Save the instruction list of a block before selecting banks.
This allows to cope with moved instructions, even if they are reordered
or splitted into multiple basic blocks.
Differential Revision: https://reviews.llvm.org/D111223
- When an unconditional branch is expanded into an indirect branch, if
there is no scavenged register, an SGPR pair needs spilling to enable
the destination PC calculation. In addition, before jumping into the
destination, that clobbered SGPR pair need restoring.
- As SGPR cannot be spilled to or restored from memory directly, the
spilling/restoring of that SGPR pair reuses the regular SGPR spilling
support but without spilling it into memory. As that spilling and
restoring points are fully controlled, we only need to spill that SGPR
into the temporary VGPR, which needs spilling into its emergency slot.
- The target-specific hook is revised to take additional restore block,
where the restoring code is filled. After that, the relaxation will
place that restore block directly before the destination block and
insert an unconditional branch in any fall-through block into the
destination block.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D106449
PromoteIntRes_MLOAD always sets the extension type to `EXTLOAD`, which
results in a sign-extended load. If the type returned by getExtensionType()
for the load being promoted is something other than `NON_EXTLOAD`, we
should instead pass this to getMaskedLoad() as the extension type.
Reviewed By: CarolineConcatto
Differential Revision: https://reviews.llvm.org/D112320
Widens the result and first input vector because they have the same size.
The subvector to be inserted is widened in the operand widen function.
Differential Revision: https://reviews.llvm.org/D112187
Add an optional bool RemoveDeadValNo argument to the
removeSegment(iterator) overload, for consistency with the other
overloads. This gives clients a way to remove dead valnos while also
getting an updated iterator returned (in the manner of vector::erase).
Use this to clean up some inefficient code in
LiveIntervals::repairOldRegInRange. NFC.
Differential Revision: https://reviews.llvm.org/D110560
ARM EHABI[1] specifies the __cxa_end_cleanup to be called after cleanup.
It will call the UnwindResume.
__cxa_begin_cleanup will be called from libcxxabi while __cxa_end_cleanup is never called.
This will trigger a termination when a foreign exception is processed while UnwindResume is called
because the global state will be wrong due to the missing __cxa_end_cleanup call.
Additional test here: D109856
[1] https://github.com/ARM-software/abi-aa/blob/main/ehabi32/ehabi32.rst#941compiler-helper-functions
Reviewed By: logan
Differential Revision: https://reviews.llvm.org/D111703
getShiftAmountTyForConstant is a special helper that changes the
shift amount to i32 if the type chosen by
TargetLowering::getShiftAmountTy can't represent all possible values.
This is needed to satisfy an assert in SelectionDAG::getNode.
It requires additional consideration to know when this helper should be used.
I'm not sure that we are always using it when we should.
This patch merges the getShiftAmountTyForConstant handling into
TargetLowering::getShiftAmountTy so we don't need to think about it
anymore.
Technically this may slightly increase compile times since the majority
of callers of getShiftAmountTy won't need this. Hopefully, this isn't
an issue in practice.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D112469
There are a few STL containers hanging around that can become DenseMaps,
SmallVectors and similar. This recovers a modest amount of compile time
performance.
While I'm here, adjust the bit layout of ValueIDNum: this was always
supposed to act like a value type, however it seems that clang doesn't
compile the comparison functions to act that way. Add a uint64_t to a
union that explicitly aliases the bitfields, so that we can compare the
whole value as a single integer.
Differential Revision: https://reviews.llvm.org/D112333
This patch is like D111627 -- instead of calculating IDF for every location
on the stack, only do it for the smallest units of interference, and copy
the PHIs for those units to any aliases.
The test added runs placeMLocPHIs directly, and tests that:
* A def of the lower 8 bits of a stack slot causes all aliasing regs to
have PHIs placed,
* It doesn't cause the equivalent location to x86's $ah, which isn't
aliased, to have a PHI placed.
Differential Revision: https://reviews.llvm.org/D112324
During register allocation, some instructions can have stack spills fused
into them. It means that when vregs are allocated on the stack we can
convert:
SETCCr %0
DBG_VALUE %0
to
SETCCm %stack.0
DBG_VALUE %stack.0
Unfortunately instruction referencing finds this harder: a store to the
stack doesn't have a specific operand number, therefore we don't substitute
the old operand for a new operand, and the location is dropped. This patch
implements a solution: just recognise the memory operand attached to an
instruction with a Special Number (TM), and record a substitution between
the old value and the new one.
This patch adds substitution code to InlineSpiller to record such fused
spills, and tracking in InstrRefBasedLDV to recognise such values, and
produce the value numbers for them. Everything to do with the movement of
stack-defined values is already handled in InstrRefBasedLDV.
Differential Revision: https://reviews.llvm.org/D111317
This patch swaps two lines -- the CurSucc reference can be invalidated
by the call to DFS.push_back, therefore that should happen last. The
usual hat-tip to asan for catching this.
This patch also swaps an ealier call to ToAdd.insert and DFS.push_back,
where a stable iterator (from successors()) is being used. This isn't
strictly necessary, but is good for consistency and avoiding readers
asking themselves why the two code portions have a different order.
(i8 X ^ 128) & (i8 X s>> 7) --> usubsat X, 128
As suggested in D112085, we can substitute 'xor' with 'add'
in this pattern, and it is logically equivalent:
https://alive2.llvm.org/ce/z/eJtWWC
We canonicalize to 'xor' in IR, but SDAG does not do that
(and it probably should not - https://llvm.org/PR52267 ), so
it is possible to see either pattern in codegen. Note that
'sub' is a another potential pattern, but that is
canonicalized to 'add' in DAGCombiner, so we don't need to
worry about that variation.
Differential Revision: https://reviews.llvm.org/D112377
This removes a condition and the corresponding FIXME comment, because
the Hexagon assertion it refers to has apparently been fixed, probably
by D76134.
NFCI. This just gives targets the opportunity to adjust latencies that
were set to 0 by the generic code because they involve "implicit pseudo"
operands.
Differential Revision: https://reviews.llvm.org/D112306
Sometimes we generate code that writes to a subregister, then spills /
restores a super-register to the stack, for example:
$eax = MOV32ri 0
MOV64mr $rsp, 1, $noreg, 16, $noreg, $rax
$rcx = MOV64rm $rsp, 1, $noreg, 8, $noreg
This patch takes a different approach: it adds another index to
MLocTracker that identifies a size/offset within a stack slot. A location
on the stack is then a pari of {FrameIndex, SlotNum}. Spilling and
restoring now involves pairing up the src/dest register numbers, and the
dest/src stack position to be transferred to/from. Location coverage
improves as a result, compile-time performance decreases, alas.
One limitation is that if a PHI occurs inside a stack slot:
DBG_PHI %stack.0, 1
We don't know how large the resulting value is, and so might have
difficulty picking which value to use. DBG_PHI might need to be augmented
in the future with such a size.
Unit tests added ensure that spills and restores correctly transfer to
positions in the Location => Value map, and that different register classes
written to the stack will correctly clobber all other positions in the
stack slot.
Differential Revision: https://reviews.llvm.org/D112133
We might be promoting a large non-power of 2 type and the new type
may need to be split. Once we split it we may have a ctlz/cttz/ctpop
instruction for the split type.
I'm also concerned that we may create large shifts with shift amounts
that are too small.
EXTRACT_SUBVECTOR indices are always constant, we don't need to check for ConstantSDNode, we should just use getConstantOperandVal which will assert for the constant.
This patch adds some unit tests for the machine-location transfer-function
building parts of InstrRefBasedLDV: i.e., test that if we feed some MIR
into the transfer-function building code, does it create the correct
transfer function.
There are a number of minor defects that get corrected in the process:
* The unit test was selecting the x86 (i.e. 32 bit) backend rather than
x86_64's 64 bit backend,
* COPY instructions weren't actually having their subregister values
correctly represented in the transfer function. Subregisters were being
defined by the COPY, rather than taking the value in the source register.
* SP aliases were at risk of being clobbered, if an SP subregister was
clobbered.
Differential Revision: https://reviews.llvm.org/D112006
Instead of returning a bool to indicate success and a separate
SDValue, return the SDValue and have the callers check if it is
null.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D112331
Expanding these requires multiple constants. If we promote during type
legalization when they'll end up getting expanded in LegalizeDAG, we'll
use larger constants. These constants may be harder to materialize.
For example, 64-bit constants on 64-bit RISCV are very expensive.
This is similar to what has already been done to BSWAP and BITREVERSE.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D112268
There is no need to return a bool and have an SDValue output
parameter. Just return the SDValue and let the caller check if it
is null.
I have another patch to add more callers of these so I thought
I'd clean up the interface first.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D112267
By expanding early it allows the shifts to be custom lowered in
LegalizeVectorOps. Then a DAG combine is able to run on them before
LegalizeDAG handles the BUILD_VECTORS for the masks used.
v16Xi8 shift lowering on X86 requires a mask to be applied to a v8i16
shift. The BITREVERSE expansion applied an AND mask before SHL ops and
after SRL ops. This was done to share the same mask constant for both shifts.
It looks like this patch allows DAG combine to remove the AND mask added
after v16i8 SHL by X86 lowering. This maintains the mask sharing that
BITREVERSE was trying to achieve. Prior to this patch it looks like
we kept the mask after the SHL instead which required an extra constant
pool or a PANDN to invert it.
This is dependent on D112248 because RISCV will end up scalarizing the BSWAP
portion of the BITREVERSE expansion if we don't disable BSWAP scalarization in
LegalizeVectorOps first.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D112254
It's better to do the ands, shifts, ors in the vector domain than
to scalarize it and do those operations on each element.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D112248
(i8 X ^ 128) & (i8 X s>> 7) --> usubsat X, 128
I haven't found a generalization of this identity:
https://alive2.llvm.org/ce/z/_sriEQ
Note: I was actually looking at the first form of the pattern in that link,
but that's part of a long chain of potential missed transforms in codegen
and IR....that I hope ends here!
The predicates for when this is profitable are a bit tricky. This version of
the patch excludes multi-use but includes custom lowering (as opposed to
legal only).
On x86 for example, we have custom lowering for some vector types, and that
uses umax and sub. So to enable that fold, we need add use checks to avoid
regressions. Even with legal-only lowering, we could see code with extra
reg move instructions for extra uses, so that constraint would have to be
eased very carefully to avoid penalties.
Differential Revision: https://reviews.llvm.org/D112085
When splitting a masked load, `GetDependentSplitDestVTs` is used to get the
MemVTs of the high and low parts. If the masked load is extended, this
may return VTs with different element types which are used to create the
high & low masked load instructions.
This patch changes `GetDependentSplitDestVTs` to ensure we return VTs with
the same element type.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D111996
With unoptimized code, we may see lots of stores and spend too much time in mergeTruncStores.
Fixes PR51827.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D111596
MachineLoop::isLoopInvariant() returns false for all VALU
because of the exec use. Check TII::isIgnorableUse() to
allow hoisting.
That unfortunately results in higher register consumption
since MachineLICM does not adequately estimate pressure.
Therefor I think it shall only be enabled after D107677 even
though it does not depend on it.
Differential Revision: https://reviews.llvm.org/D107859
As discussed in:
* https://reviews.llvm.org/D94166
* https://lists.llvm.org/pipermail/llvm-dev/2020-September/145031.html
The GlobalIndirectSymbol class lost most of its meaning in
https://reviews.llvm.org/D109792, which disambiguated getBaseObject
(now getAliaseeObject) between GlobalIFunc and everything else.
In addition, as long as GlobalIFunc is not a GlobalObject and
getAliaseeObject returns GlobalObjects, a GlobalAlias whose aliasee
is a GlobalIFunc cannot currently be modeled properly. Creating
aliases for GlobalIFuncs does happen in the wild (e.g. glibc). In addition,
calling getAliaseeObject on a GlobalIFunc will currently return nullptr,
which is undesirable because it should return the object itself for
non-aliases.
This patch refactors the GlobalIFunc class to inherit directly from
GlobalObject, and removes GlobalIndirectSymbol (while inlining the
relevant parts into GlobalAlias and GlobalIFunc). This allows for
calling getAliaseeObject() on a GlobalIFunc to return the GlobalIFunc
itself, making getAliaseeObject() more consistent and enabling
alias-to-ifunc to be properly modeled in the IR.
I exercised some judgement in the API clients of GlobalIndirectSymbol:
some were 'monomorphized' for GlobalAlias and GlobalIFunc, and
some remained shared (with the type adapted to become GlobalValue).
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D108872
This patch fixes a crash when despeculating ctlz/cttz intrinsics with
scalable-vector types. It is not safe to speculatively get the size of
the vector type in bits in case the vector type is not a fixed-length type. As
it happens this isn't required as vector types are skipped anyway.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D112141
Our fallback expansion for CTLZ/CTTZ relies on CTPOP. If CTPOP
isn't legal or custom for a vector type we would scalarize the
CTLZ/CTTZ. This is different than CTPOP itself which would use a
vector expansion.
This patch teaches expandCTLZ/CTTZ to rely on the vector CTPOP
expansion instead of scalarizing. To do this I had to add additional
checks to make sure the operations used by CTPOP expansions are all
supported. Some of the operations were already needed for the CTLZ/CTTZ
expansion.
This is a huge improvement to the RISCV which doesn't have a scalar
ctlz or cttz in the base ISA.
For WebAssembly, I've added Custom lowering to keep the scalarizing
behavior. I've also extended the scalarizing to CTPOP.
Differential Revision: https://reviews.llvm.org/D111919
Here's another performance patch for InstrRefBasedLDV: rather than
processing all variable values in a scope at a time, instead, process one
variable at a time. The benefits are twofold:
* It's easier to reason about one variable at a time in your mind,
* It improves performance, apparently from increased locality.
The downside is that the value-propagation code gets indented one level
further, plus there's some churn in the unit tests.
Differential Revision: https://reviews.llvm.org/D111799
When inserting a scalable subvector into a scalable vector through
the stack, the index to store to needs to be scaled by vscale.
Before this patch, that didn't yet happen, so it would generate the
wrong offset, thus storing a subvector to the incorrect address
and overwriting the wrong lanes.
For some insert:
nxv8f16 insert_subvector(nxv8f16 %vec, nxv2f16 %subvec, i64 2)
The offset was not scaled by vscale:
orr x8, x8, #0x4
st1h { z0.h }, p0, [sp]
st1h { z1.d }, p1, [x8]
ld1h { z0.h }, p0/z, [sp]
And is changed to:
mov x8, sp
st1h { z0.h }, p0, [sp]
st1h { z1.d }, p1, [x8, #1, mul vl]
ld1h { z0.h }, p0/z, [sp]
Differential Revision: https://reviews.llvm.org/D111633
Inspired by D111968, provide a isNegatedPowerOf2() wrapper instead of obfuscating code with (-Value).isPowerOf2() patterns, which I'm sure are likely avenues for typos.....
Differential Revision: https://reviews.llvm.org/D111998
This is purely a performance patch: InstrRefBasedLDV used to use three
DenseMaps to store variable values, two for long term storage and one as a
working set. This patch eliminates the working set, and updates the long
term storage in place, thus avoiding two DenseMap comparisons and two
DenseMap assignments, which can be expensive.
Differential Revision: https://reviews.llvm.org/D111716
This field gets assigned when the relevant object starts being used; but it
remains uninitialized beforehand. This risks introducing hard-to-detect
bugs if something changes, so zero-initialize the field.
When compiling for the RWPI relocation model the debug information is wrong:
* the debug location is described as { DW_OP_addr Var }
instead of { DW_OP_constNu Var DW_OP_bregX 0 DW_OP_plus }
* the relocation type is R_ARM_ABS32 instead of R_ARM_SBREL32
Differential Revision: https://reviews.llvm.org/D111404
This is NFC-intended for the callers. Posting in case there are
other potential users that I missed.
I would also use this from VectorCombine in a patch for:
https://llvm.org/PR52178 ( D111901 )
Differential Revision: https://reviews.llvm.org/D111891
gcc11 warns that this counter causes a signed/unsigned comaprison when it's
later compared with a SmallVector::difference_type. gcc appears to be
correct, clang does not warn one way or the other.
TargetPassConfig::addPass takes a "bool verifyAfter" argument which lets
you skip machine verification after a particular pass. Unfortunately
this is used in generic code in TargetPassConfig itself to skip
verification after a generic pass, only because some previous target-
specific pass damaged the MIR on that specific target. This is bad
because problems in one target cause lack of verification for all
targets.
This patch replaces that mechanism with a new MachineFunction property
called "FailsVerification" which can be set by (usually target-specific)
passes that are known to introduce problems. Later passes can reset it
again if they are known to clean up the previous problems.
Differential Revision: https://reviews.llvm.org/D111397
The process of widening simple vector loads attempts to use a load of a
wider vector type if the original load is sufficiently aligned to avoid
memory faults.
However this optimization is only legal when performed on fixed-length
vector types. For scalable vector types this is invalid (unless vscale
happens to be 1).
This patch does increase the likelihood of compiler crashes (from
`FindMemType` failing to find a suitable type) but this now better
matches how widening non-simple loads, insufficiently-aligned loads, and
scalable-vector stores are handled.
Patches will be introduced later by which loads and stores can be
widened on targets with support for masked or predicated operations.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D111885
We did a experiment and observed dramatic decrease on compilation time which spent on clearing kill flags.
Before:
Number of BasicBlocks:33357
Number of Instructions:162067
Number of Cleared Kill Flags:32869
Time of handling kill flags(ms):1.607509e+05
After:
Number of BasicBlocks:33357
Number of Instructions:162067
Number of Cleared Kill Flags:32869
Time of handling kill flags:3.987371e+03
Reviewed By: MatzeB
Differential Revision: https://reviews.llvm.org/D111688
After D80369, the retainedTypes in CU's should not have any subprograms
so we should not handle that case when emitting debug info.
Differential Revision: https://reviews.llvm.org/D111593
This patch is very similar to D110173 / a3936a6c19, but for variable
values rather than machine values. This is for the second instr-ref
problem, calculating the correct variable value on entry to each block.
The previous lattice based implementation was broken; we now use LLVMs
existing PHI placement utilities to work out where values need to merge,
then eliminate un-necessary ones through value propagation.
Most of the deletions here happen in vlocJoin: it was trying to pick a
location for PHIs to happen in, badly, leading to an infinite loop in the
MIR test added, where it would repeatedly switch between register
locations. The new approach is simpler: either PHIs can be eliminated, or
they can't, and the location of the value is a different problem.
Various bits and pieces move to the header so that they can be tested in
the unit tests. The DbgValue class grows a "VPHI" kind to represent
variable value PHIS that haven't been eliminated yet.
Differential Revision: https://reviews.llvm.org/D110630
Some functions get opted out of instruction referencing if they're being
compiled with no optimisations, however the LiveDebugValues pass picks one
implementation and then sticks with it through the rest of compilation.
This leads to a segfault if we encounter a function that doesn't use
instr-ref (because it's optnone, for example), but we've already decided
to use InstrRefBasedLDV which expects to be passed a DomTree.
Solution: keep both implementations around in the pass, and pick whichever
one is appropriate to the current function.
In D110173 we start using the existing LLVM IDF calculator to place PHIs as
we reconstruct an SSA form of machine-code program. Sadly that's slower
than the old (but broken) way, this patch attempts to recover some of that
performance.
The key observation: every time we def a register, we also have to def it's
register units. If we def'd $rax, in the current implementation we
independently calculate PHI locations for {al, ah, ax, eax, hax, rax}, and
they will all have the same PHI positions. Instead of doing that, we can
calculate the PHI positions for {al, ah} and place PHIs for any aliasing
registers in the same positions. Any def of a super-register has to def
the unit, and vice versa, so this is sound. It cuts down the SSA placement
we need to do significantly.
This doesn't work for stack slots, or registers we only ever read, so place
PHIs normally for those. LiveDebugValues choses to ignore writes to SP at
calls, and now have to ignore writes to SP register units too.
Differential Revision: https://reviews.llvm.org/D111627
Old versions of gcc want template specialisations to happen within the
namespace where the template lives; this is still present in gcc 5.1, which
we officially support, so it has to be worked around.
InstrRefBasedLDV used to try and determine which values are in which
registers using a lattice approach; however this is hard to understand, and
broken in various ways. This patch replaces that approach with a standard
SSA approach using existing LLVM utilities. PHIs are placed at dominance
frontiers; value propagation then eliminates un-necessary PHIs.
This patch also adds a bunch of unit tests that should cover many of the
weirder forms of control flow.
Differential Revision: https://reviews.llvm.org/D110173
This patch adds patterns to match the following with INC/DEC:
- @llvm.aarch64.sve.cnt[b|h|w|d] intrinsics + ADD/SUB
- vscale + ADD/SUB
For some implementations of SVE, INC/DEC VL is not as cheap as ADD/SUB and
so this behaviour is guarded by the "use-scalar-inc-vl" feature flag, which for SVE
is off by default. There are no known issues with SVE2, so this feature is
enabled by default when targeting SVE2.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D111441
This makes Wasm EH work with dynamic linking. So far we were only able
to handle destructors, which do not use any tags or LSDA info.
1. This uses `TargetExternalSymbol` for `GCC_except_tableN` symbols,
which points to the address of per-function LSDA info. It is more
convenient to use than `MCSymbol` because it can take additional
target flags.
2. When lowering `wasm_lsda` intrinsic, if PIC is enabled, make the
symbol relative to `__memory_base` and generate the `add` node. If
PIC is disabled, continue to use the absolute address.
3. Make tag symbols (`__cpp_exception` and `__c_longjmp`) undefined in
the backend, because it is hard to make it work with dynamic
linking's loading order. Instead, we make all tag symbols undefined
in the LLVM backend and import it from JS.
4. Add support for undefined tags to the linker.
Companion patches:
- https://github.com/WebAssembly/binaryen/pull/4223
- https://github.com/emscripten-core/emscripten/pull/15266
Reviewed By: sbc100
Differential Revision: https://reviews.llvm.org/D111388
This patch continues unblocking optimizations that are blocked by pseudo probe instrumentation.
Not exactly like DbgIntrinsics, PseudoProbe intrinsic has other attributes (such as mayread, maywrite, mayhaveSideEffect) that can block optimizations. The issues fixed are:
- Flipped default param of getFirstNonPHIOrDbg API to skip pseudo probes
- Unblocked CSE by avoiding pseudo probe from clobbering memory SSA
- Unblocked induction variable simpliciation
- Allow empty loop deletion by treating probe intrinsic isDroppable
- Some refactoring.
Reviewed By: wenlei
Differential Revision: https://reviews.llvm.org/D110847
These "dump" methods call into MachineOperand::dump, which doesn't exist
with NDEBUG, thus we croak. Disable LiveDebugValues dump methods when
NDEBUG is turned on to avoid this.
With the -early-live-intervals command line flag,
TwoAddressInstructionPass::runOnMachineFunction would call
MachineFunction::verify before returning to check the live intervals.
But there was not much benefit to doing this since -verify-machineinstrs
and LLVM_ENABLE_EXPENSIVE_CHECKS provide a more general way of
scheduling machine verification after every pass.
Also it caused problems on targets like Lanai which are marked as "not
machine verifier clean", since verification would fail for known
target-specific problems which are nothing to do with LiveIntervals.
Differential Revision: https://reviews.llvm.org/D111618
This patch shifts the InstrRefBasedLDV class declaration to a header.
Partially because it's already massive, but mostly so that I can start
writing some unit tests for it. This patch also adds the boilerplate for
said unit tests.
Differential Revision: https://reviews.llvm.org/D110165
Per discussion in https://reviews.llvm.org/D111199,
the existing btf_tag attribute will be renamed to
btf_decl_tag. This patch mostly updated the Bitcode and
DebugInfo test cases with new attribute name.
Differential Revision: https://reviews.llvm.org/D111591
This patch contains following enhancements to SrcRegMap and DstRegMap:
1 In findOnlyInterestingUse not only check if the Reg is two address usage,
but also check after commutation can it be two address usage.
2 If a physical register is clobbered, remove SrcRegMap entries that are
mapped to it.
3 In processTiedPairs, when create a new COPY instruction, add a SrcRegMap
entry only when the COPY instruction is coalescable. (The COPY src is
killed)
With these enhancements isProfitableToCommute can do better commute decision,
and finally more register copies are removed.
