If we are and the constant like 0xFFFFFFC00000, for now, we are using several
instructions to generate this 48bit constant and final an "and". However, we
could exploit it with two rotate instructions.
MB ME MB+63-ME
+----------------------+ +----------------------+
|0000001111111111111000| -> |0000000001111111111111|
+----------------------+ +----------------------+
0 63 0 63
Rotate left ME + 1 bit first, and then, mask it with (MB + 63 - ME, 63),
finally, rotate back. Notice that, we need to round it with 64 bit for the
wrapping case.
Reviewed by: ChenZheng, Nemanjai
Differential Revision: https://reviews.llvm.org/D71831
Add DestructiveBinaryImm SQSHLU patterns and tests. These patterns allow the SQSHLU instruction to match with a MOVPRFX.
Differential Revision: https://reviews.llvm.org/D76728
This patch adds PC Relative support for global values that are known at link
time. If a global value requires access through the global offset table (GOT)
it is not covered in this patch.
Differential Revision: https://reviews.llvm.org/D75280
These are needed as a counterpart for VGPR subregs even though
there are no scalar instructions which can operate 16 bit values.
When we are materializing a constant that is done into an SGPR
and that SGPR may/will be copied into a 16 bit VGPR subreg. Such
copy is illegal. There are also similar problems if a source
operand of a 16 bit VALU instruction is an SGPR. In addition
we need to get a register with a lo16 subregister of an SGPR
RC during selection and this fails as well.
All of that makes me believe we need these subregisters as a
syntactic glue.
Differential Revision: https://reviews.llvm.org/D78250
Fix for the address optimization for gathers and scatters which would in
some complex cases push out instructions not to the vector loop preheader,
but to other locations as well which lead to a scrambled order and the
compilation failing.
This patch ensures that said instructions are always pushed to the end
of the vector loop preheader.
Differential Revision: https://reviews.llvm.org/D78293
Summary:
When doing the conversion: MachineInst -> MCInst, we should ignore the
implicit operands, it will expose more opportunity for InstiAlias.
Reviewed By: steven.zhang
Differential Revision: https://reviews.llvm.org/D77118
Summary:
Changing all mnemonic to match assembly instructions to simplify mnemonic
naming rules. This time update all fixed-point arithmetic instructions.
This also corrects bswp operand type.
Reviewed By: simoll
Differential Revision: https://reviews.llvm.org/D78177
Summary:
The INLINEASM MIR instructions use immediate operands to encode the values of some operands.
The MachineInstr pretty printer function already handles those operands and prints human readable annotations instead of the immediates. This patch adds similar annotations to the output of the MIRPrinter, however uses the new MIROperandComment feature.
Reviewers: SjoerdMeijer, arsenm, efriedma
Reviewed By: arsenm
Subscribers: qcolombet, sdardis, jvesely, wdng, nhaehnle, hiraditya, jrtc27, atanasyan, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D78088
Summary:
The function in X86MCCodeEmitter has too many parameters to make it look
messy, and some parameters are unnecessary. This is the first patch to
reduce their parameters.
The follwing operations are cheap
```
unsigned Opcode = MI.getOpcode();
const MCInstrDesc &Desc = MCII.get(Opcode);
uint64_t TSFlags = Desc.TSFlags;
```
So if we pass a `MCInst`, we don't need to pass `MCInstrDesc`;
if we pass a `MCInstrDesc`, we don't need to pass `TSFlags`.
Reviewers: craig.topper, MaskRay, pengfei
Reviewed By: craig.topper
Subscribers: annita.zhang, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D78180
Summary:
The renaming is necessary to make the naming scheme uniform with other
gather/scatter load/stores SVE intrinsics.
The naming of variables and functions have been adapted to make it
explicit whether we are dealing with a scalar offset (which is
unscaled) or an index (which is scaled according to the data type of
the lanes of the vector).
Reviewers: andwar, sdesmalen, rengolin
Reviewed By: andwar
Subscribers: tschuett, hiraditya, arphaman, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77839
We have added code to correct the .localentry values on assignments. However, we
never clear the set so presumably it will still contain the (now dangling)
MCSymbol pointers across a call to finish() and reset() in the streamer.
