] ] If RHS >= Start, simplify (Start smax RHS) to RHS for trip counts.
This is the max version of D85046.
This change causes binary changes in 44 out of 237 benchmarks (out of
MultiSource/SPEC2000/SPEC2006)
Reviewed By: lebedev.ri
Differential Revision: https://reviews.llvm.org/D85189
Georgii Rymar [Mon, 10 Aug 2020 12:30:08 +0000 (15:30 +0300)]
[LLD][ELF] - Update and fix gnu-ifunc* tests.
It turns that gnu-ifunc-plt-i386.s and gnu-ifunc-plt.s tests are broken.
Initially they were implemented in D27581 and tested that `IRELATIVE` relocations
are placed after other relocations in `.rel.plt`.
Later, we started to place `IRELATIVE` relocations to `.rela.dyn` (D65651).
Also, at some point `.plt` was renamed to `.iplt` (D71520).
Now, `gnu-ifunc*` tests mentioned do not test what they intended to test initially:
they should test that `IRELATIVE` relocations are placed after other ones in
`.rela.dyn`. Also, comments needs to be updated accordingly after changes performed.
This patch updates them.
Differential revision: https://reviews.llvm.org/D85642
Kerry McLaughlin [Tue, 11 Aug 2020 11:19:42 +0000 (12:19 +0100)]
[SVE][CodeGen] Legalisation of INSERT_VECTOR_ELT for scalable vectors
When the result type of insertelement needs to be split,
SplitVecRes_INSERT_VECTOR_ELT will try to store the vector to a
stack temporary, store the element at the location of the stack
temporary plus the index, and reload the Lo/Hi parts.
This patch does the following to ensure this works for scalable vectors:
- Sets the StackID with getStackIDForScalableVectors() in CreateStackTemporary
- Adds an IsScalable flag to getMemBasePlusOffset() and scales the
offset by VScale when this is true
- Ensures the immediate is clamped correctly by clampDynamicVectorIndex
so that we don't try to use an out of range index
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D84874
Simon Pilgrim [Tue, 11 Aug 2020 11:46:48 +0000 (12:46 +0100)]
Fix Wdocumentation unknown param warnings. NFC.
Simon Pilgrim [Tue, 11 Aug 2020 11:41:29 +0000 (12:41 +0100)]
Fix Wdocumentation unknown param warning. NFC.
Haojian Wu [Tue, 11 Aug 2020 11:49:11 +0000 (13:49 +0200)]
[AST][RecoveryExpr] Don't preserve the return type if the FunctionDecl is invalid.
If a functionDecl is invalid (e.g. return type cannot be formed), int is
use as he fallback type, which may lead to some bogus diagnostics.
Reviewed By: sammccall
Differential Revision: https://reviews.llvm.org/D85714
David Stenberg [Tue, 11 Aug 2020 08:54:52 +0000 (10:54 +0200)]
[DebugInfo] Allow GNU macro extension to be read
Allow the GNU .debug_macro extension to be parsed and printed by
llvm-dwarfdump. In an upcoming patch support will be added for emitting
that format also.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D82974
David Stenberg [Tue, 11 Aug 2020 08:53:38 +0000 (10:53 +0200)]
[DebugInfo] Introduce GNU macro extension entry encodings
This is a preparatory patch for allowing the GNU .debug_macro extension,
which is a precursor to the DWARF 5 format, to be emitted by LLVM for
earlier DWARF versions.
The entries share the same encoding and behavior as in DWARF5; there are
just more entries in the DWARF 5 format. Therefore, we could have used
those existing DWARF 5 entries, but I think that explicitly referring to
the GNU macro variants makes the code more clear.
The defines that this patch introduces can be found in GCC in the dwarf2.h header:
https://gcc.gnu.org/git/?p=gcc.git;a=blob;
f=include/dwarf2.h;
h=
0b6facfd4cf4c02320c7328114231b128ab42d5e;
hb=
dccbf1e2a6e544f71b4a5795f0c79015db019fc3#l425
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D82972
David Stenberg [Tue, 11 Aug 2020 07:28:32 +0000 (09:28 +0200)]
[DebugInfo] Refactor .debug_macro checks. NFCI
Move the Dwarf version checks that determine if the .debug_macro section
should be emitted, into a DwarfDebug member. This is a preparatory
refactoring for allowing the GNU .debug_macro extension, which is a
precursor to the DWARF 5 format, to be emitted by LLVM for earlier DWARF
versions.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D82971
Gongyu Deng [Tue, 11 Aug 2020 11:19:27 +0000 (13:19 +0200)]
[lldb] thread index common completion for commands like `thread select/step-over`
1. Added a common completion completing with a list of the threads of the current process;
2. Apply the common completion above to these commands: thread
continue/info/exception/select/step-in/step-inst/step-inst-over/step-out/step-over/step-script
