Nick Desaulniers [Tue, 11 Jan 2022 18:01:25 +0000 (10:01 -0800)]
[ShrinkWrap] check for PPC's non-callee-saved LR
As pointed out in https://reviews.llvm.org/D115688#inline-1108193, we
don't want to sink the save point past an INLINEASM_BR, otherwise
prologepilog may incorrectly sink a prolog past the MBB containing an
INLINEASM_BR and into the wrong MBB.
ShrinkWrap is getting this wrong because LR is not in the list of callee
saved registers. Specifically, ShrinkWrap::useOrDefCSROrFI calls
RegisterClassInfo::getLastCalleeSavedAlias which reads
CalleeSavedAliases which was populated by
RegisterClassInfo::runOnMachineFunction by iterating the list of
MCPhysReg returned from MachineRegisterInfo::getCalleeSavedRegs.
Because PPC's LR is non-allocatable, it's NOT considered callee saved.
Add an interface to TargetRegisterInfo for such a case and use it in
Shrinkwrap to ensure we don't sink a prolog past an INLINEASM or
INLINEASM_BR that clobbers LR.
Reviewed By: jyknight, efriedma, nemanjai, #powerpc
Differential Revision: https://reviews.llvm.org/D116424
Rob Suderman [Tue, 11 Jan 2022 17:49:36 +0000 (09:49 -0800)]
[mlir][tosa] Relax tosa.apply_scale operations
Apply scale may operate on vectors, scalars, or tensors during
tiling. Relax the requirements to avoid failures.
Reviewed By: NatashaKnk
Differential Revision: https://reviews.llvm.org/D116981
William S. Moses [Tue, 11 Jan 2022 03:27:14 +0000 (22:27 -0500)]
[MLIR][SCF] Simplify scf.if by swapping regions if condition is a not
Given an if of the form, simplify it by eliminating the not and swapping the regions
scf.if not(c) {
yield origTrue
} else {
yield origFalse
}
becomes
scf.if c {
yield origFalse
} else {
yield origTrue
}
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D116990
Fangrui Song [Tue, 11 Jan 2022 17:54:53 +0000 (09:54 -0800)]
[ELF] Add RelocationScanner. NFC
Currently the way some relocation-related static functions pass around
states is clumsy. Add a Resolver class to store some states as member
variables.
Advantages:
* Avoid the parameter `InputSectionBase &sec` (this offsets the cost passing around `this` paramemter)
* Avoid the parameter `end` (Mips and PowerPC hacks)
* `config` and `target` can be cached as member variables to reduce global state accesses. (potential speedup because the compiler didn't know `config`/`target` were not changed across function calls)
* If we ever want to reduce if-else costs (e.g. `config->emachine==EM_MIPS` for non-Mips) or introduce parallel relocation scan not handling some tricky arches (PPC/Mips), we can templatize Resolver
`target` isn't used as much as `config`, so I change it to a const reference
during the migration.
There is a minor performance inprovement for elf::scanRelocations.
Reviewed By: ikudrin, peter.smith
Differential Revision: https://reviews.llvm.org/D116881
Lei Zhang [Tue, 11 Jan 2022 17:04:39 +0000 (17:04 +0000)]
[mlir][linalg] Improve pooling op iterator order consistency
All named ops list iterators for accessing output first except
pooling ops. This commit made the pooling ops consistent with
the rest.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D115520
James Y Knight [Tue, 11 Jan 2022 17:42:49 +0000 (17:42 +0000)]
Florian Hahn [Tue, 11 Jan 2022 17:30:48 +0000 (17:30 +0000)]
[IRBuilder] Introduce folder using inst-simplify, use for Or fold.
Alternative to D116817.
This introduces a new value-based folding interface for Or (FoldOr),
which takes 2 values and returns an existing Value or a constant if the
Or can be simplified. Otherwise nullptr is returned. This replaces the
more restrictive CreateOr which takes 2 constants.
This is the used to implement a folder that uses InstructionSimplify.
The logic to simplify `Or` instructions is moved there. Subsequent
patches are going to transition other CreateXXX to the more general
FoldXXX interface.
Reviewed By: nikic, lebedev.ri
Differential Revision: https://reviews.llvm.org/D116935
ksyx [Wed, 5 Jan 2022 15:20:16 +0000 (10:20 -0500)]
[clang-format] Fix SeparateDefinitionBlocks issues
Fixes https://github.com/llvm/llvm-project/issues/52976.
