The division expansions in AMDGPUCodeGenPrepare can't be relied on for
correctness, since they punt to later optimization and possibly
legalization in some cases. We still need a way to be able to write
tests for the legalizer versions of the expansion. This is mostly for
GlobalISel, since the expected optimzations is expecting aren't
implemented.
The interaction with the flag to expand 64-bit division in the IR is
pretty confusing, but these flags have different purposes.
I didn't realize we were already expanding 24/32-bit division here
already. Use the available IntegerDivision utilities. This uses loops,
so produces significantly smaller code than the inline DAG expansion.
This now requires width reductions of 64-bit divisions before
introducing the expanded loops.
This helps work around missing legalization in GlobalISel for
division, which are the only remaining core instructions that didn't
work at all.
I think this is plausibly a better implementation than exists in the
DAG, although turning it on by default misses out on the constant
value optimizations and also needs benchmarking.
This is more or less directly ported from the AMDGPU custom lowering
for FP_TO_FP16. I made a few minor fixups (using G_UNMERGE_VALUES
instead of creating shift/trunc to extract the two halves, and zexting
an inverted compare instead of select_cc).
This also does not include the fast math expansion the DAG which
converts to f32 and then to f16. I think that belongs in a
pre-legalize combine instead.
Also greatly improve i64 lowering. LegalizeIntegerTypes does the
correct narrowing if i64 isn't legal. Just workaround this for
SelectionDAG by making i64 legal and splitting in the patterns.
When SI_IF is inserted, it constrains the source register with a
register class, which was quite likely a G_ICMP. This was incorrectly
treating it as a scalar, and then applyMappingImpl would end up
producing invalid MIR since this was unexpected.
Also fix not using all VGPR sources for vcc outputs.
Load extra bits if suitably aligned. This allows using widened
3-vector loads on SI, and fixes legalization for <9 x s32> (which LSV
apparently forms frequently on lowered kernel argument lists).
Fix incorrectly treating these as legal on SI. This should emit a
64-bit store and a 32-bit store.
I think all of the load and store rules are just about complete, but
due for a rewrite.
The isNegatibleForFree/getNegatedExpression methods currently rely on a raw char value to indicate whether a negation is beneficial or not.
This patch replaces the char return value with an NegatibleCost enum to more clearly demonstrate what is implied.
It also renames isNegatibleForFree to getNegatibleCost to more accurately reflect whats going on.
Differential Revision: https://reviews.llvm.org/D74221
Summary:
Consider:
%r = call i32 @llvm.amdgcn.writelane(i32 0, i32 1, i32 2)
This produces a value that is 0 on lane 1, and 2 everywhere else; i.e.,
it is divergent.
Reported-by: Marek Olsak <Marek.Olsak@amd.com>
Reviewers: arsenm, foad, mareko
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D74400
This was creating a select on true/false values, and then comparing
that later. This produced more work for later combines, which can be
avoided by just using the boolean values. This was copied from the
original DAG expansion, which also has the same problem. This doesn't
have a observable change using SelectionDAG, but since GlobalISel is
missing these optimizations, the final code was noticeably longer.
These have nicer expansions implemented in the DAG. Ideally we would
either directly implement all of these special expansions, or stop
expanding division in the IR.
This is apparently worse than 1-byte alignment. This does not attempt
to decompose 2-byte aligned wide stores, but will stop trying to
produce them.
Also fix bug in LoadStoreVectorizer which was decreasing the alignment
and vectorizing stack accesses. It was assuming a stack object was an
alloca that could have its base alignment changed, which is not true
if the pointer is derived from a function argument.
Since natural fdiv lowering is now more conservative even with
denormals disabled, we get a slower expansion from just a plain
1.0/fdiv. Directly emit the rcp intrinsic when using it to implement
integer division to avoid a pointlessly complex sequence.
Summary:
SIInstrInfo::expandPostRAPseudo converts ENTER_WWM in-place into an
S_OR_SAVEEXEC instruction that needs certain implicit operands. Without
this patch I get errors like this that make it harder to use -stop-after
to bisect the pass pipeline:
$ llc -march=amdgcn test/CodeGen/AMDGPU/wqm.ll -stop-after=postrapseudos -o - | sed -E 's/ (from|into) custom "TargetCustom[0-9]+"//' | llc -march=amdgcn -x=mir
error: <stdin>:1295:70: missing implicit register operand 'implicit-def $scc'
renamable $sgpr2_sgpr3 = S_OR_SAVEEXEC_B64 -1, implicit-def $exec
^
Note that this error is currently only generated by MIParser but it
comes with a FIXME comment:
// FIXME: Move the implicit operand verification to the machine verifier.
Reviewers: critson, arsenm, rampitec, nhaehnle
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D74428
Summary:
Add a new method (tryParseRegister) that attempts to parse a register specification.
MASM allows the use of IFDEF <register>, as well as IFDEF <symbol>. To accommodate this, we make it possible to check whether a register specification can be parsed at the current location, without failing the entire parse if it can't.
Reviewers: thakis
Reviewed By: thakis
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73486
Summary:
As far as I know this did not affect code generation, but it did affect
the order of -debug-only=si-wqm output and the naming of autonamed
values in -print-after=si-wqm output.
Reviewers: arsenm, rampitec, nhaehnle
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, mgrang, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D74317
These are generated and do not need to have the same values.
We are defining separate subregs for R600 and GCN but then
using AMDGPU subregs on R600.
Differential Revision: https://reviews.llvm.org/D74248
Based on D72931
This adds a new feature called A16 which is enabled for gfx10.
gfx9 keeps the R128A16 feature so it can share all the instruction encodings
with gfx7/8.
Differential Revision: https://reviews.llvm.org/D73956
Vector indexing with a constant index should be folded out in the
legalizer, but this was accidentally falling through. This would
produce the indexing operation with $noreg. Handle this case as a
dynamic index just in case a bug like this happens again in the
future.
