Inspired by a recent Discourse post on undef vs. poison usage, this
series of patches should reduce the number of undefs in LLVM tests by
around 10%.
Only undef vector operands to insertelement/shufflevector have been
handled, which are by far the most common we've got.
The switchover is split into 3 fairly arbitrary clusters to make it
slightly more manageable: vector predication, fixed-length vectors,
scalable vectors.
This test shows a loop, whose preheader uses a SEW=64, LMUL=1 vector
operation. The loop body starts off with another SEW=64, LMUL=1 VADD
vector operation, before switching to a SEW=32, LMUL=1/2 vector store
instruction.
We can see that the VSETVLI insertion pass omits a VSETVLI before the
VADD (thinking it inherits its configuration from the preheader) but
does place a SEW=32, LMUL=1/2 VSETVLI before the store. This results in
a miscompilation as when the loop comes back around, the VADD is
incorrectly configured with SEW=32, LMUL=1/2.
It appears to be a bad load/store optimization, as replacing the vector
store with an SEW=32, LMUL=1/2 VADD does correctly insert a VSETVLI. The
issue is therefore possibly arising from canSkipVSETVLIForLoadStore.
Differential Revision: https://reviews.llvm.org/D118629
Do "simplifyShift" and "FoldConstantArithmetic" folds for the SSHLSAT
and USHLSAT DAG nodes.
This includes folds such as:
(shlsat undef/poison, x) -> 0
(shlsat x, undef/poison) -> undef
(shlsat x, too_large_shamt) -> undef
(shlsat 0, x) -> 0
(shlsat x, 0) -> x
(shlsat c1, c2) -> c3
Differential Revision: https://reviews.llvm.org/D118603
I have updated TargetLowering::isConstTrueVal to also consider
SPLAT_VECTOR nodes with constant integer operands. This allows the
optimisation to also work for targets that support scalable vectors.
Differential Revision: https://reviews.llvm.org/D117210
In some cases StructurizeCFG inserts i1 xor instructions to invert
predicates. Add a quick loop to clean these up afterwards if we can get
away with modifying an existing compare instruction instead.
(StructurizeCFG is generally run late in the pipeline so instcombine
does not clean them up for us.)
Differential Revision: https://reviews.llvm.org/D118623
Summary:
Add code object v5 support (deafult is still v4)
Generate metadata for implicit kernel args for the new ABI
Set the metadata version to be 1.2
Reviewers:
t-tye, b-sumner, arsenm, and bcahoon
Fixes:
SWDEV-307188, SWDEV-307189
Differential Revision:
https://reviews.llvm.org/D118272
Factoring it out so we can subsequently cache it. This should be a NFC,
however, for the float quantities, we see small errors in the least
significant digits. This is because, before, we were summing up one by
one. Now, we sum up results of sums.
This shouldn't matter for ML, and will require rework when we do
quantization (avoiding floats altogether), but meanwhile, it did require
an update to the reference file used for testing.
The patch also bumps the precision of the variables involved in this, to
reduce the error (note they are casted back to float at the end by the
SET macro, since we only work with float and not double in TF)
Differential Revision: https://reviews.llvm.org/D118659
New target SDNodes are added: AArch64ISD::MOPS_MEMSET, etc.
Each intrinsic is translated to one of these in SelectionDAGBuilder
via EmitTargetCodeForMOPS.
A custom lowering routine for INTRINSIC_W_CHAIN is added to handle
llvm.aarch64.mops.memset.tag. This takes a separate path from the common
intrinsics but ultimately ends up in the same EmitMOPS().
This is part 4/4 of a series of patches split from
https://reviews.llvm.org/D117405 to facilitate reviewing.
Patch by Tomas Matheson, Lucas Prates and Son Tuan Vu.
Differential Revision: https://reviews.llvm.org/D117764
This implements codegen for Armv8.8/9.3 Memory Operations extension (MOPS).
Any memcpy/memset/memmov intrinsics will always be emitted as a series
of three consecutive instructions P, M and E which perform the
operation. The SelectionDAG implementation is split into a separate
patch.
AArch64LegalizerInfo will now consider the following generic opcodes
if +mops is available, instead of legalising by expanding them to
libcalls: G_BZERO, G_MEMCPY_INLINE, G_MEMCPY, G_MEMMOVE, G_MEMSET
The s8 value of memset is legalised to s64 to match the pseudos.
