Since there is no "Load-and-Test-High" instruction, the 32 bit load of a
register to be compared with 0 can only be implemented with LT if the virtual
GRX32 register ends up in a low part (GR32 register).
This patch detects these cases and passes the GR32 registers (low parts) as
(soft) hints in getRegAllocationHints().
Review: Ulrich Weigand.
llvm-svn: 354935
SITargetLowering::reassociateScalarOps() does not touch constants
so that DAGCombiner::ReassociateOps() does not revert the combine.
However a global address is not a ConstantSDNode.
Switched to the method used by DAGCombiner::ReassociateOps() itself
to detect constants.
Differential Revision: https://reviews.llvm.org/D58695
llvm-svn: 354926
Original implementation can't correctly handle __m256 and __m512 types
passed by reference through stack. This patch fixes it.
Patch by Wei Xiao!
Differential Revision: https://reviews.llvm.org/D57643
llvm-svn: 354921
This patch enables the following
1) AMD family 17h "znver2" tune flag (-march, -mcpu).
2) ISAs that are enabled for "znver2" architecture.
3) For the time being, it uses the znver1 scheduler model.
4) Tests are updated.
5) Scheduler descriptions are yet to be put in place.
Reviewers: craig.topper
Differential Revision: https://reviews.llvm.org/D58343
llvm-svn: 354897
This patch aims to make sure that any such constant that can be generated
with a vector instruction (for example VGBM) is recognized as such during
legalization and kept as a target independent node through post-legalize
DAGCombining.
Two new functions named isVectorConstantLegal() and loadVectorConstant()
replace old ways of handling vector/FP constants.
A new struct named SystemZVectorConstantInfo is used to cache the results of
isVectorConstantLegal() and pass them onto loadVectorConstant().
Support for fp128 constants in the presence of FeatureVectorEnhancements1
(z14) has been added.
Review: Ulrich Weigand
https://reviews.llvm.org/D58270
llvm-svn: 354896
If SADDSAT/SSUBSAT are legal, then we can expand SADDO/SSUBO by performing a ADD/SUB and a SADDO/SSUBO and then compare the results.
I looked at doing this for UADDO/USUBO as well but as we don't have to do as many range comparisons I didn't see any/much benefit.
Differential Revision: https://reviews.llvm.org/D58637
llvm-svn: 354866
For outgoing varargs arguments, it's necessary to check the OrigAlign field
of the corresponding OutputArg entry to determine argument alignment, rather
than just computing an alignment from the argument value type. This is
because types like fp128 are split into multiple argument values, with
narrower types that don't reflect the ABI alignment of the full fp128.
This fixes the printf("printfL: %4.*Lf\n", 2, lval); testcase.
Differential Revision: https://reviews.llvm.org/D58656
llvm-svn: 354846
Summary: We shouldn't delete elements while iterating a ranged for loop.
Reviewers: dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58519
llvm-svn: 354844
Summary:
- Indent check lines to easily figure out try-catch-end structure
- Add the original C++ code the tests were genereated from
- Add a few more lines to make the structure more readable
- Rename a couple function / structures
- Add label and branch annotations to cfg-stackify-eh.ll
- Temporarily delete check lines for `test1` in `cfg-stackify-eh.ll`
because it will be updated in a later CL soon and there's no point of
making it look better here
Reviewers: dschuff
Subscribers: sunfish, sbc100, jgravelle-google, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58562
llvm-svn: 354842
Summary:
Use a custom calling convention handler for interrupts instead of fixing
up the locations in LowerMemArgument. This way, the offsets are correct
when constructed and we don't need to account for them in as many
places.
Depends on D56883
Replaces D56275
Reviewers: craig.topper, phil-opp
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D56944
llvm-svn: 354837
Try to use concat_vectors. Also remove unnecessary assert on
pointers. Fixes asserting for <4 x s16> operations and 64-bit pointers
for AMDGPU.
llvm-svn: 354828
If the LHS has known zeros, the RHS immediate will have had bits removed. So call computeKnownBits to get the known zeroes so we can handle this case.