This is based on my speculation that it is the reason we are getting
segmentation faults mentioned in https://bugs.llvm.org/show_bug.cgi?id=45366
Fixes: https://bugs.llvm.org/show_bug.cgi?id=45366
Differential revision: https://reviews.llvm.org/D78196
This moves v32i16/v64i8 to a model consistent with how we
treat integer types with avx1.
This does change the ABI for types vXi16/vXi8 vectors larger than
512 bits to pass in multiple zmms instead of multiple ymms. We'd
already hacked some code to make v64i8/v32i16 pass in zmm.
Cost model is still a bit of a mess. In some place I tried to
match existing behavior. But really we need to account for
splitting and concating costs. Cost model for shuffles is
especially pessimistic.
Differential Revision: https://reviews.llvm.org/D76212
-Consistently name the functions as split*
-Add a helper for doing the two extractSubvector calls and determining the size of the split
-Use getSplitDestVTs to get the result type for the split node.
-Move the binary and unary helper to one place in the file near the extractSubvector functions. Left the VSETCC one near LowerVSETCC since that's its only caller.
-Remove the 256/512 wrappers that just had asserts. I don't think they provided a lot of value and now with the routines called split* the call sites are more obvious what they do.
-Make the unary routine support different source and dest types to support D76212.
-Add some weaker asserts into the helpers to make up for losing the very specific asserts from the 256/512 wrappers.
Differential Revision: https://reviews.llvm.org/D78176
Summary:
No error or warning is emitted when specific reserved registers are
written to in inline assembly. Therefore, writes to the program counter
or to the frame pointer, for instance, were permitted, which could have
led to undesirable behaviour.
Example:
int foo() {
register int a __asm__("r7"); // r7 = frame-pointer in M-class ARM
__asm__ __volatile__("mov %0, r1" : "=r"(a) : : );
return a;
}
In contrast, GCC issues an error in the same scenario.
This patch detects writes to specific reserved registers in inline
assembly for ARM and emits an error in such case. The detection works
for output and input operands. Clobber operands are not handled here:
they are already covered at a later point in
AsmPrinter::emitInlineAsm(const MachineInstr *MI). The registers
covered are: program counter, frame pointer and base pointer.
This is ARM only. Therefore the implementation of other targets'
counterparts remain open to do.
Reviewers: efriedma
Reviewed By: efriedma
Subscribers: kristof.beyls, hiraditya, danielkiss, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D76848
An irreducible SCC is one which has multiple "header" blocks, i.e., blocks
with control-flow edges incident from outside the SCC. This pass converts an
irreducible SCC into a natural loop by introducing a single new header
block and redirecting all the edges on the original headers to this
new block.
This is a useful workaround for a limitation in the structurizer
which, which produces incorrect control flow in the presence of
irreducible regions. The AMDGPU backend provides an option to
enable this pass before the structurizer, which may eventually be
enabled by default.
Reviewed By: nhaehnle
Differential Revision: https://reviews.llvm.org/D77198
This restores commit 2ada8e2525.
Originally reverted with commit 44e09b59b8.
Summary:
Changing all mnemonic to match assembly instructions to simplify mnemonic
naming rules. This time update all fixed-point arithmetic instructions.
This also corrects smax/smin code generations.
Reviewed By: simoll
Differential Revision: https://reviews.llvm.org/D77856
This reverts commit 2ada8e2525.
Buildbots produced compilation errors which I was not able to quickly
reproduce locally. Need more time to investigate.
An irreducible SCC is one which has multiple "header" blocks, i.e., blocks
with control-flow edges incident from outside the SCC. This pass converts an
irreducible SCC into a natural loop by introducing a single new header
block and redirecting all the edges on the original headers to this
new block.
This is a useful workaround for a limitation in the structurizer
which, which produces incorrect control flow in the presence of
irreducible regions. The AMDGPU backend provides an option to
enable this pass before the structurizer, which may eventually be
enabled by default.
