This patch makes LSR generate better code for SystemZ in the cases of memory
intrinsics, Load->Store pairs or comparison of immediate with memory.
In order to achieve this, the following common code changes were made:
* New TTI hook: LSRWithInstrQueries(), which defaults to false. Controls if
LSR should do instruction-based addressing evaluations by calling
isLegalAddressingMode() with the Instruction pointers.
* In LoopStrengthReduce: handle address operands of memset, memmove and memcpy
as address uses, and call isFoldableMemAccessOffset() for any LSRUse::Address,
not just loads or stores.
SystemZ changes:
* isLSRCostLess() implemented with Insns first, and without ImmCost.
* New function supportedAddressingMode() that is a helper for TTI methods
looking at Instructions passed via pointers.
Review: Ulrich Weigand, Quentin Colombet
https://reviews.llvm.org/D35262https://reviews.llvm.org/D35049
llvm-svn: 308729
This adds support for the new 128-bit vector float instructions of z14.
Note that these instructions actually only operate on the f128 type,
since only each 128-bit vector register can hold only one 128-bit
float value. However, this is still preferable to the legacy 128-bit
float instructions, since those operate on pairs of floating-point
registers (so we can hold at most 8 values in registers), while the
new instructions use single vector registers (so we hold up to 32
value in registers).
Adding support includes:
- Enabling the instructions for the assembler/disassembler.
- CodeGen for the instructions. This includes allocating the f128
type now to the VR128BitRegClass instead of FP128BitRegClass.
- Scheduler description support for the instructions.
Note that for a small number of operations, we have no new vector
instructions (like integer <-> 128-bit float conversions), and so
we use the legacy instruction and then reformat the operand
(i.e. copy between a pair of floating-point registers and a
vector register).
llvm-svn: 308196
This adds support for the new 32-bit vector float instructions of z14.
This includes:
- Enabling the instructions for the assembler/disassembler.
- CodeGen for the instructions, including new LLVM intrinsics.
- Scheduler description support for the instructions.
- Update to the vector cost function calculations.
In general, CodeGen support for the new v4f32 instructions closely
matches support for the existing v2f64 instructions.
llvm-svn: 308195
This patch series adds support for the IBM z14 processor. This part includes:
- Basic support for the new processor and its features.
- Support for new instructions (except vector 32-bit float and 128-bit float).
- CodeGen for new instructions, including new LLVM intrinsics.
- Scheduler description for the new processor.
- Detection of z14 as host processor.
Support for the new 32-bit vector float and 128-bit vector float
instructions is provided by separate patches.
llvm-svn: 308194
When reusing a register for a new definition, the fast register allocator
used to insert a kill flag at the previous last use of that register to
inform later passes that this register is free between the redef and the
last use. However, this may be wrong when subregisters are involved.
Indeed, a partially redef would have trigger a kill of the full super
register, potentially wrongly marking all the other subregisters as
free. Given we don't track which lanes are still live, we cannot set the
kill flag in such case.
Note: This bug has been latent for about 7 years (r104056).
llvmg.org/PR33677
llvm-svn: 307428
We sometimes need emergency spill slots for the register scavenger.
This may be the case when code needs to access a stack slot that
has an offset of 4096 or more relative to the stack pointer.
To make that determination, processFunctionBeforeFrameFinalized
currently simply checks the total stack frame size of the current
function. But this is not enough, since code may need to access
stack slots in the caller's stack frame as well, in particular
incoming arguments stored on the stack.
This commit fixes the problem by taking argument slots into account.
llvm-svn: 306305
Csmith discovered that this function can be called with a zero argument,
in which case an assert for this triggered.
This patch also adds a guard before the other call to this function since
it was missing, although the test only covers the case where it was
discovered.
Reduced test case attached as CodeGen/SystemZ/int-cmp-54.ll.
Review: Ulrich Weigand
llvm-svn: 306287
This reverts the use of TargetLowering::prepareVolatileOrAtomicLoad
introduced by r196905. Nothing in the semantics of the "volatile"
keyword or the definition of the z/Architecture actually requires
that volatile loads are preceded by a serialization operation, and
no other compiler on the platform actually implements this.
Since we've now seen a use case where this additional serialization
causes noticable performance degradation, this patch removes it.