We were failing to find constants that were casted. I feel like the
artifact combiner should have folded the constant in the trunc before
the custom lowering, but that doesn't happen.
Reverts part of 6524a7a2b9. Since that
commit, the expansion was ignoring the actual save exec register
produced by the instruction, and looking at other instructions. I do
not understand why it was looking at other instructions, but relying
on this scan was wrong.
Fixes verifier errors after SI_IF is tail duplicated, which should be
correct to do. The results were fed into a phi, which was lowered to
the S_MOV_B64_term instructions.
Summary:
The accuracy limit to use rcp is adjusted to 1.0 ulp from 2.5 ulp.
Also, afn instead of arcp is used to allow inaccurate rcp to be used.
Reviewers:
arsenm
Differential Revision: https://reviews.llvm.org/D73588
Summary:
Current implementation of matchSwap in SIShrinkInstructions searches the entire
use_nodbg_operands set to find the possible pattern to generate v_swap instruction.
This approach will lead to a O(N^3) in compile time for SIShrinkInstructions.
But in reality, the matching pattern only exists within nearby instructions in the
same basic block. This work limits the search to a maximum of 16 instructions, and has
a linear compile time comsumption.
Reviewers:
rampitec, arsenm
Differential Revision: https://reviews.llvm.org/D74180
Allows more flexible use of buildMerge in places where
use operands are available as SrcOp since it does not
require explicit conversion to Register.
Simplify code with new buildMerge.
Differential Revision: https://reviews.llvm.org/D74223
We were executing this in a waterfall loop as a placeholder, but this
should really be converted to a MUBUF load. Also execute in a
waterfall loop if the resource isn't an SGPR. This is a case where the
DAG handling was wrong because doing the right thing was too hard.
Currently, this will mishandle 96-bit loads. There's currently no way
to track the original memory size with an MMO, so these loads will be
widened andd the resulting memory size will be 128-bits.
The type passed to lower was invalid, so I'm not sure how this was
even working before. The source and destination type also do not have
to match, so make sure to use the right ones.
Summary: This patch introduces an API for MemOp in order to simplify and tighten the client code.
Reviewers: courbet
Subscribers: arsenm, nemanjai, jvesely, nhaehnle, hiraditya, kbarton, jsji, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73964
We are using countPopulation on a LaneBitmask to determine
a number of registers it covers. This is the assumption which
does not necessarily need to be true. It is not changed but
factored into a single call SIRegisterInfo::getNumCoveredRegs().
Some other places are cleaned up with respect to assumptions
about subreg indexes values and tablegen behavior.
Differential Revision: https://reviews.llvm.org/D74177
Summary:
Current implementation of matchSwap in SIShrinkInstructions searches the entire
use_nodbg_operands set to find the possible pattern to generate v_swap instruction.
This approach will lead to a O(N^3) in compile time for SIShrinkInstructions.
But in reality, the matching pattern only exists within nearby instructions in the
same basic block. This work limits the search to a maximum of 16 instructions, and has
a linear compile time comsumption.
Reviewers:
rampitec, arsenm
Differential Revision: https://reviews.llvm.org/D74180
When multiple instructions are moved into a waterfall loop, it's
possible some of them re-use the same operands. Avoid creating
multiple sequences of readfirstlanes for them. None of the current
uses will hit this, but will be used in a future patch.
Really the intrinsic definition is wrong, but work around this
here. The DAG lowering introduces an MMO. We have to introduce a new
operation to avoid the verifier complaining about the missing mayLoad.
Use cmp ord instead of cmp_class compared to the DAG version for the
nan check, but mostly try to match the existsing pattern.
I think the sign doesn't matter for fract, so we could do a little
better with the source modifier matching.
I think this is also still broken as in D22898, but I'm leaving it
as-is for now while I don't have an SI system to test on.
Rewrite the result register pair into the expected sinigle register
format in the legalizer.
I'm also operating under the assumption that TFE doesn't apply to
stores or atomics, but don't know if this is true or not.
The mask results of these should be uniform. The trickier part is the
dummy booleans used as IR glue need to be treated as divergent. This
should make the divergence analysis results correct for the IR the DAG
is constructed from.
This should allow us to eliminate requiresUniformRegister, which has
an expensive, recursive scan over all users looking for control flow
intrinsics. This should avoid recent compile time regressions.
The 96-bit results need to be widened.
I find the interaction between LegalizerHelper and MIRBuilder somewhat
awkward. The custom legalization is called by the LegalizerHelper, but
then does not have access to the helper. You have to construct a new
helper, which then does not own the MachineIRBuilder, but does modify
it. Maybe custom legalization should be passed the helper?
The adjusted iterator range included the last we just inserted, and
don't want to process. Figure out the new iterator range before
inserting phis. This was a harmless problem, but added an unnecessary
complication for a future patch.
If we have s_pack_* instructions, legalize this to
G_BUILD_VECTOR_TRUNC from s32 elements. This is closer to how how the
s_pack_* instructions really behave.
If we don't have s_pack_ instructions, expand this by creating a merge
to s32 and bitcasting. This expands to the expected bit operations. I
think this eventually should go in a new bitcast legalize action type
in LegalizerHelper.
We already directly emit the shift operations in RegBankSelect for the
vector case. This could possibly be cleaned up, but I also may want to
defer doing this expansion to selection anyway. I'll see about that
when I try to actually match VOP3P instructions.
This breaks the selection of the build_vector since tablegen doesn't
know how to match G_BUILD_VECTOR_TRUNC yet, so just xfail it for now.
scalar_to_vector takes only one argument, not two.
The a16 tests now also check the packing of coordinates into registers
Differential Revision: https://reviews.llvm.org/D73482
This should lower the amount of used registers for gfx9.
I updated some of the changed tests with the update script because
changing them by hand is tedious.
Differential Revision: https://reviews.llvm.org/D73884
The old version might be faster on EG (RECIP_IEEE is Trans only),
but it'd need extra corner case checks.