AArch64O0PreLegalizerCombinerInfo will still be able to combine
G_MEMCPY_INLINE even if +mops is present, as it is unclear whether it is
better to generate fixed length copies or MOPS instructions for the
inline code of small or zero-sized memory operations, so we choose to be
conservative for now.
AArch64InstructionSelector will select the above as new pseudo
instructions: AArch64::MOPSMemory{Copy/Move/Set/SetTagging} These are
each expanded to a series of three instructions (e.g. SETP/SETM/SETE)
which must be emitted together during code emission to avoid scheduler
reordering.
This is part 3/4 of a series of patches split from
https://reviews.llvm.org/D117405 to facilitate reviewing.
Patch by Tomas Matheson and Son Tuan Vu
Differential Revision: https://reviews.llvm.org/D117763
This reverts commit 00bf4755e90c89963a135739218ef49c2417109f.
This change broke the emscripten builder (among other things):
https://ci.chromium.org/ui/p/emscripten-releases/builders/try/linux/b8823500584349280721/overview
Sample failure:
```
test_unistd_unlink (test_core.core0) ...
wasm-ld: error: symbol type mismatch: __stdio_write
>>> defined as WASM_SYMBOL_TYPE_FUNCTION in /usr/local/google/home/sbc/dev/wasm/emscripten/cache/sysroot/lib/wasm32-emscripten/libc-debug.a(__stdio_write.o)
>>> defined as WASM_SYMBOL_TYPE_DATA in /usr/local/google/home/sbc/dev/wasm/emscripten/cache/sysroot/lib/wasm32-emscripten/libc-debug.a(stderr.o)
```
The optimization added in D118139 causes a crash on the added test case
while trying to zero extend an vector of floats.
Fix the crash by bailing out for floating point operands.
Reviewed By: DavidTruby
Differential Revision: https://reviews.llvm.org/D118615
The spec doesn't seem to be written as if Zfh implies Zfhmin. They
seem to be separate extensions.
This patch moves the instructions from Zfhmin to be enabled with
either the Zfh or Zfhmin extensions.
Reviewed By: achieveartificialintelligence
Differential Revision: https://reviews.llvm.org/D118581
This reverts commit a6b54ddaba2d5dc0f72dcc4591c92b9544eb0016.
Apparently it is not safe to modify the condition even if it passes the
hasOneUse test, because StructurizeCFG might have other references to
the condition that are not manifest in the IR use-def chains.
Fixes a crash ('Invalid size request on a scalable vector') in visitAlloca()
when we call this function for a scalable alloca instruction, caused
by the implicit conversion of TySize to uint64_t.
This patch changes TySize to a TypeSize as returned by getTypeAllocSize()
and ensures the allocation size is multiplied by vscale for scalable vectors.
Reviewed By: sdesmalen, david-arm
Differential Revision: https://reviews.llvm.org/D118372
We already call SimplifyDemandedVectorElts using whether each vector mask element is zero/nonzero, this just extends this to also try SimplifyDemandedBits using the demanded bits mask generated from the nonzero elements.
This also requires an additional TargetLowering::SimplifyDemandedBits DemandedBits/DemandedElts wrapper.
If AMDGPUAnnotateUniformValues finds a load from a uniform pointer with
no potentially clobbering stores between the kernel entry point and the
load instruction, it adds noclobber metadata to the *address*. This is
unsafe because it can get applied to other loads in the same which do
have aliasing stores.
Differential Revision: https://reviews.llvm.org/D118458
As raised by @efriedma on D117995 - the source must not be undef/poison to demand any bits in mul(x,x) other than bit[1]
https://alive2.llvm.org/ce/z/Cxkjen
This avoids various cases where StructurizeCFG would otherwise insert an
xor i1 instruction, and it since it generally runs late in the pipeline,
instcombine does not clean up the xor-of-cmp pattern.
Differential Revision: https://reviews.llvm.org/D118478
This patches fixes the visibility and linkage information of symbols
referring to IR globals.
Emission of external declarations is now done in the first execution
of emitConstantPool rather than in emitLinkage (and a few other
places). This is the point where we have already gathered information
about used symbols (by running the MC Lower PrePass) and not yet
started emitting any functions so that any declarations that need to
be emitted are done so at the top of the file before any functions.
This changes the order of a few directives in the final asm file which
required an update to a few tests.
Reviewed By: sbc100
Differential Revision: https://reviews.llvm.org/D118122
We can use the RISCVISD::GREV encoding that swaps the bits in
each byte. This allows it to use the existing computeKnownBits
support for RISCVISD::GREV.
