It is not legal to form a phi node with token type. The generic LCSSA construction code handles this correctly - by not forming LCSSA for such cases - but the adhoc fixup implementation in LICM did not.
This was noticed in the context of PR49607, but can be demonstrated on ToT with the tweaked test case. This is not specific to gc.relocate btw, it also applies to usage of the preallocated family of intrinsics as well.
Differential Revision: https://reviews.llvm.org/D98728
Fixed section of code that iterated through a SmallDenseMap and added
instructions in each iteration, causing non-deterministic code; replaced
SmallDenseMap with MapVector to prevent non-determinism.
This reverts commit 01ac6d1587.
This caused non-deterministic compiler output; see comment on the
code review.
> This patch updates the various IR passes to correctly handle dbg.values with a
> DIArgList location. This patch does not actually allow DIArgLists to be produced
> by salvageDebugInfo, and it does not affect any pass after codegen-prepare.
> Other than that, it should cover every IR pass.
>
> Most of the changes simply extend code that operated on a single debug value to
> operate on the list of debug values in the style of any_of, all_of, for_each,
> etc. Instances of setOperand(0, ...) have been replaced with with
> replaceVariableLocationOp, which takes the value that is being replaced as an
> additional argument. In places where this value isn't readily available, we have
> to track the old value through to the point where it gets replaced.
>
> Differential Revision: https://reviews.llvm.org/D88232
This reverts commit df69c69427.
Previously we created a new node, then filled in the pieces. Now, we clone the existing node, then change the respective fields. The only change in handling is with phis since we have to handle multiple incoming edges from the same block a bit differently.
Differential Revision: https://reviews.llvm.org/D98316
A broadcast is a shufflevector where only one input is used. Because of the way we handle constants (undef is a constant), the canonical shuffle sees a meet of (some value) and (nullptr). Given this, every broadcast gets treated as a conflict and a new base pointer computation is added.
The other way to tackle this would be to change constant handling specifically for undefs, but this seems easier.
Differential Revision: https://reviews.llvm.org/D98315
RS4GC needs to rewrite the IR to ensure that every relocated pointer has an associated base pointer. The existing code isn't particularly smart about avoiding duplication of existing IR when it turns out the original pointer we were asked to materialize a base pointer for is itself a base pointer.
This patch adds a stage to the algorithm which prunes nodes proven (with a simple forward dataflow fixed point) to be base pointers from the list of nodes considered for duplication. This does require changing some of the later invariants slightly, that's probably the riskiest part of the change.
Differential Revision: D98122
Add MemorySSAWrapperPass as a dependency to MemCpyOptLegacyPass,
since MemCpyOpt now uses MemorySSA by default.
Differential Revision: https://reviews.llvm.org/D98484
This was (partially) reverted in cfe8f8e0 because the conversion from readonly to readnone in Intrinsics.td exposed a couple of problems. This change has been reworked to not need that change (via some explicit checks in client code). This is being done to address the original optimization issue and simplify the testing of the readonly changes. I'm working on that piece under 49607.
Original commit message follows:
The last two operands to a gc.relocate represent indices into the associated gc.statepoint's gc bundle list. (Effectively, gc.relocates are projections from the gc.statepoints multiple return values.)
We can use this to recognize when two gc.relocates are equivalent (and can be CSEd), even when the indices are non-equal. This is particular useful when considering a chain of multiple statepoints as it lets us eliminate all duplicate gc.relocates in a single pass.
Differential Revision: https://reviews.llvm.org/D97974
This fixes a regression from the MemDep-based implementation:
MemDep completely ignores lifetime.start intrinsics that aren't
MustAlias -- this is probably unsound, but it does mean that the
MemDep based implementation successfully eliminated memcpy's from
lifetime.start if the memcpy happens at an offset, rather than
the base address of the alloca.
Add a special case for the case where the lifetime.start spans the
whole alloca (which is pretty much the only kind of lifetime.start
that frontends ever emit), as we don't need to figure out our exact
aliasing relationship in that case, the whole alloca is dead prior
to the call.
If this doesn't cover all practically relevant cases, then it
would be possible to make use of the recently added PartialAlias
clobber offsets to make this more precise.
The added test case crashes before this fix:
```
opt: /repositories/llvm-project/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp:5172: BasicBlock::iterator (anonymous namespace)::LSRInstance::AdjustInsertPositionForExpand(BasicBlock::iterator, const (anonymous namespace)::LSRFixup &, const (anonymous namespace)::LSRUse &, llvm::SCEVExpander &) const: Assertion `!isa<PHINode>(LowestIP) && !LowestIP->isEHPad() && !isa<DbgInfoIntrinsic>(LowestIP) && "Insertion point must be a normal instruction"' failed.
```
This is fully analogous to the previous commit,
with the pointer constant replaced to be something non-null.
The comparison here can be strength-reduced,
but the second operand of the comparison happens to be identical
to the constant pointer in the `catch` case of `landingpad`.
While LSRInstance::CollectLoopInvariantFixupsAndFormulae()
already gave up on uses in blocks ending up with EH pads,
it didn't consider this case.
Eventually, `LSRInstance::AdjustInsertPositionForExpand()`
will be called, but the original insertion point it will get
is the user instruction itself, and it doesn't want to
deal with EH pads, and asserts as much.
It would seem that this basically never happens in-the-wild,
otherwise it would have been reported already,
so it seems safe to take the cautious approach,
and just not deal with such users.
If a memset destination is overwritten by a memcpy and the sizes
are exactly the same, then the memset is simply dead. We can
directly drop it, instead of replacing it with a memset of zero
size, which is particularly ugly for the case of a dynamic size.
This removes some (but not all) uses of type-less CreateGEP()
and CreateInBoundsGEP() APIs, which are incompatible with opaque
pointers.
There are a still a number of tricky uses left, as well as many
more variation APIs for CreateGEP.
As readnone function they become movable and LICM can hoist them
out of a loop. As a result in LCSSA form phi node of type token
is created. No one is ready that GCRelocate first operand is phi node
but expects to be token.
GVN test were also updated, it seems it does not do what is expected.
Test for LICM is also added.
This reverts commit f352463ade.
Relative to the previous implementation, this always uses
aliasesUnknownInst() instead of aliasesPointer() to correctly
handle atomics. The added test case was previously miscompiled.
-----
Even when MemorySSA-based LICM is used, an AST is still populated
for scalar promotion. As the AST has quadratic complexity, a lot
of time is spent in this step despite the existing access count
limit. This patch optimizes the identification of promotable stores.
The idea here is pretty simple: We're only interested in must-alias
mod sets of loop invariant pointers. As such, only populate the AST
with loop-invariant loads and stores (anything else is definitely
not promotable) and then discard any sets which alias with any of
the remaining, definitely non-promotable accesses.
If we promoted something, check whether this has made some other
accesses loop invariant and thus possible promotion candidates.
This is much faster in practice, because we need to perform AA
queries for O(NumPromotable^2 + NumPromotable*NumNonPromotable)
instead of O(NumTotal^2), and NumPromotable tends to be small.
Additionally, promotable accesses have loop invariant pointers,
for which AA is cheaper.
This has a signicant positive compile-time impact. We save ~1.8%
geomean on CTMark at O3, with 6% on lencod in particular and 25%
on individual files.
Conceptually, this change is NFC, but may not be so in practice,
because the AST is only an approximation, and can produce
different results depending on the order in which accesses are
added. However, there is at least no impact on the number of promotions
(licm.NumPromoted) in test-suite O3 configuration with this change.
Differential Revision: https://reviews.llvm.org/D89264
The isOverwrite function is making sure to identify if two stores
are fully overlapping and ideally we would like to identify all the
instances of OW_Complete as they'll yield possibly killable stores.
The current implementation is incapable of spotting instances where
the earlier store is offsetted compared to the later store, but
still fully overlapped. The limitation seems to lie on the
computation of the base pointers with the
GetPointerBaseWithConstantOffset API that often yields different
base pointers even if the stores are guaranteed to partially overlap
(e.g. the alias analysis is returning AliasResult::PartialAlias).
The patch relies on the offsets computed and cached by BatchAAResults
(available after D93529) to determine if the offsetted overlapping
is OW_Complete.
Differential Revision: https://reviews.llvm.org/D97676
Currently DSE misses cases where the size is a non-const IR value, even
if they match. For example, this means that llvm.memcpy/llvm.memset
calls are not eliminated, even if they write the same number of bytes.
This patch extends isOverwite to try to get IR values for the number of
bytes written from the analyzed instructions. If the values match,
alias checks are performed and the result is returned.
At the moment this only covers llvm.memcpy/llvm.memset. In the future,
we may enable MemoryLocation to also track variable sizes, but this
simple approach should allow us to cover the important cases in DSE.
