If the mask input to getV4X86ShuffleImm8 only refers to a single source element (+ undefs) then canonicalize to a full broadcast.
getV4X86ShuffleImm8 defaults to inline values for undefs, which can be useful for shuffle widening/narrowing but does leave SimplifyDemanded* calls thinking the shuffle depends on unnecessary elements.
I'm still investigating what we should do more generally to avoid these undemanded elements, but broadcast cases was a simpler win.
We already had CMPXCH8B feature on this CPU for the frontend so
this doesn't have much effect.
The FeatureSlowUAMem16 only matters if someone compiles with
-march=lakemont -msse which doesn't make sense, but is consistent
with all our pre-sse4.2 CPUs. Maybe the feature flag should be
FeatureFastUAMem16 and set on the newer CPUs instead.
As briefly discussed in IRC with @craig.topper,
the pass is disabled basically since it's original introduction (nov 2018)
due to known correctness issues (miscompilations),
and there hasn't been much work done to fix that.
While i won't promise that i will "fix" the pass,
i have looked at it previously, and i'm sure i won't try to fix it
if that requires actually fixing this existing code.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D84775
By repeating the Disp.isImm() check in a couple spots we can
make the normal case for immediate and for expression the same.
And then always rely on the ForceDisp32 flag to remove a later
non-zero immediate check.
This should make {disp32} pseudo prefix handling
slightly easier as we need the normal disp32 handler to handle a
immediate of 0.
We currently handle EVEX and non-EVEX separately in two places. By sinking the EVEX
check into the existing helper for CDisp8 we can simplify these two places.
Differential Revision: https://reviews.llvm.org/D84730
An initial backend patch towards fixing the various poor HADD combines (PR34724, PR41813, PR45747 etc.).
This extends isHorizontalBinOp to check if we have per-element horizontal ops (odd+even element pairs), but not in the expected serial order - in which case we build a "post shuffle mask" that we can apply to the HOP result, assuming we have fast-hops/optsize etc.
The next step will be to extend the SHUFFLE(HOP(X,Y)) combines as suggested on PR41813 - accepting more post-shuffle masks even on slow-hop targets if we can fold it into another shuffle.
Differential Revision: https://reviews.llvm.org/D83789
XBEGIN causes several based blocks to be inserted. If flags are live across it we need to make eflags live in the new basic blocks to avoid machine verifier errors.
Fixes PR46827
Reviewed By: ivanbaev
Differential Revision: https://reviews.llvm.org/D84479
By default we pick a 1 byte displacement and let relaxation enlarge it if necessary. The GNU assembler supports a pseudo prefix to basically pre-relax the instruction the larger size.
I plan to add {disp8} and {disp32} support for memory operands in another patch which is why I've included the parsing code and enum for {disp8} pseudo prefix as well.
Reviewed By: echristo
Differential Revision: https://reviews.llvm.org/D84709
In 16-bit mode we can encode a 32-bit address using 0x67 prefix.
We were failing to do this when the index register was a 32-bit
register, the base register was not present, and the displacement
fit in 16-bits.
Fixes PR46866.
Previously we just matched the logic ops and replaced with an
X86ISD::VPTERNLOG node that we would send through the normal
pattern match. But that approach couldn't handle a bitcast
between the logic ops. Extending that approach would require us
to peek through the bitcasts and emit new bitcasts to match
the types. Those new bitcasts would then have to be properly
topologically sorted.
This patch instead switches to directly emitting the
MachineSDNode and skips the normal tablegen pattern matching.
We do have to handle load folding and broadcast load folding
ourselves now. Which also means commuting the immediate control.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D83630
These cost methods don't make much sense in X86Subtarget. Make
them methods in X86's TTI and move the feature checks from the
X86Subtarget constructor into these methods.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D84594
If we lower a v2i64 shuffle to PSHUFD, we currently clamp undef elements to 0, (elements 0,1 of the v4i32) which can result in the shuffle referencing more elements of the source vector than expected, affecting later shuffle combines and KnownBits/SimplifyDemanded calls.
By ensuring we widen the undef mask element we allow getV4X86ShuffleImm8 to use inline elements as the default, which are more likely to fold.
The switch in emitNop uses 64-bit registers for nops exceeding
2 bytes. This isn't valid outside 64-bit mode. We could fix this
easily enough, but there are no users that ask for more than 2
bytes outside 64-bit mode.
Inlining the method to make the coupling between the two methods
more explicit.
If we don't care about an entire LHS/RHS of the PACK op, then can just treat it the same as undef (we don't care if it saturates) and is safe to treat as a shuffle.
This can happen if we attempt to decode as a faux shuffle before SimplifyDemandedVectorElts has been called on the PACK which should replace the source with UNDEF entirely.
ParseX86Triple already checks for 64-bit mode and produces a
static string. We can just add +sse2 to the end of that static
string. This avoids a potential reallocation when appending it
to the std::string at runtime.
This is a slight change to the behavior of tools that only use
MC layer which weren't implicitly enabling sse2 before, but will
now. I don't think we check for sse2 explicitly in any MC layer
components so this shouldn't matter in practice. And if it did
matter the new behavior is more correct.
Remove mode flags from constructor and remove calls to
ToggleFeature for the mode bits.
By adding them to the feature string we handle initializing the
mode member variables in X86Subtarget and the feature bits in
MCSubtargetInfo in one shot.
Feature64Bit is only used by a check in the X86Subtarget
constructor to ensure that the CPU selected supports 64-bit mode
when the triple is for 64-bit mode.
'generic' is the default CPU in llc and so needs to be able to
pass this check. Previously we did this by detecting the name and
adding the feature to the feature string. But there doesn't seem
to be any reason we can't just add the feature to the CPU directly.
We deprecated mpx feature in 10.0. I left this feature flag
in case someone still had IR files containing the feature
in a target-feature attribute. At the time I think I thought it
would fail the test if the feature couldn't be found. Further
review suggests that at worst it prints a message to
stderr about ignoring the feature.
SAHF/LAHF instructions are always available in 32-bit mode. Early
64-bit capable CPUs made the undefined opcodes in 64-bit mode. This
was changed on later CPUs.
We have a feature flag to control our usage of these instructions.
This feature flag is hooked up to a clang command line option
-msahf/-mno-sahf specifically to give control of the 64-bit mode
behavior.
In the backend X86Subtarget constructor we were explicitly forcing
+sahf into the feature flag string if we were not compiling for
64-bit mode. This was intended to make the predicates always allow
the instructions outside of 64-bit mode. Unfortunately, the way
it was placed into the string allowed -mno-sahf from clang to disable
SAHF instructions in 32-bit mode. This causes an assertion to fire
if you compile a floating point comparison with something like
"-march=pentium -mno-sahf" as our floating point comparison
handling on CPUs that don't support FCOMI/FUCOMI instructions
requires SAHF.
