Retrying after fixing overly aggressive load-store forwarding optimization.
Simplify Consecutive Merge Store Candidate Search
Now that address aliasing is much less conservative, push through
simplified store merging search which only checks for parallel stores
through the chain subgraph. This is cleaner as the separation of
non-interfering loads/stores from the store-merging logic.
Whem merging stores, search up the chain through a single load, and
finds all possible stores by looking down from through a load and a
TokenFactor to all stores visited. This improves the quality of the
output SelectionDAG and generally the output CodeGen (with some
exceptions).
Additional Minor Changes:
1. Finishes removing unused AliasLoad code
2. Unifies the the chain aggregation in the merged stores across
code paths
3. Re-add the Store node to the worklist after calling
SimplifyDemandedBits.
4. Increase GatherAllAliasesMaxDepth from 6 to 18. That number is
arbitrary, but seemed sufficient to not cause regressions in
tests.
This finishes the change Matt Arsenault started in r246307 and
jyknight's original patch.
Many tests required some changes as memory operations are now
reorderable. Some tests relying on the order were changed to use
volatile memory operations
Noteworthy tests:
CodeGen/AArch64/argument-blocks.ll -
It's not entirely clear what the test_varargs_stackalign test is
supposed to be asserting, but the new code looks right.
CodeGen/AArch64/arm64-memset-inline.lli -
CodeGen/AArch64/arm64-stur.ll -
CodeGen/ARM/memset-inline.ll -
The backend now generates *worse* code due to store merging
succeeding, as we do do a 16-byte constant-zero store efficiently.
CodeGen/AArch64/merge-store.ll -
Improved, but there still seems to be an extraneous vector insert
from an element to itself?
CodeGen/PowerPC/ppc64-align-long-double.ll -
Worse code emitted in this case, due to the improved store->load
forwarding.
CodeGen/X86/dag-merge-fast-accesses.ll -
CodeGen/X86/MergeConsecutiveStores.ll -
CodeGen/X86/stores-merging.ll -
CodeGen/Mips/load-store-left-right.ll -
Restored correct merging of non-aligned stores
CodeGen/AMDGPU/promote-alloca-stored-pointer-value.ll -
Improved. Correctly merges buffer_store_dword calls
CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll -
Improved. Sidesteps loading a stored value and
merges two stores
CodeGen/X86/pr18023.ll -
This test has been removed, as it was asserting incorrect
behavior. Non-volatile stores *CAN* be moved past volatile loads,
and now are.
CodeGen/X86/vector-idiv.ll -
CodeGen/X86/vector-lzcnt-128.ll -
It's basically impossible to tell what these tests are actually
testing. But, looks like the code got better due to the memory
operations being recognized as non-aliasing.
CodeGen/X86/win32-eh.ll -
Both loads of the securitycookie are now merged.
Reviewers: arsenm, hfinkel, tstellarAMD, jyknight, nhaehnle
Subscribers: wdng, nhaehnle, nemanjai, arsenm, weimingz, niravd, RKSimon, aemerson, qcolombet, dsanders, resistor, tstellarAMD, t.p.northover, spatel
Differential Revision: https://reviews.llvm.org/D14834
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289221 91177308-0d34-0410-b5e6-96231b3b80d8
Adds support for bitcasting a little endian 'small element' vector to 'large element' scalar/vector (e.g. v16i8 to v4i32 or v2i32 to i64), which is required for PR30845. We extract the knownbits for each 'small element' part and concatenate the results together.
We can add support for big endian and 'large element' scalar/vector to 'small element' vector bitcasting once we have test cases for them.
Differential Revision: https://reviews.llvm.org/D27129
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289200 91177308-0d34-0410-b5e6-96231b3b80d8
This reverts commit r288916 as it is currently causing a crasher in
Halide. Reproducer on llvm.org/PR31323. While it might be that halide is
generating invalid IR, llc shouldn't crash.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289194 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Scalar intrinsics have specific semantics about the which input's upper bits are passed through to the output. The same input is also supposed to be the input we use for the lower element when the mask bit is 0 in a masked operation. We aren't currently keeping these semantics with instruction selection.