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D98284
This patch updates the various IR passes to correctly handle dbg.values with a
DIArgList location. This patch does not actually allow DIArgLists to be produced
by salvageDebugInfo, and it does not affect any pass after codegen-prepare.
Other than that, it should cover every IR pass.
Most of the changes simply extend code that operated on a single debug value to
operate on the list of debug values in the style of any_of, all_of, for_each,
etc. Instances of setOperand(0, ...) have been replaced with with
replaceVariableLocationOp, which takes the value that is being replaced as an
additional argument. In places where this value isn't readily available, we have
to track the old value through to the point where it gets replaced.
Differential Revision: https://reviews.llvm.org/D88232
Revert 3d8f842712
Revision triggers a miscompile sinking a store incorrectly outside a
threading loop. Detected by tsan.
Reverting while investigating.
Differential Revision: https://reviews.llvm.org/D89264
This patch updates DbgVariableIntrinsics to support use of a DIArgList for the
location operand, resulting in a significant change to its interface. This patch
does not update all IR passes to support multiple location operands in a
dbg.value; the only change is to update the DbgVariableIntrinsic interface and
its uses. All code outside of the intrinsic classes assumes that an intrinsic
will always have exactly one location operand; they will still support
DIArgLists, but only if they contain exactly one Value.
Among other changes, the setOperand and setArgOperand functions in
DbgVariableIntrinsic have been made private. This is to prevent code from
setting the operands of these intrinsics directly, which could easily result in
incorrect/invalid operands being set. This does not prevent these functions from
being called on a debug intrinsic at all, as they can still be called on any
CallInst pointer; it is assumed that any code directly setting the operands on a
generic call instruction is doing so safely. The intention for making these
functions private is to prevent DIArgLists from being overwritten by code that's
naively trying to replace one of the Values it points to, and also to fail fast
if a DbgVariableIntrinsic is updated to use a DIArgList without a valid
corresponding DIExpression.
The check `tightlyNested()` in `LoopInterchange` is similar to the one in `LoopNest`.
In fact, the former misses some cases where loop-interchange is not feasible and results in incorrect behaviour.
Replacing it with the much robust version provided by `LoopNest` reduces code duplications and fixes https://bugs.llvm.org/show_bug.cgi?id=48113.
`LoopInterchange` has a weaker definition of tightly or perfectly nesting-ness than the one implemented in `LoopNest::arePerfectlyNested()`.
Therefore, `tightlyNested()` is instead implemented with `LoopNest::checkLoopsStructure` and additional checks for unsafe instructions.
Reviewed By: Whitney
Differential Revision: https://reviews.llvm.org/D97290
The MemorySSA-based implementation has been enabled without issue
for a while now, so keeping the old implementation around doesn't
seem useful anymore. This drops the MemDep-based implementation.
Differential Revision: https://reviews.llvm.org/D97877
Hello all,
I'm trying to fix unsafe propagation of poison values in and/or conditions by using
equivalent select forms (`select i1 A, i1 B, i1 false` and `select i1 A, i1 true, i1 false`)
instead.
D93065 has links to patches for this.
This patch allows unswitch to happen if the condition is in this form as well.
`collectHomogenousInstGraphLoopInvariants` is updated to keep traversal if
Root and the visiting I matches both m_LogicalOr()/m_LogicalAnd().
Other than this, the remaining changes are almost straightforward and simply replaces
Instruction::And/Or check with match(m_LogicalOr()/m_LogicalAnd()).
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D97756
When materializing an available load value, do not explicitly
materialize the undef values from dead blocks. Doing so will
will force creation of a phi with an undef operand, even if there
is a dominating definition. The phi will be folded away on
subsequent GVN iterations, but by then we may have already
poisoned MDA cache slots.
Simply don't register these values in the first place, and let
SSAUpdater do its thing.
GVN basically doesn't handle phi nodes at all. This is for a reason - we can't value number their inputs since the predecessor blocks have probably not been visited yet.
However, it also creates a significant pass ordering problem. As it stands, instcombine and simplifycfg ends up implementing CSE of phi nodes. This means that for any series of CSE opportunities intermixed with phi nodes, we end up having to alternate instcombine/simplifycfg and gvn to make progress.
This patch handles the simplest case by simply preprocessing the phi instructions in a block, and CSEing them if they are syntactically identical. This turns out to be powerful enough to handle many cases in a single invocation of GVN since blocks which use the cse'd phi results are visited after the block containing the phi. If there's a CSE opportunity in one the phi predecessors required to recognize the phi CSE opportunity, that will require a second iteration on the function. (Still within a single run of gvn though.)
Compile time wise, this could go either way. On one hand, we're potentially causing GVN to iterate over the function more. On the other, we're cutting down on iterations between two passes and potentially shrinking the IR aggressively. So, a bit unclear what to expect.
Note that this does still rely on instcombine to canonicalize block order of the phis, but that's a one time transformation independent of the values incoming to the phi.
Differential Revision: https://reviews.llvm.org/D98080
The last two operands to a gc.relocate represent indices into the associated gc.statepoint's gc bundle list. (Effectively, gc.relocates are projections from the gc.statepoints multiple return values.)
We can use this to recognize when two gc.relocates are equivalent (and can be CSEd), even when the indices are non-equal. This is particular useful when considering a chain of multiple statepoints as it lets us eliminate all duplicate gc.relocates in a single pass.
Differential Revision: https://reviews.llvm.org/D97974
(Note: Part of the reviewed change was split and landed as f352463a)
For some reason, we had been marking gc.relocates as reading memory. There's no known reason for this, and I suspect it to be a legacy of very early implementation conservatism. gc.relocate and gc.result are simply projections of the return values from the associated statepoint. Note that the LangRef has always declared them readnone.
The EarlyCSE change is simply moving the special casing from readonly to readnone handling.
As noted by the test diffs, this does allow some additional CSE when relocates are separated by stores, but since we generate gc.relocates in batches, this is unlikely to help anything in practice.
This was reviewed as part of https://reviews.llvm.org/D97974, but split at reviewer request before landing. The motivation is to enable the GVN changes in that patch.
If we have a value live over a call which is used for deopt at the call, we know that the value must be a base pointer. We can avoid potentially inserting IR to materialize a base for this value.
In it's current form, this is mostly a compile time optimization. Building the base pointer graph (and then optimizing it away again) is a relatively expensive operation. We also sometimes end up with better codegen in practice - due to failures in optimizing away the inserted base pointer propogation - but those are optimization bugs we're fixing concurrently.
The alternative to this would be to extend the base pointer inference with the ability to generally reuse multiple-base input instructions (phis and selects). That's somewhat invasive and complicated, so we're defering it a bit longer.
Differential Revision: https://reviews.llvm.org/D97885
explicitly emitting retainRV or claimRV calls in the IR
This reapplies ed4718eccb, which was reverted
because it was causing a miscompile. The bug that was causing the miscompile
has been fixed in 75805dce5f.
Original commit message:
Background:
This fixes a longstanding problem where llvm breaks ARC's autorelease
optimization (see the link below) by separating calls from the marker
instructions or retainRV/claimRV calls. The backend changes are in
https://reviews.llvm.org/D92569.
https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-autoreleasereturnvalue
What this patch does to fix the problem:
- The front-end adds operand bundle "clang.arc.attachedcall" to calls,
which indicates the call is implicitly followed by a marker
instruction and an implicit retainRV/claimRV call that consumes the
call result. In addition, it emits a call to
@llvm.objc.clang.arc.noop.use, which consumes the call result, to
prevent the middle-end passes from changing the return type of the
called function. This is currently done only when the target is arm64
and the optimization level is higher than -O0.
- ARC optimizer temporarily emits retainRV/claimRV calls after the calls
with the operand bundle in the IR and removes the inserted calls after
processing the function.
- ARC contract pass emits retainRV/claimRV calls after the call with the
operand bundle. It doesn't remove the operand bundle on the call since
the backend needs it to emit the marker instruction. The retainRV and
claimRV calls are emitted late in the pipeline to prevent optimization
passes from transforming the IR in a way that makes it harder for the
ARC middle-end passes to figure out the def-use relationship between
the call and the retainRV/claimRV calls (which is the cause of
PR31925).
- The function inliner removes an autoreleaseRV call in the callee if
nothing in the callee prevents it from being paired up with the
retainRV/claimRV call in the caller. It then inserts a release call if
claimRV is attached to the call since autoreleaseRV+claimRV is
equivalent to a release. If it cannot find an autoreleaseRV call, it
tries to transfer the operand bundle to a function call in the callee.
This is important since the ARC optimizer can remove the autoreleaseRV
returning the callee result, which makes it impossible to pair it up
with the retainRV/claimRV call in the caller. If that fails, it simply
emits a retain call in the IR if retainRV is attached to the call and
does nothing if claimRV is attached to it.