3. Correlated test case test_common_completion_thread_index.
Reviewed By: teemperor
Differential Revision: https://reviews.llvm.org/D84088
Kirill Bobyrev [Tue, 11 Aug 2020 11:24:22 +0000 (13:24 +0200)]
[clangd] Enforce trailing slash for remote index's project root
Reviewed By: hokein
Differential Revision: https://reviews.llvm.org/D85711
Benjamin Kramer [Tue, 11 Aug 2020 11:20:39 +0000 (13:20 +0200)]
[AutoUpgrade] Simplify code
No need to set the name on an instruction that's going away, just move
it from the old instruction to the new one.
Kerry McLaughlin [Tue, 11 Aug 2020 10:38:08 +0000 (11:38 +0100)]
[CodeGen] Refactor getMemBasePlusOffset & getObjectPtrOffset to accept a TypeSize
Changes the Offset arguments to both functions from int64_t to TypeSize
& updates all uses of the functions to create the offset using TypeSize::Fixed()
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D85220
Gongyu Deng [Tue, 11 Aug 2020 11:13:49 +0000 (13:13 +0200)]
[lldb] stop-hook ID common completion for commands `target stop-hook enable/disable/delete'
1. Added a common completion StopHookIDs to provide completion with a list of stop hook ids;
2. Applied the common completion to commands: `target stop-hook delete/enable/disable';
3. Added an related test case.
Reviewed By: teemperor
Differential Revision: https://reviews.llvm.org/D84123
Benjamin Kramer [Tue, 11 Aug 2020 11:00:47 +0000 (13:00 +0200)]
[X86][FPEnv] Fix a use after free
Found by asan!
Raphael Isemann [Tue, 11 Aug 2020 10:47:42 +0000 (12:47 +0200)]
[lldb] Fix unhandled switch case for GOFF in GDBRemoteCommunicationClient
Just implementing the default case that emits an error to supress the compiler
warning.
Rainer Orth [Tue, 11 Aug 2020 10:46:34 +0000 (12:46 +0200)]
[compiler-rt][ubsan][test] Fix TypeCheck/misaligned.cpp on Sparc
Two ubsan tests FAIL on Sparc:
UBSan-Standalone-sparc :: TestCases/TypeCheck/misaligned.cpp
UBSan-Standalone-sparcv9 :: TestCases/TypeCheck/misaligned.cpp
I've reported the details in Bug 47015, but it boils down to the fact that
the `s1` subtest actually incurs a fault on strict-alignment targets like
Sparc which UBSan doesn't expect.
This can be fixed like the `w1` subtest by compiling with
`-fno-sanitize-recover=alignment`.
Tested on `sparcv9-sun-solaris2.11`, `amd64-pc-solaris2.11`, and
`x86_64-pc-linux-gnu`.
Differential Revision: https://reviews.llvm.org/D85433
Kazushi (Jam) Marukawa [Mon, 10 Aug 2020 08:16:57 +0000 (17:16 +0900)]
[VE] Update bit operations
Change bitreverse/bswap/ctlz/ctpop/cttz regression tests to support i128
and signext/zeroext i32 types. This patch also change the way to support
i32 types using 64 bits VE instructions.
Reviewed By: simoll
Differential Revision: https://reviews.llvm.org/D85712
Paul Walker [Fri, 7 Aug 2020 18:19:50 +0000 (19:19 +0100)]
[SVE] Add ISD nodes for predicated integer extend inreg operations.
These are useful instructions when lowering fixed length vector
extends, so I've broken this patch out as kind of NFC like work.
Differential Revision: https://reviews.llvm.org/D85546
Simon Pilgrim [Tue, 11 Aug 2020 10:38:26 +0000 (11:38 +0100)]
[X86] Rename combineVectorPackWithShuffle -> combineHorizOpWithShuffle. NFC.
The plan is to use this for (F)HADD/SUB opcodes as well as PACKs - similar to how we use combineShuffleWithHorizOp
Simon Pilgrim [Tue, 11 Aug 2020 10:36:21 +0000 (11:36 +0100)]
[clangd] RIFF.cpp - Use logical && instead of bitwise & for padding check
Fixes PR47070
Gongyu Deng [Tue, 11 Aug 2020 10:29:25 +0000 (12:29 +0200)]
[lldb] tab completion for `target modules load -u`
1. Added a common completion ModuleUUIDs to provide a list of the UUIDs of modules for completion;
2. Added a new enumeration item eArgTypeModuleUUID to CommandArgumentType which is set as the option argument type of OptionGroupUUID;
3. Applied the module UUID completion to the argument of the type eArgTypeModuleUUID in lldb/source/Interpreter/CommandObject.cpp;
4. Added an related test case in lldb/test/API/functionalities/completion/TestCompletion.py.
Paul Walker [Mon, 10 Aug 2020 17:02:53 +0000 (18:02 +0100)]
[SVE] Lower fixed length vector integer subtract operations.
