[LV] Mark non-consecutive-like pointers non-uniform

If a memory instruction will be vectorized, but it's pointer operand is
non-consecutive-like, the instruction is a gather or scatter operation. Its
pointer operand will be non-uniform. This should fix PR31671.

Reference: https://llvm.org/bugs/show_bug.cgi?id=31671
Differential Revision: https://reviews.llvm.org/D28819

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292254 91177308-0d34-0410-b5e6-96231b3b80d8

[WebAssembly] Add triple support for the new wasm object format

Differential Revision: https://reviews.llvm.org/D26701

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292252 91177308-0d34-0410-b5e6-96231b3b80d8

Rename scalar_promote.ll to scalar-promote.ll and scalar_promote-unwind.ll to scalar-promote-unwind.ll. NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292251 91177308-0d34-0410-b5e6-96231b3b80d8

Refactor out LoopInfo computation so that it can be used by
other test cases.

Summary: Refactor out LoopInfo computation so that it can be
used by other test cases.

So i am changing this test proactively for later commit, which will use
this function.

Reviewers: sanjoy, hfinkel

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D28778

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292250 91177308-0d34-0410-b5e6-96231b3b80d8

[EarlyCSE] Don't DSE across readnone functions that may throw

Summary: Depends on D28740

Reviewers: dberlin, chandlerc, hfinkel, majnemer

Subscribers: mcrosier, llvm-commits

Differential Revision: https://reviews.llvm.org/D28741

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292249 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] add tests for shl nsw + icmp sle; NFC

We want to handle these cases similarly to icmp sgt, so add the tests for it.
See: https://reviews.llvm.org/D28406

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292248 91177308-0d34-0410-b5e6-96231b3b80d8

[TLI] Appease spurious MSVC warning using llvm_unreachable. NFC.

r292188 confused MSVC because of the combined lack of a default
case and return statement.

Move the unreachable outside of the NumLibFuncs case, to make it
obvious that all cases should be handled.

llvm_unreachable is __declspec(noreturn), so I'm assuming this
does appease MSVC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292246 91177308-0d34-0410-b5e6-96231b3b80d8

Remove an overeager assert from r288844.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292244 91177308-0d34-0410-b5e6-96231b3b80d8

Revert r291640 change to fold X86 comparison with atomic_load_add.

Even with the fix from r291630, this still causes problems. I get
widespread assertion failures in the Swift runtime's WeakRefCount::increment()
function. I sent a reduced testcase in reply to the commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292242 91177308-0d34-0410-b5e6-96231b3b80d8

[PM] Teach the LoopPassManager to automatically canonicalize loops by
runnig LCSSA over them prior to running the loop pipeline.

This also teaches the loop PM to verify that LCSSA form is preserved
throughout the pipeline's run across the loop nest.

Most of the test updates just leverage this new functionality. One has to be
relaxed with the new PM as IVUsers is less powerful when it sees LCSSA input.

Differential Revision: https://reviews.llvm.org/D28743

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292241 91177308-0d34-0410-b5e6-96231b3b80d8

[ValueTracking] recognize a 'not' of an assumed condition as false

Also, add the corresponding match to the AssumptionCache's 'Affected Values' list.

Differential Revision: https://reviews.llvm.org/D28485

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292239 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] Fold ((C1 OP zext(X)) & C2) -> zext((C1 OP X) & C2)

This further extends r292179 to support additional binary operators
beyond subtraction.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292238 91177308-0d34-0410-b5e6-96231b3b80d8

Revert r292231.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292237 91177308-0d34-0410-b5e6-96231b3b80d8

[X86][AVX512] Add all_of/any_of avx512vl tests

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292235 91177308-0d34-0410-b5e6-96231b3b80d8

[ValueTracking] Extend known bits to understand @llvm.bitreverse.

