Summary:
Darwin dynamic linker can handle weak symbols in ConstDataSection.
ReadonReadOnlyWithRel symbols should be emitted in ConstDataSection
instead of normal DataSection.
rdar://problem/39298457
Reviewers: dexonsmith, kledzik
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D45472
llvm-svn: 329752
In the presence of variable-sized stack objects, we always picked the
base pointer when resolving frame indices if it was available.
This makes us hit an assert where we can't reach the emergency spill
slot if it's too far away from the base pointer. Since on AArch64 we
decide to place the emergency spill slot at the top of the frame, it
makes more sense to use FP to access it.
The changes here don't affect only emergency spill slots but all the
frame indices. The goal here is to try to choose between FP, BP and SP
so that we minimize the offset and avoid scavenging, or worse, asserting
when trying to access a slot allocated by the scavenger.
Previously discussed here: https://reviews.llvm.org/D40876.
Differential Revision: https://reviews.llvm.org/D45358
llvm-svn: 329691
Summary:
For OS type AMDPAL, the scratch descriptor is loaded from offset 0 of
the GIT, whose 32 bit pointer is in s0 (s8 for gfx9 merged shaders).
This commit fixes that to use offset 0x10 instead of offset 0 for a
compute shader, per the PAL ABI spec.
V2: Ensure s0 (s8 for gfx9 merged shader) is marked live-in when loading
scratch descriptor from GIT.
Reviewers: kzhuravl, nhaehnle, timcorringham
Subscribers: kzhuravl, wdng, yaxunl, t-tye, llvm-commits, dstuttard, nhaehnle, arsenm
Differential Revision: https://reviews.llvm.org/D44468
Change-Id: I93dffa647758e37f613bb5e0dfca840d82e6d26f
llvm-svn: 329690
This cleans up a number of operations that only claimed te use EFLAGS
due to using DF. But no instructions which we think of us setting EFLAGS
actually modify DF (other than things like popf) and so this needlessly
creates uses of EFLAGS that aren't really there.
In fact, DF is so restrictive it is pretty easy to model. Only STD, CLD,
and the whole-flags writes (WRFLAGS and POPF) need to model this.
I've also somewhat cleaned up some of the flag management instruction
definitions to be in the correct .td file.
Adding this extra register also uncovered a failure to use the correct
datatype to hold X86 registers, and I've corrected that as necessary
here.
Differential Revision: https://reviews.llvm.org/D45154
llvm-svn: 329673
Prefer to use the 32-bit AND with immediate instead.
Primarily I'm doing this to ensure that immediates created by shrinkAndImmediate will always get absorbed into the AND. But I do believe this would be a reduction in the number of uops that need to execute. Ideally we should shrink the 'and' and the 'load' during DAG combine to re-enable the fold.
Fixes PR37063.
llvm-svn: 329667
The key idea is to lower COPY nodes populating EFLAGS by scanning the
uses of EFLAGS and introducing dedicated code to preserve the necessary
state in a GPR. In the vast majority of cases, these uses are cmovCC and
jCC instructions. For such cases, we can very easily save and restore
the necessary information by simply inserting a setCC into a GPR where
the original flags are live, and then testing that GPR directly to feed
the cmov or conditional branch.
However, things are a bit more tricky if arithmetic is using the flags.
This patch handles the vast majority of cases that seem to come up in
practice: adc, adcx, adox, rcl, and rcr; all without taking advantage of
partially preserved EFLAGS as LLVM doesn't currently model that at all.
There are a large number of operations that techinaclly observe EFLAGS
currently but shouldn't in this case -- they typically are using DF.
Currently, they will not be handled by this approach. However, I have
never seen this issue come up in practice. It is already pretty rare to
have these patterns come up in practical code with LLVM. I had to resort
to writing MIR tests to cover most of the logic in this pass already.
I suspect even with its current amount of coverage of arithmetic users
of EFLAGS it will be a significant improvement over the current use of
pushf/popf. It will also produce substantially faster code in most of
the common patterns.
This patch also removes all of the old lowering for EFLAGS copies, and
the hack that forced us to use a frame pointer when EFLAGS copies were
found anywhere in a function so that the dynamic stack adjustment wasn't
a problem. None of this is needed as we now lower all of these copies
directly in MI and without require stack adjustments.
Lots of thanks to Reid who came up with several aspects of this
approach, and Craig who helped me work out a couple of things tripping
me up while working on this.
Differential Revision: https://reviews.llvm.org/D45146
llvm-svn: 329657
LowerIntUnary as its name says has an assert for integer types. But for the bitcast case one side might be an FP type.
Rather than making sure the function really works for fp types and renaming it. Just do really basic splitting directly. The LowerIntUnary has the advantage that it can peek through BUILD_VECTOR because every other call is during Lowering. But these calls are during legalization and will be followed by a DAG combine round.
Revert some change to LowerVectorIntUnary that were originally made just to make these two calls work even in pure integer cases.
This was found purely by compiling the avx512f-builtins.c test from clang so I've copied over the offending function from that.
llvm-svn: 329616
This is a code size win in code that takes offseted addresses
frequently, such as C++ constructors that typically need to compute
an offseted address of a vtable. It reduces the size of Chromium for
Android's .text section by 46KB, or 56KB with ThinLTO (which exposes
more opportunities to use a direct access rather than a GOT access).
Because the addend range is limited in COFF and Mach-O, this is
enabled for ELF only.
Differential Revision: https://reviews.llvm.org/D45199
llvm-svn: 329611
In somes cases fast-isel fails to remove the and/shifts and uses blends or conditional moves.