This gives correct corner case behaviour and saves a register.
Fixes OCL CTS sqrt test (1-thread, scalar) on Turks.
Reviewer: arsenm
Differential Revision: https://reviews.llvm.org/D74017
This was incorrectly rounding up to the next power of 2. v4f32 was
rounding up to v8f32, which was just wrong. There are also v3i16/v3f16
available in MVT, so we don't even need to round the f16 cases
anymore. Additionally, this field is really an EVT so we don't even
need to consider this.
Also switch some asserts to return invalid. We should have an IR
verifier for these intrinsic return types, but for now it's better to
not assert on IR that passes the verifier.
This should also probably be fixed to consider that dmask is really
eliminating some of the loaded components.
https://reviews.llvm.org/D72312 introduced an infinite loop which involves
DAGCombiner::visitFMA and AMDGPUTargetLowering::performFNegCombine.
fma( a, fneg(b), fneg(c) ) => fneg( fma (a, b, c) ) => fma( a, fneg(b), fneg(c) ) ...
This only breaks with types where 'isFNegFree' returns flase, e.g. v4f32.
Reproducing the issue also needs the attribute 'no-signed-zeros-fp-math',
and no source mods allowed on one of the users of the Op.
This fix makes changes to indicate that it is not free to negate a fma if it
has users with source mods.
Differential Revision: https://reviews.llvm.org/D73939
Prepare to accurately track the future denormal-fp-math attribute
changes. The way to actually set these separately is not wired in yet.
This is just a mechanical change, and mostly still assumes the input
and output mode match. This should be refined for some cases. For
example, fcanonicalize lowering should use the flushing variant if
either input or output flushing is enabled
The usage of the Imm out argument from SelectSMRDOffset is pretty
confusing. Stop trying to reject CI immediates in the case where the
offset field can be used. It's not an illegal way to encode the
immediate, so just prefer the better encoding pattern with
AddedComplexity.
We probably don't even really need the different opcodes for the
different offset types anymore, but that will be more work to cleanup.
The SMRD non-buffer load patterns could also use a cleanup to be done
separately.
Start using a new strategy with a combination of merge and unmerges.
This allows scalarizing before lowering, which in cases like
<2 x s128> avoids producing giant illegal shifts.
Summary:
This is patch is part of a series to introduce an Alignment type.
See this thread for context: http://lists.llvm.org/pipermail/llvm-dev/2019-July/133851.html
See this patch for the introduction of the type: https://reviews.llvm.org/D64790
Reviewers: courbet
Subscribers: arsenm, dschuff, jyknight, sdardis, nemanjai, jvesely, nhaehnle, sbc100, jgravelle-google, hiraditya, aheejin, kbarton, fedor.sergeev, asb, rbar, johnrusso, simoncook, sabuasal, niosHD, jrtc27, MaskRay, zzheng, edward-jones, atanasyan, rogfer01, MartinMosbeck, brucehoult, the_o, PkmX, jocewei, jsji, Jim, lenary, s.egerton, pzheng, sameer.abuasal, apazos, luismarques, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73885
Summary:
D68092 introduced a new SIRemoveShortExecBranches optimization pass and
broke some graphics shaders. The problem is that it was removing
branches over KILL pseudo instructions, and the fix is to explicitly
check for that in mustRetainExeczBranch.
Reviewers: critson, arsenm, nhaehnle, cdevadas, hakzsam
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73771
Summary:
Virtual registers that are undef have an empty LiveInterval at this
point, which means beginIndex() and endIndex() cannot be used. We
only need those indices to determine the range in which to scan for
affected other NSA instructions, and undef operands cannot contribute
to that range.
Reviewers: arsenm, rampitec, mareko
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73831
This fixes legalizations of global stores > 128-bits. It seems work is
needed on how this split actually occurs. For example, we get the
right code for s160, with an s128 and s32 load, but get 5 s32 loads
for <5 x s32>.
Summary: This is a first step before changing the types to llvm::Align and introduce functions to ease client code.
Reviewers: courbet
Subscribers: arsenm, sdardis, nemanjai, jvesely, nhaehnle, hiraditya, kbarton, jrtc27, atanasyan, jsji, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73785
Try out using combine definition rules.
This really should be a post-legalizer combine, but the combiner pass
is currently pre-legalize. Most of the target combines are really
post-legalize, so we should probably move the pass.
I believe this also fixes bugs with CI 32-bit handling, which was
incorrectly skipping offsets that look like signed 32-bit values. Also
validate the offsets are dword aligned before folding.
There's not much value to this separate node from the intrinsic. Make
the operand structure the same as the intrinsic, so we can reuse the
same pattern for GlobalISel.
Summary:
Added file headers for files which implement iterative lightweight scheduling
strategies. Which is basically an exercise which I undertook in order to get
used to LLVM development process.
Reviewers: arsenm, vpykhtin, cdevadas
Reviewed By: vpykhtin
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, javed.absar, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73417
On targets that don't have the normal packed f16 layout, handle these
during legalization. Directly modify the register types. We can infer
this was a d16 load based on the mem operand size during selection.
A16 operands should possibly be handled here as well, but don't worry
about that yet.
This trivially avoids violating the constant bus restriction.
Previously this was allowing one SGPR in the first source
operand, which technically also avoided violating this for most
operations (but not for special cases reading vcc).
We do need to write some new, smarter operand folds to pick the
optimal SGPR to use in some kind of post-isel fold, but that's purely
an optimization.
I was originally thinking we would pick which operands should be SGPRs
in RegBankSelect, but I think this isn't really manageable. There
would be additional complexity to handle every G_* instruction, and
then any nontrivial instruction patterns would need to know when to
avoid violating it, which is likely to be very error prone.
I think having all inputs being canonically copies to VGPRs will
simplify the operand folding logic. The current folding we do is
backwards, and only considers one operand at a time, relative to
operands it already has. It therefore poorly handles the case where
there is already a constant bus operand user. If all operands are
copies, it's somewhat simpler to consider all input operands at once
to choose the optimal constant bus user.