Differential Revision: https://reviews.llvm.org/D28780

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292233 91177308-0d34-0410-b5e6-96231b3b80d8

[lit] Limit parallelism of sanitizer tests on Darwin [llvm part]

Running lit tests and unit tests of ASan and TSan on macOS has very bad performance when running with a high number of threads. This is caused by xnu (the macOS kernel), which currently doesn't handle mapping and unmapping of sanitizer shadow regions (reserved VM which are several terabytes large) very well. The situation is so bad that increasing the number of threads actually makes the total testing time larger. The macOS buildbots are affected by this. Note that we can't easily limit the number of sanitizer testing threads without affecting the rest of the tests.

This patch adds a special "group" into lit, and limits the number of concurrently running tests in this group. This helps solve the contention problem, while still allowing other tests to run in full, that means running lit with -j8 will still with 8 threads, and parallelism is only limited in sanitizer tests.

Differential Revision: https://reviews.llvm.org/D28420

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292231 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] reduce indent; NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292230 91177308-0d34-0410-b5e6-96231b3b80d8

Recommit r292214 "[Support/Compression] - Change zlib API to return Error instead of custom status"

No any changes, will follow up with D28807 commit containing APLi change for clang
to fix build issues happened.

Original commit message:
[Support/Compression] - Change zlib API to return Error instead of custom status.

Previously API returned custom enum values.
Patch changes it to return Error with string description.
That should help users to report errors in universal way.

Differential revision: https://reviews.llvm.org/D28684

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292226 91177308-0d34-0410-b5e6-96231b3b80d8

[AMDGPU] Assembler: fix v_mac_f16 immediates

Reviewers: vpykhtin, artem.tamazov, tstellarAMD

Subscribers: arsenm, kzhuravl, wdng, nhaehnle, yaxunl, tony-tye

Differential Revision: https://reviews.llvm.org/D28802

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292224 91177308-0d34-0410-b5e6-96231b3b80d8

[X86][SSE] Tests showing horizontal all_of/any_of of vector comparison results

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292223 91177308-0d34-0410-b5e6-96231b3b80d8

[llvm-objdump tests] Copy the inputs of tests closer to tests.

Summary:
Tests under tools/llvm-objdump should not use inputs from Object. Copied the
required inputs and aligned the new tests to be more consistent with the existing
tests in this respect.

Reviewers: ioeric

Reviewed By: ioeric

Subscribers: davide, djasper, cfe-commits

Differential Revision: https://reviews.llvm.org/D28799

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292222 91177308-0d34-0410-b5e6-96231b3b80d8

Revert r292214 "[Support/Compression] - Change zlib API to return Error instead of custom status."

It broked clang:
http://lab.llvm.org:8080/green//job/clang-stage1-cmake-RA-incremental_build/34218/consoleFull#46141505449ba4694-19c4-4d7e-bec5-911270d8a58c

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292217 91177308-0d34-0410-b5e6-96231b3b80d8

BrainF example: fixing output buffering issue
Differential Revision: https://reviews.llvm.org/D27824

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292216 91177308-0d34-0410-b5e6-96231b3b80d8

[Support/Compression] - Change zlib API to return Error instead of custom status.

Previously API returned custom enum values.
Patch changes it to return Error with string description.
That should help users to report errors in universal way.

Differential revision: https://reviews.llvm.org/D28684

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292214 91177308-0d34-0410-b5e6-96231b3b80d8

[XRay][Arm] Repair XRay table emission on Arm32 and add tests to identify such problem earlier

Summary:
Emission of XRay table was occasionally disabled for Arm32, but this bug was not then detected because earlier (also by mistake) testing of XRay was occasionally disabled on 32-bit Arm targets. This patch should fix that problem and detect such problems in the future.
This patch is one of a series, see also
- https://reviews.llvm.org/D28623

Reviewers: rengolin, dberris

Reviewed By: dberris

Subscribers: llvm-commits, aemerson, rengolin, dberris, iid_iunknown

Differential Revision: https://reviews.llvm.org/D28624

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292210 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine][X86][AVX] Add DemandedElts support for VPERMILPD/VPERMILPS instructions

Simplify a vpermilvar shuffle mask based on the elements of the mask that are actually demanded.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292209 91177308-0d34-0410-b5e6-96231b3b80d8