- SCCP refrains from replacing the return value of a call with a
constant value if the call has the operand bundle. This ensures the
call always has at least one user (the call to
@llvm.objc.clang.arc.noop.use).
- This patch also fixes a bug in replaceUsesOfNonProtoConstant where
multiple operand bundles of the same kind were being added to a call.
Future work:
- Use the operand bundle on x86-64.
- Fix the auto upgrader to convert call+retainRV/claimRV pairs into
calls with the operand bundles.
rdar://71443534
Differential Revision: https://reviews.llvm.org/D92808
Same dangling probes are redundant since they all have the same semantic that is to rely on the counts inference tool to get reasonable count for the same original block. Therefore, there's no need to keep multiple copies of them. I've seen jump threading created tons of redundant dangling probes that slowed down the compiler dramatically. Other optimization passes can also result in redundant probes though without an observed impact so far.
This change removes block-wise redundant dangling probes specifically introduced by jump threading. To support removing redundant dangling probes caused by all other passes, a final function-wise deduplication is also added.
An 18% size win of the .pseudo_probe section was seen for SPEC2017. No performance difference was observed.
Differential Revision: https://reviews.llvm.org/D97482
This change fixes a couple places where the pseudo probe intrinsic blocks optimizations because they are not naturally removable. To unblock those optimizations, the blocking pseudo probes are moved out of the original blocks and tagged dangling, instead of allowing pseudo probes to be literally removed. The reason is that when the original block is removed, we won't be able to sample it. Instead of assigning it a zero weight, moving all its pseudo probes into another block and marking them dangling should allow the counts inference a chance to assign them a more reasonable weight. We have not seen counts quality degradation from our experiments.
The optimizations being unblocked are:
1. Removing conditional probes for if-converted branches. Conditional probes are tagged dangling when their homing branch arms are folded so that they will not be over-counted.
2. Unblocking jump threading from removing empty blocks. Pseudo probe prevents jump threading from removing logically empty blocks that only has one unconditional jump instructions.
3. Unblocking SimplifyCFG and MIR tail duplicate to thread empty blocks and blocks with redundant branch checks.
Since dangling probes are logically deleted, they should not consume any samples in LTO postLink. This can be achieved by setting their distribution factors to zero when dangled.
Reviewed By: wmi
Differential Revision: https://reviews.llvm.org/D97481
This is an attempt to improve handling of partial overlaps in case of unaligned begin\end.
Existing implementation just bails out if it encounters such cases. Even when it doesn't I believe existing code checking alignment constraints is not quite correct. It tries to ensure alignment of the "later" start/end offset while should be preserving relative alignment between earlier and later start/end.
The idea behind the change is simple. When start/end is not aligned as we wish instead of bailing out let's adjust it as necessary to get desired alignment.
I'll update with performance results as measured by the test-suite...it's still running...
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D93530
See pr46990(https://bugs.llvm.org/show_bug.cgi?id=46990). LICM should not sink store instructions to loop exit blocks which cross coro.suspend intrinsics. This breaks semantic of coro.suspend intrinsic which return to caller directly. Also this leads to use-after-free if the coroutine is freed before control returns to the caller in multithread environment.
This patch disable promotion by check whether loop contains coro.suspend intrinsics.
This is a resubmit of D86190.
Disabling LICM for loops with coroutine suspension is a better option not only for correctness purpose but also for performance purpose.
In most cases LICM sinks memory operations. In the case of coroutine, sinking memory operation out of the loop does not improve performance since coroutien needs to get data from the frame anyway. In fact LICM would hurt coroutine performance since it adds more entries to the frame.
Differential Revision: https://reviews.llvm.org/D96928
Probably should have done this before landing, but I forgot.
Basic idea is to avoid using the SCEV predicate when it doesn't buy us anything. Also happens to set us up for handling non-add recurrences in the future if desired.
LSR goes to some lengths to schedule IV increments such that %iv and %iv.next never need to overlap. This is fairly fundamental to LSRs cost model. LSR assumes that an addrec can be represented with a single register. If %iv and %iv.next have to overlap, then that assumption does not hold.
The bug - which this patch is fixing - is that LSR only does this scheduling for IVs which it inserts, but it's cost model assumes the same for existing IVs that it reuses. It will rewrite existing IV users such that the no-overlap property holds, but will not actually reschedule said IV increment.
As you can see from the relatively lack of test updates, this doesn't actually impact codegen much. The main reason for doing it is to make a follow up patch series which improves post-increment use and scheduling easier to follow.
Differential Revision: https://reviews.llvm.org/D97219
This caused miscompiles of Chromium tests for iOS due clobbering of live
registers. See discussion on the code review for details.
> Background:
>
> This fixes a longstanding problem where llvm breaks ARC's autorelease
> optimization (see the link below) by separating calls from the marker
> instructions or retainRV/claimRV calls. The backend changes are in
> https://reviews.llvm.org/D92569.
>
> https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-autoreleasereturnvalue
>
> What this patch does to fix the problem:
>
> - The front-end adds operand bundle "clang.arc.attachedcall" to calls,
> which indicates the call is implicitly followed by a marker
> instruction and an implicit retainRV/claimRV call that consumes the
> call result. In addition, it emits a call to
> @llvm.objc.clang.arc.noop.use, which consumes the call result, to
> prevent the middle-end passes from changing the return type of the
> called function. This is currently done only when the target is arm64
> and the optimization level is higher than -O0.
>
> - ARC optimizer temporarily emits retainRV/claimRV calls after the calls
> with the operand bundle in the IR and removes the inserted calls after
> processing the function.
>
> - ARC contract pass emits retainRV/claimRV calls after the call with the
> operand bundle. It doesn't remove the operand bundle on the call since
> the backend needs it to emit the marker instruction. The retainRV and
> claimRV calls are emitted late in the pipeline to prevent optimization
> passes from transforming the IR in a way that makes it harder for the
> ARC middle-end passes to figure out the def-use relationship between
> the call and the retainRV/claimRV calls (which is the cause of
> PR31925).
>
> - The function inliner removes an autoreleaseRV call in the callee if
> nothing in the callee prevents it from being paired up with the
> retainRV/claimRV call in the caller. It then inserts a release call if
> claimRV is attached to the call since autoreleaseRV+claimRV is
> equivalent to a release. If it cannot find an autoreleaseRV call, it
> tries to transfer the operand bundle to a function call in the callee.
> This is important since the ARC optimizer can remove the autoreleaseRV
> returning the callee result, which makes it impossible to pair it up
> with the retainRV/claimRV call in the caller. If that fails, it simply
> emits a retain call in the IR if retainRV is attached to the call and
> does nothing if claimRV is attached to it.
>
> - SCCP refrains from replacing the return value of a call with a
> constant value if the call has the operand bundle. This ensures the
> call always has at least one user (the call to
> @llvm.objc.clang.arc.noop.use).
>
> - This patch also fixes a bug in replaceUsesOfNonProtoConstant where
> multiple operand bundles of the same kind were being added to a call.
>
> Future work:
>
> - Use the operand bundle on x86-64.
>
> - Fix the auto upgrader to convert call+retainRV/claimRV pairs into
> calls with the operand bundles.
>
> rdar://71443534
>
> Differential Revision: https://reviews.llvm.org/D92808
This reverts commit ed4718eccb.
Even when MemorySSA-based LICM is used, an AST is still populated
for scalar promotion. As the AST has quadratic complexity, a lot
of time is spent in this step despite the existing access count
limit. This patch optimizes the identification of promotable stores.
The idea here is pretty simple: We're only interested in must-alias
mod sets of loop invariant pointers. As such, only populate the AST
with loop-invariant loads and stores (anything else is definitely
not promotable) and then discard any sets which alias with any of
the remaining, definitely non-promotable accesses.
If we promoted something, check whether this has made some other
accesses loop invariant and thus possible promotion candidates.
This is much faster in practice, because we need to perform AA
queries for O(NumPromotable^2 + NumPromotable*NumNonPromotable)
instead of O(NumTotal^2), and NumPromotable tends to be small.
Additionally, promotable accesses have loop invariant pointers,
for which AA is cheaper.
This has a signicant positive compile-time impact. We save ~1.8%
geomean on CTMark at O3, with 6% on lencod in particular and 25%
on individual files.
Conceptually, this change is NFC, but may not be so in practice,
because the AST is only an approximation, and can produce
different results depending on the order in which accesses are
added. However, there is at least no impact on the number of promotions
(licm.NumPromoted) in test-suite O3 configuration with this change.
Differential Revision: https://reviews.llvm.org/D89264
This now analyzes calls to both intrinsics and functions.
For intrinsics, grab the ones we know and care about (mem* family) and
analyze the arguments.
For calls, use TLI to get more information about the libcalls, then
analyze the arguments if known.