Reviewed By: nhaehnle
Differential Revision: https://reviews.llvm.org/D77198
Summary:
Only first-faulting and non-faulting loads read/update the FFR register
and hence only the corresponding SDNodes should be decorated with
`SDNPOptInGlue` and `SDNPOutGlue`. This patch:
* removes SDNPOptInGlue from regular loads stores (FFR is not read)
* adds SDNPOutGlue to first-faulting and non-faulting loads (FFR is
both read and updated)
Differential Revision: https://reviews.llvm.org/D77724
The pass was incorrectly reverting back to a "T" when something wrote
to VPR inside a "E" block. This is not the correct behaviour, the
predicate should stay the same.
Differential Revision: https://reviews.llvm.org/D77798
It can be used to avoid passing the begin and end of a range.
This makes the code shorter and it is consistent with another
wrappers we already have.
Differential revision: https://reviews.llvm.org/D78016
Summary:
Creates the SVEIntrinsicOpts pass. In this patch, the pass tries
to remove unnecessary reinterpret intrinsics which convert to
and from svbool_t (llvm.aarch64.sve.convert.[to|from].svbool)
For example, the reinterprets below are redundant:
%1 = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> %a)
%2 = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %1)
The pass also looks for ptest intrinsics and phi instructions where
the operands are being needlessly converted to and from svbool_t.
Reviewers: sdesmalen, andwar, efriedma, cameron.mcinally, c-rhodes, rengolin
Reviewed By: efriedma
Subscribers: mgorny, tschuett, kristof.beyls, hiraditya, rkruppe, psnobl, danielkiss, cfe-commits, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D76078
The _GLOBAL_OFFSET_TABLE_ in SysVr4 ELF is conventionally the base of the
.got or .got.prel sections. Expressions such as _GLOBAL_OFFSET_TABLE_
- (.L1 +8) are used in assembler code to calculate offsets into the .got.
At present MC outputs a R_ARM_REL32 with respect to the
_GLOBAL_OFFSET_TABLE_ symbol, whereas gas outputs a R_ARM_BASE_PREL
relocation with respect to the _GLOBAL_OFFSET_TABLE_ symbol. While both are
correct the R_ARM_REL32 depends on the value of the _GLOBAL_OFFSET_TABLE_
symbol, wheras te R_ARM_BASE_PREL relocation is idependent of the symbol.
The R_ARM_BASE_PREL is therefore slightly more robust to linker's that may
not follow the conventional placement of _GLOBAL_OFFSET_TABLE_; for example
LLD for some time defined _GLOBAL_OFFSET_TABLE_ to 0.
Differential Revision: https://reviews.llvm.org/D46319
Summary:
This iz pattern is a special pattern of im pattern. This im pattern
has been supported by https://reviews.llvm.org/D77769, so removing
iz pattern as a continuous patch.
Reviewed By: simoll
Differential Revision: https://reviews.llvm.org/D77770
Ports the existing DAG combines, minus the simplify demanded bits
which seems to have no equivalent now. Without these, this isn't
particularly helpful in most of the IR sample cases.
Summary:
In CFGStackify, `fixUnwindMismatches` function fixes unwind destination
mismatches created by `try` marker placement. For example,
```
try
...
call @qux ;; This should throw to the caller!
catch
...
end
```
When `call @qux` is supposed to throw to the caller, it is possible that
it is wrapped inside a `catch` so in case it throws it ends up unwinding
there incorrectly. (Also it is possible `call @qux` is supposed to
unwind to another `catch` within the same function.)
To fix this, we wrap this inner `call @qux` with a nested
`try`-`catch`-`end` sequence, and within the nested `catch` body, branch
to the right destination:
```
block $l0
try
...
try ;; new nested try
call @qux
catch ;; new nested catch
local.set n ;; store exnref to a local
br $l0
end
catch
...
end
end
local.get n ;; retrieve exnref back
rethrow ;; rethrow to the caller
```
The previous algorithm placed the nested `try` right before the `call`.
But it is possible that there are stackified instructions before the
call from which the call takes arguments.