Update the release tester for MIPS. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292208 91177308-0d34-0410-b5e6-96231b3b80d8

Remove pid_t usage from llvm-xray

This type is not available on windows.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292206 91177308-0d34-0410-b5e6-96231b3b80d8

AMDGPU: Add replacement export intrinsics

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292205 91177308-0d34-0410-b5e6-96231b3b80d8

[bpf] error when unknown bpf helper is called

Emit error when BPF backend sees a call to a global function or to an external symbol.
The kernel verifier only allows calls to predefined helpers from bpf.h
which are defined in 'enum bpf_func_id'. Such calls in assembler must
look like 'call [1-9]+' where number matches bpf_func_id.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292204 91177308-0d34-0410-b5e6-96231b3b80d8

[utils] Add libc++ and libc++abi config to llvm-lit

This allows us to use bin/llvm-lit to run individual libc++ and
libc++abi tests without having to explicitly specify the site config
paths, similar to other projects.

Differential Revision: https://reviews.llvm.org/D28733

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292203 91177308-0d34-0410-b5e6-96231b3b80d8

[AVX-512] Add support for taking a bitcast between a SUBV_BROADCAST and VSELECT and moving it to the input of the SUBV_BROADCAST if it will help with using a masked operation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292201 91177308-0d34-0410-b5e6-96231b3b80d8

[AVX-512] Add test cases showing missed opportunities to fold subvector broadcasts with a mask operation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292200 91177308-0d34-0410-b5e6-96231b3b80d8

llc: Update link components

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292198 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] Don't DSE across readnone functions that may throw

Summary: Depends on D28740

Reviewers: dberlin, chandlerc, hfinkel, majnemer

Subscribers: mcrosier, llvm-commits

Differential Revision: https://reviews.llvm.org/D28742

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292197 91177308-0d34-0410-b5e6-96231b3b80d8

llc: Update LLVMBuild

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292196 91177308-0d34-0410-b5e6-96231b3b80d8

llc: Initialize more passes

Targets can add these. Initialize them so -print-before/-print-after
etc. work.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292195 91177308-0d34-0410-b5e6-96231b3b80d8

[Orc][RPC] Return unsupported rpc function errors from the non-retry cases in
negotiateFunction.

These cases were accidentally left out of r292055, resulting in a less
descriptive ECError being returned on these paths.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292193 91177308-0d34-0410-b5e6-96231b3b80d8

Revert "[TLI] Robustize SDAG proto checking by merging it into TLI."

This reverts commit r292189, as it causes issues on SystemZ bots.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292191 91177308-0d34-0410-b5e6-96231b3b80d8

[TLI] Robustize SDAG proto checking by merging it into TLI.

SelectionDAGBuilder recognizes libfuncs using some homegrown
parameter type-checking.

Use TLI instead, removing another heap of redundant code.

This isn't strictly NFC, as the SDAG code was too lax.
Concretely, this means changes are required to two tests:
- calling a non-variadic function via a variadic prototype isn't OK;
  it just happens to work on x86_64 (but not on, e.g., aarch64).
- mempcpy has a size_t parameter;  the SDAG code accepts any integer
  type, which meant using i32 on x86_64 worked.

I don't think it's worth supporting either of these (IMO) broken
testcases.  Instead, fix them to be more correct.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292189 91177308-0d34-0410-b5e6-96231b3b80d8

[TLI] Add prototype checking for all remaining LibFuncs.

This is another step towards unifying all LibFunc prototype checks.
This work started in r267758 (D19469);  add the remaining checks.

Also add a unittest that checks each libfunc declared with a known-valid
and known-invalid prototype.  New libfuncs added in the future are
required to have prototype checking in place; the known-valid test will
fail otherwise.