```
auto-init.c:4:7: remark: Call to memset inserted by -ftrivial-auto-var-init. Memory operation size: 4096 bytes. [-Rpass-missed=annotation-remarks]
int var[1024];
^
```
Differential Revision: https://reviews.llvm.org/D97489
This adds support for analyzing the instruction with the !annotation
"auto-init" in order to generate a more user-friendly remark.
For now, support the store size, and whether it's atomic/volatile.
Example:
```
auto-init.c:4:7: remark: Store inserted by -ftrivial-auto-var-init.Store size: 4 bytes. [-Rpass-missed=annotation-remarks]
int var;
^
```
Differential Revision: https://reviews.llvm.org/D97412
Using the !annotation metadata, emit remarks pointing to code added by
`-ftrivial-auto-var-init` that survived the optimizer.
Example:
```
auto-init.c:4:7: remark: Initialization inserted by -ftrivial-auto-var-init. [-Rpass-missed=annotation-remarks]
int buf[1024];
^
```
The tests are testing various situations like calls/stores/other
instructions, with debug locations, and extra debug information on
purpose: more patches will come to improve the reporting to make it more
user-friendly, and these tests will show how the reporting evolves.
Differential Revision: https://reviews.llvm.org/D97405
Support reassociation for min/max. With that we should be able to transform min(min(a, b), c) -> min(min(a, c), b) if min(a, c) is already available.
Reviewed By: mkazantsev
Differential Revision: https://reviews.llvm.org/D88287
The **IsGuaranteedLoopInvariant** function is making sure to check if the
incoming pointer is guaranteed to be loop invariant, therefore I think
the case where the pointer is defined in the entry block of a function
automatically guarantees the pointer to be loop invariant, as the entry
block of a function cannot have predecessors or be part of a loop.
I implemented this small patch and tested it using
**ninja check-llvm-unit** and **ninja check-llvm**. I added a contained test
file that shows the problem and used **opt -O3 -debug** on it to make sure
the case is not currently handled (in fact the debug log is showing that
the DSE pass is bailing out when testing if the killer store is able to
clobber the dead store).
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D96979
ICMP_NE predicates cannot be directly represented as constraint. But we
can use ICMP_UGT instead ICMP_NE for %x != 0.
See https://alive2.llvm.org/ce/z/XlLCsW
If the call is readnone, then there may not be any MemoryAccess
associated with the call. Bail out in that case.
This fixes the issue reported at
https://reviews.llvm.org/D94376#2578312.
When cloning instructions during jump threading, also clone and
adapt any declared scopes. This is primarily important when
threading loop exits, because we'll end up with two dominating
scope declarations in that case (at least after additional loop
rotation). This addresses a loose thread from
https://reviews.llvm.org/rG2556b413a7b8#975012.
Differential Revision: https://reviews.llvm.org/D97154
This enables use of MemorySSA instead of MemDep in MemCpyOpt. To
allow this without significant compile-time impact, the MemCpyOpt
pass is moved directly before DSE (in the cases where this was not
already the case), which allows us to reuse the existing MemorySSA
analysis.
Unlike the MemDep-based implementation, the MemorySSA-based MemCpyOpt
can also perform simple optimizations across basic blocks.
Differential Revision: https://reviews.llvm.org/D94376
In both ADCE and BDCE (via DemandedBits) we should not remove
instructions that are not guaranteed to return. This issue was
pointed out by fhahn in the recent llvm-dev thread.
Differential Revision: https://reviews.llvm.org/D96993
This fixes https://bugs.llvm.org/show_bug.cgi?id=49185
When `NDEBUG` is not set, `LPMUpdater` checks if the added loops have the same parent loop as the current one in `addSiblingLoops`.
If multiple loop passes are executed through `LoopPassManager`, `U.ParentL` will be the same across all passes.
However, the parent loop might change after running a loop pass, resulting in assertion failures in subsequent passes.
This patch resets `U.ParentL` after running individual loop passes in `LoopPassManager`.
Reviewed By: asbirlea, ychen
Differential Revision: https://reviews.llvm.org/D96727
SROA does not correctly account for offsets in TBAA/TBAA struct metadata.
This patch creates functionality for generating new MD with the corresponding
offset and updates SROA to use this functionality.
Differential Revision: https://reviews.llvm.org/D95826
This is the preliminary patch of converting `LoopInterchange` pass to a loop-nest pass and has no intended functional change.
Changes that are not loop-nest related are split to D96650.
Reviewed By: Whitney
Differential Revision: https://reviews.llvm.org/D96644
This adds a new flag -lsr-preferred-addressing-mode to override the target's
preferred addressing mode. It replaces flag -lsr-backedge-indexing, which is
equivalent to preindexed addressing that is one of the options that
-lsr-preferred-addressing-mode accepts.
Differential Revision: https://reviews.llvm.org/D96855
This is a follow up D96600 and cleans up most calls to
getPreferredAddresingMode. I.e., we really don't need to query the same things
again and again, but get the preferred addressing mode once for each loop. So
this should be a lot friendlier for compile times, especially if we start
implementing getPreferredAddresingMode.
Differential Revision: https://reviews.llvm.org/D96772
This is a split patch of D96644.
Explicitly pass both `InnerLoop` and `OuterLoop` to function `processLoop` to remove the need to swap elements in loop list and allow making loop list an `ArrayRef`.
Also, fix inconsistent spellings of `OuterLoopId` and `Inner Loop Id` in debug log.
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D96650
Loop canonicalization may end up deleting blocks from CFG. And
Scalar Evolution may still keep cached referenced to those blocks
unless updated properly.
This refactors shouldFavorPostInc() and shouldFavorBackedgeIndex() into
getPreferredAddressingMode() so that we have one interface to steer LSR in
generating the preferred addressing mode.
Differential Revision: https://reviews.llvm.org/D96600
explicitly emitting retainRV or claimRV calls in the IR
Background:
This fixes a longstanding problem where llvm breaks ARC's autorelease
optimization (see the link below) by separating calls from the marker
instructions or retainRV/claimRV calls. The backend changes are in
https://reviews.llvm.org/D92569.
https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-autoreleasereturnvalue
What this patch does to fix the problem:
- The front-end adds operand bundle "clang.arc.attachedcall" to calls,
which indicates the call is implicitly followed by a marker
instruction and an implicit retainRV/claimRV call that consumes the
call result. In addition, it emits a call to
@llvm.objc.clang.arc.noop.use, which consumes the call result, to
prevent the middle-end passes from changing the return type of the
called function. This is currently done only when the target is arm64
and the optimization level is higher than -O0.
- ARC optimizer temporarily emits retainRV/claimRV calls after the calls
with the operand bundle in the IR and removes the inserted calls after
processing the function.
- ARC contract pass emits retainRV/claimRV calls after the call with the
operand bundle. It doesn't remove the operand bundle on the call since
the backend needs it to emit the marker instruction. The retainRV and
claimRV calls are emitted late in the pipeline to prevent optimization
passes from transforming the IR in a way that makes it harder for the
ARC middle-end passes to figure out the def-use relationship between
the call and the retainRV/claimRV calls (which is the cause of
PR31925).
- The function inliner removes an autoreleaseRV call in the callee if
nothing in the callee prevents it from being paired up with the
retainRV/claimRV call in the caller. It then inserts a release call if
claimRV is attached to the call since autoreleaseRV+claimRV is
equivalent to a release. If it cannot find an autoreleaseRV call, it
tries to transfer the operand bundle to a function call in the callee.
This is important since the ARC optimizer can remove the autoreleaseRV
returning the callee result, which makes it impossible to pair it up
with the retainRV/claimRV call in the caller. If that fails, it simply
emits a retain call in the IR if retainRV is attached to the call and
does nothing if claimRV is attached to it.
- SCCP refrains from replacing the return value of a call with a
constant value if the call has the operand bundle. This ensures the
call always has at least one user (the call to
@llvm.objc.clang.arc.noop.use).
- This patch also fixes a bug in replaceUsesOfNonProtoConstant where
multiple operand bundles of the same kind were being added to a call.
Future work:
- Use the operand bundle on x86-64.
- Fix the auto upgrader to convert call+retainRV/claimRV pairs into
calls with the operand bundles.
rdar://71443534
Differential Revision: https://reviews.llvm.org/D92808
This reverts commit b7d870eae7 and the
subsequent fix "[Polly] Fix build after AssumptionCache change (D96168)"
(commit e6810cab09).
It caused indeterminism in the output, such that e.g. the
polly-x86_64-linux buildbot failed accasionally.
This patch improves the index management during constraint building.
Previously, the code rejected constraints which used values that were not
part of Value2Index, but after combining the coefficients of the new
indices were 0 (if ShouldAdd was 0).
In those cases, no new indices need to be added. Instead of adding to
Value2Index directly, add new indices to the NewIndices map. The caller
can then check if it needs to add any new indices.
This enables checking constraints like `a + x <= a + n` to `x <= n`,
even if there is no constraint for `a` directly.