Since the failure mode for constant bus violations is now a verifier
error and not an selection failure, this moves towards a place where
we can turn on the fallback mode. The SGPR copy folding optimizations
can be left for later.
Summary:
The code was assuming in a few places that if there was only one exit
from the function that it was a normal return, which is invalid. It
could be an infinite loop, in which case we still need to insert the
usual fake edge so that the null export happens. This fixes shaders that
end with an infinite loop that discards.
Reviewers: arsenm, nhaehnle, critson
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D71192
This is passed to legalizeCustom, but not intrinsic. Also remove the
MRI argument, since you can get that from the MachineIRBuilder.
I'm not sure why MachineIRBuilder has a private observer member, and
this is passed separately.
Summary:
The initialization of RegisterBank needs to be done only once. The
logic of AlreadyInit has data race, use llvm::call_once instead.
This is continuing work of D73587.
Reviewers: arsenm, tstellar, ronlieb, efriedma, apazos, nhaehnle
Reviewed By: nhaehnle
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73604
Fixes selection for scalar G_SMULH/G_UMULH. Also switches to using
tablegen selected add/sub, which switch to the signed version of the
opcode. This matches the current DAG behavior. We can't drop the
manual selection for add/sub yet, because it's still both for VALU
add/sub and for G_PTR_ADD.
Manually select this is as a tablegen workraound. Both SelectionDAG
and GlobalISel end up misplacing the copy to m0 when both instructions
in the output need it. Neither considers that both output instructions
depend on m0. I don't know of any other pattern we need to handle this
case, so it's less effort to just workaround this for now.
Summary:
Due to the fact that kill is just a normal intrinsic, even though it's
supposed to terminate the thread, we can end up with provably infinite
loops that are actually supposed to end successfully. The
AMDGPUUnifyDivergentExitNodes pass breaks up these loops, but because
there's no obvious place to make the loop branch to, it just makes it
return immediately, which skips the exports that are supposed to happen
at the end and hangs the GPU if all the threads end up being killed.
While it would be nice if the fact that kill terminates the thread were
modeled in the IR, I think that the structurizer as-is would make a mess if we
did that when the kill is inside control flow. For now, we just add a null
export at the end to make sure that it always exports something, which fixes
the immediate problem without penalizing the more common case. This means that
we sometimes do two "done" exports when only some of the threads enter the
discard loop, but from tests the hardware seems ok with that.
This fixes dEQP-VK.graphicsfuzz.while-inside-switch with radv.
Reviewers: arsenm, nhaehnle
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70781
SIMachineScheduler uses isHighLatencyInstruction with the same
sematincs, but TargetInstrInfo has virtual isHighLatencyDef
method, so override it instead.
Added FLAT to the list of high latency opcodes and a check for
mayLoad since stores are not technically high latency in terms
of data dependency.
This change did not produce any visible impact on our tests.
Differential Revision: https://reviews.llvm.org/D73582
This should be no problem to support with a pattern, but it turns out
there are just too many yaks to shave. The main problem is in the DAG
emitter, which I have no desire to sink effort into fixing.
If we had a bit to disable patterns in the DAG importer, fixing the
GlobalISelEmitter is more manageable.
Summary:
The code was assuming in a few places that if there was only one exit
from the function that it was a normal return, which is invalid. It
could be an infinite loop, in which case we still need to insert the
usual fake edge so that the null export happens. This fixes shaders that
end with an infinite loop that discards.
Reviewers: arsenm, nhaehnle, critson
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D71192
Summary:
Due to the fact that kill is just a normal intrinsic, even though it's
supposed to terminate the thread, we can end up with provably infinite
loops that are actually supposed to end successfully. The
AMDGPUUnifyDivergentExitNodes pass breaks up these loops, but because
there's no obvious place to make the loop branch to, it just makes it
return immediately, which skips the exports that are supposed to happen
at the end and hangs the GPU if all the threads end up being killed.
While it would be nice if the fact that kill terminates the thread were
modeled in the IR, I think that the structurizer as-is would make a mess if we
did that when the kill is inside control flow. For now, we just add a null
export at the end to make sure that it always exports something, which fixes
the immediate problem without penalizing the more common case. This means that
we sometimes do two "done" exports when only some of the threads enter the
discard loop, but from tests the hardware seems ok with that.
This fixes dEQP-VK.graphicsfuzz.while-inside-switch with radv.
Reviewers: arsenm, nhaehnle
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70781
Summary:
This removes a couple of unnecessary isReg checks, now that
memOpsHaveSameBasePtr can handle FI operands, but is otherwise NFC.
Reviewers: arsenm, rampitec
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73485
This is how it should've been and brings it more in line with
std::string_view. There should be no functional change here.
This is mostly mechanical from a custom clang-tidy check, with a lot of
manual fixups. It uncovers a lot of minor inefficiencies.
This doesn't actually modify StringRef yet, I'll do that in a follow-up.
Summary:
Up to gfx9, writes to vcc_lo and vcc_hi by instructions like
v_readlane and v_readfirstlane do not update vccz to reflect the new
value of vcc. Fix it by reusing part of the existing vccz bug handling
code, which inserts an "s_mov_b64 vcc, vcc" instruction to restore vccz
just before an instruction that needs the correct value.
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D69661
Trivial type predicates should be moved into the tablegen pattern
itself, and not checked inside complex patterns. This eliminates a
redundant complex pattern, and fixes select source modifiers for
GlobalISel.
I have further patches which fully handle select in tablegen and
remove all of the C++ selection, although it requires the ugliness to
support the entire range of legal register types.
We want to have more load/store clustering but we also want
to maintain low register pressure which are oposit targets.
Allow scheduler to reschedule regions without mutations
applied if we hit a register limit.