```
try
...
i32.const 5
call @qux ;; This should throw to the caller!
catch
...
end
```
In this case we have to place `try` before those stackified
instructions.
```
block $l0
try
...
try ;; this should go *before* 'i32.const 5'
i32.const 5
call @qux
catch
local.set n
br $l0
end
catch
...
end
end
local.get n
rethrow
```
We correctly handle this in the first normal `try` placement phase
(`placeTryMarker` function), but failed to handle this in this
`fixUnwindMismatches`.
Reviewers: dschuff
Subscribers: sbc100, jgravelle-google, hiraditya, sunfish, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77950
in the same section.
This allows specifying BasicBlock clusters like the following example:
!foo
!!0 1 2
!!4
This places basic blocks 0, 1, and 2 in one section in this order, and
places basic block #4 in a single section of its own.
Since 1725f28841, this should check
isFMADLegalForFAddFSub rather than the the plain isOperationLegal.
This would assert in a subset of cases due to an oddity in how FMAD is
selected. We will allow FMA formation pre-legalize, but not FMAD even
in cases where it would be valid.
The current hook requires passing in the root fadd/fsub. However, in
this distributed case, this would be far more complicated to pass in
the relevant operand. AMDGPU doesn't get any value from the node, and
only needs the type and is the only implementor, so I'm not sure why
we have this complexity. Just rename and expand the assert to avoid
the more complicated checks spread through the distribution logic.
The shuffle decoding is used by X86ISelLowering and
MCTargetDesc/X86InstComments. The latter used to be in a
separate InstPrinter library. The Utils library existed to allow
InstPrinter and CodeGen to share the shuffle decoding. Since
X86InstComments now lives in the MCTargetDesc, which CodeGen
already depends on, we can sink the shuffle decoding there as well.
Differential Revision: https://reviews.llvm.org/D77980
Summary:
Fold (shift (shift X, C2), C1) -> (shift X, (C1 + C2)) for logical as
well as arithmetic shifts. This is needed to prevent regressions from
an upcoming funnel shift expansion change.
While we're here, fold (VSRAI -1, C) -> -1 too.
Reviewers: RKSimon, craig.topper
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77300
Improve the chances of folding the writemask into the combined shuffle by scaling a wider shuffle mask to match the root's original type.
This creates a few minor issues with variable shuffles, preventing combines of shuffles because of the more limited support binary shuffle types. In most cases we're probably better off combining the shuffles and losing the writemask fold, but this isn't always going to be true.
Summary:
Updated CallPromotionUtils and impacted sites. Parameters that are
expected to be non-null, and return values that are guranteed non-null,
were replaced with CallBase references rather than pointers.
Left FIXME in places where more changes are facilitated by CallBase, but
aren't CallSites: Instruction* parameters or return values, for example,
where the contract that they are actually CallBase values.
Reviewers: davidxl, dblaikie, wmi
Reviewed By: dblaikie
Subscribers: arsenm, jvesely, nhaehnle, eraman, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77930
As discussed in PR36617:
https://bugs.llvm.org/show_bug.cgi?id=36617#c13
...we can avoid the likely slow round-trip from XMM to GPR to XMM
by using the vector versions of the convert instructions.
Based on experimental results from recent Intel/AMD chips, we don't
need to worry about triggering denorm stalls while operating on
garbage data in the high lanes with convert instructions, so this is
expected to always be as good or better perf than the scalar
instruction equivalent. FP exceptions are also not a concern because
strict code should not be using the regular SDAG opcodes.
Differential Revision: https://reviews.llvm.org/D77895
A lot of vectorized code doesn't use masks so we shouldn't penalize them by not doing shuffle combining on avx512 targets.
I've added support for VALIGNQ/VALIGND and 512-bit SHUF128 to prevent some regressions. I also prevented recombining 256-bit SHUF128 to PERM2X128. We may not need to add 256-bit SHUF128 support, but I don't think I found any cases requiring that in my testing.