As discussed in:
https://llvm.org/PR49055
We invert instcombine's add->or transform here
because it makes it easier to identify factorization
transforms like the mul in the motivating test.
This extends the logic added with:
https://reviews.llvm.org/rG70472f3https://reviews.llvm.org/rG93f3d7f
(I intentionally kept the formatting fix in this patch
to provide more context about the calling logic.)
PR49043 exposed a problem when it comes to RAUW llvm.assumes. While
D96106 would fix it for GVNSink, it seems a more general concern. To
avoid future problems this patch moves away from the vector of weak
reference model used in the assumption cache. Instead, we track the
llvm.assume calls with a callback handle which will remove itself from
the cache if the call is deleted.
Fixes PR49043.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D96168
emitting retainRV or claimRV calls in the IR
This reapplies 3fe3946d9a without the
changes made to lib/IR/AutoUpgrade.cpp, which was violating layering.
Original commit message:
Background:
This patch makes changes to the front-end and middle-end that are
needed to fix a longstanding problem where llvm breaks ARC's autorelease
optimization (see the link below) by separating calls from the marker
instructions or retainRV/claimRV calls. The backend changes are in
https://reviews.llvm.org/D92569.
https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-autoreleasereturnvalue
What this patch does to fix the problem:
- The front-end adds operand bundle "clang.arc.rv" to calls, which
indicates the call is implicitly followed by a marker instruction and
an implicit retainRV/claimRV call that consumes the call result. In
addition, it emits a call to @llvm.objc.clang.arc.noop.use, which
consumes the call result, to prevent the middle-end passes from changing
the return type of the called function. This is currently done only when
the target is arm64 and the optimization level is higher than -O0.
- ARC optimizer temporarily emits retainRV/claimRV calls after the calls
with the operand bundle in the IR and removes the inserted calls after
processing the function.
- ARC contract pass emits retainRV/claimRV calls after the call with the
operand bundle. It doesn't remove the operand bundle on the call since
the backend needs it to emit the marker instruction. The retainRV and
claimRV calls are emitted late in the pipeline to prevent optimization
passes from transforming the IR in a way that makes it harder for the
ARC middle-end passes to figure out the def-use relationship between
the call and the retainRV/claimRV calls (which is the cause of
PR31925).
- The function inliner removes an autoreleaseRV call in the callee if
nothing in the callee prevents it from being paired up with the
retainRV/claimRV call in the caller. It then inserts a release call if
the call is annotated with claimRV since autoreleaseRV+claimRV is
equivalent to a release. If it cannot find an autoreleaseRV call, it
tries to transfer the operand bundle to a function call in the callee.
This is important since ARC optimizer can remove the autoreleaseRV
returning the callee result, which makes it impossible to pair it up
with the retainRV/claimRV call in the caller. If that fails, it simply
emits a retain call in the IR if the implicit call is a call to
retainRV and does nothing if it's a call to claimRV.
Future work:
- Use the operand bundle on x86-64.
- Fix the auto upgrader to convert call+retainRV/claimRV pairs into
calls annotated with the operand bundles.
rdar://71443534
Differential Revision: https://reviews.llvm.org/D92808
emitting retainRV or claimRV calls in the IR
Background:
This patch makes changes to the front-end and middle-end that are
needed to fix a longstanding problem where llvm breaks ARC's autorelease
optimization (see the link below) by separating calls from the marker
instructions or retainRV/claimRV calls. The backend changes are in
https://reviews.llvm.org/D92569.
https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-autoreleasereturnvalue
What this patch does to fix the problem:
- The front-end adds operand bundle "clang.arc.rv" to calls, which
indicates the call is implicitly followed by a marker instruction and
an implicit retainRV/claimRV call that consumes the call result. In
addition, it emits a call to @llvm.objc.clang.arc.noop.use, which
consumes the call result, to prevent the middle-end passes from changing
the return type of the called function. This is currently done only when
the target is arm64 and the optimization level is higher than -O0.
- ARC optimizer temporarily emits retainRV/claimRV calls after the calls
with the operand bundle in the IR and removes the inserted calls after
processing the function.
- ARC contract pass emits retainRV/claimRV calls after the call with the
operand bundle. It doesn't remove the operand bundle on the call since
the backend needs it to emit the marker instruction. The retainRV and
claimRV calls are emitted late in the pipeline to prevent optimization
passes from transforming the IR in a way that makes it harder for the
ARC middle-end passes to figure out the def-use relationship between
the call and the retainRV/claimRV calls (which is the cause of
PR31925).
- The function inliner removes an autoreleaseRV call in the callee if
nothing in the callee prevents it from being paired up with the
retainRV/claimRV call in the caller. It then inserts a release call if
the call is annotated with claimRV since autoreleaseRV+claimRV is
equivalent to a release. If it cannot find an autoreleaseRV call, it
tries to transfer the operand bundle to a function call in the callee.
This is important since ARC optimizer can remove the autoreleaseRV
returning the callee result, which makes it impossible to pair it up
with the retainRV/claimRV call in the caller. If that fails, it simply
emits a retain call in the IR if the implicit call is a call to
retainRV and does nothing if it's a call to claimRV.
Future work:
- Use the operand bundle on x86-64.
- Fix the auto upgrader to convert call+retainRV/claimRV pairs into
calls annotated with the operand bundles.
rdar://71443534
Differential Revision: https://reviews.llvm.org/D92808
This patch extends the condition collection logic to allow adding
conditions from pre-headers to loop headers, by allowing cases where the
target block dominates some of its predecessors.
A == B map to A >= B && A <= B
(https://alive2.llvm.org/ce/z/_dwxKn).
This extends the constraint construction to return a list of
constraints, which can be used to properly de-compose nested AND & OR.
If the incoming block to a phi node is an EH pad, then we will
materialize into an EH pad, which is not supposed to happen. To fix
this, I added a check to see if incoming block of a phi node is an EH
pad before using it as the insertion point.
Differential Revision: https://reviews.llvm.org/D95019
Instead of using ConstraintSystem::negate when adding new constraints,
flip the condition in IR.
The main advantage is that EQ predicates can be represented by 2
constraints, which makes negating based on the constraint tricky. The IR
condition can easily negated.
If we determine that the invariant path through the loop has no effects,
we can directly branch to the exit block, instead to unswitching first.
Besides avoiding some extra work (unswitching first, then deleting the
loop again) this allows to be more aggressive than regular unswitching
with respect to cost-modeling. This approach should always be be
desirable.
This is similar in spirit to D93734, just that it uses the previously
added checks for loop-unswitching.
I tried to add the required no-op checks from scratch, as we only check
a subset of the loop. There is potential to unify the checks with
LoopDeletion, at the cost of adding a predicate whether a block should
be considered.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D95468
This patch fixes updating MemorySSA if the header contains memory
defs that do not clobber a duplicated instruction. We need to find the
first defining access outside the loop body and use that as defining
access of the duplicated instruction.
This fixes a crash caused by bee486851c.
GCC warning:
```
/llvm-project/llvm/lib/Transforms/Scalar/ScalarizeMaskedMemIntrin.cpp: In function ‘void scalarizeMaskedStore(llvm::CallInst*, llvm::DomTreeUpdater*, bool&)’:
/llvm-project/llvm/lib/Transforms/Scalar/ScalarizeMaskedMemIntrin.cpp:295:15: warning: variable ‘IfBlock’ set but not used [-Wunused-but-set-variable]
295 | BasicBlock *IfBlock = CI->getParent();
| ^~~~~~~
/llvm-project/llvm/lib/Transforms/Scalar/ScalarizeMaskedMemIntrin.cpp: In function ‘void scalarizeMaskedScatter(llvm::CallInst*, llvm::DomTreeUpdater*, bool&)’:
/llvm-project/llvm/lib/Transforms/Scalar/ScalarizeMaskedMemIntrin.cpp:555:15: warning: variable ‘IfBlock’ set but not used [-Wunused-but-set-variable]
555 | BasicBlock *IfBlock = CI->getParent();
| ^~~~~~~
```
This de-pessimizes the arguably more usual case of no masked mem intrinsics,
and gets rid of one more Dominator Tree recalculation.
As per llvm/test/CodeGen/X86/opt-pipeline.ll,
there's one more Dominator Tree recalculation left, we could get rid of.
This patch adds additional checks to avoid partial unswitching
in cases where it won't be profitable, e.g. because the path directly
exits the loop anyways.
Loop peeling removes conditions from loop bodies that become invariant
after a small number of iterations. When triggered, this leads to fewer
compares and possibly PHIs in loop bodies, enabling further
optimizations. The current cost-model of loop peeling should be quite
conservative/safe, i.e. only peel if a condition in the loop becomes
known after peeling.