To fix this, this commit restricts the feature flag to only apply to
64-bit mode by ignoring the flag outside 64-bit mode in
X86Subtarget::hasLAHFSAHF(). This way we don't need to mess with
the feature string at all.
The input to these functions is a StringRef. We then convert it
to a std::string. Then maybe replace with "generic". I think we
can just overwrite the incoming StringRef with "generic" if needed
and then pass it along without creating any std::string.
There's no reason to involve the hassle of a virtual method targets
have to override for a simple boolean.
Not sure exactly what's going on with Mips, but it seems to define its
own totally separate handler classes.
For a long time, the InstCombine pass handled target specific
intrinsics. Having target specific code in general passes was noted as
an area for improvement for a long time.
D81728 moves most target specific code out of the InstCombine pass.
Applying the target specific combinations in an extra pass would
probably result in inferior optimizations compared to the current
fixed-point iteration, therefore the InstCombine pass resorts to newly
introduced functions in the TargetTransformInfo when it encounters
unknown intrinsics.
The patch should not have any effect on generated code (under the
assumption that code never uses intrinsics from a foreign target).
This introduces three new functions:
TargetTransformInfo::instCombineIntrinsic
TargetTransformInfo::simplifyDemandedUseBitsIntrinsic
TargetTransformInfo::simplifyDemandedVectorEltsIntrinsic
A few target specific parts are left in the InstCombine folder, where
it makes sense to share code. The largest left-over part in
InstCombineCalls.cpp is the code shared between arm and aarch64.
This allows to move about 3000 lines out from InstCombine to the targets.
Differential Revision: https://reviews.llvm.org/D81728
getTargetShuffleMask is used by the various "SimplifyDemanded" folds so we can't assume that the bypassed extract_subvector can be safely simplified - getFauxShuffleMask performs a more general decode that allows us to more safely catch many of these cases so the impact is minimal.
fma reassoc A, B, C --> fadd (fmul A, B), C (when target has no FMA hardware)
C/C++ code may use explicit fma() calls (which become LLVM fma
intrinsics in IR) but then gets compiled with -ffast-math or similar.
For targets that do not have FMA hardware, we don't want to go out to
the math library for a precise but slow FMA result.
I tried this as a generic DAGCombine, but it caused infinite looping
on more than 1 other target, so there's likely some over-reaching fma
formation happening.
There's also a potential intersection of strict FP with fast-math here.
Deferring to current behavior for that case (assuming that strict-ness
overrides fast-ness).
Differential Revision: https://reviews.llvm.org/D83981
I meant to do this in D83913, but missed it while updating the
feature list.
Interestingly I think this is disabling the postRA scheduler. But
it does match our default 64-bit behavior.
Reviewed By: echristo
Differential Revision: https://reviews.llvm.org/D83996
We use a SmallString<512> and attempted to reserve enough space
for CPU plus Features, but that doesn't account for all the things
that get added to the string.
Reorder the string so the shortest things go first which shouldn't
exceed the small size. Finally add the feature string at the end
which might be long. This should ensure at most one heap allocation
without needing to use reserve.
I don't know if this matters much in practice, but I was looking
into something else that will require more code here and noticed
the odd reserve call.
There was a lot of duplicate code here for checking the VT and
subtarget. Moving it into a helper avoids that.
It also fixes a bug that combineAdd reused Op0/Op1 after a call
to isHorizontalBinOp may have changed it. The new helper function
has its own local version of Op0/Op1 that aren't shared by other
code.
Fixes PR46455.
Reviewed By: spatel, bkramer
Differential Revision: https://reviews.llvm.org/D83971
Alternative to D83897. I believe the big change here is that I removed slow unaligned memory 16
Down side that it may adversely effect tuning if someone explicitly targets -march=pentium4 and expects pentium4 tuned code. Of course pentium4 is so old our default behavior with the previous settings may not have been the best either.
Reviewed By: echristo, RKSimon
Differential Revision: https://reviews.llvm.org/D83913
Previously we only accepted a 32-bit source with a 64-bit dest.
Accepting 64-bit as well is more consistent with gas behavior. I
think maybe we should accept 16 bit register as well, but I'm not
sure.
This patch handles CFI with basic block sections, which unlike DebugInfo does
not support ranges. The DWARF standard explicitly requires emitting separate
CFI Frame Descriptor Entries for each contiguous fragment of a function. Thus,
the CFI information for all callee-saved registers (possibly including the
frame pointer, if necessary) have to be emitted along with redefining the
Call Frame Address (CFA), viz. where the current frame starts.
CFI directives are emitted in FDE’s in the object file with a low_pc, high_pc
specification. So, a single FDE must point to a contiguous code region unlike
debug info which has the support for ranges. This is what complicates CFI for
basic block sections.
Now, what happens when we start placing individual basic blocks in unique
sections:
* Basic block sections allow the linker to randomly reorder basic blocks in the
address space such that a given basic block can become non-contiguous with the
original function.
* The different basic block sections can no longer share the cfi_startproc and
cfi_endproc directives. So, each basic block section should emit this
independently.
* Each (cfi_startproc, cfi_endproc) directive will result in a new FDE that
caters to that basic block section.
* Now, this basic block section needs to duplicate the information from the
entry block to compute the CFA as it is an independent entity. It cannot refer
to the FDE of the original function and hence must duplicate all the stuff that
is needed to compute the CFA on its own.
* We are working on a de-duplication patch that can share common information in
FDEs in a CIE (Common Information Entry) and we will present this as a follow up
patch. This can significantly reduce the duplication overhead and is
particularly useful when several basic block sections are created.
* The CFI directives are emitted similarly for registers that are pushed onto
the stack, like callee saved registers in the prologue. There are cfi
directives that emit how to retrieve the value of the register at that point
when the push happened. This has to be duplicated too in a basic block that is
floated as a separate section.
Differential Revision: https://reviews.llvm.org/D79978
This fixes warnings raised by Clang's new -Wsuggest-override, in preparation for enabling that warning in the LLVM build. This patch also removes the virtual keyword where redundant, but only in places where doing so improves consistency within a given file. It also removes a couple unnecessary virtual destructor declarations in derived classes where the destructor inherited from the base class is already virtual.
Differential Revision: https://reviews.llvm.org/D83709
Summary:
Add support for MASM STRUCT casting field accessors: (<TYPE> PTR <value>).<field>
Since these are operands, we add them to X86AsmParser. If/when we extend MASM support to other architectures (e.g., ARM), we will need similar changes there as well.
Reviewed By: thakis
Differential Revision: https://reviews.llvm.org/D83346
Bit 7 of the index controls zeroing, the other bits are ignored when bit 7 is set. Shuffle lowering was using 128 and shuffle combining was using 255. Seems like we should be consistent.