Differential Revision: https://reviews.llvm.org/D85665
Gongyu Deng [Tue, 11 Aug 2020 10:20:43 +0000 (12:20 +0200)]
[lldb] move the frame index completion into a common completion and apply it to `thread backtrace -s`
Commands frame select and thread backtrace -s can be completed in the same way.
Moved the dedicated completion of frame select into a common completion and
apply it to the both commands, along with the test modified.
David Zarzycki [Tue, 11 Aug 2020 09:56:58 +0000 (05:56 -0400)]
Add missing `-o -` to a recent test
Caught with a build-system that remounts the source directory read-only.
Gongyu Deng [Tue, 11 Aug 2020 09:12:11 +0000 (11:12 +0200)]
[lldb] tab completion for `target modules search-paths insert`
Dedicated completion for the command `target modules search-paths insert` with a test case.
Reviewed By: JDevlieghere
Differential Revision: https://reviews.llvm.org/D83309
Kai Nacke [Tue, 11 Aug 2020 09:27:56 +0000 (05:27 -0400)]
[NFC] Fix typo in comment.
Twelvth -> Twelfth
Kai Nacke [Fri, 19 Jun 2020 11:43:35 +0000 (13:43 +0200)]
[SystemZ/ZOS] Add binary format goff and operating system zos to the triple
Adds the binary format goff and the operating system zos to the triple
class. goff is selected as default binary format if zos is choosen as
operating system. No further functionality is added.
Reviewers: efriedma, tahonermann, hubert.reinterpertcast, MaskRay
Reviewed By: efriedma, tahonermann, hubert.reinterpertcast
Differential Revision: https://reviews.llvm.org/D82081
Dávid Bolvanský [Tue, 11 Aug 2020 09:21:15 +0000 (11:21 +0200)]
[Docs] Fixed missing closing quote character
Florian Hahn [Tue, 11 Aug 2020 09:15:20 +0000 (11:15 +0200)]
[SLP] Make sure instructions are ordered when computing spill cost.
The entries in VectorizableTree are not necessarily ordered by their
position in basic blocks. Collect them and order them by dominance so
later instructions are guaranteed to be visited first. For instructions
in different basic blocks, we only scan to the beginning of the block,
so their order does not matter, as long as all instructions in a basic
block are grouped together. Using dominance ensures a deterministic order.
The modified test case contains an example where we compute a wrong
spill cost (2) without this patch, even though there is no call between
any instruction in the bundle.
This seems to have limited practical impact, .e.g on X86 with a recent
Intel Xeon CPU with -O3 -march=native -flto on MultiSource,SPEC2000,SPEC2006
there are no binary changes.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D82444
Gongyu Deng [Tue, 11 Aug 2020 08:32:55 +0000 (10:32 +0200)]
[lldb] type language common completion
1. Added a new common completion TypeLanguages to provide a list of supporting languages;
2. Bound the completion to eArgTypeLanguage;
3. Added a related test case.
Dávid Bolvanský [Tue, 11 Aug 2020 09:05:06 +0000 (11:05 +0200)]
[InstCombine] ~(~X + Y) -> X - Y
Proof:
https://alive2.llvm.org/ce/z/4xharr
Solves PR47051
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D85593
Florian Hahn [Tue, 11 Aug 2020 08:30:31 +0000 (09:30 +0100)]
[SCEVExpander] Add helper to clean up instrs inserted while expanding.
SCEVExpander already tracks which instructions have been inserted n
InsertedValues/InsertedPostIncValues. This patch adds an additional
vector to collect the instructions in insertion order. This can then be
used to remove exactly the instructions inserted by the expander.
This replaces ExpandedValuesCleaner, which in some cases might remove
values not inserted by the expander (e.g. if a value was dead before
insertion and is then used during expansion).
Reviewed By: lebedev.ri
Differential Revision: https://reviews.llvm.org/D84327
Gongyu Deng [Tue, 11 Aug 2020 08:10:08 +0000 (10:10 +0200)]
[lldb] tab completion for 'command delete/unalias'
Provided dedicated tab completions for `command delete/unalias`.
