Unlike xN/wN, the size of vN is genuinely ambiguous in the assembly, so we
should try to infer what was intended from the type. But only down to 64-bits
(vN can never represent sN, hN or bN).
llvm-svn: 269132
Summary:
This implements the lowering of the X constraint on
AArch64.
The default behaviour of the X constraint lowering is to
restrict it to "f". This is a problem because the "f"
constraint is not implemented on AArch64 and would be too
restrictive anyway. Therefore, the AArch64 hook will
lower this to "w" (if the operand is a floating point or
vector) or "r" otherwise.
The implementation is similar with the one added for
ARM (r267411).
This is the AArch64 side of the fix for http://llvm.org/PR26493
Reviewers: rengolin
Subscribers: aemerson, rengolin, llvm-commits, t.p.northover
Differential Revision: http://reviews.llvm.org/D19967
llvm-svn: 268907
Summary:
If a function needs to allocate both callee-save stack memory and local
stack memory, we currently decrement/increment the SP in two steps:
first for the callee-save area, and then for the local stack area. This
changes the code to allocate them both at once at the very beginning/end
of the function. This has two benefits:
1) there is one fewer sub/add micro-op in the prologue/epilogue
2) the stack adjustment instructions act as a scheduling barrier, so
moving them to the very beginning/end of the function increases post-RA
scheduler's ability to move instructions (that only depend on argument
registers) before any of the callee-save stores
This change can cause an increase in instructions if the original local
stack SP decrement could be folded into the first store to the stack.
This occurs when the first local stack store is to stack offset 0. In
this case we are trading off one more sub instruction for one fewer sub
micro-op (along with benefits (2) and (3) above).
Reviewers: t.p.northover
Subscribers: aemerson, rengolin, mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D18619
llvm-svn: 268746
If we don't, values that aren't precisely representable in f16 could
be used as-is in a promoted f32 operation, which would produce
incorrect results.
AArch64 had the correct behavior; add a focused test.
Fixes http://llvm.org/PR26871
llvm-svn: 268700
This patch adds support for estimating the square root, its reciprocal and
division or reciprocal using the combiner generic reciprocal machinery.
llvm-svn: 268539
transferSuccessors() would LoadCmpBB a successor of DoneBB,
whereas it should be a successor of the original MBB.
Follow-up to r266339.
Unfortunately, it's tricky to catch this in the verifier.
llvm-svn: 267779
log2(Mask) is smaller than 32, we must use the 32-bit variant because the 64-bit
variant cannot encode it. Therefore, set the subreg part accordingly.
[AArch64] Fix optimizeCondBranch logic.
The opcode for the optimized branch does not depend on the size
of the activate bits in the AND masks, but the AND opcode itself.
Indeed, we need to use a X or W variant based on the AND variant
not based on whether the mask fits into the related variant.
Otherwise, we may end up using the W variant of the optimized branch
for 64-bit register inputs!
This fixes the last make check verifier issues for AArch64: PR27479.
llvm-svn: 267465
The original patch caused crashes because it could derefence a null pointer
for SelectionDAGTargetInfo for targets that do not define it.
Evaluates fmul+fadd -> fmadd combines and similar code sequences in the
machine combiner. It adds support for float and double similar to the existing
integer implementation. The key features are:
- DAGCombiner checks whether it should combine greedily or let the machine
combiner do the evaluation. This is only supported on ARM64.
- It gives preference to throughput over latency: the heuristic used is
to combine always in loops. The targets decides whether the machine
combiner should optimize for throughput or latency.
- Supports for fmadd, f(n)msub, fmla, fmls patterns
- On by default at O3 ffast-math
llvm-svn: 267328
The opcode for the optimized branch does not depend on the size
of the activate bits in the AND masks, but the AND opcode itself.
Indeed, we need to use a X or W variant based on the AND variant
not based on whether the mask fits into the related variant.
Otherwise, we may end up using the W variant of the optimized branch
for 64-bit register inputs!
This fixes the last make check verifier issues for AArch64: PR27479.
llvm-svn: 267206
Avoid quadratic complexity in unusually large basic blocks by limiting
the size of the ready lists.
Differential Revision: http://reviews.llvm.org/D19349
llvm-svn: 267189
We used to simply set the kill flags to true when transforming a scalar
instruction to a vector one.
SrcScalar1 = copy SrcVector1
... = opScalar SrcScalar1
=>
SrcScalar1 = copy SrcVector1
... = opVector SrcVector1<kill>
This is obviously wrong. The proper update consists in:
1. Propagate the kill status from the copy to the new opVector
2. Reset the kill status on the copy, since the live-range of
SrcVector1 got extended.
This fixes some of the machine verifier errors for AArch64 with make check.
llvm-svn: 267180
Evaluates fmul+fadd -> fmadd combines and similar code sequences in the
machine combiner. It adds support for float and double similar to the existing
integer implementation. The key features are:
- DAGCombiner checks whether it should combine greedily or let the machine
combiner do the evaluation. This is only supported on ARM64.
- It gives preference to throughput over latency: the heuristic used is
to combine always in loops. The targets decides whether the machine
combiner should optimize for throughput or latency.
- Supports for fmadd, f(n)msub, fmla, fmls patterns
- On by default at O3 ffast-math
llvm-svn: 267098
This test used to write a .s file until r266971 fixed that. But on most bots,
the .s file still exists. Add an rm statement to clean up the bots. In a few
days, this statement can go away again.
llvm-svn: 267095
AArch64InstrInfo::optimizeCompareInstr has bug PR27158 which causes generation of incorrect code.
A compare instruction is substituted with another instruction which does not
produce the same flags as the original compare instruction.
This patch contains:
1. Fix of the bug.
2. A regression test in MIR.
3. A new test to check that SUBS is replaced by SUB.
Differential Revision: http://reviews.llvm.org/D18838
llvm-svn: 266969
Both AArch64 and ARM support llvm.<arch>.thread.pointer intrinsics that
just return the thread pointer. I have a pending patch that does the same
for SystemZ (D19054), and there are many more targets that could benefit
from one.
This patch merges the ARM and AArch64 intrinsics into a single target
independent one that will also be used by subsequent targets.
Differential Revision: http://reviews.llvm.org/D19098
llvm-svn: 266818
With this change, ideally IR pass can always generate llvm.stackguard
call to get the stack guard; but for now there are still IR form stack
guard customizations around (see getIRStackGuard()). Future SSP
customization should go through LOAD_STACK_GUARD.
There is a behavior change: stack guard values are not CSEed anymore,
since we should never reuse the value in case that it has been spilled (and
corrupted). See ssp-guard-spill.ll. This also cause the change of stack
size and codegen in X86 and AArch64 test cases.
Ideally we'd like to know if the guard created in llvm.stackprotector() gets
spilled or not. If the value is spilled, discard the value and reload
stack guard; otherwise reuse the value. This can be done by teaching
register allocator to know how to rematerialize LOAD_STACK_GUARD and
force a rematerialization (which seems hard), or check for spilling in
expandPostRAPseudo. It only makes sense when the stack guard is a global
variable, which requires more instructions to load. Anyway, this seems to go out
of the scope of the current patch.
llvm-svn: 266806
This improves AA in the MI schduler when reason about paired instructions.
Phabricator Revision: http://reviews.llvm.org/D17098
PR26358
llvm-svn: 266462
Currently each Function points to a DISubprogram and DISubprogram has a
scope field. For member functions the scope is a DICompositeType. DIScopes
point to the DICompileUnit to facilitate type uniquing.
Distinct DISubprograms (with isDefinition: true) are not part of the type
hierarchy and cannot be uniqued. This change removes the subprograms
list from DICompileUnit and instead adds a pointer to the owning compile
unit to distinct DISubprograms. This would make it easy for ThinLTO to
strip unneeded DISubprograms and their transitively referenced debug info.
Motivation
----------
Materializing DISubprograms is currently the most expensive operation when
doing a ThinLTO build of clang.
We want the DISubprogram to be stored in a separate Bitcode block (or the
same block as the function body) so we can avoid having to expensively
deserialize all DISubprograms together with the global metadata. If a
function has been inlined into another subprogram we need to store a
reference the block containing the inlined subprogram.
Attached to https://llvm.org/bugs/show_bug.cgi?id=27284 is a python script
that updates LLVM IR testcases to the new format.
http://reviews.llvm.org/D19034
<rdar://problem/25256815>
llvm-svn: 266446
Summary:
Without MMOs, the callee-save load/store instructions were treated as
volatile by the MI post-RA scheduler and AArch64LoadStoreOptimizer.
Reviewers: t.p.northover, mcrosier
Subscribers: aemerson, rengolin, mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D17661
llvm-svn: 266439
Perform store clustering just like load clustering. This change add
StoreClusterMutation in machine-scheduler. To control StoreClusterMutation,
added enableClusterStores() in TargetInstrInfo.h. This is enabled only on
AArch64 for now.
