In some cases the order that we hoist instructions in means that when rehoisting
(which uses the same order as hoisting) we can rehoist to a block A, then a
block B, then block A again. This currently causes an assertion failure as it
expects that when changing the hoist point it only ever moves to a block that
dominates the hoist point being moved from.
Fix this by moving the re-hoist point when it doesn't dominate the dominator of
hoisted instruction, or in other words when it wouldn't dominate the uses of
the instruction being rehoisted.
Differential Revision: https://reviews.llvm.org/D55266
llvm-svn: 350408
OptimizeAutoreleaseRVCall skips optimizing llvm.objc.autoreleaseReturnValue if it
sees a user which is llvm.objc.retainAutoreleasedReturnValue, and if they have
equivalent arguments (either identical or equivalent PHIs). It then assumes that
ObjCARCOpt::OptimizeRetainRVCall will optimize the pair instead.
Trouble is, ObjCARCOpt::OptimizeRetainRVCall doesn't know about equivalent PHIs
so optimizes in a different way and we are left with an unoptimized llvm.objc.autoreleaseReturnValue.
This teaches ObjCARCOpt::OptimizeRetainRVCall to also understand PHI equivalence.
rdar://problem/47005143
Reviewed By: ahatanak
Differential Revision: https://reviews.llvm.org/D56235
llvm-svn: 350284
The caller to EraseInstruction had this conditional:
// ARC calls with null are no-ops. Delete them.
if (IsNullOrUndef(Arg))
but the assert inside EraseInstruction only allowed ConstantPointerNull and not
undef or bitcasts.
This adds support for both of these cases.
rdar://problem/47003805
llvm-svn: 350261
If an instruction has no demanded bits, remove it directly during BDCE,
instead of leaving it for something else to clean up.
Differential Revision: https://reviews.llvm.org/D56185
llvm-svn: 350257
The final piece of IR-level analysis to allow this was committed with:
rL350188
Using the intrinsics should improve transforms based on cost models
like vectorization and inlining.
The backend should be prepared too, so we can now canonicalize more
sequences of shift/logic to the intrinsics and know that the end
result should be equal or better to the original code even if the
target does not have an actual rotate instruction.
llvm-svn: 350199
To make it more obvious which part of the transformation is carried
out by BDCE. Also drop the CHECK-IO lines which only run -instsimplify
as they don't really seem meaningful if the main check doesn't run
-instsimplify either.
llvm-svn: 350189
This (mostly) fixes https://bugs.llvm.org/show_bug.cgi?id=39771.
BDCE currently detects instructions that don't have any demanded bits
and replaces their uses with zero. However, if an instruction has
multiple uses, then some of the uses may be dead (have no demanded bits)
even though the instruction itself is still live. This patch extends
DemandedBits/BDCE to detect such uses and replace them with zero.
While this will not immediately render any instructions dead, it may
lead to simplifications (in the motivating case, by converting a rotate
into a simple shift), break dependencies, etc.
The implementation tries to strike a balance between analysis power and
complexity/memory usage. Originally I wanted to track demanded bits on
a per-use level, but ultimately we're only really interested in whether
a use is entirely dead or not. I'm using an extra set to track which uses
are dead. However, as initially all uses are dead, I'm not storing uses
those user is also dead. This case is checked separately instead.
The previous attempt to land this lead to miscompiles, because cases
where uses were initially dead but were later found to be live during
further analysis were not always correctly removed from the DeadUses
set. This is fixed now and the added test case demanstrates such an
instance.
Differential Revision: https://reviews.llvm.org/D55563
llvm-svn: 350188
Summary:
Existing LIR recognizes CTLZ where shifting input variable right until it is zero. (Shift-Until-Zero idiom)
This commit:
1. Augments Shift-Until-Zero idiom to recognize CTTZ where input variable is shifted left.
2. Prepare for BitScan idiom recognition.
Patch by Yuanfang Chen (tabloid.adroit)
Reviewers: craig.topper, evstupac
Reviewed By: craig.topper
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D55876
llvm-svn: 350074
This patch teaches LoopSimplifyCFG to remove dead exiting edges
from loops.
Differential Revision: https://reviews.llvm.org/D54025
Reviewed By: fedor.sergeev
llvm-svn: 350049
Summary:
BasicAA has special logic for unescaped allocas, which normally applies
equally well to dynamic and static allocas. However, llvm.stackrestore
has the power to end the lifetime of dynamic allocas, without referring
to them directly.
stackrestore is already marked with the most conservative memory
modification attributes, but because the alloca is not escaped, the
normal logic produces incorrect results. I think BasicAA needs a special
case here to teach it about the relationship between dynamic allocas and
stackrestore.
Fixes PR40118
Reviewers: gbiv, efriedma, george.burgess.iv
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D55969
llvm-svn: 349945
If we found unsafe dependences other than 'unknown', we already know at
compile time that they are unsafe and the runtime checks should always
fail. So we can avoid generating them in those cases.
This should have no negative impact on performance as the runtime checks
that would be created previously should always fail. As a sanity check,
I measured the test-suite, spec2k and spec2k6 and there were no regressions.