Differential Revision: https://reviews.llvm.org/D58475
llvm-svn: 354811
Fixes https://bugs.llvm.org/show_bug.cgi?id=40325 by zero extending
(and x, 1) the condition before branching on it.
To avoid regressing trivial cases, I'm combining emission of cmp+br
sequences for the single-use + same block case (similar to what we
do in x86). icmpbr1.ll still regresses due to the cross-bb usage
of the condition.
Differential Revision: https://reviews.llvm.org/D58576
llvm-svn: 354808
Support undef shuffle mask indices in the shuffle(concat_vectors, concat_vectors) -> concat_vectors fold
Differential Revision: https://reviews.llvm.org/D58585
llvm-svn: 354793
This adds a few extra Thumb1 opcodes to improve the peephole opimisers
ability to remove redundant cmp instructions. tADC and tSBC require
a small fixup to prevent MOVS being moved past the instruction, giving
the wrong flags.
Differential Revision: https://reviews.llvm.org/D58281
llvm-svn: 354791
Avoid ADD/SUB instruction duplication by reusing the X86ISD::ADD/SUB results.
Includes ADD commutation - I tried to include NEG+SUB SUB commutation as well but this causes regressions as we don't have good combine coverage to simplify X86ISD::SUB.
Differential Revision: https://reviews.llvm.org/D58597
llvm-svn: 354771
More or less all the instructions defined in the v8.2a full-fp16
extension are defined as UNPREDICTABLE if you put them in an IT block
(Thumb) or use with any condition other than AL (ARM). LLVM didn't
know that, and was happy to conditionalise them.
In order to force these instructions to count as not predicable, I had
to make a small Tablegen change. The code generation back end mostly
decides if an instruction was predicable by looking for something it
can identify as a predicate operand; there's an isPredicable bit flag
that overrides that check in the positive direction, but nothing that
overrides it in the negative direction.
(I considered the alternative approach of actually removing the
predicate operand from those instructions, but thought that it would
be more painful overall for instructions differing only in data type
to have different shapes of operand list. This way, the only code that
has to notice the difference is the if-converter.)
So I've added an isUnpredicable bit alongside isPredicable, and set
that bit on the right subset of FP16 instructions, and also on the
VSEL, VMAXNM/VMINNM and VRINT[ANPM] families which should be
unpredicable for all data types.
I've included a couple of representative regression tests, both of
which previously caused an fp16 instruction to be conditionalised in
ARM state and (with -arm-no-restrict-it) to be put in an IT block in
Thumb.
Reviewers: SjoerdMeijer, t.p.northover, efriedma
Reviewed By: efriedma
Subscribers: jdoerfert, javed.absar, kristof.beyls, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D57823
llvm-svn: 354768
Summary:
Fast selection of llvm fptoi & fptrunc instructions is not handled well about
VSX instruction support.
We'd use VSX float convert integer instruction instead of non-vsx float convert
integer instruction if the operand register class is VSSRC or VSFRC because i32
and i64 are mapped to VSSRC and VSFRC correspondingly if VSX feature is
openeded.
For float trunc instruction, we do this silimar work like float convert integer
instruction to try to use VSX instruction.
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D58430
llvm-svn: 354762
Its proving tricky to combine shuffles across multiple vector sizes, so for now I'm adding this more specific combine - the pattern is common enough to be worth it as a first step.
llvm-svn: 354757
Summary:
Previously we used BLENDPS/BLENDPD but that puts the blend in the FP domain. Under optsize, the two address instruction pass can cause blendps/blendpd to commute to blendps/blendpd. But we probably shouldn't do that if the original type was a integer. So use pblendw instead.
Reviewers: spatel, RKSimon
Reviewed By: RKSimon
Subscribers: jdoerfert, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58574
llvm-svn: 354755
Summary:
When promoting the over flow vector for these ops we should use the target's desired setcc result type. This way a v8i32 result type will use a v8i32 overflow vector instead of a v8i16 overflow vector. A v8i16 overflow vector will cause LegalizeDAG/LegalizeVectorOps to have to use v8i32 and truncate to v8i16 in its expansion. By doing this in type legalization instead, we get the truncate into the DAG earlier and give DAG combine more of a chance to optimize it.
