G_VAARG can be expanded similiar to SelectionDAG::expandVAArg through
LegalizerHelper::lower. This patch implements the lowering through this
style of expansion.
The expansion gets the head of the va_list by loading the pointer to
va_list. Then, the head of the list is adjusted depending on argument
alignment information. This gives a pointer to the element to be read
out of the va_list. Next, the head of the va_list is bumped to the next
element in the list. The new head of the list is stored back to the
original pointer to the head of the va_list so that subsequent G_VAARG
instructions get the next element in the list. Lastly, the element is
loaded from the alignment adjusted pointer constructed earlier.
This change is stacked on #73062.
Cleanup work towards removing the method Type::getPointerTo.
If a call to Type::getPointerTo is used solely to support an unneeded
pointer-cast, remove the call entirely.
This changes the fadd legalization to handle fp16 types, and treats more types
as legal so that the backend can produce the correct patterns. This is
currently a missing identity fold for `fadd x -0.0 -> x`
This fixes cases when SizeInBits is a multiple of the narrow size.
If SizeBits is equal to NarrowTy size, the first block would create an
illegal G_SEXT_INREG where the the extension size is equal to the type.
I tried to turn it into G_TRUNC+G_SEXT, but that just turned back into
G_SEXT_INREG causing an infinite loop. So punt to the splitting case.
In the for loop we should copy when the part ends on SizeInBits. In that
case there is no G_SEXT_INREG needed for partial. But we should note
that register in PartialExtensionReg for the first full part to use.
If the part starts on SizeInBits then we should do an AShr of
PartialExtensionReg.
We should only get to the G_SEXT_INREG case if the SizeInBits is in the
middle of the part.
It seems TypeSize is currently broken in the sense that:
TypeSize::Fixed(4) + TypeSize::Scalable(4) => TypeSize::Fixed(8)
without failing its assert that explicitly tests for this case:
assert(LHS.Scalable == RHS.Scalable && ...);
The reason this fails is that `Scalable` is a static method of class
TypeSize,
and LHS and RHS are both objects of class TypeSize. So this is
evaluating
if the pointer to the function Scalable == the pointer to the function
Scalable,
which is always true because LHS and RHS have the same class.
This patch fixes the issue by renaming `TypeSize::Scalable` ->
`TypeSize::getScalable`, as well as `TypeSize::Fixed` to
`TypeSize::getFixed`,
so that it no longer clashes with the variable in
FixedOrScalableQuantity.
The new methods now also better match the coding standard, which
specifies that:
* Variable names should be nouns (as they represent state)
* Function names should be verb phrases (as they represent actions)
visitJumpTable is called on FinishBasicBlock. At that time, getCurSDLoc
will always return SDLoc without DebugLoc since CurInst was set to
nullptr after visiting each instruction.
This patch passes SDLoc to buildJumpTable when visiting SwitchInst so
that visitJumpTable can use it later.
Similar to #70635, this expands the handling of integer to fp
conversions. The code is very similar to the float->integer conversions
with types handled oppositely. There are some extra unhandled cases
which require more handling for ASR operations.
The @llvm.amdgcn.cs.chain intrinsic is essentially a call. The call
parameters are bundled up into 2 intrinsic arguments, one for those that
should go in the SGPRs (the 3rd intrinsic argument), and one for those
that should go in the VGPRs (the 4th intrinsic argument). Both will
often be some kind of aggregate.
Both instruction selection frameworks have some internal representation
for intrinsics (G_INTRINSIC[_WITH_SIDE_EFFECTS] for GlobalISel,
ISD::INTRINSIC_[VOID|WITH_CHAIN] for DAGISel), but we can't use those
because aggregates are dissolved very early on during ISel and we'd lose
the inreg information. Therefore, this patch shortcircuits both the
IRTranslator and SelectionDAGBuilder to lower this intrinsic as a call
from the very start. It tries to use the existing infrastructure as much
as possible, by calling into the code for lowering tail calls.
This has already gone through a few rounds of review in Phab:
Differential Revision: https://reviews.llvm.org/D153761
This reverts commit 28ae42e662.
The assert in getIConstantVRegVal was not updated for this change.
The ValAndVReg->VReg == VReg check fails if any look through happens.
RISC-V was the only target using the lookthrough functionality, but I'm
not sure it was needed so I'm removing that too.
Now that we have more types handled for zext/sext and trunc, it is
possible to get more types working for the vector float to integer
conversions. This patch adds fp16, widening and narrowing vector support
to handle more types. The smaller types wil be expanded to the size of
the larger element type. A couple of case require more awkward truncates
to get working as they go from illegal to illegal types.