Reviewers: Ayal, anemet, hsaito
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D55798
llvm-svn: 349794
As discussed on D55894, this replaces the existing PADDS/PSUBUS intrinsics with the the sadd/ssub.sat generic intrinsics and moves the tests out of the x86 subfolder.
PR40110 has been raised to fix the regression with constant folding vectors containing undef elements.
llvm-svn: 349759
Summary:
This allows expanding {7,11,13,14,15,21,22,23,25,26,27,28,29,30,31}-byte memcmp
in just two loads on X86. These were previously calling memcmp.
Reviewers: spatel, gchatelet
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D55263
llvm-svn: 349731
The current llvm.mem.parallel_loop_access metadata has a problem in that
it uses LoopIDs. LoopID unfortunately is not loop identifier. It is
neither unique (there's even a regression test assigning the some LoopID
to multiple loops; can otherwise happen if passes such as LoopVersioning
make copies of entire loops) nor persistent (every time a property is
removed/added from a LoopID's MDNode, it will also receive a new LoopID;
this happens e.g. when calling Loop::setLoopAlreadyUnrolled()).
Since most loop transformation passes change the loop attributes (even
if it just to mark that a loop should not be processed again as
llvm.loop.isvectorized does, for the versioned and unversioned loop),
the parallel access information is lost for any subsequent pass.
This patch unlinks LoopIDs and parallel accesses.
llvm.mem.parallel_loop_access metadata on instruction is replaced by
llvm.access.group metadata. llvm.access.group points to a distinct
MDNode with no operands (avoiding the problem to ever need to add/remove
operands), called "access group". Alternatively, it can point to a list
of access groups. The LoopID then has an attribute
llvm.loop.parallel_accesses with all the access groups that are parallel
(no dependencies carries by this loop).
This intentionally avoid any kind of "ID". Loops that are clones/have
their attributes modifies retain the llvm.loop.parallel_accesses
attribute. Access instructions that a cloned point to the same access
group. It is not necessary for each access to have it's own "ID" MDNode,
but those memory access instructions with the same behavior can be
grouped together.
The behavior of llvm.mem.parallel_loop_access is not changed by this
patch, but should be considered deprecated.
Differential Revision: https://reviews.llvm.org/D52116
llvm-svn: 349725
Creating the IR builder, then modifying the CFG, leads to an IRBuilder
where the BB and insertion point are inconsistent, so new instructions
have the wrong parent.
Modified an existing test because the test wasn't covering anything
useful (the "invoke" was not actually an invoke by the time we hit the
code in question).
Differential Revision: https://reviews.llvm.org/D55729
llvm-svn: 349693
This (mostly) fixes https://bugs.llvm.org/show_bug.cgi?id=39771.
BDCE currently detects instructions that don't have any demanded bits
and replaces their uses with zero. However, if an instruction has
multiple uses, then some of the uses may be dead (have no demanded bits)
even though the instruction itself is still live. This patch extends
DemandedBits/BDCE to detect such uses and replace them with zero.
While this will not immediately render any instructions dead, it may
lead to simplifications (in the motivating case, by converting a rotate
into a simple shift), break dependencies, etc.
The implementation tries to strike a balance between analysis power and
complexity/memory usage. Originally I wanted to track demanded bits on
a per-use level, but ultimately we're only really interested in whether
a use is entirely dead or not. I'm using an extra set to track which uses
are dead. However, as initially all uses are dead, I'm not storing uses
those user is also dead. This case is checked separately instead.
The test case has a couple of cases that are not simplified yet. In
particular, we're only looking at uses of instructions right now. I think
it would make sense to also extend this to arguments. Furthermore
DemandedBits doesn't yet know some of the tricks that InstCombine does
for the demanded bits or bitwise or/and/xor in combination with known
bits information.
Differential Revision: https://reviews.llvm.org/D55563
llvm-svn: 349674
There's a mismatch internally about how we are handling these patterns.
We count loads as cheapToScalarize(), but then we don't actually
scalarize them, so that can leave extra instructions compared to where
we started when scalarizing other ops. If it's cheapToScalarize, then
we should be scalarizing.
llvm-svn: 349560
Clang uses weak linkage for objc runtime functions when they are not available on the platform.
The intrinsic has this linkage so we just need to pass that on to the runtime call.
llvm-svn: 349559
For performance reasons, clang set nonlazybind on these functions. Now that we
are using intrinsics instead of runtime calls, we should set this attribute when
creating the runtime functions.
llvm-svn: 349558
This patch adds a VectorizationSafetyStatus enum, which will be extended
in a follow up patch to distinguish between 'safe with runtime checks'
and 'known unsafe' dependences.
Reviewers: anemet, anna, Ayal, hsaito
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D54892
llvm-svn: 349556
The first test claims to show that the vectorizer will
generate a vector load/loop, but then this file runs
other passes which might scalarize that op. I'm removing
instcombine from the RUN line here to break that dependency.
Also, I'm generating full checks to make it clear exactly
what the vectorizer has done.
llvm-svn: 349554
SelectionDAG currently changes these intrinsics to function calls, but that won't work
for other ISel's. Also we want to eventually support nonlazybind and weak linkage coming
from the front-end which we can't do in SelectionDAG.
llvm-svn: 349552
We're moving ARC optimisation and ARC emission in clang away from runtime methods
and towards intrinsics. This is the part which actually uses the intrinsics in the ARC
optimizer when both analyzing the existing calls and emitting new ones.