This patch changes shuffle combining to use 128 to match lowering.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D83587
peekThroughOneUseBitcasts checks the use count of the operand of the bitcast. Not the bitcast itself. So I think that means we need to do any outside haseOneUse checks before calling the function not after.
I was working on another patch where I misused the function and did a very quick audit to see if I there were other similar mistakes.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D83598
Truncations lowered as shuffles of multiple (concatenated) vectors often leave us with lane-crossing shuffles that feed a PACKSS/PACKUS, if both shuffles are fed from the same 2 vector sources, then we can PACK the sources directly and shuffle the result instead.
This is currently limited to whole i128 lanes in a 256-bit vector, but we can extend this if the need arises (but I'm not seeing many examples in real world code).
If we don't immediately lower the vector shift, the splat
constant vector we created may get turned into a constant pool
load before we get around to lowering the shift. This makes it
a lot more difficult to create a shift by constant. Sometimes we
fail to see through the constant pool at all and end up trying
to lower as if it was a variable shift. This requires custom
handling and may create an unsupported vselect on pre-sse-4.1
targets. Since we're after LegalizeVectorOps we are unable to
legalize the unsupported vselect as that code is in LegalizeVectorOps
rather than LegalizeDAG.
So calling LowerShift immediately ensures that we get see the
splat constant.
Fixes PR46527.
Differential Revision: https://reviews.llvm.org/D83455
Technically a VSELECT expects a vector of all 1s or 0s elements
for its condition. But we aren't guaranteeing that the sign bit
and the non sign bits match in these locations. So we should use
BLENDV which is more relaxed.
Differential Revision: https://reviews.llvm.org/D83447
If we're extracting a subvector from a shuffle that is shuffling entire subvectors we can peek through and extract the subvector from the shuffle source instead.
This helps remove some cases where concat_vectors(extract_subvector(),extract_subvector()) legalizations has resulted in BLEND/VPERM2F128 shuffles of the subvectors.
Summary:
Almost all uses of these iterators, including implicit ones, really
only need the const variant (as it should be). The only exception is
in NewGVN, which changes the order of dominator tree child nodes.
Change-Id: I4b5bd71e32d71b0c67b03d4927d93fe9413726d4
Reviewers: arsenm, RKSimon, mehdi_amini, courbet, rriddle, aartbik
Subscribers: wdng, Prazek, hiraditya, kuhar, rogfer01, rriddle, jpienaar, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, aartbik, liufengdb, stephenneuendorffer, Joonsoo, grosul1, vkmr, Kayjukh, jurahul, msifontes, cfe-commits, llvm-commits
Tags: #clang, #mlir, #llvm
Differential Revision: https://reviews.llvm.org/D83087
vselect ((X & Pow2C) == 0), LHS, RHS --> vselect ((shl X, C') < 0), RHS, LHS
Follow-up to D83073 - the non-splat mask cases where we actually see an
improvement are quite limited from what I can tell. AVX1 needs multiply
and blend capabilities and AVX2 needs vector shift and blend capabilities.
The intersection of those 2 constraints is only vectors with 32-bit or
64-bit elements.
XOP is/was better.
Differential Revision: https://reviews.llvm.org/D83181
Summary:
Add support for user-defined types to MasmParser, including initialization and field access.
Known issues:
- Omitted entry initializers (e.g., <,0>) do not work consistently for nested structs/arrays.
- Size checking/inference for values with known types is not yet implemented.
- Some ml64.exe syntaxes for accessing STRUCT fields are not recognized.
- `[<register>.<struct name>].<field>`
- `[<register>[<struct name>.<field>]]`
- `(<struct name> PTR [<register>]).<field>`
- `[<variable>.<struct name>].<field>`
- `(<struct name> PTR <variable>).<field>`
Reviewed By: thakis
Differential Revision: https://reviews.llvm.org/D75306
This patch creates a clang flag to enable SESES. This flag also ensures that
lvi-cfi is on when using seses via clang.
SESES should use lvi-cfi to mitigate returns and indirect branches.
The flag to enable the SESES functionality only without lvi-cfi is now
-x86-seses-enable-without-lvi-cfi to warn users part of the mitigation is not
enabled if they use this flag. This is useful in case folks want to see the
cost of SESES separate from the LVI-CFI.
Reviewed By: sconstab
Differential Revision: https://reviews.llvm.org/D79910
We were checking the VBROADCAST_LOAD element size against the extraction destination size instead of the extracted vector element size - PEXTRW/PEXTB have implicit zext'ing so have i32 destination sizes for v8i16/v16i8 vectors, resulting in us extracting from the wrong part of a load.
This patch bails from the fold if the vector element sizes don't match, and we now use the target constant extraction code later on like the pre-AVX2 targets, fixing the test case.
Found by internal fuzzing tests.
Use SESES as the fallback at O0 where the optimized LVI pass isn't desired due
to its effect on build times at O0.
I updated the LVI tests since this changes the code gen for the tests touched in the parent revision.
This is a follow up to the comments I made here: https://reviews.llvm.org/D80964
Hopefully we can continue the discussion here.
Also updated SESES to handle LFENCE instructions properly instead of adding
redundant LFENCEs. In particular, 1) no longer add LFENCE if the current
instruction being processed is an LFENCE and 2) no longer add LFENCE if the
instruction right before the instruction being processed is an LFENCE
Reviewed By: sconstab
Differential Revision: https://reviews.llvm.org/D82037
In the test based on PR46586:
https://bugs.llvm.org/show_bug.cgi?id=46586
...we are inserting 16-bits into the high element of the vector, shuffling it
to element 0, and extracting 32-bits. But xmm1 was never initialized, so the
top 16-bits of the extract are undef without this patch.
(It seems like we could do better than this by recognizing that we only demand
a subsection of the build vector, but I want to make sure we fix the
miscompile 1st.)
This path is only used for pre-SSE4.1, and simpler patterns get squashed
somewhere along the way, so the test still includes a 'urem' as it did in the
original test from the bug report.
Differential Revision: https://reviews.llvm.org/D83319
When an argument has 'byval' attribute and should be
passed on the stack according calling convention,
a stack copy would be emitted twice. This will cause
the real value will be put into stack where the pointer
should be passed.
Differential Revision: https://reviews.llvm.org/D83175
Probably not super important since there are no real CPUs with
avx512vl and not avx512bw. But vpternlog should be better than
vblendvb.
I do wonder if we should use vpternlog even with BWI. We
currently use vblendmb or vpblendmw by putting the mask into a GPR
and moving it to a k-register. But I don't think we hoist the
GPR to k-register copy in machine LICM. Using VPTERNLOG would use
a constant pool load, but has the advantage that we're pretty good
at hoisting and rematerializing those.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D83156
VPBLENDVB is multiple uops while VPTERNLOG is a single uop. So
we should use that instead.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D83155
Using PACK for truncations leaves us with intermediate shuffles that can be tricky to remove while the truncation tree is being formed.