This patch corrects this by introducing new scalar FMA ISD nodes that indicate whether operand 1(one of the multiply inputs) or operand 3(the additon/subtraction input) should pass thru its upper bits.
We use this information to select 213/132 form for the operand 1 version and the 231 form for the operand 3 version.
We also use this information to suppress combining FNEG operations on the passthru input since semantically the passthru bits aren't negated. This is stronger than the earlier check added for a user being SELECTS so we can remove that.
This fixes PR30913.
Reviewers: delena, zvi, v_klochkov
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D27144
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289190 91177308-0d34-0410-b5e6-96231b3b80d8
These were selecting directly to the VOP2 form instead
of VOP3 like the i32 instructions. Fixes regressions in
future commits where an immediate isn't folded because it was
initially used for the second operand.
Because uniform 16-bit operations are promoted to i32, it's
difficult to get a simple testcase where this matters. Fold
failures in SIFoldOperands here tend to be hidden by commute
and fold in SIShrinkInstructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289189 91177308-0d34-0410-b5e6-96231b3b80d8
The motivating example is:
extern int patatino;
int goo() {
int x = 0;
for (int i = 0; i < 1000000; ++i) {
x *= patatino;
}
return x;
}
Currently SCCP will not realize that this function returns always zero,
therefore will try to unroll and vectorize the loop at -O3 producing an
awful lot of (useless) code. With this change, it will just produce:
0000000000000000 <g>:
xor %eax,%eax
retq
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289175 91177308-0d34-0410-b5e6-96231b3b80d8
Supporting them properly is a reasonably complex chunk of work, so to allow bot
testing before then we should at least be able to fall back to DAG ISel.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289150 91177308-0d34-0410-b5e6-96231b3b80d8
Currently SCCP folds the value to -1, while ConstantProp folds to
0. This changes SCCP to do what ConstantFolding does.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289147 91177308-0d34-0410-b5e6-96231b3b80d8
We were falsely claiming that we had an LSDA for the relevant EH
personality before this change, which could lead to the EH machinery
interpreting random adjacent data as an LSDA.
Fixes PR31317
This change is safe because cleanups can't contain exception handlers
today. We do these things to maintain that invariant:
- C++ destructors are naturally out-of-line
- __finally blocks are outlined in clang
- LLVM's inliner will not inline EH constructs into cleanups
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289101 91177308-0d34-0410-b5e6-96231b3b80d8
Not having this legal led to combine failures, resulting
in dumb things like bitcasts of constants not being folded
away.
The only reason I'm leaving the v_mov_b32 hack that f32
already uses is to avoid madak formation test regressions.
PeepholeOptimizer has an ordering issue where the immediate
fold attempt is into the sgpr->vgpr copy instead of the actual
use. Running it twice avoids that problem.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289096 91177308-0d34-0410-b5e6-96231b3b80d8
The correct commutable opcode was set to itself, so this
was simply swapping the operands to commute instead of also
changing the opcode to v_subrev_u16.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289093 91177308-0d34-0410-b5e6-96231b3b80d8
Multiple metadata values for records such as opencl.ocl.version, llvm.ident
and similar are created after linking several modules. For some of them, notably
opencl.ocl.version, this creates semantic problem because we cannot tell which
version of OpenCL the composite module conforms.
Moreover, such repetitions of identical values often create a huge list of
unneeded metadata, which grows bitcode size both in memory and stored on disk.
It can go up to several Mb when linked against our OpenCL library. Lastly, such
long lists obscure reading of dumped IR.
The pass unifies metadata after linking.
Differential Revision: https://reviews.llvm.org/D25381
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289092 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Attaching !absolute_symbol to a global variable does two things:
1) Marks it as an absolute symbol reference.
2) Specifies the value range of that symbol's address.
Teach the X86 backend to allow absolute symbols to appear in place of
immediates by extending the relocImm and mov64imm32 matchers. Start using
relocImm in more places where it is legal.
