This is a reimplementation of the optimization removed in D75964. The actual spill/fill optimization is handled by D76013, this one just worries about reducing the stackmap section size itself by eliminating redundant entries. As noted in the comments, we could go a lot further here, but avoiding the degenerate invoke case as we did before is probably "enough" in practice.
Differential Revision: https://reviews.llvm.org/D76021
Summary:
callbr's indirect branches aren't expected to be taken, so reduce their
probabilities to 0 while increasing the default destination to 1. This
allows some code improvements through block placement.
Reviewers: nickdesaulniers
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D72656
We just removed a broken duplicate elimination algorithm in D75964, and after landed that it occurred to me that duplicate elimination is simply CSE. SelectionDAG has a build in CSE, so why wasn't that triggering? Well, it turns out we were overly conservative in the memory states for our reloads and CSE (rightly) considers the incoming memory state for a load part of the identity of the load.
By loosening the chain and allowing reordering, we also allow CSE. As shown in the test case, doing iterative CSE as we go is enough to eliminate duplicate stores in later statepoints as well. We key our (block local) slot map by SDValue, so commoning a previous pair of loads at construction time means we also common following stores.
Differential Revision: https://reviews.llvm.org/D76013
This patch reuses the existing MatchRotate ROTL/ROTR rotation pattern code to also recognize the more general FSHL/FSHR funnel shift patterns when we have constant shift amounts.
Differential Revision: https://reviews.llvm.org/D75114
A downstream test case (see included reduced test) revealed that we have a bug in how we handle duplicate relocations. If we have the same SDValue relocated twice, and that value happens to be a constant (such as null), we only export one of the two llvm::Values. Exporting on a per llvm::Value basis is required to allow lowering of gc.relocates in following basic blocks (e.g. invokes). Without it, we end up with a use of an undefined vreg and bad things happen.
Rather than fixing the optimization - which appears to be hard - I propose we simply remove it. There are no tests in tree that change with this code removed. If we find out later that this did matter for something, we can reimplement a variation of this in CodeGenPrepare to catch the easy cases without complicating the lowering code.
Thanks to Denis and Serguei who did all the hard work of figuring out what went wrong here. The patch is by far the easy part. :)
Differential Revision: https://reviews.llvm.org/D75964
For i32 and i64 cases, X86ISD::SHLD/SHRD are close enough to ISD::FSHL/FSHR that we can use them directly, we just need to account for the operand commutation for SHRD.
The i16 SHLD/SHRD case is annoying as the shift amount is modulo-32 (vs funnel shift modulo-16), so I've added X86ISD::FSHL/FSHR equivalents, which matches the generic implementation in all other terms.
Something I'm slightly concerned with is that ISD::FSHL/FSHR legality is controlled by the Subtarget.isSHLDSlow() feature flag - we don't normally use non-ISA features for this but it allows the DAG combines to continue to operate after legalization in a lot more cases.
The X86 *bits.ll changes are all affected by the same issue - we now have a "FSHR(-1,-1,amt) -> ROTR(-1,amt) -> (-1)" simplification that reduces the dependencies enough for the branch fall through code to mess up.
Differential Revision: https://reviews.llvm.org/D75748
For pre-AVX512 targets, combine binary shuffles to X86ISD::VPERM2X128 if possible. This mainly helps optimize the blend(extract_subvector(x,1),y) pattern.
At some point soon we're going to have make a decision about when to combine AVX512 shuffles more aggressively - we bail out if there is any change in element size (to protect predicate mask merging) which means we miss out on a lot of optimizations.
If we don't need the upper subvector elements of the BLENDI node then use a smaller vector size.
This causes a couple of minor regressions in insertelement-ones.ll which are more examples of PR26018; given how cheap allones generation is I don't consider that a showstopper, just an annoyance (and there's plenty of other poor codegen cases in that file).
If we're inserting a scalar that is smaller than the element
size of the final VT, the value of the extra bits doesn't matter.
Previously we any_extended in the scalar domain before inserting.
This patch changes this to use a broadcast of the original
scalar type and then a bitcast to the final type. This might
enable the use of a broadcast load.
This recovers regressions from 07d68c24aa
and 9fcd212e2f without relying on
alignment of the load.
Differential Revision: https://reviews.llvm.org/D75835
This was selecting VBROADCASTW which turned the 8-bit load into
a 16-bit load if it happened to be 2 byte aligned.
I have a plan to fix the regression with a follow up patch
which I'll post shortly.
In 172eee9c, we tried to avoid these by modelling the callee as
internally resetting the stack pointer.
However, for the majority of functions with reserved stack frames, this
would lead LLVM to emit extra SP adjustments to undo the callee's
internal adjustment. This lead us to fix the problem further on down the
pipeline in eliminateCallFramePseudoInstr. In 5b79e603d3, I added
use a heuristic to try to detect when the adjustment would be
unreachable.
This heuristic is imperfect, and when exception handling is involved, it
fails to fire. The new test is an example of this. Simply throwing an
exception with an active cleanup emits dead SP adjustments after the
throw. Not only are they dead, but if they were executed, they would be
incorrect, so they are confusing.
This change essentially reverts 172eee9c and makes the 5b79e603d3
heuristic responsible for preventing unreachable stack adjustments. This
means we may emit unreachable stack adjustments for functions using EH
with unreserved call frames, but that is not very many these days. Back
in 2016 when this change was added, we were focused on 32-bit, which we
observed to have fewer reserved frames.