This fold helps pull out the PERMQ case which is one of the most common, avoiding some costly lane-crossing shuffles.
A future patch will begin adding more general shuffle folding, which we should be able to use for HADD/HSUB as well.
We canonicalize patterns like:
%s = lshr i32 %a0, 1
%t = trunc i32 %s to i1
to:
%a = and i32 %a0, 2
%c = icmp ne i32 %a, 0
...in IR, but the bit-shifting original sequence may be better for x86 vector codegen.
I tried several variants of the transform, and it's tricky to not induce regressions.
In particular, I did not find a way to cleanly handle non-splat constants, so I've left
that as a TODO item here (currently negative tests for those are included). AVX512
resulted in some diffs, but didn't look meaningful, so I left that out too. Some of
the 256-bit AVX1 diffs are questionable, but close enough that they are probably
insignificant.
Differential Revision: https://reviews.llvm.org/D83073.
I think this mostly looks ok. The only weird thing I noticed was
a couple rotate vXi8 tests picked up an extra logic op where we have
(and (or (and), (andn)), X). Previously we matched the (or (and), (andn))
to vpternlog, but now we match the (and (or), X) and leave the and/andn
unmatched.
We consider v32i16/v64i8 to be legal types on avx512f, but we
don't have most operations until avx512bw. But we can use
and/or/xor operations. So try those before splitting.
This is especially helpful since we turn some ands with constant
masks into shuffles in early DAG combines. So we should make sure
we recover those back to AND.
The comments here indicate that we prefer to promote the shifts
instead of allowing rotate to be pattern matched. But we weren't
taking into account whether 512-bit registers are enabled or
whethever we have vpsllvw/vpsrlvw instructions.
splatvar_rotate_v32i8 is a slight regrssion, but the other cases
are neutral or improved.
Summary:
When parsing 64-bit MASM, treat memory operands with unspecified base register as RIP-based.
Documented in several places, including https://software.intel.com/en-us/articles/introduction-to-x64-assembly: "Unfortunately, MASM does not allow this form of opcode, but other assemblers like FASM and YASM do. Instead, MASM embeds RIP-relative addressing implicitly."
Reviewed By: thakis
Differential Revision: https://reviews.llvm.org/D73227
D82257/rG3521ecf1f8a3 was incorrectly sign-extending a constant vector from the lsb, this is fine if all the constant elements are 'allsignbits' in the active bits, but if only some of the elements are, then we are corrupting the constant values for those elements.
This fix ensures we sign extend from the msb of the active/demanded bits instead.
This function picks X86 opcode name based on type, masking,
and whether not a load or broadcast has been folded using multiple
switch statements. The contents of the switches mostly just vary in
a few characters in the instruction name. So use some macros to
build the instruction names to reduce the repetiveness.
If we're masking the result of an OR-reduction before comparing against zero, we can fold this into the PTEST() / MOVMSK(CMPEQ()) codegen by pre-masking the source value.
This works particularly well on PTEST which performs the AND as part of its operation, but the MOVMSK variant also benefits for non-V2I64 cases.
Fixes PR44781
If the shuffle is a blend and one input is a 0 vector, we should prefer AND over PSHUFB since its available on more execution ports.
Differential Revision: https://reviews.llvm.org/D82798
This was producing reg = xor undef reg, undef reg. This looks similar
to a use of a value to define itself, and I want to disallow undef
uses for SSA virtual registers. If this were to use implicit_def,
there's no guarantee the two operands end up using the same register
(I think no guarantee exists even if the two operands start out as the
same register, but this was violated when I switched this to use an
explicit implicit_def). The MOV32r0 pseudo evidently exists to handle
this case, so use it instead. This was more work than I expected for
the 64-bit case, but I didn't see any helper for materializing a
64-bit 0.
If a constant is only allsignbits in the demanded/active bits, then sign extend it to an allsignbits bool pattern for OR/XOR ops.
This also requires SimplifyDemandedBits XOR handling to be modified to call ShrinkDemandedConstant on any (non-NOT) XOR pattern to account for non-splat cases.
Next step towards fixing PR45808 - with this patch we now get a <-1,-1,0,0> v4i64 constant instead of <1,1,0,0>.
Differential Revision: https://reviews.llvm.org/D82257
Pre-commit for D82257, this adds a DemandedElts arg to ShrinkDemandedConstant/targetShrinkDemandedConstant which will allow future patches to (optionally) add vector support.
We already fold (v2i64 scalar_to_vector(aext)) -> (v2i64 bitcast(v4i32 scalar_to_vector(x))), this adds support for similar aextload cases and also handles v2f64 cases that wrap the i64 extension behind bitcasts.
Fixes the remaining issue with PR39016
Generalize the vector operand extraction code for shuffle/pack ops - we can assume that the vector operands are the same width as the result, and any non-vector values can be reused directly in the smaller width op.
Summary:
A while ago I implemented the functionality to lower Microsoft __ptr32
and __ptr64 pointers, which are stored as 32-bit and 64-bit pointer
and are extended/truncated to the appropriate pointer size when
dereferenced.
This patch adds an addrspacecast to cast from the __ptr32/__ptr64
pointer to a default address space when dereferencing.
Bug: https://bugs.llvm.org/show_bug.cgi?id=42359
Reviewers: hans, arsenm, RKSimon
Subscribers: wdng, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D81517
These features implicitly enabled XSAVE in the frontend, but not
the backend. Disabling XSAVE in the frontend disabled XSAVEOPT, but
not the other 2. Nothing happened in the backend.
The PREFETCHW instruction was originally part of the 3DNow. But
it was given its own CPUID bit on later CPUs just before 3DNow
was deprecated.
We were setting the -mprfchw flag if -m3dnow was passed or the CPU
supported 3dnow unless -mno-prfchw was passed. But -march=native
on a CPU without the PRFCHW CPUID bit set will pass -mno-prfchw.
So -march=k8 will behave differently than -march=native on a K8
for example.
So remove this implicit setting from the frontend and instead
enable the backend to use PREFETCHW if 3dnow OR prfchw is enabled.
Also enable PRFCHW flag on amdfam10/barcelona which seems to be
where this CPUID bit was introduced. That CPU also supported
3dnow.
The PREFETCHW instruction was originally part of the 3DNow. But
it was given its own CPUID bit on later CPUs just before 3DNow
was deprecated.
We were setting the -mprfchw flag if -m3dnow was passed or the CPU
supported 3dnow unless -mno-prfchw was passed. But -march=native
on a CPU without the PRFCHW CPUID bit set will pass -mno-prfchw.
So -march=k8 will behave differently than -march=native on a K8
for example.
So remove this implicit setting from the frontend and instead
enable the backend to use PREFETCHW if 3dnow OR prfchw is enabled.