Differential Revision: https://reviews.llvm.org/D108731
PHI elimination updates LiveVariables info as described here:
// We only need to update the LiveVariables kill of SrcReg if this was the
// last PHI use of SrcReg to be lowered on this CFG edge and it is not live
// out of the predecessor. We can also ignore undef sources.
Unfortunately if the last use also happened to be an undef use then it
would fail to update the LiveVariables at all. Fix this by not counting
undef uses in the VRegPHIUse map.
Thanks to Mikael Holmén for the test case!
Differential Revision: https://reviews.llvm.org/D111552
This moves the registry higher in the LLVM library dependency stack.
Every client of the target registry needs to link against MC anyway to
actually use the target, so we might as well move this out of Support.
This allows us to ensure that Support doesn't have includes from MC/*.
Differential Revision: https://reviews.llvm.org/D111454
This reverts commit 3e8d2008f7.
The code removed in this commit is actually required for extracting
fixed types from illegal scalable types, hence this commit causes
assertion failures in such extracts.
This was using MachineFunction::createExternalSymbolName() before, which seems
reasonable, but in fact this is freed before the asm emitter which tries to access
the function name string. Switching it to use the string returned by the attribute
seems to fix the problem.
In order to not generate an unnecessary G_CTLZ, I extended the constant folder
in the CSEMIRBuilder to handle G_CTLZ. I also added some extra handing of
vector constants too. It seems we don't have any support for doing constant
folding of vector constants, so the tests show some other useless G_SUB
instructions too.
Differential Revision: https://reviews.llvm.org/D111036
The shift libcalls have a shift amount parameter of MVT::i32, but
sometimes ExpandIntRes_Shift may be called with a node whose
second operand is a type that is larger than that. This leads to
an ABI mismatch, and for example causes a spurious zeroing of
a register in RV32 for 64-bit shifts. Note that at present regular
shift intstructions already have their shift amount operand adapted
at SelectionDAGBuilder::visitShift time, and funnelled shifts bypass that.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D110508
This change fixes a bug where the compiler generates a prologue_end
for line 0 locs. That is because line 0 is not associated with any
source location, so there should not be a prolgoue_end at a location
that doesn't correspond to a source location.
There were some LLVM tests that were explicitly checking for line 0
prologue_end's as well since I believe that to be incorrect, I had to
change those tests as well.
Patch by Shubham Rastogi!
Differential Revision: https://reviews.llvm.org/D110740
In a couple of places machine verification was disabled for no apparent
reason, probably just because an "addPass(..., false)" line was cut and
pasted from elsewhere.
After this patch the only remaining place where machine verification is
disabled in the generic TargetPassConfig code, is after addPreEmitPass.
When PHIElimination adds kills after lowering PHIs to COPYs it knows
that some instructions after the inserted COPY might use the same
SrcReg, but it was only looking at the terminator instructions at the
end of the block, not at other instructions like INLINEASM_BR that can
appear after the COPY insertion point.
Since we have already called findPHICopyInsertPoint, which knows about
INLINEASM_BR, we might as well reuse the insertion point that it
calculated when looking for instructions that might use SrcReg.
This fixes a machine verification failure if you force machine
verification to run after PHIElimination (currently it is disabled for
other reasons) when running
test/CodeGen/X86/callbr-asm-phi-placement.ll.
Differential Revision: https://reviews.llvm.org/D110834
Print this basic block flag as inlineasm-br-indirect-target and parse
it. This allows you to write MIR test cases for INLINEASM_BR. The test
case I added is one that I wanted to precommit anyway for D110834.
Differential Revision: https://reviews.llvm.org/D111291
Based on the reasoning of D53903, register operands of DBG_VALUE are
invariably treated as RegState::Debug operands. This change enforces
this invariant as part of MachineInstr::addOperand so that all passes
emit this flag consistently.
RegState::Debug is inconsistently set on DBG_VALUE registers throughout
LLVM. This runs the risk of a filtering iterator like
MachineRegisterInfo::reg_nodbg_iterator to process these operands
erroneously when not parsed from MIR sources.
This issue was observed in the development of the llvm-mos fork which
adds a backend that relies on physical register operands much more than
existing targets. Physical RegUnit 0 has the same numeric encoding as
$noreg (indicating an undef for DBG_VALUE). Allowing debug operands into
the machine scheduler correlates $noreg with RegUnit 0 (i.e. a collision
of register numbers with different zero semantics). Eventually, this
causes an assert where DBG_VALUE instructions are prohibited from
participating in live register ranges.
Reviewed By: MatzeB, StephenTozer
Differential Revision: https://reviews.llvm.org/D110105
Fix the calculation of ReplacedAllUntiedUses when any of the tied defs
are early-clobber. The effect of this is to fix the placement of kill
flags on an instruction like this (from @f2 in
test/CodeGen/SystemZ/asm-18.ll):
INLINEASM &"stepb $1, $2" [attdialect], $0:[regdef-ec:GRH32Bit], def early-clobber %3:grh32bit, $1:[reguse tiedto:$0], killed %4:grh32bit(tied-def 3), $2:[reguse:GRH32Bit], %4:grh32bit
After TwoAddressInstruction without this patch:
%3:grh32bit = COPY killed %4:grh32bit
INLINEASM &"stepb $1, $2" [attdialect], $0:[regdef-ec:GRH32Bit], def early-clobber %3:grh32bit, $1:[reguse tiedto:$0], %3:grh32bit(tied-def 3), $2:[reguse:GRH32Bit], %4:grh32bit
Note that the COPY kills %4, even though there is a later use of %4 in
the INLINEASM. This fails machine verification if you force it to run
after TwoAddressInstruction (currently it is disabled for other
reasons).
After TwoAddressInstruction with this patch:
%3:grh32bit = COPY %4:grh32bit
INLINEASM &"stepb $1, $2" [attdialect], $0:[regdef-ec:GRH32Bit], def early-clobber %3:grh32bit, $1:[reguse tiedto:$0], %3:grh32bit(tied-def 3), $2:[reguse:GRH32Bit], %4:grh32bit
Differential Revision: https://reviews.llvm.org/D110848
To better reflect the meaning of the now-disambiguated {GlobalValue,
GlobalAlias}::getBaseObject after breaking off GlobalIFunc::getResolverFunction
(D109792), the function is renamed to getAliaseeObject.
This reverts c7f16ab3e3 / r109694 - which
suggested this was done to improve consistency with the gdb test suite.
Possible that at the time GCC did not canonicalize integer types, and so
matching types was important for cross-compiler validity, or that it was
only a case of over-constrained test cases that printed out/tested the
exact names of integer types.
In any case neither issue seems to exist today based on my limited
testing - both gdb and lldb canonicalize integer types (in a way that
happens to match Clang's preferred naming, incidentally) and so never
print the original text name produced in the DWARF by GCC or Clang.
This canonicalization appears to be in `integer_types_same_name_p` for
GDB and in `TypeSystemClang::GetBasicTypeEnumeration` for lldb.
(I tested this with one translation unit defining 3 variables - `long`,
`long (*)()`, and `int (*)()`, and another translation unit that had
main, and a function that took `long (*)()` as a parameter - then
compiled them with mismatched compilers (either GCC+Clang, or
Clang+(Clang with this patch applied)) and no matter the combination,
despite the debug info for one CU naming the type "long int" and the
other naming it "long", both debuggers printed out the name as "long"
and were able to correctly perform overload resolution and pass the
`long int (*)()` variable to the `long (*)()` function parameter)
Did find one hiccup, identified by the lldb test suite - that CodeView
was relying on these names to map them to builtin types in that format.
So added some handling for that in LLVM. (these could be split out into
separate patches, but seems small enough to not warrant it - will do
that if there ends up needing any reverti/revisiting)
Differential Revision: https://reviews.llvm.org/D110455
Currently the max alignment representable is 1GB, see D108661.
Setting the align of an object to 4GB is desirable in some cases to make sure the lower 32 bits are clear which can be used for some optimizations, e.g. https://crbug.com/1016945.
This uses an extra bit in instructions that carry an alignment. We can store 15 bits of "free" information, and with this change some instructions (e.g. AtomicCmpXchgInst) use 14 bits.
We can increase the max alignment representable above 4GB (up to 2^62) since we're only using 33 of the 64 values, but I've just limited it to 4GB for now.
The one place we have to update the bitcode format is for the alloca instruction. It stores its alignment into 5 bits of a 32 bit bitfield. I've added another field which is 8 bits and should be future proof for a while. For backward compatibility, we check if the old field has a value and use that, otherwise use the new field.
Updating clang's max allowed alignment will come in a future patch.
Reviewed By: hans
Differential Revision: https://reviews.llvm.org/D110451
Currently the max alignment representable is 1GB, see D108661.
Setting the align of an object to 4GB is desirable in some cases to make sure the lower 32 bits are clear which can be used for some optimizations, e.g. https://crbug.com/1016945.
This uses an extra bit in instructions that carry an alignment. We can store 15 bits of "free" information, and with this change some instructions (e.g. AtomicCmpXchgInst) use 14 bits.
We can increase the max alignment representable above 4GB (up to 2^62) since we're only using 33 of the 64 values, but I've just limited it to 4GB for now.
The one place we have to update the bitcode format is for the alloca instruction. It stores its alignment into 5 bits of a 32 bit bitfield. I've added another field which is 8 bits and should be future proof for a while. For backward compatibility, we check if the old field has a value and use that, otherwise use the new field.
Updating clang's max allowed alignment will come in a future patch.
Reviewed By: hans
Differential Revision: https://reviews.llvm.org/D110451
Currently the max alignment representable is 1GB, see D108661.
Setting the align of an object to 4GB is desirable in some cases to make sure the lower 32 bits are clear which can be used for some optimizations, e.g. https://crbug.com/1016945.
This uses an extra bit in instructions that carry an alignment. We can store 15 bits of "free" information, and with this change some instructions (e.g. AtomicCmpXchgInst) use 14 bits.
We can increase the max alignment representable above 4GB (up to 2^62) since we're only using 33 of the 64 values, but I've just limited it to 4GB for now.
The one place we have to update the bitcode format is for the alloca instruction. It stores its alignment into 5 bits of a 32 bit bitfield. I've added another field which is 8 bits and should be future proof for a while. For backward compatibility, we check if the old field has a value and use that, otherwise use the new field.
Updating clang's max allowed alignment will come in a future patch.
Reviewed By: hans
Differential Revision: https://reviews.llvm.org/D110451
As described on D111049, we're trying to remove the <string> dependency from error handling and replace uses of report_fatal_error(const std::string&) with the Twine() variant which can be forward declared.
We were previously silently generating incorrect code when extracting a
fixed-width vector from a scalable vector. This is worse than crashing,
since the user will have no indication that this is currently unsupported
behaviour. I have fixed the code to only perform DAG combines when safe
to do so, i.e. the input and output vectors are both fixed-width or
both scalable.
Test added here:
CodeGen/AArch64/sve-extract-scalable-vector.ll
Differential revision: https://reviews.llvm.org/D110624
If these blocks are unreachable, then we can discard all of the instructions.
However, keep the block around because it may have an address taken or the
block may have a stale reference from a PHI somewhere. Instead of finding
those PHIs and fixing them up, just leave the block empty.
Differential Revision: https://reviews.llvm.org/D111201
As described on D111049, we're trying to remove the <string> dependency from error handling and replace uses of report_fatal_error(const std::string&) with the Twine() variant which can be forward declared.
We can use the raw_string_ostream::str() method to perform the implicit flush() and return a reference to the std::string container that we can then wrap inside Twine().
hasLessThanNumFused and fuseInstructionPair are useful for
DAG mutations similar to MacroFusion, but which cannot use
MacroFusion as a whole (such as fusing non-dependent instruction).
Reviewed By: MatzeB
Differential Revision: https://reviews.llvm.org/D111070
Change-Id: I3a5d56aba0471d45ef64cebb9b724030e2eae2f3
Update the LiveVariables analysis after the special handling for
unspillable terminators which was added in D91358. This is just enough
to fix some "Block should not be in AliveBlocks" / "Block missing from
AliveBlocks" errors in the codegen test suite when machine verification
is forced to run after PHIElimination (currently it is disabled).
Differential Revision: https://reviews.llvm.org/D110939
An important part of the instruction referencing solution is that we
identify all the registers that values move between before we then compute
an SSA-like function from the machine code, and from the variable
intrinsics. DBG_PHIs weren't causing all the subregisters of their operands
to be tracked; this patch forces that to happen.
The practical implications were that not enough space is allocated for
storing values when analysing the function -- asan will crash on the
attached test case with an unpatched compiler. Non-asan llc's will produce
a DBG_VALUE $noreg, where it should be $dil.
Differential Revision: https://reviews.llvm.org/D109064
Deriving NoAlias based on having the same index in two BaseIndexOffset
expressions seemed weird (and as shown in the added unittest the
correctness of doing so depended on undocumented pre-conditions that
the user of BaseIndexOffset::computeAliasing would need to take care
of.
This patch removes the code that dereived NoAlias based on indices
being the same. As a compensation, to avoid regressions/diffs in
various lit test, we also add a new check. The new check derives
NoAlias in case the two base pointers are based on two different
GlobalValue:s (neither of them being a GlobalAlias).
Reviewed By: niravd
Differential Revision: https://reviews.llvm.org/D110256
This fixes a bug detected in DAGCombiner when using global alias
variables. Here is an example:
@foo = global i16 0, align 1
@aliasFoo = alias i16, i16 * @foo
define i16 @bar() {
...
store i16 7, i16 * @foo, align 1
store i16 8, i16 * @aliasFoo, align 1
...
}
BaseIndexOffset::computeAliasing would incorrectly derive NoAlias
for the two accesses in the example above, resulting in DAGCombiner
miscompiles.
This patch fixes the problem by a defensive approach letting
BaseIndexOffset::computeAliasing return false, i.e. that the aliasing
couldn't be determined, when comparing two global values and at least
one is a GlobalAlias. In the future we might improve this with a
deeper analysis to look at the aliasee for the GlobalAlias etc. But
that is a bit more complicated considering that we could have
'local_unnamed_addr' and situations with several 'alias' variables.
Fixes PR51878.
Differential Revision: https://reviews.llvm.org/D110064
Enable verification of live intervals immediately after computing them
(when -early-live-intervals is used) and fix a problem that that
provokes: currently the verifier insists that a segment that ends at an
early-clobber slot must be followed by another segment starting at the
same slot. But before TwoAddressInstruction runs, the equivalent
condition is: a segment that ends at an early-clobber slot must have its
last use tied to an early-clobber def. That condition is harder to check
here, so for now just disable this check until tied operands have been
rewritten.
Differential Revision: https://reviews.llvm.org/D111065
This is a port of the feature that allows the StackProtector pass to omit
checking code for stack canary checks, and rely on SelectionDAG to do it at a
later stage. The reasoning behind this seems to be to prevent the IR checking
instructions from hindering tail-call optimizations during codegen.
Here we allow GlobalISel to also use that scheme. Doing so requires that we
do some analysis using some factored-out code to determine where to generate
code for the epilogs.
Not every case is handled in this patch since we don't have support for all
targets that exercise different stack protector schemes.
Differential Revision: https://reviews.llvm.org/D98200
This reverts commit d95cd81141.
The selector sometimes leaves unreachable blocks unselected because it uses a
postorder traversal for the block ordering.
With the trap intrinsics now being emitted, these blocks are no longer empty and
the unselected G_INTRINSIC instructions survive past selection. To fix this,
keep track of which blocks are selected and later delete any blocks that weren't
selected.
This patch makes instruction-referencing accepts an additional scenario
where values can be read from physical registers at the start of blocks. As
far as I was aware, this only happened:
* With arguments in the entry block,
* With constant physical registers,
To which this patch adds a third case:
* With exception-handling landing-pad blocks
In the attached test: the operand of the dbg.value traces back to the
"landingpad" instruction, which becomes some copies from physregs. Right
now, that's deemed unacceptable, and the assertion fires. The fix is to
just accept this scenario; this is a case where the value in question is
defined by a register and a position, not by an instruction that defines
it. Reading it with a DBG_PHI is the correct behaviour, there isn't a
non-copy instruction that we can refer to.
Differential Revision: https://reviews.llvm.org/D109005
The delayed stack protector feature which is currently used for SDAG (and thus
allows for more commonly generating tail calls) depends on being able to extract
the tail call into a separate return block. To do this it also has to extract
the vreg->physreg copies that set up the call's arguments, since if it doesn't
then the call inst ends up using undefined physregs in it's new spliced block.
SelectionDAG implementations can do this because they delay emitting register
copies until *after* the stack arguments are set up. GISel however just
processes and emits the arguments in IR order, so stack arguments always end up
last, and thus this breaks the code that looks for any register arg copies that
precede the call instruction.
This patch adds a thunk argument to the assignValueToReg() and custom assignment
hooks. For outgoing arguments, register assignments use this return param to
return a thunk that does the actual generating of the copies. We collect these
until all the outgoing stack assignments have been done and then execute them,
so that the copies (and perhaps some artifacts like G_SEXTs) are placed after
any stores.
Differential Revision: https://reviews.llvm.org/D110610
Also remove some redundancy because the source and result
types of any multiply are always the same.
Differential Revision: https://reviews.llvm.org/D110926
We were previously just ignoring unreachable, but targets like Darwin want to
keep unreachable instructions as traps.
Differential Revision: https://reviews.llvm.org/D110603
Stop using APInt constructors and methods that were soft-deprecated in
D109483. This fixes all the uses I found in llvm, except for the APInt
unit tests which should still test the deprecated methods.
Differential Revision: https://reviews.llvm.org/D110807
This patch re-introduces the fix in the commit https://github.com/llvm/llvm-project/commit/66b0cebf7f736 by @yrnkrn
> In DwarfEHPrepare, after all passes are run, RewindFunction may be a dangling
>
> pointer to a dead function. To make sure it's valid, doFinalization nullptrs
> RewindFunction just like the constructor and so it will be found on next run.
>
> llvm-svn: 217737
It seems that the fix was not migrated to `DwarfEHPrepareLegacyPass`.
This patch also updates `llvm/test/CodeGen/X86/dwarf-eh-prepare.ll` to include `-run-twice` to exercise the cleanup. Without this patch `llvm-lit -v llvm/test/CodeGen/X86/dwarf-eh-prepare.ll` fails with
```
-- Testing: 1 tests, 1 workers --
FAIL: LLVM :: CodeGen/X86/dwarf-eh-prepare.ll (1 of 1)
******************** TEST 'LLVM :: CodeGen/X86/dwarf-eh-prepare.ll' FAILED ********************
Script:
--
: 'RUN: at line 1'; /home/arakaki/build/llvm-project/main/bin/opt -mtriple=x86_64-linux-gnu -dwarfehprepare -simplifycfg-require-and-preserve-domtree=1 -run-twice < /home/arakaki/repos/watch/llvm-project/llvm/test/CodeGen/X86/dwarf-eh-prepare.ll -S | /home/arakaki/build/llvm-project/main/bin/FileCheck /home/arakaki/repos/watch/llvm-project/llvm/test/CodeGen/X86/dwarf-eh-prepare.ll
--
Exit Code: 2
Command Output (stderr):
--
Referencing function in another module!
call void @_Unwind_Resume(i8* %ehptr) #1
; ModuleID = '<stdin>'
void (i8*)* @_Unwind_Resume
; ModuleID = '<stdin>'
in function simple_cleanup_catch
LLVM ERROR: Broken function found, compilation aborted!
PLEASE submit a bug report to https://bugs.llvm.org/ and include the crash backtrace.
Stack dump:
0. Program arguments: /home/arakaki/build/llvm-project/main/bin/opt -mtriple=x86_64-linux-gnu -dwarfehprepare -simplifycfg-require-and-preserve-domtree=1 -run-twice -S
1. Running pass 'Function Pass Manager' on module '<stdin>'.
2. Running pass 'Module Verifier' on function '@simple_cleanup_catch'
#0 0x000056121b570a2c llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) /home/arakaki/repos/watch/llvm-project/llvm/lib/Support/Unix/Signals.inc:569:0
#1 0x000056121b56eb64 llvm::sys::RunSignalHandlers() /home/arakaki/repos/watch/llvm-project/llvm/lib/Support/Signals.cpp:97:0
#2 0x000056121b56f28e SignalHandler(int) /home/arakaki/repos/watch/llvm-project/llvm/lib/Support/Unix/Signals.inc:397:0
#3 0x00007fc7e9b22980 __restore_rt (/lib/x86_64-linux-gnu/libpthread.so.0+0x12980)
#4 0x00007fc7e87d3fb7 raise /build/glibc-S7xCS9/glibc-2.27/signal/../sysdeps/unix/sysv/linux/raise.c:51:0
#5 0x00007fc7e87d5921 abort /build/glibc-S7xCS9/glibc-2.27/stdlib/abort.c:81:0
#6 0x000056121b4e1386 llvm::raw_svector_ostream::raw_svector_ostream(llvm::SmallVectorImpl<char>&) /home/arakaki/repos/watch/llvm-project/llvm/include/llvm/Support/raw_ostream.h:674:0
#7 0x000056121b4e1386 llvm::report_fatal_error(llvm::Twine const&, bool) /home/arakaki/repos/watch/llvm-project/llvm/lib/Support/ErrorHandling.cpp:114:0
#8 0x000056121b4e1528 (/home/arakaki/build/llvm-project/main/bin/opt+0x29e3528)
#9 0x000056121adfd03f llvm::raw_ostream::operator<<(llvm::StringRef) /home/arakaki/repos/watch/llvm-project/llvm/include/llvm/Support/raw_ostream.h:218:0
FileCheck error: '<stdin>' is empty.
FileCheck command line: /home/arakaki/build/llvm-project/main/bin/FileCheck /home/arakaki/repos/watch/llvm-project/llvm/test/CodeGen/X86/dwarf-eh-prepare.ll
--
********************
********************
Failed Tests (1):
LLVM :: CodeGen/X86/dwarf-eh-prepare.ll
Testing Time: 0.22s
Failed: 1
```
Reviewed By: loladiro
Differential Revision: https://reviews.llvm.org/D110979
One of the cases identified in PR45116 - we don't need to limit extracted loads to ABI alignment, we can use allowsMemoryAccess - which tests using getABITypeAlign, but also checks if a target permits (fast) misaligned memory loads by checking allowsMisalignedMemoryAccesses as a fallback.
I've also cleaned up the alignment calculation code - if we have a constant extraction index then the alignment can be based on an offset from the original vector load alignment, but for non-constant indices we should assume the worst (single element alignment only).
Differential Revision: https://reviews.llvm.org/D110486
In collectTiedOperands, when handling an undef use that is tied to a
def, constrain the dst reg with the actual register class of the src
reg, instead of with the register class from the instructions's
MCInstrDesc. This makes a difference in some AMDGPU test cases like
this, before:
%16:sgpr_96 = INSERT_SUBREG undef %15:sgpr_96_with_sub0_sub1(tied-def 0), killed %11:sreg_64_xexec, %subreg.sub0_sub1
After, without this patch:
undef %16.sub0_sub1:sgpr_96 = COPY killed %11:sreg_64_xexec
This fails machine verification if you force it to run after
TwoAddressInstruction (currently it is disabled) with:
*** Bad machine code: Invalid register class for subregister index ***
- function: s_load_constant_v3i32_align4
- basic block: %bb.0 (0xa011a88)
- instruction: undef %16.sub0_sub1:sgpr_96 = COPY killed %11:sreg_64_xexec
- operand 0: undef %16.sub0_sub1:sgpr_96
Register class SGPR_96 does not fully support subreg index 4
After, with this patch:
undef %16.sub0_sub1:sgpr_96_with_sub0_sub1 = COPY killed %11:sreg_64_xexec
See also svn r159120 which introduced the code to handle tied undef
uses.
Differential Revision: https://reviews.llvm.org/D110944
Enabling this does not show any problems in check-llvm in an
LLVM_ENABLE_EXPENSIVE_CHECKS build.
Differential Revision: https://reviews.llvm.org/D110703
LiveVariables does not examine the contents of bundles, so
MachineVerifier should not expect it to know about kill flags on
operands of instructions inside a bundle.
With this fix we can enable machine verification after running the
LiveVariables analysis. Doing this does not show any problems in
check-llvm in an LLVM_ENABLE_EXPENSIVE_CHECKS build.