Differential Revision: https://reviews.llvm.org/D73386
Use intermediate instructions, unlike with buffer stores. This is
necessary because of the need to have an internal way to distinguish
between signed and unsigned extloads. This introduces some duplication
and near duplication with the buffer store selection path. The store
handling should maybe be moved into legalization to match and
eliminate the duplication.
Try to keep simple v2s16 cases as-is. This will more naturally map to
how the VOP3P op_sel modifiers work compared to the expansion
involving bitcasts and bitshifts.
This could maybe try harder with wider source vector types, although
that could be handled with a pre-legalize combine.
This reverts commit fe23ed2c68.
It was never really clear this was responsible for the performance
regressions that caused this to be reverted. It's been a long time,
and we need to have scalar patterns for this to get GlobalISel
working.
For some reason the flat/global atomics end up in the generated
matcher table in a different order from SelectionDAG. Use
AddedComplexity to prefer checking for global atomics first.
Summary:
This is in preparation for getMemOperandsWithOffset returning more base
operands.
Depends on D73455.
Reviewers: arsenm, rampitec
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73456
Summary:
This is in preparation for getMemOperandsWithOffset returning more base
operands.
Depends on D73454.
Reviewers: arsenm, rampitec
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73455
Summary:
This improves merging of sequences like:
store a, ptr + 4
store b, ptr + 8
store c, ptr + 12
store d, ptr + 16
store e, ptr + 20
store f, ptr
Prior to this patch the basic block was scanned in order to find instructions
to merge and the above sequence would be transformed to:
store4 <a, b, c, d>, ptr + 4
store e, ptr + 20
store r, ptr
With this change, we now sort all the candidate merge instructions by their offset,
so instructions are visited in offset order rather than in the order they appear
in the basic block. We now transform this sequnce into:
store4 <f, a, b, c>, ptr
store2 <d, e>, ptr + 16
Another benefit of this change is that since we have sorted the mergeable lists
by offset, we can easily check if an instruction is mergeable by checking the
offset of the instruction that becomes before or after it in the sorted list.
Once we determine an instruction is not mergeable we can remove it from the list
and avoid having to do the more expensive mergeablilty checks.
Reviewers: arsenm, pendingchaos, rampitec, nhaehnle, vpykhtin
Reviewed By: arsenm, nhaehnle
Subscribers: kerbowa, merge_guards_bot, kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D65966
Scheduler sends NumLoads argument into shouldClusterMemOps()
one less the actual cluster length. So for 2 instructions
it will pass just 1. Correct this number.
This is NFC for in tree targets.
Differential Revision: https://reviews.llvm.org/D73292
We are relying on atrificial DAG edges inserted by the
MemOpClusterMutation to keep loads and stores together in the
post-RA scheduler. This does not work all the time since it
allows to schedule a completely independent instruction in the
middle of the cluster.
Removed the DAG mutation and added pass to bundle already
clustered instructions. These bundles are unpacked before the
memory legalizer because it does not work with bundles but also
because it allows to insert waitcounts in the middle of a store
cluster.
Removing artificial edges also allows a more relaxed scheduling.
Differential Revision: https://reviews.llvm.org/D72737
Currently BE allows only a little load narrowing because
of the fear it will produce sub-dword ext loads. However,
we can always allow narrowing if we are shrinking one
multi-dword load to another multi-dword load.
In particular we were unable to reduce s_load_dwordx8 into
s_load_dwordx4 if identity shuffle was used to extract
low 4 dwords.
Differential Revision: https://reviews.llvm.org/D73133
Summary:
Enable the new diveregence analysis by default for AMDGPU.
Resubmit with test updates since GPUDA was causing failures on Windows.
Reviewers: rampitec, nhaehnle, arsenm, thakis
Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D73315
Summary:
This is a follow up on https://reviews.llvm.org/D71473#inline-647262.
There's a caveat here that `Align(1)` relies on the compiler understanding of `Log2_64` implementation to produce good code. One could use `Align()` as a replacement but I believe it is less clear that the alignment is one in that case.
Reviewers: xbolva00, courbet, bollu
Subscribers: arsenm, dylanmckay, sdardis, nemanjai, jvesely, nhaehnle, hiraditya, kbarton, jrtc27, atanasyan, jsji, Jim, kerbowa, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D73099
Summary:
RCP has the accuracy limit. If FDIV fpmath require high accuracy rcp may not
meet the requirement. However, in DAG lowering, fpmath information gets lost,
and thus we may generate either inaccurate rcp related computation or slow code
for fdiv.
In patch implements fdiv optimizations in the AMDGPUCodeGenPrepare, which could
exactly know !fpmath.
FastUnsafeRcpLegal: We determine whether it is legal to use rcp based on
unsafe-fp-math, fast math flags, denormals and fpmath
accuracy request.
RCP Optimizations:
1/x -> rcp(x) when fast unsafe rcp is legal or fpmath >= 2.5ULP with
denormals flushed.
a/b -> a*rcp(b) when fast unsafe rcp is legal.
Use fdiv.fast:
a/b -> fdiv.fast(a, b) when RCP optimization is not performed and
fpmath >= 2.5ULP with denormals flushed.
1/x -> fdiv.fast(1,x) when RCP optimization is not performed and
fpmath >= 2.5ULP with denormals.
Reviewers:
arsenm
Differential Revision:
https://reviews.llvm.org/D71293
Since register classes go up to 1024, 32 elements, all masks bits are
needed and a 32-bit shift by 32 is illegal. We didn't have any
instructions theoretically using a 32 element VGPR before
d1dbb5e471
The other 3-op patterns should also be theoretically handled, but
currently there's a bug in the inferred pattern complexity.
I'm not sure what the error handling strategy should be for potential
constant bus violations. I think the correct strategy is to never
produce mixed SGPR and VGPR operands in a typical VOP instruction,
which will trivially avoid them. However, it's possible to still have
hand written MIR (or erroneously transformed code) with these
operands. When these fold, the restriction will be violated. We
currently don't have any verifiers for reg bank legality. For now,
just ignore the restriction.