But once masking gets involved, fast-isel aborts on the mask portion and we DAG combine more thorougly.
llvm-svn: 329604
While it appears to be correct information based on Intel's optimization manual and Agner's data, it causes perf regressions on a couple of the benchmarks in our internal list.
llvm-svn: 329593
Author: Samuel Pitoiset
ds_read_b128 and ds_write_b128 have been recently enabled
under the amdgpu-ds128 option because the performance benefit
is unclear.
Though, using 128-bit loads/stores for the local address space
appears to introduce regressions in tessellation shaders. Not
sure what is broken, but as ds_read_b128/ds_write_b128 are not
enabled by default, just introduce a global option and enable
128-bit only if requested (until it's fixed/used correctly).
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105464
llvm-svn: 329591
Without the fast math flags, the llvm.experimental.vector.reduce.fadd/fmul intrinsic expansions must be expanded in order.
This patch scalarizes the reduction, applying the accumulator at the start of the sequence: ((((Acc + Scl[0]) + Scl[1]) + Scl[2]) + ) ... + Scl[NumElts-1]
Differential Revision: https://reviews.llvm.org/D45366
llvm-svn: 329585
This patch fixes an issue exposed on the SystemZ build bots when committing
https://reviews.llvm.org/rL327856. The hoisting was temporarily disabled with
an option. This patch now re-enables hoisting and checks that we only hoist a
store instruction when all its operands are either constant caller preserved
registers or immediates.
Differential Revision: https://reviews.llvm.org/D45286
llvm-svn: 329577
Recommitting r329283, third time lucky...
If the SRL node is only used by an AND, we may be able to set the
ExtVT to the width of the mask, making the AND redundant. To support
this, another check has been added in isLegalNarrowLoad which queries
whether the load is valid.
Differential Revision: https://reviews.llvm.org/D41350
llvm-svn: 329551
Summary:
Currently MachineLoopInfo is used in only two places:
1) for computing IsBasicBlockInsideInnermostLoop field of MCCodePaddingContext, and it is never used.
2) in emitBasicBlockLoopComments, which is called only if `isVerbose()` is true.
Despite that, we currently have a dependency on MachineLoopInfo, which makes
pass manager to compute it and MachineDominator Tree. This patch removes the
use (1) and makes the use (2) lazy, thus avoiding some redundant
recomputations.
Reviewers: opaparo, gadi.haber, rafael, craig.topper, zvi
Subscribers: rengolin, javed.absar, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D44812
llvm-svn: 329542
Summary:
Cmov and setcc previously used WriteALU, but on Intel processors at least they are more restricted than basic ALU ops.
This patch adds new SchedWrites for them and removes the InstRWs. I had to leave some InstRWs for CMOVA/CMOVBE and SETA/SETBE because those have an extra uop relative to the other condition codes on Intel CPUs.
The test changes are due to fixing a missing ZnAGU dependency on the memory form of setcc.
Reviewers: RKSimon, andreadb, GGanesh
Reviewed By: RKSimon
Subscribers: GGanesh, llvm-commits
Differential Revision: https://reviews.llvm.org/D45380
llvm-svn: 329539
Summary:
This removes the InstRWs for BLENDVPS/PD in favor of WriteFVarBlend. The latency listed was 3 cycles but WriteFVarBlend is defined as 1 cycle latency. The 1 cycle latency matches Agner Fog's data.
The patterns were missing the VEX forms which is why there are no test changes. We don't test "-mcpu=znver1 -mattr=-avx"
Reviewers: RKSimon, GGanesh
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D44841
llvm-svn: 329538
In our real world application, we found the following optimization is missed in DAGCombiner
(zext (and/or/xor (shl/shr (load x), cst), cst)) -> (and/or/xor (shl/shr (zextload x), (zext cst)), (zext cst))
If the user of original zext is an add, it may enable further lea optimization on x86.
This patch add a new function CombineZExtLogicopShiftLoad to do this optimization.
Differential Revision: https://reviews.llvm.org/D44402
llvm-svn: 329516
Previously we used a custom lowering for this because of the AVX1 splitting requirement. But we can do the split during DAG combine if we check the types and subtarget
llvm-svn: 329510
Should fix UBSan bot by also checking there's no "uwtable" attribute
before skipping. Otherwise the unwind table will be useless since its
moves expect CSRs to actually be preserved.
A noreturn nounwind function can be expected to never return in any way, and by
never returning it will also never have to restore any callee-saved registers
for its caller. This makes it possible to skip spills of those registers during
function entry, saving some stack space and time in the process. This is rather
useful for embedded targets with limited stack space.
Should fix PR9970.
Patch mostly by myeisha (pmb).
llvm-svn: 329494
Previously HalfTy was not handled which would either trigger an assertion,
or result in array initialized with garbage.
Differential Revision: https://reviews.llvm.org/D45391
llvm-svn: 329463
v2f16 is a special case in NVPTX. v4f16 may be loaded as a pair of v2f16
and that was not previously handled correctly by tryLDGLDU()
Differential Revision: https://reviews.llvm.org/D45339
llvm-svn: 329456
Summary:
This patch implements a tablegen-driven Instruction Compression
mechanism for generating RISCV compressed instructions
(C Extension) from the expanded instruction form.
This tablegen backend processes CompressPat declarations in a
td file and generates all the compile-time and runtime checks
required to validate the declarations, validate the input
operands and generate correct instructions.
The checks include validating register operands, immediate
operands, fixed register operands and fixed immediate operands.