Differential Revision: https://reviews.llvm.org/D110700
Enabling this does not show any problems in check-llvm in an
LLVM_ENABLE_EXPENSIVE_CHECKS build.
Differential Revision: https://reviews.llvm.org/D110697
Enabling this does not show any problems in check-llvm in an
LLVM_ENABLE_EXPENSIVE_CHECKS build.
Differential Revision: https://reviews.llvm.org/D110695
Machine verification after DetectDeadLanes has been disabled since the
pass was first added in D18427, but I guess this was just due to copy-
and-paste. Enabling it does not show any problems in check-llvm in an
LLVM_ENABLE_EXPENSIVE_CHECKS build.
Differential Revision: https://reviews.llvm.org/D110689
LLVM (llvmorg-14-init) under Debian sid using latest gcc (Debian
10.3.0-9) 10.3.0 fails due to ambiguous overload on operators == and !=:
/root/src/llvm/src/llvm/tools/obj2yaml/elf2yaml.cpp:212:22:
error: ambiguous overload for 'operator!='
(operand types are 'llvm::ELFYAML::ELF_SHF' and 'int')
/root/src/llvm/src/llvm/tools/obj2yaml/elf2yaml.cpp:204:32:
error: ambiguous overload for 'operator!='
(operand types are 'const llvm::yaml::Hex64' and 'int')
/root/src/llvm/src/llvm/lib/CodeGen/LiveDebugValues/VarLocBasedImpl.cpp:629:35:
error: ambiguous overload for 'operator=='
(operand types are 'const uint64_t' {aka 'const long unsigned int'} and
'llvm::Register')
Reviewed by: StephenTozer, jmorse, Higuoxing
Differential Revision: https://reviews.llvm.org/D109534
Some vectors require both widening and promotion for their legalization.
This case is not yet handled in getCopyToPartsVector and falls back
on scalarizing by default. BBecause scalable vectors can't easily be
scalarised, we need to implement this in two separate stages:
1. Widen the vector.
2. Promote the vector.
As part of this patch, PromoteIntRes_CONCAT_VECTORS also needed to be
made scalable aware. Instead of falling back on scalarizing the vector
(fixed-width only), each sub-part of the CONCAT vector is promoted,
and the operation is performed on the type with the widest element type,
finally truncating the result to the promoted result type.
Differential Revision: https://reviews.llvm.org/D110646
While these functions are only used in one location in upstream,
it has been reused in multiple downstreams. Restore this file to
a globally visibile location (outside of APInt.h) to eliminate
donwstream breakage and enable potential future reuse.
Additionally, this patch renames types and cleans up
clang-tidy issues.
If the old instructions mentioned a subreg that the new instructions do
not, remove the subrange for that subreg.
For example, in TwoAddressInstructionPass::eliminateRegSequence, if a
use operand in the REG_SEQUENCE has the undef flag then we don't
generate a copy for it so after the elimination there should be no live
interval at all for the corresponding subreg of the def.
This is a small step towards switching TwoAddressInstructionPass over
from LiveVariables to LiveIntervals. Currently this path is only tested
if you explicitly enable -early-live-intervals.
Differential Revision: https://reviews.llvm.org/D110542
The legalizer handles this by breaking up an EXTRACT_SUBVECTOR into
smaller parts, and combines those together, padding the result with
UNDEF vectors, e.g.
nxv6i64 extract_subvector(nxv12i64, 6)
<->
nxv8i64 concat(
nxv2i64 extract_subvector(nxv16i64, 6)
nxv2i64 extract_subvector(nxv16i64, 8)
nxv2i64 extract_subvector(nxv16i64, 10)
nxv2i64 undef)
Reviewed By: frasercrmck, david-arm
Differential Revision: https://reviews.llvm.org/D110253
Machine verification after RemoveRedundantDebugValues has been disabled
since the pass was first added in D105279, but I guess this was just due
to copy-and-paste. Enabling it does not show any problems in check-llvm
in an LLVM_ENABLE_EXPENSIVE_CHECKS build.
Differential Revision: https://reviews.llvm.org/D110688
Comment says:
// If the operand is larger than the shift count type but the shift
// count type has enough bits to represent any shift value ...
It clearly talks about the shifted operand, not the shift-amount operand,
but the comparison is performed against Log2_32_Ceil(Op2.getValueSizeInBits())
where Op2 is the shift amount operand. This comparison also doesn't make
sense in the context of the previous one (ShiftsSize > Op2Size) because
Op2Size == Op2.getValueSizeInBits(). Fix to use Op1.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D110509
Similar to what SDAG does when it sees a smulo/umulo against 2
(see: `DAGCombiner::visitMULO`)
This pattern is fairly common in Swift code AFAICT.
Here's an example extracted from a Swift testcase:
https://godbolt.org/z/6cT8Mesx7
Differential Revision: https://reviews.llvm.org/D110662
To avoid using the AST when emitting diagnostics, split the "dontcall"
attribute into "dontcall-warn" and "dontcall-error", and also add the
frontend attribute value as the LLVM attribute value. This gives us all
the information to report diagnostics we need from within the IR (aside
from access to the original source).
One downside is we directly use LLVM's demangler rather than using the
existing Clang diagnostic pretty printing of symbols.
Previous revisions didn't properly declare the new dependencies.
Reviewed By: nickdesaulniers
Differential Revision: https://reviews.llvm.org/D110364
The ARM backend was explicitly setting global binding on the personality
symbol. This was added without any comment in a7ec2dcefd, which
introduced EHABI support (back in 2011). None of the other backends do
anything equivalent, as far as I can tell.
This causes problems when attempting to wrap the personality symbol.
Wrapped symbols are marked as weak inside LTO to inhibit IPO (see
https://reviews.llvm.org/D33621). When we wrap the personality symbol,
it initially gets weak binding, and then the ARM backend attempts to
change the binding to global, which causes an error in MC because of
attempting to change the binding of a symbol from non-global to global
(the error was added in https://reviews.llvm.org/D90108).
Simply drop the ARM backend's explicit global binding setting to fix
this. This matches all the other backends, and a large internal
application successfully linked and ran with this change, so it
shouldn't cause any problems. Test via LLD, since wrapping is required
to exhibit the issue.
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D110609
To avoid using the AST when emitting diagnostics, split the "dontcall"
attribute into "dontcall-warn" and "dontcall-error", and also add the
frontend attribute value as the LLVM attribute value. This gives us all
the information to report diagnostics we need from within the IR (aside
from access to the original source).
One downside is we directly use LLVM's demangler rather than using the
existing Clang diagnostic pretty printing of symbols.
Reviewed By: nickdesaulniers
Differential Revision: https://reviews.llvm.org/D110364
Call RemoveMachineInstrFromMaps before erasing instrs.
repairIntervalsInRange will do this for you after erasing the
instruction, but it's not safe to rely on it because assertions in
SlotIndexes::removeMachineInstrFromMaps refer to fields in the erased
instruction.
This fixes asan buildbot failures caused by D110335.
This patch emits DW_TAG_namelist and DW_TAG_namelist_item for fortran
namelist variables. DICompositeType is extended to support this fortran
feature.
Reviewed By: aprantl
Differential Revision: https://reviews.llvm.org/D108553
In TwoAddressInstructionPass::processTiedPairs, update subranges of the
live interval for RegB as well as the main range.
This is a small step towards switching TwoAddressInstructionPass over
from LiveVariables to LiveIntervals. Currently this path is only tested
if you explicitly enable -early-live-intervals.
Differential Revision: https://reviews.llvm.org/D110526
After TwoAddressInstructionPass calls
TargetInstrInfo::convertToThreeAddress, improve the LiveIntervals repair
to cope with convertToThreeAddress creating more than one new
instruction.
This mostly seems to benefit X86. For example in
test/CodeGen/X86/zext-trunc.ll it converts:
%4:gr32 = ADD32rr %3:gr32(tied-def 0), %2:gr32, implicit-def dead $eflags
to:
undef %6.sub_32bit:gr64 = COPY %3:gr32
undef %7.sub_32bit:gr64_nosp = COPY %2:gr32
%4:gr32 = LEA64_32r killed %6:gr64, 1, killed %7:gr64_nosp, 0, $noreg
Differential Revision: https://reviews.llvm.org/D110335
This patch adds a generic DAGCombine for vector-predicated (VP) nodes.
Those for which we can determine that no vector element is active can be
replaced by either undef or, for reductions, the start value.
This is tested rather trivially at the IR level, where it's possible
that we want to teach instcombine to perform this optimization.
However, we can also see the zero-evl case arise during SelectionDAG
legalization, when wide VP operations can be split into two and the
upper operation emerges as trivially false.
It's possible that we could perform this optimization "proactively"
(both on legal vectors and before splitting) and reduce the width of an
operation and insert it into a larger undef vector:
```
v8i32 vp_add x, y, mask, 4
->
v8i32 insert_subvector (v8i32 undef), (v4i32 vp_add xsub, ysub, mask, 4), i32 0
```
This is somewhat analogous to similar vector narrow/widening
optimizations, but it's unclear at this point whether that's beneficial
to do this for VP ops for any/all targets.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D109148
One of the cases identified in PR45116 - we don't need to limit store narrowing to ABI alignment, we can use allowsMemoryAccess - which tests using getABITypeAlign, but also checks if a target permits (fast) misaligned memory access by checking allowsMisalignedMemoryAccesses as a fallback.
Call RemoveMachineInstrFromMaps before erasing instrs.
repairIntervalsInRange will do this for you after erasing the
instruction, but it's not safe to rely on it because assertions in
SlotIndexes::removeMachineInstrFromMaps refer to fields in the erased
instruction.
This fixes asan buildbot failures caused by D110328.
In repairIntervalsInRange, if the new instructions refer to subregs but
the old instructions did not, make sure any existing live interval for
the superreg is updated to have subranges. Also skip repairing any range
that we have recalculated from scratch, partly for efficiency but also
to avoids some cases that repairOldRegInRange can't handle.
The existing test/CodeGen/AMDGPU/twoaddr-regsequence.mir provides some
test coverage for this change: when TwoAddressInstructionPass converts
REG_SEQUENCE into subreg copies, the live intervals will now get
subranges and MachineVerifier will verify that the subranges are
correct. Unfortunately MachineVerifier does not complain if the
subranges are not present, so the test also passed before this patch.
This patch also fixes ~800 of the ~1500 failures in the whole CodeGen
lit test suite when -early-live-intervals is forced on.
Differential Revision: https://reviews.llvm.org/D110328
The fix applied in D23303 "LiveIntervalAnalysis: fix a crash in repairOldRegInRange"
was over-zealous. It would bail out when the end of the range to be
repaired was in the middle of the first segment of the live range of
Reg, which was always the case when the range contained a single def of
Reg.
This patch fixes it as suggested by Matthias Braun in post-commit review
on the original patch, and tests it by adding -early-live-intervals to
a selection of existing lit tests that now pass.
(Note that D23303 was originally applied to fix a crash in
SILoadStoreOptimizer, but that is now moot since D23814 updated
SILoadStoreOptimizer to run before scheduling so it no longer has to
update live intervals.)
Differential Revision: https://reviews.llvm.org/D110238
Unrevert with some changes to the tests:
- Add -verify-machineinstrs to check for remaining problems in live
interval support in TwoAddressInstructionPass.
- Drop test/CodeGen/AMDGPU/extract-load-i1.ll since it suffers from
some of those remaining problems.
The fix applied in D23303 "LiveIntervalAnalysis: fix a crash in repairOldRegInRange"
was over-zealous. It would bail out when the end of the range to be
repaired was in the middle of the first segment of the live range of
Reg, which was always the case when the range contained a single def of
Reg.
This patch fixes it as suggested by Matthias Braun in post-commit review
on the original patch, and tests it by adding -early-live-intervals to
a selection of existing lit tests that now pass.
(Note that D23303 was originally applied to fix a crash in
SILoadStoreOptimizer, but that is now moot since D23814 updated
SILoadStoreOptimizer to run before scheduling so it no longer has to
update live intervals.)
Differential Revision: https://reviews.llvm.org/D110238
This patch adds codegen support for lowering the vector-predicated
reduction intrinsics to RVV instructions. The process is similar to that
of the other reduction intrinsics, save for the fact that every VP
reduction has a start value. We reuse the existing custom "VL" nodes,
adding extra patterns where required to handle non-true masks.
To support these nodes, the `RISCVISD::VECREDUCE_*_VL` nodes have been
given an explicit "merge" operand. This is to faciliate the VP
reductions, where we must be careful to ensure that even if no operation
is performed (when VL=0) we still produce the start value. The RVV
reductions don't update the destination register under these conditions,
so we tie the splatted start value to the output register.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D107657
This simplifies the API and addresses a FIXME in
TwoAddressInstructionPass::convertInstTo3Addr.
Differential Revision: https://reviews.llvm.org/D110229
Neither of these passes modify the CFG, allowing us to preserve DomTree
and LoopInfo across them by using setPreservesCFG.
Differential Revision: https://reviews.llvm.org/D110161
The pass uses different cost kinds to estimate "old" and "interleaved" costs:
default cost kind for all targets override `getInterleavedMemoryOpCost()` is
`TCK_SizeAndLatency`. Although at the moment estimated `TCK_Latency` costs are
equal to `TCK_SizeAndLatency`, (so the change is NFC) it may change in future.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D110100
Currenlty PseudoProbeInserter is a pass conditioned on a target switch. It works well with a single clang invocation. It doesn't work so well when the backend is called separately (i.e, through the linker or llc), where user has always to pass -pseudo-probe-for-profiling explictly. I'm making the pass a default pass that requires no command line arg to trigger, but will be actually run depending on whether the CU comes with `llvm.pseudo_probe_desc` metadata.
Reviewed By: wenlei
Differential Revision: https://reviews.llvm.org/D110209
Note that MIRCanonicalizerID is declared in
llvm/include/llvm/CodeGen/Passes.h, which MIRCanonicalizerPass.cpp
includes.
Identified with readability-redundant-declaration.
This code seems untested and is likely obsolete, because this case
should already be handled by the code that legalizes the result type
of EXTRACT_SUBVECTOR.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D110061
This is required to codegen something like:
<vscale x 8 x i16> @llvm.experimental.vector.insert(<vscale x 8 x i16> %vec,
<vscale x 2 x i16> %subvec,
i64 %idx)
where the output vector is legal, but the input vector needs promoting.
It implements this by performing the whole operation on the promoted type,
and then truncating the result.
Reviewed By: david-arm, craig.topper
Differential Revision: https://reviews.llvm.org/D110059
Most of the code wasn't yet scalable safe, although most of the
code conceptually just works for scalable vectors. This change
makes the algorithm work on ElementCount, where appropriate,
and leaves the fixed-width only code to use `getFixedNumElements`.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D110058
And always print it.
This makes some LLVM diagnostics match up better with Clang's diagnostics.
Updated some AMDGPU uses of DiagnosticInfoResourceLimit and now we print
better diagnostics for those.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D110204
This requires a minor change to CodeGenPrepare to ensure that
shouldSinkOperands will be called for And.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D110106
When using instructions which have a MetadataAsValue argument
(e.g. some target-specific intrinsics) MD canonicalization strips
internal MDNodes with a single ConstantAsMetadata child. That
prevented IRTranslator from the proper translation of such a calls.
Add generic helper function that matches constant splat. It has option to
match constant splat with undef (some elements can be undef but not all).
Add util function and matcher for G_FCONSTANT splat.
Differential Revision: https://reviews.llvm.org/D104410
For artifacts excluding G_TRUNC/G_SEXT, which have IR counterparts, we don't
seem to have debug users of defs. However, in the legalizer we're always calling
MachineInstr::eraseFromParentAndMarkDBGValuesForRemoval() which is expensive.
In some rare cases, this contributes significantly to unreasonably long compile
times when we have lots of artifact combiner activity.
To verify this, I added asserts to that function when it actually replaced a debug
use operand with undef for these artifacts. On CTMark with both -O0 and -Os and
debug info enabled, I didn't see a single case where it triggered.
In my measurements I saw around a 0.5% geomean compile-time improvement on -g -O0
for AArch64 with this change.
Differential Revision: https://reviews.llvm.org/D109750
Add eraseInstr(s) utility functions. Before deleting an instruction
collects its use instructions. After deletion deletes use instructions
that became trivially dead.
This patch clears all dead instructions in existing legalizer mir tests.
Differential Revision: https://reviews.llvm.org/D109154
Rework getConstantstVRegValWithLookThrough in order to make it clear if we
are matching integer/float constant only or any constant(default).
Add helper functions that get DefVReg and APInt/APFloat from constant instr
getIConstantVRegValWithLookThrough: integer constant, only G_CONSTANT
getFConstantVRegValWithLookThrough: float constant, only G_FCONSTANT
getAnyConstantVRegValWithLookThrough: either G_CONSTANT or G_FCONSTANT
Rename getConstantVRegVal and getConstantVRegSExtVal to getIConstantVRegVal
and getIConstantVRegSExtVal. These now only match G_CONSTANT as described
in comment.
Relevant matchers now return both DefVReg and APInt/APFloat.
Replace existing uses of getConstantstVRegValWithLookThrough and
getConstantVRegVal with new helper functions. Any constant match is
only required in:
ConstantFoldBinOp: for constant argument that was bit-cast of float to int
getAArch64VectorSplat: AArch64::G_DUP operands can be any constant
amdgpu select for G_BUILD_VECTOR_TRUNC: operands can be any constant
In other places use integer only constant match.
Differential Revision: https://reviews.llvm.org/D104409
getMetadata() currently uses a weird API where it populates a
structure passed to it, and optionally merges into it. Instead,
we can return the AAMDNodes and provide a separate merge() API.
This makes usages more compact.
Differential Revision: https://reviews.llvm.org/D109852
Introduce a new command-line flag `-swift-async-fp={auto|always|never}`
that controls how code generation sets the Swift extended async frame
info bit. There are three possibilities:
* `auto`: which determines how to set the bit based on deployment target, either
statically or dynamically via `swift_async_extendedFramePointerFlags`.
* `always`: the default, always set the bit statically, regardless of deployment
target.
* `never`: never set the bit, regardless of deployment target.
Patch by Doug Gregor <dgregor@apple.com>
Reviewed By: doug.gregor
Differential Revision: https://reviews.llvm.org/D109392
If a loop count was initially represented by a 32b unsigned int in C
then the hardware-loop pass can recognise the loop guard and insert
the llvm.test.set.loop.iterations intrinsic. If this was instead a
unsigned short/char then clang inserts a zext instruction to expand
the loop count to an i32. This patch adds the necessary pattern
matching to enable the use of lvm.test.set.loop.iterations in those
cases.
Patch by: sherwin-dc
Differential Revision: https://reviews.llvm.org/D109631
New field `elements` is added to '!DIImportedEntity', representing
list of aliased entities.
This is needed to dump optimized debugging information where all names
in a module are imported, but a few names are imported with overriding
aliases.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D109343
The doc comment for isPredecessor says:
Returns true if \p DefMI precedes \p UseMI or they are the same
instruction.
And dominates relies on that behavior for its own:
Returns true if \p DefMI dominates \p UseMI. By definition an
instruction dominates itself.
Make both statements correct by fixing isPredecessor.
Found by inspection.
The fmul is a canonicalizing operation, and fneg is not so this would
break denormals that need flushing and also would not quiet signaling
nans. Fold to fsub instead, which is also canonicalizing.
This simple heuristic uses the estimated live range length combined
with the number of registers in the class to switch which heuristic to
use. This was taking the raw number of registers in the class, even
though not all of them may be available. AMDGPU heavily relies on
dynamically reserved numbers of registers based on user attributes to
satisfy occupancy constraints, so the raw number is highly misleading.
There are still a few problems here. In the original testcase that
made me notice this, the live range size is incorrect after the
scheduler rearranges instructions, since the instructions don't have
the original InstrDist offsets. Additionally, I think it would be more
appropriate to use the number of disjointly allocatable registers in
the class. For the AMDGPU register tuples, there are a large number of
registers in each tuple class, but only a small fraction can actually
be allocated at the same time since they all overlap with each
other. It seems we do not have a query that corresponds to the number
of independently allocatable registers. Relatedly, I'm still debugging
some allocation failures where overlapping tuples seem to not be
handled correctly.
The test changes are mostly noise. There are a handful of x86 tests
that look like regressions with an additional spill, and a handful
that now avoid a spill. The worst looking regression is likely
test/Thumb2/mve-vld4.ll which introduces a few additional
spills. test/CodeGen/AMDGPU/soft-clause-exceeds-register-budget.ll
shows a massive improvement by completely eliminating a large number
of spills inside a loop.
Ignore dbg instructions when collecting stack slot markers. This is
to make sure the coloring is invariant regarding presence of dbg
instructions (even in cases when the dbg instructions might be
badly placed in the input).
Differential Revision: https://reviews.llvm.org/D109758
APInt is used to describe a bit mask in a variety of value tracking and demanded bits/elts functions.
When traversing through dst/src operands, we have a number of places where these masks need to widened/narrowed to translate through bitcasts, reductions etc. to a different type.
This patch add a APIntOps::ScaleBitMask common helper, adds unit test coverage, and updates a number of cases to use the the helper instead of their own implementation.
This came up on D109065 where we currently have to add yet another implementation of the same code.
Differential Revision: https://reviews.llvm.org/D109683
This extends the custom lowering for extending loads on
fixed length vectors in SVE to support masked extending loads.
The existing tests for correct behaviour of masked extending loads
exhibit bad code generation due to the legalistaion of i1 vectors.
They have been left as-is and new tests have been added that do not
exhibit this behaviour.
Differential Revision: https://reviews.llvm.org/D108200
Rather than inspecting the pointer element type, use the access
type of the load/store/atomicrmw/cmpxchg.
In the process of doing this, simplify the logic by storing the
address + type in MemoryUses, rather than an Instruction + Operand
pair (which was then used to fetch the address).
Always use the byval/inalloca/preallocated type (which is required
nowadays), don't fall back on the pointer element type.
This requires adding Function::getParamPreallocatedType() to
mirror the CallBase API, so that the templated code can work with
both.
This patch implements legalization of EXTRACT_SUBVECTOR for the case
where the result needs promoting, and the input type requires widening.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D109509
This patch implements legalization of EXTRACT_SUBVECTOR for the case
where the result needs promoting, and the input type is either legal
or requires splitting.
The idea is that the operation is broken down into simpler steps,
by first extracting a smaller subvector until the input vector
becomes legal or requires promotion.
Reviewed By: CarolineConcatto
Differential Revision: https://reviews.llvm.org/D109313
Soft deprecrate isNullValue/isAllOnesValue and update in tree
callers. This matches the changes to the APInt interface from
D109483.
Reviewed By: lattner
Differential Revision: https://reviews.llvm.org/D109535
Follow up to suggestions in D109103 via hans:
I think UnreachableDefault (or UnreachableFallthrough) would be a
better name now, since it doesn't just omit the range check, it also
omits the last bit test.
Reviewed By: hans
Differential Revision: https://reviews.llvm.org/D109455
This renames the primary methods for creating a zero value to `getZero`
instead of `getNullValue` and renames predicates like `isAllOnesValue`
to simply `isAllOnes`. This achieves two things:
1) This starts standardizing predicates across the LLVM codebase,
following (in this case) ConstantInt. The word "Value" doesn't
convey anything of merit, and is missing in some of the other things.
2) Calling an integer "null" doesn't make any sense. The original sin
here is mine and I've regretted it for years. This moves us to calling
it "zero" instead, which is correct!
APInt is widely used and I don't think anyone is keen to take massive source
breakage on anything so core, at least not all in one go. As such, this
doesn't actually delete any entrypoints, it "soft deprecates" them with a
comment.
Included in this patch are changes to a bunch of the codebase, but there are
more. We should normalize SelectionDAG and other APIs as well, which would
make the API change more mechanical.
Differential Revision: https://reviews.llvm.org/D109483
This moves one mid-size function out of line, inlines the
trivial tcAnd/tcOr/tcXor/tcComplement methods into their only
caller, and moves the magic/umagic functions into SelectionDAG
since they are implementation details of its algorithm. This
also removes the unit tests for magic, but these are already
tested in the divide lowering logic for various targets.
This also upgrades some C style comments to C++.