The patch still leaves in the serialization before atomic loads,
which is now implemented directly in lowerATOMIC_LOAD. (This also
seems overkill, but that can be addressed separately.)
llvm-svn: 306117
The isBarrier/isTerminator flags have been removed from the SystemZ trap
instructions, so that tests do not fail with EXPENSIVE_CHECKS. This was just
an issue at -O0 and did not affect code output on benchmarks.
(Like Eli pointed out: "targets are split over whether they consider their
"trap" a terminator; x86, AArch64, and NVPTX don't, but ARM, MIPS, PPC, and
SystemZ do. We should probably try to be consistent here.". This is still the
case, although SystemZ has switched sides).
SystemZ now returns true in isMachineVerifierClean() :-)
These Generic tests have been modified so that they can be run with or without
EXPENSIVE_CHECKS: CodeGen/Generic/llc-start-stop.ll and
CodeGen/Generic/print-machineinstrs.ll
Review: Ulrich Weigand, Simon Pilgrim, Eli Friedman
https://bugs.llvm.org/show_bug.cgi?id=33047https://reviews.llvm.org/D34143
llvm-svn: 306106
Summary:
This change enables the sin(x) cos(x) -> sincos(x) optimization on GNU
target triples. This optimization was being inhibited when -ffast-math
wasn't set because sincos in GLibC does not set errno, while sin and cos
do. However, this optimization will only run if the attributes on the
sin/cos calls include readnone, which is how clang represents the fact
that it doesn't care about the errno values set by these functions (via
the -fno-math-errno flag).
Reviewers: hfinkel, bogner
Subscribers: mcrosier, javed.absar, llvm-commits, paul.redmond
Differential Revision: https://reviews.llvm.org/D32921
llvm-svn: 305204
Prior to this patch we used to not touch the LiveRegMatrix while doing
live-range splitting. In other words, when live-range splitting was
occurring, the LiveRegMatrix was not reflecting the changes.
This is generally fine because it means the query to the LiveRegMatrix
will be conservately correct. However, when decisions are taken based on
what is going to happen on the interferences (e.g., when we spill a
register and know that it is going to be available for another one), we
might hit an assertion that the color used for the assignment is still
in use.
This patch makes sure the changes on the live-ranges are properly
reflected in the LiveRegMatrix, so the assertions don't break.
An alternative could have been to remove the assertion, but it would
make the invariants of the code and the general reasoning more
complicated in my opnion.
http://llvm.org/PR33057
llvm-svn: 304603
Summary:
In SelectionDAG, when a store is immediately chained to another store
to the same address, elide the first store as it has no observable
effects. This is causes small improvements dealing with intrinsics
lowered to stores.
Test notes:
* Many testcases overwrite store addresses multiple times and needed
minor changes, mainly making stores volatile to prevent the
optimization from optimizing the test away.
* Many X86 test cases optimized out instructions associated with
associated with va_start.
* Note that test_splat in CodeGen/AArch64/misched-stp.ll no longer has
dependencies to check and can probably be removed and potentially
replaced with another test.
Reviewers: rnk, john.brawn
Subscribers: aemerson, rengolin, qcolombet, jyknight, nemanjai, nhaehnle, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D33206
llvm-svn: 303198
Llvm-stress discovered that a COPY may end up in ExpandPostRA::LowerCopy()
with an undef source operand. It is not possible for the target to handle
this, as this flag is not passed to TII->copyPhysReg().
This patch solves this by treating such a COPY as an identity COPY.
Review: Matthias Braun
https://reviews.llvm.org/D32892
llvm-svn: 302877
This method must return a valid register class, or the list-ilp isel
scheduler will crash. For MVT::Untyped nullptr was previously returned, but
now ADDR128BitRegClass is returned instead. This is needed just as long as
list-ilp (and probably also list-hybrid) is still there.
Review: Ulrich Weigand, A Trick
https://reviews.llvm.org/D32802
llvm-svn: 302649
When a 128 bit COPY is lowered into two instructions, an impl-use operand of
the super-reg should be added to each new instruction in case one of the
sub-regs is undefined.
Review: Ulrich Weigand
llvm-svn: 302146
A test case was found with llvm-stress that caused DAGCombiner to crash
when compiling for an older subtarget without vector support.
SystemZTargetLowering::combineTruncateExtract() should do nothing for older
subtargets.
This check was placed in canTreatAsByteVector(), which also helps in a few
other places.
Review: Ulrich Weigand
llvm-svn: 299763
When DAGCombiner visits a SIGN_EXTEND_INREG of a BUILD_VECTOR with
constant operands, a new BUILD_VECTOR node will be created transformed
constants.