We also have to fix unrolling to use the scalar setcc result type for the scalarized operation, and convert it to the required vector element type after the scalar operation. We have to observe the vector boolean contents when doing this conversion. The previous code was just taking the scalar result and putting it in the vector. But for X86 and AArch64 that would have only put a the boolean value in bit 0 of the element and left all other bits in the element 0. We need to ensure all bits in the element are the same. I'm using a select with constants here because that's what setcc unrolling in LegalizeVectorOps used.
Reviewers: spatel, RKSimon, nikic
Reviewed By: nikic
Subscribers: javed.absar, kristof.beyls, dmgreen, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58567
llvm-svn: 354753
There's likely a missed IR canonicalization for at least 1 of these
patterns. Otherwise, we wouldn't have needed the pattern-matching
enhancement in D57516.
Note that -- unlike usubo added with D57789 -- the TLI hook for
this transform defaults to 'on'. So if there's any perf fallout
from this, targets should look at how they're lowering the uaddo
node in SDAG and/or override that hook.
The x86 diffs suggest that there's some missing pattern-matching
for forming inc/dec.
This should fix the remaining known problems in:
https://bugs.llvm.org/show_bug.cgi?id=40486https://bugs.llvm.org/show_bug.cgi?id=31754
llvm-svn: 354746
The new instruciton might have less operands than the original instruction. If we don't resample, the next loop iteration might read an operand that doesn't exist.
X86 can commute blends to movss/movsd which reduces from 4 operands to 3. This happened in the test case that caused r354363 & company to be reverted. A reduced version of that has been committed here.
Really this whole checking for more commutable operands is a little fragile. It assumes that the new instructions operands are the same order and positions as the original except for the pair that was swapped. I don't know of anything that breaks this assumption today, but I've left a fixme. Fixing this will likely require an interface change.
llvm-svn: 354738
r354648 was a follow up to fix a regression "[X86] Add a DAG combine for (aext_vector_inreg (aext_vector_inreg X)) -> (aext_vector_inreg X) to fix a regression from my previous commit."
These were reverted in r354713 as their context depended on other patches that were reverted for a bug.
llvm-svn: 354734
Fixes https://bugs.llvm.org/show_bug.cgi?id=40805 introduced by
patterns added in D53676.
I'm removing the patterns entirely here, as they are not correct
in the general case. If necessary something more specific can be
added in the future.
Differential Revision: https://reviews.llvm.org/D58575
llvm-svn: 354733
Even on AVX1 we can pretty cheaply (VPERM2F128+VSHUFPD) permute a single v4f64/v4i64 input (on AVX2 its just a single VPERMPD), followed by a BLENDPD.
llvm-svn: 354729
Conversion from ConstantSDNode to MachineInstr sign extends immediates from their APInt representation to int64_t.
This commit makes sure we do the same for commuting. The tests changes show how this improves CSE. This issue was made worse by the MachineCSE using commuteInstruction to undo a commute. So we virtually guarantee the sign extend from isel would be lost.
The improved CSE also occurred with r354363, but that was reverted. I'm working to undo the revert, but wanted to get this fix in while it was easy to see the results.
llvm-svn: 354724
r354363 caused https://crbug.com/934963#c1, which has a plain C reduced
test case.
I also had to revert some dependent changes:
- r354648
- r354647
- r354640
- r354511
llvm-svn: 354713
If the the input type will be promoted to 128 bits its better to put a sign_extend_inreg/and in the 128 bit register before the split occurs. Otherwise we end up doing it on each half in the wider register.
Some of the overflow arithmetic tests are regressions, but I think we can make some improvement using getSetccResultType in DAG combine and/or type legalization.
llvm-svn: 354709
If v8i64 isn't a legal type but v4i64 is, these will be split and then each half will get their input promoted and become an any_extend_vector_inreg/punpckhwd + any_extend + and/sign_extend_inreg.