Also enable PRFCHW flag on amdfam10/barcelona which seems to be
where this CPUID bit was introduced. That CPU also supported
3dnow.
The main interface has been migrated to Align already but a few backends where broadening the type from Align to MaybeAlign.
This patch makes sure all implementations conform to the public API.
Differential Revision: https://reviews.llvm.org/D82465
Previously we just updated a map and moved on. But it possible
we cached known bits information with the vreg that can be used by
another basic block. If the other basic block has a different view
of the VT these known bits won't make sense.
By emitting a copy we ensure we have different vregs before and
after the bitcast. This prevents the known bits from being used
with the wrong type.
Differential Revision: https://reviews.llvm.org/D82517
Summary:
Get back `const` partially lost in one of recent changes.
Additionally specify explicit qualifiers in few places.
Reviewers: samparker
Reviewed By: samparker
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82383
Eric Christopher informed me that FastISel memcpy handling creates
load/store instructions without mem operands. We should fix that,
but I doubt that's the only case of missed mem operands so seems
better to be defensive here.
I don't have a test case yet, but I'll try to add one if i get a
test from Eric.
This caused a Chromium test to miscompile. See discussion on the Phabricator
review.
> This patch extends MatchVectorAllZeroTest to handle OR vector reduction patterns where the result is compared against zero.
>
> Fixes PR45378
>
> Differential Revision: https://reviews.llvm.org/D81547
This reverts 057c9c7ee0
Following on from this RFC[0] from a while back, this is the first patch towards
implementing variadic debug values.
This patch specifically adds a set of functions to MachineInstr for performing
operations specific to debug values, and replacing uses of the more general
functions where appropriate. The most prevalent of these is replacing
getOperand(0) with getDebugOperand(0) for debug-value-specific code, as the
operands corresponding to values will no longer be at index 0, but index 2 and
upwards: getDebugOperand(x) == getOperand(x+2). Similar replacements have been
added for the other operands, along with some helper functions to replace
oft-repeated code and operate on a variable number of value operands.
[0] http://lists.llvm.org/pipermail/llvm-dev/2020-February/139376.html<Paste>
Differential Revision: https://reviews.llvm.org/D81852
We have many cases where we call SimplifyMultipleUseDemandedBits and demand specific vector elements, but all the bits from them - this adds a helper wrapper to handle this.
If a collection of interconnected phi nodes is only ever loaded, stored
or bitcast then we can convert the whole set to the bitcast type,
potentially helping to reduce the number of register moves needed as the
phi's are passed across basic block boundaries. This has to be done in
CodegenPrepare as it naturally straddles basic blocks.
The alorithm just looks from phi nodes, looking at uses and operands for
a collection of nodes that all together are bitcast between float and
integer types. We record visited phi nodes to not have to process them
more than once. The whole subgraph is then replaced with a new type.
Loads and Stores are bitcast to the correct type, which should then be
folded into the load/store, changing it's type.
This comes up in the biquad testcase due to the way MVE needs to keep
values in integer registers. I have also seen it come up from aarch64
partner example code, where a complicated set of sroa/inlining produced
integer phis, where float would have been a better choice.
I also added undef and extract element handling which increased the
potency in some cases.
This adds it with an option that defaults to off, and disabled for 32bit
X86 due to potential issues around canonicalizing NaNs.
Differential Revision: https://reviews.llvm.org/D81827
Pulled out from the ongoing work on D66004, currently we don't do a good job of simplifying variable shuffle masks that have already lowered to constant pool entries.
This patch adds SimplifyDemandedVectorEltsForTargetShuffle (a custom x86 helper) to first try SimplifyDemandedVectorElts (which we already do) and then constant pool simplification to help mark undefined elements.
To prevent lowering/combines infinite loops, we only handle basic constant pool loads instead of creating new BUILD_VECTOR nodes for lowering - e.g. we don't try to convert them to broadcast/vzext_load - there might be some benefit to this but if so I'd rather we come up with some way to reuse existing code than reimplement a lot of BUILD_VECTOR code.
Differential Revision: https://reviews.llvm.org/D81791
We were missing the modrm byte this instruction has according
to current Intel SDM. Experiments with gcc indicate that different
modrm values are chosen based on 2 operands so I've added those
as well.
I think our previous implementation was based on an older behavior of
binutils that has since been changed.
These are documented as using modrm byte of 0xe8, 0xf0, and 0xf8
respectively. But hardware ignore bits 2:0. So 0xe9-0xef is treated
the same as 0xe8. Similar for the other two.
Fixing this required adding 8 new formats to the X86 instructions
to convey this information. Could have gotten away with 3, but
adding all 8 made for a more logical conversion from format to
modrm encoding.
I renumbered the format encodings to keep the register modrm
formats grouped together.
Without SSE41 we don't have the PCMPEQQ instruction, making cmp-with-zero reductions more complicated than necessary. We can compare as vXi32 (PCMPEQD) and tweak the MOVMSK comparison to test upper/lower DWORD comparisons.
This pre-fixes something that occurs with null tests for vectors of (64-bit) pointers such as in PR35129.
This was passing in all the parameters needed to construct a
LegalizerHelper in the custom legalization, when it's simpler to just
pass in the existing helper.
This is slightly more annoying to use in the common case where you
don't need the legalizer helper, but we could add back the common
parameters back in addition to the helper.
I didn't propagate this to all the internal target changes that this
logically implies, but did update a sample one for
legalizeMinNumMaxNum.
This is in preparation for moving AMDGPU load/store legalization
entirely into custom lowering. The current set of legalization actions
is really constraining and not really capable of expressing all the
actions needed to legalize loads/stores. In particular there's no way
to express when the memory access itself needs to change size vs. the
result type. There's also a lot of redundancy since the same
split/widen actions need to be applied in both vector and scalar
cases. All of the sub-cases logically belong as steps in the legalizer
helper, but it will be easier to consider everything at once in custom
lowering.
This also enables running the AArch64 SLSHardening pass with GlobalISel,
so add a test for that.
Differential Revision: https://reviews.llvm.org/D81403
In order to support hot-patching, we need to make sure the first emitted instruction in a function is a two-byte+ op. This is already the case on x86_64, which seems to always emit two-byte+ ops. However on 32-bit targets this wasn't the case.
PATCHABLE_OP now lowers to a XCHG AX, AX, (66 90) like MSVC does. However when targetting pentium3 (/arch:SSE) or i386 (/arch:IA32) targets, we generate MOV EDI,EDI (8B FF) like MSVC does. This is for compatiblity reasons with older tools that rely on this two byte pattern.
Differential Revision: https://reviews.llvm.org/D81301
This patch extends MatchVectorAllZeroTest to handle OR vector reduction patterns where the result is compared against zero.