Differential Revision: https://reviews.llvm.org/D109476
This is a port of a combine which matches a pattern where a wide type scalar
value is stored by several narrow stores. It folds it into a single store or
a BSWAP and a store if the targets supports it.
Assuming little endian target:
i8 *p = ...
i32 val = ...
p[0] = (val >> 0) & 0xFF;
p[1] = (val >> 8) & 0xFF;
p[2] = (val >> 16) & 0xFF;
p[3] = (val >> 24) & 0xFF;
=>
*((i32)p) = val;
On CTMark AArch64 -Os this results in a good amount of savings:
Program before after diff
SPASS 412792 412788 -0.0%
kc 432528 432512 -0.0%
lencod 430112 430096 -0.0%
consumer-typeset 419156 419128 -0.0%
bullet 475840 475752 -0.0%
tramp3d-v4 367760 367628 -0.0%
clamscan 383388 383204 -0.0%
pairlocalalign 249764 249476 -0.1%
7zip-benchmark 570100 568860 -0.2%
sqlite3 287628 286920 -0.2%
Geomean difference -0.1%
Differential Revision: https://reviews.llvm.org/D109419
Otherwise we end up with an extra conditional jump, following by an
unconditional jump off the end of a function. ie.
bb.0:
BT32rr ..
JCC_1 %bb.4 ...
bb.1:
BT32rr ..
JCC_1 %bb.2 ...
JMP_1 %bb.3
bb.2:
...
bb.3.unreachable:
bb.4:
...
Should be equivalent to:
bb.0:
BT32rr ..
JCC_1 %bb.4 ...
JMP_1 %bb.2
bb.1:
bb.2:
...
bb.3.unreachable:
bb.4:
...
This can occur since at the higher level IR (Instruction) SwitchInsts
are required to have BBs for default destinations, even when it can be
deduced that such BBs are unreachable.
For most programs, this isn't an issue, just wasted instructions since the
unreachable has been statically proven.
The x86_64 Linux kernel when built with CONFIG_LTO_CLANG_THIN=y fails to
boot though once D106056 is re-applied. D106056 makes it more likely
that correlation-propagation (CVP) can deduce that the default case of
SwitchInsts are unreachable. The x86_64 kernel uses a binary post
processor called objtool, which emits this warning:
vmlinux.o: warning: objtool: cfg80211_edmg_chandef_valid()+0x169: can't
find jump dest instruction at .text.cfg80211_edmg_chandef_valid+0x17b
I haven't debugged precisely why this causes a failure at boot time, but
fixing this very obvious jump off the end of the function fixes the
warning and boot problem.
Link: https://bugs.llvm.org/show_bug.cgi?id=50080
Fixes: https://github.com/ClangBuiltLinux/linux/issues/679
Fixes: https://github.com/ClangBuiltLinux/linux/issues/1440
Reviewed By: hans
Differential Revision: https://reviews.llvm.org/D109103
When handling register spill for indirect debug value LiveDebugValues pass doesn't add
DW_OP_deref operator which may in some cases cause debugger to return value address, instead
of value while machine register holding that address is spilled.
Differential revision: https://reviews.llvm.org/D109142
This patch extends the preliminary support for vector-predicated (VP)
operation legalization to include promotion of illegal integer vector
types.
Integer promotion of binary VP operations is relatively simple and
piggy-backs on the non-VP logic, but passing the two extra mask and VP
operands through to the promoted operation.
Tests have been added to the RISC-V target to cover the basic scenarios
for integer promotion for both fixed- and scalable-vector types.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D108288
In preparation for passing the MCSubtargetInfo (STI) through to writeNops
so that it can use the STI in operation at the time, we need to record the
STI in operation when a MCAlignFragment may write nops as padding. The
STI is currently unused, a further patch will pass it through to
writeNops.
There are many places that can create an MCAlignFragment, in most cases
we can find out the STI in operation at the time. In a few places this
isn't possible as we are in initialisation or finalisation, or are
emitting constant pools. When possible I've tried to find the most
appropriate existing fragment to obtain the STI from, when none is
available use the per module STI.
For constant pools we don't actually need to use EmitCodeAlign as the
constant pools are data anyway so falling through into it via an
executable NOP is no better than falling through into data padding.
This is a prerequisite for D45962 which uses the STI to emit the
appropriate NOP for the STI. Which can differ per fragment.
Note that involves an interface change to InitSections. It is now
called initSections and requires a SubtargetInfo as a parameter.
Differential Revision: https://reviews.llvm.org/D45961
Legalizing G_MUL for non-standard types (like i33) generated an error. Putting
minScalar and maxScalar instead of clampScalar. Also using new rule, instead
of widening to the next power of 2, widen to the next multiple of the passed
argument (32 in this case), so instead of widening i65 to i128, we widen it to
i96.
Patch by: Mateja Marjanovic
Differential Revision: https://reviews.llvm.org/D109228
Add implementation for the legalization of G_ROTL and G_ROTR machine
instructions. They are very similar to funnel shift instructions, the only
difference is funnel shifts have 3 operands, whereas rotate instructions have
two operands, the first being the register that is being rotated and the second
being the number of shifts. The legalization of G_ROTL/G_ROTR is just lowering
them into funnel shift instructions if they are legal.
Patch by: Mateja Marjanovic
Differential Revision: https://reviews.llvm.org/D105347
Legalize G_MEMCPY, G_MEMMOVE, G_MEMSET and G_MEMCPY_INLINE.
Corresponding intrinsics are replaced by a loop that uses loads/stores in
AMDGPULowerIntrinsics pass unless their length is a constant lower then
MemIntrinsicExpandSizeThresholdOpt (default 1024). Any G_MEM* instruction that
reaches legalizer should have a const length argument and should be expanded
into appropriate number of loads + stores.
Differential Revision: https://reviews.llvm.org/D108357
This patch adds support for the vector-predicated `VP_STORE` and
`VP_LOAD` nodes. We do this in the same way we lower `MSTORE` and
`MLOAD`: to regular load/store instructions via intrinsics.
One necessary change was made to `SelectionDAGLegalize` so that
`VP_STORE` nodes' operation actions are taken from the stored "value"
operands, in the same vein as `STORE` or `MSTORE`.
Reviewed By: craig.topper, rogfer01
Differential Revision: https://reviews.llvm.org/D108999
This patch adds support for the `VP_SCATTER` and `VP_GATHER` nodes by
lowering them to RVV's `vsox`/`vlux` instructions, respectively. This
process is almost identical to the existing `MSCATTER`/`MGATHER` support.
One extra change was made to `SelectionDAGLegalize` so that
`VP_SCATTER`'s operation action is derived from its stored "value"
operand rather than its return type (which is always the chain).
Reviewed By: craig.topper, rogfer01
Differential Revision: https://reviews.llvm.org/D108987
In case of a virtual register tied to a phys-def, the register class needs to
be computed. Make sure that this works generally also with fast regalloc by
using TLI.getRegClassFor() whenever possible, and make only the case of
'Untyped' use getMinimalPhysRegClass().
Fixes https://bugs.llvm.org/show_bug.cgi?id=51699.
Review: Ulrich Weigand
Differential Revision: https://reviews.llvm.org/D109291
Given a select_cc producing a constant and a invertion of the constant
for a comparison more than zero, we can produce an xor with ashr
instead, which produces smaller code. The ashr either sets all bits or
clear all bits depending on if the value is negative. This is then xor'd
with the constant to optionally negate the value.
https://alive2.llvm.org/ce/z/DTFaBZ
This includes a OneUseCheck on the Cmp, which seems to make thinks a
little worse and will be removed in a followup.
Differential Revision: https://reviews.llvm.org/D109149
Pulled out of D109149, this folds set_cc seteq (ashr X, BW-1), -1 ->
set_cc setlt X, 0 to prevent some regressions later on when folding
select_cc setgt X, -1, C, ~C -> xor (ashr X, BW-1), C
Differential Revision: https://reviews.llvm.org/D109214
The temporary object was used as a workaround when the target parser may
change STI. D14346 made the MCSubtargetInfo argument to
createMCAsmParser const, so we no longer need the temporary object.
This allows constructing a MemDesc from a MachineMemoryOperand, a pattern that starts to show up more frequently.
Reviewed By: foad
Differential Revision: https://reviews.llvm.org/D109161
This adds the following combines:
```
x = ... 0 or 1
c = icmp eq x, 1
->
c = x
```
and
```
x = ... 0 or 1
c = icmp ne x, 0
->
c = x
```
When the target's true value for the relevant types is 1.
This showed up in the following situation:
https://godbolt.org/z/M5jKexWTW
SDAG currently supports the `ne` case, but not the `eq` case. This can probably
be further generalized, but I don't feel like thinking that hard right now.
This gives some minor code size improvements across the board on CTMark at
-Os for AArch64. (0.1% for 7zip and pairlocalalign in particular.)
Differential Revision: https://reviews.llvm.org/D109130
This add support for SjLj using Wasm exception handling instructions:
https://github.com/WebAssembly/exception-handling/blob/master/proposals/exception-handling/Exceptions.md
This does not yet support the mixed use of EH and SjLj within a
function. It will be added in a follow-up CL.
This currently passes all SjLj Emscripten tests for wasm0/1/2/3/s,
except for the below:
- `test_longjmp_standalone`: Uses Node
- `test_dlfcn_longjmp`: Uses NodeRAWFS
- `test_longjmp_throw`: Mixes EH and SjLj
- `test_exceptions_longjmp1`: Mixes EH and SjLj
- `test_exceptions_longjmp2`: Mixes EH and SjLj
- `test_exceptions_longjmp3`: Mixes EH and SjLj
Reviewed By: dschuff, tlively
Differential Revision: https://reviews.llvm.org/D108960
The semantics of tail predication loops means that the value of LR as an
instruction is executed determines the predicate. In other words:
mov r3, #3
DLSTP lr, r3 // Start tail predication, lr==3
VADD.s32 q0, q1, q2 // Lanes 0,1 and 2 are updated in q0.
mov lr, #1
VADD.s32 q0, q1, q2 // Only first lane is updated.
This means that the value of lr cannot be spilled and re-used in tail
predication regions without potentially altering the behaviour of the
program. More lanes than required could be stored, for example, and in
the case of a gather those lanes might not have been setup, leading to
alignment exceptions.
This patch adds a new lr predicate operand to MVE instructions in order
to keep a reference to the lr that they use as a tail predicate. It will
usually hold the zeroreg meaning not predicated, being set to the LR phi
value in the MVETPAndVPTOptimisationsPass. This will prevent it from
being spilled anywhere that it needs to be used.
A lot of tests needed updating.
Differential Revision: https://reviews.llvm.org/D107638
Please refer to
https://lists.llvm.org/pipermail/llvm-dev/2021-September/152440.html
(and that whole thread.)
TLDR: the original patch had no prior RFC, yet it had some changes that
really need a proper RFC discussion. It won't be productive to discuss
such an RFC, once it's actually posted, while said patch is already
committed, because that introduces bias towards already-committed stuff,
and the tree is potentially in broken state meanwhile.
While the end result of discussion may lead back to the current design,
it may also not lead to the current design.
Therefore i take it upon myself
to revert the tree back to last known good state.
This reverts commit 4c4093e6e3.
This reverts commit 0a2b1ba33a.
This reverts commit d9873711cb.
This reverts commit 791006fb8c.
This reverts commit c22b64ef66.
This reverts commit 72ebcd3198.
This reverts commit 5fa6039a5f.
This reverts commit 9efda541bf.
This reverts commit 94d3ff09cf.
This patch extends D107904's introduction of vector-predicated (VP)
operation legalization to include vector splitting.
When the result of a binary VP operation needs splitting, all of its
operands are split in kind. The two operands and the mask are split as
usual, and the vector-length parameter EVL is "split" such that the low
and high halves each execute the correct number of elements.
Tests have been added to the RISC-V target to show splitting several
scenarios for fixed- and scalable-vector types. Without support for
`umax` (e.g. in the `B` extension) the generated code starts to branch.
Ideally a cost model would prevent their insertion in the first place.
Through these tests many opportunities for better codegen can be seen:
combining known-undef VP operations and for constant-folding operations
on `ISD::VSCALE`, to name but a few.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D107957
I believe, the profitability reasoning here is correct
"sub"reg is already located within the 0'th subreg of wider reg,
so if we have suvector insertion at index 0 into undef,
then it's always free do to.
After this, D109065 finally avoids the regression in D108382.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D109074
A new LLVM specific TAG DW_TAG_LLVM_annotation is added.
The name is suggested by Paul Robinson ([1]).
Currently, this tag is used to output __attribute__((btf_tag("string")))
annotations in dwarf. The following is an example for a global
variable with two btf_tag attributes:
0x0000002a: DW_TAG_variable
DW_AT_name ("g1")
DW_AT_type (0x00000052 "int")
DW_AT_external (true)
DW_AT_decl_file ("/tmp/home/yhs/work/tests/llvm/btf_tag/t.c")
DW_AT_decl_line (8)
DW_AT_location (DW_OP_addr 0x0)
0x0000003f: DW_TAG_LLVM_annotation
DW_AT_name ("btf_tag")
DW_AT_const_value ("tag1")
0x00000048: DW_TAG_LLVM_annotation
DW_AT_name ("btf_tag")
DW_AT_const_value ("tag2")
0x00000051: NULL
In the future, DW_TAG_LLVM_annotation may encode other type
of non-string const value.
[1] https://lists.llvm.org/pipermail/llvm-dev/2021-June/151250.html
Differential Revision: https://reviews.llvm.org/D106621
When RA eliminated a dead def it can either immediately delete
the instruction itself or replace it with KILL to defer the
actual removal. If this instruction has a virtual register use
killing the register it will shrink the LI of the use. However,
if the LI covers the instruction and extends beyond it the
shrink will not happen. In fact that is impossible to shrink
such use because of the KILL still using it.
If later the LI of the use will be split at the KILL and the
KILL itself is eliminated after that point the new live segment
ends up at an invalid slot index.
This extremely rare condition was hit after D106408 which has
enabled rematerialization of such instructions. The replacement
with KILL is only done for rematerialized defs which became dead
and such rematerialization did not generally happen before.
The patch deletes an instruction immediately if it is a result
of rematerialization and has such use. An alternative would be
to prohibit a split at a KILL instruction, but it looks like it
is better to split a live range rather then keeping a killed
instruction just in case it can be rematerialized further.
Fixes PR51655.
Differential Revision: https://reviews.llvm.org/D108951
As noted in the comments in D108227, using G_FPTOSI produces wrong results for
G_ISNAN. Drop the G_FPTOSI and perform the operation on integer types.
Elsewhere in LLVM, a bitcast would be the appropriate choice (as it is in SDAG).
GlobalISel does not distinguish between integer and FP types, so a bitcast would
be meaningless here.
Followup to D99355: SDAG support for vector-predicated load/store/gather/scatter.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D105871
Instead of splitting off the fp16 to float conversion and generating
a libcall, we should split the operation into fp16 to float and float
to integer operations. This will allow the float to integer conversion
to go through any custom handling the target has. If the target doesn't
have custom handling then we should come back to ExpandIntRes_FP_TO_SINT/
ExpandIntRes_FP_TO_UINT automatically to create the libcall.
This avoids generating libcalls on 32-bit X86. These library functions may
not exist in 32-bit libgcc. At least for LLVM, we never generate them when
hardware floating point instructions are available.
Differential Revision: https://reviews.llvm.org/D108933
When expanding a SMULFIXSAT ISD node (usually originating from
a smul.fix.sat intrinsic) we've applied some optimizations for
the special case when the scale is zero. The idea has been that
it would be cheaper to use an SMULO instruction (if legal) to
perform the multiplication and at the same time detect any overflow.
And in case of overflow we could use some SELECT:s to replace the
result with the saturated min/max value. The only tricky part
is to know if we overflowed on the min or max value, i.e. if the
product is positive or negative. Unfortunately the implementation
has been incorrect as it has looked at the product returned by the
SMULO to determine the sign of the product. In case of overflow that
product is truncated and won't give us the correct sign bit.
This patch is adding an extra XOR of the multiplication operands,
which is used to determine the sign of the non truncated product.
This patch fixes PR51677.
Reviewed By: lebedev.ri
Differential Revision: https://reviews.llvm.org/D108938
The backend generally uses 64-bit immediates (e.g. what
MachineOperand::getImm() returns), so use that for analyzeCompare()
and optimizeCompareInst() as well. This avoids truncation for
targets that support immediates larger 32-bit. In particular, we
can avoid the bugprone value normalization hack in the AArch64
target.
This is a followup to D108076.
Differential Revision: https://reviews.llvm.org/D108875
Md5 hashing is expansive. Using a hash map to look up already computed GUID for dwarf names. Saw a 2% build time improvement on an internal large application.
Reviewed By: wenlei
Differential Revision: https://reviews.llvm.org/D108722
The check for whether a rotate is possible occurs before the
memory legality checks for the integer type. So it's possible we
decide we can use a rotate, but then fail the legality checks. If
that happens we should not fall back to a vector type. This triggers
an assertion in the rotate handling when it finds a vector type
instead of an integer type.
In theory we could use a shufflevector in place of the rotate, but
right now I'd just like to fix the crash.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D108839
If we encounter a new debug value, describing the same parameter,
we should stop tracking the parameter's Entry Value. At that point,
in some cases, the Transfer which uses the parameter's Entry Value,
is already emitted. Thanks to the RemoveRedundantDebugValues pass,
many problems with incorrect instruction order and number of DBG_VALUEs
are fixed. However, we still cannot rely on the rule that each new
debug value is set by the previous non-debug instruction in Machine
Basic Block.
When new parameter debug value triggers removal of Backup Entry Value
for the same parameter, do the cleanup of Transfers emitted from Backup
Entry Values. Get the Transfer Instruction which created the new debug
value and search for debug values already emitted from the to-be-deleted
Backup Entry Value and attached to the Transfer Instruction. If found,
delete the Transfer and remove "primary" Entry Value Var Loc from
OpenRanges.
This patch fixes PR47628.
Patch by Nikola Tesic.
Differential revision: https://reviews.llvm.org/D106856
This patch emits DW_TAG_namelist and DW_TAG_namelist_item for fortran
namelist variables. DICompositeType is extended to support this fortran
feature.
Reviewed By: aprantl
Differential Revision: https://reviews.llvm.org/D108553
Without this change only the preferred fusion opcode is tested
when attempting to combine FMA operations.
If both FMA and FMAD are available then FMA ops formed prior to
legalization will not be merged post legalization as FMAD becomes
the preferred fusion opcode.
Reviewed By: foad
Differential Revision: https://reviews.llvm.org/D108619
We can halve the number of mask constants by masking before shl
and after srl.
This can reduce the number of mov immediate or constant
materializations. Or reduce the number of constant pool loads
for X86 vectors.
I think we might be able to do something similar for bswap. I'll
look at it next.
Differential Revision: https://reviews.llvm.org/D108738
In the combination of addressing modes, when replacing the matched phi nodes,
sometimes the phi node to be replaced has been modified. For example,
there’s matcher set [A, B] and [C, A], which will have cyclic dependency:
A is replaced by B and C will be replaced by A. Because we tried to match new phi node
to another new phi node, we should ignore new phi nodes when mapping new phi node to old one.
Reviewed By: skatkov
Differential Revision: https://reviews.llvm.org/D108635
Emscripten SjLj and (soon-to-be-added) Wasm SjLj transformation share
many steps:
1. Initialize `setjmpTable` and `setjmpTableSize` in the entry BB
2. Handle `setjmp` callsites
3. Handle `longjmp` callsites
4. Cleanup and update SSA
1, 3, and 4 are identical for Emscripten SjLj and Wasm SjLj. Only the
step 2 is different. This CL extracts the current Emscripten SjLj's
longjmp callsites handling into a function. The reason to make this a
separate CL is, without this, the diff tool cannot compare things well
in the presence of moved code and added code in the followup Wasm SjLj
CL, and it ends up mixing them together, making the diff unreadable.
Also fixes some typos and variable names. So far we've been calling the
buffer argument to `setjmp` and `longjmp` `jmpbuf`, but the name used in
the man page for those functions is `env`, so updated them to be
consistent.
Reviewed By: tlively
Differential Revision: https://reviews.llvm.org/D108728
The plan was to use `wasm.catch.exn` intrinsic to catch exceptions and
add `wasm.catch.longjmp` intrinsic, that returns two values (setjmp
buffer and return value), later to catch longjmps. But because we
decided not to use multivalue support at the moment, we are going to use
one intrinsic that returns a single value for both exceptions and
longjmps. And even if it's not for that, I now think the naming of
`wasm.catch.exn` is a little weird, because the intrinsic can still take
a tag immediate, which means it can be used for anything, not only
exceptions, as long as that returns a single value.
This partially reverts D107405.
Reviewed By: tlively
Differential Revision: https://reviews.llvm.org/D108683
This is another bug exposed by https://llvm.org/PR51612
(and the one that triggered the initial assertion) in the report.
That example was suppressed with:
985b48f183
...but these would still crash because we created nodes
like UADDO without the expected 2 output values.
There are 2 bugs here:
1. We were not checking uses of operand 2 (the false value of the select).
2. We were not checking for multiple uses of nodes that produce >1 result.
Correcting those is enough to avoid the crash in the reduced test based on:
https://llvm.org/PR51612
The additional use check on operand 0 (the condition value of the select)
should not strictly be necessary because we are only replacing one use
with another (whether it makes performance sense to do the transform with
that pattern is not clear). But as noted in the TODO, changing that
uncovers another bug.
Note: there's at least one more bug here - we aren't propagating EVTs
correctly, but I plan to fix that in another patch.
Add support for the GNU C style __attribute__((error(""))) and
__attribute__((warning(""))). These attributes are meant to be put on
declarations of functions whom should not be called.
They are frequently used to provide compile time diagnostics similar to
_Static_assert, but which may rely on non-ICE conditions (ie. relying on
compiler optimizations). This is also similar to diagnose_if function
attribute, but can diagnose after optimizations have been run.
While users may instead simply call undefined functions in such cases to
get a linkage failure from the linker, these provide a much more
ergonomic and actionable diagnostic to users and do so at compile time
rather than at link time. Users instead may be able use inline asm .err
directives.
These are used throughout the Linux kernel in its implementation of
BUILD_BUG and BUILD_BUG_ON macros. These macros generally cannot be
converted to use _Static_assert because many of the parameters are not
ICEs. The Linux kernel still needs to be modified to make use of these
when building with Clang; I have a patch that does so I will send once
this feature is landed.
To do so, we create a new IR level Function attribute, "dontcall" (both
error and warning boil down to one IR Fn Attr). Then, similar to calls
to inline asm, we attach a !srcloc Metadata node to call sites of such
attributed callees.
The backend diagnoses these during instruction selection, while we still
know that a call is a call (vs say a JMP that's a tail call) in an arch
agnostic manner.
The frontend then reconstructs the SourceLocation from that Metadata,
and determines whether to emit an error or warning based on the callee's
attribute.
Link: https://bugs.llvm.org/show_bug.cgi?id=16428
Link: https://github.com/ClangBuiltLinux/linux/issues/1173
Reviewed By: aaron.ballman
Differential Revision: https://reviews.llvm.org/D106030
InstrRefBasedLDV is marginally slower than VarlocBasedLDV when analysing
optimised code -- however, it's much slower when analysing code compiled
-O0.
To avoid this: don't use instruction referencing for -O0 functions. In the
"pure" case of unoptimised code, this won't really harm the debugging
experience because most variables won't have been promoted off the stack,
so can't go missing. It becomes more complicated when optimised code is
inlined into functions marked optnone; however these are rare, and as -O0
doesn't run many optimisations there should be little damage to the debug
experience as a result.
I've taken the opportunity to refactor testing for instruction-referencing
into a MachineFunction method, which seems the most appropriate place to
put it.
Differential Revision: https://reviews.llvm.org/D108585
For ISD::EXTRACT_SUBVECTOR, its second operand must be a constant
multiple of the known-minimum vector length of the result type.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D107795
Stack slot colouring adds "weight" to slots if a non-dbg-value instruction
refers to it. This, unfortunately, means that DBG_PHI instructions can have
an effect on codegen. The fix is very simple, replace isDebugValue with
isDebugInstr.