- Make no formatting for macros
- Attach comment with definition headers
- Make no change on use of empty lines at block start/end
- Fix misrecognition of keyword namespace
Differential Revision: https://reviews.llvm.org/D116663
Reviewed By: MyDeveloperDay, HazardyKnusperkeks, curdeius
Philip Reames [Tue, 11 Jan 2022 17:23:33 +0000 (09:23 -0800)]
[instsimplify] Add a comment and test for a highly confusing case
LLVM GN Syncbot [Tue, 11 Jan 2022 17:12:28 +0000 (17:12 +0000)]
[gn build] Port
f77d115cc136
Matthias Braun [Tue, 28 Sep 2021 00:57:22 +0000 (17:57 -0700)]
X86InstrInfo: Support immediates that are +1/-1 different in optimizeCompareInstr
This is a re-commit of
e2c7ee0743592e39274e28dbe0d0c213ba342317 which
was reverted in
a2a58d91e82db38fbdf88cc317dcb3753d79d492 and
ea81cea8163a1a0e54df42103ee1c657bbf03791. This includes a fix to
consistently check for EFLAGS being live-out. See phabricator
review.
Original Summary:
This extends `optimizeCompareInstr` to re-use previous comparison
results if the previous comparison was with an immediate that was 1
bigger or smaller. Example:
CMP x, 13
...
CMP x, 12 ; can be removed if we change the SETg
SETg ... ; x > 12 changed to `SETge` (x >= 13) removing CMP
Motivation: This often happens because SelectionDAG canonicalization
tends to add/subtract 1 often when optimizing for fallthrough blocks.
Example for `x > C` the fallthrough optimization switches true/false
blocks with `!(x > C)` --> `x <= C` and canonicalization turns this into
`x < C + 1`.
Differential Revision: https://reviews.llvm.org/D110867
Craig Topper [Tue, 11 Jan 2022 16:55:45 +0000 (08:55 -0800)]
[RISCV] Add DAG combine to fold (fp_to_int (ffloor X)) -> (fcvt X, rdn)
Similar for ceil, trunc, round, and roundeven. This allows us to use
static rounding modes to avoid a libcall.
This optimization is done for AArch64 as isel patterns.
RISCV doesn't have instructions for ceil/floor/trunc/round/roundeven
so the operations don't stick around until isel to enable a pattern
match. Thus I've implemented a DAG combine.
We only handle XLen types except i32 on RV64. i32 will be type
legalized to a RISCVISD node. All other types will be type legalized
to XLen and maintain the FP_TO_SINT/UINT ISD opcode.
Reviewed By: asb
Differential Revision: https://reviews.llvm.org/D116771
Sven van Haastregt [Tue, 11 Jan 2022 16:54:19 +0000 (16:54 +0000)]
[SPIR-V] Drop double quote from test pattern
When spirv-link is found, it won't match a leading `"`. This fixes
the test added by commit
dbb8d086377b ("[SPIR-V] Add linking using
spirv-link.", 2022-01-11).
Jan Svoboda [Tue, 11 Jan 2022 16:17:07 +0000 (17:17 +0100)]
[clang] Move `ApplyHeaderSearchOptions` from Frontend to Lex
In D116750, the `clangFrontend` library was added as a dependency of `LexTests` in order to make `clang::ApplyHeaderSearchOptions()` available. This increased the number of TUs the test depends on.
This patch moves the function into `clangLex` and removes dependency of `LexTests` on `clangFrontend`.
Reviewed By: thakis
Differential Revision: https://reviews.llvm.org/D117024
Siva Chandra Reddy [Tue, 11 Jan 2022 05:24:57 +0000 (05:24 +0000)]
[libc][NFC] Move sys/mman entrypoints to the default build configs.
Specifically, mmap and munmap have been moved to the default build list
of entrypoints. To support this, certain deps and includes have been
adjusted. The use of errno in some cases has been updated.
Christian Sigg [Tue, 11 Jan 2022 12:13:17 +0000 (13:13 +0100)]
Mark arith.minf, arith.maxf as commutative.
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D117010
Philip Reames [Tue, 11 Jan 2022 16:40:03 +0000 (08:40 -0800)]
[GlobalsModRef] Apply indirect-global rule to all globals initialized from noalias calls
Extend the existing malloc-family specific optimization to all noalias calls. This allows us to handle allocation wrappers, and removes a dependency on a lib-func check in favor of generic attribute usage.
Differential Revision: https://reviews.llvm.org/D116980
Simon Pilgrim [Tue, 11 Jan 2022 16:40:24 +0000 (16:40 +0000)]
[X86] Apply clang-format to X86TargetLowering::isVectorShiftByScalarCheap
Fix indentation
Philip Reames [Tue, 11 Jan 2022 16:37:41 +0000 (08:37 -0800)]
[DSE] Style improvements after 3cef3cf - remove redundant dyn_casts [NFC]
I'd been working on exactly the same patch when Nikita landed his, so this patch is basically the style diff between the two. :)
Dimitry Andric [Tue, 11 Jan 2022 16:30:19 +0000 (17:30 +0100)]
[Nomination] Adding Intel representatives to security group
We would like to nominate Andy Kaylor and Sergey Maslov to join the LLVM security group as a representative of Intel. Both are members of the Intel compiler team, and would like to register as vendor contacts. Intel packages and distributes LLVM-based toolchains as part of our compiler products. As such, we would like to be aware of any security vulnerability found in the compiler, and would like to contribute to the resolution of such issues.