It might be worth triggering a DAG fallback on verifier error.
R600 relies on this behaviour.
Fixes: 6e18266aa4 ('Partially revert D61491 "AMDGPU: Be explicit about whether the high-word in SI_PC_ADD_REL_OFFSET is 0"')
Fixes ~100 piglit regressions since 6e18266
Differential Revision: https://reviews.llvm.org/D72991
The pattern is also mishandled by the generated matcher, so workaround
this as in the DAG path.
The existing DAG tests aren't particularly targeted to just this one
intrinsic. These also end up differing in scheduling from SGPR->VGPR
operand constraint copies.
The waterfall utility function blindly inserts a phi for every def in
the loop. We don't need this one to be preserved for every
iteration. Saves an extra phi and copy inside the loop body.
The result and source vector are going to be tied, so these need to be
the same bank.
The inserted value also needs to be broken down based on the result
bank, not the inserted value itself.
Handle dynamic vector extracts that use an index that's an add of a
constant offset into moving the base subregister of the indexing
operation.
Force the add into the loop in regbankselect, which will be recognized
when selected.
DAGCombiner does this, but divisions expanded here miss this
optimization. Since 67aa18f165,
divisions have been expanded here and missed out on this
optimization. Avoids test regressions in a future patch.
The generic BaseMemOpClusterMutation calls into TargetInstrInfo to
analyze the address of each load/store instruction, and again to decide
whether two instructions should be clustered. Previously this had to
represent each address as a single base operand plus a constant byte
offset. This patch extends it to support any number of base operands.
The old target hook getMemOperandWithOffset is now a convenience
function for callers that are only prepared to handle a single base
operand. It calls the new more general target hook
getMemOperandsWithOffset.
The only requirements for the base operands returned by
getMemOperandsWithOffset are:
- they can be sorted by MemOpInfo::Compare, such that clusterable ops
get sorted next to each other, and
- shouldClusterMemOps knows what they mean.
One simple follow-on is to enable clustering of AMDGPU FLAT instructions
with both vaddr and saddr (base register + offset register). I've left
a FIXME in the code for this case.
Differential Revision: https://reviews.llvm.org/D71655
This using the wrong result register, and dropping the result entirely
for v2f16. This would fail to select on the scalar case. I believe it
was also mishandling packed/unpacked subtargets.
Summary:
Check that a s_cbranch_execz is not a loop exit before removing it.
As the pass is generating infinite loops.
Reviewers: cdevadas, arsenm, nhaehnle
Reviewed By: nhaehnle
Subscribers: kzhuravl, jvesely, wdng, yaxunl, tpr, t-tye, hiraditya, kerbowa, llvm-commits, dstuttard, foad
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D72997
The current implementation of skip insertion (SIInsertSkip) makes it a
mandatory pass required for correctness. Initially, the idea was to
have an optional pass. This patch inserts the s_cbranch_execz upfront
during SILowerControlFlow to skip over the sections of code when no
lanes are active. Later, SIRemoveShortExecBranches removes the skips
for short branches, unless there is a sideeffect and the skip branch is
really necessary.
This new pass will replace the handling of skip insertion in the
existing SIInsertSkip Pass.
Differential revision: https://reviews.llvm.org/D68092
In GlobalISel we may in some unfortunate circumstances generate PHIs with
operands that are on separate banks. If-conversion doesn't currently check for
that case and ends up generating a CSEL on AArch64 with incorrect register
operands.
Differential Revision: https://reviews.llvm.org/D72961
Pointers of unrecognized address spaces shoudl be treated as
global-like pointers. Even if loads and stores of them aren't handled,
dumb operations that just operate on the bits should work.
These names have been changed from CamelCase to camelCase, but there were
many places (comments mostly) that still used the old names.
This change is NFC.
There's no reason to introduce a new, unnaturally sized value
here. This has a chance to produce worse code with
legalization. Avoids regression in a future patch.
a785209bc2 switched to using a pseudos instead of manually tying
operands on the regular instruction. The VGPR indexing mode path
should have the same problems that change attempted to avoid, so these
should use the same strategy.
Use a single pseudo for the VGPR indexing mode and movreld paths, and
expand it based on the subtarget later. These have essentially the
same constraints, reading the index from m0.
Switch from using an offset to the subregister index directly, instead
of computing an offset and re-adding it back. Also add missing pseudos
for existing register class sizes.
Document some high level strategies that should be used for register
bank selection. The constant bus restriction section hasn't actually
been implemented yet.
Except AMDGPU/R600RegisterInfo (a bunch of MIR tests seem to have
problems), every target overrides it with true. PostMachineScheduler
requires livein information. Not providing it can cause assertion
failures in ScheduleDAGInstrs::addSchedBarrierDeps().
Currently the lowering for i16 image coordinates asserts on gfx10. I'm
somewhat confused by this though. The feature is missing from the
gfx10 feature lists, but the a16 bit appears to be present in the
manual for MIMG instructions.
This case can be handled as a regular selection pattern, so move it
out of the weird post-isel folding code which doesn't have an exactly
equivalent place in GlobalISel.
I think it doesn't make much sense to do this optimization here
though, and it would be more useful in instcombine. There's not really
any new information that will be gained during lowering since these
inputs were known from the beginning.
This does produce slightly different code. Now a unique IMPLICIT_DEF
is emitted for each of the implicit_def operands, rather than reusing
the same one.
I'm mildly worried about potentially reordering exp/exp_done with
IntrWriteMem on the intrinsic.
Requires hacking out the illegal type on SI, so manually select that
case during lowering.
Summary:
InlineResult is used both in APIs assessing whether a call site is
inlinable (e.g. llvm::isInlineViable) as well as in the function
inlining utility (llvm::InlineFunction). It means slightly different
things (can/should inlining happen, vs did it happen), and the
implicit casting may introduce ambiguity (casting from 'false' in
InlineFunction will default a message about hight costs,
which is incorrect here).