This change also add support for unscaled stores which were not handled in
getMemOpBaseRegImmOfs().
llvm-svn: 266437
FastRegAlloc works only at the basic-block level and spills all live-out
registers. Unfortunately for a stack-based cmpxchg near the spill slots, this
can perpetually clear the exclusive monitor, which means the cmpxchg will never
succeed.
I believe the only way to handle this within LLVM is by expanding the loop
post-regalloc. We don't want this in general because it severely limits the
optimisations that can be done, so we limit this to -O0 compilations.
It's an ugly hack, and about the one good point in the whole mess is that we
can treat all cmpxchg operations in the most naive way possible (seq_cst, no
clrex faff) without affecting correctness.
Should fix PR25526.
llvm-svn: 266339
It is very likely that the swiftself parameter is alive throughout most
functions function so putting it into a callee save register should
avoid spills for the callers with only a minimum amount of extra spills
in the callees.
Currently the generated code is correct but unnecessarily spills and
reloads arguments passed in callee save registers, I will address this
in upcoming patches.
This also adds a missing check that for tail calls the preserved value
of the caller must be the same as the callees parameter.
Differential Revision: http://reviews.llvm.org/D19007
llvm-svn: 266251
Disable LDP/STP for quads on Exynos M1 as they are not as efficient as pairs
of regular LDR/STR.
Patch by Abderrazek Zaafrani <a.zaafrani@samsung.com>.
llvm-svn: 266223
This patch fixes a bug (PR26827) when using anti-aliasing in store
merging. This sets the chain users of the component stores to point to
the new store instead of the component stores chain parent.
Reviewers: jyknight
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D18909
llvm-svn: 266217
two fixes with one about error verify-regalloc reported, and
another about live range update of phi after rematerialization.
r265547:
Replace analyzeSiblingValues with new algorithm to fix its compile
time issue. The patch is to solve PR17409 and its duplicates.
analyzeSiblingValues is a N x N complexity algorithm where N is
the number of siblings generated by reg splitting. Although it
causes siginificant compile time issue when N is large, it is also
important for performance since it removes redundent spills and
enables rematerialization.
To solve the compile time issue, the patch removes analyzeSiblingValues
and replaces it with lower cost alternatives containing two parts. The
first part creates a new spill hoisting method in postOptimization of
register allocation. It does spill hoisting at once after all the spills
are generated instead of inside every instance of selectOrSplit. The
second part queries the define expr of the original register for
rematerializaiton and keep it always available during register allocation
even if it is already dead. It deletes those dead instructions only in
postOptimization. With the two parts in the patch, it can remove
analyzeSiblingValues without sacrificing performance.
Patches on top of r265547:
r265610 "Fix the compare-clang diff error introduced by r265547."
r265639 "Fix the sanitizer bootstrap error in r265547."
r265657 "InlineSpiller.cpp: Escap \@ in r265547. [-Wdocumentation]"
Differential Revision: http://reviews.llvm.org/D15302
Differential Revision: http://reviews.llvm.org/D18934
Differential Revision: http://reviews.llvm.org/D18935
Differential Revision: http://reviews.llvm.org/D18936
llvm-svn: 266162
Summary:
In getUnderlyingObjectsForInstr(): Don't give up on instructions with
multiple MMOs, instead look through all the MMOs and if they all meet
the conservative criteria previously used for single MMO instructions,
then return all of the underlying objects derived from the MMOs.
The change to ScheduleDAGInstrs::buildSchedGraph() is needed to avoid
the case where multiple underlying objects are present and are related
in such a way that successive iterations of the loop end up adding a
dependency from an instruction to itself.
Reviewers: atrick, hfinkel
Subscribers: MatzeB, mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D18093
llvm-svn: 266084
It caused PR27275: "ARM: Bad machine code: Using an undefined physical register"
Also reverting the following commits that were landed on top:
r265610 "Fix the compare-clang diff error introduced by r265547."
r265639 "Fix the sanitizer bootstrap error in r265547."
r265657 "InlineSpiller.cpp: Escap \@ in r265547. [-Wdocumentation]"
llvm-svn: 265790
when DenseMap growed and moved memory. I verified it fixed the bootstrap
problem on x86_64-linux-gnu but I cannot verify whether it fixes
the bootstrap error on clang-ppc64be-linux. I will watch the build-bot
result closely.
Replace analyzeSiblingValues with new algorithm to fix its compile
time issue. The patch is to solve PR17409 and its duplicates.
analyzeSiblingValues is a N x N complexity algorithm where N is
the number of siblings generated by reg splitting. Although it
causes siginificant compile time issue when N is large, it is also
important for performance since it removes redundent spills and
enables rematerialization.
To solve the compile time issue, the patch removes analyzeSiblingValues
and replaces it with lower cost alternatives containing two parts. The
first part creates a new spill hoisting method in postOptimization of
register allocation. It does spill hoisting at once after all the spills
are generated instead of inside every instance of selectOrSplit. The
second part queries the define expr of the original register for
rematerializaiton and keep it always available during register allocation
even if it is already dead. It deletes those dead instructions only in
postOptimization. With the two parts in the patch, it can remove
analyzeSiblingValues without sacrificing performance.
Differential Revision: http://reviews.llvm.org/D15302
llvm-svn: 265547
Bionic has a defined thread-local location for the stack protector
cookie. Emit a direct load instead of going through __stack_chk_guard.
llvm-svn: 265481
Presently, CodeGenPrepare deletes all nearly empty (only phi and branch)
basic blocks. This pass can delete loop preheaders which frequently creates
critical edges. A preheader can be a convenient place to spill registers to
the stack. If the entrance to a loop body is a critical edge, then spills
may occur in the loop body rather than immediately before it. This patch
protects loop preheaders from deletion in CodeGenPrepare even if they are
nearly empty.
Since the patch alters the CFG, it affects a large number of test cases.
In most cases, the changes are merely cosmetic (basic blocks have different
names or instruction orders change slightly). I am somewhat concerned about
the test/CodeGen/Mips/brdelayslot.ll test case. If the loop preheader is not
deleted, then the MIPS backend does not take advantage of a branch delay
slot. Consequently, I would like some close review by a MIPS expert.
The patch also partially subsumes D16893 from George Burgess IV. George
correctly notes that CodeGenPrepare does not actually preserve the dominator
tree. I think the dominator tree was usually not valid when CodeGenPrepare
ran, but I am using LoopInfo to mark preheaders, so the dominator tree is
now always valid before CodeGenPrepare.
Author: Tom Jablin (tjablin)
Reviewers: hfinkel george.burgess.iv vkalintiris dsanders kbarton cycheng
http://reviews.llvm.org/D16984
llvm-svn: 265397
We can only perform a tail call to a callee that preserves all the
registers that the caller needs to preserve.
This situation happens with calling conventions like preserver_mostcc or
cxx_fast_tls. It was explicitely handled for fast_tls and failing for
preserve_most. This patch generalizes the check to any calling
convention.
Related to rdar://24207743
Differential Revision: http://reviews.llvm.org/D18680
llvm-svn: 265329
time issue. The patch is to solve PR17409 and its duplicates.
analyzeSiblingValues is a N x N complexity algorithm where N is
the number of siblings generated by reg splitting. Although it
causes siginificant compile time issue when N is large, it is also
important for performance since it removes redundent spills and
enables rematerialization.
To solve the compile time issue, the patch removes analyzeSiblingValues
and replaces it with lower cost alternatives containing two parts. The
first part creates a new spill hoisting method in postOptimization of
register allocation. It does spill hoisting at once after all the spills
are generated instead of inside every instance of selectOrSplit. The
second part queries the define expr of the original register for
rematerializaiton and keep it always available during register allocation
even if it is already dead. It deletes those dead instructions only in
postOptimization. With the two parts in the patch, it can remove
analyzeSiblingValues without sacrificing performance.
Differential Revision: http://reviews.llvm.org/D15302
llvm-svn: 265309
This mostly cosmetic patch moves the DebugEmissionKind enum from DIBuilder
into DICompileUnit. DIBuilder is not the right place for this enum to live
in — a metadata consumer should not have to include DIBuilder.h.
I also added a Verifier check that checks that the emission kind of a
DICompileUnit is actually legal.
http://reviews.llvm.org/D18612
<rdar://problem/25427165>
llvm-svn: 265077
Summary:
This change will allow loads with imp-def to be clustered in machine-scheduler pass.
areMemAccessesTriviallyDisjoint() can also handle loads with imp-def.