Fixes PR45064
Reviewed By: hans
Differential Revision: https://reviews.llvm.org/D75712
As noted on D75114, if both arguments of a funnel shift are consecutive loads we are missing the opportunity to combine them into a single load.
Differential Revision: https://reviews.llvm.org/D75624
The new behavior matches GNU objdump. A pair of angle brackets makes tests slightly easier.
`.foo:` is not unique and thus cannot be used in a `CHECK-LABEL:` directive.
Without `-LABEL`, the CHECK line can match the `Disassembly of section`
line and causes the next `CHECK-NEXT:` to fail.
```
Disassembly of section .foo:
0000000000001634 .foo:
```
Bdragon: <> has metalinguistic connotation. it just "feels right"
Reviewed By: rupprecht
Differential Revision: https://reviews.llvm.org/D75713
Previously we tried to promote these to xmm/ymm/zmm by promoting
in the X86CallingConv.td file. But this breaks when we run out
of xmm/ymm/zmm registers and need to fall back to memory. We end
up trying to create a non-sensical scalar to vector. This lead
to an assertion. The new tests in avx512-calling-conv.ll all
trigger this assertion.
Since we really want to treat these types like we do on avx2,
it seems better to promote them before the calling convention
code gets involved. Except when the calling convention is one
that passes the vXi1 type in a k register.
The changes in avx512-regcall-Mask.ll are because we indicated
that xmm/ymm/zmm types should be passed indirectly for the
Win64 ABI before we go to the common lines that promoted the
vXi1 types. This caused the promoted types to be picked up by
the default calling convention code. Now we promote them earlier
so they get passed indirectly as though they were xmm/ymm/zmm.
Differential Revision: https://reviews.llvm.org/D75154
I believe this is the correct fix for D75506 rather than disabling all commuting. We can still commute the remaining two sources.
Differential Revision:m https://reviews.llvm.org/D75526
https://reviews.llvm.org/D42848 only handled CFA related cfi directives but
didn't handle CSR related cfi. The patch adds the CSR part. Basically it reuses
the framework created in D42848. For each basicblock, the patch tracks which
CSR set have been saved at its CFG predecessors's exits, and compare the CSR
set with the set at its previous basicblock's exit (The previous block is the
block laid before the current block). If the saved CSR set at its previous
basicblock's exit is larger, .cfi_restore will be inserted.
The patch also generates proper .cfi_restore in epilogue to make sure the
saved CSR set is consistent for the incoming edges of each block.
Differential Revision: https://reviews.llvm.org/D74303
As the test case shows if there is an ExtractValueInst in the Ret block, function dupRetToEnableTailCallOpts can't duplicate it into the block containing call. So later no tail call is generated in CodeGen.
This patch adds the ExtractValueInst handling code in function dupRetToEnableTailCallOpts and FoldReturnIntoUncondBranch, and later tail call can be generated for this case.
Differential Revision: https://reviews.llvm.org/D74242
If we go with D75412, we no longer depend on the scalar type directly. So we don't need to avoid using i64. We already have AVX1 fallback patterns with i32 and i64 scalar types so we don't need to avoid using integer types on AVX1.
Differential Revision: https://reviews.llvm.org/D75413
Summary:
Follow up from D75377. If the subvector is byte sized and the
index is aligned to the subvector size, we can shrink the load.
Reviewers: spatel, RKSimon
Reviewed By: RKSimon
Subscribers: dbabokin, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75434
The build_vector needs to be the only user of the data, but the
chain will likely have another use. So we can't make sure the
build_vector is the only user of the node.
We should be careful to allow count of re-materialization of operands to be less
then number of physical registers.
STATEPOINT instruction has a variable number of operands and potentially very big.
So re-materialization for all operands is disabled at the moment if restrict-statepoint-remat is true.
The patch relaxes the re-materialization restriction for STATEPOINT instruction allowing it for
fixed operands. Specifically it is about call target.
Reviewers: reames
Reviewed By: reames
Subscribers: llvm-commits, qcolombet, hiraditya
Differential Revision: https://reviews.llvm.org/D75335
The address calculation for the offset assumes that you can calculate the offset by multiplying the index by the store size of the element. But that only works if the element's store size is exactly its real size since we store vectors tightly packed in memory. There are improvements we could make to this like special casing extracting element 0. I think we could also handle cases where the extracted VT is byte sized and the index is aligned with the extract element count.
Differential Revision: https://reviews.llvm.org/D75377
Select_cc isn't used by all targets. X86 doesn't have optimizations
for it.
Since we already know the input to the sint_to_fp/uint_to_fp is
a setcc we can just emit a plain select using that setcc as the
condition. Other DAG combines can turn that into a select_cc on
targets that support it.
Differential Revision: https://reviews.llvm.org/D75415
We get the simple cases of this via demanded elements and other folds,
but that doesn't work if the values have >1 use, so add a dedicated
match for the pattern.
We already have this transform in IR, but it doesn't help the
motivating x86 tests (based on PR42024) because the shuffles don't
exist until after legalization and other combines have happened.
The AArch64 test shows a minimal IR example of the problem.
Differential Revision: https://reviews.llvm.org/D75348
Simon Pilgrim