The regression test contains a scenario that reproduces this problem; I've
represented both normal-debug mode and instr-ref debug mode instructions
in comment lines prefixed with AAAAAA and BBBBBB, and un-comment them with
sed to test that the two different modes produce the same behaviour.
Differential Revision: https://reviews.llvm.org/D108627
The sext_inreg_of_load combine did not have the isLegalOrBeforeLegalizer check,
leading to the generation of potentially illegal G_SEXTLOADs when run after legalization.
Reviewed By: foad
Differential Revision: https://reviews.llvm.org/D108626
On some AMDGPU subtargets, copying to and from AGPR registers using another
AGPR register is not possible. A intermediate VGPR register is needed for AGPR
to AGPR copy. This is an issue when machine copy propagation forwards a
COPY $agpr, replacing a COPY $vgpr which results in $agpr = COPY $agpr. It is
removing a cross class copy that may have been optimized by previous passes and
potentially creating an unoptimized cross class copy later on.
To avoid this issue, check CrossCopyRegClass if a different register class will
be needed for the copy. If so then avoid forwarding the copy when the
destination does not match the desired register class and if the original copy
already matches the desired register class.
Issue seen while attempting to optimize another AGPR to AGPR issue:
Live-ins: $agpr0
$vgpr0 = COPY $agpr0
$agpr1 = V_ACCVGPR_WRITE_B32 $vgpr0
$agpr2 = COPY $vgpr0
$agpr3 = COPY $vgpr0
$agpr4 = COPY $vgpr0
After machine-cp:
$vgpr0 = COPY $agpr0
$agpr1 = V_ACCVGPR_WRITE_B32 $vgpr0
$agpr2 = COPY $agpr0
$agpr3 = COPY $agpr0
$agpr4 = COPY $agpr0
Machine-cp propagated COPY $agpr0 to replace $vgpr0 creating 3 AGPR to AGPR
copys. Later this creates a cross-register copy from AGPR->VGPR->AGPR for each
copy when the prior VGPR->AGPR copy was already optimal.
Reviewed By: lkail, rampitec
Differential Revision: https://reviews.llvm.org/D108011
Currently isReallyTriviallyReMaterializableGeneric() implementation
prevents rematerialization on any virtual register use on the grounds
that is not a trivial rematerialization and that we do not want to
extend liveranges.
It appears that LRE logic does not attempt to extend a liverange of
a source register for rematerialization so that is not an issue.
That is checked in the LiveRangeEdit::allUsesAvailableAt().
The only non-trivial aspect of it is accounting for tied-defs which
normally represent a read-modify-write operation and not rematerializable.
The test for a tied-def situation already exists in the
/CodeGen/AMDGPU/remat-vop.mir,
test_no_remat_v_cvt_f32_i32_sdwa_dst_unused_preserve.
The change has affected ARM/Thumb, Mips, RISCV, and x86. For the targets
where I more or less understand the asm it seems to reduce spilling
(as expected) or be neutral. However, it needs a review by all targets'
specialists.
Differential Revision: https://reviews.llvm.org/D106408
One of the cases identified in PR45116 - we don't need to limit load combines to ABI alignment, we can use allowsMemoryAccess - which tests using getABITypeAlign, but also checks if a target permits (fast) misaligned memory loads by checking allowsMisalignedMemoryAccesses as a fallback.
One of the cases identified in PR45116 - we don't need to limit load combines (in this case for fp->int load/store copies) to ABI alignment, we can use allowsMemoryAccess - which tests using getABITypeAlign, but also checks if a target permits (fast) misaligned memory loads by checking allowsMisalignedMemoryAccesses as a fallback.
Differential Revision: https://reviews.llvm.org/D108318
One of the cases identified in PR45116 - we don't need to limit load combines (in this case for ISD::BUILD_PAIR) to ABI alignment, we can use allowsMemoryAccess - which tests using getABITypeAlign, but also checks if a target permits (fast) misaligned memory loads by checking allowsMisalignedMemoryAccesses as a fallback.
This helps in particular for 32-bit X86 cases loading 64-bit size data, reducing codegen diffs vs x86_64.
Differential Revision: https://reviews.llvm.org/D108307
D106408 enables rematerialization of instructions with virtual
register uses. That has uncovered the bug in the allUsesAvailableAt
implementation: https://bugs.llvm.org/show_bug.cgi?id=51516.
In the majority of cases canRematerializeAt() called to check if
an instruction can be rematerialized before the given UseIdx.
However, SplitEditor::enterIntvAtEnd() calls it to rematerialize
an instruction at the end of a block passing LIS.getMBBEndIdx()
into the check. In the testcase from the bug it has attempted to
rematerialize ADDXri after STRXui in bb.17. The use operand %55
of the ADD is killed by the STRX but that is undetected by the check
because it adjusts passed UseIdx to the reg slot, before the kill.
The value is dead at the index passed to the check however.
This change uses a later of passed UseIdx and its reg slot. This
shall be correct because if are checking an availability of operands
before an instruction that instruction cannot be the one defining
these operands. If we are checking for late rematerialization we
are really interested if operands live past the instruction.
The bug is not exploitable without D106408 but needed to reland
reverted D106408.
Differential Revision: https://reviews.llvm.org/D108475
Basically the same as G_LROUND. Handles the llvm.llround family of intrinsics.
Also add a helper function to the MachineVerifier for checking if all of the
(virtual register) operands of an instruction are scalars. Seems like a useful
thing to have.
Differential Revision: https://reviews.llvm.org/D108429
Issue Details:
The addresses for SEH tables for Windows are incorrect as 1 was unconditionally being added to all addresses. +1 is required for the SEH end address (as it is exclusive), but the SEH start addresses is inclusive and so should be used as-is.
In the IP2State tables, the addresses are +1 for AMD64 to handle the return address for a call being after the actual call instruction but are as-is for ARM and ARM64 as the `StateFromIp` function in the VC runtime automatically takes this into account and adjusts the address that is it looking up.
Fix Details:
* Split the `getLabel` function into two: `getLabel` (used for the SEH start address and ARM+ARM64 IP2State addresses) and `getLabelPlusOne` (for the SEH end address, and AMD64 IP2State addresses).
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D107784
This patch makes InstrRefBasedLDV "safe" to work with DBG_VALUE_LISTs. It
doesn't actually interpret them, but it recognises that they specify
variable locations and avoids propagating false locations, which is better
than the current state. Observe the attached tes
* We avoid propagating DBG_VALUE_LISTs into successor blocks, as they're
not "currently" supported,
* We don't propagate other variable locations across DBG_VALUE_LISTs,
because we know that the variable location is terminated by the
DBG_VALUE_LIST.
Differential Revision: https://reviews.llvm.org/D108143
This patch removes an assertion, and adds a regression test showing why the
assertion is broken.
For context, LocIdx is a key/index number for machine locations, so that we
can describe locations as a single integer and ignore whether they're on
the stack, in registers or otherwise. Back when InstrRefBasedLDV was added,
I happened to bake in a "special" zero number for various reasons, which
Vedant identified as undesirable in this review comment:
https://reviews.llvm.org/D83047#inline-765495 . I subsequently removed that
special zero number, but it looks like I didn't delete this assertion at
the time, which assumes that a zero LocIdx is invalid.
The attached test shows that this assertion is reachable on valid code --
on x86 $rsp always gets the LocIdx number zero, and if you transfer a
variable value into it, InstrRefBasedLDV crashes on that assertion. The
code might be a bit wild to be storing variables to $rsp like that, however
we shouldn't crash on it.
Differential Revision: https://reviews.llvm.org/D108134
Produce remarks when atomic instructions are expanded into hardware instructions
in SIISelLowering.cpp. Currently, these remarks are only emitted for atomic fadd
instructions.
Differential Revision: https://reviews.llvm.org/D108150
Translate the `@llvm.lround.*` family to G_LROUND via
`IRTranslator::translateSimpleIntrinsic`.
Differential Revision: https://reviews.llvm.org/D108418
For some reductions like G_VECREDUCE_OR on AArch64, we need to scalarize
completely if the source is <= 64b. This change adds support for that in
the legalizer. If the source has a pow-2 num elements, then we can do
a tree reduction using the scalar operation in the individual elements.
Otherwise, we just create a sequential chain of operations.
For AArch64, we only need to scalarize if the input is <64b. If it's great than
64b then we can first do a fewElements step to 64b, taking advantage of vector
instructions until we reach the point of scalarization.
I also had to relax the verifier checks for reductions because the intrinsics
support <1 x EltTy> types, which we lower to scalars for GlobalISel.
Differential Revision: https://reviews.llvm.org/D108276
This changes the lowering of saddsat and ssubsat so that instead of
using:
r,o = saddo x, y
c = setcc r < 0
s = c ? INTMAX : INTMIN
ret o ? s : r
into using asr and xor to materialize the INTMAX/INTMIN constants:
r,o = saddo x, y
s = ashr r, BW-1
x = xor s, INTMIN
ret o ? x : r
https://alive2.llvm.org/ce/z/TYufgD
This seems to reduce the instruction count in most testcases across most
architectures. X86 has some custom lowering added to compensate for
cases where it can increase instruction count.
Differential Revision: https://reviews.llvm.org/D105853
Previously we pre-calculated this and cached it for every
instruction in the function. Most of the calculated results will
never be used. So instead calculate it only on the first use, and
then cache it.
The cache was originally added to fix a compile time issue which
caused r216066 to be reverted.
This change exposed that we weren't pre-computing the Value for
Arguments. I've explicitly disabled that for now as it seemed to
regress some tests on AArch64 which has sext built into its compare
instructions.
Spotted while investigating how to improve heuristics to work better
with RISCV preferring sign extend for unsigned compares for i32 on RV64.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D107976
This patch adds the beginnings of more thorough support in the
legalizers for vector-predicated (VP) operations.
The first step is the ability to widen illegal vectors. The more
complicated scenario in which the result/operands need widening but the
mask doesn't has not been handled here. That would require a lot of code
without an in-tree target on which to test it.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D107904
This patch implements Flow Sensitive Sample FDO (FSAFDO) profile
loader. We have two profile loaders for FS profile,
one before RegAlloc and one before BlockPlacement.
To enable it, when -fprofile-sample-use=<profile> is specified,
add "-enable-fs-discriminator=true \
-disable-ra-fsprofile-loader=false \
-disable-layout-fsprofile-loader=false"
to turn on the FS profile loaders.
Differential Revision: https://reviews.llvm.org/D107878
Translate the `@llvm.isnan` intrinsic to G_ISNAN when we see it.
This is pretty much the same as the associated SelectionDAGBuilder code. Main
difference is that we don't expand it here. It makes more sense to do that
during legalization in GlobalISel. GlobalISel will just legalize the generated
illegal types.
Differential Revision: https://reviews.llvm.org/D108226
Add a generic opcode equivalent to the `llvm.isnan` intrinsic +
MachineVerifier support for it.
We need an opcode here because we may want target-specific lowering later on.
Differential Revision: https://reviews.llvm.org/D108222
It was introduced in 1a6dc92 and only enabled on PowerPC/AMDGPU. That
should be enabled for all targets.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D108010
This patch adds vector-predicated ("VP") reduction intrinsics corresponding to
each of the existing unpredicated `llvm.vector.reduce.*` versions. Unlike the
unpredicated reductions, all VP reductions have a start value. This start value
is returned when the no vector element is active.
Support for expansion on targets without native vector-predication support is
included.
This patch is based on the ["reduction
slice"](https://reviews.llvm.org/D57504#1732277) of the LLVM-VP reference patch
(https://reviews.llvm.org/D57504).
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D104308
Combine two G_PTR_ADDs, but keep the register bank of the constant.
That way, the combine can be used in post-regbank-select combines.
Introduce two helper methods in CombinerHelper, getRegBank and
setRegBank that get and set an optional register bank to a register.
That way, they can be used before and after register bank selection.
Differential Revision: https://reviews.llvm.org/D103326
In current implementation, the instruction to be sunk will be inserted before the target instruction without considering the def-use tree,
which may case Instruction does not dominate all uses error. We need to choose a suitable location to insert according to the use chain
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D107262
This reapplies 54a61c94f9, its follow up in 547b712500, which were
reverted 95fe61e639. Original commit message:
VarLoc based LiveDebugValues will abandon variable location propagation if
there are too many blocks and variable assignments in the function. If it
didn't, and we had (say) 1000 blocks and 1000 variables in scope, we'd end
up with 1 million DBG_VALUEs just at the start of blocks.
Instruction-referencing LiveDebugValues should honour this limitation too
(because the same limitation applies to it). Hoist the relevant command
line options into LiveDebugValues.cpp and pass it down into the
implementation classes as an argument to ExtendRanges. I've duplicated all
the run-lines in live-debug-values-cutoffs.mir to have an
instruction-referencing flavour.
Differential Revision: https://reviews.llvm.org/D107823
Basic block pointer is dereferenced unconditionally for MBBs with
hasAddressTaken property.
MBBs might have hasAddressTaken property without reference to BB.
Backend developers must assign fake BB to MBB to workaround this issue
and it should be fixed.
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D108092
Currently isReallyTriviallyReMaterializableGeneric() implementation
prevents rematerialization on any virtual register use on the grounds
that is not a trivial rematerialization and that we do not want to
extend liveranges.
It appears that LRE logic does not attempt to extend a liverange of
a source register for rematerialization so that is not an issue.
That is checked in the LiveRangeEdit::allUsesAvailableAt().
The only non-trivial aspect of it is accounting for tied-defs which
normally represent a read-modify-write operation and not rematerializable.
The test for a tied-def situation already exists in the
/CodeGen/AMDGPU/remat-vop.mir,
test_no_remat_v_cvt_f32_i32_sdwa_dst_unused_preserve.
The change has affected ARM/Thumb, Mips, RISCV, and x86. For the targets
where I more or less understand the asm it seems to reduce spilling
(as expected) or be neutral. However, it needs a review by all targets'
specialists.
Differential Revision: https://reviews.llvm.org/D106408
Check if a remateralizable nstruction does not have any virtual
register uses. Even though rematerializable RA might not actually
rematerialize it in this scenario. In that case we do not want to
hoist such instruction out of the loop in a believe RA will sink
it back if needed.
This already has impact on AMDGPU target which does not check for
this condition in its isTriviallyReMaterializable implementation
and have instructions with virtual register uses enabled. The
other targets are not impacted at this point although will be when
D106408 lands.
Differential Revision: https://reviews.llvm.org/D107677
SwitchInst should have a void result type.
Add a check to the verifier to catch this error.
Reviewed By: samparker
Differential Revision: https://reviews.llvm.org/D108084
Follow-up to D107068, attempt to fold nested concat_vectors/undefs, as long as both the vector and inner subvector types are legal.
This exposed the same issue in ARM's MVE LowerCONCAT_VECTORS_i1 (raised as PR51365) and AArch64's performConcatVectorsCombine which both assumed concat_vectors only took 2 subvector operands.
Differential Revision: https://reviews.llvm.org/D107597
VarLoc based LiveDebugValues will abandon variable location propagation if
there are too many blocks and variable assignments in the function. If it
didn't, and we had (say) 1000 blocks and 1000 variables in scope, we'd end
up with 1 million DBG_VALUEs just at the start of blocks.
Instruction-referencing LiveDebugValues should honour this limitation too
(because the same limitation applies to it). Hoist the relevant command
line options into LiveDebugValues.cpp and pass it down into the
implementation classes as an argument to ExtendRanges. I've duplicated all
the run-lines in live-debug-values-cutoffs.mir to have an
instruction-referencing flavour.
Differential Revision: https://reviews.llvm.org/D107823
visitEXTRACT_SUBVECTOR can sometimes create illegal BITCASTs when
removing "redundant" INSERT_SUBVECTOR operations. This patch adds
an extra check to ensure such combines only occur after operation
legalisation if any resulting BITBAST is itself legal.
Differential Revision: https://reviews.llvm.org/D108086
This is a fairly common pattern:
```
%mask = G_CONSTANT iN <mask val>
%add = G_ADD %lhs, %rhs
%and = G_AND %add, %mask
```
We have combines to eliminate G_AND with a mask that does nothing.
If we combined the above to this:
```
%mask = G_CONSTANT iN <mask val>
%narrow_lhs = G_TRUNC %lhs
%narrow_rhs = G_TRUNC %rhs
%narrow_add = G_ADD %narrow_lhs, %narrow_rhs
%ext = G_ZEXT %narrow_add
%and = G_AND %ext, %mask
```
We'd be able to take advantage of those combines using the trunc + zext.
For this to work (or be beneficial in the best case)
- The operation we want to narrow then widen must only be used by the G_AND
- The G_TRUNC + G_ZEXT must be free
- Performing the operation at a narrower width must not produce a different
value than performing it at the original width *after masking.*
Example comparison between SDAG + GISel: https://godbolt.org/z/63jzb1Yvj
At -Os for AArch64, this is a 0.2% code size improvement on CTMark/pairlocalign.
Differential Revision: https://reviews.llvm.org/D107929
AttributeList::hasAttribute() is confusing, use clearer methods like
hasParamAttr()/hasRetAttr().
Add hasRetAttr() since it was missing from AttributeList.
We may use several COPY instructions to copy the needed sub-registers
during split. But the way we split the lanes during the COPYs may be
different from the subranges of the old register. This would fail when we
extend the subranges of the new register because the LaneMasks do not
match exactly between subranges of new register and old register.
Since we are bundling the COPYs, I think there is no need to further refine the
subranges of the new register based on the set of LaneMasks of the inserted COPYs.
I am not sure if there will be further breaking cases. But as the subranges of
new register are created based on the LaneMasks of the subranges of old register,
it will be highly possible we will always find an exact LaneMask match.
We can think about how to make the extendPHIKillRanges() work for
subrange mask mismatch case if we meet more such cases in the future.
The test case was from D105065 by @arsenm.
Differential Revision: https://reviews.llvm.org/D107829
This patch adds Pass1 of MIRADDFSDiscriminatorsPass before register
allocation, and Pass2 of MIRAddFSDiscriminatorsPass before
Block-Placement. This is still under --enable-fs-discrmininator
option (default false).
This would reduce the turn-around time for FSAFDO transition.
Differential Revision: https://reviews.llvm.org/D104579
The introduction of `SHF_GNU_RETAIN` has caused massive problems on Solaris.
Initially, as reported in Bug 49437, it caused dozens of testsuite failures
on both sparc and x86. The objects were marked as `ELFOSABI_NONE`, but
`SHF_GNU_RETAIN` is a GNU extension. In the native Solaris ABI, that flag
(in the range for OS-specific values) is `SHF_SUNW_ABSENT` with a
completely different semantics, which confuses Solaris `ld` very much.
Later, the objects became (correctly) marked `ELFOSABI_GNU`, which Solaris
`ld` doesn't support, causing it to SEGV and break the build. The linker
is currently being hardened to not accept non-native OS ABIs to avoid this.
The need for linker support is already documented in
`clang/include/clang/Basic/AttrDocs.td`, but not currently checked.
This patch avoids all this by not emitting `SHF_GNU_RETAIN` on Solaris at all.
Tested on `amd64-pc-solaris2.11`, `sparcv9-sun-solaris2.11`, and
`x86_64-pc-linux-gnu`.
Differential Revision: https://reviews.llvm.org/D107747
We were calling find and then using operator[]. Instead keep the
iterator from find and use it to get the value.
Just happened to notice while investigating how we decide what extends
to use between basic blocks.
Some files still contained the old University of Illinois Open Source
Licence header. This patch replaces that with the Apache 2 with LLVM
Exception licence.
Differential Revision: https://reviews.llvm.org/D107528
This patch refactors / simplifies salvageDebugInfoImpl(). The goal
here is to simplify the implementation of coro::salvageDebugInfo() in
a followup patch.
1. Change the return value to I.getOperand(0). Currently users of
salvageDebugInfoImpl() assume that the first operand is
I.getOperand(0). This patch makes this information explicit. A
nice side-effect of this change is that it allows us to salvage
expressions such as add i8 1, %a in the future.
2. Factor out the creation of a DIExpression and return an array of
DIExpression operations instead. This change allows users that
call salvageDebugInfoImpl() in a loop to avoid the costly
creation of temporary DIExpressions and to defer the creation of
a DIExpression until the end.
This patch does not change any functionality.
rdar://80227769
Differential Revision: https://reviews.llvm.org/D107383
We may call lowerRelativeReference in MC to determine whether target
supports this lowering. We should return nullptr instead of crashing
when we haven't implemented the real lowering.
Reviewed By: hubert.reinterpretcast
Differential Revision: https://reviews.llvm.org/D107830
If a G_SHL is fed by a G_CONSTANT, the lower and upper bits of the source can be
shifted individually by the constant shift amount.
However in case the shift amount came from a G_TRUNC(G_CONSTANT), the generic shift legalization
code was used, producing intermediate shifts that are potentially illegal on some targets.
This change teaches narrowScalarShift to look through G_TRUNCs and G_*EXTs.
Reviewed By: paquette
Differential Revision: https://reviews.llvm.org/D89100
This patch is a revert of e08f205f5c. In that patch, DW_TAG_subprograms
were permitted to be referenced across CU boundaries, to improve stack
trace construction using call site information. Unfortunately, as
documented in PR48790, the way that subprograms are "owned" by dwarf units
is sufficiently complicated that subprograms end up in unexpected units,
invalidating cross-unit references.
There's no obvious way to easily fix this, and several attempts have
failed. Revert this to ensure correct DWARF is always emitted.
Three tests change in addition to the reversion, but they're all very
light alterations.
Differential Revision: https://reviews.llvm.org/D107076
We should use MachineInstr::eraseFromParentAndMarkDBGValuesForRemoval()
instead of eraseFromParent().
We should probably use that in other places too but fix this issue which
affects clang bootstrap builds for now.
This commit adds the isnan intrinsic and provides a default expansion
for it in the SDAG. However, it makes the assumption that types
it operates on are IEEE-compliant types. This is not always the case.
An example of that is PPC "double double" which has a representation
that
- Does not need to conform to IEEE requirements for isnan as it is
not an IEEE-compliant type
- Does not have a representation that allows for straightforward
reinterpreting as an integer and use of integer operations
The result was that this commit broke __builtin_isnan for ppc_fp128
making many valid numeric values report a NaN.
This patch simply changes the expansion to always expand to unordered
comparison (regardless of whether FP exceptions are tracked). This
is inline with previous semantics.
This isn't optimal, but prevents crashing when the libcall isn't
available. It just calculates the full product and makes sure the high bits
match the sign of the low half. Each of the pieces should go through their own
type legalization.
This can make D107420 unnecessary.
Needs tests, but I wanted to start discussion about D107420.
Reviewed By: FreddyYe
Differential Revision: https://reviews.llvm.org/D107581
This is recommit of the patch 16ff91ebcc,
reverted in 0c28a7c990 because it had
an error in call of getFastMathFlags (base type should be FPMathOperator
but not Instruction). The original commit message is duplicated below:
Clang has builtin function '__builtin_isnan', which implements C
library function 'isnan'. This function now is implemented entirely in
clang codegen, which expands the function into set of IR operations.
There are three mechanisms by which the expansion can be made.
* The most common mechanism is using an unordered comparison made by
instruction 'fcmp uno'. This simple solution is target-independent
and works well in most cases. It however is not suitable if floating
point exceptions are tracked. Corresponding IEEE 754 operation and C
function must never raise FP exception, even if the argument is a
signaling NaN. Compare instructions usually does not have such
property, they raise 'invalid' exception in such case. So this
mechanism is unsuitable when exception behavior is strict. In
particular it could result in unexpected trapping if argument is SNaN.
* Another solution was implemented in https://reviews.llvm.org/D95948.
It is used in the cases when raising FP exceptions by 'isnan' is not
allowed. This solution implements 'isnan' using integer operations.
It solves the problem of exceptions, but offers one solution for all
targets, however some can do the check in more efficient way.
* Solution implemented by https://reviews.llvm.org/D96568 introduced a
hook 'clang::TargetCodeGenInfo::testFPKind', which injects target
specific code into IR. Now only SystemZ implements this hook and it
generates a call to target specific intrinsic function.
Although these mechanisms allow to implement 'isnan' with enough
efficiency, expanding 'isnan' in clang has drawbacks:
* The operation 'isnan' is hidden behind generic integer operations or
target-specific intrinsics. It complicates analysis and can prevent
some optimizations.