Fixes PR45378
Differential Revision: https://reviews.llvm.org/D81547
Most of the wrappers exist to print the memory size in Intel syntax
and then call the printMemReference. But printanymem/printopaquemem
don't print anything extra in Intel syntax so just drop them.
Pull the lowering code out of LowerVectorAllZeroTest (and rename it MatchVectorAllZeroTest).
We should be able to reuse this in combineVectorSizedSetCCEquality as well.
Another cleanup to simplify D81547.
Reduce by splitting the vector until we reach the target size for PTEST/MOVMSK_PCMPEQ. There might be some cases where AVX512 can perform this with 512-bit vectors but so far I haven't encountered any such pattern that reaches LowerVectorAllZeroTest.
Prep work for D81547
> relocImm was a complexPattern that handled both ConstantSDNode
> and X86Wrapper. But it was only applied selectively because using
> it would cause patterns to be not importable into FastISel or
> GlobalISel. So it only got applied to flag setting instructions,
> stores, RMW arithmetic instructions, and rotates.
>
> Most of the test changes are a result of making patterns available
> to GlobalISel or FastISel. The absolute-cmp.ll change is due to
> this fixing a pattern ordering issue to make an absolute symbol
> match to an 8-bit immediate before trying a 32-bit immediate.
>
> I tried to use PatFrags to reduce the repetition, but I was getting
> errors from TableGen.
This caused "Invalid EmitNode" assertions, see the llvm-commits thread for
discussion.
matchScalarReduction should return all its source vectors with the same type, so we can safely perform the OR reduction with the original type.
So we just need to bitcast for PTEST/PCMPEQB with the final reduced vector.
Have BasicTTI call the base implementation so that both agree on the
default behaviour, which the default being a cost of '1'. This has
required an X86 specific implementation as it seems to be very
reliant on those instructions being free. Changes are also made to
AMDGPU so that their implementations distinguish between cost kinds,
so that the unrolling isn't affected. PowerPC also has its own
implementation to prevent changes to the reg-usage vectorizer test.
The cost model test changes now reflect that ret instructions are not
generally free.
Differential Revision: https://reviews.llvm.org/D79164
When checking for an enum function attribute, use hasFnAttribute()
rather than hasAttribute() at FunctionIndex, because it is
significantly faster (and more concise to boot).
Reduce XMM->GPR traffic by performing bitops on the vectors, and using a single MOVMSK call.
This requires us to use vectors of the same size and element width, but we can mix fp/int type equivalents with suitable bitcasting.
We never codegen them so this doesn't matter in practice. But
sometimes someone comes along and tries to use these flags
for something else. LIke the Load Value Inject inline assembly
handling.
If the input to the bitcast is a sign bit test, it makes sense to
directly use vpmovmskb or vmovmskps/pd. This removes the need to
copy the sign bits to a k-register and then to a GPR.
Fixes PR46200.
Differential Revision: https://reviews.llvm.org/D81327
This pass has no dependencies on other passes so conditionally
including it in the pipeline doens't do much. Just move it the
pass itself to keep it isolated.
A lot of what EVEX->VEX does is equivalent to what the
prioritization in the assembly parser does. When an AVX mnemonic
is used without any EVEX features or XMM16-31, the parser will
pick the VEX encoding.
Since codegen doesn't go through the parser, we should also
use VEX instructions when we can so that the code coming out of
integrated assembler matches what you'd get from outputing an
assembly listing and parsing it.
The pass early outs if AVX isn't enabled and uses TSFlags to
check for EVEX instructions before doing the more costly table
lookups. Hopefully that's enough to keep this from impacting
-O0 compile times.
relocImm was a complexPattern that handled both ConstantSDNode
and X86Wrapper. But it was only applied selectively because using
it would cause patterns to be not importable into FastISel or
GlobalISel. So it only got applied to flag setting instructions,
stores, RMW arithmetic instructions, and rotates.
Most of the test changes are a result of making patterns available
to GlobalISel or FastISel. The absolute-cmp.ll change is due to
this fixing a pattern ordering issue to make an absolute symbol
match to an 8-bit immediate before trying a 32-bit immediate.
I tried to use PatFrags to reduce the repetition, but I was getting
errors from TableGen.
Noticed while trying to cleanup D66004 - if a shuffle operand came from a scalar, we're better off using INSERTPS vs UNPCKLPS as this is more likely to load fold later on. It also matches our existing BUILD_VECTOR lowering.
We can extend this to other PINSRB/D/Q/W cases in the future as the need arises.
The spec for these says they need 0xf3 but also mentions REP
before the mnemonic. But I don't think its fair to users to make
them write REP first. And gas doesn't make them. objdump seems to
disassemble with or without the prefix and just prints any 0xf3
as REP.
'NP' means that the instruction is not recognized with a 66, F2 or F3
prefix. It will either #UD or decode to a different instruction.
All of the cases are here should fall into the #UD variety since
we should be detecting the collision with other instructions when
we build the disassembler tables.
Convert shift+or bool vector patterns into CONCAT_VECTORS if we know this will be lowered to KUNPCK (which requires 16+ vector elements).
Fixes PR32547
AVX512 mask types are often bitcasted to scalar integers for various ops before being bitcast back to be used as a predicate. In many cases we can avoid these KMASK<->GPR transfers and perform equivalent operations on the mask unit.
If the destination mask type is legal, and we can confirm that the scalar op originally came from a mask/vector/float/double type then we should try to avoid the scalar entirely.
This avoids some codegen issues noticed while working on PTEST/MOVMSK improvements.
Partially fixes PR32547 - we don't create a KUNPCK yet, but OR(X,KSHIFTL(Y)) can be handled in a separate patch.
Differential Revision: https://reviews.llvm.org/D81548
By moving target-independent code from
llvm/lib/Target/X86/X86IndirectThunks.cpp
to
llvm/include/llvm/CodeGen/IndirectThunks.h
Differential Revision: https://reviews.llvm.org/D81401
This appears to have been added when In64BitMode was added to a
bunch of instructions that don't have register operands. When an
instruction uses a register the parser will prevent a 64-bit
register from being parsed on a 32-bit target. But with only
memory and immediate operands this doesn't happen.
TEST64ri32 does have a register operand so the issue the predicate
was supposed to fix doesn't apply.
@nikic raised an issue on D75936 that the added complexity to the O0 pipeline was causing noticeable slowdowns for `-O0` builds. This patch addresses the issue by adding a pass with equal security properties, but without any optimizations (and more importantly, without the need for expensive analysis dependencies).
Reviewers: nikic, craig.topper, mattdr
Reviewed By: craig.topper, mattdr
Differential Revision: https://reviews.llvm.org/D80964
This is a cheap instruction. It's better to repeat it than to do
two separate operations.
There are probably more cases like this, but this one was reported
as a regression in our internal benchmarking.