Differential Revision: https://reviews.llvm.org/D28030

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292188 91177308-0d34-0410-b5e6-96231b3b80d8

[TLI] Alphabetize some of the prototype check switch. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292187 91177308-0d34-0410-b5e6-96231b3b80d8

[unittests] Alphabetize cmake file list. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292186 91177308-0d34-0410-b5e6-96231b3b80d8

[bpf] error when BPF stack size exceeds 512 bytes

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292180 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] Fold ((C1-zext(X)) & C2) -> zext((C1-X) & C2)

This is valid if C2 fits within the bitwidth of X thanks to two's
complement modulo arithmetic.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292179 91177308-0d34-0410-b5e6-96231b3b80d8

Add comment to test file I forgot to save

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292178 91177308-0d34-0410-b5e6-96231b3b80d8

SimplifyLibCalls: Remove checks for fabs

Use the intrinsic instead of emitting the libcall which
will be replaced by the intrinsic.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292176 91177308-0d34-0410-b5e6-96231b3b80d8

AMDGPU: Remove dead pattern

This is the unsafe conversion pattern, but not guarded by
an unsafe math check. It is also already done in LegalizeDAG.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292173 91177308-0d34-0410-b5e6-96231b3b80d8

SimplifyLibCalls: Replace fabs libcalls with intrinsics

Add missing fabs(fpext) optimzation that worked with the call,
and also fixes it creating a second fpext when there were multiple
uses.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292172 91177308-0d34-0410-b5e6-96231b3b80d8

[Object] Fixup permissions of input files.

They just need to be read/dumped, so no need to set the exec
bit on any of them. NFCI, I guess.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292171 91177308-0d34-0410-b5e6-96231b3b80d8

[llvm-objdump] Dump PT_NOTE as part of -p.

PR: 31641

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292170 91177308-0d34-0410-b5e6-96231b3b80d8

[llvm-objdump] Dump PT_GNU_RELRO as part of -p.

PR: 31641

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292169 91177308-0d34-0410-b5e6-96231b3b80d8

[llvm-objdump] Dump PT_OPENBSD_{BOOTDATA,RANDOMIZE,WXNEEDED}.

PR: 31641

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292167 91177308-0d34-0410-b5e6-96231b3b80d8

Add missing header to see if that clears up the build

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292166 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine][AVX] Tests showing missed opportunities to pass demanded elts through a permilpd/permilps shuffle mask

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292165 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] use m_APInt instead of faking it

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292164 91177308-0d34-0410-b5e6-96231b3b80d8

Attempt to fix the MSVC build by using llvm::errc instead of std::errc

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292163 91177308-0d34-0410-b5e6-96231b3b80d8

ADMGPU/EG,CM: Implement _noret global atomics

_RTN versions will be a lot more complicated

Differential Revision: https://reviews.llvm.org/D28067

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292162 91177308-0d34-0410-b5e6-96231b3b80d8

[XRay] Implement the `llvm-xray graph` subcommand

Here we define the `graph` subcommand which generates a graph from the function
call information and uses it to present the call information graphically with
additional annotations.

Reviewers: dblaikie, dberris

Differential Revision: https://reviews.llvm.org/D27243

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292156 91177308-0d34-0410-b5e6-96231b3b80d8

Attempt to workaround MSVC build issue where I suspect an enum class constant 0 is considered a possible null pointer

I can't reproduce this so far with web compilers, so throwing this at
the bots to see if it sticks.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292155 91177308-0d34-0410-b5e6-96231b3b80d8

[PowerPC] Expand ISEL instruction into if-then-else sequence.

Generally, the ISEL is expanded into if-then-else sequence, in some
cases (like when the destination register is the same with the true
or false value register), it may just be expanded into just the if
or else sequence.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292154 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] fix names in canEvaluateShiftedShift(); NFC

It's not clear what 'First' and 'Second' mean, so use 'Inner' and 'Outer'
to match foldShiftedShift() and add comments with formulas, so it's easier
to see what's going on.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292153 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] use m_APInt to allow shift-shift folds for vectors with splat constants

Some existing 'FIXME' tests are still not folded because of splat holes in value tracking.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292151 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] add tests to show missed vector folds; NFC

The shift-shift possibilities became easier to see after:
https://reviews.llvm.org/rL292145

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292150 91177308-0d34-0410-b5e6-96231b3b80d8

PR31650: Refer to enum constant when initializing llvm::None constant

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292149 91177308-0d34-0410-b5e6-96231b3b80d8

[NVPTX] Add blank line to NVPTXUsage.rst to appease the Sphinx.