* IR can be created by tools other than clang, in this case treatment
of 'isnan' has to be duplicated in that tool.
Another issue with the current implementation of 'isnan' comes from the
use of options '-ffast-math' or '-fno-honor-nans'. If such option is
specified, 'fcmp uno' may be optimized to 'false'. It is valid
optimization in general, but it results in 'isnan' always returning
'false'. For example, in some libc++ implementations the following code
returns 'false':
std::isnan(std::numeric_limits<float>::quiet_NaN())
The options '-ffast-math' and '-fno-honor-nans' imply that FP operation
operands are never NaNs. This assumption however should not be applied
to the functions that check FP number properties, including 'isnan'. If
such function returns expected result instead of actually making
checks, it becomes useless in many cases. The option '-ffast-math' is
often used for performance critical code, as it can speed up execution
by the expense of manual treatment of corner cases. If 'isnan' returns
assumed result, a user cannot use it in the manual treatment of NaNs
and has to invent replacements, like making the check using integer
operations. There is a discussion in https://reviews.llvm.org/D18513#387418,
which also expresses the opinion, that limitations imposed by
'-ffast-math' should be applied only to 'math' functions but not to
'tests'.
To overcome these drawbacks, this change introduces a new IR intrinsic
function 'llvm.isnan', which realizes the check as specified by IEEE-754
and C standards in target-agnostic way. During IR transformations it
does not undergo undesirable optimizations. It reaches instruction
selection, where is lowered in target-dependent way. The lowering can
vary depending on options like '-ffast-math' or '-ffp-model' so the
resulting code satisfies requested semantics.
Differential Revision: https://reviews.llvm.org/D104854
Fixes issue where late materialized constants can be more strictly
aligned then their containing csect.
Differential Revision: https://reviews.llvm.org/D103103
We don't have real demanded bits support for MULHU, but we can
still use the known bits based constant folding support at the end
of SimplifyDemandedBits to simplify a MULHU. This helps with cases
where we know the LHS and RHS have enough leading zeros so that
the high multiply result is always 0.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D106471
IR typically creates INSERT_SUBVECTOR patterns as a widening of the subvector with undefs to pad to the destination size, followed by a shuffle for the actual insertion - SelectionDAGBuilder has to do something similar for shuffles when source/destination vectors are different sizes.
This combine attempts to recognize these patterns by looking for a shuffle of a subvector (from a CONCAT_VECTORS) that starts at a modulo of its size into an otherwise identity shuffle of the base vector.
This uncovered a couple of target-specific issues as we haven't often created INSERT_SUBVECTOR nodes in generic code - aarch64 could only handle insertions into the bottom of undefs (i.e. a vector widening), and x86-avx512 vXi1 insertion wasn't keeping track of undef elements in the base vector.
Fixes PR50053
Differential Revision: https://reviews.llvm.org/D107068
It's entirely possible (because it actually happened) for a bool
variable to end up with a 256-bit DW_AT_const_value. This came about
when a local bool variable was initialized from a bitfield in a
32-byte struct of bitfields, and after inlining and constant
propagation, the variable did have a constant value. The sequence of
optimizations had it carrying "i256" values around, but once the
constant made it into the llvm.dbg.value, no further IR changes could
affect it.
Technically the llvm.dbg.value did have a DIExpression to reduce it
back down to 8 bits, but the compiler is in no way ready to emit an
oversized constant *and* a DWARF expression to manipulate it.
Depending on the circumstances, we had either just the very fat bool
value, or an expression with no starting value.
The sequence of optimizations that led to this state did seem pretty
reasonable, so the solution I came up with was to invent a DWARF
constant expression folder. Currently it only does convert ops, but
there's no reason it couldn't do other ops if that became useful.
This broke three tests that depended on having convert ops survive
into the DWARF, so I added an operator that would abort the folder to
each of those tests.
Differential Revision: https://reviews.llvm.org/D106915
Instructions that produceSameValue produce same values for operands with
same index. matchEqualDefs used to return true for any two values from
different instructions that produce same values. Fix this by checking if
values are defined by operands with the same index.
Differential Revision: https://reviews.llvm.org/D107362
The LegalizeAction for this node should follow the logic for
`VECREDUCE_SEQ_FADD` and be determined using the vector operand's type.
here isn't an in-tree target that makes use of this, but I think it's safe to
say this is how it should behave, should a target want to customize the action
for this node.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D107478
to `lib/CodeGen/CommandFlags.cpp`. It can replace
-x86-experimental-pref-loop-alignment=.
The loop alignment is only used by MachineBlockPlacement.
The implementation uses a new `llvm::TargetOptions` for now, as
an IR function attribute/module flags metadata may be overkill.
This is the llvm part of D106701.
This allows special constants like to 0 to be recognized. It's also
expected by isel patterns if a target had a mulh with immediate instructions.
The commuting done by tablegen won't commute patterns with immediates since it
expects DAGCombine to have done it.
Reviewed By: foad
Differential Revision: https://reviews.llvm.org/D107486
This attempts to make more of RDA aware of potentially overlapping
subregisters. Some of this was already in place, with it iterating
through MCRegUnitIterators. This also replaces calls to
LiveRegs.contains(..) with !LiveRegs.available(..), and updates the
isValidRegUseOf and isValidRegDefOf to search subregs.
Differential Revision: https://reviews.llvm.org/D107351
Clang has builtin function '__builtin_isnan', which implements C
library function 'isnan'. This function now is implemented entirely in
clang codegen, which expands the function into set of IR operations.
There are three mechanisms by which the expansion can be made.
* The most common mechanism is using an unordered comparison made by
instruction 'fcmp uno'. This simple solution is target-independent
and works well in most cases. It however is not suitable if floating
point exceptions are tracked. Corresponding IEEE 754 operation and C
function must never raise FP exception, even if the argument is a
signaling NaN. Compare instructions usually does not have such
property, they raise 'invalid' exception in such case. So this
mechanism is unsuitable when exception behavior is strict. In
particular it could result in unexpected trapping if argument is SNaN.
* Another solution was implemented in https://reviews.llvm.org/D95948.
It is used in the cases when raising FP exceptions by 'isnan' is not
allowed. This solution implements 'isnan' using integer operations.
It solves the problem of exceptions, but offers one solution for all
targets, however some can do the check in more efficient way.
* Solution implemented by https://reviews.llvm.org/D96568 introduced a
hook 'clang::TargetCodeGenInfo::testFPKind', which injects target
specific code into IR. Now only SystemZ implements this hook and it
generates a call to target specific intrinsic function.
Although these mechanisms allow to implement 'isnan' with enough
efficiency, expanding 'isnan' in clang has drawbacks:
* The operation 'isnan' is hidden behind generic integer operations or
target-specific intrinsics. It complicates analysis and can prevent
some optimizations.
* IR can be created by tools other than clang, in this case treatment
of 'isnan' has to be duplicated in that tool.
Another issue with the current implementation of 'isnan' comes from the
use of options '-ffast-math' or '-fno-honor-nans'. If such option is
specified, 'fcmp uno' may be optimized to 'false'. It is valid
optimization in general, but it results in 'isnan' always returning
'false'. For example, in some libc++ implementations the following code
returns 'false':
std::isnan(std::numeric_limits<float>::quiet_NaN())
The options '-ffast-math' and '-fno-honor-nans' imply that FP operation
operands are never NaNs. This assumption however should not be applied
to the functions that check FP number properties, including 'isnan'. If
such function returns expected result instead of actually making
checks, it becomes useless in many cases. The option '-ffast-math' is
often used for performance critical code, as it can speed up execution
by the expense of manual treatment of corner cases. If 'isnan' returns
assumed result, a user cannot use it in the manual treatment of NaNs
and has to invent replacements, like making the check using integer
operations. There is a discussion in https://reviews.llvm.org/D18513#387418,
which also expresses the opinion, that limitations imposed by
'-ffast-math' should be applied only to 'math' functions but not to
'tests'.
To overcome these drawbacks, this change introduces a new IR intrinsic
function 'llvm.isnan', which realizes the check as specified by IEEE-754
and C standards in target-agnostic way. During IR transformations it
does not undergo undesirable optimizations. It reaches instruction
selection, where is lowered in target-dependent way. The lowering can
vary depending on options like '-ffast-math' or '-ffp-model' so the
resulting code satisfies requested semantics.
Differential Revision: https://reviews.llvm.org/D104854
- Rename `wasm.catch` intrinsic to `wasm.catch.exn`, because we are
planning to add a separate `wasm.catch.longjmp` intrinsic which
returns two values.
- Rename several variables
- Remove an unnecessary parameter from `canLongjmp` and `isEmAsmCall`
from LowerEmscriptenEHSjLj pass
- Add `-verify-machineinstrs` in a test for a safety measure
- Add more comments + fix some errors in comments
- Replace `std::vector` with `SmallVector` for cases likely with small
number of elements
- Renamed `EnableEH`/`EnableSjLj` to `EnableEmEH`/`EnableEmSjLj`: We are
soon going to add `EnableWasmSjLj`, so this makes the distincion
clearer
Reviewed By: tlively
Differential Revision: https://reviews.llvm.org/D107405
Previously we would emit constant pool entries for ldr inline asm at the
very end of AsmPrinter::doFinalization(). However, if we're emitting
dwarf aranges, that would end all sections with aranges. Then if we have
constant pool entries to be emitted in those same sections, we'd hit an
assert that the section has already been ended.
We want to emit constant pool entries before emitting dwarf aranges.
This patch splits out arm32/64's constant pool entry emission into its
own MCTargetStreamer virtual method.
Fixes PR51208
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D107314
We had some similar hasOneUse/isNON_EXTLoad early-outs spread out over different parts of the method - we should pull them all together.
Noticed while triaging PR45116
This adds handling for two cases:
1. A scalable vector where the element type is promoted.
2. A scalable vector where the element count is odd (or more generally,
not divisble by the element count of the part type).
(Some element types still don't work; for example, <vscale x 2 x i128>,
or <vscale x 2 x fp128>.)
Differential Revision: https://reviews.llvm.org/D105591
We might want to use info from GC strategy in middle end analysis.
The motivation for this is provided in D99135: we may want to ask
a GC if it's going to work with a given pointer (currently this code
makes naive check by the method name).
Differetial Revision: https://reviews.llvm.org/D100559
Reviewed By: reames
If all demanded elements of the BUILD_VECTOR pass a isGuaranteedNotToBeUndefOrPoison check, then we can treat this specific demanded use of the BUILD_VECTOR as guaranteed not to be undef or poison either.
Differential Revision: https://reviews.llvm.org/D107174
This patch legalizes the Machine Value Type introduced in D94096 for loads
and stores. A new target hook named getAsmOperandValueType() is added which
maps i512 to MVT::i64x8. GlobalISel falls back to DAG for legalization.
Differential Revision: https://reviews.llvm.org/D94097
Adds MVT::i64x8, a Machine Value Type needed for lowering inline assembly
operands which materialize a sequence of eight general purpose registers.
Differential Revision: https://reviews.llvm.org/D94096
This could be smarter by picking an ideal type, or at least splitting
the vector in half first. Also handles lower for non-power-of-2,
non-extending vector loads.
Currently this just avoids failing to legalize some odd vector AMDGPU
tests, but is a step towards removing the split logic from the
NarrowScalar logic.
The code for splitting an unaligned access into 2 pieces is
essentially the same as for splitting a non-power-of-2 load for
scalars. It would be better to pick an optimal memory access size and
directly use it, but splitting in half is what the DAG does.
As-is this fixes handling of some unaligned sextload/zextloads for
AMDGPU. In the future this will help drop the ugly abuse of
narrowScalar to handle splitting unaligned accesses.
This patch prevents GlobalISel from optimizing out redundant branch
instructions when compiling without optimizations.
The motivating example is code like the following common pattern in
Swift, where users expect to be able to set a breakpoint on the early
exit:
public func f(b: Bool) {
guard b else {
return // I would like to set a breakpoint here.
}
...
}
The patch modifies two places in GlobalISEL: The first one is in
IRTranslator.cpp where the removal of redundant branches is made
conditional on the optimization level. The second one is in
AArch64InstructionSelector.cpp where an -O0 *only* optimization is
being removed.
Disabling these optimizations increases code size at -O0 by
~8%. However, doing so improves debuggability, and debug builds are
the primary reason why developers compile without optimizations. We
thus concluded that this is the right trade-off.
rdar://79515454
This tenatively reapplies the patch without modifications, the LLDB
test that has blocked this from landing previously has since been
modified to hopefully no longer be sensitive to this change.
Differential Revision: https://reviews.llvm.org/D105238
If a target lists both a subreg and a superreg in a callee-saved
register mask, the prolog will spill both aliasing registers. Instead,
don't spill the subreg if a superreg is being spilled. This case is hit by the
PowerPC SPE code, as well as a modified RISC-V backend for CHERI I maintain out
of tree.
Reviewed By: jhibbits
Differential Revision: https://reviews.llvm.org/D73170
This transform was added with D58874, but there were no tests for overflow ops.
We need to change this one way or another because it can crash as shown in:
https://llvm.org/PR51238
Note that if there are no uses of an overflow op's bool overflow result, we
reduce it to a regular math op, so we continue to fold that case either way.
If we have uses of both the math and the overflow bool, then we are likely
not saving anything by creating an independent sub instruction as seen in
the test diffs here.
This patch makes the behavior in SDAG consistent with what we do in
instcombine AFAICT.
Differential Revision: https://reviews.llvm.org/D106983
Function findBestLoopTopHelper tries to find a new loop top block which can also
fall through to OldTop, but it's impossible if OldTop is not a chain header, so
it should exit immediately.
Differential Revision: https://reviews.llvm.org/D106329
When we have a terminator sequence (i.e. a tailcall or return),
MIIsInTerminatorSequence is used to work out where the preceding ABI-setup
instructions end, i.e. the parts that were glued to the terminator
instruction. This allows LLVM to split blocks safely without having to
worry about ABI stuff.
The function only ignores DBG_VALUE instructions, meaning that the two
debug instructions I recently added can end terminator sequences early,
causing various MachineVerifier errors. This patch promotes the test for
debug instructions from "isDebugValue" to "isDebugInstr", thus avoiding any
debug-info interfering with this function.
Differential Revision: https://reviews.llvm.org/D106660
This patch adds a peephole optimization `SETCC(FREEZE(x),const)` => `FREEZE(SETCC(x,const))`
if the SETCC is only used by BRCOND.
Combined with `BRCOND(FREEZE(X)) => BRCOND(X)`, this leads to a nice improvement in the generated assembly when x is a masked loaded value.
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D105344
- This patch consists of the bare basic code needed in order to generate some assembly for the z/OS target.
- Only the .text and the .bss sections are added for now.
- The relevant MCSectionGOFF/Symbol interfaces have been added. This enables us to print out the GOFF machine code sections.
- This patch enables us to add simple lit tests wherever possible, and contribute to the testing coverage for the z/OS target
- Further improvements and additions will be made in future patches.
Reviewed By: tmatheson
Differential Revision: https://reviews.llvm.org/D106380
Avoid several crashes when DBG_INSTR_REF and DBG_PHI instructions are fed
to the instruction scheduler. DBG_INSTR_REFs should be treated like
DBG_LABELs, and just ignored for the purpose of scheduling [0].
DBG_PHIs however behave much more like DBG_VALUEs: they refer to register
operands, and if some register defs get shuffled around during instruction
scheduling, there's a risk that the debug instr will refer to the wrong
value. There's already a facility for updating DBG_VALUEs to reflect this;
add DBG_PHI to the list of instructions that it will update.
[0] Suboptimal, but it's what instr scheduling does right now.
Differential Revision: https://reviews.llvm.org/D106663
When working out which instruction defines a value, the
instruction-referencing variable location code has a few special cases for
physical registers:
* Arguments are never defined by instructions,
* Constant physical registers always read the same value, are never def'd
This patch adds a third case for the llvm.frameaddress intrinsics: you can
read the framepointer in any block if you so choose, and use it as a
variable location, as shown in the added test.
This rather violates one of the assumptions behind instruction referencing,
that LLVM-ir shouldn't be able to read from an arbitrary register at some
arbitrary point in the program. The solution for now is to just emit a
DBG_PHI that reads the register value: this works, but if we wanted to do
something clever with DBG_PHIs in the future then this would probably get
in the way. As it stands, this patch avoids a crash.
Differential Revision: https://reviews.llvm.org/D106659
This patch builds on top of D106575 in which scalable-vector splats were
supported in `ISD::matchBinaryPredicate`. It teaches the DAGCombiner how
to perform a variety of the pre-existing saturating add/sub combines on
scalable-vector types.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D106652
This adds support for the case where
WideSize = DstSize + K * SrcSize
In this case, we can pad the G_MERGE_VALUES instruction with K extra undef
values with width SrcSize. Then the destination can be handled via
widenScalarDst.
Differential Revision: https://reviews.llvm.org/D106814
Use it AArch64 post-legal combiner. These don't always get folded because when
the instructions are created the constants are obscured by artifacts.
Differential Revision: https://reviews.llvm.org/D106776
Dominator trees were previously used for an optimization related to
`wasm.lsda` but the optimization was removed in D97309. Currently
dominators are not doing anything in this pass. Also removes some
`include` lines without which it compiles.
Reviewed By: tlively
Differential Revision: https://reviews.llvm.org/D106811
This fixes an assert firing when compiling code which involves 128 bit
integrals.
This would trigger runtime checks similar to this:
```
Assertion failed: getMinSignedBits() <= 64 && "Too many bits for int64_t", file llvm/include/llvm/ADT/APInt.h, line 1646
```
To get around this, we just saturate those big values.
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D105320
During tail duplication, SSA values may be updated and have their uses
replaced with a virtual register, and any debug instructions that use
that value are deleted. This patch fixes the implementation of the debug
instruction deletion to work correctly for debug instructions that use
the SSA value multiple times, by batching deletions so that we don't
attempt to delete the same instruction twice.
Differential Revision: https://reviews.llvm.org/D106557
Late in SelectionDAG we join up instruction numbers with their defining
instructions, if it couldn't be done during the main part of SelectionDAG.
One exception is function arguments, where we have to point a DBG_PHI
instruction at the incoming live register, as they don't have a defining
instruction. This patch adds another exception, for constant physregs, like
aarch64 has.
It may seem wasteful to use two instructions where we could use a single
DBG_VALUE, however the whole point of instruction referencing is to
decouple the identification of values from the specification of where
variable location ranges start.
(Part of my aarch64 work to ease adoption of instruction referencing, as
in the meta comment on D104520)
Differential Revision: https://reviews.llvm.org/D104520
This patch extends support for (scalable-vector) splats in the
DAGCombiner via the `ISD::matchBinaryPredicate` function, which enable a
variety of simple combines of constants.
Users of this function may now have to distinguish between
`BUILD_VECTOR` and `SPLAT_VECTOR` vector operands. The way of dealing
with this in-tree follows the approach added for
`ISD::matchUnaryPredicate` implemented in D94501.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D106575
to encode the constants for DW_AT_data_member_location.
Summary: In DWARF v3, DW_FORM_data4/8 in
DW_AT_data_member_location are interpreted as location
list pointers. Interpreting constants as pointers is
not expected, so we use DW_FORM_udata to encode the
constants.
Reviewed By: probinson
Differential Revision: https://reviews.llvm.org/D105687
If value tracking can confirm that the cttz/ctlz source is known non-zero then we don't need to create a branch (which DAG will struggle to recover from).
Differential Revision: https://reviews.llvm.org/D106685
I've setup the basic framework for the isGuaranteedNotToBeUndefOrPoison call and updated DAGCombiner::visitFREEZE to use it, further Opcodes can be handled when we have test coverage.
I'm not aware of any vector test freeze coverage so the DemandedElts (and the Depth) args are not being used yet - but they are in place.
SelectionDAG::isGuaranteedNotToBePoison wrappers have also been added.
Differential Revision: https://reviews.llvm.org/D106668
This adds custom lowering for truncating stores when operating on
fixed length vectors in SVE. It also includes a DAG combine to
fold extends followed by truncating stores into non-truncating
stores in order to prevent this pattern appearing once truncating
stores are supported.
Currently truncating stores are not used in certain cases where
the size of the vector is larger than the target vector width.
Differential Revision: https://reviews.llvm.org/D104471
Reland of 31859f896.
This change implements new DAG notes GLOBAL_GET/GLOBAL_SET, and
lowering methods for load and stores of reference types from IR
globals. Once the lowering creates the new nodes, tablegen pattern
matches those and converts them to Wasm global.get/set.
Reviewed By: tlively
Differential Revision: https://reviews.llvm.org/D104797
This is part of a patch series working towards the ability to make
SourceLocation into a 64-bit type to handle larger translation units.
!srcloc is generated in clang codegen, and pulled back out by llvm
functions like AsmPrinter::emitInlineAsm that need to report errors in
the inline asm. From there it goes to LLVMContext::emitError, is
stored in DiagnosticInfoInlineAsm, and ends up back in clang, at
BackendConsumer::InlineAsmDiagHandler(), which reconstitutes a true
clang::SourceLocation from the integer cookie.
Throughout this code path, it's now 64-bit rather than 32, which means
that if SourceLocation is expanded to a 64-bit type, this error report
won't lose half of the data.
The compiler will tolerate both of i32 and i64 !srcloc metadata in
input IR without faulting. Test added in llvm/MC. (The semantic
accuracy of the metadata is another matter, but I don't know of any
situation where that matters: if you're reading an IR file written by
a previous run of clang, you don't have the SourceManager that can
relate those source locations back to the original source files.)
Original version of the patch by Mikhail Maltsev.
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D105491
Prior to this patch, it skipped the instruction defining VNI when checking if the tainted lanes are used.
In the given example, VRGATHER is an illegal instruction because its DstReg overlaps with SrcReg.
Therefore we need to check the defining instruction as well when there is an earlyclobber constraint.
Reviewed By: qcolombet
Differential Revision: https://reviews.llvm.org/D105684
The coalescer does not check if register uses are available
at the point of rematerialization. If it attempts to rematerialize
an instruction with such uses it can end up with use without a def.
LiveRangeEdit does such check during rematerialization, so just
call LiveRangeEdit::allUsesAvailableAt() to avoid the problem.
Differential Revision: https://reviews.llvm.org/D106396
The existing rule about the operand type is strange. Instead, just say
the operand is a TargetConstant with the right width. (Legalization
ignores TargetConstants, so it doesn't matter if that width is legal.)
Highlights:
1. I had to substantially rewrite the AArch64 isel patterns to expect a
TargetConstant. Nothing too exotic, but maybe a little hairy. Maybe
worth considering a target-specific node with some dagcombines instead
of this complicated nest of isel patterns.
2. Our behavior on RV32 for vectors of i64 has changed slightly. In
particular, we correctly preserve the width of the arithmetic through
legalization. This changes the DAG a bit. Maybe room for
improvement here.
3. I explicitly defined the behavior around overflow. This is necessary
to make the DAGCombine transforms legal, and I don't think it causes any
practical issues.
Differential Revision: https://reviews.llvm.org/D105673
This patch allows iterating typed enum via the ADT/Sequence utility.
It also changes the original design to better separate concerns:
- `StrongInt` only deals with safe `intmax_t` operations,
- `SafeIntIterator` presents the iterator and reverse iterator
interface but only deals with safe `StrongInt` internally.
- `iota_range` only deals with `SafeIntIterator` internally.
This design ensures that operations are always valid. In particular,
"Out of bounds" assertions fire when:
- the `value_type` is not representable as an `intmax_t`
- iterator operations make internal computation underflow/overflow
- the internal representation cannot be converted back to `value_type`
Differential Revision: https://reviews.llvm.org/D106279
We have SelectionDAG patterns for 8 & 16-bit atomic operations, but they
assume the value types will have been legalized to 32-bits. So this adds
the ability to widen them to both AArch64 & generic GISel
infrastructure.
In the textual format, `noduplicates` means no COMDAT/section group
deduplication is performed. Therefore, if both sets of sections are retained, and
they happen to define strong external symbols with the same names,
there will be a duplicate definition linker error.
In PE/COFF, the selection kind lowers to `IMAGE_COMDAT_SELECT_NODUPLICATES`.