Reviewed By: teemperor
Differential Revision: https://reviews.llvm.org/D81128
Raphael Isemann [Tue, 11 Aug 2020 08:26:01 +0000 (10:26 +0200)]
[lldb][NFC] Fix warning in Thread::AutoCompleteThreadPlans
Gongyu Deng [Tue, 11 Aug 2020 08:02:51 +0000 (10:02 +0200)]
[lldb] tab completion for `thread plan discard`
Dedicated completion for the command `thread plan discard` with a corresponding
test case.
Reviewed By: teemperor
Differential Revision: https://reviews.llvm.org/D83234
Balázs Kéri [Tue, 11 Aug 2020 07:03:22 +0000 (09:03 +0200)]
[Analyzer] Improve invalid dereference bug reporting in DereferenceChecker.
Report undefined pointer dereference in similar way as null pointer dereference.
Reviewed By: NoQ
Differential Revision: https://reviews.llvm.org/D84520
Raphael Isemann [Tue, 11 Aug 2020 08:04:16 +0000 (10:04 +0200)]
[lldb][NFC] Remove unused custom reimplementation of realpath for Windows
No one is calling this function it seems and according to
https://bugs.llvm.org/show_bug.cgi?id=47088 this can leak memory, so let's just
remove it:
Quote from the bug report:
> Before return on line 146, the memory allocated on line 130 is not freed.
Reviewed By: amccarth
Differential Revision: https://reviews.llvm.org/D85633
Sam Parker [Fri, 7 Aug 2020 08:49:09 +0000 (09:49 +0100)]
[RDA] Fix DBG_VALUE issues
We skip debug instructions in RDA so we cannot attempt to look them
up in our instruction map without causing a crash. But some of the
methods select the last instruction in the block and this
instruction may be a debug instruction... So, use getLastNonDebugInstr
instead of calling back on a MachineBasicBlock.
MachineBasicBlock iterators have also been updated to use
instructionsWithoutDebug so we can avoid the manual checks for debug
instructions.
Differential Revision: https://reviews.llvm.org/D85658
Gongyu Deng [Tue, 11 Aug 2020 07:59:30 +0000 (09:59 +0200)]
[lldb] tab completion for `disassemble -F`
1.Added a new common completion DisassemblyFlavors;
2. Bound DisassemblyFlavors to argument of type eArgTypeDisassemblyFlavor in
CommandObject.cpp;
3. Added a related test case.
Gongyu Deng [Tue, 11 Aug 2020 07:50:28 +0000 (09:50 +0200)]
[lldb] tab completion for `watchpoint set variable`
1. Applied the common completion `eVariablePathCompletion` to command
`watchpoint set variable`;
2. Added a related test case.
Reviewed By: teemperor, JDevlieghere
Differential Revision: https://reviews.llvm.org/D84177
Kirill Bobyrev [Tue, 11 Aug 2020 07:48:54 +0000 (09:48 +0200)]
[clangd] Improve diagnostics in dexp interface
When running dexp in remote mode without --project-root it shuts down
with an assertion. This is not the desired behaviour: instruct user on
how to run it properly when the configuration is incorrect.
Richard Barton [Tue, 11 Aug 2020 07:41:57 +0000 (08:41 +0100)]
Revert "[flang] Version information in flang/f18"
Patch breaks out of tree builds and fails its regression test on buildbots.
This reverts commit
6e67a74ed73e60a9609f2666e9ac9f44c63c3e0f.
Juneyoung Lee [Tue, 11 Aug 2020 01:15:18 +0000 (10:15 +0900)]
[LazyValueInfo] Let getEdgeValueLocal look into freeze instructions
This patch makes getEdgeValueLocal more precise when a freeze instruction is
given, by adding support for freeze into constantFoldUser
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D84629
Eric Christopher [Tue, 11 Aug 2020 06:57:33 +0000 (23:57 -0700)]
Test requires a debug build to pass.
Shinji Okumura [Tue, 11 Aug 2020 06:52:02 +0000 (15:52 +0900)]
[Attributor][NFC] Connect AAPotentialValues with AAValueSimplify
This patch enables `AAValueSimplify` to use information from `AAPotentialValues`
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D85668
Craig Topper [Tue, 11 Aug 2020 06:13:40 +0000 (23:13 -0700)]
[X86] Custom match X86ISD::VPTERNLOG in X86ISelDAGToDAG in order to reduce isel patterns.
By factoring out the end of tryVPTERNLOG, we can use the same code
to directly match X86ISD::VPTERNLOG. This allows us to remove
around 3-4K worth of X86GenDAGISel.inc.