Differential Revision: https://reviews.llvm.org/D55348
Reviewers: ahatanak
llvm-svn: 349534
Checking whether a number has a certain number of trailing / leading
zeros means checking whether it is of the form XXXX1000 / 0001XXXX,
which can be done with an and+icmp.
Related to https://bugs.llvm.org/show_bug.cgi?id=28668. As a next
step, this can be extended to non-equality predicates.
Differential Revision: https://reviews.llvm.org/D55745
llvm-svn: 349530
When using clang with `-fno-unroll-loops` (implicitly added with `-O1`),
the LoopUnrollPass is not not added to the (legacy) pass pipeline. This
also means that it will not process any loop metadata such as
llvm.loop.unroll.enable (which is generated by #pragma unroll or
WarnMissedTransformationsPass emits a warning that a forced
transformation has not been applied (see
https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20181210/610833.html).
Such explicit transformations should take precedence over disabling
heuristics.
This patch unconditionally adds LoopUnrollPass to the optimizing
pipeline (that is, it is still not added with `-O0`), but passes a flag
indicating whether automatic unrolling is dis-/enabled. This is the same
approach as LoopVectorize uses.
The new pass manager's pipeline builder has no option to disable
unrolling, hence the problem does not apply.
Differential Revision: https://reviews.llvm.org/D55716
llvm-svn: 349509
This modifies the IPO pass so that it respects any explicit function
address space specified in the data layout.
In targets with nonzero program address spaces, all functions should, by
default, be placed into the default program address space.
This is required for Harvard architectures like AVR. Without this, the
functions will be marked as residing in data space, and thus not be
callable.
This has no effect to any in-tree official backends, as none use an
explicit program address space in their data layouts.
Patch by Tim Neumann.
llvm-svn: 349469
When splitting up an alloca's uses we were dropping any explicit
alignment tags, which means they default to the ABI-required default
alignment and this can cause miscompiles if the real value was smaller.
Also refactor the TBAA metadata into a parent class since it's shared by
both children anyway.
llvm-svn: 349465
Now, that we have funnel shift intrinsics, it should be safe to convert this form of rotate to it.
In the worst case (a target that doesn't have rotate instructions), we will expand this into a
branch-less sequence of ALU ops (neg/and/and/lshr/shl/or) in the backend, so it's still very
likely to be a perf improvement over the original code.
The motivating source code pattern for this is shown in:
https://bugs.llvm.org/show_bug.cgi?id=34924
Background:
I looked at several different options before deciding where to try this - instcombine, simplifycfg,
CGP - because it doesn't fit cleanly anywhere AFAIK.
The backend (CGP, SDAG, GlobalIsel?) is too late for what we're trying to accomplish. We want to
have the IR converted before we reach things like vectorization because the reduced code can make a
loop much simpler to transform.
Technically, this could be included in instcombine, but it's a large pattern match that includes
control-flow, so it just felt wrong to stuff into there (although I have a draft of that patch).
Similarly, this could be part of simplifycfg, but all of this pattern matching is a stretch.
So we're left with our relatively new dumping ground for homeless transforms: aggressive-instcombine.
This only runs at -O3, but that seems like a reasonable limitation given that source code has many
options to avoid this pattern (including the recently added clang intrinsics for rotates).
I'm including a PhaseOrdering test because we require the teamwork of 3 passes (aggressive-instcombine,
instcombine, simplifycfg) to get this into the minimal IR form that we want. That test shows a bug
with the new pass manager that's independent of this change (but it will be masked if we canonicalize
harder to funnel shift intrinsics in instcombine).
Differential Revision: https://reviews.llvm.org/D55604
llvm-svn: 349396
The problem is shown specifically for a case with vector multiply here:
https://bugs.llvm.org/show_bug.cgi?id=40032
...and this might mask the original backend bug for ARM shown in:
https://bugs.llvm.org/show_bug.cgi?id=39967
As the test diffs here show, we were (and probably still aren't) doing
these kinds of transforms in a principled way. We are producing more or
equal wide instructions than we started with in some cases, so we still
need to restrict/correct other transforms from overstepping.
If there are perf regressions from this change, we can either carve out
exceptions to the general IR rules, or improve the backend to do these
transforms when we know the transform is profitable. That's probably
similar to a change like D55448.
Differential Revision: https://reviews.llvm.org/D55744
llvm-svn: 349389
If a saturating add/sub has one constant operand, then we can
determine the possible range of outputs it can produce, and simplify
an icmp comparison based on that.
The implementation is based on a similar existing mechanism for
simplifying binary operator + icmps.
Differential Revision: https://reviews.llvm.org/D55735
llvm-svn: 349369
If a saturating add/sub with a constant operand is compared to
another constant, we should be able to determine that the condition
is always true/false in some cases (but currently don't).
llvm-svn: 349261
The current code relies on LeaderUseCount to determine if we can remove
an SSA copy, but in that the LeaderUseCount does not refer to the SSA
copy. If a SSA copy is a dominating leader, we use the operand as dominating
leader instead. This means we removed a user of a ssa copy and we should
decrement its use count, so we can remove the ssa copy once it becomes dead.