Fixes:

  Warning, treated as error:
  /home/buildbot/llvm-build-dir/llvm-sphinx-docs/llvm/src/docs/NVPTXUsage.rst:333:
  ERROR: Error in "code-block" directive:
  maximum 1 argument(s) allowed, 17 supplied.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292148 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] refactor shift-of-shift folds; NFCI

Reduces code duplication and makes it easier to extend these folds for vectors.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292145 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine][SSE] Tests showing missed opportunities to pass demanded elts through a packss/packus truncation

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292144 91177308-0d34-0410-b5e6-96231b3b80d8

[llvm-xray] Fix android build

std::to_string is not available in the android ndk. Using llvm::to_string
instead.

Committing as obvious.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292143 91177308-0d34-0410-b5e6-96231b3b80d8

[AArch64] Falkor supports Rounding Double Multiply Add/Subtract instructions.

Falkor only partially implements the ARMv8.1a extensions, so this patch
refactors the support for the SQRDML[A|S]H instruction into a separate
feature.

Differential Revision: https://reviews.llvm.org/D28681

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292142 91177308-0d34-0410-b5e6-96231b3b80d8

Revert r292132: [globalisel] Tablegen-erate current Register Bank Information'...

Several buildbots encountered a crash in tablegen when building this commit.
Reverting while I investigate the cause.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292136 91177308-0d34-0410-b5e6-96231b3b80d8

Fix use-after-free bug in AffectedValueCallbackVH::allUsesReplacedWith

When transferring affected values in the cache from an old value, identified by
the value of the current callback, to the specified new value we might need to
insert a new entry into the DenseMap which constitutes the cache. Doing so
might delete the current callback object. Move the copying logic into a new
function, a member of the assumption cache itself, so that we don't run into UB
should the callback handle itself be removed mid-copy.

Differential Revision: https://reviews.llvm.org/D28749

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292133 91177308-0d34-0410-b5e6-96231b3b80d8

[globalisel] Tablegen-erate current Register Bank Information

Summary:
Adds a RegisterBank tablegen class that can be used to declare the register
banks and an associated tablegen pass to generate the necessary code.

Reviewers: t.p.northover, ab, rovka, qcolombet

Subscribers: aditya_nandakumar, rengolin, kristof.beyls, vkalintiris, mgorny, dberris, llvm-commits, rovka

Differential Revision: https://reviews.llvm.org/D27338

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292132 91177308-0d34-0410-b5e6-96231b3b80d8

Revert "[PowerPC] Expand ISEL instruction into if-then-else sequence."

This reverts commit 1d0e0374438ca6e153844c683826ba9b82486bb1.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292131 91177308-0d34-0410-b5e6-96231b3b80d8

[SelectionDAG] Add knownbits support for BITREVERSE

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292130 91177308-0d34-0410-b5e6-96231b3b80d8

[PowerPC] Expand ISEL instruction into if-then-else sequence.

Generally, the ISEL is expanded into if-then-else sequence, in some
cases (like when the destination register is the same with the true
or false value register), it may just be expanded into just the if
or else sequence.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292128 91177308-0d34-0410-b5e6-96231b3b80d8

DWARFDebugInfoTest.cpp: Don't use ArrayRef with initializer. It was allocated locally.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292127 91177308-0d34-0410-b5e6-96231b3b80d8

[X86][SSE] Test showing missing BITREVERSE knownbits support

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292118 91177308-0d34-0410-b5e6-96231b3b80d8

[mips] Correct c.cond.fmt instruction definition.

Permit explicit $fcc<X> operand in c.cond.fmt instruction.

Add c.cond.fmt to the MIPS to microMIPS instruction mapping table.

Check that $fcc1 - $fcc7 are unusable for MIPS-I to MIPS-III for
c.cond.fmt, bc1t, bc1f.