If we probe *after* each static stack allocation, we need to probe *before* each
dynamic stack allocation. Provide a scheme to describe the possible scenario.
Thanks a lot to @jonpa for motivating this fix.
Differential Revision: https://reviews.llvm.org/D81067
Add the remaining arithmetic opcodes into the generic implementation
of getUserCost and then call this from getInstructionThroughput. Most
of the backends have been modified to return the base implementation
for cost kinds other RecipThroughput. The outlier here is AMDGPU
which already uses getArithmeticInstrCost for all the cost kinds.
This change means that most of the opcodes can be removed from that
backends implementation of getUserCost.
Differential Revision: https://reviews.llvm.org/D80992
As shown in PR46237:
https://bugs.llvm.org/show_bug.cgi?id=46237
The size-savings win for hoisting an 8-bit ALU immediate (intentionally
excluding store constants) requires extreme conditions; it may not even
be possible when including REX prefix bytes on x86-64.
I did draft a version of this patch that included use counts after the
loop, but I suspect that accounting is not working as expected. I think
that is because the number of constant uses are changing as we select
instructions (for example as we transform shl/add into LEA).
Differential Revision: https://reviews.llvm.org/D81468
Add cases for icmp, fcmp and select into the switch statement of the
generic getUserCost implementation with getInstructionThroughput then
calling into it. The BasicTTI and backend implementations have be set
to return a default value (1) when a cost other than throughput is
being queried.
Differential Revision: https://reviews.llvm.org/D80550
LowerSELECT sees the CMP with 0 and wants to use a trick with SUB
and SBB. But we can use the flags from the BSF/TZCNT.
Fixes PR46203.
Differential Revision: https://reviews.llvm.org/D81312
This combine tries shrink a vzmovl if its input is an
insert_subvector. This patch improves it to turn
(vzmovl (bitcast (insert_subvector))) into
(insert_subvector (vzmovl (bitcast))) potentially allowing the
bitcast to be folded with a load.
We can pad the v2f32 with 0s up to v8f32 and use a v8f32->v8i64
operation. This is what we end up with on non-strict nodes except
we don't pad with 0s since we don't care about exceptions.
In the sign splat case, we can fold PMOVMSKB(PACKSSBW(LO(X), HI(X))) -> PMOVMSKB(BITCAST_v32i8(X)) without introducing a signmask + comparison (which unlike for any_of won't fold into a single TEST).
Handle MOVMSK 'allof' comparisons (X86ISD::SUB X, AllBitsMask) as well as 'anyof' patterns.
This allows us to handle these patterns in the MOVMSK(BITCAST(X)) pattern to fix PR37087.
As shown on PR37087, if we have a MOVMSK(BICAST(X)) from a wider vector, then by using MOVMSK from the wider type (32/64-bit elements) we can improve the chances of further combines with SimplifyDemandedBits/Elts and on some targets (skylake) can be more efficient.
Shifts are supposed to always shift in zeros or sign bits regardless of their inputs. It's possible the input value may have been replaced with undef by SimplifyDemandedBits, but the shift in zeros are still demanded.
This issue was reported to me by ispc from 10.0. Unfortunately their failing test does not fail on trunk. Seems to be because the shl is optimized out earlier now and doesn't become VSHLI.
ispc bug https://github.com/ispc/ispc/issues/1771
Differential Revision: https://reviews.llvm.org/D81212
Motivating examples are seen in the PhaseOrdering tests based on:
https://bugs.llvm.org/show_bug.cgi?id=43953#c2 - if we have
intrinsics there, some pass can fold them.
The intrinsics are still named "experimental" at this point, but
if there is no fallout from this patch, that will be a good
indicator that it is safe to finalize them.
Differential Revision: https://reviews.llvm.org/D80867
An initial patch adding combineSetCCMOVMSK to simplify MOVMSK and its vector input based on the comparison of the MOVMSK result.
This first stage just adds support for some simple MOVMSK(PACKSSBW()) cases where we remove the PACKSS if we're comparing ne/eq zero (any_of patterns), allowing us to directly compare against the v8i16 source vector(s) bitcasted to v16i8, with suitable masking to take into account of which signbits are valid.
Future combines could peek through further PACKSS, target shuffles, handle all_of patterns (ne/eq -1), optimize to a PTEST op, etc.
Differential Revision: https://reviews.llvm.org/D81171
Use getMemoryOpCost from the generic implementation of getUserCost
and have getInstructionThroughput return the result of that for loads
and stores.
This also means that the X86 implementation of getUserCost can be
removed with the functionality folded into its getMemoryOpCost.
Differential Revision: https://reviews.llvm.org/D80984
I think these are left over from when we used to type legalize
v2f32 loads using bitcast+scalar_to_vec+loadi64 on 64-bit targets.
These days we use loadf64. If this becomes a problem a better
solution would be a DAG combine to turn it into scalar_to_vec+loadf64.
If we're only demanding the (shifted) sign bits of the shift source value, then we can use the value directly.
This handles SimplifyDemandedBits/SimplifyMultipleUseDemandedBits for both ISD::SHL and X86ISD::VSHLI.
Differential Revision: https://reviews.llvm.org/D80869
-Fix one place where we had a X86vzload64 but should have had
X86vzload32.
-Make sure all patterns that have scalar_to_vector+loadi64 also
have scalar_to_vector+f64 to match 32-bit codegen.
-Add some bitcasts that were missing from patterns.
-Make sure that if we have a scalar_to_vector+load pattern
we also have a vzload pattern.
We probably need some better canonicalization to avoid having
so many patterns.
Instead of using a fake call and metadata to temporarily represent a probed
static alloca, use a pseudo instruction.
This is inspired by the SystemZ approach proposed in https://reviews.llvm.org/D78717.
Differential Revision: https://reviews.llvm.org/D80641
We looked through a truncate to get to the load. So we should be
deleting the truncate first.
There is a check that the node is really unused before deleting
so this didn't cause a functional issue.
Summary:
While clustering mem ops, AMDGPU target needs to consider number of clustered bytes
to decide on max number of mem ops that can be clustered. This patch adds support to pass
number of clustered bytes to target mem ops clustering logic.
Reviewers: foad, rampitec, arsenm, vpykhtin, javedabsar
Reviewed By: foad
Subscribers: MatzeB, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, javed.absar, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D80545
Previously we walked the users of any vector binop looking for
more binops with the same opcode or phis that eventually ended up
in a reduction. While this is simple it also means visiting the
same nodes many times since we'll do a forward walk for each
BinaryOperator in the chain. It was also far more general than what
we have tests for or expect to see.
This patch replaces the algorithm with a new method that starts at
extract elements looking for a horizontal reduction. Once we find
a reduction we walk through backwards through phis and adds to
collect leaves that we can consider for rewriting.
We only consider single use adds and phis. Except for a special
case if the Add is used by a phi that forms a loop back to the
Add. Including other single use Adds to support unrolled loops.