If we instead recognize the input will be promoted we can emit the and/sign_extend_inreg first in a 128 bit register. Then we can sign_extend/zero_extend one half and pshufd+sign_extend/zero_extend the other half.
llvm-svn: 354708
This patch factor out the function hasViableTopFallthrough from rotateLoop. It is also enhanced. Original code checks only if there is a block can be placed before current loop top. This patch also checks if the loop top is the most possible successor of its predecessor. The attached test case shows its effect.
Differential Revision: https://reviews.llvm.org/D58393
llvm-svn: 354682
As discussed in:
D56864
D58197
Always use the narrow (128-bit) instruction when possible.
We already had the signed int version of this transform.
llvm-svn: 354675
Filling a delay slot in 32bit jump instructions with a 16bit instruction
can cause issues. According to the documentation such an operation is
unpredictable.
This patch adds opcode Mips::PseudoIndirectBranch_MM alongside
Mips::PseudoIndirectBranch and other instructions that are expanded to jr
instruction and do not allow a 16bit instruction in their delay slots.
Patch by Mirko Brkusanin.
Differential Revision: https://reviews.llvm.org/D58507
llvm-svn: 354672
This patch adds LazyValueInfo to LowerSwitch to compute the range of the
value being switched over and reduce the size of the tree LowerSwitch
builds to lower a switch.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D58096
llvm-svn: 354670
This adds a number of missing Thumb1 opcodes so that the peephole optimiser can
remove redundant CMP instructions.
Reapplying this after the first attempt broke non-thumb1 code as the t2ADDri
instruction can be used with frame indices. In thumb1 we use tADDframe.
Differential Revision: https://reviews.llvm.org/D57833
llvm-svn: 354667
This is exactly the same as arm mode, so for the instruction selector
tests we just extract them to a new file and run with the same checks
for both arm and thumb mode.
For the legalizer we need to update the tests for soft float a bit, but
only because BL and tBL are slightly different. We could be pedantic and
check that we get a well-formed BL for arm mode and a tBL for thumb, but
for the purposes of the legalizer test it's sufficient to just skip over
the predicate operands in the checks. Also note that we have the
pedantic checks in the divmod test, so we're covered.
llvm-svn: 354665
When we need to merge two adjacent loads the AND mask for the low piece was still sized for the full src element size. But we didn't have that many bits. The upper bits are already zero due to the SRL. So we can skip the AND if we're going to combine with the high bits.
We do need an AND to clear out any bits from the high part. We were anding the high part before combining with the low part, but it looks like ANDing after the OR gets better results.
So we can just emit the final AND after the optional concatentation is done. That will handling skipping before the OR and get rid of extra high bits after the OR.
llvm-svn: 354655
If the scalar type doesn't divide evenly into the WideVT then the code will need to take some bits from adjacent scalar loads and combine them.
But most of our testing is for i1 element type which always divides evenly.
llvm-svn: 354653
Otherwise we end up creating extract_vector_elts that then each need to have their input promoted. This can lead to truncates needing to be emitted for each of those.
But we already emitted any_extends when we legalized the extract_subvector. So now we have pairs of any_extend+trunc that partially cancel. But depending on how DAGCombiner visits them we can get weird results.
By promoting the input at the same time we can create only a single any_extend or truncate.
There's one regression in the vector-narrow-binop.ll case, but that looks easy to fix with a follow up patch.
llvm-svn: 354647
This was done in r321424 to prevent scheduling from reordering things. But now that we model FPCW as a dependency, I don't think the same scheduling we were trying to prevent can occur.
llvm-svn: 354628
This is a follow-up to D56864.
If we're extracting from a non-zero index before casting to FP,
then shuffle the vector and optionally narrow the vector before doing the cast:
cast (extelt V, C) --> extelt (cast (extract_subv (shuffle V, [C...]))), 0
This might be enough to close PR39974:
https://bugs.llvm.org/show_bug.cgi?id=39974
Differential Revision: https://reviews.llvm.org/D58197
llvm-svn: 354619
The trap instruction is intercepted by various runtime environments,
and instead of a crash it creates confusion.