Reviewers: mcrosier, jmolloy, t.p.northover
Subscribers: aemerson, rengolin, mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D18665
llvm-svn: 265051
Summary:
This change will handle missing store pair opportunity where the first store
instruction stores zero followed by the non-zero store. For example, this change
will convert :
str wzr, [x8]
str w1, [x8, #4]
into:
stp wzr, w1, [x8]
Reviewers: jmolloy, t.p.northover, mcrosier
Subscribers: flyingforyou, aemerson, rengolin, mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D18570
llvm-svn: 265021
When merging stores in DAGCombiner, add check to ensure that no
dependenices exist that would cause the construction of a cycle in our
DAG. This may happen if one store has a data dependence on another
instruction (e.g. a load) which itself has a (chain) dependence on
another store being merged. These stores cannot be merged safely and
doing so results in a cycle that is discovered in LegalizeDAG.
This test is only done in cases where Antialias analysis is used (UseAA)
as non-AA store merge candidates will be merged logically after all
loads which have been checked to not alias.
Reviewers: ahatanak, spatel, niravd, arsenm, hfinkel, tstellarAMD, jyknight
Subscribers: llvm-commits, tberghammer, danalbert, srhines
Differential Revision: http://reviews.llvm.org/D18336
llvm-svn: 264461
This patch adds unscaled loads and sign-extend loads to the TII
getMemOpBaseRegImmOfs API, which is used to control clustering in the MI
scheduler. This is done to create more opportunities for load pairing. I've
also added the scaled LDRSWui instruction, which was missing from the scaled
instructions. Finally, I've added support in shouldClusterLoads for clustering
adjacent sext and zext loads that too can be paired by the load/store optimizer.
Differential Revision: http://reviews.llvm.org/D18048
llvm-svn: 263819
When the SP in not changed because of realignment/VLAs etc., we restore the SP
by using the previous value of SP and not the FP. Breaking the dependency will
help in cases when the epilog of a callee is close to the epilog of the caller;
for then "sub sp, fp, #" depends on the load restoring the FP in the epilog of
the callee.
http://reviews.llvm.org/D18060
Patch by Aditya Kumar and Evandro Menezes.
llvm-svn: 263458
Summary:
Peephole optimization that generates a single TBZ/TBNZ instruction
for test and branch sequences like in the example below. This handles
the cases that miss folding of AND into TBZ/TBNZ during ISelLowering of BR_CC
Examples:
and w8, w8, #0x400
cbnz w8, L1
to
tbnz w8, #10, L1
Reviewers: MatzeB, jmolloy, mcrosier, t.p.northover
Subscribers: aemerson, rengolin, mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D17942
llvm-svn: 263136
This change adds a support for a preserve_most calling convention to the AArch64 backend, similar to how it was done for X86-64.
There is also a subsequent patch on top of this one to add a tail-calls support for this calling convention.
Differential Revision: http://reviews.llvm.org/D18016
llvm-svn: 263092
This test failed in some ARM bots after a divmod change because it was
running on a native llc, instead of targeted one. This makes sure the test
is target-specific (as intended), and also copies to ARM and AArch64
directories. If it is also supposed to work on other architectures, I'll
leave as an exercise to the respective maintainers.
llvm-svn: 262620
Summary:
This change enables frame pointer elimination in non-leaf functions.
The -fomit-frame-pointer option still needs to be used when compiling
via clang (or an equivalent method of not setting the
'no-frame-pointer-elim*' function attributes if generating llvm IR via
some other method) to take advantage of this optimization.
This change should be NFC when compiling via clang without
-fomit-frame-pointer.
Reviewers: t.p.northover
Subscribers: aemerson, rengolin, tberghammer, qcolombet, llvm-commits, danalbert, mcrosier, srhines
Differential Revision: http://reviews.llvm.org/D17730
llvm-svn: 262495
Revert r262248 in an attempt to fix the clang-native-aarch64-full
bot and to investigate a performance regression in
SingleSource/Benchmarks/CoyoteBench/huffbench
llvm-svn: 262388
Summary:
Both the hardware and LLVM have changed since 2012.
Now, load-based heuristic don't show big differences any more on OoO cores.
There is no notable regressons and improvements on spec2000/2006. (Cortex-A57, Core i5).
Reviewers: spatel, zansari
Differential Revision: http://reviews.llvm.org/D16836
llvm-svn: 261809
Summary:
Fix a bug in epilog generation where the incoming stack arguments were
not being popped for fastcc functions when -tailcallopt was passed.
Reviewers: t.p.northover, mcrosier, jmolloy, rengolin
Subscribers: aemerson, rengolin, mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D16894
llvm-svn: 261650
Summary:
Without this, this command
$ llvm-run llc -stop-after machine-cp -o - <( echo '' )
outputs an error, because we close stdout twice -- once when closing the
file opened for "-o", and again when closing outs().
Also clarify in the outs() definition that you can't ever call it if you
want to open your own raw_fd_ostream on stdout.
Reviewers: jroelofs, tstellarAMD
Subscribers: jholewinski, qcolombet, dsanders, llvm-commits
Differential Revision: http://reviews.llvm.org/D17422
llvm-svn: 261286
After r261154, we were only clearing flags if the known-zero register was
originally live-in to the basic block, but we have to do it even if not when
more than one COPY has been eliminated, otherwise the user of the first COPY
may still have <kill> marked.
E.g.
BB#N:
%X0 = COPY %XZR
STRXui %X0<kill>, <fi#0>
%X0 = COPY %XZR
STRXui %X0<kill>, <fi#1>
We can eliminate both copies, X0 is not live-in, but we must clear the kill on
the first store.
Unfortunately, I've been unable to come up with a non-fragile test for this.
I've only seen it in the wild with regalloc-created spills, and attempts to
reproduce that in a reasonable way run afoul of COPY coalescing. Even volatile
asm clobbers were moved around. Should fix the aarch64 bot though.
llvm-svn: 261175
Mostly, this fixes the bug that if the CBZ guaranteed Xn but Wn was used, we
didn't sort out the use-def chain properly.
I've also made it check more than just the last instruction for a compatible
CBZ (so it can cope without fallthroughs). I'd have liked to do that
separately, but it's helps writing the test.
Finally, I removed some custom loops in favour of MachineInstr helpers and
refactored the control flow to flatten it and avoid possibly quadratic
iterations in blocks with many copies. NFC for these, just a general tidy-up.
llvm-svn: 261154
Summary:
This change will add a pass to remove unnecessary zero copies in target blocks
of cbz/cbnz instructions. E.g., the copy instruction in the code below can be
removed because the cbz jumps to BB1 when x0 is zero :
BB0:
cbz x0, .BB1
BB1:
mov x0, xzr
Jun
Reviewers: gberry, jmolloy, HaoLiu, MatzeB, mcrosier
Subscribers: mcrosier, mssimpso, haicheng, bmakam, llvm-commits, aemerson, rengolin
Differential Revision: http://reviews.llvm.org/D16203
llvm-svn: 261004
Summary:
Before this change, callee-save registers would be rounded up to even
pairs of GPRs and FPRs. This change eliminates these extra padding
load/stores, though it does keep the stack allocation the same size
unless both the GPR and FPR sets have an odd size, in which case one
full pair stack slot (16 bytes) is saved.
This optimization cannot currently be done for MachO targets since they
rely on a fast-path .debug_frame equivalent that can only encode
callee-save registers as pairs.
Reviewers: t.p.northover, rengolin, mcrosier, jmolloy
Subscribers: aemerson, rengolin, mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D17000
llvm-svn: 260689
This patch allows the mixing of scaled and unscaled load/stores to form
load/store pairs.
This is a reapplication of r259812, which had an incorrect assert. The
test_stur_str_no_assert() test is a reduced version of the issue hit in
the AArch64 self-host.
PR24465
llvm-svn: 260523
Summary:
Fix case where a pre-inc/dec load/store would not be formed if the
add/sub that forms the inc/dec part of the operation was the first
instruction in the block being examined.
Reviewers: mcrosier, jmolloy, t.p.northover, junbuml
Subscribers: aemerson, rengolin, mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D16785
llvm-svn: 260275
If a range has a lower bound of 0, add an AssertZext from the
nearest floor power of two.
This allows operations with some workitem intrinsics with known
maximum ranges to use fast 24-bit multiplies.
llvm-svn: 260109
This patch allows the mixing of scaled and unscaled load/stores to form
load/store pairs.
PR24465
http://reviews.llvm.org/D12116
Many thanks to Ahmed and Michael for fixes and code review.
This is a reapplication of r246769 and r259790. The tramp3d failure was caused
by an incorrect refactoring in the patch. Specifically, we weren't always
properly clearing the SExtIdx flag.
llvm-svn: 259812
During instruction selection, the AArch64 backend can recognise the
following pattern and generate an [U|S]MADDL instruction, i.e. a
multiply of two 32-bit operands with a 64-bit result:
(mul (sext i32), (sext i32))
However, when one of the operands is constant, the sign extension
gets folded into the constant in SelectionDAG::getNode(). This means
that the instruction selection sees this:
(mul (sext i32), i64)
...which doesn't match the pattern. Sign-extension and 64-bit
multiply instructions are generated, which are slower than one 32-bit
multiply.