Ultimately, I want to narrow the Adds, Phis, and final reduction
based on the partial reduction we're doing. I still haven't
figured out exactly what that looks like yet. But restricting
the types of graphs we expect to handle seemed like a good first
step. As does having all the leaves and the reduction at once.
Differential Revision: https://reviews.llvm.org/D79971
This matches what we do for the full sized vector ops at the start of combineX86ShufflesRecursively, and helps getFauxShuffleMask extract more INSERT_SUBVECTOR patterns.
Try to prevent future node creation issues (as detailed in PR45974) by making the SelectionDAG reference const, so it can still be used for analysis, but not node creation.
As detailed on PR45974 and D79987, getFauxShuffleMask is creating nodes on the fly to create shuffles with inputs the same size as the result, causing problems for hasOneUse() checks in later simplification stages.
Currently only combineX86ShufflesRecursively benefits from these widened inputs so I've begun moving the functionality there, and out of getFauxShuffleMask. This allows us to remove the widening from VBROADCAST and *EXTEND* faux shuffle cases.
This just leaves the INSERT_SUBVECTOR case in getFauxShuffleMask still creating nodes, which will require more extensive refactoring.
We already had a DAG combine for (mmx (bitconvert (i64 (extractelement v2i64))))
to MOVDQ2Q.
Remove patterns for MMX_MOVQ2DQrr/MMX_MOVDQ2Qrr that use
scalar_to_vector/extractelement involving i64 scalar type with
v2i64 and x86mmx.
The instruction is defined to only produce high result if both
destinations are the same. We can exploit this to avoid
unnecessarily clobbering a register.
In order to hide this from register allocation we use a pseudo
instruction and expand the result during MCInst creation.
Differential Revision: https://reviews.llvm.org/D80500
-Replace some ifs that should be impossible with asserts.
-Use X86::AddrDisp and X86::AddrNumOperands to make code more readable
-Use X86II::isKMasked/isKMergeMasked to do some operand skipping to remove or simplify switches
Let the codegen recognized the nomerge attribute and disable branch folding when the attribute is given
Differential Revision: https://reviews.llvm.org/D79537
Large code model doesn't mean anything to 32-bit mode. But nothing
prevents it from being set. Ignore to avoid generating 64-bit mode
only instructions.
Differential Revision: https://reviews.llvm.org/D80768
Only 64-bit bits will be loaded, not the whole 128 bits. We can
just combine it to plain mmx load. This has the side effect of
enabling isel load folding for it.
This part of my desire to get rid of isel patterns that shrink loads.
LowerConstantPool passes a nullptr into classifyLocalReference. The medium code model handling for PIC will try to deference it using isa. This patch switches to isa_and_nonnull.
Differential Revision: https://reviews.llvm.org/D80763
Summary:
Propagate memory operands when folding load instructions into instructions that directly operate on memory.
The original revision has been split. See D80140 for the other part of the changes.
Reviewers: craig.topper, rnk, lebedev.ri, efriedma
Reviewed By: craig.topper
Subscribers: lebedev.ri, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D80062
Looking back over gcc and icc behavior it looks like icc does
use mulx32 on 32-bit targets and mulx64 on 64-bit targets. It's
also used when dividing i32 by constant on 32-bit targets and
i64 by constant on 64-bit targets.
gcc uses it multiplies producing a 64 bit result on 32-bit targets
and 128-bit results on a 64-bit target. gcc does not appear to use
it for division by constant.
After this patch clang is closer to the icc behavior. This
basically reverts d1c61861dd, but
there were no strong feelings at the time.
Fixes PR45518.
Differential Revision: https://reviews.llvm.org/D80498
ffmpeg/libavcodec/x86/h264_cabac.c inline assembly may produce
movzb 1280(%rbx, %r12), %r12
After D80608, llvm-mc errors:
error: unknown use of instruction mnemonic without a size suffix
If we are using PTEST to check 'allsign bits' vector elements we can use MOVMSK to extract the signbits directly and perform the comparison on the scalar value.
For vXi16 cases, as we don't have a MOVMSK for this type, we must mask each signbit out of a PMOVMSKB v2Xi8 result, which folds into the TEST comparison.
If this allows us to remove a vector op (via the SimplifyMultipleUseDemandedBits call) this is consistently faster than a PTEST (https://godbolt.org/z/ziJUst).
I'm investigating whether we ever get regressions without the SimplifyMultipleUseDemandedBits call, even if this means we don't remove a vector op, but that has exposed some other poor codegen issues that I'm still investigating and would have to wait for a later patch.
Suggested on PR42035 to avoid unnecessary ashr(x,bw-1)/pcmpgt(0,x) sign splat patterns feeding into ptest.
Differential Revision: https://reviews.llvm.org/D80563
There's more code for calling CombineTo and replacing the nodes
that I'd like to share, but its complicated by the getNode call
in the middle that needs to be specific to each opcode.
While there are also make sure we recursively delete the load
we're replacing. It eventually gets removed by a RemoveDeadNodes
call at the end of DAG combine, but we should be more eager about
it. We were inconsistently doing this in some places but not all.
Isel match that instead of the intrinsic. Similar to what we do
for avx512.
Trying to move more intrinsics to target specific ISD opcodes.
Hoping to add DAG combines to shrink simple loads going into
scalar intrinsics that only read 32 or 64 bits.
Summary:
Some instruction like VPMULDQ is NOT the variant of VPMULD but a new
one.
So we should make sure the suffix matcher only works for memory variant
that has the same size with the suffix.
Currently we only check for SSE/AVX* instructions, because many legacy
instructions didn't declare the alias instructions of their variants.
Differential Revision: https://reviews.llvm.org/D80608
Add the remaining cast instruction opcodes to the base implementation
of getUserCost and directly return the result. This allows
getInstructionThroughput to return getUserCost for the casts. This
has required changes to PPC and SystemZ because they implement
getUserCost and/or getCastInstrCost with adjustments for vector
operations. Adjusts have also been made in the remaining backends
that implement the method so that they still produce a cost of zero
or one for cost kinds other than throughput.
Differential Revision: https://reviews.llvm.org/D79848
Recommitting most of the remaining changes from
259eb619ff, but excluding the call to
getUserCost from getInstructionThroughput. Though there's still no
test changes, I doubt that this is an NFC...
With the two getIntrinsicInstrCosts folded into one, now fold in the
scalar/code-size orientated getIntrinsicCost. The remaining scalar
intrinsics were memcpy, cttz and ctlz which now have special handling
in the BasicTTI implementation.
This had required a change in the AMDGPU backend for fabs as it
should always be 'free'. I've also changed the X86 backend to return
the BaseT implementation when the CostKind isn't RecipThroughput.
Differential Revision: https://reviews.llvm.org/D80012
Simon Pilgrim