Please let us know if anything is missing from the nomination.
Reviewed By: apilipenko, dim, george.burgess.iv, kristof.beyls, mattdr, nikhgupt, probinson, peter.smith, pietroalbini, steveklabnik
Differential Revision: https://reviews.llvm.org/D115657
Florian Hahn [Tue, 11 Jan 2022 16:11:22 +0000 (16:11 +0000)]
[InstSimplify] Fold inbounds GEP to poison if base is undef.
D92270 updated constant expression folding to fold inbounds GEP to
poison if the base is undef. Apply the same logic to SimplifyGEPInst.
The justification is that we can choose an out-of-bounds pointer as base
pointer.
Reviewed By: nikic, lebedev.ri
Differential Revision: https://reviews.llvm.org/D117015
Simon Atanasyan [Tue, 11 Jan 2022 16:06:40 +0000 (19:06 +0300)]
[mips][lld] Add test case to check symbol index reading on mips64el. NFC
Simon Atanasyan [Tue, 11 Jan 2022 05:56:26 +0000 (08:56 +0300)]
[mips] Use `push_back` to insert element at the end of a container. NFC
Simon Pilgrim [Tue, 11 Jan 2022 15:11:06 +0000 (15:11 +0000)]
[X86] Tag existing shuffle test case as PR53124
The poor AVX1 codegen matches the core issue raised in PR53124
Roman Lebedev [Tue, 11 Jan 2022 15:23:02 +0000 (18:23 +0300)]
[SCEV] `getSequentialMinMaxExpr()`: look into `umin` when deduplicating operands
We could just merge all umin into umin_seq, but that is likely
a pessimization, so don't do that, but pretend that we did
for the purpose of deduplication.
Roman Lebedev [Tue, 11 Jan 2022 15:25:53 +0000 (18:25 +0300)]
[NFC][SCEV] More tests with operand-wise redundant operands of umin of umin_seq
Alexandre Ganea [Tue, 11 Jan 2022 15:36:46 +0000 (10:36 -0500)]
[compiler-rt] Silence warnings when building with MSVC
Differential Revision: https://reviews.llvm.org/D116872
Louis Dionne [Mon, 10 Jan 2022 22:14:01 +0000 (17:14 -0500)]
[libc++] Use TEST_HAS_NO_UNICODE instead of _LIBCPP_HAS_NO_UNICODE in the test suite
Differential Revision: https://reviews.llvm.org/D116973
Kirill Stoimenov [Wed, 5 Jan 2022 17:14:40 +0000 (17:14 +0000)]
[ASan] Driver changes to always link-in asan_static library.
This enables the changes from D116182.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D116670
Nikita Popov [Tue, 11 Jan 2022 15:00:13 +0000 (16:00 +0100)]
[GlobalStatus] Look through non-constexpr casts
analyzeGlobal() looks through non-constexpr cast instructions when
looking for users. However, this particular place only strips the
casts again if they are constexprs. We should be looking through all
casts here.
Roman Lebedev [Tue, 11 Jan 2022 14:33:15 +0000 (17:33 +0300)]
[NFC][SCEV] Add more tests for umin_seq with redundant operands
Nico Weber [Tue, 11 Jan 2022 14:47:28 +0000 (09:47 -0500)]
[gn build] (manually) port
8503c688d555
Nikita Popov [Tue, 11 Jan 2022 14:34:16 +0000 (15:34 +0100)]
[GlobalOpt] Regenerate test checks (NFC)
Jan Svoboda [Tue, 11 Jan 2022 13:47:03 +0000 (14:47 +0100)]
[clang][lex] Keep references to `DirectoryLookup` objects up-to-date
The elements of `SearchPath::SearchDirs` are being referenced to by their indices. This proved to be error-prone: `HeaderSearch::SearchDirToHSEntry` was accidentally not being updated in `HeaderSearch::AddSearchPath()`. This patch fixes that by referencing `SearchPath::SearchDirs` elements by their address instead, which is stable thanks to the bump-ptr-allocation strategy.
Reviewed By: ahoppen
Differential Revision: https://reviews.llvm.org/D116750
Benjamin Kramer [Tue, 11 Jan 2022 14:05:45 +0000 (15:05 +0100)]
[mlir][linalg] Use cast instead of dyn_cast that's always dereferenced
This turns a random nullptr deref into an assertion failure in case
`tensor::registerInferTypeOpInterfaceExternalModels` isn't called.
Roman Lebedev [Tue, 11 Jan 2022 13:45:22 +0000 (16:45 +0300)]
[SCEV] `getSequentialMinMaxExpr()`: keep only the first instance of an operand
Having the same operand more than once doesn't change the outcome here,
neither reduction-wise nor poison-wise.