As previously proposed on llvm-dev:
http://lists.llvm.org/pipermail/llvm-dev/2016-October/105800.html
Differential Revision: https://reviews.llvm.org/D25878
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289087 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
LC can currently select scalar load for uniform memory access
basing on readonly memory address space only. This restriction
originated from the fact that in HW prior to VI vector and scalar caches
are not coherent. With MemoryDependenceAnalysis we can check that the
memory location corresponding to the memory operand of the LOAD is not
clobbered along the all paths from the function entry.
Reviewers: rampitec, tstellarAMD, arsenm
Subscribers: wdng, arsenm, nhaehnle
Differential Revision: https://reviews.llvm.org/D26917
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289076 91177308-0d34-0410-b5e6-96231b3b80d8
ConstantFolding tried to cast one of the scalar indices to a vector
type. Instead, use the vector type only for the first index (which
is the only one allowed to be a vector) and use its scalar type
otherwise.
Fixes PR31250.
Reviewers: majnemer
Differential Revision: https://reviews.llvm.org/D27389
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289073 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Without the fix to isFrameOffsetLegal to consider the instruction's
immediate offset, the new test case hits the corresponding assertion in
resolveFrameIndex, because the LocalStackSlotAllocation pass re-uses a
different base register.
With only the fix to isFrameOffsetLegal, code quality reduces in a bunch of
places because frame base registers are added where they're not needed.
This is addressed by properly implementing needsFrameBaseReg, which also
helps to avoid unnecessary zero frame indices in a bunch of other places.
Fixes piglit glsl-1.50/execution/variable-indexing/gs-output-array-vec4-index-wr.shader_test
Reviewers: arsenm, tstellarAMD
Subscribers: qcolombet, kzhuravl, wdng, yaxunl, tony-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D27344
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289048 91177308-0d34-0410-b5e6-96231b3b80d8
When trying to vectorize trees that start at insertelement instructions
function tryToVectorizeList() uses vectorization factor calculated as
MinVecRegSize/ScalarTypeSize. But sometimes it does not work as tree
cost for this fixed vectorization factor is too high.
Patch tries to improve the situation. It tries different vectorization
factors from max(PowerOf2Floor(NumberOfVectorizedValues),
MinVecRegSize/ScalarTypeSize) to MinVecRegSize/ScalarTypeSize and tries
to choose the best one.
Differential Revision: https://reviews.llvm.org/D27215
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289043 91177308-0d34-0410-b5e6-96231b3b80d8
The relocations for `DIEEntry::EmitValue` were wrong for Win64
(emitting FK_Data_4 instead of FK_SecRel_4). This corrects that
oversight so that the DWARF data is correct in Win64 COFF files.
Fixes PR15393.
Patch by Jameson Nash <jameson@juliacomputing.com> based on a patch
by David Majnemer.
Differential Revision: https://reviews.llvm.org/D21731
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289013 91177308-0d34-0410-b5e6-96231b3b80d8
Replace @progbits in the section directive with %progbits, because "@" starts a comment on arm/thumb.
Use b.w branch instruction.
Use .thumb_function and .thumb_set for proper arm/thumb interwork. This way jumptable entry addresses on thumb have bit 0 set (correctly). This does not affect CFI check math, because the address of the jumptable start also has that bit set.
This does not work on thumbv5, because it does not support b.w, and the linker would not insert a veneer (trampoline?) to extend the range of b.n. We may need to do full-range plt-style jumptables on thumbv54, which are 12 bytes per entry. Another option is "push lr; bl; pop pc" (4 bytes) but that needs unwinding instructions, etc.
Differential Revision: https://reviews.llvm.org/D27499
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@289008 91177308-0d34-0410-b5e6-96231b3b80d8
The fix committed in r288851 doesn't cover all the cases.
In particular, if we have an instruction with side effects
which has a no non-dbg use not depending on the bits, we still
perform RAUW destroying the dbg.value's first argument.
Prevent metadata from being replaced here to avoid the issue.
Differential Revision: https://reviews.llvm.org/D27534
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@288987 91177308-0d34-0410-b5e6-96231b3b80d8
ConstantExpr instances were emitting code into the current block rather than
the entry block. This meant they didn't necessarily dominate all uses, which is
clearly wrong.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@288985 91177308-0d34-0410-b5e6-96231b3b80d8
