Unpredicated FP-multiply of SVE vector with a vector-element given by
vector[index], for example:
fmul z0.s, z1.s, z2.s[0]
which performs an unpredicated FP-multiply of all 32-bit elements in
'z1' with the first element from 'z2'.
This patch adds restricted register classes for SVE vectors:
ZPR_3b (only z0..z7 are allowed) - for indexed vector of 16/32-bit elements.
ZPR_4b (only z0..z15 are allowed) - for indexed vector of 64-bit elements.
Reviewers: rengolin, fhahn, SjoerdMeijer, samparker, javed.absar
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D48823
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336205 91177308-0d34-0410-b5e6-96231b3b80d8
Now that D45806 has landed, we can re-enable support for MIN_SIGNED_VALUE in the sdiv by pow2-constant code
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336198 91177308-0d34-0410-b5e6-96231b3b80d8
This is the last significant change suggested in PR37806:
https://bugs.llvm.org/show_bug.cgi?id=37806#c5
...though there are several follow-ups noted in the code comments
in this patch to complete this transform.
It's possible that a binop feeding a select-shuffle has been eliminated
by earlier transforms (or the code was just written like this in the 1st
place), so we'll fail to match the patterns that have 2 binops from:
D48401,
D48678,
D48662,
D48485.
In that case, we can try to materialize identity constants for the remaining
binop to fill in the "ghost" lanes of the vector (where we just want to pass
through the original values of the source operand).
I added comments to ConstantExpr::getBinOpIdentity() to show planned follow-ups.
For now, we only handle the 5 commutative integer binops (add/mul/and/or/xor).
Differential Revision: https://reviews.llvm.org/D48830
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336196 91177308-0d34-0410-b5e6-96231b3b80d8
This adds the following system registers:
- RAS registers,
- MPAM registers,
- Activitiy monitor registers,
- Trace Extension registers,
- Timing insensitivity of data processing instructions,
- Enhanced Support for Nested Virtualization.
Differential Revision: https://reviews.llvm.org/D48871
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336193 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
When salvaging a dbg.declare/dbg.addr we should not add
DW_OP_stack_value to the DIExpression
(see test/Transforms/InstCombine/salvage-dbg-declare.ll).
Consider this example
%vla = alloca i32, i64 2
call void @llvm.dbg.declare(metadata i32* %vla, metadata !1, metadata !DIExpression())
Instcombine will turn it into
%vla1 = alloca [2 x i32]
%vla1.sub = getelementptr inbounds [2 x i32], [2 x i32]* %vla, i64 0, i64 0
call void @llvm.dbg.declare(metadata [2 x i32]* %vla1.sub, metadata !19, metadata !DIExpression())
If the GEP can be eliminated, then the dbg.declare will be salvaged
and we should get
%vla1 = alloca [2 x i32]
call void @llvm.dbg.declare(metadata [2 x i32]* %vla1, metadata !19, metadata !DIExpression())
The problem was that salvageDebugInfo did not recognize dbg.declare
as being indirect (%vla1 points to the value, it does not hold the
value), so we incorrectly got
call void @llvm.dbg.declare(metadata [2 x i32]* %vla1, metadata !19, metadata !DIExpression(DW_OP_stack_value))
I also made sure that llvm::salvageDebugInfo and
DIExpression::prependOpcodes do not add DW_OP_stack_value to
the DIExpression in case no new operands are added to the
DIExpression. That way we avoid to, unneccessarily, turn a
register location expression into an implicit location expression
in some situations (see test11 in test/Transforms/LICM/sinking.ll).
Reviewers: aprantl, vsk
Reviewed By: aprantl, vsk
Subscribers: JDevlieghere, llvm-commits
Differential Revision: https://reviews.llvm.org/D48837
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336191 91177308-0d34-0410-b5e6-96231b3b80d8
unswitching loops.
Original patch trying to address this was sent in D47624, but that
didn't quite handle things correctly. There are two key principles used
to select whether and how to invalidate SCEV-cached information about
loops:
1) We must invalidate any info SCEV has cached before unswitching as we
may change (or destroy) the loop structure by the act of unswitching,
and make it hard to recover everything we want to invalidate within
SCEV.
2) We need to invalidate all of the loops whose CFGs are mutated by the
unswitching. Notably, this isn't the *entire* loop nest, this is
every loop contained by the outermost loop reached by an exit block
relevant to the unswitch.
And we need to do this even when doing trivial unswitching.
I've added more focused tests that directly check that SCEV starts off
with imprecise information and after unswitching (and simplifying
instructions) re-querying SCEV will produce precise information. These
tests also specifically work to check that an *outer* loop's information
becomes precise.
However, the testing here is still a bit imperfect. Crafting test cases
that reliably fail to be analyzed by SCEV before unswitching and succeed
afterward proved ... very, very hard. It took me several hours and
careful work to build these, and I'm not optimistic about necessarily
coming up with more to cover more elaborate possibilities. Fortunately,
the code pattern we are testing here in the pass is really
straightforward and reliable.