Limit this to SSE41 - AVX1 targets to avoid UNPCKL(PSHUFB,PSHUFB), pre-SSE41 we don't have PACKUSDW/BLENDW and with AVX2 we can perform this as PERMQ(PSHUFB()).
Allows pow2 mask tests to avoid an unnecessary constant load.
Noticed while investigating how to extend MatchVectorAllZeroTest to support more allof/anyof patterns.
This would allow pow2 mask tests to avoid an unnecessary constant load.
Noticed while investigating how to extend MatchVectorAllZeroTest to support more allof/anyof patterns.
Without AVX512 (which can efficiently extend/truncate to vXi16/vXi32), unpacking/packing to vXi16 is more efficient that relying on the (uops-heavy) PBLENDV shift expansion
Now the backend promotes mask vector to an i8 vector and extract element from that. We could bitcast to a widen element vector, and extract from it to GPR, then use I instruction to extract the certain bit.
Differential Revision: https://reviews.llvm.org/D117389
Currently, the clang.arc.attachedcall bundle takes an optional function
argument. Depending on whether the argument is present, calls with this
bundle have the following semantics:
- on x86, with the argument present, the call is lowered to:
call _target
mov rax, rdi
call _objc_retainAutoreleasedReturnValue
- on AArch64, without the argument, the call is lowered to:
bl _target
mov x29, x29
and the objc runtime call is expected to be emitted separately.
That's because, on x86, the objc runtime checks for both the mov and
the call on x86, and treats the combination as the ARC autorelease elision
marker.
But on AArch64, it only checks for the dedicated NOP marker, as that's
historically been sufficiently unique. Thanks to that, the runtime call
wasn't required to be adjacent to the NOP marker, so it wasn't emitted
as part of the bundle sequence.
This patch unifies both architectures: on AArch64, we now emit all
3 instructions for the bundle. This guarantees that the runtime call
is adjacent to the marker in the sequence, and that's information the
runtime can use to further optimize this.
This helps simplify some of the handling, in particular
BundledRetainClaimRVs, which no longer needs to know whether the bundle
is sufficient or not: it now always should be.
Note that this does not include an AutoUpgrade for the nullary bundles,
as they are only produced in ObjCContract as part of the obj/asm emission
pipeline, and are not expected to be in bitcode.
Differential Revision: https://reviews.llvm.org/D118214
We were creating a truncate with the default for the type, but for
VP intrinsics we have a VL that we should use.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D118406
If we have a vector FP division with a splatted divisor, use
getVectorMinNumElements when scaling the num of uses by splat factor.
For AArch64 the combine kicks in for the <vscale x 4 x float> case since it's
above the fdiv threshold (3) when scaling num uses by splat factor, but the
codegen is worse (splat + vector fdiv + vector fmul) than the <vscale x 2 x
double> case (splat + vector fdiv).
If the combine could be converted into a scalar FP division by
scalarizeBinOpOfSplats it may be cheaper, but it looks like this is predicated
on the isExtractVecEltCheap TLI function which is implemented for x86 but not
AArch64. Perhaps for now combineRepeatedFPDivisors should only scale num uses
by splat if the division can be converted into scalar op.
Reviewed By: sdesmalen
Differential Revision: https://reviews.llvm.org/D118343
`__gnu_h2f_ieee` and `__gnu_f2h_ieee` are introduce by ARM and set that as
default name for fp16 and fp32 conversion in LLVM.
However RISC-V GCC using default naming scheme for that, which is
`__extendhfsf2` and `__truncsfhf2` for that, that cause runtime ABI
incompatible issue.
Although we didn't have formal runtime ABI spec to specify those naming
convention yet, but I think it would be great to fix the incompatible
issue first.
And I've plan to create a runtime ABI spec undere psABI spec this year.
Reviewed By: asb
Differential Revision: https://reviews.llvm.org/D118207
extract_vec_elt (load X), C --> scalar load (X+C)
As noted in the comment, DAGCombiner has this fold -- and the code in this
patch is adapted from DAGCombiner::scalarizeExtractedVectorLoad() -- but
x86 should benefit even if the loaded vector has other uses as long as we
apply some other x86-specific conditions. The motivating example from #50310
is shown in vec_int_to_fp.ll.
Fixes#50310
Differential Revision: https://reviews.llvm.org/D118376