Differential Revision: https://reviews.llvm.org/D77928
-Drop llvm:: on MVT::i32
-Use getValueType instead of getSimpleValueType for an equality
check just cause its shorter and doesn't matter.
-Don't create a const SDValue & since its cheap to copy.
-Remove explicit case from MVT enum to EVT.
-Add message to assert.
Selection would fail after the post legalize combiner put an illegal
zextload back together.
The base combiner has parameter to only allow legal operations, but
they appear to not be used. I also don't see a nice way to remove a
single entry from all_combines, so just hack around this.
As proposed in D77881, we'll have the related widening operation,
so this name becomes too vague.
While here, change the function signature to take an 'int' rather
than 'size_t' for the scaling factor, add an assert for overflow of
32-bits, and improve the documentation comments.
This is the same as what was done to the CallLoweringInfo in
TargetLowering.h in r309159.
This is just a step on the way to replacing this with CallBase.
There was another issue introduced by this commit that the OP
initially missed. Namely, for functions that are free to use
R2 as a callee-saved register, we emit a TOC expression based
on the address of the GEP label without emitting the GEP label.
Since we only emit such expressions for the large code model, this
issue only surfaced there.
I have confirmed that with this fix, the kernel build is successful
with target "all".
Summary:
Remove usages of asserting vector getters in Type in preparation for the
VectorType refactor. The existence of these functions complicates the
refactor while adding little value.
Reviewers: sdesmalen, efriedma, jonpa
Reviewed By: sdesmalen
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77265
Remove a number of includes that aren't necessary (nor are we relying on the remaining includes to provide the declarations), we just needed a llvm::Instruction forward declaration.
This exposed a couple of source files that were implicitly replying on the includes for their use of llvm::SmallSet or std::set, requiring local includes to be added there instead.
These will be widened in the DAG. In the meanwhile early
widening prevents otherwise possible vectorization of
such loads.
Differential Revision: https://reviews.llvm.org/D77835
If we're inserting into v2i8/v4i8/v8i8/v2i16/v4i16 style sub-128bit vectors ensure we don't use the SK_PermuteTwoSrc cost of the legalized value type - this is a followup to rG12c629ec6c59 which added equivalent sub-128bit shuffle costs
In LowerFP_TO_INTForReuse, when emitting `stfiwx`, alignment of 4 is
set for the `MachineMemOperand`, but RLI(ReuseLoadInfo)'s alignment is
not updated for following loads.
It's related to failed alignment check reported in
https://bugs.llvm.org/show_bug.cgi?id=45297
Differential Revision: https://reviews.llvm.org/D77624
The transformation currently does not differentiate between explicit
and implicit kills. However, it is not valid to later simply clear
an implicit kill flag since the kill could be due to a call or return.
Fixes: https://bugs.llvm.org/show_bug.cgi?id=45374
Summary:
D74269 added debug info to newly created instructions, including calls
to `malloc` and `free`, by taking debug info from existing instructions
around, whose debug info may or may not be empty.
But there are cases debug info is required by the IR verifier: when both
the caller and the callee functions have DISubprograms, meaning we
already have declarations to `malloc` or `free` with a DISubprogram
attached, newly created calls to `malloc` and `free` should have
non-empty debug info. This patch creates a non-empty dummy debug info in
this case to those calls to make the IR verifier pass.
Fixes https://bugs.llvm.org/show_bug.cgi?id=45461.
Reviewers: dschuff
Subscribers: aprantl, sbc100, jgravelle-google, hiraditya, sunfish, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77784
This is a fix for the previous patch 6c4b40def7.
In some cases it may be possible to have the compiler produce st_other=1 without
the compiler using mcpu=future which should not be the case. This patch adds a
guard to make sure that if we are using st_other=1 then we are also compiling
for future CPU.
This is similar to what I recently did for getArithmeticReductionCost.
I'm trying to account for the narrowing from 512->256->128 as we go.
I've also added a new helper method getMinMaxCost that tries to
handle the cases where we have native min/max instructions and
fall back to cmp+select when we don't.
Differential Revision: https://reviews.llvm.org/D76634
QingShan Zhang