Thanks to Max Kazantsev for the initial work on this as well as the
review, and to Hal Finkel for helping me talk through approaches to test
this stuff even if it didn't come to much.
Differential Revision: https://reviews.llvm.org/D47624
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336183 91177308-0d34-0410-b5e6-96231b3b80d8
This patch changes order of transform in InstCombineCompares to avoid
performing transforms based on ranges which produce complex bit arithmetics
before more simple things (like folding with constants) are done. See PR37636
for the motivating example.
Differential Revision: https://reviews.llvm.org/D48584
Reviewed By: spatel, lebedev.ri
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336172 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
When we import an alias (which will import a copy of the aliasee), but
aren't going to import the aliasee directly, the distributed backend
index will not contain the aliasee summary. Handle this in the summary
assembly printer by printing "null" as the aliasee.
Reviewers: davidxl, dexonsmith
Subscribers: mehdi_amini, inglorion, eraman, steven_wu, llvm-commits
Differential Revision: https://reviews.llvm.org/D48699
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336160 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
In the individual index files emitted for distributed ThinLTO backends,
the module path ids are not contiguous. Assign slots to module paths in
order to handle this better and also to get contiguous numbering in the
summary assembly.
Reviewers: davidxl, dexonsmith
Subscribers: mehdi_amini, inglorion, eraman, llvm-commits, steven_wu
Differential Revision: https://reviews.llvm.org/D48698
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336148 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Comment on Transforms/LoopVersioning/incorrect-phi.ll: With the change
SCEV is able to prove that the loop doesn't wrap-self (due to zext i16
to i64), disabling the entire loop versioning pass. Removed the zext and
just use i64.
Reviewers: sanjoy
Subscribers: jlebar, hiraditya, javed.absar, bixia, llvm-commits
Differential Revision: https://reviews.llvm.org/D48409
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336140 91177308-0d34-0410-b5e6-96231b3b80d8
LLVM doesn't guarantee anything about the high bits of a register holding
an i1 value at the IR level, so don't translate LLVM IR i1 values directly
into WebAssembly conditional branch operands. WebAssembly's conditional
branches do demand all 32 bits be valid.
Fixes PR38019.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336138 91177308-0d34-0410-b5e6-96231b3b80d8
Add registers still missing after r328016 (D43353):
- for bits 15-8 of SI, DI, BP, SP (*H), and R8-R15 (*BH),
- for bits 31-16 of R8-R15 (*WH).
Thanks to Craig Topper for pointing it out.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336134 91177308-0d34-0410-b5e6-96231b3b80d8
Summary: It is common to have the following min/max pattern during the intermediate stages of SLP since we only optimize at the end. This patch tries to catch such patterns and allow more vectorization.
%1 = extractelement <2 x i32> %a, i32 0
%2 = extractelement <2 x i32> %a, i32 1
%cond = icmp sgt i32 %1, %2
%3 = extractelement <2 x i32> %a, i32 0
%4 = extractelement <2 x i32> %a, i32 1
%select = select i1 %cond, i32 %3, i32 %4
Author: FarhanaAleen
Reviewed By: ABataev, RKSimon, spatel
Differential Revision: https://reviews.llvm.org/D47608
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336130 91177308-0d34-0410-b5e6-96231b3b80d8
This extends D48485 to allow another pair of binops (add/or) to be combined either
with or without a leading shuffle:
or X, C --> add X, C (when X and C have no common bits set)
Here, we need value tracking to determine that the 'or' can be reversed into an 'add',
and we've added general infrastructure to allow extending to other opcodes or moving
to where other passes could use that functionality.
Differential Revision: https://reviews.llvm.org/D48662
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336128 91177308-0d34-0410-b5e6-96231b3b80d8
On darwin, all virtual sections have zerofill type, and having a
.zerofill directive in a non-virtual section is not allowed. Instead of
asserting, show a nicer error.
In order to use the equivalent of .zerofill in a non-virtual section,
the usage of .zero of .space is required.
This patch replaces the assert with an error.
Differential Revision: https://reviews.llvm.org/D48517
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336127 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
This adds a new -no-weak flag to nm to hide weak symbols in its output.
This also adds a -W alias for this which is analogous to -U.
Patch by Keith Smiley
Reviewers: kastiglione, enderby, compnerd
Reviewed By: kastiglione
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D48751
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336126 91177308-0d34-0410-b5e6-96231b3b80d8
Alternate opcode handling only supports binary operators, these tests demonstrate a missed opportunity to vectorize ceil/floor calls
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336125 91177308-0d34-0410-b5e6-96231b3b80d8
Use an -implicit-check-not to make sure an error which should not occur
in fact does not occur before the first CHECK line.
Suggested by Paul Robinson in post-commit feedback for r335897.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336123 91177308-0d34-0410-b5e6-96231b3b80d8
Similarily, don't fold fp128 loads into SSE instructions if the load isn't aligned. Unless we're targeting an AMD CPU that doesn't check alignment on arithmetic instructions.