Llvm-stress found a case where the new BUILD_VECTOR had constant operands
of an illegal type, because the (legal) element type is in fact not a legal
scalar type.
This patch changes this so that the new BUILD_VECTOR has the same operand
type as the old one.
Review: Eli Friedman, Nirav Dave
https://bugs.llvm.org//show_bug.cgi?id=32422
llvm-svn: 299540
Since LOCR only accepts GR32 virtual registers, its operands must be copied
into this regclass in insertSelect(), when an LOCR is built. Otherwise, the
case where the source operand was GRX32 will produce invalid IR.
Review: Ulrich Weigand
llvm-svn: 299220
Even on older subtargets that lack vector support, there may be vector values
with just one element in the input program. These are converted during DAG
legalization to scalar values.
The pre-legalize SystemZ DAGCombiner methods should in this circumstance not
touch these nodes. This patch adds a check for this in
SystemZTargetLowering::combineEXTRACT_VECTOR_ELT().
Review: Ulrich Weigand
llvm-svn: 299213
Previously, PromoteIntRes_TRUNCATE() did not handle the case where
the operand needs widening, which resulted in llvm_unreachable().
This patch adds the needed handling, along with a test case.
Review: Eli Friedman, Simon Pilgrim.
https://reviews.llvm.org/D31077
llvm-svn: 298357
The def operand of the new LG/LD should have the old def operands
flags and subreg index.
New test: test/CodeGen/SystemZ/fold-memory-op-impl.ll
Review: Ulrich Weigand
llvm-svn: 298341
New SystemZ tests for the improved codegen of vector compare and select,
including cases with a logical combination of two compares.
Review: Ulrich Weigand.
https://reviews.llvm.org/D29489
llvm-svn: 298049
If one of the subregs of the 128 bit reg is undefined when splitMove() splits
a store into two instructions, a use of an undefined physical register
results.
To remedy this, an implicit use of the super register is added onto both new
instructions, along with propagated kill and undef flags.
This was discovered with llvm-stress, and that test case is attached as
test/CodeGen/SystemZ/splitMove_undefReg_mverifier.ll
Thanks to Matthias Braun for helping with a nice explanation.
Review: Ulrich Weigand
llvm-svn: 298047
Recommiting with compiler time improvements
Recommitting after fixup of 32-bit aliasing sign offset bug in DAGCombiner.
* Simplify Consecutive Merge Store Candidate Search
Now that address aliasing is much less conservative, push through
simplified store merging search and chain alias analysis 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.
When 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 the output
Codegen (save perhaps for some ARM cases where we correctly constructs
wider loads, but then promotes them to float operations which appear
but requires more expensive constant generation).
Some minor peephole optimizations to deal with improved SubDAG shapes (listed below)
Additional Minor Changes:
1. Finishes removing unused AliasLoad code
2. Unifies 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 seems sufficient to not cause regressions in
tests.
5. Remove Chain dependencies of Memory operations on CopyfromReg
nodes as these are captured by data dependence
6. Forward loads-store values through tokenfactors containing
{CopyToReg,CopyFromReg} Values.
7. Peephole to convert buildvector of extract_vector_elt to
extract_subvector if possible (see
CodeGen/AArch64/store-merge.ll)
8. Store merging for the ARM target is restricted to 32-bit as
some in some contexts invalid 64-bit operations are being
generated. This can be removed once appropriate checks are
added.
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, improving load-store forwarding. One test in
particular is worth noting:
CodeGen/PowerPC/ppc64-align-long-double.ll - Improved load-store
forwarding converts a load-store pair into a parallel store and
a memory-realized bitcast of the same value. However, because we
lose the sharing of the explicit and implicit store values we
must create another local store. A similar transformation
happens before SelectionDAG as well.
Reviewers: arsenm, hfinkel, tstellarAMD, jyknight, nhaehnle
llvm-svn: 297695
This patch causes compile times for some patterns to explode. I have
a (large, unreduced) test case that slows down by more than 20x and
several test cases slow down by 2x. I'm sending some of the test cases
directly to Nirav and following up with more details in the review log,
but this should unblock anyone else hitting this.
llvm-svn: 296862
Recommiting after fixup of 32-bit aliasing sign offset bug in DAGCombiner.
* Simplify Consecutive Merge Store Candidate Search
Now that address aliasing is much less conservative, push through
simplified store merging search and chain alias analysis 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.