Reviewers: seanbruno, zoran.jovanovic, vkalintiris

Differential Revision: https://reviews.llvm.org/D24510

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292117 91177308-0d34-0410-b5e6-96231b3b80d8

[SelectionDAG] Add support for BITREVERSE constant folding

We were relying on constant folding of the legalized instructions to do what constant folding we had previously

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292114 91177308-0d34-0410-b5e6-96231b3b80d8

[X86][SSE] Tests showing missing BITREVERSE constant folding

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292111 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine][SSE] Add DemandedElts support for PSHUFB instructions

Simplify a pshufb shuffle mask based on the elements of the mask that are actually demanded.

Differential Revision: https://reviews.llvm.org/D28745

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292101 91177308-0d34-0410-b5e6-96231b3b80d8

[X86] Merge the disassemblers handling of the different TYPE_RELs by getting the size information from the ENCODING field. NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292096 91177308-0d34-0410-b5e6-96231b3b80d8

[X86] Reduce the number of operand 'types' the disassembler needs to deal with. NFCI

We were frequently checking for a list of types and the different types
conveyed no real information. So lump them together explicitly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292095 91177308-0d34-0410-b5e6-96231b3b80d8

[AVX-512] Teach the disassembler about all of the EVEX gather and scatter instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292094 91177308-0d34-0410-b5e6-96231b3b80d8

[AVX-512] Begin giving the disassembler a way to recognize that VSIB is a different encoding than regular addressing modes.

This part first teaches it not to check error if EVEX.V2 is used by a VSIB instruction.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292093 91177308-0d34-0410-b5e6-96231b3b80d8

Fix typos. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292092 91177308-0d34-0410-b5e6-96231b3b80d8

[AVX-512] Add more gather/scatter encoding test cases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292089 91177308-0d34-0410-b5e6-96231b3b80d8

[AVX-512] Correct memory operand size for VPGATHERQPS and VPGATHERQD
with ZMM index. Similar for SCATTER and the prefetch gather and scatter
instructions.

Fixes PR31618.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292088 91177308-0d34-0410-b5e6-96231b3b80d8

[AVX-512] Fix register class in one of the gather/scatter memory operands so that all 32 bit registers can be allowed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292087 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] add tests to show missed vector folds; NFC

Also, add comments and remove bogus comment.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292082 91177308-0d34-0410-b5e6-96231b3b80d8

Empty line. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292081 91177308-0d34-0410-b5e6-96231b3b80d8

Use getLoopLatch in place of isLoopSimplifyForm

Summary:
Use getLoopLatch in place of isLoopSimplifyForm. we do not need
to know whether the loop has a preheader nor dedicated exits.

Reviewers: hfinkel, sanjoy, atrick, mkuper

Subscribers: mzolotukhin, llvm-commits

Differential Revision: https://reviews.llvm.org/D28724

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292078 91177308-0d34-0410-b5e6-96231b3b80d8

[CostModel][X86] Fix AVX512BW vector shift costs for vXi16 types

We already have patterns in place to support 128/256-bit shifts without AVX512VL

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292077 91177308-0d34-0410-b5e6-96231b3b80d8

[CostModel][X86] Drop separate AVX512VL checks - they match existing AVX512 costs

Keep the tests though.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292076 91177308-0d34-0410-b5e6-96231b3b80d8

[CostModel][X86] Update vector shift tests to correctly check by non-constant uniform values.

Use shuffle( scslar_to_vector, zeroinitializer) pattern instead of shuffle( vec, zeroinitializer)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292075 91177308-0d34-0410-b5e6-96231b3b80d8

Delete a dead argument. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292074 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine] fix formatting; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292073 91177308-0d34-0410-b5e6-96231b3b80d8

[InstCombine][SSE] Tests showing missed opportunities to pass demanded elts through a pshufb shuffle mask

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292072 91177308-0d34-0410-b5e6-96231b3b80d8

[NVPTX] Add fptosi tests to convert-fp.ll.

These seem to have been left off by accident.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292071 91177308-0d34-0410-b5e6-96231b3b80d8