Add a pattern to match this and generate the correct instruction, for
both signed and unsigned multiplies.
Patch by Chris Diamand!
llvm-svn: 259800
This patch allows the mixing of scaled and unscaled load/stores to form
load/store pairs.
PR24465
http://reviews.llvm.org/D12116
Many thanks to Ahmed and Michael for fixes and code review.
This is a reapplication of r246769, which was reverted in r246782 due to a
test-suite failure. I'm unable to reproduce the issue at this time.
llvm-svn: 259790
Recommited, after some fixing with test cases.
Updated test cases:
test/CodeGen/AArch64/arm64-misched-memdep-bug.ll
test/CodeGen/AArch64/tailcall_misched_graph.ll
Temporarily disabled test cases:
test/CodeGen/AMDGPU/split-vector-memoperand-offsets.ll
test/CodeGen/PowerPC/ppc64-fastcc.ll (partially updated)
test/CodeGen/PowerPC/vsx-fma-m.ll
test/CodeGen/PowerPC/vsx-fma-sp.ll
http://reviews.llvm.org/D8705
Reviewers: Hal Finkel, Andy Trick.
llvm-svn: 259673
Summary:
This is an extension to the existing implementation of r242436 which
restricts to only select inputs. This version fixes missed opportunities
in pr26084 by attempting to lower conditional compare sequences of
and/or trees with setcc leafs. This will additionaly handle the case
when a tree with select input is not a conjunction-disjunction tree
but some of the sub trees are conjunction-disjunction trees.
Reviewers: jmolloy, t.p.northover, mcrosier, MatzeB
Subscribers: mcrosier, llvm-commits, junbuml, haicheng, mssimpso, gberry
Differential Revision: http://reviews.llvm.org/D16291
llvm-svn: 259387
Since we only have pair - not single - nontemporal store instructions,
we have to extract the high part into a separate register to be able
to use them.
When the initial nontemporal codegen support was added, I wrote the
extract using the nonsensical UBFX [0,32[.
Use the correct LSR form instead.
llvm-svn: 259134
Summary:
findBetterNeighborChains does not handle volatile or indexed stores.
However, it did not check when adding stores to ChainedStores.
Reviewers: arsenm
Differential Revision: http://reviews.llvm.org/D16463
llvm-svn: 259024
For historic reasons, the behavior of .align differs between targets.
Fortunately, there are alternatives, .p2align and .balign, which make the
interpretation of the parameter explicit, and which behave consistently across
targets.
This patch teaches MC to use .p2align instead of .align, so that people reading
code for multiple architectures don't have to remember which way each platform
does its .align directive.
Differential Revision: http://reviews.llvm.org/D16549
llvm-svn: 258750
Some of the conditions necessary to produce ccmp sequences were only
checked in recursive calls to emitConjunctionDisjunctionTree() after
some of the earlier expressions were already built. Move all checks over
to isConjunctionDisjunctionTree() so they are all checked before we
start emitting instructions.
Also rename some variable to better reflect their usage.
llvm-svn: 258605
The current behavior is incorrect, as the two CCs returned by
changeFPCCToAArch64CC, intended to be OR'ed, are instead used
in an AND ccmp chain.
Consider:
define i32 @t(float %a, float %b, float %c, float %d, i32 %e, i32 %f) {
%cc1 = fcmp one float %a, %b
%cc2 = fcmp olt float %c, %d
%and = and i1 %cc1, %cc2
%r = select i1 %and, i32 %e, i32 %f
ret i32 %r
}
Assuming (%a < %b) and (%c < %d); we used to do:
fcmp s0, s1 # nzcv <- 1000
orr w8, wzr, #0x1 # w8 <- 1
csel w9, w8, wzr, mi # w9 <- 1
csel w8, w8, w9, gt # w8 <- 1
fcmp s2, s3 # nzcv <- 1000
cset w9, mi # w9 <- 1
tst w8, w9 # (w8 & w9) == 1, so: nzcv <- 0000
csel w0, w0, w1, ne # w0 <- w0
We now do:
fcmp s2, s3 # nzcv <- 1000
fccmp s0, s1, #0, mi # mi, so: nzcv <- 1000
fccmp s0, s1, #8, le # !le, so: nzcv <- 1000
csel w0, w0, w1, pl # !pl, so: w0 <- w1
In other words, we transformed:
(c < d) && ((a < b) || (a > b))
into:
(c < d) && (a u>= b) && (a u<= b)
whereas, per De Morgan's, we wanted:
(c < d) && !((a u>= b) && (a u<= b))
Note that this problem doesn't occur in the test-suite.
changeFPCCToAArch64CC produces disjunct CCs; here, one -> mi/gt.
We can't represent that in the fccmp chain; it can't express
arbitrary OR sequences, as one comment explains:
In general we can create code for arbitrary "... (and (and A B) C)"
sequences. We can also implement some "or" expressions, because
"(or A B)" is equivalent to "not (and (not A) (not B))" and we can
implement some negation operations. [...] However there is no way
to negate the result of a partial sequence.
Instead, introduce changeFPCCToANDAArch64CC, which produces the
conjunct cond codes:
- (a one b)
== ((a olt b) || (a ogt b))
== ((a ord b) && (a une b))
- (a ueq b)
== ((a uno b) || (a oeq b))
== ((a ule b) && (a uge b))
Note that, at first, one might think that, when PushNegate is true,
we should use the disjunct CCs, in effect doing:
(a || b)
= !(!a && !(b))
= !(!a && !(b1 || b2)) <- changeFPCCToAArch64CC(b, b1, b2)
= !(!a && !b1 && !b2)
However, we can take advantage of the fact that the CC is already
negated, which lets us avoid special-casing PushNegate and doing
the simpler to reason about:
(a || b)
= !(!a && (!b))
= !(!a && (b1 && b2)) <- changeFPCCToANDAArch64CC(!b, b1, b2)
= !(!a && b1 && b2)
This makes both emitConditionalCompare cases behave identically,
and produces correct ccmp sequences for the 2-CC fcmps.
llvm-svn: 258533
Summary:
SETCC with f16 vectors has OperationAction set to Expand but still gets
lowered to FCM* intrinsics based on its result type. This patch skips
lowering of VSETCC if the operand is an f16 vector.
v4 and v8 tests included.
Reviewers: ab, jmolloy
Subscribers: srhines, llvm-commits
Differential Revision: http://reviews.llvm.org/D15361
llvm-svn: 258471
When we have a single basic block, the explicit copy-back instructions should
be inserted right before the terminator. Before this fix, they were wrongly
placed at the beginning of the basic block.
I will commit fixes to other platforms as well.
PR26136
llvm-svn: 257929
Summary:
This pass may modify the Cmp operands. However, the flag reg may be used by both the branch and CSEL.
Modifying CMP will have side effect on CSEL.
Reviewers: t.p.northover
Subscribers: llvm-commits, aemerson, rengolin
Differential Revision: http://reviews.llvm.org/D16147
llvm-svn: 257844
Previous implementation in http://reviews.llvm.org/D10522
created external references to __emutls_v.* variables.
Such references are inaccurate and cannot be handled by
all linkers, e.g. Android dynamic and gold linkers for aarch64.
Now a new LowerEmuTLS pass to go through all global variables,
and add emutls_v.* and emutls_t.* variables.
These __emutls* variables have the same linkage and
visibility as the associated user defined TLS variable.
Also removed old code that dump __emutls* variables in AsmPrinter.cpp,
and updated TLS unit tests.
Differential Revision: http://reviews.llvm.org/D15300
llvm-svn: 257718
This is a recommit of r257253 which was reverted in r257270.
Previous testcase can make failure on some targets due to using opt with O3 option.
Original Summary:
Merge MBBICommon and MBBI's MMOs.
Differential Revision: http://reviews.llvm.org/D15990
llvm-svn: 257317
Summary:
In buildSchedGraph(), when adding memory dependencies for loads, move
the call to adjustChainDeps() after the call to
addChainDependency(AliasChain) to handle the case where
addChainDependency(AliasChain) ends up not adding a dependency and
instead putting the SU on the RejectMemNodes list. The call to
adjustChainDeps() must be done after the call to addChainDependency() in
order to process the SU added to the RejectMemNodes list to create
memory dependencies for it.
Reviewers: hfinkel, atrick, jonpa, resistor
Subscribers: mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D15927
llvm-svn: 256950
Summary:
There are a number of files in the tree which have been accidentally checked in with DOS line endings. Convert these to native line endings.
There are also a few files which have DOS line endings on purpose, and I have set the svn:eol-style property to 'CRLF' on those.