When 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 the output
Codegen (save perhaps for some ARM cases where we correctly constructs
wider loads, but then promotes them to float operations which appear
but requires more expensive constant generation).
Some minor peephole optimizations to deal with improved SubDAG shapes (listed below)
Additional Minor Changes:
1. Finishes removing unused AliasLoad code
2. Unifies 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 seems sufficient to not cause regressions in
tests.
5. Remove Chain dependencies of Memory operations on CopyfromReg
nodes as these are captured by data dependence
6. Forward loads-store values through tokenfactors containing
{CopyToReg,CopyFromReg} Values.
7. Peephole to convert buildvector of extract_vector_elt to
extract_subvector if possible (see
CodeGen/AArch64/store-merge.ll)
8. Store merging for the ARM target is restricted to 32-bit as
some in some contexts invalid 64-bit operations are being
generated. This can be removed once appropriate checks are
added.
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, improving load-store forwarding. One test in
particular is worth noting:
CodeGen/PowerPC/ppc64-align-long-double.ll - Improved load-store
forwarding converts a load-store pair into a parallel store and
a memory-realized bitcast of the same value. However, because we
lose the sharing of the explicit and implicit store values we
must create another local store. A similar transformation
happens before SelectionDAG as well.
Reviewers: arsenm, hfinkel, tstellarAMD, jyknight, nhaehnle
llvm-svn: 296476
Recommiting after fixup of 32-bit aliasing sign offset bug in DAGCombiner.
* Simplify Consecutive Merge Store Candidate Search
Now that address aliasing is much less conservative, push through
simplified store merging search and chain alias analysis 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.
When 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 the output
Codegen (save perhaps for some ARM cases where we correctly constructs
wider loads, but then promotes them to float operations which appear
but requires more expensive constant generation).
Some minor peephole optimizations to deal with improved SubDAG shapes (listed below)
Additional Minor Changes:
1. Finishes removing unused AliasLoad code
2. Unifies 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 seems sufficient to not cause regressions in
tests.
5. Remove Chain dependencies of Memory operations on CopyfromReg
nodes as these are captured by data dependence
6. Forward loads-store values through tokenfactors containing
{CopyToReg,CopyFromReg} Values.
7. Peephole to convert buildvector of extract_vector_elt to
extract_subvector if possible (see
CodeGen/AArch64/store-merge.ll)
8. Store merging for the ARM target is restricted to 32-bit as
some in some contexts invalid 64-bit operations are being
generated. This can be removed once appropriate checks are
added.
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, improving load-store forwarding. One test in
particular is worth noting:
CodeGen/PowerPC/ppc64-align-long-double.ll - Improved load-store
forwarding converts a load-store pair into a parallel store and
a memory-realized bitcast of the same value. However, because we
lose the sharing of the explicit and implicit store values we
must create another local store. A similar transformation
happens before SelectionDAG as well.
Reviewers: arsenm, hfinkel, tstellarAMD, jyknight, nhaehnle
llvm-svn: 296252
This re-applies commit r292189, reverted in r292191.
SelectionDAGBuilder recognizes libfuncs using some homegrown
parameter type-checking.
Use TLI instead, removing another heap of redundant code.
This isn't strictly NFC, as the SDAG code was too lax.
Concretely, this means changes are required to a few tests:
- calling a non-variadic function via a variadic prototype isn't OK;
it just happens to work on x86_64 (but not on, e.g., aarch64).
- mempcpy has a size_t parameter; the SDAG code accepts any integer
type, which meant using i32 on x86_64 worked.
- a handful of SystemZ tests check the SDAG support for lax prototype
checking: Ulrich agrees on removing them.
I don't think it's worth supporting any of these (IMO) invalid
testcases. Instead, fix them to be more meaningful.
llvm-svn: 294028
Summary:
llc would hit a fatal error for errors in inline assembly. The
diagnostics message is now printed.
Reviewers: rengolin, MatzeB, javed.absar, anemet
Reviewed By: anemet
Subscribers: jyknight, nemanjai, llvm-commits
Differential Revision: https://reviews.llvm.org/D29408
llvm-svn: 293999
Recommiting after fixing X86 inc/dec chain bug.
* Simplify Consecutive Merge Store Candidate Search
Now that address aliasing is much less conservative, push through
simplified store merging search and chain alias analysis 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.