Fixes PR38804.
Reviewers: efriedma, davide
Reviewed By: davide
Differential Revision: https://reviews.llvm.org/D51595
llvm-svn: 349217
When converting dbg.declares, if the described value is a [s|z]ext,
refer to the ext directly instead of referring to its operand.
This fixes a narrowing bug (the debugger got the sign of a variable
wrong, see llvm.org/PR35400).
The main reason to refer to the ext's operand was that an optimization
may remove the ext itself, leading to a dropped variable. Now that
InstCombine has been taught to use replaceAllDbgUsesWith (r336451), this
is less of a concern. Other passes can/should adopt this API as needed
to fix dropped variable bugs.
Differential Revision: https://reviews.llvm.org/D51813
llvm-svn: 349214
Optimization transformations are intentionally disabled by the 'optnone'
function attribute. Therefore do not warn if transformation metadata is
still present.
Using the legacy pass manager structure, the `skipFunction` method takes
care for the optnone attribute (already called before this patch). For
the new pass manager, there is no equivalent, so we check for the
'optnone' attribute manually.
Differential Revision: https://reviews.llvm.org/D55690
llvm-svn: 349184
The `changeToCall` function did not preserve the invoke's metadata.
Currently, there is probably no metadata that depends on being applied
on a CallInst or InvokeInst. Therefore we can replace the instruction's
metadata.
This fixes http://llvm.org/PR39994
Suggested-by: Moritz Kreutzer <moritz.kreutzer@siemens.com>
Differential Revision: https://reviews.llvm.org/D55666
llvm-svn: 349170
ProfileSampleAccurate is used to indicate the profile has exact match to the
code to be optimized.
Previously ProfileSampleAccurate is handled in ProfileSummaryInfo::isColdCallSite
and ProfileSummaryInfo::isColdBlock. A better solution is to initialize function
entry count to 0 when ProfileSampleAccurate is true, so we don't have to handle
ProfileSampleAccurate in multiple places.
Differential Revision: https://reviews.llvm.org/D55660
llvm-svn: 349088
Currently memcpyopt optimizes cases like
memset(a, byte, N);
memcpy(b, a, M);
to
memset(a, byte, N);
memset(b, byte, M);
if M <= N. Often this allows further simplifications down the line,
which drop the first memset entirely.
This patch extends this optimization for the case where M > N, but we
know that the bytes a[N..M] are undef due to alloca/lifetime.start.
This situation arises relatively often for Rust code, because Rust does
not initialize trailing structure padding and loves to insert redundant
memcpys. This also fixes https://bugs.llvm.org/show_bug.cgi?id=39844.
The previous version of this patch did not perform dependency checking
properly: While the dependency is checked at the position of the memset,
the used size must be that of the memcpy. Previously the size of the
memset was used, which missed modification in the region
MemSetSize..CopySize, resulting in miscompiles. The added tests cover
variations of this issue.
Differential Revision: https://reviews.llvm.org/D55120
llvm-svn: 349078
The actual type of the first argument of the @dbg intrinsic
doesn't really matter as we're setting it to `undef`, but the
bitcode reader is picky about `void` types.
llvm-svn: 349069
For anyone curious, the first test example is illustrative of a real code idiom produced by branching on the result of a three way comparison.
llvm-svn: 348997
When loops are deleted, we don't keep track of variables modified inside
the loops, so the DI will contain the wrong value for these.
e.g.
int b() {
int i;
for (i = 0; i < 2; i++)
;
patatino();
return a;
-> 6 patatino();
7 return a;
8 }
9 int main() { b(); }
(lldb) frame var i
(int) i = 0
We mark instead these values as unavailable inserting a
@llvm.dbg.value(undef to make sure we don't end up printing an incorrect
value in the debugger. We could consider doing something fancier,
for, e.g. constants, in the future.
PR39868.
rdar://problem/46418795)
Differential Revision: https://reviews.llvm.org/D55299
llvm-svn: 348988
This fixes https://bugs.llvm.org/show_bug.cgi?id=39908.
The evaluateGEPOffsetExpression() function simplifies GEP offsets for
use in comparisons against zero, basically by converting X*Scale+Offset==0
to X+Offset/Scale==0 if Scale divides Offset. However, before this is done,
Offset is masked down to the pointer size. This results in incorrect
results for negative Offsets, because we basically end up dividing the
32-bit offset *zero* extended to 64-bit bits (rather than sign extended).
Fix this by explicitly sign extending the truncated value.
Differential Revision: https://reviews.llvm.org/D55449
llvm-svn: 348987
As mentioned in D55604, there are 2 bugs here:
1. The new pass manager is speculating wildly by default.
2. The old pass manager is not converting this to funnel shift.
llvm-svn: 348980
Indices for getelementptr can be signed so we should use
getMinSignedBits instead of getActiveBits here. The function later calls
getSExtValue to get the int64_t value, which also checks
getMinSignedBits.
This fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=11647.
Reviewers: mssimpso, efriedma, davide
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D55536
llvm-svn: 348957
When multiple loop transformation are defined in a loop's metadata, their order of execution is defined by the order of their respective passes in the pass pipeline. For instance, e.g.