The name describes the corollary instead of the immediate semantics. The name
can cause confusion to other binary formats (ELF, wasm) which have implemented/
want to implement the "no deduplication" selection kind. Rename it to be clearer.
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D106319
ACC registers are a combination of four consecutive vector registers.
If the vector registers are assigned first this often forces a number
of copies to appear just before the ACC register is created. If the ACC
register is assigned first then fewer copies are generated when the vector
registers are assigned.
This patch tries to force the register allocator to assign the ACC registers first
and then the UACC registers and then the vector pair registers. It does this
by changing the priority of the register classes.
This patch also adds hints to help the register allocator assign UACC registers from
known ACC registers and vector pair registers from known UACC registers.
Reviewed By: nemanjai
Differential Revision: https://reviews.llvm.org/D105854
This patch fixes a clearly-broken function that I absent-mindedly bodged
many months ago.
Over in D85749 I landed the substituteDebugValuesForInst, that creates
substitution records for all the def operands from one debug-labelled
instruction to the new one. Unfortunately it would crash if the two
instructions had different numbers of operands; I tried to fix this in
537f0fbe82 by adding a "max operand" parameter to the method, but then
didn't actually change the loop bound to take account of this. It passed
all the tests because.... well there wasn't any real test coverage of this
method.
This patch fixes up the loop to be bounded by the MaxOperand bound; and
adds test coverage for the x86-fixup-LEAs calls to this method, so that
it's actually tested.
Differential Revision: https://reviews.llvm.org/D105820
Although this combine checks that there's no load folding barriers between
the loads that it's trying to merge, it was inserting the load at the
MIRBuilder's default insertion point, which is the G_OR use inst.
This was causing a miscompile in the test suite's
SingleSource/Regression/C/gcc-c-torture/execute/GCC-C-execute-bswap-2
Differential Revision: https://reviews.llvm.org/D106251
RISCV would prefer a sign extended constant since that works better
with our constant materialization. We have an existing TLI hook we
use to control sign extension of setcc operands in type legalization.
That hook happens to do the right check we need here, but might be
straying from its original purpose. With only RISCV defining this
hook in tree, I wasn't sure if it was worth adding another hook
with identical behavior.
This is an alternative to D105785 where I tried to handle this in
the RISCV backend by not creating ANY_EXTENDs in some places.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D105918
This reverts commit 2a419a0b99.
The result of a shufflevector must not propagate poison from any element
other than the one noted in the shuffle mask.
The regressions outside of fptoui-may-overflow.ll can probably be
recovered some other way; for example, using isGuaranteedNotToBePoison.
See discussion on https://reviews.llvm.org/D106053 for more background.
Differential Revision: https://reviews.llvm.org/D106222
I'm going to extend the functionality started in D106058 so move the folds into their own method to reduce the amount of code in DAGCombiner::visitSELECT
llvm::KnownBits::byteSwap() and reverse() don't modify in-place, so
we weren't actually computing anything. This was causing a miscompile on an
arm64 stage2 bootstrap clang build.
s56 stores are broken down into s32 + s24 stores. During this step
both of those new stores use an anyextended s64 value, resulting in
truncating stores. With s56, the s24 requires another lower step to
make it legal, and we were crashing because we didn't expect non-pow-2
stores to also be truncating as well.
Differential Revision: https://reviews.llvm.org/D106183
This patch transforms the sequence
lea (reg1, reg2), reg3
sub reg3, reg4
to two sub instructions
sub reg1, reg4
sub reg2, reg4
Similar optimization can also be applied to LEA/ADD sequence.
The modifications to TwoAddressInstructionPass is to ensure the operands of ADD
instruction has expected order (the dest register of LEA should be src register
of ADD).
Differential Revision: https://reviews.llvm.org/D104684
Add an assertion that we've calling MaskedElementsAreZero with a vector op and that the DemandedElts arg is a matching width.
Makes the error a lot easier to grok when something else accidentally gets used.
If you attach __attribute__((optnone)) to a function when using
optimisations, that function will use fast-isel instead of the usual
SelectionDAG method. This is a problem for instruction referencing,
because it means DBG_VALUEs of virtual registers will be created,
triggering some safety assertions in LiveDebugVariables. Those assertions
exist to detect exactly this scenario, where an unexpected piece of code is
generating virtual register references in instruction referencing mode.
Fix this by transforming the DBG_VALUEs created by fast-isel into
half-formed DBG_INSTR_REFs, after which they get patched up in
finalizeDebugInstrRefs. The test modified adds a fast-isel mode to the
instruction referencing isel test.
Differential Revision: https://reviews.llvm.org/D105694
Since we're still building on top of the MVT based infrastructure, we
need to track the pointer type/address space on the side so we can end
up with the correct pointer LLTs when interpreting CCValAssigns.
This adds some level of type safety, allows helper functions to be added for
specific opcodes for free, and also allows us to succinctly check for class
membership with the usual dyn_cast/isa/cast functions.
To start off with, add variants for the different load/store operations with some
places using it.
Differential Revision: https://reviews.llvm.org/D105751
Similar to the folds performed in InstCombinerImpl::foldSelectOpOp, this attempts to push a select further up to help merge a pair of binops.
I'm primarily interested in select(cond,add(x,y),add(x,z)) folds to help expose pointer math (see https://bugs.llvm.org/show_bug.cgi?id=51069 etc.) but I've tried to use the more generic isBinOp().
Differential Revision: https://reviews.llvm.org/D106058
The linker can sometimes drop the do_not_dead_strip if it can't associate the
atom with a symbol (the other place to specify no dead-stripping in MachO
files).
This patch adds the forward scan for finding redundant DBG_VALUEs.
This analysis aims to remove redundant DBG_VALUEs by going forward
in the basic block by considering the first DBG_VALUE as a valid
until its first (location) operand is not clobbered/modified.
For example:
(1) DBG_VALUE $edi, !"var1", ...
(2) <block of code that does affect $edi>
(3) DBG_VALUE $edi, !"var1", ...
...
in this case, we can remove (3).
Differential Revision: https://reviews.llvm.org/D105280
This patch uses AtomicExpandPass to implement quadword lock free atomic operations. It adopts the method introduced in https://reviews.llvm.org/D47882, which expand atomic operations post RA to avoid spilling that might prevent LL/SC progress.
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D103614
Any def of EXEC prevents rematerialization of any VOP instruction
because of the physreg use. Create a callback to check if the
physreg use can be ingored to allow rematerialization.
Differential Revision: https://reviews.llvm.org/D105836
This is mostly a minor convenience, but the pattern seems frequent
enough to be worthwhile (and we'll probably add more uses in the
future).
Differential Revision: https://reviews.llvm.org/D105850
This new MIR pass removes redundant DBG_VALUEs.
After the register allocator is done, more precisely, after
the Virtual Register Rewriter, we end up having duplicated
DBG_VALUEs, since some virtual registers are being rewritten
into the same physical register as some of existing DBG_VALUEs.
Each DBG_VALUE should indicate (at least before the LiveDebugValues)
variables assignment, but it is being clobbered for function
parameters during the SelectionDAG since it generates new DBG_VALUEs
after COPY instructions, even though the parameter has no assignment.
For example, if we had a DBG_VALUE $regX as an entry debug value
representing the parameter, and a COPY and after the COPY,
DBG_VALUE $virt_reg, and after the virtregrewrite the $virt_reg gets
rewritten into $regX, we'd end up having redundant DBG_VALUE.
This breaks the definition of the DBG_VALUE since some analysis passes
might be built on top of that premise..., and this patch tries to fix
the MIR with the respect to that.
This first patch performs bacward scan, by trying to detect a sequence of
consecutive DBG_VALUEs, and to remove all DBG_VALUEs describing one
variable but the last one:
For example:
(1) DBG_VALUE $edi, !"var1", ...
(2) DBG_VALUE $esi, !"var2", ...
(3) DBG_VALUE $edi, !"var1", ...
...
in this case, we can remove (1).
By combining the forward scan that will be introduced in the next patch
(from this stack), by inspecting the statistics, the RemoveRedundantDebugValues
removes 15032 instructions by using gdb-7.11 as a testbed.
Differential Revision: https://reviews.llvm.org/D105279
Currently we are resolving lane/subregister conflict by visiting
instructions sequentially in current block to see whether there is any
use of the tainted lanes. To save compile time, we are not doing further
check in successor blocks. This sounds reasonable without subgregister liveness.
But since we have added subregister liveness tracking capability to
register coalescer, we can easily determine whether we have subregister
liveness conflict by checking subranges. This would help coalescing more
COPYs for target that enables subregister liveness tracking.
Reviewed by: arsenm, qcolombet
Differential Revision: https://reviews.llvm.org/D104509
Previously we reliedy on pseudo probe descriptors to look up precomputed GUID during probe emission for inlined probes. Since we are moving to always using unique linkage names, GUID for functions can be computed in place from dwarf names. This eliminates the need of importing pseudo probe descs in thinlto, since those descs should be emitted by the original modules.
This significantly reduces thinlto memory footprint in some extreme case where the number of imported modules for a single module is massive.
Test Plan:
Reviewed By: wenlei
Differential Revision: https://reviews.llvm.org/D105248
AMDGPU normally spills SGPRs to VGPRs. Previously, since all register
classes are handled at the same time, this was problematic. We don't
know ahead of time how many registers will be needed to be reserved to
handle the spilling. If no VGPRs were left for spilling, we would have
to try to spill to memory. If the spilled SGPRs were required for exec
mask manipulation, it is highly problematic because the lanes active
at the point of spill are not necessarily the same as at the restore
point.
Avoid this problem by fully allocating SGPRs in a separate regalloc
run from VGPRs. This way we know the exact number of VGPRs needed, and
can reserve them for a second run. This fixes the most serious
issues, but it is still possible using inline asm to make all VGPRs
unavailable. Start erroring in the case where we ever would require
memory for an SGPR spill.
This is implemented by giving each regalloc pass a callback which
reports if a register class should be handled or not. A few passes
need some small changes to deal with leftover virtual registers.
In the AMDGPU implementation, a new pass is introduced to take the
place of PrologEpilogInserter for SGPR spills emitted during the first
run.
One disadvantage of this is currently StackSlotColoring is no longer
used for SGPR spills. It would need to be run again, which will
require more work.
Error if the standard -regalloc option is used. Introduce new separate
-sgpr-regalloc and -vgpr-regalloc flags, so the two runs can be
controlled individually. PBQB is not currently supported, so this also
prevents using the unhandled allocator.
This fixes not respecting signext/zeroext in these cases. In the
anyext case, this avoids a larger merge with undef and should be a
better canonical form.
This should also handle this if a merge is needed, but I'm not aware
of a case where that can happen. In a future change this will also
allow AMDGPU to drop some custom code without introducing regressions.
Generalize the existing eq/ne case using `extractParts`. The original code only
handled narrowings for types of width 2n->n. This generalization allows for any
type that can be broken down by `extractParts`.
General overview is:
- Loop over each narrow-sized part and do exactly what the 2-register case did.
- Loop over the leftover-sized parts and do the same thing
- Widen the leftover-sized XOR results to the desired narrow size
- OR that all together and then do the comparison against 0 (just like the old
code)
This shows up a lot when building clang for AArch64 using GlobalISel, so it's
worth fixing. For the sake of simplicity, this doesn't handle the non-eq/ne
case yet.
Also remove the code in this case that notifies the observer; we're just going
to delete MI anyway so talking to the observer shouldn't be necessary.
Differential Revision: https://reviews.llvm.org/D105161
Let other parts of legalization handle the rest of the node, this allows
re-use of existing optimizations elsewhere.
Differential Revision: https://reviews.llvm.org/D105624
This adds custom lowering for truncating stores when operating on
fixed length vectors in SVE. It also includes a DAG combine to
fold extends followed by truncating stores into non-truncating
stores in order to prevent this pattern appearing once truncating
stores are supported.
Currently truncating stores are not used in certain cases where
the size of the vector is larger than the target vector width.
Differential Revision: https://reviews.llvm.org/D104471
The test case here hits machine verifier problems. There are volatile
long loads that the results of do not get used, loading into two dead
registers. IfCvt will predicate them and as it does will add implicit
uses of the predicating registers due to thinking they are live in. As
nothing has used the register, the machine verifier disagrees that they
are really live and we end up with a failure.
The registers come from Pristine regs that LivePhysRegs counts as live.
This patch adds a addLiveInsNoPristines method to be used instead in
IfCvt, so that only really live in regs need to be added as implicit
operands.
Differential Revision: https://reviews.llvm.org/D90965
The original motivation for this was to implement moreElementsVector of shuffles
on AArch64, which resulted in complex sequences of artifacts like unmerge(unmerge(concat...))
which the combiner couldn't handle. It seemed here that the better option,
instead of writing ever-more-complex combines, was to have a way to find
the original "non-artifact" source registers for a given definition, walking
through arbitrary expressions of unmerge/concat/insert. As long as the bits
aren't extended or truncated, this is a pretty simple algorithm that avoids
the need for lots of combines and instead jumps straight to the final result
we want.
I've only used this new technique in 2 places within tryCombineUnmerge, using it
in more general situations resulted in infinite loops in AMDGPU. So for now
it's used when we would otherwise fail to combine and that seems to work.
In order to support looking through G_INSERTs, I also had to add it as an
artifact in isArtifact(), which caused a whole lot of issues in tests. AMDGPU
started infinite looping since full legalization of G_INSERT doensn't seem to
be there. To work around this, I've temporarily added a CLI option to use the
old behaviour so that the MIR tests will still run and terminate.
Other minor changes include no longer making >128b G_MERGE/UNMERGE legal.
We never had isel support for that anyway and it was a remnant of the legacy
legalizer rules. However being legal prevented the combiner from checking if it
was dead and deleting them.
Differential Revision: https://reviews.llvm.org/D104355
`LegalizerHelper::insertParts` uses `extractGCDType` on registers split into
a desired type and a smaller leftover type. This is used to populate a list
of registers. Each register in the list will have the same type as returned by
`extractGCDType`.
If we have
- `ResultTy` = s792
- `PartTy` = s64
- `LeftoverTy` = s24
When we call `extractGCDType`, we'll end up with two different types appended
to the list:
Part: gcd(792, 64, 24) => s8
Leftover: gcd(792, 24, 24) => s24
When this happens, we'll hit an assert while trying to build a G_MERGE_VALUES.
This patch changes the code for the leftover type so that we reuse the GCD from
the desired type.
e.g.
Leftover: gcd(792, 8, 24) => s8
https://llvm.godbolt.org/z/137Kqxj6j
Differential Revision: https://reviews.llvm.org/D105674
This to protect against non-sensical instruction sequences being assembled,
which would either cause asserts/crashes further down, or a Wasm module being output that doesn't validate.
Unlike a validator, this type checker is able to give type-errors as part of the parsing process, which makes the assembler much friendlier to be used by humans writing manual input.
Because the MC system is single pass (instructions aren't even stored in MC format, they are directly output) the type checker has to be single pass as well, which means that from now on .globaltype and .functype decls must come before their use. An extra pass is added to Codegen to collect information for this purpose, since AsmPrinter is normally single pass / streaming as well, and would otherwise generate this information on the fly.
A `-no-type-check` flag was added to llvm-mc (and any other tools that take asm input) that surpresses type errors, as a quick escape hatch for tests that were not intended to be type correct.
This is a first version of the type checker that ignores control flow, i.e. it checks that types are correct along the linear path, but not the branch path. This will still catch most errors. Branch checking could be added in the future.
Differential Revision: https://reviews.llvm.org/D104945
LLVM provides target hooks to recognise stack spill and restore
instructions, such as isLoadFromStackSlot, and it also provides post frame
elimination versions such as isLoadFromStackSlotPostFE. These are supposed
to return the store-source and load-destination registers; unfortunately on
X86, the PostFE recognisers just return "1", apparently to signify "yes
it's a spill/load". This patch alters the hooks to correctly return the
store-source and load-destination registers:
This is really useful for debug-info as we it helps follow variable values
as they move on/off the stack. There should be no codegen changes: the only
other users of these PostFE target hooks are MachineInstr::getRestoreSize
and MachineInstr::getSpillSize, which don't attempt to interpret the
returned register location.
While we're here, delete the (InstrRef) LiveDebugValues heuristic that
tries to find the spill source register by looking for a killed reg -- we
should be able to rely on the target hooks for that. This involves
temporarily turning off a n InstrRef LivedDebugValues test on aarch64
(patch to re-enable it is in D104521).
Differential Revision: https://reviews.llvm.org/D105428
We keep a record of substitutions between debug value numbers post-isel,
however we never actually look them up until the end of compilation. As a
result, there's nothing gained by the collection being a std::map. This
patch downgrades it to being a vector, that's then sorted at the end of
compilation in LiveDebugValues.
Differential Revision: https://reviews.llvm.org/D105029
C++23 will make these conversions ambiguous - so fix them to make the
codebase forward-compatible with C++23 (& a follow-up change I've made
will make this ambiguous/invalid even in <C++23 so we don't regress
this & it generally improves the code anyway)
SelectionDAG's equivalents in ISD::InputArg/OutputArg track the
original argument index. Mips relies on this, and its currently
reinventing its own parallel CallLowering infrastructure which tracks
these indexes on the side. Add this to help move towards deleting the
custom mips handling.
This is a cleanup patch -- we're now able to support all flavours of
variable location in instruction referencing mode. This patch updates
various tests for debug instructions to be broader: numerous code paths
try to ignore debug isntructions, and they now have to ignore the
additional DBG_PHI and DBG_INSTR_REFs that we can generate.
A small amount of rework happens for LiveDebugVariables: as we don't need
to track live intervals through regalloc any more, we can get away with
unlinking debug instructions before regalloc, then re-inserting them after.
Note that this isn't (yet) true of DBG_VALUE_LISTs, they still have to go
through live interval tracking.
In SelectionDAG, add a helper lambda that emits half-formed DBG_INSTR_REFs
for arguments in instr-ref mode, DBG_VALUE otherwise. This is one of the
final locations where DBG_VALUEs are emitted for vreg arguments.
X86InstrInfo now un-sets the debug instr number on SUB instructions that
get mutated into CMP instructions. As the instruction no longer computes a
subtraction, we can't use it for variable locations.
Differential Revision: https://reviews.llvm.org/D88898
This patch prevents GlobalISel from optimizing out redundant branch
instructions when compiling without optimizations.
The motivating example is code like the following common pattern in
Swift, where users expect to be able to set a breakpoint on the early
exit:
public func f(b: Bool) {
guard b else {
return // I would like to set a breakpoint here.
}
...
}
The patch modifies two places in GlobalISEL: The first one is in
IRTranslator.cpp where the removal of redundant branches is made
conditional on the optimization level. The second one is in
AArch64InstructionSelector.cpp where an -O0 *only* optimization is
being removed.
Disabling these optimizations increases code size at -O0 by
~8%. However, doing so improves debuggability, and debug builds are
the primary reason why developers compile without optimizations. We
thus concluded that this is the right trade-off.
rdar://79515454
Differential Revision: https://reviews.llvm.org/D105238
We already have reassociation code for Adds and Ors separately in DAG
combiner, this adds it for the combination of the two where Ors act like
Adds. It reassociates (add (or (x, c), y) -> (add (add (x, y), c)) where
we know that the Ors operands have no common bits set, and the Or has
one use.
Differential Revision: https://reviews.llvm.org/D104765
This patch emits DBG_INSTR_REFs for two remaining flavours of variable
locations that weren't supported: copies, and inter-block VRegs. There are
still some locations that must be represented by DBG_VALUE such as
constants, but they're mostly independent of optimisations.
For variable locations that refer to values defined in different blocks,
vregs are allocated before isel begins, but the defining instruction
might not exist until late in isel. To get around this, emit
DBG_INSTR_REFs in a "half done" state, where the first operand refers to a
VReg. Then at the end of isel, patch these back up to refer to
instructions, using the finalizeDebugInstrRefs method.
Copies are something that I complained about the original RFC, and I
really don't want to have to put instruction numbers on copies. They don't
define a value: they move them. To address this isel, salvageCopySSA
interprets:
* COPYs,
* SUBREG_TO_REG,
* Anything that isCopyInstr thinks is a copy.
And follows chains of copies back to the defining instruction that they
read from. This relies on any physical registers that COPYs read being
defined in the same block, or being entry-block arguments. For the former
we can put an instruction number on the defining instruction; for the
latter we can drop a DBG_PHI that reads the incoming value.
Differential Revision: https://reviews.llvm.org/D88896
This patch fixes an issue which occurred in CodeGenPrepare and
HWAddressSanitizer, which both at some point create a map of Old->New
instructions and update dbg.value uses of these. They did this by
iterating over the dbg.value's location operands, and if an instance of
the old instruction was found, replaceVariableLocationOp would be
called on that dbg.value. This would cause an error if the same operand
appeared multiple times as a location operand, as the first call to
replaceVariableLocationOp would update all uses of the old instruction,
invalidating the old iterator and eventually hitting an assertion.
This has been fixed by no longer iterating over the dbg.value's location
operands directly, but by first collecting them into a set and then
iterating over that, ensuring that we never attempt to replace a
duplicated operand multiple times.
Differential Revision: https://reviews.llvm.org/D105129
This reverts commit 8cd35ad854.
It breaks `TestMembersAndLocalsWithSameName.py` on GreenDragon and
Mikael Holmén points out in D104827 that bitcode files created with the
patch cannot be parsed with binaries built before it.
We're trying to match a few pointer computation patterns here for
re-association opportunities.
1) Isolating a constant operand to be on the RHS, e.g.:
G_PTR_ADD(BASE, G_ADD(X, C)) -> G_PTR_ADD(G_PTR_ADD(BASE, X), C)
2) Folding two constants in each sub-tree as long as such folding
doesn't break a legal addressing mode.
G_PTR_ADD(G_PTR_ADD(BASE, C1), C2) -> G_PTR_ADD(BASE, C1+C2)
AArch64 code size improvements on CTMark with -Os:
Program before after diff
pairlocalalign 251048 251044 -0.0%
consumer-typeset 421820 421812 -0.0%
kc 431348 431320 -0.0%
SPASS 413404 413300 -0.0%
clamscan 384396 384220 -0.0%
tramp3d-v4 370640 370412 -0.1%
lencod 432096 431772 -0.1%
bullet 479400 478796 -0.1%
sqlite3 288504 288072 -0.1%
7zip-benchmark 573796 570768 -0.5%
Geomean difference -0.1%
Differential Revision: https://reviews.llvm.org/D105069
Add a flag so that target can choose to use AsmParser for parsing inline asm.
And set the flag by default for AIX.
-no-intergrated-as will override this default if specified explicitly.
Reviewed By: #powerpc, shchenz
Differential Revision: https://reviews.llvm.org/D105314
Fixes bugs [[ https://bugs.llvm.org/show_bug.cgi?id=50580 | 50580 ]] and [[ https://bugs.llvm.org/show_bug.cgi?id=49446 | 49446 ]]
When compiling with -g "DBG_VALUE <reg>" instructions are added in the MIR, if such a instruction is inserted between instructions that use <reg> then MachineCopyPropagation invalidates <reg> , this causes some copies to not be propagated and causes differences in code generation (ex bugs 50580 and 49446 ). DBG_VALUE instructions should be ignored since they don't actually modify the register.
Reviewed By: lkail
Differential Revision: https://reviews.llvm.org/D104394
Reland of 31859f896.
This change implements new DAG notes GLOBAL_GET/GLOBAL_SET, and
lowering methods for load and stores of reference types from IR
globals. Once the lowering creates the new nodes, tablegen pattern
matches those and converts them to Wasm global.get/set.
Differential Revision: https://reviews.llvm.org/D104797
Previously we used the vector type, but we're loading/storing
invididual elements so I think only element alignment should matter.
Noticed while looking at the code for something else so I don't
have a test case.
Differential Revision: https://reviews.llvm.org/D105220
In `IRTranslator::translateGetElementPtr`, when we run into a vector gep with
some scalar operands, we try to normalize those operands using
`buildSplatVector`.
This is fine except for when the getelementptr has a <1 x N> type. In that case
it is treated as a scalar. If we run into one of these then every call to
```
// With VectorWidth = 1
LLT::fixed_vector(VectorWidth, PtrTy)
```
will assert.
Here's an example (equivalent to the added testcase):
https://godbolt.org/z/hGsTnMYdW
To get around this, this patch adds a variable, `WantSplatVector`, which
is true when our vector type ought to actually be represented using a vector.