We must keep the first instance specifically though.
Roman Lebedev [Tue, 11 Jan 2022 13:24:29 +0000 (16:24 +0300)]
[SCEV] Add test for umin_seq with duplicate operands
Anastasia Stulova [Tue, 11 Jan 2022 13:45:33 +0000 (13:45 +0000)]
[SPIR-V] Remove unused variable
Florian Hahn [Tue, 11 Jan 2022 13:43:53 +0000 (13:43 +0000)]
[InstSimplify] Add additional GEP tests with undef bases.
David Spickett [Wed, 3 Nov 2021 12:10:40 +0000 (12:10 +0000)]
[lldb] Remove non address bits from memory read arguments
Addresses on AArch64 can have top byte tags, memory tags and pointer
authentication signatures in the upper bits.
While testing memory tagging I found that memory read couldn't
read a range if the two addresses had different tags. The same
could apply to signed pointers given the right circumstance.
(lldb) memory read mte_buf_alt_tag mte_buf+16
error: end address (0x900fffff7ff8010) must be greater than the start
address (0xa00fffff7ff8000).
Or it would try to read a lot more memory than expected.
(lldb) memory read mte_buf mte_buf_alt_tag+16
error: Normally, 'memory read' will not read over 1024 bytes of data.
error: Please use --force to override this restriction just once.
error: or set target.max-memory-read-size if you will often need a
larger limit.
Fix this by removing non address bits before we calculate the read
range. A test is added for AArch64 Linux that confirms this by using
the top byte ignore feature.
This means that if you do read with a tagged pointer the output
does not include those tags. This is potentially confusing but I think
overall it's better that we don't pretend that we're reading memory
from a range that the process is unable to map.
(lldb) p ptr1
(char *) $4 = 0x3400fffffffff140 "\x80\xf1\xff\xff\xff\xff"
(lldb) p ptr2
(char *) $5 = 0x5600fffffffff140 "\x80\xf1\xff\xff\xff\xff"
(lldb) memory read ptr1 ptr2+16
0xfffffffff140: 80 f1 ff ff ff ff 00 00 38 70 bc f7 ff ff 00 00 ........8p......
Reviewed By: omjavaid, danielkiss
Differential Revision: https://reviews.llvm.org/D103626
Anastasia Stulova [Tue, 11 Jan 2022 11:21:30 +0000 (11:21 +0000)]
[SPIR-V] Add linking using spirv-link.
Add support of linking files compiled into SPIR-V objects
using spirv-link.
Command line inteface examples:
clang --target=spirv64 test1.cl test2.cl
clang --target=spirv64 test1.cl -o test1.o
clang --target=spirv64 test1.o test2.cl -o test_app.out
This works independently from the SPIR-V generation method
(via an external tool or an internal backend) and applies
to either approach that is being used.
Differential Revision: https://reviews.llvm.org/D116266
Pavel Labath [Mon, 10 Jan 2022 15:34:39 +0000 (16:34 +0100)]
[lldb/qemu] Implement GetMmapArgumentList
By forwarding it to the host platform.
Roman Lebedev [Tue, 11 Jan 2022 12:51:43 +0000 (15:51 +0300)]
[SCEV] Reenable umin_seq support and fix the `computeSCEVAtScope()`
This reverts commit
f62f47f5e1f641b41d3b7d593c058ebec2883534.
Roman Lebedev [Tue, 11 Jan 2022 12:33:53 +0000 (15:33 +0300)]
[NFC][SCEV] Add reproducers for umin_seq crashes
As reported in https://reviews.llvm.org/D116766#3233042
Florian Hahn [Tue, 11 Jan 2022 12:35:55 +0000 (12:35 +0000)]
[LSR] Use pointer args instead of undef for uglygep*.ll tests.
Make the test more robust by replacing undef by pointer arguments. This
ensures that the GEPs cannot be folded to undef.
David Green [Tue, 11 Jan 2022 12:33:53 +0000 (12:33 +0000)]
Revert "[Clang][AArch64][ARM] PMUv3.4 Option Added"
It turns out this is conflating a few different PMU extensions. And on
Arm ended up breaking M-Profile code generation. Reverting for the
moment whilst we sort out the details.
This reverts commit
d17fb46e894501568a1bf3b11a5d920817444630.