Example:
class CompressPat<dag input, dag output> {
dag Input = input;
dag Output = output;
list<Predicate> Predicates = [];
}
let Predicates = [HasStdExtC] in {
def : CompressPat<(ADD GPRNoX0:$rs1, GPRNoX0:$rs1, GPRNoX0:$rs2),
(C_ADD GPRNoX0:$rs1, GPRNoX0:$rs2)>;
}
The result is an auto-generated header file
'RISCVGenCompressEmitter.inc' which exports two functions for
compressing/uncompressing MCInst instructions, plus
some helper functions:
bool compressInst(MCInst& OutInst, const MCInst &MI,
const MCSubtargetInfo &STI,
MCContext &Context);
bool uncompressInst(MCInst& OutInst, const MCInst &MI,
const MCRegisterInfo &MRI,
const MCSubtargetInfo &STI);
The clients that include this auto-generated header file and
invoke these functions can compress an instruction before emitting
it, in the target-specific ASM or ELF streamer, or can uncompress
an instruction before printing it, when the expanded instruction
format aliases is favored.
The following clients were added to implement compression\uncompression
for RISCV:
1) RISCVAsmParser::MatchAndEmitInstruction:
Inserted a call to compressInst() to compresses instructions
parsed by llvm-mc coming from an ASM input.
2) RISCVAsmPrinter::EmitInstruction:
Inserted a call to compressInst() to compress instructions that
were lowered from Machine Instructions (MachineInstr).
3) RVInstPrinter::printInst:
Inserted a call to uncompressInst() to print the expanded
version of the instruction instead of the compressed one (e.g,
add s0, s0, a5 instead of c.add s0, a5) when -riscv-no-aliases
is not passed.
This patch squashes D45119, D42780 and D41932. It was reviewed in smaller patches by
asb, efriedma, apazos and mgrang.
Reviewers: asb, efriedma, apazos, llvm-commits, sabuasal
Reviewed By: sabuasal
Subscribers: mgorny, eraman, asb, rbar, johnrusso, simoncook, jordy.potman.lists, apazos, niosHD, kito-cheng, shiva0217, zzheng
Differential Revision: https://reviews.llvm.org/D45385
llvm-svn: 329455
Summary:
The 'strong' StackProtector heuristic takes into consideration call instructions.
Certain intrinsics, such as lifetime.start, can cause the
StackProtector to protect functions that do not need to be protected.
Specifically, a volatile variable, (not optimized away), but belonging to a stack
allocation will encourage a llvm.lifetime.start to be inserted during
compilation. Because that intrinsic is a 'call' the strong StackProtector
will see that the alloca'd variable is being passed to a call instruction, and
insert a stack protector. In this case the intrinsic isn't really lowered to a
call. This can cause unnecessary stack checking, at the cost of additional
(wasted) CPU cycles.
In the future we should rely on TargetTransformInfo::isLoweredToCall, but as of
now that routine considers all intrinsics as not being lowerable. That needs
to be corrected, and such a change is on my list of things to get moving on.
As a side note, the updated stack-protector-dbginfo.ll test always seems to
pass. I never see the dbg.declare/dbg.value reaching the
StackProtector::HasAddressTaken, but I don't see any code excluding dbg
intrinsic calls either, so I think it's the safest thing to do.
Reviewers: void, timshen
Reviewed By: timshen
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D45331
llvm-svn: 329450
Summary:
This patch removes InstRW overrides for basic arithmetic/logic instructions. To do this I've added the store address port to RMW. And used a WriteSequence to make the latency additive. It does not cover ADC/SBB because they have different latency.
Apparently we were inconsistent about whether the store has latency or not thus the test changes.
I've also left out Sandy Bridge because the load latency there is currently 4 cycles and should be 5.
Reviewers: RKSimon, andreadb
Reviewed By: andreadb
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D45351
llvm-svn: 329416
As mentioned on D44647, this patch increases the default memory latency to +5cy , which more closely matches what most custom cases are doing for reg-mem instructions.
I've bumped LoadLatency, ReadAfterLd and WriteLoad values to 5cy to be consistent.
As Sandy Bridge is currently our default generic model, this affects a lot of scheduling tests...
Differential Revision: https://reviews.llvm.org/D44654
llvm-svn: 329388
MFI.LocalFrameSize was not serialized.
It is usually set from LocalStackSlotAllocation, so if that pass doesn't
run it is impossible do deduce it from the stack objects. Until now, this
information was lost.
llvm-svn: 329382
VSX D-form load/store instructions of POWER9 require the offset be a multiple of 16 and a helper`isOffsetMultipleOf` is used to check this.
So far, the helper handles FrameIndex + offset case, but not handling FrameIndex without offset case. Due to this, we are missing opportunities to exploit D-form instructions when accessing an object or array allocated on stack.
For example, x-form store (stxvx) is used for int a[4] = {0}; instead of d-form store (stxv). For larger arrays, D-form instruction is not used when accessing the first 16-byte. Using D-form instructions reduces register pressure as well as instructions.
Differential Revision: https://reviews.llvm.org/D45079
llvm-svn: 329377
This is the 32-bit mode version of LEAVE64. It should be at least somewhat similar to LEAVE64.
The Sandy Bridge version was missing a load port use.
llvm-svn: 329347
Currently it is 6. If the "feature" was not used, report dummy
hidden argument. Otherwise it does not match the kernarg size
reported in the kernel header.
Differential Revision: https://reviews.llvm.org/D45129
llvm-svn: 329341
We were forcing the latency of these instructions to 5 cycles, but every other scheduler model had them as 1 cycle. I'm sure I didn't get everything, but this gets a big portion.
llvm-svn: 329339
This patch adds a way for users to create their own custom sections to
be added to wasm files. At the LLVM IR layer, they are defined through
the "wasm.custom_sections" named metadata. The expected use case for
this is bindings generators such as wasm-bindgen.
Patch by Dan Gohman
Differential Revision: https://reviews.llvm.org/D45297
llvm-svn: 329315
A noreturn nounwind function can be expected to never return in any way, and by
never returning it will also never have to restore any callee-saved registers
for its caller. This makes it possible to skip spills of those registers during
function entry, saving some stack space and time in the process. This is rather
useful for embedded targets with limited stack space.