#pragma clang loop unroll_and_jam(enable)
#pragma clang loop distribute(enable)
is the same as
#pragma clang loop distribute(enable)
#pragma clang loop unroll_and_jam(enable)
and will try to loop-distribute before Unroll-And-Jam because the LoopDistribute pass is scheduled after UnrollAndJam pass. UnrollAndJamPass only supports one inner loop, i.e. it will necessarily fail after loop distribution. It is not possible to specify another execution order. Also,t the order of passes in the pipeline is subject to change between versions of LLVM, optimization options and which pass manager is used.
This patch adds 'followup' attributes to various loop transformation passes. These attributes define which attributes the resulting loop of a transformation should have. For instance,
!0 = !{!0, !1, !2}
!1 = !{!"llvm.loop.unroll_and_jam.enable"}
!2 = !{!"llvm.loop.unroll_and_jam.followup_inner", !3}
!3 = !{!"llvm.loop.distribute.enable"}
defines a loop ID (!0) to be unrolled-and-jammed (!1) and then the attribute !3 to be added to the jammed inner loop, which contains the instruction to distribute the inner loop.
Currently, in both pass managers, pass execution is in a fixed order and UnrollAndJamPass will not execute again after LoopDistribute. We hope to fix this in the future by allowing pass managers to run passes until a fixpoint is reached, use Polly to perform these transformations, or add a loop transformation pass which takes the order issue into account.
For mandatory/forced transformations (e.g. by having been declared by #pragma omp simd), the user must be notified when a transformation could not be performed. It is not possible that the responsible pass emits such a warning because the transformation might be 'hidden' in a followup attribute when it is executed, or it is not present in the pipeline at all. For this reason, this patche introduces a WarnMissedTransformations pass, to warn about orphaned transformations.
Since this changes the user-visible diagnostic message when a transformation is applied, two test cases in the clang repository need to be updated.
To ensure that no other transformation is executed before the intended one, the attribute `llvm.loop.disable_nonforced` can be added which should disable transformation heuristics before the intended transformation is applied. E.g. it would be surprising if a loop is distributed before a #pragma unroll_and_jam is applied.
With more supported code transformations (loop fusion, interchange, stripmining, offloading, etc.), transformations can be used as building blocks for more complex transformations (e.g. stripmining+stripmining+interchange -> tiling).
Reviewed By: hfinkel, dmgreen
Differential Revision: https://reviews.llvm.org/D49281
Differential Revision: https://reviews.llvm.org/D55288
llvm-svn: 348944
For SampleFDO, when a callsite doesn't appear in the profile, it will not be marked as cold callsite unless the option -profile-sample-accurate is specified.
But profile-sample-accurate doesn't cover function isFunctionColdInCallGraph which is used to decide whether a function should be put into text.unlikely section, so even if the user knows the profile is accurate and specifies profile-sample-accurate, those functions not appearing in the sample profile are still not be put into text.unlikely section right now.
The patch fixes that.
Differential Revision: https://reviews.llvm.org/D55567
llvm-svn: 348940
Without this check, we hit an assertion in getZExtValue, if the constant
value does not fit into an uint64_t.
As getZExtValue returns an uint64_t, should we update
getAggregateElement to take an uin64_t as well?
This fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=6109.
Reviewers: efriedma, craig.topper, spatel
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D55547
llvm-svn: 348906
Summary:
Enable suspend point simplification for cases where:
* coro.save and coro.suspend are in different basic blocks
* where there are intervening intrinsics
Reviewers: modocache, tks2103, lewissbaker
Reviewed By: modocache
Subscribers: EricWF, llvm-commits
Differential Revision: https://reviews.llvm.org/D55160
llvm-svn: 348897
Struct types may have leading zero-size elements like [0 x i32], in
which case the "real" element at offset 0 will not necessarily coincide
with the 0th element of the aggregate. ConstantFoldLoadThroughBitcast()
wants to drill down the element at offset 0, but currently always picks
the 0th aggregate element to do so. This patch changes the code to find
the first non-zero-size element instead, for the struct case.
The motivation behind this change is https://github.com/rust-lang/rust/issues/48627.
Rust is fond of emitting [0 x iN] separators between struct elements to
enforce alignment, which prevents constant folding in this particular case.
The additional tests with [4294967295 x [0 x i32]] check that we don't
end up unnecessarily looping over a large number of zero-size elements
of a zero-size array.
Differential Revision: https://reviews.llvm.org/D55169
llvm-svn: 348895
It's currently not safe to outline landingpad instructions (see
llvm.org/PR39917). Like @llvm.eh.typeid.for, the order and content of
previous landingpad instructions in a function alters the lowering of
subsequent landingpads by renumbering type info ID's. Outlining a
landingpad therefore breaks exception handling & unwinding.
llvm-svn: 348870
call iM movmsk(sext <N x i1> X) --> zext (bitcast <N x i1> X to iN) to iM
This has the potential to create less-than-8-bit scalar types as shown in
some of the test diffs, but it looks like the backend knows how to deal
with that in these patterns. This is the simple part of the fix suggested in:
https://bugs.llvm.org/show_bug.cgi?id=39927
Differential Revision: https://reviews.llvm.org/D55529
llvm-svn: 348862
These involve cases where certain uses are dead by means of having
no demanded bits, even though the used instruction still has demanded
bits when other uses are taken into account. BDCE currently does not
simplify such cases.
llvm-svn: 348861
Summary:
When eliminating a dead argument or return value in a function with
local linkage, all uses, including in dbg.value intrinsics, would be
replaced with null constants. This would mean that, for example for an
integer argument, the debug info would incorrectly express that the
value is 0. Instead, replace all uses with undef to indicate that the
argument/return value is optimized out.