When 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 the output
Codegen (save perhaps for some ARM cases where we correctly constructs
wider loads, but then promotes them to float operations which appear
but requires more expensive constant generation).
Some minor peephole optimizations to deal with improved SubDAG shapes (listed below)
Additional Minor Changes:
1. Finishes removing unused AliasLoad code
2. Unifies 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 seems sufficient to not cause regressions in
tests.
5. Remove Chain dependencies of Memory operations on CopyfromReg
nodes as these are captured by data dependence
6. Forward loads-store values through tokenfactors containing
{CopyToReg,CopyFromReg} Values.
7. Peephole to convert buildvector of extract_vector_elt to
extract_subvector if possible (see
CodeGen/AArch64/store-merge.ll)
8. Store merging for the ARM target is restricted to 32-bit as
some in some contexts invalid 64-bit operations are being
generated. This can be removed once appropriate checks are
added.
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, improving load-store forwarding. One test in
particular is worth noting:
CodeGen/PowerPC/ppc64-align-long-double.ll - Improved load-store
forwarding converts a load-store pair into a parallel store and
a memory-realized bitcast of the same value. However, because we
lose the sharing of the explicit and implicit store values we
must create another local store. A similar transformation
happens before SelectionDAG as well.
Reviewers: arsenm, hfinkel, tstellarAMD, jyknight, nhaehnle
llvm-svn: 293893
When choosing the best successor for a block, ordinarily we would have preferred
a block that preserves the CFG unless there is a strong probability the other
direction. For small blocks that can be duplicated we now skip that requirement
as well, subject to some simple frequency calculations.
Differential Revision: https://reviews.llvm.org/D28583
llvm-svn: 293716
Previously, we would hit UB (or the ISD::DELETED_NODE assert) if we
happened to replace a node during UpdateChains, because it would be
left in the list we were iterating over. This nulls out the pointer
when that happens so that we can avoid the issue.
Fixes llvm.org/PR31710
llvm-svn: 293522
In case of a SIGN/ZERO_EXTEND of an incomplete vector type (using only a
partial number of available vector elements), WidenVecRes_Convert() used to
resort to scalarization.
This patch adds a handling of the (common) case where an input vector can be
found of same width as the widened result vector, by converting the node to
SIGN/ZERO_EXTEND_VECTOR_INREG.
Review: Eli Friedman
llvm-svn: 293268
* Simplify Consecutive Merge Store Candidate Search
Now that address aliasing is much less conservative, push through
simplified store merging search and chain alias analysis 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.
When 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 the output
Codegen (save perhaps for some ARM cases where we correctly constructs
wider loads, but then promotes them to float operations which appear
but requires more expensive constant generation).
Some minor peephole optimizations to deal with improved SubDAG shapes (listed below)
Additional Minor Changes:
1. Finishes removing unused AliasLoad code
2. Unifies 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 seems sufficient to not cause regressions in
tests.
5. Remove Chain dependencies of Memory operations on CopyfromReg
nodes as these are captured by data dependence
6. Forward loads-store values through tokenfactors containing
{CopyToReg,CopyFromReg} Values.
7. Peephole to convert buildvector of extract_vector_elt to
extract_subvector if possible (see
CodeGen/AArch64/store-merge.ll)
8. Store merging for the ARM target is restricted to 32-bit as
some in some contexts invalid 64-bit operations are being
generated. This can be removed once appropriate checks are
added.
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, improving load-store forwarding. One test in
particular is worth noting:
CodeGen/PowerPC/ppc64-align-long-double.ll - Improved load-store
forwarding converts a load-store pair into a parallel store and
a memory-realized bitcast of the same value. However, because we
lose the sharing of the explicit and implicit store values we
must create another local store. A similar transformation
happens before SelectionDAG as well.
Reviewers: arsenm, hfinkel, tstellarAMD, jyknight, nhaehnle
llvm-svn: 293184
During post-RA pseudo expansion, an 'undef' flag of the source operand should
be propagated by emitGRX32Move().
Review: Ulrich Weigand
llvm-svn: 292353
This reverts commit ada6595a526d71df04988eb0a4b4fe84df398ded.
This needs a simple probability check because there are some cases where it is
not profitable.
llvm-svn: 291695
When choosing the best successor for a block, ordinarily we would have preferred
a block that preserves the CFG unless there is a strong probability the other
direction. For small blocks that can be duplicated we now skip that requirement
as well.
Differential revision: https://reviews.llvm.org/D27742
llvm-svn: 291609