Reviewers: joerg, aaron.ballman
Subscribers: aaron.ballman, sanjoy, dsanders, llvm-commits
Differential Revision: http://reviews.llvm.org/D15848
llvm-svn: 256707
This is a recommit of r256004 which was reverted in r256160. The issue was the
incorrect promotion for half and byte loads transformed into mov instructions.
This fix will replace half and byte type loads only with bit field extracts.
Original commit message:
This change promotes load instructions which directly read from stored by
replacing them with mov instructions. If the store is wider than the load,
the load will be replaced with a bitfield extract.
For example :
STRWui %W1, %X0, 1
%W0 = LDRHHui %X0, 3
becomes
STRWui %W1, %X0, 1
%W0 = UBFMWri %W1, 16, 31
llvm-svn: 256249
This patch adds to the target description two additional patterns for matching
extract-extend operations to SMOV. The patterns catch the v16i8-to-i64 and
v8i16-to-i64 cases. The existing patterns miss these cases because the
extracted elements must first be legalized to i32, resulting in any_extend
nodes.
This was originally implemented as a DAG combine (r255895), but was reverted
due to failing out-of-tree tests.
llvm-svn: 256176
Disable post-ra scheduler for perturbed tests to appease the bots and to
preserve the history of the tests.
http://reviews.llvm.org/D15652
llvm-svn: 256158
This change promotes load instructions which directly read from stores by
replacing them with mov instructions. If the store is wider than the load,
the load will be replaced with a bitfield extract.
For example :
STRWui %W1, %X0, 1
%W0 = LDRHHui %X0, 3
becomes
STRWui %W1, %X0, 1
%W0 = UBFMWri %W1, 16, 31
llvm-svn: 256004
This patch adds a DAG combine for (any_extend (extract_vector_elt v, i)) ->
(extract_vector_elt v, i). The combine enables us to better match some SMOV
patterns.
Differential Revision: http://reviews.llvm.org/D15515
llvm-svn: 255895
The access function has a short entry and a short exit, the initialization
block is only run the first time. To improve the performance, we want to
have a short frame at the entry and exit.
We explicitly handle most of the CSRs via copies. Only the CSRs that are not
handled via copies will be in CSR_SaveList.
Frame lowering and prologue/epilogue insertion will generate a short frame
in the entry and exit according to CSR_SaveList. The majority of the CSRs will
be handled by register allcoator. Register allocator will try to spill and
reload them in the initialization block.
We add CSRsViaCopy, it will be explicitly handled during lowering.
1> we first set FunctionLoweringInfo->SplitCSR if conditions are met (the target
supports it for the given machine function and the function has only return
exits). We also call TLI->initializeSplitCSR to perform initialization.
2> we call TLI->insertCopiesSplitCSR to insert copies from CSRsViaCopy to
virtual registers at beginning of the entry block and copies from virtual
registers to CSRsViaCopy at beginning of the exit blocks.
3> we also need to make sure the explicit copies will not be eliminated.
The target independent portion was committed as r255353.
rdar://problem/23557469
Differential Revision: http://reviews.llvm.org/D15341
llvm-svn: 255821
PR25763 demonstrated an issue with D14683 - vector comparison constant folding only works for i1 results, so we need to split off the sign-extension of the result to the required type. Luckily this can be done with the existing type legalization code.
llvm-svn: 255289
Otherwise, we think that most types that look like they'd fit in a
legal vector type are legal (so, basically, *any* vector type with a
size between 33 and 128 bits, I think, since we use pow2 alignment;
e.g., v2i25, v3f32, ...).
DataLayout::getTypeAllocSize rounds up based on alignment.
When checking for target intrinsic legality, that's not what we want:
if rounding makes a difference, the type isn't legal, and the
target intrinsics shouldn't be used, as they are always assumed legal.
One could make the argument that alloc size is ultimately the most
relevant here, since we're dealing with LD/ST intrinsics. That's only
true if we did legalize them though; that's a problem for another day.
Use DataLayout::getTypeSizeInBits instead of getTypeAllocSizeInBits.
Type::getSizeInBits can't be used because that'd gratuitously break
pointer vector support.
Some of these uses are currently fine, because we only hit them when
the type is already known legal (e.g., r114454). Update them for
consistency. It's faster to avoid the rounding anyway!
llvm-svn: 255089
Summary:
This fixes failure when trying to select
insertelement <4 x half> undef, half %a, i64 0
which gets transformed to a scalar_to_vector node.
The accompanying v4 and v8 tests fail instruction selection without this
patch.
Reviewers: ab, jmolloy
Subscribers: srhines, llvm-commits
Differential Revision: http://reviews.llvm.org/D15322
llvm-svn: 255072
In the case of a conditional branch without a preceding cmp we used to emit
a "and; cmp; b.eq/b.ne" sequence, use tbz/tbnz instead.
Differential Revision: http://reviews.llvm.org/D15122
llvm-svn: 254621
The ARM ARM is clear that 128-bit loads are only guaranteed to have been atomic
if there has been a corresponding successful stxp. It's less clear for AArch32, so
I'm leaving that alone for now.
llvm-svn: 254524
We mustn't introduce a shift of exactly 64-bits for any inputs, since that's an
UNDEF value (and worse, it's not what you want with the natural Arch64
implementation).
The generated code is pretty horrific, but I couldn't come up with an obviously
better alternative (if the amount is constant EXTR could help). Turns out
128-bit shifts are just nasty.
rdar://22491037
llvm-svn: 254475
Summary:
When not useful bits, BitWidth becomes 0 and APInt will not be happy.
See https://llvm.org/bugs/show_bug.cgi?id=25571
We can just mark the operand as IMPLICIT_DEF is none bits of it is used.
Reviewers: t.p.northover, jmolloy
Subscribers: gberry, jmolloy, mgrang, aemerson, llvm-commits, rengolin
Differential Revision: http://reviews.llvm.org/D14803
llvm-svn: 254440
Summary:
Add and instructions immediately after loads that only have their low
bits used, assuming that the (and (load x) c) will be matched as a
extload and the ands/truncs fed by the extload will be removed by isel.
Reviewers: mcrosier, qcolombet, ab
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14584
llvm-svn: 253722
The included test only checks for a compiler crash for now. Several people are
facing this issue, so we first resolve the crash, and will increase shrinkwrap's
coverage later in a follow-up patch.
llvm-svn: 253718
This change merges adjacent zero stores into a wider single store.
For example :
strh wzr, [x0]
strh wzr, [x0, #2]
becomes
str wzr, [x0]
This will fix PR25410.
llvm-svn: 253711
This change extends r251438 to handle more narrow load promotions
including byte type, unscaled, and signed. For example, this change will
convert :
ldursh w1, [x0, #-2]
ldurh w2, [x0, #-4]
into
ldur w2, [x0, #-4]
asr w1, w2, #16
and w2, w2, #0xffff
llvm-svn: 253577
Note, this was reviewed (and more details are in) http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151109/312083.html
These intrinsics currently have an explicit alignment argument which is
required to be a constant integer. It represents the alignment of the
source and dest, and so must be the minimum of those.
This change allows source and dest to each have their own alignments
by using the alignment attribute on their arguments. The alignment
argument itself is removed.
There are a few places in the code for which the code needs to be
checked by an expert as to whether using only src/dest alignment is
safe. For those places, they currently take the minimum of src/dest
alignments which matches the current behaviour.
For example, code which used to read:
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 500, i32 8, i1 false)
will now read:
call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 8 %dest, i8* align 8 %src, i32 500, i1 false)
For out of tree owners, I was able to strip alignment from calls using sed by replacing:
(call.*llvm\.memset.*)i32\ [0-9]*\,\ i1 false\)
with:
$1i1 false)
and similarly for memmove and memcpy.
I then added back in alignment to test cases which needed it.
A similar commit will be made to clang which actually has many differences in alignment as now
IRBuilder can generate different source/dest alignments on calls.
In IRBuilder itself, a new argument was added. Instead of calling:
CreateMemCpy(Dst, Src, getInt64(Size), DstAlign, /* isVolatile */ false)
you now call
CreateMemCpy(Dst, Src, getInt64(Size), DstAlign, SrcAlign, /* isVolatile */ false)
There is a temporary class (IntegerAlignment) which takes the source alignment and rejects
implicit conversion from bool. This is to prevent isVolatile here from passing its default
parameter to the source alignment.
Note, changes in future can now be made to codegen. I didn't change anything here, but this
change should enable better memcpy code sequences.
Reviewed by Hal Finkel.
llvm-svn: 253511
If a section is rw, it is irrelevant if the dynamic linker will write to
it or not.
It looks like llvm implemented this because gcc was doing it. It looks
like gcc implemented this in the hope that it would put all the
relocated items close together and speed up the dynamic linker.