For example, see PR47671, where loop peeling enables vectorization by
removing a PHI the vectorizer does not understand. Granted, the
loop-vectorizer could also be taught about constant PHIs, but loop
peeling is likely to enable other optimizations as well.
This has an impact on quite a few benchmarks from
MultiSource/SPEC2000/SPEC2006 on X86 with -O3 -flto, for example
Same hash: 186 (filtered out)
Remaining: 51
Metric: loop-vectorize.LoopsVectorized
Program base patch diff
test-suite...ve-susan/automotive-susan.test 8.00 9.00 12.5%
test-suite...nal/skidmarks10/skidmarks.test 35.00 31.00 -11.4%
test-suite...lications/sqlite3/sqlite3.test 41.00 43.00 4.9%
test-suite...s/ASC_Sequoia/AMGmk/AMGmk.test 25.00 26.00 4.0%
test-suite...006/450.soplex/450.soplex.test 88.00 89.00 1.1%
test-suite...TimberWolfMC/timberwolfmc.test 120.00 119.00 -0.8%
test-suite.../CINT2006/403.gcc/403.gcc.test 215.00 216.00 0.5%
test-suite...006/447.dealII/447.dealII.test 957.00 958.00 0.1%
test-suite...ternal/HMMER/hmmcalibrate.test 75.00 75.00 0.0%
Same hash: 186 (filtered out)
Remaining: 51
Metric: loop-vectorize.LoopsAnalyzed
Program base patch diff
test-suite...ks/Prolangs-C/agrep/agrep.test 440.00 434.00 -1.4%
test-suite...nal/skidmarks10/skidmarks.test 312.00 308.00 -1.3%
test-suite...marks/7zip/7zip-benchmark.test 6399.00 6323.00 -1.2%
test-suite...lications/minisat/minisat.test 134.00 135.00 0.7%
test-suite...rks/FreeBench/pifft/pifft.test 295.00 297.00 0.7%
test-suite...TimberWolfMC/timberwolfmc.test 1879.00 1869.00 -0.5%
test-suite...pplications/treecc/treecc.test 689.00 691.00 0.3%
test-suite...T2000/300.twolf/300.twolf.test 1593.00 1597.00 0.3%
test-suite.../Benchmarks/Bullet/bullet.test 1394.00 1392.00 -0.1%
test-suite...ications/JM/ldecod/ldecod.test 1431.00 1429.00 -0.1%
test-suite...6/464.h264ref/464.h264ref.test 2229.00 2230.00 0.0%
test-suite...lications/sqlite3/sqlite3.test 2590.00 2589.00 -0.0%
test-suite...ications/JM/lencod/lencod.test 2732.00 2733.00 0.0%
test-suite...006/453.povray/453.povray.test 3395.00 3394.00 -0.0%
Note the -11% regression in number of loops vectorized for skidmarks. I
suspect this corresponds to the fact that those loops are gone now (see
the reduction in number of loops analyzed by LV).
Reviewed By: lebedev.ri
Differential Revision: https://reviews.llvm.org/D88471
Fixes an infinite loop encountered in GVN.
GVN will delay PRE if it encounters critical edges, attempt to split
them later via calls to SplitCriticalEdge(), then restart.
The caller of GVN::splitCriticalEdges() assumed a return value of true
meant that critical edges were split, that the IR had changed, and that
PRE should be re-attempted, upon which we loop infinitely.
This was exposed after D88438, by compiling the Linux kernel for s390,
but the test case is reproducible on x86.
Fixes: https://github.com/ClangBuiltLinux/linux/issues/1261
Reviewed By: void
Differential Revision: https://reviews.llvm.org/D94996
If i change it to AssertingVH instead, a number of existing tests fail,
which means we don't consistently remove from the set when deleting blocks,
which means newly-created blocks may happen to appear in that set
if they happen to occupy the same memory chunk as did some block
that was in the set originally.
There are many places where we delete blocks,
and while we could probably consistently delete from LoopHeaders
when deleting a block in transforms located in SimplifyCFG.cpp itself,
transforms located elsewhere (Local.cpp/BasicBlockUtils.cpp) also may
delete blocks, and it doesn't seem good to teach them to deal with it.
Since we at most only ever delete from LoopHeaders,
let's just delegate to WeakVH to do that automatically.
But to be honest, personally, i'm not sure that the idea
behind LoopHeaders is sound.
This builds on the restricted after initial revert form of D93906, and adds back support for breaking backedges of inner loops. It turns out the original invalidation logic wasn't quite right, specifically around the handling of LCSSA.
When breaking the backedge of an inner loop, we can cause blocks which were in the outer loop only because they were also included in a sub-loop to be removed from both loops. This results in the exit block set for our original parent loop changing, and thus a need for new LCSSA phi nodes.
This case happens when the inner loop has an exit block which is also an exit block of the parent, and there's a block in the child which reaches an exit to said block without also reaching an exit to the parent loop.
(I'm describing this in terms of the immediate parent, but the problem is general for any transitive parent in the nest.)
The approach implemented here involves a potentially expensive LCSSA rebuild. Perf testing during review didn't show anything concerning, but we may end up needing to revert this if anyone encounters a practical compile time issue.
Differential Revision: https://reviews.llvm.org/D94378
Similar to binary operators like fadd/fmul/fsub, propagate shape info
through unary operators (fneg is the only one?).
Differential Revision: https://reviews.llvm.org/D95252
The existing code did not deal with atomic loads correctly. Such loads
are represented as MemoryDefs. Bail out on any MemoryAccess that is not
a MemoryUse.
In LoopInterchange, `findInnerReductionPhi()` looks for reduction
variables, which cannot be constants. Update it to return early in that
case.
This also addresses a blocker for removing use-lists from ConstantData,
whose users could be spread across arbitrary modules in the same
LLVMContext.
Differential Revision: https://reviews.llvm.org/D94712
This patch applies the idea from D93734 to LoopUnswitch.
It adds support for unswitching on conditions that are only
invariant along certain paths through a loop.
In particular, it targets conditions in the loop header that
depend on values loaded from memory. If either path from
the true or false successor through the loop does not modify
memory, perform partial loop unswitching.
That is, duplicate the instructions feeding the condition in the pre-header.
Then unswitch on the duplicated condition. The condition is now known
in the unswitched version for the 'invariant' path through the original loop.
On caveat of this approach is that one of the loops created can be partially
unswitched again. To avoid this behavior, `llvm.loop.unswitch.partial.disable`
metadata is added to the unswitched loops, to avoid subsequent partial
unswitching.
If that's the approach to go, I can move the code handling the metadata kind
into separate functions.
This increases the cases we unswitch quite a bit in SPEC2006/SPEC2000 &
MultiSource. It also allows us to eliminate a dead loop in SPEC2017's omnetpp
```
Tests: 236
Same hash: 170 (filtered out)
Remaining: 66
Metric: loop-unswitch.NumBranches
Program base patch diff
test-suite...000/255.vortex/255.vortex.test 2.00 23.00 1050.0%
test-suite...T2006/401.bzip2/401.bzip2.test 7.00 55.00 685.7%
test-suite :: External/Nurbs/nurbs.test 5.00 26.00 420.0%
test-suite...s-C/unix-smail/unix-smail.test 1.00 3.00 200.0%
test-suite.../Prolangs-C++/ocean/ocean.test 1.00 3.00 200.0%
test-suite...tions/lambda-0.1.3/lambda.test 1.00 3.00 200.0%
test-suite...yApps-C++/PENNANT/PENNANT.test 2.00 5.00 150.0%
test-suite...marks/Ptrdist/yacr2/yacr2.test 1.00 2.00 100.0%
test-suite...lications/viterbi/viterbi.test 1.00 2.00 100.0%
test-suite...plications/d/make_dparser.test 12.00 24.00 100.0%
test-suite...CFP2006/433.milc/433.milc.test 14.00 27.00 92.9%
test-suite.../Applications/lemon/lemon.test 7.00 12.00 71.4%
test-suite...ce/Applications/Burg/burg.test 6.00 10.00 66.7%
test-suite...T2006/473.astar/473.astar.test 16.00 26.00 62.5%
test-suite...marks/7zip/7zip-benchmark.test 78.00 121.00 55.1%
```
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D93764
The pass has dependency on 'TargetTransformInfoWrapperPass', but the
corresponding call to INITIALIZE_PASS_DEPENDENCY was missing.
Differential Revision: https://reviews.llvm.org/D94916
Just like llvm.assume, there are a lot of cases where we can just ignore llvm.experimental.noalias.scope.decl.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D93042
D84108 exposed a bad interaction between inlining and loop-rotation
during regular LTO, which is causing notable regressions in at least
CINT2006/473.astar.
The problem boils down to: we now rotate a loop just before the vectorizer
which requires duplicating a function call in the preheader when compiling
the individual files ('prepare for LTO'). But this then prevents further
inlining of the function during LTO.