Should fix PR9970.
Patch by myeisha (pmb).
llvm-svn: 329287
It's failing on the bots and I'm not sure why.
This reverts:
[X86] Synchronize the SchedRW on some EVEX instructions with their VEX equivalents.
[X86] Use WriteFShuffle256 for VEXTRACTF128 to be consistent with VEXTRACTI128 which uses WriteShuffle256.
[X86] Remove some InstRWs for plain store instructions on Sandy Bridge.
[X86] Auto-generate complete checks. NFC
llvm-svn: 329256
We were forcing the latency of these instructions to 5 cycles, but every other scheduler model had them as 1 cycle. I'm sure I didn't get everything, but this gets a big portion.
llvm-svn: 329252
The implementation of shadow call stack on aarch64 is quite different to
the implementation on x86_64. Instead of reserving a segment register for
the shadow call stack, we reserve the platform register, x18. Any function
that spills lr to sp also spills it to the shadow call stack, a pointer to
which is stored in x18.
Differential Revision: https://reviews.llvm.org/D45239
llvm-svn: 329236
Summary:
These new image intrinsics contain the texture type as part of
their name and have each component of the address/coordinate as
individual parameters.
This is a preparatory step for implementing the A16 feature, where
coordinates are passed as half-floats or -ints, but the Z compare
value and texel offsets are still full dwords, making it difficult
or impossible to distinguish between A16 on or off in the old-style
intrinsics.
Additionally, these intrinsics pass the 'texfailpolicy' and
'cachectrl' as i32 bit fields to reduce operand clutter and allow
for future extensibility.
v2:
- gather4 supports 2darray images
- fix a bug with 1D images on SI
Change-Id: I099f309e0a394082a5901ea196c3967afb867f04
Reviewers: arsenm, rampitec, b-sumner
Subscribers: kzhuravl, wdng, yaxunl, dstuttard, tpr, llvm-commits, t-tye
Differential Revision: https://reviews.llvm.org/D44939
llvm-svn: 329166
Summary:
When an i1-value is defined inside of a loop and used outside of it, we
cannot simply use the SGPR bitmask from the loop's last iteration.
There are also useful and correct cases of an i1-value being copied between
basic blocks, e.g. when a condition is computed outside of a loop and used
inside it. The concept of dominators is not sufficient to capture what is
going on, so I propose the notion of "lane-dominators".
Fixes a bug encountered in Nier: Automata.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=103743
Change-Id: If37b969ddc71d823ab3004aeafb9ea050e45bd9a
Reviewers: arsenm, rampitec
Subscribers: kzhuravl, wdng, mgorny, yaxunl, dstuttard, tpr, llvm-commits, t-tye
Differential Revision: https://reviews.llvm.org/D40547
llvm-svn: 329164
Recommitting rL321259. Previosuly this caused an issue with PPCBE but
I didn't receieve a reproducer and didn't have the time to follow up.
If the issue appears again, please provide a reproducer so I can fix
it.
Original commit message:
If the SRL node is only used by an AND, we may be able to set the
ExtVT to the width of the mask, making the AND redundant. To support
this, another check has been added in isLegalNarrowLoad which queries
whether the load is valid.
Differential Revision: https://reviews.llvm.org/D41350
llvm-svn: 329160
Summary:
The ShadowCallStack pass instruments functions marked with the
shadowcallstack attribute. The instrumented prolog saves the return
address to [gs:offset] where offset is stored and updated in [gs:0].
The instrumented epilog loads/updates the return address from [gs:0]
and checks that it matches the return address on the stack before
returning.
Reviewers: pcc, vitalybuka
Reviewed By: pcc
Subscribers: cryptoad, eugenis, craig.topper, mgorny, llvm-commits, kcc
Differential Revision: https://reviews.llvm.org/D44802
llvm-svn: 329139
This commit is similar to r329120, but uses the existing getUsesRedZone() function
in X86MachineFunctionInfo. This teaches the outliner to look at whether or not a
function *truly* uses a redzone instead of just the noredzone attribute on a
function.
Thus, after this commit, it's possible to outline from x86 without using
-mno-red-zone and still get outlining results.
This also adds a new test for the new redzone behaviour.
llvm-svn: 329134
Summary: There are no packed instructions for min3 or max3. So, performMinMaxCombine should not optimize vectors of f16 to min3/max3.
Author: FarhanaAleen
Reviewed By: arsenm
Subscribers: llvm-commits, AMDGPU
Differential Revision: https://reviews.llvm.org/D45219
llvm-svn: 329131
This patch adds a hasRedZone() function to AArch64MachineFunctionInfo. It
returns true if the function is known to use a redzone, false if it is known
to not use a redzone, and no value otherwise.
This removes the requirement to pass -mno-red-zone when outlining for AArch64.
https://reviews.llvm.org/D45189
llvm-svn: 329120
The linkage type on outlined functions was private before. This meant that if
you set a breakpoint in an outlined function, the debugger wouldn't be able to
give a sane name to the outlined function.
This commit changes the linkage type to internal and updates any tests that
relied on the prefixes on the names of outlined functions.
llvm-svn: 329116
We don't constant fold any of these, but we could...but if we
do, we must produce the right answer.
Unlike the IR fptosi instruction or its DAG node counterpart
ISD::FP_TO_SINT, these are not undef for an out-of-range input.
llvm-svn: 329100
fact use regular expression syntax to use regular expressions.
Should restore the bots. Sorry for the noise on this test.
Thanks to Philip for spotting the bug!
llvm-svn: 329057
This adds the basic test cases from all the EFLAGS bugs in more direct
forms. It also switches to generated check lines, and includes both
32-bit and 64-bit variations.