There are two problem with this:
* It doesn't work. Both bfd and gold will map .data.rel to .data and
concatenate the input sections in the order they are seen.
* If we want a feature like that, it can be implemented directly in the
linker since it knowns where the dynamic relocations are.
llvm-svn: 253436
SELECT_CC has the nasty property of having operands with unrelated
types. So if you do something like:
f32 = select_cc f16, f16, f32, f32, cc
You'd only look for the action for <select_cc, f32>, but never f16.
If the types are all legal, but the op isn't (as for f16 on AArch64,
or for f128 on x86_64/AArch64?), then you get into trouble.
For f128, we have softenSetCCOperands to handle this case.
Similarly, for f16, we can directly promote the CC operands.
llvm-svn: 253344
The way prelink used to work was
* The compiler decides if a given section only has relocations that
are know to point to the same DSO. If so, it names it
.data.rel.ro.local<something>.
* The static linker puts all of these together.
* The prelinker program assigns addresses to each library and resolves
the local relocations.
There are many problems with this:
* It is incompatible with address space randomization.
* The information passed by the compiler is redundant. The linker
knows if a given relocation is in the same DSO or not. If could sort
by that if so desired.
* There are newer ways of speeding up DSO (gnu hash for example).
* Even if we want to implement this again in the compiler, the previous
implementation is pretty broken. It talks about relocations that are
"resolved by the static linker". If they are resolved, there are none
left for the prelinker. What one needs to track is if an expression
will require only dynamic relocations that point to the same DSO.
At this point it looks like the prelinker is an historical curiosity.
For example, fedora has retired it because it failed to build for two
releases
(http://pkgs.fedoraproject.org/cgit/prelink.git/commit/?id=eb43100a8331d91c801ee3dcdb0a0bb9babfdc1f)
This patch removes support for it. That is, it stops printing the
".local" sections.
llvm-svn: 253280
Something I missed from Hal's review, rightly pointed out by Ben Kramer - we should make sure the expansion is properly checked as it can be easy for bugs to creep in.
I've checked the scalar i8 expansion here and the vector i8 expansion in a previous commit.
llvm-svn: 253024
Richard Trieu noted that UBSan detected an overflowing shift, and the obvious fix caused a crash.
What was happening was that the shiftee (1U) was indeed too small for the possible range of shifts it had to handle, but also we were using "VT.getSizeInBits()" to get the maximum type bitwidth, but we wanted "VT.getScalarSizeInBits()" to get the vector lane size instead of the entire vector size.
Use an APInt for the shift and VT.getScalarSizeInBits().
llvm-svn: 253023
Several backends have instructions to reverse the order of bits in an integer. Conceptually matching such patterns is similar to @llvm.bswap, and it was mentioned in http://reviews.llvm.org/D14234 that it would be best if these patterns were matched in InstCombine instead of reimplemented in every different target.
This patch introduces an intrinsic @llvm.bitreverse.i* that operates similarly to @llvm.bswap. For plumbing purposes there is also a new ISD node ISD::BITREVERSE, with simple expansion and promotion support.
The intention is that InstCombine's BSWAP detection logic will be extended to support BITREVERSE too, and @llvm.bitreverse intrinsics emitted (if the backend supports lowering it efficiently).
llvm-svn: 252878
Summary:
Don't fold
(zext (and (load x), cst)) -> (and (zextload x), (zext cst))
if
(and (load x) cst)
will match as a zextload already and has additional users.
For example, the following IR:
%load = load i32, i32* %ptr, align 8
%load16 = and i32 %load, 65535
%load64 = zext i32 %load16 to i64
store i32 %load16, i32* %dst1, align 4
store i64 %load64, i64* %dst2, align 8
used to produce the following aarch64 code:
ldr w8, [x0]
and w9, w8, #0xffff
and x8, x8, #0xffff
str w9, [x1]
str x8, [x2]
but with this change produces the following aarch64 code:
ldrh w8, [x0]
str w8, [x1]
str x8, [x2]
Reviewers: resistor, mcrosier
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14340
llvm-svn: 252789
For big-endian targets, when we merge two halfword loads into a word load, the
order of the halfwords in the loaded value is reversed compared to
little-endian, so the load-store optimiser needs to swap the destination
registers.
This does not affect merging of two word loads, as we use ldp, which treats the
memory as two separate 32-bit words.
llvm-svn: 252597
AArch64 has the ability to use the top 8-bits of an "address" for extra
information, with the memory subsystem automatically masking them off for loads
and stores. When that's happening, we can sometimes skip masks on memory
operations in the compiler.
However, this requires the host OS and support stack to preserve those bits so
it can't be enabled everywhere. In principle iOS 8.0 and above do take the
required precautions and but we'll put it under a flag for now.
llvm-svn: 252573
Summary:
This matches the sum-of-absdiff patterns emitted by the vectoriser using log2 shuffles.
Relies on D14207 to be able to match the `extract_subvector(..., 0)`
Reviewers: t.p.northover, jmolloy
Subscribers: aemerson, llvm-commits, rengolin
Differential Revision: http://reviews.llvm.org/D14208
llvm-svn: 252465
We used to try to constant-fold them to i32 immediates.
Given that fast-isel doesn't otherwise support vNi1, when selecting
the result users, we'd fallback to SDAG anyway.
However, if the users were in another block, we'd insert broken
cross-class copies (GPR32 to FPR64).
Give up, let SDAG agree with itself on a vNi1 legalization strategy.
llvm-svn: 252364
The benefit from converting narrow loads into a wider load (r251438) could be
micro-architecturally dependent, as it assumes that a single load with two bitfield
extracts is cheaper than two narrow loads. Currently, this conversion is
enabled only in cortex-a57 on which performance benefits were verified.
llvm-svn: 252316
Previously, subprograms contained a metadata reference to the function they
described. Because most clients need to get or set a subprogram for a given
function rather than the other way around, this created unneeded inefficiency.
For example, many passes needed to call the function llvm::makeSubprogramMap()
to build a mapping from functions to subprograms, and the IR linker needed to
fix up function references in a way that caused quadratic complexity in the IR
linking phase of LTO.
This change reverses the direction of the edge by storing the subprogram as
function-level metadata and removing DISubprogram's function field.
Since this is an IR change, a bitcode upgrade has been provided.
Fixes PR23367. An upgrade script for textual IR for out-of-tree clients is
attached to the PR.
Differential Revision: http://reviews.llvm.org/D14265
llvm-svn: 252219
This recommits r250719, which caused a failure in SPEC2000.gcc
because of the incorrect insert point for the new wider load.
Convert two halfword loads into a single 32-bit word load with bitfield extract
instructions. For example :
ldrh w0, [x2]
ldrh w1, [x2, #2]
becomes
ldr w0, [x2]
ubfx w1, w0, #16, #16
and w0, w0, #ffff
llvm-svn: 251438
Summary: After D13851 landed, we saw backend crashes when compiling the reduced test case included in this patch. The right fix seems to be to allow these vector types for expansion in instruction selection.
Reviewers: rengolin, t.p.northover
Subscribers: RKSimon, t.p.northover, aemerson, llvm-commits, rengolin
Differential Revision: http://reviews.llvm.org/D14082
llvm-svn: 251401
Convert two halfword loads into a single 32-bit word load with bitfield extract
instructions. For example :
ldrh w0, [x2]
ldrh w1, [x2, #2]
becomes
ldr w0, [x2]
ubfx w1, w0, #16, #16
and w0, w0, #ffff
llvm-svn: 250719
This basic combine was surprisingly missing.
AMDGPU legalizes many operations in terms of 32-bit vector components,
so not doing this results in many extra copies and subregister extracts
that need to be cleaned up later.
InstCombine already does this for the hasOneUse case. The target hook
is to fix a handful of tests which break (e.g. ARM/vmov.ll) which turn
from a vector materialize repeated immediate instruction to a constant
vector load with more scalar copies from it.
llvm-svn: 250129
Without an additional check for NEON, the compiler crashes during
legalization of NEON ldN/stN.
Differential Revision: http://reviews.llvm.org/D13508
llvm-svn: 249550
Unscaled load/store combining has been enabled since the initial ARM64 port. No
need for a redundance run. Also, add CHECK-LABEL directives.
llvm-svn: 248945
Previously, the index was constrained to the size of the memory operation for
no apparent reason. This change removes that constraint so that we can form
pre-index instructions with any valid offset.
llvm-svn: 248931
BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change:
Pros:
1. It uses only a half space of the current one.
2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer.
Cons:
1. Constructing a probability using arbitrary numerator and denominator needs additional calculations.
2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio.
One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern.
Differential revision: http://reviews.llvm.org/D12603
llvm-svn: 248633
This is a redo of D7208 ( r227242 - http://llvm.org/viewvc/llvm-project?view=revision&revision=227242 ).