This patch tries to resolve this issue by making LoopRotate more
conservative with respect to rotating loops that have inline-able calls
during the 'prepare for LTO' stage.
I think this change intuitively improves the current situation in
general. Loop-rotate tries hard to avoid creating headers that are 'too
big'. At the moment, it assumes all inlining already happened and the
cost of duplicating a call is equal to just doing the call. But with LTO,
inlining also happens during full LTO and it is possible that a previously
duplicated call is actually a huge function which gets inlined
during LTO.
From the perspective of LV, not much should change overall. Most loops
calling user-provided functions won't get vectorized to start with
(unless we can infer that the function does not touch memory, has no
other side effects). If we do not inline the 'inline-able' call during
the LTO stage, we merely delayed loop-rotation & vectorization. If we
inline during LTO, chances should be very high that the inlined code is
itself vectorizable or the user call was not vectorizable to start with.
There could of course be scenarios where we inline a sufficiently large
function with code not profitable to vectorize, which would have be
vectorized earlier (by scalarzing the call). But even in that case,
there probably is no big performance impact, because it should be mostly
down to the cost-model to reject vectorization in that case. And then
the version with scalarized calls should also not be beneficial. In a way,
LV should have strictly more information after inlining and make more
accurate decisions (barring cost-model issues).
There is of course plenty of room for things to go wrong unexpectedly,
so we need to keep a close look at actual performance and address any
follow-up issues.
I took a look at the impact on statistics for
MultiSource/SPEC2000/SPEC2006. There are a few benchmarks with fewer
loops rotated, but no change to the number of loops vectorized.
Reviewed By: sanwou01
Differential Revision: https://reviews.llvm.org/D94232
This is not nice, but it's the best transient solution possible,
and is better than just duplicating the whole function.
The problem is, this function is widely used,
and it is not at all obvious that all the users
could be painlessly switched to operate on DomTreeUpdater,
and somehow i don't feel like porting all those users first.
This function is one of last three that not operate on DomTreeUpdater.
Summary:
Set the default for the option enabling memory ssa use in the loop sink
pass to true for the new pass manager.
Author: Jamie Schmeiser <schmeise@ca.ibm.com>
Reviewed By: asbirlea (Alina Sbirlea)
Differential Revision: https://reviews.llvm.org/D92486
In places where we calculate costs using TTI.getXXXCost() interfaces
I have changed the code to use InstructionCost instead of unsigned.
The change is non functional since InstructionCost behaves in the
same way as an integer for valid costs. Currently the getXXXCost()
functions used in this file do not return invalid costs.
See this patch for the introduction of the type: https://reviews.llvm.org/D91174
See this thread for context: http://lists.llvm.org/pipermail/llvm-dev/2020-November/146408.html
Differential revision: https://reviews.llvm.org/D94484
Added a utility function in Value class to print block name and use
block labels for unnamed blocks.
Changed LICM to call this function in its debug output.
Patch by Xiaoqing Wu <xiaoqing_wu@apple.com>
Differential Revision: https://reviews.llvm.org/D93577
This boils down to how we deal with early-increment iterator
over function's basic blocks: not only we need to early-increment,
after that we also need to skip all the blocks
that are scheduled for removal, as per DomTreeUpdater.
Thus supporting lazy DomTreeUpdater mode,
where the domtree updates (and thus block removals)
aren't applied immediately, but are delayed
until last possible moment.
This patch fixes a bug that could result in miscompiles (at least
in an OOT target). The problem could be seen by adding checks that
the DominatorTree used in BasicAliasAnalysis and ValueTracking was
valid (e.g. by adding DT->verify() call before every DT dereference
and then running all tests in test/CodeGen).
Problem was that the LegacyPassManager calculated "last user"
incorrectly for passes such as the DominatorTree when not telling
the pass manager that there was a transitive dependency between
the different analyses. And then it could happen that an incorrect
dominator tree was used when doing alias analysis (which was a pretty
serious bug as the alias analysis result could be invalid).
Fixes: https://bugs.llvm.org/show_bug.cgi?id=48709
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D94138
This is a resubmit of dd6bb367 (which was reverted due to stage2 build failures in 7c63aac), with the additional restriction added to the transform to only consider outer most loops.
As shown in the added test case, ensuring LCSSA is up to date when deleting an inner loop is tricky as we may actually need to remove blocks from any outer loops, thus changing the exit block set. For the moment, just avoid transforming this case. I plan to return to this case in a follow up patch and see if we can do better.
Original commit message follows...
The basic idea is that if SCEV can prove the backedge isn't taken, we can go ahead and get rid of the backedge (and thus the loop) while leaving the rest of the control in place. This nicely handles cases with dispatch between multiple exits and internal side effects.
Differential Revision: https://reviews.llvm.org/D93906
Add support for mixed pre/post CFG views.
Update usages of the MemorySSAUpdater to use the new DT API by
requesting the DT updates to be done by the MSSAUpdater.
Differential Revision: https://reviews.llvm.org/D93371
Currently, LoopDeletion does skip loops that have sub-loops, but this
means we currently fail to remove some no-op loops.
One example are inner loops with live-out values. Those cannot be
removed by itself. But the containing loop may itself be a no-op and the
whole loop-nest can be deleted.
The legality checks do not seem to rely on analyzing inner-loops only
for correctness.
With LoopDeletion being a LoopPass, the change means that we now
unfortunately need to do some extra work in parent loops, by checking
some conditions we already checked. But there appears to be no
noticeable compile time impact:
http://llvm-compile-time-tracker.com/compare.php?from=02d11f3cda2ab5b8bf4fc02639fd1f4b8c45963e&to=843201e9cf3b6871e18c52aede5897a22994c36c&stat=instructions
This changes patch leads to ~10 more loops being deleted on
MultiSource, SPEC2000, SPEC2006 with -O3 & LTO
This patch is also required (together with a few others) to eliminate a
no-op loop in omnetpp as discussed on llvm-dev 'LoopDeletion / removal of
empty loops.' (http://lists.llvm.org/pipermail/llvm-dev/2020-December/147462.html)
This change becomes relevant after removing potentially infinite loops
is made possible in 'must-progress' loops (D86844).
Note that I added a function call with side-effects to an outer loop in
`llvm/test/Transforms/LoopDeletion/update-scev.ll` to preserve the
original spirit of the test.
Reviewed By: reames
Differential Revision: https://reviews.llvm.org/D93716
From C11 and C++11 onwards, a forward-progress requirement has been
introduced for both languages. In the case of C, loops with non-constant
conditionals that do not have any observable side-effects (as defined by
6.8.5p6) can be assumed by the implementation to terminate, and in the
case of C++, this assumption extends to all functions. The clang
frontend will emit the `mustprogress` function attribute for C++
functions (D86233, D85393, D86841) and emit the loop metadata
`llvm.loop.mustprogress` for every loop in C11 or later that has a
non-constant conditional.
This patch modifies LoopDeletion so that only loops with
the `llvm.loop.mustprogress` metadata or loops contained in functions
that are required to make progress (`mustprogress` or `willreturn`) are
checked for observable side-effects. If these loops do not have an
observable side-effect, then we delete them.
Loops without observable side-effects that do not satisfy the above
conditions will not be deleted.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D86844
Loop strength reduction tries to recover debug variable values by looking
for simple offsets from PHI values. In really extreme conditions there may
be an offset used that won't fit in an int64_t, hitting an APInt assertion.
This patch adds a regression test and adjusts the equivalent value
collecting code to filter out any values where the offset can't be
represented by an int64_t. This means that for very large integers with
very large offsets, the variable location will become undef, which is the
same behaviour as before 2a6782bb9f / D87494.
Differential Revision: https://reviews.llvm.org/D94016
Notably, this doesn't switch *every* case, remaining cases
don't actually pass sanity checks in non-permissve mode,
and therefore require further analysis.
Note that SimplifyCFG still defaults to not preserving DomTree by default,
so this is effectively a NFC change.
This reverts commit dd6bb367d1.
Multi-stage builders are showing an assertion failure w/LCSSA not being preserved on entry to IndVars. Reason isn't clear, reverting while investigating.
The basic idea is that if SCEV can prove the backedge isn't taken, we can go ahead and get rid of the backedge (and thus the loop) while leaving the rest of the control in place. This nicely handles cases with dispatch between multiple exits and internal side effects.
Differential Revision: https://reviews.llvm.org/D93906
This patch makes Scalarizer to use poison as insertelement's placeholder.
It contains two changes in Scalarizer.cpp, and the both changes does not change the semantics of the optimized program.
It is because the placeholder value (poison) is already completely hidden by following insertelement instructions.
The first change at visitBitCastInst() creates poison vector of MidTy and consecutively inserts FanIn times,
which is # of elems of MidTy.