Should fix PR38001
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336121 91177308-0d34-0410-b5e6-96231b3b80d8
We currently don't any-extend vararg parameters before storing them to the stack
locations on Darwin. However, SelectionDAG however does this, and so user code
is in the wild which inadvertently relies on this extension. This can manifest
in cases where the value stored is (int)0, but the actual parameter is interpreted
by va_arg as a pointer, and so not extending to 64 bits causes the callee to
load additional undefined bits.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336120 91177308-0d34-0410-b5e6-96231b3b80d8
We were only doing this for basic blends, despite shuffle lowering now being good enough to handle more complex blends. This means that the two v8i16 splat shifts are performed in parallel instead of serially as the general shift case.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336113 91177308-0d34-0410-b5e6-96231b3b80d8
We have special case support for 2 shift values for basic blends, but irregular shift patterns end up using the generic lowering, despite shuffle lowering being good enough to handle more complex blends.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336112 91177308-0d34-0410-b5e6-96231b3b80d8
This was my mistake for only running test/MC/X86 and test/CodeGen/X86.
Arguably .word should be removed from this test, as it is not supported
universally.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336107 91177308-0d34-0410-b5e6-96231b3b80d8
Due to current limitations in constant analysis, we need flags
on add or mul to show propagation for the potential transform
suggested in these tests (no other binops currently report
identity constants).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336101 91177308-0d34-0410-b5e6-96231b3b80d8
This version contains a fix to add values for which the state in ParamState change
to the worklist if the state in ValueState did not change. To avoid adding the
same value multiple times, mergeInValue returns true, if it added the value to
the worklist. The value is added to the worklist depending on its state in
ValueState.
Original message:
For comparisons with parameters, we can use the ParamState lattice
elements which also provide constant range information. This improves
the code for PR33253 further and gets us closer to use
ValueLatticeElement for all values.
Also, as we are using the range information in the solver directly, we
do not need tryToReplaceWithConstantRange afterwards anymore.
Reviewers: dberlin, mssimpso, davide, efriedma
Reviewed By: mssimpso
Differential Revision: https://reviews.llvm.org/D43762
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336098 91177308-0d34-0410-b5e6-96231b3b80d8
We were always using the opcodes of the first 2 scalars for the costs of the alternate opcode + shuffle. This made sense when we used SK_Alternate and opcodes were guaranteed to be alternating, but this fails for the more general SK_Select case.
This fix exposes an issue demonstrated by the fmul_fdiv_v4f32_const test - the SLM model has v4f32 fdiv costs which are more than twice those of the f32 scalar cost, meaning that the cost model determines that the vectorization is not performant. Unfortunately it completely ignores the fact that the fdiv by a constant will be changed into a fmul by InstCombine for a much lower cost vectorization. But at least we're seeing this now...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336095 91177308-0d34-0410-b5e6-96231b3b80d8
Increment/decrement vector by multiple of predicate constraint
element count.
The variants added by this patch are:
- INCH, INCW, INC
and (saturating):
- SQINCH, SQINCW, SQINCD
- UQINCH, UQINCW, UQINCW
- SQDECH, SQINCW, SQINCD
- UQDECH, UQINCW, UQINCW
For example:
incw z0.s, all, mul #4
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336090 91177308-0d34-0410-b5e6-96231b3b80d8
This change fixes the issue that arises when we duplicate condition from
the predecessor block. If the condition's arguments are not considered alive
across the blocks, fast regalloc gets confused and starts generating reloads
from the slots that have never been spilled to. This change also leads to
smaller code given that, unlike on architectures with condition codes, on
Mips we can branch directly on register value, thus we gain nothing by
duplication.
Patch by Dragan Mladjenovic.
Differential Revision: https://reviews.llvm.org/D48642
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336084 91177308-0d34-0410-b5e6-96231b3b80d8
These patches were previously reverted as they led to
buildbot time-outs caused by large switch statement in
printAliasInstr when using UBSan and O3. The issue has
been addressed with a workaround (r335525).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336079 91177308-0d34-0410-b5e6-96231b3b80d8
For the below case, pre-inc prep think it's a good candidate to use pre-inc for the bucket, but 64bit integer load/store update (pre-inc) instruction on Power requires the displacement field should be DS-form (4's multiple). Since it can't satisfy the constraint, we have to do some fix ups later. As below, the original load/stores could be well-form, it makes things worse.
unsigned long long result = 0;
unsigned long long foo(char *p, unsigned long long n) {
for (unsigned long long i = 0; i < n; i++) {
unsigned long long x1 = *(unsigned long long *)(p - 50000 + i);
unsigned long long x2 = *(unsigned long long *)(p - 61024 + i);
unsigned long long x3 = *(unsigned long long *)(p - 62048 + i);
unsigned long long x4 = *(unsigned long long *)(p - 64096 + i);
result *= x1 * x2 * x3 * x4;
}
return result;
}
Patch by jedilyn(Kewen Lin).
Differential Revision: https://reviews.llvm.org/D48813
--This line, and those below, will be ignored--
M lib/Target/PowerPC/PPCLoopPreIncPrep.cpp
A test/CodeGen/PowerPC/preincprep-i64-check.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@336074 91177308-0d34-0410-b5e6-96231b3b80d8
Sanjay Patel