No functionality changing here, just setting things up to have a nice
clean asm diff in my EFLAGS patch.
llvm-svn: 329056
do explicit scrubbing of the offsets of stack spills and reloads.
You can always turn this off in order to test specific stack slot usage.
We were already hiding most of this, but the new logic hides it more
generically. Notably, we should effectively hide stack slot churn in
functions that have a frame pointer now, and should also hide it when
changing a function from stack pointer to frame pointer. That transition
already changes enough to be clearly noticed in the test case diff,
showing *every* spill and reload is really noisy without benefit. See
the test case I ran this on as a classic example.
llvm-svn: 329055
Commit 37962a331c77 ("bpf: Improve expanding logic in LowerSELECT_CC")
intended to improve code quality for certain jmp conditions. The
commit, however, has a couple of issues:
(1). In code, just swap is not enough, ConditionalCode CC
should also be swapped, otherwise incorrect code will
be generated.
(2). The ConditionalCode swap should be subject to
getHasJmpExt(). If getHasJmpExt() is False, certain
conditional codes will not be supported and swap
may generate incorrect code.
The original goal for this patch is to optimize jmp operations
which does not have JmpExt turned on. If JmpExt is on,
better code could be generated. For example, the test
select_ri.ll is introduced to demonstrate the optimization.
The same result can be achieved with -mcpu=v2 flag.
Signed-off-by: Yonghong Song <yhs@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
llvm-svn: 329043
Just adds basic block labels and tidies up where comments go in the test
case and then generates fresh CHECK lines with the script. This way, the
check lines are much easier to maintain. They were already close to this
but not quite there.
llvm-svn: 329040
If a load follows a store and reloads data that the store has written to memory, Intel microarchitectures can in many cases forward the data directly from the store to the load, This "store forwarding" saves cycles by enabling the load to directly obtain the data instead of accessing the data from cache or memory.
A "store forward block" occurs in cases that a store cannot be forwarded to the load. The most typical case of store forward block on Intel Core microarchiticutre that a small store cannot be forwarded to a large load.
The estimated penalty for a store forward block is ~13 cycles.
This pass tries to recognize and handle cases where "store forward block" is created by the compiler when lowering memcpy calls to a sequence
of a load and a store.
The pass currently only handles cases where memcpy is lowered to XMM/YMM registers, it tries to break the memcpy into smaller copies.
breaking the memcpy should be possible since there is no atomicity guarantee for loads and stores to XMM/YMM.
Differential revision: https://reviews.llvm.org/D41330
Change-Id: Ib48836ccdf6005989f7d4466fa2035b7b04415d9
llvm-svn: 328973
fptosi / fptoui round towards zero, and that's the same behavior as ISD::FTRUNC,
so replace a pair of casts with the equivalent node. We don't have to account for
special cases (NaN, INF) because out-of-range casts are undefined.
Differential Revision: https://reviews.llvm.org/D44909
llvm-svn: 328921
Summary:
It seems many CPUs don't implement this instruction as well as the other vector multiplies. Often using a multi uop flow. Silvermont in particular has a 7 uop flow with 11 cycle throughput. Sandy Bridge implements it as a single uop with 5 cycle latency and 1 cycle throughput. But Haswell and later use 2 uops with 10 cycle latency and 2 cycle throughput.
This patch adds a new X86SchedWritePair we can use to tag this instruction separately. I've provided correct information for Silvermont, Btver2, and Sandy Bridge. I've removed the InstRWs for SandyBridge. I've left Haswell/Broadwell/Skylake InstRWs in place because I wasn't sure how to account for the different load latency between 128 and 256 bits. I also left Znver1 InstRWs in place because the existing values don't match Agner's spreadsheet.
I also left a FIXME in the SandyBridge model because it being used for the "generic" model is too optimistic for the 256/512-bit versions since those are multiple uops on all known CPUs.
Reviewers: RKSimon, GGanesh, courbet
Reviewed By: RKSimon
Subscribers: gchatelet, gbedwell, andreadb, llvm-commits
Differential Revision: https://reviews.llvm.org/D44972
llvm-svn: 328914
Two memory instructions with a dependency only on the address register
between the two (the first one of them being post-incrememnt) can be
packetized together after the offset on the second was updated to the
incremement value. Make sure that the new offset is valid for the
instruction.
llvm-svn: 328897
The code has bugs dealing with -0.0.
Since D44550 introduced FABS pattern folding in InstCombine,
this patch removes the now-redundant code that causes
https://bugs.llvm.org/show_bug.cgi?id=36600.
Patch by Mikhail Dvoretckii!
Differential Revision: https://reviews.llvm.org/D44683
llvm-svn: 328872
Summary:
The phase attempts to transform operations that extract a portion of a value
into an SDWA src operand in cases where that value is used only once. It
was not prepared for this use to be the preserved portion of a value for
dst:UNUSED_PRESERVE, resulting in a crash or assert.
This change either rejects the illegal SDWA attempt, or in the case where
dst:WORD_1 and the src_sel would be WORD_0, removes the unneeded
extract instruction.
Reviewers: arsenm, #amdgpu
Reviewed By: arsenm, #amdgpu
Subscribers: arsenm, kzhuravl, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D44364
llvm-svn: 328856
MachineCopyPropagation::CopyPropagateBlock has a bunch of special
handling for COPY instructions. This handling assumes that COPY
instructions do not modify the source of the copy; this is wrong if
the COPY destination overlaps the source.
To fix the bug, check explicitly for this situation, and fall back to
the generic instruction handling.
This bug can't happen for most register classes because they don't
have this sort of overlap, but there are a few register classes
where this is possible. The testcase uses the AArch64 QQQQ register
class.