The second change at ScalarizerVisitor::finish() creates poison with Op->getType(), and it is filled with
Count insertelements.
The test diffs show that the poison value is never exposed after insertelements.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D93989
Test clang/test/Misc/loop-opt-setup.c fails when executed in Release.
This reverts commit 6f1503d598.
Reviewed By: SureYeaah
Differential Revision: https://reviews.llvm.org/D93956
From C11 and C++11 onwards, a forward-progress requirement has been
introduced for both languages. In the case of C, loops with non-constant
conditionals that do not have any observable side-effects (as defined by
6.8.5p6) can be assumed by the implementation to terminate, and in the
case of C++, this assumption extends to all functions. The clang
frontend will emit the `mustprogress` function attribute for C++
functions (D86233, D85393, D86841) and emit the loop metadata
`llvm.loop.mustprogress` for every loop in C11 or later that has a
non-constant conditional.
This patch modifies LoopDeletion so that only loops with
the `llvm.loop.mustprogress` metadata or loops contained in functions
that are required to make progress (`mustprogress` or `willreturn`) are
checked for observable side-effects. If these loops do not have an
observable side-effect, then we delete them.
Loops without observable side-effects that do not satisfy the above
conditions will not be deleted.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D86844
As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used
instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`.
Let's update them.
Actually, it would have been more natural if the patches were made in this order:
(1) let them use unary CreateShuffleVector first
(2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793)
The order is swapped, but in terms of correctness it is still fine.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D93923
This patch adds support for select form of and/or.
Currently there is an ongoing effort for moving towards using `select a, b, false` instead of `and i1 a, b` and
`select a, true, b` instead of `or i1 a, b` as well.
D93065 has links to relevant changes.
Alive2 proof: (undef input was disabled due to timeout :( )
- and: https://alive2.llvm.org/ce/z/AgvFbQ
- or: https://alive2.llvm.org/ce/z/KjLJyb
Differential Revision: https://reviews.llvm.org/D93935
The function FoldSingleEntryPHINodes() is changed to return if
it has changed IR or not. This return value is used by RS4GC to
set the MadeChange flag respectively.
Reviewed By: reames
Differential Revision: https://reviews.llvm.org/D93810
This patch updates GVN to correctly return the modified status, if PRE
is performed on indices. It fixes a crash when building the test-suite
with EXPENSIVE_CHECKS and LTO.
EarlyCSE's handleBranchCondition says:
```
// If the condition is AND operation, we can propagate its operands into the
// true branch. If it is OR operation, we can propagate them into the false
// branch.
```
This holds for the corresponding select patterns as well.
This is a part of an ongoing work for disabling buggy select->and/or transformations.
See llvm.org/pr48353 and D93065 for more context
Proof:
and: https://alive2.llvm.org/ce/z/MQWodU
or: https://alive2.llvm.org/ce/z/9GLbB_
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D93842
This patch makes GVN recognize `select c1, c2, false` as well as `select c1, true, c2`
branch condition and propagate equality from these.
See llvm.org/pr48353, D93065
Differential Revision: https://reviews.llvm.org/D93841
While `%x.curr` is always safe to compute, because `LoopBackedgeTakenCount`
will always be smaller than `bitwidth(X)`, i.e. we never get poison,
rewriting `%x.next` is more complicated, however, because `X << LoopTripCount`
will be poison iff `LoopTripCount == bitwidth(X)` (which will happen
iff `BitPos` is `bitwidth(x) - 1` and `X` is `1`).
So unless we know that isn't the case (as alive2 notes, we know it's safe
to do iff shift had no-wrap flags, or bitpos does not indicate signbit,
or we know that %x is never `1`), we'll need to emit an alternative,
safe IR, by either just shifting the `%x.curr`, or conditionally selecting
between the computed `%x.next` and `0`..
Former IR looks better so let's do that.
While there, ensure that we don't drop no-wrap flags from said shift.
The current state of the transform is still not enough to support
my motivational pattern, because it has one more "induction variable".
I have delayed posting this patch, because originally even just rewriting
the loop as countable wasn't enough to nicely transform my motivational pattern,
because i expected that extra IV to be rewritten afterwards,
but it wasn't happening until i fixed that in D91800.
So, this patch allows the 'left-shift until bittest' loop idiom
as long as the inserted ops are cheap,
and lifts any and all extra use checks on the instructions.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D92754
If the bitmask is for sign bit, instcombine would have canonicalized
the pattern into a proper sign bit check. Supporting that is still
simple, but requires a bit of a roundtrip - we first have to use
`decomposeBitTestICmp()`, and the rest again just works.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D91726
The handing of the case where the mask is a constant is trivial,
if said constant is a power of two, the bit in question is log2(mask),
rest just works.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D91725
The motivation here is the following inner loop in fp16/fp24 -> fp32 expander,
that runs as part of the floating-point DNG decompression in RawSpeed library:
cd380bb9a2/src/librawspeed/decompressors/DeflateDecompressor.cpp (L112-L115)
```
while (!(fp32_fraction & (1 << 23))) {
fp32_exponent -= 1;
fp32_fraction <<= 1;
}
```
(https://godbolt.org/z/r13YMh)
As one might notice, that loop is currently uncountable, and that whole code stays scalar.
Yet, it is rather trivial to make that loop countable:
https://godbolt.org/z/do8WMz
and we can prove that via alive2:
https://alive2.llvm.org/ce/z/7vQnji (ha nice, isn't it?)
... and that allow for the whole fp16->fp32 code to vectorize:
https://godbolt.org/z/7hYr13
Now, while i'd love to get there, i feel like i should take it in steps.
For now, this introduces support for the most basic case,
where the bit position is known as a variable,
and the loop *will* go away (has no live-outs other than the recurrence,
no extra instructions in the loop).
I have added sufficient (i believe) test coverage,
and alive2 is happy with those transforms.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D91038
Current approach doesn't work well in cases when multiple paths are predicted to be "cold". By "cold" paths I mean those containing "unreachable" instruction, call marked with 'cold' attribute and 'unwind' handler of 'invoke' instruction. The issue is that heuristics are applied one by one until the first match and essentially ignores relative hotness/coldness
of other paths.
New approach unifies processing of "cold" paths by assigning predefined absolute weight to each block estimated to be "cold". Then we propagate these weights up/down IR similarly to existing approach. And finally set up edge probabilities based on estimated block weights.
One important difference is how we propagate weight up. Existing approach propagates the same weight to all blocks that are post-dominated by a block with some "known" weight. This is useless at least because it always gives 50\50 distribution which is assumed by default anyway. Worse, it causes the algorithm to skip further heuristics and can miss setting more accurate probability. New algorithm propagates the weight up only to the blocks that dominates and post-dominated by a block with some "known" weight. In other words, those blocks that are either always executed or not executed together.
In addition new approach processes loops in an uniform way as well. Essentially loop exit edges are estimated as "cold" paths relative to back edges and should be considered uniformly with other coldness/hotness markers.
Reviewed By: yrouban
Differential Revision: https://reviews.llvm.org/D79485
This is a follow-up patch of D87045.
The patch implements "loop-nest mode" for `LPMUpdater` and `FunctionToLoopPassAdaptor` in which only top-level loops are operated.
`createFunctionToLoopPassAdaptor` decides whether the returned adaptor is in loop-nest mode or not based on the given pass. If the pass is a loop-nest pass or the pass is a `LoopPassManager` which contains only loop-nest passes, the loop-nest version of adaptor is returned; otherwise, the normal (loop) version of adaptor is returned.
Reviewed By: Whitney
Differential Revision: https://reviews.llvm.org/D87531
MSSA DSE starts at a killing store, finds an earlier store and
then checks that the earlier store is not read along any paths
(without being killed first). However, it uses the memory location
of the killing store for that, not the earlier store that we're
attempting to eliminate.
This has a number of problems:
* Mismatches between what BasicAA considers aliasing and what DSE
considers an overwrite (even though both are correct in isolation)
can result in miscompiles. This is PR48279, which D92045 tries to
fix in a different way. The problem is that we're using a location
from a store that is potentially not executed and thus may be UB,
in which case analysis results can be arbitrary.
* Metadata on the killing store may be used to determine aliasing,
but there is no guarantee that the metadata is valid, as the specific
killing store may not be executed. Using the metadata on the earlier
store is valid (it is the store we're removing, so on any execution
where its removal may be observed, it must be executed).
* The location is imprecise. For full overwrites the killing store
will always have a location that is larger or equal than the earlier
access location, so it's beneficial to use the earlier access
location. This is not the case for partial overwrites, in which
case either location might be smaller. There is some room for
improvement here.
Using the earlier access location means that we can no longer cache
which accesses are read for a given killing store, as we may be
querying different locations. However, it turns out that simply
dropping the cache has no notable impact on compile-time.
Differential Revision: https://reviews.llvm.org/D93523
Philip Reames