The change renames the type to a more generic name, and disables
implicit constructors.
Reviewers: eraman, davidxl
Reviewed By: davidxl
Subscribers: kerbowa, arsenm, jvesely, nhaehnle, eraman, hiraditya, haicheng, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D72744
Bitcast only really applies between scalars and vectors. Implement as
an unmerge and remerge. The test needs to tolerate failure since one
of the unmerges currently fails to legalize.
The 16 bank LDS case is complicated due to using multiple
instructions. If I attempt to write a pattern for it, the generated
selector incorrectly places the copy to m0 after the first
instruction, so that needs to be separately addressed.
Also fix not gluing the copy to m0 to the second operation in the
second half of the 16 bank lowering.
The current implementation of skip insertion (SIInsertSkip) makes it a
mandatory pass required for correctness. Initially, the idea was to
have an optional pass. This patch inserts the s_cbranch_execz upfront
during SILowerControlFlow to skip over the sections of code when no
lanes are active. Later, SIRemoveShortExecBranches removes the skips
for short branches, unless there is a sideeffect and the skip branch is
really necessary.
This new pass will replace the handling of skip insertion in the
existing SIInsertSkip Pass.
Differential revision: https://reviews.llvm.org/D68092
Summary:
For builds with LLVM_BUILD_LLVM_DYLIB=ON and BUILD_SHARED_LIBS=OFF
this change makes all symbols in the target specific libraries hidden
by default.
A new macro called LLVM_EXTERNAL_VISIBILITY has been added to mark symbols in these
libraries public, which is mainly needed for the definitions of the
LLVMInitialize* functions.
This patch reduces the number of public symbols in libLLVM.so by about
25%. This should improve load times for the dynamic library and also
make abi checker tools, like abidiff require less memory when analyzing
libLLVM.so
One side-effect of this change is that for builds with
LLVM_BUILD_LLVM_DYLIB=ON and LLVM_LINK_LLVM_DYLIB=ON some unittests that
access symbols that are no longer public will need to be statically linked.
Before and after public symbol counts (using gcc 8.2.1, ld.bfd 2.31.1):
nm before/libLLVM-9svn.so | grep ' [A-Zuvw] ' | wc -l
36221
nm after/libLLVM-9svn.so | grep ' [A-Zuvw] ' | wc -l
26278
Reviewers: chandlerc, beanz, mgorny, rnk, hans
Reviewed By: rnk, hans
Subscribers: merge_guards_bot, luismarques, smeenai, ldionne, lenary, s.egerton, pzheng, sameer.abuasal, MaskRay, wuzish, echristo, Jim, hiraditya, michaelplatings, chapuni, jholewinski, arsenm, dschuff, jyknight, dylanmckay, sdardis, nemanjai, jvesely, javed.absar, sbc100, jgravelle-google, aheejin, kbarton, fedor.sergeev, asb, rbar, johnrusso, simoncook, apazos, sabuasal, niosHD, jrtc27, zzheng, edward-jones, mgrang, atanasyan, rogfer01, MartinMosbeck, brucehoult, the_o, PkmX, jocewei, kristina, jsji, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D54439
Summary:
- `dead-mi-elimination` assumes MIR in the SSA form and cannot be
arranged after phi elimination or DeSSA. It's enhanced to handle the
dead register definition by skipping use check on it. Once a register
def is `dead`, all its uses, if any, should be `undef`.
- Re-arrange the DIE in RA phase for AMDGPU by placing it directly after
`detect-dead-lanes`.
- Many relevant tests are refined due to different register assignment.
Reviewers: rampitec, qcolombet, sunfish
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D72709
This change allows to model the height of the instruction
within a bundle for latency adjustment purposes.
Differential Revision: https://reviews.llvm.org/D72669
We do not have InstrItinerary so generic getInstLatency() was always
defaulting to return 1 cycle. We need to use TargetSchedModel instead
to compute an instruction's latency.
Differential Revision: https://reviews.llvm.org/D72655
The current users of the waterfall loop utility functions do not make
use of the restored original insert point. The insertion is either
done, or they set the insert point somewhere else. A future change
will want to insert instructions after the waterfall loop, but
figuring out the point after the loop is more difficult than ensuring
the insert point is there after the loop.
The branch target needs to be changed depending on whether there is an
unconditional branch or not.
Loops also need to be similarly fixed, but compiling a simple testcase
end to end requires another set of patches that aren't upstream yet.
The argument is llvm::null() everywhere except llvm::errs() in
llvm-objdump in -DLLVM_ENABLE_ASSERTIONS=On builds. It is used by no
target but X86 in -DLLVM_ENABLE_ASSERTIONS=On builds.
If we ever have the needs to add verbose log to disassemblers, we can
record log with a member function, instead of passing it around as an
argument.
Summary:
- SI Whole Quad Mode phase is replacing WQM pseudo instructions with v_mov instructions.
While this is necessary for the special handling of moving results out of WWM live ranges,
it is not necessary for WQM live ranges. The result is a v_mov from a register to itself after every
WQM operation. This change uses a COPY psuedo in these cases, which allows the register
allocator to coalesce the moves away.
Reviewers: tpr, dstuttard, foad, nhaehnle
Reviewed By: nhaehnle
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D71386
This reverts the AMDGPU DAG mutation implemented in D72487 and gives
a more general way of adjusting BUNDLE operand latency.
It also replaces FixBundleLatencyMutation with adjustSchedDependency
callback in the AMDGPU, fixing not only successor latencies but
predecessors' as well.
Differential Revision: https://reviews.llvm.org/D72535
If an SGPR vector is indexed with a VGPR, the actual indexing will be
done on the SGPR and produce an SGPR. A copy needs to be inserted
inside the waterwall loop to the VGPR result.
The current implementation assumes there is an instruction associated
with the transform, but this is not the case for
timm/TargetConstant/immarg values. These transforms should directly
operate on a specific MachineOperand in the source
instruction. TableGen would assert if you attempted to define an
equivalent GISDNodeXFormEquiv using timm when it failed to find the
instruction matcher.