Egor Zhdan [Wed, 29 Dec 2021 22:02:21 +0000 (22:02 +0000)]
[Clang][Sema] Fix attribute mismatch warning for ObjC class properties
If a class declares an instance property, and an inheritor class declares a class property with the same name, Clang Sema currently treats the latter as an overridden property, and compares the attributes of the two properties to check for a mismatch. The resulting diagnostics might be misleading, since neither of the properties actually overrides the another one.
rdar://
86018435
Differential Revision: https://reviews.llvm.org/D116412
Nikita Popov [Tue, 11 Jan 2022 12:01:28 +0000 (13:01 +0100)]
[CodeGen] Avoid deprecated Address constructor
David Sherwood [Thu, 16 Dec 2021 08:57:18 +0000 (08:57 +0000)]
[SVE][CodeGen] Use splice instruction when lowering VECTOR_SPLICE
For certain negative indices passed to the VECTOR_SPLICE operation
we can actually directly use the SVE splice instruction by creating
the appropriate predicate. The predicate needs to be constructed in
such a way that all but the last -idx elements are false. We can do
this efficiently using a combination of 'ptrue' (with the appropriate
fixed pattern, e.g. vl1, vl2, etc.) and 'rev'. The advantage of using
these instructions to generate the predicate is they do not set any
flags, unlike the whilelo instruction. This is critical when the splice
operation is in a loop, since we want MachineLICM to hoist the
predicate generation out of the loop.
Differential Revision: https://reviews.llvm.org/D115863
Tim Northover [Fri, 7 Jan 2022 13:49:18 +0000 (13:49 +0000)]
ARM: make FastISel & GISel pass -1 to ADJCALLSTACKUP to signal no callee pop.
The interface for these instructions changed with support for mandatory tail
calls, and now -1 indicates the CalleePopAmount argument is not valid.
Unfortunately I didn't realise FastISel or GISel did calls at the time so
didn't update them.
Simon Pilgrim [Tue, 11 Jan 2022 11:12:12 +0000 (11:12 +0000)]
[SemaTemplateInstantiate] Use cast<> instead of dyn_cast<> to avoid dereference of nullptr
The pointer is always dereferenced immediately below, so assert the cast is correct instead of returning nullptr
Nikita Popov [Tue, 11 Jan 2022 11:27:23 +0000 (12:27 +0100)]
[MemoryBuiltins] Remove unused isOpNewLikeFn() (NFC)
This function is no longer used since
2cafbcb560d9e6e2300941d088e754b01d56595b.
Nikita Popov [Tue, 11 Jan 2022 11:25:39 +0000 (12:25 +0100)]
[MemoryBuiltins] Remove unused isStrdupLikeFn() function (NFC)
This function is no longer used after
dcbc91f40c2e6ff578667020f7c6a05c25149865.
Nikita Popov [Tue, 11 Jan 2022 11:08:44 +0000 (12:08 +0100)]
[DSE] Check for noalias calls rather than alloc functions
For these "visible on unwind/ret" checks we only care about the
fact that no other code has access to the pointer (unless it
escapes). A noalias call is sufficient for this, it does not
have to be a known allocation function.
This is basically the same change as D116728, but for DSE rather
than LICM.
Florian Hahn [Tue, 11 Jan 2022 11:18:28 +0000 (11:18 +0000)]
[LSR] Remove duplicated test address-space-loop.ll.
llvm/test/Transforms/LoopStrengthReduce/uglygep-address-space.ll has
exactly the same checks and input. Remove the duplicated test.
Matthias Springer [Tue, 11 Jan 2022 11:04:55 +0000 (20:04 +0900)]
[mlir][linalg][bufferize] Fix CallOp bufferization
Previously, CallOps did not have any aliasing OpResult/OpOperand pairs. Therefore, CallOps were mostly ignored by the analysis and buffer copies were not inserted when necessary.
This commit introduces the following changes:
* Function bbArgs writable by default. A function can now be bufferized without inspecting its callers.
* Callers must introduce buffer copies of function arguments when necessary. If a function is external, the caller must conservatively assume that a function argument is modified by the callee after bufferization. If the function is not external, the caller inspects the callee to determine if a function argument is modified.
Differential Revision: https://reviews.llvm.org/D116457
Nikita Popov [Tue, 11 Jan 2022 10:55:59 +0000 (11:55 +0100)]
[DSE] Add additional tests for noalias calls (NFC)
Currently this is special-cased to TLI alloc functions only.
Haojian Wu [Mon, 10 Jan 2022 14:19:05 +0000 (15:19 +0100)]
Reland "[AST] Add RParen loc for decltype AutoTypeloc."
Reland 55d96ac and 37ec65e with a clang-tidy fix.
Hans Wennborg [Mon, 10 Jan 2022 18:45:13 +0000 (19:45 +0100)]
[ADT] Add an in-place version of toHex()
and use that to simplify MD5's hex string code which was previously
using a string stream, as well as Clang's
CGDebugInfo::computeChecksum().
Differential revision: https://reviews.llvm.org/D116960
Nikita Popov [Tue, 11 Jan 2022 10:49:08 +0000 (11:49 +0100)]
[DSE] Make test more robust (NFC)
If the allocation is not captured, then all the stores before the
ret are dead anyway.
Hans Wennborg [Tue, 11 Jan 2022 09:08:25 +0000 (10:08 +0100)]
[ADT] Use a lookup table in hexdigit() and call that from toHex()
A lookup table, which toHex() was using, seems like the better approach.