Also, make sure that metadata uses of return values are rewritten even
if there are no non-metadata uses of the value.
As a bit of historical curiosity, the code that emitted null constants
was introduced in the initial check-in of the pass in 2003, before
'undef' values even existed in LLVM.
This fixes PR23260.
Reviewers: dblaikie, aprantl, vsk, djtodoro
Reviewed By: aprantl
Subscribers: llvm-commits
Tags: #debug-info
Differential Revision: https://reviews.llvm.org/D55513
llvm-svn: 348837
Summary: The comment says we need 3 extracts and a select at the end. But didn't we just account for the select in the vector cost above. Aren't we just extracting the single element after taking the min/max in the vector register?
Reviewers: RKSimon, spatel, ABataev
Reviewed By: RKSimon
Subscribers: javed.absar, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D55480
llvm-svn: 348739
Currently memcpyopt optimizes cases like
memset(a, byte, N);
memcpy(b, a, M);
to
memset(a, byte, N);
memset(b, byte, M);
if M <= N. Often this allows further simplifications down the line,
which drop the first memset entirely.
This patch extends this optimization for the case where M > N, but we
know that the bytes a[N..M] are undef due to alloca/lifetime.start.
This situation arises relatively often for Rust code, because Rust does
not initialize trailing structure padding and loves to insert redundant
memcpys. This also fixes https://bugs.llvm.org/show_bug.cgi?id=39844.
For the implementation, I'm reusing a bit of code for a similar existing
optimization (direct memcpy of undef). I've also added memset support to
MemDepAnalysis GetLocation -- Instead, getPointerDependencyFrom could be
used, but it seems to make more sense to add this to GetLocation and thus
make the computation cachable.
Differential Revision: https://reviews.llvm.org/D55120
llvm-svn: 348645
The splitting pass uses its 'unlikelyExecuted' predicate to statically
decide which blocks are cold.
- Do not treat noreturn calls as if they are cold unless they are actually
marked cold. This is motivated by functions like exit() and longjmp(), which
are not beneficial to outline.
- Do not treat inline asm as an outlining barrier. In practice asm("") is
frequently used to inhibit basic block merging; enabling outlining in this case
results in substantial memory savings.
- Treat invokes of cold functions as cold.
As a drive-by, remove the 'exceptionHandlingFunctions' predicate, because it's
no longer needed. The pass can identify & outline blocks dominated by EH pads,
so there's no need to special-case __cxa_begin_catch etc.
Differential Revision: https://reviews.llvm.org/D54244
llvm-svn: 348640
Algorithm: Identify maximal cold regions and put them in a worklist. If
a candidate region overlaps with another, discard it. While the worklist
is full, remove a single-entry sub-region from the worklist and attempt
to outline it. By the non-overlap property, this should not invalidate
parts of the domtree pertaining to other outlining regions.
Testing: LNT results on X86 are clean. With test-suite + externals, llvm
outlines 134KB pre-patch, and 352KB post-patch (+ ~2.6x). The file
483.xalancbmk/src/Constants.cpp stands out as an extreme case where llvm
outlines over 100 times in some functions (mostly EH paths). There was
not a significant performance impact pre vs. post-patch.
Differential Revision: https://reviews.llvm.org/D53887
llvm-svn: 348639
We were overcounting the number of arithmetic operations needed at each level before we reach a legal type. We were using the full vector type for that level, but we are going to split the input vector at that level in half. So the effective arithmetic operation cost at that level is half the width.
So for example on 8i32 on an sse target. Were were calculating the cost of an 8i32 op which is likely 2 for basic integer. Then after the loop we count 2 more v4i32 ops. For a total arith cost of 4. But if you look at the assembly there would only be 3 arithmetic ops.
There are still more bugs in this code that I'm going to work on next. The non pairwise code shouldn't count extract subvectors in the loop. There are no extracts, the types are split in registers. For pairwise we need to use 2 two src permute shuffles.
Differential Revision: https://reviews.llvm.org/D55397
llvm-svn: 348621
DemandedBits and BDCE currently only support scalar integers. This
patch extends them to also handle vector integer operations. In this
case bits are not tracked for individual vector elements, instead a
bit is demanded if it is demanded for any of the elements. This matches
the behavior of computeKnownBits in ValueTracking and
SimplifyDemandedBits in InstCombine.
Unlike the previous iteration of this patch, getDemandedBits() can now
again be called on arbirary (sized) instructions, even if they don't
have integer or vector of integer type. (For vector types the size of the
returned mask will now be the scalar size in bits though.)