Differential Revision: https://reviews.llvm.org/D44911
llvm-svn: 328851
While the stack access instructions don't care about
alignment > 4, some transformations on the pointer calculation
do make assumptions based on knowing the low bits of a pointer
are 0. If a stack object ends up being accessed through its
absolute address (relative to the kernel scratch wave offset),
the addressing expression may depend on the stack frame being
properly aligned. This was breaking in a testcase due to the
add->or combine.
I think some of the SP/FP handling logic is still backwards,
and overly simplistic to support all of the stack features.
Code which tries to modify the SP with inline asm for example
or variable sized objects will probably require redoing this.
llvm-svn: 328831
These instructions have the memory operand before the register operand. So we need to put ReadDefault for all the load ops first. Then the ReadAfterLd
Differential Revision: https://reviews.llvm.org/D44838
llvm-svn: 328823
DWARF v5 specifies that the root file (also given in the DW_AT_name
attribute of the compilation unit DIE) should be emitted explicitly to
the line table's list of files. This makes the line table more
independent of the .debug_info section.
We emit the new syntax only for DWARF v5 and later.
Fixes the bug found by asan. Also XFAIL the new test for Darwin, which
is stuck on DWARF v2, and fix up other tests so they stop failing on
Windows. Last but not least, don't break "clang -g" of an assembler
file that has .file directives in it.
Differential Revision: https://reviews.llvm.org/D44054
llvm-svn: 328805
Summary: Mark CFG is preserved since this pass do not make any change in CFG.
Reviewers: sebpop, mzolotukhin, mcrosier
Reviewed By: mzolotukhin
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D44845
llvm-svn: 328727
These instructions have been around for a long time, but we
haven't supported intrinsics for them. The "new" versions use
the CSx register for the start of the buffer instead of the K
field in the Mx register.
We need to use pseudo instructions for these instructions until
after register allocation. The problem is that these instructions
allocate a M0/CS0 or M1/CS1 pair. But, we can't generate code for
the CSx set-up until after register allocation when the Mx
register has been fixed for the instruction.
There is a related clang patch.
Patch by Brendon Cahoon.
llvm-svn: 328724
This commit simplifies the call outlining logic by removing references to the
Function associated with the callee. To do this, it requires that valid
callee save info is available to the outliner.
llvm-svn: 328719
This reverts commit r328676.
Commit r328676 broke the -no-integrated-as flag necessary to build Linux kernel with Clang:
$ cat t.c
void foo() {}
$ clang -no-integrated-as -c t.c -g
/tmp/t-dcdec5.s: Assembler messages:
/tmp/t-dcdec5.s:8: Error: file number less than one
clang-7.0: error: assembler command failed with exit code 1 (use -v to see invocation)
llvm-svn: 328699
This reverts commit 0daf86291d3aa04d3cc280cd0ef24abdb0174981.
It was causing an assert in test/CodeGen/AMDGPU/amdpal.ll only on a
release-with-asserts build. I will resubmit the change when I have fixed
that.
Change-Id: If270594eba27a7dc4076bdeab3fa8e6bfda3288a
llvm-svn: 328695
Follow up patch of r328313 to support the UseVMOVSR constraint. Removed
some unneeded instructions from the test and removed some stray
comments.
Differential Revision: https://reviews.llvm.org/D44941
llvm-svn: 328691
Summary:
RegisterCoalescer::removePartialRedundancy tries to hoist B = A from
BB0/BB2 to BB1:
BB1:
...
BB0/BB2: ----
B = A; |
... |
A = B; |
|-------
|
It does so if a number of conditions are fulfilled. However, it failed
to check if B was used by any of the terminators in BB1. Since we must
insert B = A before the terminators (since it's not a terminator itself),
this means that we could erroneously insert a new definition of B before a
use of it.
Reviewers: wmi, qcolombet
Reviewed By: wmi
Subscribers: MatzeB, llvm-commits, sdardis
Differential Revision: https://reviews.llvm.org/D44918
llvm-svn: 328689
DWARF v5 specifies that the root file (also given in the DW_AT_name
attribute of the compilation unit DIE) should be emitted explicitly to
the line table's list of files. This makes the line table more
independent of the .debug_info section.
Fixes the bug found by asan. Also XFAIL the new test for Darwin, which
is stuck on DWARF v2, and fix up other tests so they stop failing on
Windows. Last but not least, don't break "clang -g" of an assembler
file that has .file directives in it.
Differential Revision: https://reviews.llvm.org/D44054
llvm-svn: 328676
If an ADRP appears with, say, a CPI operand, we shouldn't outline it.
This moves the check for unsafe operands so that it occurs before the special-case
for ADRPs. Also add a test for outlining ADRPs.
llvm-svn: 328674
Summary:
For OS type AMDPAL, the scratch descriptor is loaded from offset 0 of
the GIT, whose 32 bit pointer is in s0 (s8 for gfx9 merged shaders).
This commit fixes that to use offset 0x10 instead of offset 0 for a
compute shader, per the PAL ABI spec.
Reviewers: kzhuravl, nhaehnle, timcorringham
Subscribers: kzhuravl, wdng, yaxunl, t-tye, llvm-commits, dstuttard, nhaehnle, arsenm
Differential Revision: https://reviews.llvm.org/D44468
Change-Id: I93dffa647758e37f613bb5e0dfca840d82e6d26f
llvm-svn: 328673
If a given split type unit does not have source locations, don't have
it refer to the split line table.
If no split type unit refers to the split line table, don't emit the
line table at all.
This will save a little space on rare occasions, but also refactors
things a bit to improve which class is responsible for what.
Responding to review comments on r326395.
Differential Revision: https://reviews.llvm.org/D44220
llvm-svn: 328670
Summary:
Rev 327580 "[CodeGen] Use MIR syntax for MachineMemOperand printing"
broke -print-machineinstrs for us on AMDGPU, because we have custom
pseudo source values, and MIR serialization does not implement that.