Having two implementations is redundant, so put the lookup table in
hexdigit() and make toHex() call that.
Differential revision: https://reviews.llvm.org/D116960
Simon Pilgrim [Mon, 10 Jan 2022 17:44:18 +0000 (17:44 +0000)]
[SemaOverload] compareConversionFunctions - use castAs<> instead of getAs<> to avoid dereference of nullptr
The pointer is dereferenced immediately below, so assert the cast is correct instead of returning nullptr
Simon Pilgrim [Mon, 10 Jan 2022 17:31:35 +0000 (17:31 +0000)]
[SemaOverload] Use castAs<> instead of getAs<> to avoid dereference of nullptr
The pointer is always dereferenced inside BuildSimilarlyQualifiedPointerType, so assert the cast is correct instead of returning nullptr
Martin Storsjö [Tue, 11 Jan 2022 10:22:42 +0000 (12:22 +0200)]
[clang] [test] Fix clang-cl unused argument tests on paths that start with /U
This reinstates a test that was temporarily removed in
e26bbae30218a35d76a79fe90b0e41dd0f71b779, in a form that works on
Darwin.
Use -LD instead of -link as a linker argument that is unused when
compiling, that produces warnings normally. -LD can be placed anywhere
in the command line, so that the command line ends with "-- %s", making
paths starting with /U correctly interpreted as paths, not options.
Nikita Popov [Tue, 11 Jan 2022 09:55:50 +0000 (10:55 +0100)]
[LICM] Regenerate test checks (NFC)
Julian Gross [Tue, 11 Jan 2022 10:04:15 +0000 (11:04 +0100)]
[MLIR] Update allocs to memref.allocs in documentation.
Changed the remaining appearances of alloc to memref.alloc in several
documentation sections, since they lead to misunderstandings, if they
are used.
Differential Revision: https://reviews.llvm.org/D116999
wangpc [Tue, 11 Jan 2022 10:19:05 +0000 (18:19 +0800)]
[RISCV] Generate 32 bits jumptable entries when code model is small
The code can only address the whole RV32 address space or the lower 2 GiB
of the RV64 address space in small code model, so 32 bits entry is enough.
Cache hit ratio and code size have some improvements.
Reviewed By: asb
Differential Revision: https://reviews.llvm.org/D116435
Diana Picus [Wed, 22 Dec 2021 09:06:37 +0000 (09:06 +0000)]
[flang] Add tests for converting arrays and refs to arrays. NFC
Cover more of the code paths from LLVMTypeConverter::convertPointerLike
and LLVMTypeConverter::convertSequenceType.
Differential Revision: https://reviews.llvm.org/D116927
Sam McCall [Tue, 11 Jan 2022 10:01:02 +0000 (11:01 +0100)]
[clangd] Save more getFileID in Selection
This saves about 10% of SelectionVisitor::pop().
Florian Hahn [Tue, 11 Jan 2022 09:40:21 +0000 (09:40 +0000)]
[SCEVExpander] Use IntToPtr for temporary instruction.
Use PtrToInt instead Add when creating temporary instructions. The add
might get folded away with more sophisticated folding.
David Sherwood [Fri, 17 Dec 2021 09:39:21 +0000 (09:39 +0000)]
[IR] Change vector.splice intrinsic to reject out-of-bounds indices
I've changed the definition of the experimental.vector.splice
instrinsic to reject indices that are known to be or possibly
out-of-bounds. In practice, this means changing the definition so that
the index is now only valid in the range [-VL, VL-1] where VL is the
known minimum vector length. We use the vscale_range attribute to
take the minimum vscale value into account so that we can permit
more indices when the attribute is present.
The splice intrinsic is currently only ever generated by the vectoriser,
which will never attempt to splice vectors with out-of-bounds values.
Changing the definition also makes things simpler for codegen since we
can always assume that the index is valid.
This patch was created in response to review comments on D115863
Differential Revision: https://reviews.llvm.org/D115933
Sam McCall [Tue, 11 Jan 2022 09:15:12 +0000 (10:15 +0100)]
[clangd] Small optimization in SelectionTree
This seems to be strictly faster in all cases. Before fixing D116978 it
was one of the hot paths, and may become one again.
Martin Storsjö [Tue, 11 Jan 2022 09:06:29 +0000 (11:06 +0200)]
[clang] [test] Remove newly added tests that fail on Darwin
These tests were recently added in
50ec1306d060e46e0d53c9f5d8a052e1b0d10d3b. The clang-cl invocations
interpret the source path, %s, which begins with /Users, as a cl
option /U. Normally this is worked around by passing -- before
the arguments that must be interpreted as files, not options, but
in the current test, the -link option must be passed last, after any
file names.
Florian Hahn [Tue, 11 Jan 2022 08:37:29 +0000 (08:37 +0000)]
[LoopVersioning] Check the full generated runtime checks in lcssa.ll.