When it's false, we'll translate as a scalar. This checks if `VectorWidth > 1`.
This fixes this bug:
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=35496
Differential Revision: https://reviews.llvm.org/D105316
Inserting into a smaller-than-legal scalable vector would result in an
internal compiler error. For example, inserting a <vscale x 4 x i8> into
a <vscale x 8 x i8> (both illegal vector types for SVE) would cause a
crash.
This crash was happening because there was no code to promote (legalise)
the result of an INSERT_SUBVECTOR node.
This patch implements PromoteIntRes_INSERT_SUBVECTOR, which legalises
the ISD node. This is currently done by going through memory. This is
necessary because of the requirement that the SubVec parameter of the
INSERT_SUBVECTOR node must be smaller than the Vec parameter, which
means that INSERT_SUBVECTOR cannot always have a legal result/operand
types.
Co-Authored-by: Joe Ellis <joe.ellis@arm.com>
Differential Revision: https://reviews.llvm.org/D102766
Added in 47c3fe2a22, we sometimes need to describe a variable value
substitution with a subregister qualifier, to say that "the value is the
lower 32 bits of this 64 bit register def" for example. That then needs
support during LiveDebugValues to interpret the subregister qualifiers,
which is what this patch adds.
Whenever we encounter a DBG_INSTR_REF and find its value by using a
substitution, collect any subregister qualifiers seen. Then, accumulate the
effects of the qualifiers to work out what offset and what size should be
extracted from the defined register. Finally, for the target ValueIDNum,
extract whatever subregister is in the correct position
Currently, describing a subregister field of a larger value that has been
spilt to the stack, is unimplemented.
Differential Revision: https://reviews.llvm.org/D88894
Since gather lowering can now lower to nodes that may need expansion via
the vector legalizer, do MGATHER lowering via vector legalizer.
Additionally, as part of adding passthru support for fixed typed
gathers, fix passthru support for scalable types.
Depends on D104910
Differential Revision: https://reviews.llvm.org/D104217
Very late in compilation, backends like X86 will perform optimisations like
this:
$cx = MOV16rm $rax, ...
->
$rcx = MOV64rm $rax, ...
Widening the load from 16 bits to 64 bits. SEeing how the lower 16 bits
remain the same, this doesn't affect execution. However, any debug
instruction reference to the defined operand now refers to a 64 bit value,
nto a 16 bit one, which might be unexpected. Elsewhere in codegen, there's
often this pattern:
CALL64pcrel32 @foo, implicit-def $rax
%0:gr64 = COPY $rax
%1:gr32 = COPY %0.sub_32bit
Where we want to refer to the definition of $eax by the call, but don't
want to refer the copies (they don't define values in the way
LiveDebugValues sees it). To solve this, add a subregister field to the
existing "substitutions" facility, so that we can describe a field within
a larger value definition. I would imagine that this would be used most
often when a value is widened, and we need to refer to the original,
narrower definition.
Differential Revision: https://reviews.llvm.org/D88891
This patch changes return type of tryCandidate from void to bool:
1. Methods in some targets already follow this convention.
2. This would help if some target wants to re-use generic code.
3. It looks more intuitive if these try-method returns the same type.
We may need to change return type of them from bool to some enum
further, to make it less confusing.
Reviewed By: foad
Differential Revision: https://reviews.llvm.org/D103951
This patch adds support to the instruction-referencing LiveDebugValues
implementation for emitting entry values. The instruction referencing
implementations tracking by value rather than location means that we can
get around two of the issues with VarLocs. DBG_VALUE instructions that
re-assign the same value to a variable are no longer a problem, because we
can "see through" to the value being assigned. We also don't need to do
anything special during the dataflow stages: the "variable value problem"
doesn't need to know whether a value is available most of the time, and the
times it deoes need to know are always when entry values need to be
terminated.
The patch modifies the "TransferTracker" class, adding methods to identify
when a variable ias an entry value candidate, and when a machine value is
an entry value. recoverAsEntryValue tests these two things and emits an
entry-value expression if they're true. It's used when we clobber or
otherwise lose a value and can't find a replacement location for the value
it contained.
Differential Revision: https://reviews.llvm.org/D88406
This enables proper lowering of non-byte sized loads. We still aren't
faithfully preserving memory types everywhere, so the legality checks
still only consider the size.
Previously we didn't preserve the memory type and had to blindly
interpret a number of bytes. Now that non-byte memory accesses are
representable, we can handle these correctly.
Ported from DAG version (minus some weird special case i1 legality
checking which I don't fully understand, and we don't have a way to
query for)
For now, this is NFC and the test changes are placeholders. Since the
legality queries are still relying on byte-flattened memory sizes, the
legalizer can't actually see these non-byte accesses. This keeps this
change self contained without merging it with the larger patch to
switch to LLT memory queries.
This will currently accept the old number of bytes syntax, and convert
it to a scalar. This should be removed in the near future (I think I
converted all of the tests already, but likely missed a few).
Not sure what the exact syntax and policy should be. We can continue
printing the number of bytes for non-generic instructions to avoid
test churn and only allow non-scalar types for generic instructions.
This will currently print the LLT in parentheses, but accept parsing
the existing integers and implicitly converting to scalar. The
parentheses are a bit ugly, but the parser logic seems unable to deal
without either parentheses or some keyword to indicate the start of a
type.
In various circumstances, when we clobber a register there may be
alternative locations that the value is live in. The classic example would
be a value loaded from the stack, and then clobbered: the value is still
available on the stack. InstrRefBasedLDV was coping with this at block
starts where it's forced to pick a location, however it wasn't searching
for alternative locations when values were clobbered.
This patch notifies the "Transfer Tracker" object when clobbers occur, and
it's able to find alternatives and issue DBG_VALUEs for that location. See:
the added test.
Differential Revision: https://reviews.llvm.org/D88405
When clamping the index for a memory access to a stacked vector we must
take into account the entire type being accessed, not just assume that
we are accessing only a single element.
Differential Revision: https://reviews.llvm.org/D105016
GlobalISel is relying on regular MachineMemOperands to track all of
the memory properties of accesses. Just the raw byte size is
insufficent to disambiguate all situations. For example, if we need to
split an unaligned extending load, we need to know the number of bits
in the original source value and can't infer it from the result
type. This is also a problem for extending vector loads.
This does decrease the maximum representable size from the full
uint64_t bytes to a maximum of 16-bits. No in tree testcases hit this,
other than places using UINT64_MAX for unknown sizes. This may be an
issue for G_MEMCPY and co., although they can just use unknown size
for large static sizes. This also has potential for backend abuse by
relying on the type when it really shouldn't be relevant after
selection.
This does not include the necessary MIR printer/parser changes to
represent this.
We were trying to expand these if they were going to be expanded
in op legalization so that we generated the minimum number of
operations. We failed to take into account that NVT could be
promoted to another legal type in op legalization.
Hoping this fixes the issue on the VE target reported as a follow
up to D96681. The check line changes were taken from before
1e46b6f401 so this patch does
appear to improve some cases that had previously regressed.
This patch reads machine value numbers from DBG_PHI instructions (marking
where SSA PHIs used to be), and matches them up with DBG_INSTR_REF
instructions that refer to them. Essentially they are two separate parts of
a DBG_VALUE: the place to read the value (register and program position),
and where the variable is assigned that value.
Sometimes these DBG_PHIs can be duplicated, usually by tail duplication.
This corresponds to the SSA structure of the program being destroyed, and
the original PHI being split. When this happens: run LLVMs standard
SSAUpdater utility, to work out what values should appear in which blocks.
The majority of this patch is boilerplate to make use of SSAUpdater.
If there are any additional PHIs on the path between multiple DBG_PHIs and
their using DBG_INSTR_REF, their existance is validated, just in case a
value gets clobbered along the way (see dbg-phis-with-loops.mir for
several examples).
Differential Revision: https://reviews.llvm.org/D86814
- Add standalone metadata parsing support so that machine metadata nodes
could be populated before and accessed during MIR is parsed.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D103282
Add UNIQUED and DISTINCT properties in Metadata.def and use them to
implement restrictions on the `distinct` property of MDNodes:
* DIExpression can currently be parsed from IR or read from bitcode
as `distinct`, but this property is silently dropped when printing
to IR. This causes accepted IR to fail to round-trip. As DIExpression
appears inline at each use in the canonical form of IR, it cannot
actually be `distinct` anyway, as there is no syntax to describe it.
* Similarly, DIArgList is conceptually always uniqued. It is currently
restricted to only appearing in contexts where there is no syntax for
`distinct`, but for consistency it is treated equivalently to
DIExpression in this patch.
* DICompileUnit is already restricted to always being `distinct`, but
along with adding general support for the inverse restriction I went
ahead and described this in Metadata.def and updated the parser to be
general. Future nodes which have this restriction can share this
support.
The new UNIQUED property applies to DIExpression and DIArgList, and
forbids them to be `distinct`. It also implies they are canonically
printed inline at each use, rather than via MDNode ID.
The new DISTINCT property applies to DICompileUnit, and requires it to
be `distinct`.
A potential alternative change is to forbid the non-inline syntax for
DIExpression entirely, as is done with DIArgList implicitly by requiring
it appear in the context of a function. For example, we would forbid:
!named = !{!0}
!0 = !DIExpression()
Instead we would only accept the equivalent inlined version:
!named = !{!DIExpression()}
This essentially removes the ability to create a `distinct` DIExpression
by construction, as there is no syntax for `distinct` inline. If this
patch is accepted as-is, the result would be that the non-canonical
version is accepted, but the following would be an error and produce a diagnostic:
!named = !{!0}
; error: 'distinct' not allowed for !DIExpression()
!0 = distinct !DIExpression()
Also update some documentation to consistently use the inline syntax for
DIExpression, and to describe the restrictions on `distinct` for nodes
where applicable.
Reviewed By: StephenTozer, t-tye
Differential Revision: https://reviews.llvm.org/D104827
This intrinsic blocks floating point transformations by the optimizer.
Author: Pengfei
Reviewed By: LuoYuanke, Andy Kaylor, Craig Topper, kpn
Differential Revision: https://reviews.llvm.org/D99675
This patch relands https://reviews.llvm.org/D104454, but fixes some failing
builds on Mac OS which apparently has a different definition for size_t,
that caused 'ambiguous operator overload' for the implicit conversion
of TypeSize to a scalar value.
This reverts commit b732e6c9a8.
Peephole optimizer should not be introducing sub-reg definitions
as they are illegal in machine SSA phase. This patch modifies
the optimizer to not emit sub-register definitions.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D103408
Adds legalizer, register bank select, and instruction
select support for G_SBFX and G_UBFX. These opcodes generate
scalar or vector ALU bitfield extract instructions for
AMDGPU. The instructions allow both constant or register
values for the offset and width operands.
The 32-bit scalar version is expanded to a sequence that
combines the offset and width into a single register.
There are no 64-bit vgpr bitfield extract instructions, so the
operations are expanded to a sequence of instructions that
implement the operation. If the width is a constant,
then the 32-bit bitfield extract instructions are used.
Moved the AArch64 specific code for creating G_SBFX to
CombinerHelper.cpp so that it can be used by other targets.
Only bitfield extracts with constant offset and width values
are handled currently.
Differential Revision: https://reviews.llvm.org/D100149
A combination of features ^ that lead to a mismatch of expectations
about how a subprogram definition DIE would be produced with/without a
declaration when taking full -g debug info and inlining it into a -gmlt
CU - specifically when using Split DWARF that doesn't support cross-CU
references, so we have to put the -g debug info into the -gmlt CU, which
gets confusing about which mode is respected.
This patch comes down on respecting the CU the debug info is emitted
into, rather than preserving the full debug info when it's emitted into
the gmlt CU.
This ports the AArch64 SABD and USBD over to DAG Combine, where they can be
used by more backends (notably MVE in a follow-up patch). The matching code
has changed very little, just to handle legal operations and types
differently. It selects from (ABS (SUB (EXTEND a), (EXTEND b))), producing
a ubds/abdu which is zexted to the original type.
Differential Revision: https://reviews.llvm.org/D91937
This add as a fold of sub(0, splat(sub(0, x))) -> splat(x). This can
come up in the lowering of right shifts under AArch64, where we generate
a shift left of a negated number.
Differential Revision: https://reviews.llvm.org/D103755
This change is NFC upstream. We pass in the loop's block to the kernel
rewriter explicitly, instead of assuming it's the loop's top block. This
change is made for downstream targets where this assumption doesn't hold.
Differential Revision: https://reviews.llvm.org/D104811
To reflect that the size may be scalable, a TypeSize is returned
instead of an unsigned. In places where the result is used,
it currently relies on an implicit cast of TypeSize -> uint64_t,
which asserts that the type is not scalable.
This patch is NFC for fixed-width vectors.
Reviewed By: aemerson
Differential Revision: https://reviews.llvm.org/D104454
This is a mechanical change. This actually also renames the
similarly named methods in the SmallString class, however these
methods don't seem to be used outside of the llvm subproject, so
this doesn't break building of the rest of the monorepo.
We don't constant fold based on demanded bits elsewhere in
SimplifyDemandedBits, so I don't think we should shrink them either.
The affected ARM test changes because a constant become non-opaque
and eventually enabled some constant folding. This no longer happens.
I checked and InstCombine is able to simplify this test. I'm not sure exactly
what it was trying to test.
Reviewed By: lebedev.ri, dmgreen
Differential Revision: https://reviews.llvm.org/D104832
This also adds new interfaces for the fixed- and scalable case:
* LLT::fixed_vector
* LLT::scalable_vector
The strategy for migrating to the new interfaces was as follows:
* If the new LLT is a (modified) clone of another LLT, taking the
same number of elements, then use LLT::vector(OtherTy.getElementCount())
or if the number of elements is halfed/doubled, it uses .divideCoefficientBy(2)
or operator*. That is because there is no reason to specifically restrict
the types to 'fixed_vector'.
* If the algorithm works on the number of elements (as unsigned), then
just use fixed_vector. This will need to be fixed up in the future when
modifying the algorithm to also work for scalable vectors, and will need
then need additional tests to confirm the behaviour works the same for
scalable vectors.
* If the test used the '/*Scalable=*/true` flag of LLT::vector, then
this is replaced by LLT::scalable_vector.
Reviewed By: aemerson
Differential Revision: https://reviews.llvm.org/D104451
This is a partial reapply of the original commit and the followup commit
that were previously reverted; this reapply also includes a small fix
for a potential source of non-determinism, but also has a small change
to turn off variadic debug value salvaging, to ensure that any future
revert/reapply steps to disable and renable this feature do not risk
causing conflicts.
Differential Revision: https://reviews.llvm.org/D91722
This reverts commit 386b66b2fc.
Having type symmetry with these is somewhat necessary when implementing support for 192-bit values.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D104621
Stats added:
1. NumCleanupLandingPadsUnreachable: how many cleanup landing pads were optimized as unreachable
1. NumCleanupLandingPadsRemaining: how many cleanup landing pads remain
1. NumNoUnwind: Number of functions with nounwind attribute
1. NumUnwind: Number of functions with unwind attribute
DwarfEHPrepare is always run a single time as part of `TargetPassConfig::addISelPasses()` which makes it an ideal place near the end of the pipeline to record this information.
Example output from clang built with exceptions cumulative during thinLTO backend (NumCleanupLandingPadsUnreachable was not incremented):
"dwarfehprepare.NumCleanupLandingPadsRemaining": 123660,
"dwarfehprepare.NumNoUnwind": 323836,
"dwarfehprepare.NumUnwind": 472893,
Reviewed By: wenlei
Differential Revision: https://reviews.llvm.org/D104161
This optimization pre-promotes the input and constants for a
switch instruction to a legal type so that all the generated compares
share the same extend. Since RISCV prefers sext for i32 to i64
extends, we should honor that to use sext.w instead of a pair
of shifts.
Reviewed By: jrtc27
Differential Revision: https://reviews.llvm.org/D104612
When inserting UnregisterFn, if there is a musttail call, we must insert before the call so that we don't break the musttail call contract.
Reviewed By: wenlei
Differential Revision: https://reviews.llvm.org/D104807
When inserting UnregisterFn, if there is a musttail call, we must insert before the call so that we don't break the musttail call contract.
Differential Revision: https://reviews.llvm.org/D104807
The is from discussion in https://reviews.llvm.org/D104247#inline-993387
The contract and reassoc flags shouldn't imply each other .
All the aggressive fsub fusion reassociate operations,
we should guard them with reassoc flag check.
Reviewed By: mcberg2017
Differential Revision: https://reviews.llvm.org/D104723
Summary:
generate eh_info when vector registers are saved according to the traceback table.
struct eh_info_t {
unsigned version; /* EH info version 0 */
#if defined(64BIT)
char _pad[4]; /* padding */
#endif
unsigned long lsda; /* Pointer to Language Specific Data Area */
unsigned long personality; /* Pointer to the personality routine */
};
the value of lsda and personality is zero when the number of vector registers saved is large zero and there is not personality of the function
Reviewers: Jason Liu
Differential Revision: https://reviews.llvm.org/D103651
This patch aims to add the scalable property to LLT. The rest of the
patch-series changes the interfaces to take/return ElementCount and
TypeSize, which both have the ability to represent the scalable property.
The changes are mostly mechanical and aim to be non-functional changes
for fixed-width vectors.
For scalable vectors some unit tests have been added, but no effort has
been put into making any of the GlobalISel algorithms work with scalable
vectors yet. That will be left as future work.
The work is split into a series of 5 patches to make reviews easier.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D104450
Since this method can apply to cmpxchg operations, make sure it's clear
what value we're actually retrieving. This will help ensure we don't
accidentally ignore the failure ordering of cmpxchg in the future.
We could potentially introduce a getOrdering() method on AtomicSDNode
that asserts the operation isn't cmpxchg, but not sure that's
worthwhile.
Differential Revision: https://reviews.llvm.org/D103338
According to IR LangRef, the FMF flag:
contract
Allow floating-point contraction (e.g. fusing a multiply followed by an
addition into a fused multiply-and-add).
reassoc
Allow reassociation transformations for floating-point instructions.
This may dramatically change results in floating-point.
My understanding is that these two flags shouldn't imply each other,
as we might have a SDNode that can be reassociated with others, but
not contractble.
eg: We may want following fmul/fad/fsub to freely reassoc, but don't
want fma being generated here.
%F = fmul reassoc double %A, %B ; <double> [#uses=1]
%G = fmul reassoc double %C, %D ; <double> [#uses=1]
%H = fadd reassoc double %F, %G ; <double> [#uses=1]
%I = fsub reassoc double %H, %E ; <double> [#uses=1]
Before https://reviews.llvm.org/D45710, `reassoc` flag actually
did not imply isContratable either.
The current implementation also only check the flag in fadd node,
ignoring fmul node, this patch update that as well.
Reviewed By: spatel, qiucf
Differential Revision: https://reviews.llvm.org/D104247
Fixes a minor bug when trying to iterate through use operands when
updating debug use operands.
Extends a test to include above.
Differential Revision: https://reviews.llvm.org/D104576
TypePromotion is meant to be a generic pass and doesn't reference
any ARM intrinsics so it shouldn't include IntrinsicsARM.h.
The other Intrinsic related headers appear to be unneeded as well.
- Distinct metadata needs generating in the codegen to attach correct
AAInfo on the loads/stores after lowering, merging, and other relevant
transformations.
- This patch adds 'MachhineModuleSlotTracker' to help assign slot
numbers to these newly generated unnamed metadata nodes.
- To help 'MachhineModuleSlotTracker' track machine metadata, the
original 'SlotTracker' is rebased from 'AbstractSlotTrackerStorage',
which provides basic interfaces to create/retrive metadata slots. In
addition, once LLVM IR is processsed, additional hooks are also
introduced to help collect machine metadata and assign them slot
numbers.
- Finally, if there is any such machine metadata, 'MIRPrinter' outputs
an additional 'machineMetadataNodes' field containing all the
definition of those nodes.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D103205
As a follow-up to https://reviews.llvm.org/D104129, I'm cleaning up the danling probe related code in both the compiler and llvm-profgen.
I'm seeing a 5% size win for the pseudo_probe section for SPEC2017 and 10% for Ciner. Certain benchmark such as 602.gcc has a 20% size win. No obvious difference seen on build time for SPEC2017 and Cinder.
Reviewed By: wenlei
Differential Revision: https://reviews.llvm.org/D104477
The Interleave Access pass will convert shuffle(binop(load, load)) to
binop(shuffle(load), shuffle(load)), in order to create more
interleaving load patterns (VLD2/3/4) that might have been messed up by
instcombine. As shown in D104247 we were missing copying IR flags to the
new instruction though, which should just be kept the same as the
original instruction.
Differential Revision: https://reviews.llvm.org/D104255
This can be seen as a follow up to commit 0ee439b705,
that changed the second argument of __powidf2, __powisf2 and
__powitf2 in compiler-rt from si_int to int. That was to align with
how those runtimes are defined in libgcc.
One thing that seem to have been missing in that patch was to make
sure that the rest of LLVM also handle that the argument now depends
on the size of int (not using the si_int machine mode for 32-bit).
When using __builtin_powi for a target with 16-bit int clang crashed.
And when emitting libcalls to those rtlib functions, typically when
lowering @llvm.powi), the backend would always prepare the exponent
argument as an i32 which caused miscompiles when the rtlib was
compiled with 16-bit int.
The solution used here is to use an overloaded type for the second
argument in @llvm.powi. This way clang can use the "correct" type
when lowering __builtin_powi, and then later when emitting the libcall
it is assumed that the type used in @llvm.powi matches the rtlib
function.
One thing that needed some extra attention was that when vectorizing
calls several passes did not support that several arguments could
be overloaded in the intrinsics. This patch allows overload of a
scalar operand by adding hasVectorInstrinsicOverloadedScalarOpd, with
an entry for powi.
Differential Revision: https://reviews.llvm.org/D99439
This only applies to FastIsel. GlobalIsel seems to sidestep
the issue.
This fixes https://bugs.llvm.org/show_bug.cgi?id=46996
One of the things we do in llvm is decide if a type needs
consecutive registers. Previously, we just checked if it
was an array or not.
(plus an SVE specific check that is not changing here)
This causes some confusion when you arbitrary IR like:
```
%T1 = type { double, i1 };
define [ 1 x %T1 ] @foo() {
entry:
ret [ 1 x %T1 ] zeroinitializer
}
```
We see it is an array so we call CC_AArch64_Custom_Block
which bails out when it sees the i1, a type we don't want
to put into a block.
This leaves the location of the double in some kind of
intermediate state and leads to odd codegen. Which then crashes
the backend because it doesn't know how to implement
what it's been asked for.
You get this:
```
renamable $d0 = FMOVD0
$w0 = COPY killed renamable $d0
```
Rather than this:
```
$d0 = FMOVD0
$w0 = COPY $wzr
```
The backend knows how to copy 64 bit to 64 bit registers,
but not 64 to 32. It can certainly be taught how but the real
issue seems to be us even trying to assign a register block
in the first place.
This change makes the logic of
AArch64TargetLowering::functionArgumentNeedsConsecutiveRegisters
a bit more in depth. If we find an array, also check that all the
nested aggregates in that array have a single member type.
Then CC_AArch64_Custom_Block's assumption of a type that looks
like [ N x type ] will be valid and we get the expected codegen.
New tests have been added to exercise these situations. Note that
some of the output is not ABI compliant. The aim of this change is
to simply handle these situations and not to make our processing
of arbitrary IR ABI compliant.
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D104123
We create flag variable "__llvm_fs_discriminator__" in the binary
to indicate that FSAFDO hierarchical discriminators are used.
This variable might be GC'ed by the linker since it is not explicitly
reference. I initially added the var to the use list in pass
MIRFSDiscriminator but it did not work. It turned out the used global
list is collected in lowering (before MIR pass) and then emitted in
the end of pass pipeline.
Here I add the variable to the use list in IR level's AddDiscriminators
pass. The machine level code is still keep in the case IR's
AddDiscriminators is not invoked. If this is the case, this just use
-Wl,--export-dynamic-symbol=__llvm_fs_discriminator__
to force the emit.
Differential Revision: https://reviews.llvm.org/D103988
Neubauer, Sebastian