This commit at least restores the functionality of -print-machineinstrs,
even if it does not properly implement the missing MIR serialization
functionality.
Differential Revision: https://reviews.llvm.org/D44871
Change-Id: I44961c0b90bf6d48c01484ed7a4e466fd300db66
llvm-svn: 328668
Currently MOVMSK instructions use the WriteVecLogic class, which is a very poor choice given that MOVMSK involves a SSE->GPR transfer.
Differential Revision: https://reviews.llvm.org/D44924
llvm-svn: 328664
Before this was not done if the function had no calls in it. This
is still a possible issue with any callable function, regardless
of calls present.
llvm-svn: 328659
The combine on a select of a load only triggers for
addrspace 0, and discards the MachinePointerInfo. The
conservative default needs to be used for this.
llvm-svn: 328652
In a function, s5 is used as the frame base SGPR. If a function
is calling another function, during the call sequence
it is copied to a preserved SGPR and restored.
Before it was possible for the scheduler to move stack operations
before the restore of s5, since there's nothing to associate
a frame index access with the restore.
Add an implicit use of s5 to the adjcallstack pseudo which ends
the call sequence to preven this from happening. I'm not 100%
satisfied with this solution, but I'm not sure what else would be
better.
llvm-svn: 328650
%tmp = bitcast i32* %arg to i8*
%tmp1 = getelementptr inbounds i8, i8* %tmp, i32 0
- %tmp2 = load i8, i8* %tmp, align 1
+ %tmp2 = load i8, i8* %tmp1, align 1
This doesn't change the semantics of the tests but makes use of %tmp1 which was originally intended.
llvm-svn: 328642
On Hexagon "x = y" is a syntax used in most instructions, and is not
treated as a directive.
Differential Revision: https://reviews.llvm.org/D44256
llvm-svn: 328635
Currently CRC32 instructions use the WriteFAdd class, this patch splits them off into their own, at the moment it is still mostly just a duplicate of WriteFAdd but it can now be tweaked on a target by target basis.
Differential Revision: https://reviews.llvm.org/D44647
llvm-svn: 328582
Summary:
Re-lands r328386 and r328443, reverting r328482.
Incorporates fixes from @mstorsjo in D44876 (thanks!) so that small
parameters in i8 and i16 do not end up in the SysV register parameters
(EDI, ESI, etc).
I added tests for how we receive small parameters, since that is the
important part. It's always safe to store more bytes than will be read,
but the assumptions you make when loading them are what really matter.
I also tested this by self-hosting clang and it passed tests on win64.
Reviewers: mstorsjo, hans
Subscribers: hiraditya, mstorsjo, llvm-commits
Differential Revision: https://reviews.llvm.org/D44900
llvm-svn: 328570
Give the bit count instructions their own scheduler classes instead of forcing them into existing classes.
These were mostly overridden anyway, but I had to add in costs from Agner for silvermont and znver1 and the Fam16h SoG for btver2 (Jaguar).
Differential Revision: https://reviews.llvm.org/D44879
llvm-svn: 328566
Legalize and emit code for quad-precision floating point operation xscvdpqp
and add option to guard the quad precision operation support.
Differential Revision: https://reviews.llvm.org/D44746
llvm-svn: 328558
A new function getOpcodeForSpill should now be the only place to get
the opcode for a given spilled register.
Differential Revision: https://reviews.llvm.org/D43086
llvm-svn: 328556
The pipeliner is not adding a dependence edge for a loop carried
dependence, and ends up scheduling a load from iteration n prior
to an aliased store in iteration n-1.
The code that adds the loop carried dependences in the pipeliner
doesn't check if the memory objects for loads and stores are
"identified" (i.e., distinct) objects. If they are not, then the
code that adds the dependences needs to be conservative. The
objects can be used to check dependences only when they are
distinct objects.
The code that checks for loop carried dependences has been updated
to classify loads and stores that are not identified as "unknown"
values. A store with an "unknown" value can potentially create
a loop carried dependence with any pending load.
Patch by Brendon Cahoon.
llvm-svn: 328547
The patch contains severals changes needed to pipeline an example
that was transformed so that a Phi with a subreg is converted to
copies.
The pipeliner wasn't working for a couple of reasons.
- The RecMII was 3 instead of 2 due to the extra copies.
- Copy instructions contained a latency of 1.
- The node order algorithm was not choosing the best "bottom"
node, which caused an instruction to be scheduled that had a
predecessor and successor already scheduled.
- Updated the Hexagon Machine Scheduler to check if the node is
latency bound when adding the cost for a 0-latency dependence.
The RecMII was 3 because the computation looks at the number of
nodes in the recurrence. The extra copy is an extra node but
it shouldn't increase the latency. The new RecMII computation
looks at the latency of the instructions in the recurrence. We
changed the latency of the dependence of a copy to 0. The latency
computation for the copy also checks the use of the copy (similar
to a reg_sequence).
The node order algorithm was not choosing the last instruction
in the recurrence for a bottom up traversal. This was when the
last instruction is a copy. A check was added when choosing the
instruction to check for NodeNum if the maxASAP is the same. This
means that the scheduler will not end up with another node in
the recurrence that has both a predecessor and successor already
scheduled.
The cost computation in Hexagon Machine Scheduler adds cost when
an instruction can be packetized with a zero-latency instruction.
We should only do this if the schedule is latency bound.
Patch by Brendon Cahoon.
llvm-svn: 328542
The pipeliner is asserting because the serialization step that
occurs at the end is deleting an instruction. The assert
occurs later on because there is a use without a definition.