G_GLOBAL_VALUE should be lowered into an absolute address if
`-codemodel=small` is used or into a PC-relative if `-codemodel=medium`
is used.
PR #68380 tried to create special instructions to do this, but I don't
see why we need to do that.
The pre-index matcher just needs some small heuristics to make sure it
doesn't cause regressions. Apart from that it's a simple change, since
the only difference is an immediate operand of '1' vs '0' in the
instruction.
Previously they were passed by non-const reference. No in tree target
modifies the values.
This makes it possible to call assignValueToAddress from
assignCustomValue without a const_cast. For example in this patch
https://github.com/llvm/llvm-project/pull/69138.
…tVRegSExtVal
The implementation of these methods uses
getIConstantVRegValWithLookThrough with LookThroughInstrs argument set
to false. By adding the optional argument to getIConstantVRegVal and
getIConstantVRegSExtVal we can take advantage of the already built look
through functionality.
Update `LegalizerHelper::widenScalarMulo` to not create a mulo if we aren't going to use the overflow flag. This prevents needing to legalize the widened operation. This generates better code when we need to make a libcall for multiply.
G_TRUNC will get lowered into trunc(merge(trunc(unmerge),
trunc(unmerge))) if the source is larger than 128 bits or the truncation
is more than half of the current bit size.
Now mirrors ZEXT/SEXT code more closely for vector types.
The change implements support of the intrinsics `get_fpmode`,
`set_fpmode` and `reset_fpmode` in Global Instruction Selector. Now they
are lowered into library function calls.
Differential Revision: https://reviews.llvm.org/D158260
The `BlockFrequency` class abstracts `uint64_t` frequency values. Use it
more consistently in various APIs and disable implicit conversion to
make usage more consistent and explicit.
- Use `BlockFrequency Freq` parameter for `setBlockFreq`,
`getProfileCountFromFreq` and `setBlockFreqAndScale` functions.
- Return `BlockFrequency` in `getEntryFreq()` functions.
- While on it change some `const BlockFrequency& Freq` parameters to
plain `BlockFreqency Freq`.
- Mark `BlockFrequency(uint64_t)` constructor as explicit.
- Add missing `BlockFrequency::operator!=`.
- Remove `uint64_t BlockFreqency::getMaxFrequency()`.
- Add `BlockFrequency BlockFrequency::max()` function.
Replace some uses of `Type::getPointerTo` via 2 ways
* Remove entirely if it's only used to support an unnecessary bitcast
(remove the bitcast as well).
* Replace with `PointerType::get`/`PointerType::getUnqual`
NFC opaque pointer clean-up effort.
The legalizer currently generates lots of G_AND artifacts.
For example between boolean uses and defs there is always a G_AND with a mask of 1, but when the target uses ZeroOrOneBooleanContents, this is unnecessary.
Currently these artifacts have to be removed using post-legalize combines.
Omitting these artifacts at their source in the artifact combiner has a few advantages:
- We know that the emitted G_AND is very likely to be useless, so our KnownBits call is likely worth it.
- The G_AND and G_CONSTANT can interrupt e.g. G_UADDE/... sequences generated during legalization of wide adds which makes it harder to detect these sequences in the instruction selector (e.g. useful to prevent unnecessary reloading of AArch64 NZCV register).
- This cleans up a lot of legalizer output and even improves compilation-times.
AArch64 CTMark geomean: `O0` -5.6% size..text; `O0` and `O3` ~-0.9% compilation-time (instruction count).
Since this introduces KnownBits into code-paths used by `O0`, I reduced the default recursion depth.
This doesn't seem to make a difference in CTMark, but should prevent excessive recursive calls in the worst case.
Reviewed By: aemerson
Differential Revision: https://reviews.llvm.org/D159140
There isn't a test for this yet since the combines aren't used atm, but it will
be tested as part of a future commit. I'm just making this a separate change
tidyness reasons.
Given a list of constraints for InlineAsm (ex. "imr") I'm looking to
modify the order in which they are chosen. Before doing so, I noticed a
fair
amount of logic is duplicated between SelectionDAGISel and GlobalISel
for this.
That is because SelectionDAGISel is also trying to lower immediates
during selection. If we detangle these concerns into:
1. choose the preferred constraint
2. attempt to lower that constraint
Then we can slide down the list of constraints until we find one that
can be lowered. That allows the implementation to be shared between
instruction selection frameworks.
This makes it so that later I might only need to adjust the priority of
constraints in one place, and have both selectors behave the same.