This reapplies r354606 with a fix.
llvm-svn: 354611
This fold can occur during legalization, so it can fight with promotion
to the larger type. It apparently takes a special sequence and subtarget
to avoid more basic simplifications that would hide the problem.
But there's a bigger question raised here: why does distributeTruncateThroughAnd()
even exist? It duplicates functionality from a more minimal pattern that we
already have. But getting rid of this function requires some preliminary steps.
https://bugs.llvm.org/show_bug.cgi?id=40793
llvm-svn: 354594
This adds a number of missing Thumb1 opcodes so that the peephole optimiser can
remove redundant CMP instructions.
Differential Revision: https://reviews.llvm.org/D57833
llvm-svn: 354564
During type promotion, sometimes we convert negative an add with a
negative constant into a sub with a positive constant. The loop that
performs this transformation has two issues:
- it iterates over a set, causing non-determinism.
- it breaks, instead of continuing, when it finds the first
non-negative operand.
Differential Revision: https://reviews.llvm.org/D58452
llvm-svn: 354557
Previously if we couldn't derive a prototype for a "no-prototype"
function from C we would leave it as is:
void foo(...)
With this change we instead give is an empty signature and remove
the "no-prototype" attribute.
This fixes the current wasm waterfall test failure.
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D58488
llvm-svn: 354544
and greendragon also passes, but some other bots fail for reasons I don't understand.
The only difference I can see between these tests is it's missing an -O0
If this doesn't work I'll revert and continue investigating.
llvm-svn: 354529
When we can't determine with certainty the signature of a function
import we pick the fist signature we find rather than error'ing out.
The resulting program might not do what is expected since we might pick
the wrong signature. However since undefined behavior in C to use the
same function with different signatures this seems better than refusing
to compile such programs.
Fixes PR40472
Differential Revision: https://reviews.llvm.org/D58304
llvm-svn: 354523
This change makes some basic type combinations for G_SHUFFLE_VECTOR legal, and
implements them with a very pessimistic TBL2 instruction in the selector.
For TBL2, support is also needed to generate constant pool entries and load from
them in order to materialize the mask register.
Currently supports <2 x s64> and <4 x s32> result types.
Differential Revision: https://reviews.llvm.org/D58466
llvm-svn: 354521
We can currently remove the mask if the immediate has all ones in the LSBs, but if the LHS of the AND is known zero, then the immediate might have had bits removed.
A similar issue also occurs with shifts and rotates. I'm preparing a common fix for all of them.
llvm-svn: 354520
If the LHS has known zeros, then the RHS immediate mask might have been simplified to remove those bits.
This patch adds a call to computeKnownBits to get the known zeroes to handle that possibility. I left an early out to skip the call if all of the demanded bits are set in the mask.
Differential Revision: https://reviews.llvm.org/D58464
llvm-svn: 354514
Second part of https://bugs.llvm.org/show_bug.cgi?id=40442.
This adds an extra UnrollVectorOverflowOp() method to SDAG, because
the general UnrollOverflowOp() method can't deal with multiple results.
Additionally we need to expand UMULO/SMULO during vector op
legalization, as it may result in unrolling, which may need additional
type legalization.
Differential Revision: https://reviews.llvm.org/D57997
llvm-svn: 354513
This avoids depending on the peephole pass to do load folding.
Also adds some load folding for some insert_subvector patterns that use blend.
All of this was found by temporarily adding TB_NO_FORWARD to the blend immediate entries in the load folding tables.
I've added -disable-peephole to some of the affected tests from that experiment to ensure we're testing isel patterns.
llvm-svn: 354511
If the bit position has known zeros in it, then the AND immediate will likely be optimized to remove bits.
This can prevent GetDemandedBits from recognizing that the AND is unnecessary.
llvm-svn: 354501
If the bit position has known zeros in it, then the AND immediate will likely be optimized to remove bits.
This can prevent GetDemandedBits from recognizing that the AND is unnecessary.
llvm-svn: 354498
Jonas Paulsson