The patch was reverted because an AArch64 target could infinite loop after the change in DAGCombiner
to merge vector stores. That happened because AArch64's allowsMisalignedMemoryAccesses() wasn't telling
the truth. It reported all unaligned memory accesses as fast, but then split some 128-bit unaligned
accesses up in performSTORECombine() because they are slow.
This patch attempts to fix the problem in AArch's allowsMisalignedMemoryAccesses() while preserving
existing (perhaps questionable) lowering behavior.
The x86 test shows that store merging is working as intended for a target with fast 32-byte unaligned
stores.
Differential Revision: http://reviews.llvm.org/D12635
llvm-svn: 248622
This patch changes the order of GEPs generated by Splitting GEPs
pass, specially when one of the GEPs has constant and the base is
loop invariant, then we will generate the GEP with constant first
when beneficial, to expose more cases for LICM.
If originally Splitting GEP generate the following:
do.body.i:
%idxprom.i = sext i32 %shr.i to i64
%2 = bitcast %typeD* %s to i8*
%3 = shl i64 %idxprom.i, 2
%uglygep = getelementptr i8, i8* %2, i64 %3
%uglygep7 = getelementptr i8, i8* %uglygep, i64 1032
...
Now it genereates:
do.body.i:
%idxprom.i = sext i32 %shr.i to i64
%2 = bitcast %typeD* %s to i8*
%3 = shl i64 %idxprom.i, 2
%uglygep = getelementptr i8, i8* %2, i64 1032
%uglygep7 = getelementptr i8, i8* %uglygep, i64 %3
...
For no-loop cases, the original way of generating GEPs seems to
expose more CSE cases, so we don't change the logic for no-loop
cases, and only limit our change to the specific case we are
interested in.
llvm-svn: 248420
In the comparison failure block of a cmpxchg expansion, the initial
ldrex/ldxr will not be followed by a matching strex/stxr.
On ARM/AArch64, this unnecessarily ties up the execution monitor,
which might have a negative performance impact on some uarchs.
Instead, release the monitor in the failure block.
The clrex instruction was designed for this: use it.
Also see ARMARM v8-A B2.10.2:
"Exclusive access instructions and Shareable memory locations".
Differential Revision: http://reviews.llvm.org/D13033
llvm-svn: 248291
The C standard has historically not specified whether or not these functions should raise the inexact flag. Traditionally on Darwin, these functions *did* raise inexact, and the llvm lowerings followed that conventions. n1778 (C bindings for IEEE-754 (2008)) clarifies that these functions should not set inexact. This patch brings the lowerings for arm64 and x86 in line with the newly specified behavior. This also lets us fold some logic into TD patterns, which is nice.
Differential Revision: http://reviews.llvm.org/D12969
llvm-svn: 248266
Summary:
For bitfield insert OR matching, check both operands for larger pattern
first before checking for smaller pattern.
Add pattern for unsigned bitfield insert-in-zero done with SHL+AND.
Resolves PR21631.
Reviewers: jmolloy, t.p.northover
Subscribers: aemerson, rengolin, llvm-commits, mcrosier
Differential Revision: http://reviews.llvm.org/D12908
llvm-svn: 248006
- Strenghten the logic to be sure we hoist the restore point out of the current
loop. (The fixes a bug with infinite loop, added as part of the patch.)
- Walk over the exit blocks of the current loop to conver to the desired restore
point in one iteration of the update loop.
llvm-svn: 247958
Summary:
The BUILD_VECTOR node will truncate its operators to match the
type. We need to take this into account when constant folding -
we need to perform a truncation before constant folding the elements.
This is because the upper bits can change the result, depending on
the operation type (for example this is the case for min/max).
This change also adds a regression test.
Reviewers: jmolloy
Subscribers: jmolloy, llvm-commits
Differential Revision: http://reviews.llvm.org/D12697
llvm-svn: 247265
First, we need to teach isFrameOffsetLegal about STNP.
It already knew about the STP/LDP variants, but those were probably
never exercised, because it's only the load/store optimizer that
generates STP/LDP, and the only user of the method is frame lowering,
which runs earlier.
The STP/LDP cases were wrong: they didn't take into account the fact
that they return two results, not one, so the immediate offset will be
the 4th operand, not the 3rd.
Follow-up to r247234.
llvm-svn: 247236
We could go through the load/store optimizer and match STNP where
we would have matched a nontemporal-annotated STP, but that's not
reliable enough, as an opportunistic optimization.
Insetad, we can guarantee emitting STNP, by matching them at ISel.
Since there are no single-input nontemporal stores, we have to
resort to some high-bits-extracting trickery to generate an STNP
from a plain store.
Also, we need to support another, LDP/STP-specific addressing mode,
base + signed scaled 7-bit immediate offset.
For now, only match the base. Let's make it smart separately.
Part of PR24086.
llvm-svn: 247231
This patch allows the mixing of scaled and unscaled load/stores to form
load/store pairs.
PR24465
http://reviews.llvm.org/D12116
Many thanks to Ahmed and Michael for fixes and code review.
llvm-svn: 246769
This patch uses the metadata defined in D12341 to avoid creating an unpredictable branch.
Differential Revision: http://reviews.llvm.org/D12342
llvm-svn: 246692
This matches the ARM behavior. In both cases, the register is part
of the optional Performance Monitors extension, so, add the feature,
and enable it for the A-class processors we support.
Differential Revision: http://reviews.llvm.org/D12425
llvm-svn: 246555
Currently, when edge weights are assigned to edges that are created when lowering switch statement, the weight on the edge to default statement (let's call it "default weight" here) is not considered. We need to distribute this weight properly. However, without value profiling, we have no idea how to distribute it. In this patch, I applied the heuristic that this weight is evenly distributed to successors.
For example, given a switch statement with cases 1,2,3,5,10,11,20, and every edge from switch to each successor has weight 10. If there is a binary search tree built to test if n < 10, then its two out-edges will have weight 4x10+10/2 = 45 and 3x10 + 10/2 = 35 respectively (currently they are 40 and 30 without considering the default weight). Each distribution (which is 5 here) will be stored in each SwitchWorkListItem for further distribution.
There are some exceptions:
For a jump table header which doesn't have any edge to default statement, we don't distribute the default weight to it.
For a bit test header which covers a contiguous range and hence has no edges to default statement, we don't distribute the default weight to it.
When the branch checks a single value or a contiguous range with no edge to default statement, we don't distribute the default weight to it.
In other cases, the default weight is evenly distributed to successors.
Differential Revision: http://reviews.llvm.org/D12418
llvm-svn: 246522
The ISelLowering code turned insertion turned the element for the
lowest lane of a BUILD_VECTOR into an INSERT_SUBREG, this prohibited
the patterns for SCALAR_TO_VECTOR(Load) to match later. Restrict this
to cases without a load argument.
Reported in rdar://22223823
Differential Revision: http://reviews.llvm.org/D12467
llvm-svn: 246462
As a follow-up to r246098, require `DISubprogram` definitions
(`isDefinition: true`) to be 'distinct'. Specifically, add an assembler
check, a verifier check, and bitcode upgrading logic to combat testcase
bitrot after the `DIBuilder` change.
While working on the testcases, I realized that
test/Linker/subprogram-linkonce-weak-odr.ll isn't relevant anymore. Its
purpose was to check for a corner case in PR22792 where two subprogram
definitions match exactly and share the same metadata node. The new
verifier check, requiring that subprogram definitions are 'distinct',
precludes that possibility.
I updated almost all the IR with the following script:
git grep -l -E -e '= !DISubprogram\(.* isDefinition: true' |
grep -v test/Bitcode |
xargs sed -i '' -e 's/= \(!DISubprogram(.*, isDefinition: true\)/= distinct \1/'
Likely some variant of would work for out-of-tree testcases.
llvm-svn: 246327
more than 2 instructions.
I introduced this regression a while back and did not noticed it because I
somehow forgot to push the initial test cases for the pass!
Fix that as well!
llvm-svn: 246239
When producing conditional compare sequences for or operations we need
to negate the operands and the finally tested flags. The thing is if we negate
the finally tested flags this equals a logical negation of all previously
emitted expressions. There was a case missing where we have to order OR
expressions so they get emitted first.
This fixes http://llvm.org/PR24459
llvm-svn: 245641
Create CMP;CCMP sequences from and/or trees does not gain us anything if
the and/or tree is materialized to a GP register anyway. While most of
the code already checked for hasOneUse() there was one important case
missing.
llvm-svn: 245640
We are already falling back to SelectionDAG when encountering an shift with UB.
This adds the same checks for shifts with UB that get folded into arithmetic or
logical operations.