This makes improvements in follow-up patches more obvious.
Diego Caballero [Tue, 11 Jan 2022 08:30:33 +0000 (08:30 +0000)]
[mlir][GPU] Fix attribute name of DL specification
D115722 added a DL spec to GPU modules. It happens that the DL
default interface implementation is sensitive to the name of the
DL spec attribute. This patch is fixing the name of the attribute
to be the expected one.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D116956
Nikita Popov [Tue, 4 Jan 2022 14:24:14 +0000 (15:24 +0100)]
[CodeGen] Make element type in emitArrayDestroy() predictable
When calling emitArrayDestroy(), the pointer will usually have
ConvertTypeForMem(EltType) as the element type, as one would expect.
However, globals with initializers sometimes don't use the same
types as values normally would, e.g. here the global uses
{ double, i32 } rather than %struct.T as element type.
Add an early cast to the global destruction path to avoid this
special case. The cast would happen lateron anyway, it only gets
moved to an earlier point.
Differential Revision: https://reviews.llvm.org/D116219
Zi Xuan Wu [Tue, 11 Jan 2022 08:13:21 +0000 (16:13 +0800)]
[CSKY] Add CSKYConstantIslands Pass to lift or duplicate constant pool entry
Loading constants inline is expensive on CSKY and it's in general better
to place the constant nearby in code space and then it can be loaded with a
simple 16/32 bit load instruction like lrw.
It needs lift or duplicates constant pool entry to make constant nearby so that lrw instruction can reach.
wangpc [Tue, 11 Jan 2022 07:52:00 +0000 (15:52 +0800)]
[RISCV] Override TargetLowering::BuildSDIVPow2 to generate SELECT
When `Zbt` is enabled, we can generate SELECT for division by power
of 2, so that there is no data dependency.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D114856
Martin Storsjö [Sat, 1 Jan 2022 13:54:29 +0000 (15:54 +0200)]
[clang] Add --start-no-unused-arguments/--end-no-unused-arguments to silence some unused argument warnings
When passing a set of flags to configure defaults for a specific
target (similar to the cmake settings `CLANG_DEFAULT_RTLIB`,
`CLANG_DEFAULT_UNWINDLIB`, `CLANG_DEFAULT_CXX_STDLIB` and
`CLANG_DEFAULT_LINKER`, but without hardcoding them in the binary),
some of the flags may cause warnings (e.g. `-stdlib=` when compiling C
code). Allow requesting selectively ignoring unused arguments among
some of the arguments on the command line, without needing to resort
to `-Qunused-arguments` or `-Wno-unused-command-line-argument`.
Fix up the existing diagnostics.c testcase. It was added in
response to PR12181 to fix handling of
`-Werror=unused-command-line-argument`, but the command line option
in the test (`-fzyzzybalubah`) now triggers "error: unknown argument"
instead of the intended warning. Change it into a linker input
(`-lfoo`) which triggers the intended diagnostic. Extend the
existing test case to check more cases and make sure that it keeps
testing the intended case.
Add testing of the new option to this existing test.
Differential Revision: https://reviews.llvm.org/D116503
LLVM GN Syncbot [Tue, 11 Jan 2022 06:50:34 +0000 (06:50 +0000)]
[gn build] Port
8d23b7420c92
Konstantin Varlamov [Tue, 11 Jan 2022 06:49:37 +0000 (22:49 -0800)]
[libc++][ranges] Implement `uninitialized_copy{,_n}` and `uninitialized_move{,_n}`.
Also implement `in_out_result` which is a prerequisite.
Differential Revision: https://reviews.llvm.org/D116023
Esme-Yi [Tue, 11 Jan 2022 06:17:16 +0000 (06:17 +0000)]
[NFC][XCOFF][llvm-readobj] replace binaries with YAMLs (only tests for Symbols).
Summary: Since yaml2obj now supports converting AuxSymbols, we can use YAMLs as input for symbol-related tests of llvm-readobj instead of binaries.
Reviewed By: jhenderson, shchenz
Differential Revision: https://reviews.llvm.org/D114434
John Regehr [Tue, 11 Jan 2022 05:24:23 +0000 (22:24 -0700)]
allow llvm-reduce, if asked, to run its set of passes more than once, taking longer to finish but also potentially resulting in a smaller reduced file.
Fangrui Song [Tue, 11 Jan 2022 04:16:02 +0000 (20:16 -0800)]
[ELF] Move OffsetGetter before some static functions. NFC
to prepare for D116881.
Fangrui Song [Tue, 11 Jan 2022 03:39:14 +0000 (19:39 -0800)]
[lld-macho] Change some global pointers to unique_ptr
Similar to D116143. My x86-64 `lld` is ~8KiB smaller.