The added LoopVectorize test case shows a case which triggered an
assertion failure with the previous attempt, because getDemandedBits()
was called on a pointer-typed instruction.
Differential Revision: https://reviews.llvm.org/D55297
llvm-svn: 348602
This patch introduces a new instinsic `@llvm.experimental.widenable_condition`
that allows explicit representation for guards. It is an alternative to using
`@llvm.experimental.guard` intrinsic that does not contain implicit control flow.
We keep finding places where `@llvm.experimental.guard` is not supported or
treated too conservatively, and there are 2 reasons to that:
- `@llvm.experimental.guard` has memory write side effect to model implicit control flow,
and this sometimes confuses passes and analyzes that work with memory;
- Not all passes and analysis are aware of the semantics of guards. These passes treat them
as regular throwing call and have no idea that the condition of guard may be used to prove
something. One well-known place which had caused us troubles in the past is explicit loop
iteration count calculation in SCEV. Another example is new loop unswitching which is not
aware of guards. Whenever a new pass appears, we potentially have this problem there.
Rather than go and fix all these places (and commit to keep track of them and add support
in future), it seems more reasonable to leverage the existing optimizer's logic as much as possible.
The only significant difference between guards and regular explicit branches is that guard's condition
can be widened. It means that a guard contains (explicitly or implicitly) a `deopt` block successor,
and it is always legal to go there no matter what the guard condition is. The other successor is
a guarded block, and it is only legal to go there if the condition is true.
This patch introduces a new explicit form of guards alternative to `@llvm.experimental.guard`
intrinsic. Now a widenable guard can be represented in the CFG explicitly like this:
%widenable_condition = call i1 @llvm.experimental.widenable.condition()
%new_condition = and i1 %cond, %widenable_condition
br i1 %new_condition, label %guarded, label %deopt
guarded:
; Guarded instructions
deopt:
call type @llvm.experimental.deoptimize(<args...>) [ "deopt"(<deopt_args...>) ]
The new intrinsic `@llvm.experimental.widenable.condition` has semantics of an
`undef`, but the intrinsic prevents the optimizer from folding it early. This form
should exploit all optimization boons provided to `br` instuction, and it still can be
widened by replacing the result of `@llvm.experimental.widenable.condition()`
with `and` with any arbitrary boolean value (as long as the branch that is taken when
it is `false` has a deopt and has no side-effects).
For more motivation, please check llvm-dev discussion "[llvm-dev] Giving up using
implicit control flow in guards".
This patch introduces this new intrinsic with respective LangRef changes and a pass
that converts old-style guards (expressed as intrinsics) into the new form.
The naming discussion is still ungoing. Merging this to unblock further items. We can
later change the name of this intrinsic.
Reviewed By: reames, fedor.sergeev, sanjoy
Differential Revision: https://reviews.llvm.org/D51207
llvm-svn: 348593
The current algorithm that collects live/dead/inloop blocks relies on some invariants
related to RPO and PO traversals. In particular, the important fact it requires is that
the only loop's latch is the first block in PO traversal. It also relies on fact that during
RPO we visit all prececessors of a block before we visit this block (backedges ignored).
If a loop has irreducible non-loop cycle inside, both these assumptions may break.
This patch adds detection for this situation and prohibits the terminator folding
for loops with irreducible CFG.
We can in theory support this later, for this some algorithmic changes are needed.
Besides, irreducible CFG is not a frequent situation and we can just don't bother.
Thanks @uabelho for finding this!
Differential Revision: https://reviews.llvm.org/D55357
Reviewed By: skatkov
llvm-svn: 348567
DemandedBits and BDCE currently only support scalar integers. This
patch extends them to also handle vector integer operations. In this
case bits are not tracked for individual vector elements, instead a
bit is demanded if it is demanded for any of the elements. This matches
the behavior of computeKnownBits in ValueTracking and
SimplifyDemandedBits in InstCombine.
The getDemandedBits() method can now only be called on instructions that
have integer or vector of integer type. Previously it could be called on
any sized instruction (even if it was not particularly useful). The size
of the return value is now always the scalar size in bits (while
previously it was the type size in bits).
Differential Revision: https://reviews.llvm.org/D55297
llvm-svn: 348549
Partial Redundancy Elimination of GEPs prevents CodeGenPrepare from
sinking the addressing mode computation of memory instructions back
to its uses. The problem comes from the insertion of PHIs, which
confuse CGP and make it bail.
I've autogenerated the check lines of an existing test and added a
store instruction to demonstrate the motivation behind this change.
The store is now using the gep instead of a phi.
Differential Revision: https://reviews.llvm.org/D55009
llvm-svn: 348496
This reverts commit r348457.
The original commit causes clang to crash when doing an instrumented
build with a new pass manager. Reverting to unbreak our integrate.
llvm-svn: 348484
I was finally able to quantify what i thought was missing in the fix,
it was vector constants. If we have a scalar (and %x, -1),
it will be instsimplified before we reach this code,
but if it is a vector, we may still have a -1 element.
Thus, we want to avoid the fold if *at least one* element is -1.
Or in other words, ignoring the undef elements, no sign bits
should be set. Thus, m_NonNegative().