Specially recognize SDNodeXForms on timm, and pass the operand index
to the render function.
Ideally this would be a separate render function type that looks like
void renderFoo(MachineInstrBuilder, const MachineOperand&), but this
proved to be somewhat mechanically painful. Add an optional operand
index which will only be passed if the transform should only look at
the one source operand.
Theoretically it would also be possible to only ever pass the
MachineOperand, and the existing renderers would check the parent. I
think that would be somewhat ugly for the standard usage which may
want to inspect other operands, and I also think MachineOperand should
eventually not carry a pointer to the parent instruction.
Use it in one sample pattern. This isn't a great example, since the
transform exists to satisfy DAG type constraints. This could also be
avoided by just changing the MachineInstr's arbitrary choice of
operand type from i16 to i32. Other patterns have nontrivial uses, but
this serves as the simplest example.
One flaw this still has is if you try to use an SDNodeXForm defined
for imm, but the source pattern uses timm, you still see the "Failed
to lookup instruction" assert. However, there is now a way to avoid
it.
Only PPC seems to be using it, and only checks some simple cases and
doesn't distinguish between FP. Just switch to using LLT to simplify
use from GlobalISel.
When these arguments are broken down by the EVT based callbacks, the
pointer information is lost. Hack around this by coercing the register
types to be the expected pointer element type when building the
remerge operations.
This should be legal, but will require future selection work. 16-bit
shift amounts were already removed from being legal, but this didn't
adjust the transformation rules.
This doesn't enable any new imports yet, but moves the fmed patterns
from failing on this to hitting the "complex suboperand referenced
more than once" limitation in tablegen.
`APFLoat::convertFromString` returns `Expected` result, which must be
"checked" if the LLVM_ENABLE_ABI_BREAKING_CHECKS preprocessor flag is
set.
To mark an `Expected` result as "checked" we must consume the `Error`
within.
In many cases, we are only interested in knowing if an error occured,
without the need to examine the error info. This is achieved, easily,
with the `errorToBool()` API.
printInst prints a branch/call instruction as `b offset` (there are many
variants on various targets) instead of `b address`.
It is a convention to use address instead of offset in most external
symbolizers/disassemblers. This difference makes `llvm-objdump -d`
output unsatisfactory.
Add `uint64_t Address` to printInst(), so that it can pass the argument to
printInstruction(). `raw_ostream &OS` is moved to the last to be
consistent with other print* methods.
The next step is to pass `Address` to printInstruction() (generated by
tablegen from the instruction set description). We can gradually migrate
targets to print addresses instead of offsets.
In any case, downstream projects which don't know `Address` can pass 0 as
the argument.
Reviewed By: jhenderson
Differential Revision: https://reviews.llvm.org/D72172
We have a lot of complex pattern variants that just set the source
modifiers that are really handled, and then set the output modifiers
to 0. We're unlikely to ever match output modifiers from the use
instruction side, and we already match clamp/omod in a separate pass.
This solves selection failures with generated selection patterns,
which would fail due to inferring the SGPR reg bank for virtual
registers with a set register class instead of VCC bank. Use
instruction selection would constrain the virtual register to a
specific class, so when the def was selected later the bank no longer
was set to VCC.
Remove the SCC reg bank. SCC isn't directly addressable, so it
requires copying from SCC to an allocatable 32-bit register during
selection, so these might as well be treated as 32-bit SGPR values.
Now any scalar boolean value that will produce an outupt in SCC should
be widened during RegBankSelect to s32. Any s1 value should be a
vector boolean during selection. This makes the vcc register bank
unambiguous with a normal SGPR during selection.
Summary of how this should now work:
- G_TRUNC is always a no-op, and never should use a vcc bank result.
- SALU boolean operations should be promoted to s32 in RegBankSelect
apply mapping
- An s1 value means vcc bank at selection. The exception is for
legalization artifacts that use s1, which are never VCC. All other
contexts should infer the VCC register classes for s1 typed
registers. The LLT for the register is now needed to infer the
correct register class. Extensions with vcc sources should be
legalized to a select of constants during RegBankSelect.
- Copy from non-vcc to vcc ensures high bits of the input value are
cleared during selection.
- SALU boolean inputs should ensure the inputs are 0/1. This includes
select, conditional branches, and carry-ins.
There are a few somewhat dirty details. One is that G_TRUNC/G_*EXT
selection ignores the usual register-bank from register class
functions, and can't handle truncates with VCC result banks. I think
this is OK, since the artifacts are specially treated anyway. This
does require some care to avoid producing cases with vcc. There will
also be no 100% reliable way to verify this rule is followed in
selection in case of register classes, and violations manifests
themselves as invalid copy instructions much later.
Standard phi handling also only considers the bank of the result
register, and doesn't insert copies to make the source banks
match. This doesn't work for vcc, so we have to manually correct phi
inputs in this case. We should add a verifier check to make sure there
are no phis with mixed vcc and non-vcc register bank inputs.
There's also some duplication with the LegalizerHelper, and some code
which should live in the helper. I don't see a good way to share
special knowledge about what types to use for intermediate operations
depending on the bank for example. Using the helper to replace
extensions with selects also seems somewhat awkward to me.
Another issue is there are some contexts calling
getRegBankFromRegClass that apparently don't have the LLT type for the
register, but I haven't yet run into a real issue from this.
This also introduces new unnecessary instructions in most cases, since
we don't yet try to optimize out the zext when the source is known to
come from a compare.
This would complain about invalid legalizer rules otherwise.
Mark some operations as unsupported for AMDGPU. This currently seems
to produce the same legalize error as when no rules are defined, but
eventually this should produce a proper user facing error.
The existing test only covered one case for r600. The use of
mul_legacy also looks suspicious to me, but leave it for now. The
patterns are also not making use of source modifiers.
Fangrui Song