Reviewed By: keith
Differential Revision: https://reviews.llvm.org/D116902
Fangrui Song [Tue, 11 Jan 2022 03:36:24 +0000 (19:36 -0800)]
[lld-macho][test] Add missing coverage for archive/dylib resolution after D115092
When `file->fetch(sym)` is replaced with a no-op, no test fails.
The new test catches the case.
Reviewed By: #lld-macho, oontvoo
Differential Revision: https://reviews.llvm.org/D116916
Chenbing.Zheng [Tue, 11 Jan 2022 03:00:09 +0000 (03:00 +0000)]
[RISCV] Block vmsgeu.vi with 0 immediate in Isel
For vmsgeu.vi with 0, we know this is always true. So we can replace
it with vmset.m (unmasked) or vmset.m+vmand.mm (masked).
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D116584
jacquesguan [Fri, 7 Jan 2022 08:35:23 +0000 (16:35 +0800)]
[RISCV] Select vl op to X0 when it is equal to ~0.
Now the backend will select ~0 vl to a register and load instruction, we could use X0 to replace it.
Differential Revision: https://reviews.llvm.org/D116798
jacquesguan [Fri, 7 Jan 2022 08:53:02 +0000 (16:53 +0800)]
[RISCV] Add precommit test for select vl op that equal to ~0.
Precommit test for D116798
Differential Revision: https://reviews.llvm.org/D116799
Jim Lin [Tue, 11 Jan 2022 02:16:01 +0000 (10:16 +0800)]
[M68k][test][NFC] Reorder the tests for MxBTST_MI and MxBTST_MR
Follow the order of definitions in M68kInstrBits.td
Haocong.Lu [Tue, 11 Jan 2022 02:32:25 +0000 (02:32 +0000)]
[RISCV] Use shift for zero extension when Zbb and Zbp are not enabled
Now AND is used for zero extension when both Zbb and Zbp are not enabled.
It may be better to use shift operation if the trailing ones mask exceeds simm12.
This patch optimzes LUI+ADDI+AND to SLLI+SRLI.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D116720
Nick Desaulniers [Tue, 11 Jan 2022 02:34:30 +0000 (18:34 -0800)]
git-clang-format HEAD~
Nick Desaulniers [Tue, 11 Jan 2022 02:31:29 +0000 (18:31 -0800)]
[TargetLowering] precommit refactor from D115688 NFC
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
jacquesguan [Sat, 25 Dec 2021 10:09:27 +0000 (18:09 +0800)]
[RISCV] Use vmv.s.x to build one element splat vector.
When we want to create an splat vector that only the first element is initialized, we could use vmv.s.x or vfmv.s.f to build it.
Differential Revision: https://reviews.llvm.org/D116277
Philip Reames [Tue, 11 Jan 2022 02:17:34 +0000 (18:17 -0800)]
Partial revert of 82fb4f4
Two crashes have been reported. This change disables the new logic while leaving the new node in tree. Hopefully, that's enough to allow investigation without breakage while avoiding massive churn.
Philip Reames [Tue, 11 Jan 2022 00:15:39 +0000 (16:15 -0800)]
Delete a stale comment
Chuanqi Xu [Tue, 11 Jan 2022 02:15:46 +0000 (10:15 +0800)]
[AST] lookup in parent DeclContext for transparent DeclContext
The compiler would crash if we lookup for name in transparent decl
context. See the tests attached for example.
I think this should make sense since the member declared in transparent
DeclContext are semantically defined in the enclosing (non-transparent)
DeclContext, this is the definition for transparent DeclContext.
Reviewed By: erichkeane
Differential Revision: https://reviews.llvm.org/D116792
Craig Topper [Tue, 11 Jan 2022 02:07:12 +0000 (18:07 -0800)]
[SelectionDAG] Add additional documentation to FP_TO_S/UINT_SAT to justify D116870. NFC
As noted in post-commit review, the value of the bits between
the saturating type and result type were not defined defined. Define
them to be sign extended for FP_TO_SINT_SAT and zero extended for
FP_TO_UINT_SAT.
Chuanqi Xu [Tue, 11 Jan 2022 01:54:57 +0000 (09:54 +0800)]
[AST] Don't consider 'ExportDecl' when calculating DeclContext 'Encloses'
This mimics the style of
90010c2e1 (Don't consider 'LinkageSpec' when
calculating DeclContext 'Encloses'). Since ExportDecl and LinkageSpec
are transparent DeclContext, they share some similarity.
Reviewed By: erichkeane
Differential Revision: https://reviews.llvm.org/D116911
Mehdi Amini [Tue, 11 Jan 2022 01:26:44 +0000 (01:26 +0000)]
Add a `qualified` directive to the Op, Attribute, and Type declarative assembly format
This patch introduces a new directive that allow to parse/print attributes and types fully
qualified.
This is a follow-up to
ee0908703d29 which introduces the eliding of the `!dialect.mnemonic` by default and allows to force to fully qualify each type/attribute
individually.
Differential Revision: https://reviews.llvm.org/D116905