This fixes rdar://problem/22345295.
llvm-svn: 245499
The current code normalizes select(C0, x, select(C1, x, y)) towards
select(C0|C1, x, y) if the targets prefers that form. This patch adds an
additional rule that if the select(C1, x, y) part already exists in the
function then we want to normalize into the other direction because the
effects of reusing the existing value are bigger than transforming into
the target preferred form.
This addresses regressions following r238793, see also:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20150727/290272.html
Differential Revision: http://reviews.llvm.org/D11616
llvm-svn: 245350
These only get generated if the target supports them. If one of the variants is not legal and the other is, and it is safe to do so, the other variant will be emitted.
For example on AArch32 (V8), we have scalar fminnm but not fmin.
Fix up a couple of tests while we're here - one now produces better code, and the other was just plain wrong to start with.
llvm-svn: 245196
For cases where we TRUNCATE and then ZERO_EXTEND to a larger size (often from vector legalization), see if we can mask the source data and then ZERO_EXTEND (instead of after a ANY_EXTEND). This can help avoid having to generate a larger mask, and possibly applying it to several sub-vectors.
(zext (truncate x)) -> (zext (and(x, m))
Includes a minor patch to SystemZ to better recognise 8/16-bit zero extension patterns from RISBG bit-extraction code.
This is the first of a number of minor patches to help improve the conversion of byte masks to clear mask shuffles.
Differential Revision: http://reviews.llvm.org/D11764
llvm-svn: 245160
We canonicalize V64 vectors to V128 through insert_subvector: the other
FMLA/FMLS/FMUL/FMULX patterns match that already, but this one doesn't,
so we'd fail to match fmls and generate fneg+fmla instead.
The vector equivalents are already tested and functional.
llvm-svn: 245107
Spotted by Ahmed - in r244594 I inadvertently marked f16 min/max as legal.
I've reverted it here, and marked min/max on scalar f16's as promote. I've also added a testcase. The test just checks that the compiler doesn't fall over - it doesn't create fmin nodes for f16 yet.
llvm-svn: 245035
These tests relied on -enable-no-nans-fp-math, whereas soon they'll take their no-nans hint
from the FCMP instruction itself, so split the no-nans stuff out into its own test.
Also do a slight rejig of instruction order. The old FMIN/MAX backend matching had to deal with looking through casts, which it never did particularly well. Now, instcombine will recognize such patterns and canonicalize the cast outside the select. So modify the test inputs to assume that instcombine has already run.
llvm-svn: 244913
Now that we can properly promote mismatched FCOPYSIGNs (r244858), we
can mark the FP_ROUND on the result as truncating, to expose folding.
FCOPYSIGN doesn't change anything but the sign bit, so
(fp_round (fcopysign (fpext a), b))
is equivalent to (modulo the sign bit):
(fp_round (fpext a))
which is a no-op.
llvm-svn: 244862
We can lower them using our cool tricks if we fpext/fptrunc the second
input, like we do for f32/f64.
Follow-up to r243924, r243926, and r244858.
llvm-svn: 244860
We don't care about its type, and there's even a combine that'll fold
away the FP_EXTEND if we let it run. However, until it does, we'll have
something broken like:
(f32 (fp_extend (f64 v)))
Scalar f16 follow-up to r243924.
llvm-svn: 244858
r242520 was reverted in r244313 as the expected behaviour of the alias
attribute in C is that the alias has the same size as the aliasee. However
we can re-introduce adding the size on the alias when the aliasee does not,
from a source code or object perspective, exist as a discrete entity. This
happens when the aliasee is not a symbol, or when that symbol is private.
Differential Revision: http://reviews.llvm.org/D11943
llvm-svn: 244752
On Mach-O emitting aliases for the variables that make up a MergedGlobals
variable can cause problems when linking with dead stripping enabled so don't
do that, except for external variables where we must emit an alias.
llvm-svn: 244748
Other objects can never reference the MergedGlobals symbol so external linkage
is never needed. Using private instead of internal linkage means the object is
more similar to what it looks like when global merging is not enabled, with
the only difference being that the merged variables are addressed indirectly
relative to the start of the section they are in.
Also add aliases for merged variables with internal linkage, as this also makes
the object be more like what it is when they are not merged.
Differential Revision: http://reviews.llvm.org/D11942
llvm-svn: 244615
Lower Intrinsic::aarch64_neon_fmin/fmax to fminnum/fmannum and match that instead. Minimal functional change:
- Extra tests added because coverage of scalar fminnm/fmaxnm instructions was nonexistant.
- f16 test updated because now we actually generate scalar fminnm/fmaxnm we no longer need to bail out to a libcall!
llvm-svn: 244595
frame setup instruction.
This commit ensures that the stack map lowering code in FastISel adds an
appropriate number of immediate operands to the frame setup instruction.
The previous code added just one immediate operand, which was fine for a target
like AArch64, but on X86 the ADJCALLSTACKDOWN64 instruction needs two explicit
operands. This caused the machine verifier to report an error when the old code
added just one.
Reviewers: Juergen Ributzka
Differential Revision: http://reviews.llvm.org/D11853
llvm-svn: 244508
I looked into adding a warning / error for this to FileCheck, but there doesn't
seem to be a good way to avoid it triggering on the instances of it in RUN lines.
llvm-svn: 244481
When we are not emitting the condition for the branch, because the condition is
in another BB or SDAG did the selection for us, then we have to mask the flag in
the register with AND.
This is required when the condition comes from a truncate, because SDAG only
truncates down to a legal size of i32.
This fixes rdar://problem/22161062.
llvm-svn: 244291
This reverts commit r243198 and 243304.
Turns out this wasn't the correct fix for this problem. It works only within
FastISel, but fails when the truncate is selected by SDAG.
llvm-svn: 244287
points.
There is an infinite loop that can occur in Shrink Wrapping while searching
for the Save/Restore points.
Part of this search checks whether the save/restore points are located in
different loop nests and if so, uses the (post) dominator trees to find the
immediate (post) dominator blocks. However, if the current block does not have
any immediate (post) dominators then this search will result in an infinite
loop. This can occur in code containing an infinite loop.
The modification checks whether the immediate (post) dominator is different from
the current save/restore block. If it is not, then the search terminates and the
current location is not considered as a valid save/restore point for shrink wrapping.
Phabricator: http://reviews.llvm.org/D11607
llvm-svn: 244247
There's a bunch of code in LowerFCOPYSIGN that does smart lowering, and
is actually already vector-aware; let's use it instead of scalarizing!
The only interesting change is that for v2f32, we previously always used
use v4i32 as the integer vector type.
Use v2i32 instead, and mark FCOPYSIGN as Custom.
llvm-svn: 243926
We used to legalize it like it's any other binary operations. It's not,
because it accepts mismatched operand types. Because of that, we used
to hit various asserts and miscompiles.
Specialize vector legalizations to, in the worst case, unroll, or, when
possible, to just legalize the operand that needs legalization.
Scalarization isn't covered, because I can't think of a target where
some but not all of the 1-element vector types are to be scalarized.
llvm-svn: 243924
Since r241097, `DIBuilder` has only created distinct `DICompileUnit`s.
The backend is liable to start relying on that (if it hasn't already),
so make uniquable `DICompileUnit`s illegal and automatically upgrade old
bitcode. This is a nice cleanup, since we can remove an unnecessary
`DenseSet` (and the associated uniquing info) from `LLVMContextImpl`.
Almost all the testcases were updated with this script:
git grep -e '= !DICompileUnit' -l -- test |
grep -v test/Bitcode |
xargs sed -i '' -e 's,= !DICompileUnit,= distinct !DICompileUnit,'
I imagine something similar should work for out-of-tree testcases.
llvm-svn: 243885
Remove the fake `DW_TAG_auto_variable` and `DW_TAG_arg_variable` tags,
using `DW_TAG_variable` in their place Stop exposing the `tag:` field at
all in the assembly format for `DILocalVariable`.
Most of the testcase updates were generated by the following sed script:
find test/ -name "*.ll" -o -name "*.mir" |
xargs grep -l 'DILocalVariable' |
xargs sed -i '' \
-e 's/tag: DW_TAG_arg_variable, //' \
-e 's/tag: DW_TAG_auto_variable, //'
There were only a handful of tests in `test/Assembly` that I needed to
update by hand.
(Note: a follow-up could change `DILocalVariable::DILocalVariable()` to
set the tag to `DW_TAG_formal_parameter` instead of `DW_TAG_variable`
(as appropriate), instead of having that logic magically in the backend
in `DbgVariable`. I've added a FIXME to that effect.)
llvm-svn: 243774
Summary:
Favor the extended reg patterns over the shifted reg patterns that match
only the operand shift and not the full sign/zero extend and shift.
Reviewers: jmolloy, t.p.northover
Subscribers: mcrosier, aemerson, llvm-commits, rengolin
Differential Revision: http://reviews.llvm.org/D11569
llvm-svn: 243753
Tim Northover