The problem occurs when an instruction defines a value used
by a REQ_SEQUENCE and that value is used by a COPY instruction.
The latencies between these instructions are zero, so they are
put in to the same packet. The serialization code is unable to
handle this correctly, and ends up putting the REG_SEQUENCE
before its definition.
There is special code in the serialization step that attempts
to handle zero-cost instructions (phis, copy, reg_sequence)
differently than regular instructions. Unfortunately, this means
the order does not come out correct.
This patch simplifies the code by changing the seperate steps for
handling zero-cost and regular instructions. Only phis are
handled separate now, since they should occurs first. Then, this
patch adds checks to make use the MoveUse is set to the smallest
value if there are multiple uses in a cycle.
Patch by Brendon Cahoon.
llvm-svn: 328540
The code in orderDepdences that looks at the order dependences between
instructions was processing all the successor and predecessor order
dependences. However, we really only want to check for an order dependence
for instructions scheduled in the same cycle.
Also, fixed how the pipeliner handles output dependences. An output
dependence is also a potential loop carried dependence. The pipeliner
didn't handle this case properly so an invalid schedule could be created
that allowed an output dependence to be scheduled in the next iteration
at the same cycle.
Patch by Brendon Cahoon.
llvm-svn: 328516
When the definition of a phi is used by a phi in the next iteration,
the pipeliner was assuming that the definition is processed first.
Because of the assumption, an incorrect phi name was used. This patch
has a check to see if the phi definition has been processed already.
Patch by Brendon Cahoon.
llvm-svn: 328510
The pipeliner needs to be conservative when updating the memoperands
of instructions in the epilog. Previously, the pipeliner was changing
the offset of the memoperand based upon the scheduling stage. However,
that is incorrect when control flow branches around the kernel code.
The bug enabled a load and store to the same stack offset to be swapped.
This patch fixes the bug by updating the size of the memoperands to be
UINT_MAX. This conservative value means that dependences will be created
between other loads and stores.
Patch by Brendon Cahoon.
llvm-svn: 328508
This broke Chromium (see crbug.com/825748). It looks like mstorsjo's follow-up
patch at D44876 fixes this, but let's revert back to green for now until that's
ready to land.
(Also reverts r328443.)
> Both GCC and MSVC only look at the low byte of a boolean when it is
> passed.
llvm-svn: 328482
Summary:
LLVM defaults to the newer .init_array/.fini_array scheme for static
constructors rather than the less desirable .ctors/.dtors (the UseCtors
flag defaults to false). This wasn't being respected in the RISC-V
backend because it fails to call TargetLoweringObjectFileELF::InitializeELF with the the appropriate
flag for UseInitArray.
This patch fixes this by implementing RISCVELFTargetObjectFile and overriding its Initialize method to call
InitializeELF(TM.Options.UseInitArray).
Reviewers: asb, apazos
Reviewed By: asb
Subscribers: mgorny, rbar, johnrusso, simoncook, jordy.potman.lists, sabuasal, niosHD, kito-cheng, shiva0217, llvm-commits
Differential Revision: https://reviews.llvm.org/D44750
llvm-svn: 328433
These nodes only use the lower 32 bits of their inputs so we can use SimplifyDemandedBits to simplify them.
Differential Revision: https://reviews.llvm.org/D44375
llvm-svn: 328405
Add two additional implicit arguments for OpenCL for the AMDGPU target using the AMDHSA runtime to support device enqueue.
Differential Revision: https://reviews.llvm.org/D44697
llvm-svn: 328351
- Remove use of the opencl and amdopencl environment member of the target triple for the AMDGPU target.
- Use function attribute to communicate to the AMDGPU backend to add implicit arguments for OpenCL kernels for the AMDHSA OS.
Differential Revision: https://reviews.llvm.org/D43736
llvm-svn: 328349
HexagonGenMux would collapse pairs of predicated transfers if it assumed
that the predicated .new forms cannot be created. Turns out that generating
mux is preferable in almost all cases.
Introduce an option -hexagon-gen-mux-threshold that controls the minimum
distance between the instruction defining the predicate and the later of
the two transfers. If the distance is closer than the threshold, mux will
not be generated. Set the threshold to 0 by default.
llvm-svn: 328346
This patch fixes PR36658, "Constant pool entry out of range!" in Thumb1 mode.
In ARMConstantIslands::optimizeThumb2JumpTables() in Thumb1 mode,
adjustBBOffsetsAfter() is not calculating postOffset correctly by
properly accounting for the padding that is required for the constant pool
that immediately follows the jump table branch instruction.
Reviewers: t.p.northover, eli.friedman
Reviewed By: t.p.northover
Subscribers: chrib, tstellar, javed.absar, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D44709
llvm-svn: 328341
This patch adds functions to allow MachineLICM to hoist invariant stores.
Currently, MachineLICM does not hoist any store instructions, however
when storing the same value to a constant spot on the stack, the store
instruction should be considered invariant and be hoisted. The function
isInvariantStore iterates each operand of the store instruction and checks
that each register operand satisfies isCallerPreservedPhysReg. The store
may be fed by a copy, which is hoisted by isCopyFeedingInvariantStore.
This patch also adds the PowerPC changes needed to consider the stack
register as caller preserved.
Differential Revision: https://reviews.llvm.org/D40196
llvm-svn: 328326
Loads and stores can only shift the offset register by the size of the value
being loaded, but currently the DAGCombiner will reduce the width of the load
if it's followed by a trunc making it impossible to later combine the shift.
Solve this by implementing shouldReduceLoadWidth for the AArch64 backend and
make it prevent the width reduction if this is what would happen, though do
allow it if reducing the load width will let us eliminate a later sign or zero
extend.
Differential Revision: https://reviews.llvm.org/D44794
llvm-svn: 328321
Steven Wu