A follow-up for rL348181
https://bugs.llvm.org/show_bug.cgi?id=39861
llvm-svn: 348462
This patch teaches LoopSimplifyCFG to delete loop blocks that have
become unreachable after terminator folding has been done.
Differential Revision: https://reviews.llvm.org/D54023
Reviewed By: anna
llvm-svn: 348457
This change caused SEGVs in instcombine. (The r347934 change seems to me to be a
precipitating cause, not a root cause. Details are on the llvm-commits thread
for r347934.)
llvm-svn: 348426
Extracting from a splat constant is always handled by InstSimplify.
Move the test for this from InstCombine to InstSimplify to make
sure that stays true.
llvm-svn: 348423
Treat terminators which resume exception propagation as returning instructions
(at least, for the purposes of marking outlined functions `noreturn`). This is
to avoid inserting traps after calls to outlined functions which unwind.
rdar://46129950
llvm-svn: 348404
Summary: debug intrinsics might be marked norecurse to enable the caller function to be norecurse and optimized if needed. This avoids code gen optimisation differences when -g is used, as in globalOpt.cpp:processInternalGlobal checks.
Reviewers: chandlerc, jmolloy, aprantl
Reviewed By: aprantl
Subscribers: aprantl, llvm-commits
Differential Revision: https://reviews.llvm.org/D55187
llvm-svn: 348381
The tests here are based on the motivating cases from D54827.
More background:
1. We don't get these cases in general with SimplifyCFG because the root
of the pattern match is an icmp, not a branch. I'm not sure how often
we encounter this pattern vs. the seemingly more likely case with
branches, but I don't see evidence to leave the minimal pattern
unoptimized.
2. This has a chance of increasing compile-time because we're using a
ValueTracking call to handle the match. The motivating cases could be
handled with a simpler pair of calls to isImpliedTrueByMatchingCmp/
isImpliedFalseByMatchingCmp, but I saw that we have a more
comprehensive wrapper around those, so we might as well use it here
unless there's evidence that it's significantly slower.
3. Ideally, we'd handle the fold to constants in InstSimplify, but as
with the existing code here, we could extend this to handle cases
where the result is not a constant, but a new combined predicate.
That would mean splitting the logic across the 2 passes and possibly
duplicating the pattern-matching cost.
4. As mentioned in D54827, this seems like the kind of thing that should
be handled in Correlated Value Propagation, but that pass is currently
limited to dealing with instructions with constant operands, so extending
this bit of InstCombine is the smallest/easiest way to get these patterns
optimized.
llvm-svn: 348367
Ideally, we would fold all of these in InstSimplify in a
similar way to rL347896, but this is a bit awkward when
we're trying to simplify a compare directly because the
ValueTracking API expects the compare as an input, but
in InstSimplify, we just have the operands of the compare.
Given that we can do transforms besides just simplifications,
we might as well just extend the code in InstCombine (which
already does simplifications with constant operands).
llvm-svn: 348312
If a PHI node out of extracted region has multiple incoming values from it,
split this PHI on two parts. First PHI has incomings only from region and
extracts with it (they are placed to the separate basic block that added to the
list of outlined), and incoming values in original PHI are replaced by first
PHI. Similar solution is already used in CodeExtractor for PHIs in entry block
(severSplitPHINodes method). It covers PR39433 bug.
Patch by Sergei Kachkov!
Differential Revision: https://reviews.llvm.org/D55018
llvm-svn: 348205
When we have a shuffle that extends a source vector with undefs
and then do some binop on that, we must make sure that the extra
elements remain undef with that binop if we reverse the order of
the binop and shuffle.
'or' is probably the easiest example to show the bug because
'or C, undef --> -1' (not undef). But there are other
opcode/constant combinations where this is true as shown by
the 'shl' test.
llvm-svn: 348191
These are the baseline tests for D54827.
Patch based on code originally written by: @yinyuefengyi (luo xionghu)
Differential Revision: https://reviews.llvm.org/D54994
llvm-svn: 348151
Add '-k 1' to 'sort -b' calls in SimpleLoopUnswitch tests, as required
for sort implementation on NetBSD. The '-b' modifier is ineffective
if specified without any key. Per the manpage:
Note that the -b option has no effect unless key fields are specified.
Differential Revision: https://reviews.llvm.org/D55168
llvm-svn: 348097
We were adding the entire scalarization extraction cost for reductions, which returns the total cost of extracting every element of a vector type.
For reductions we don't need to do this - we just need to extract the 0'th element after the reduction pattern has completed.
Fixes PR37731
Rebased and reapplied after being reverted in rL347541 due to PR39774 - which was fixed by D54955/rL347759 and D55017/rL347997
Differential Revision: https://reviews.llvm.org/D54585
llvm-svn: 348076
Extend ssub.sat(X, C) -> sadd.sat(X, -C) canonicalization to also
support non-splat vector constants. This is done by generalizing
the implementation of the isNotMinSignedValue() helper to return
true for constants that are non-splat, but don't contain any
signed min elements.
Differential Revision: https://reviews.llvm.org/D55011
llvm-svn: 348072
Summary:
Follow up to D54270, which allowed importing of var args functions
unless they called va_start. As pointed out in the post-commit comments
on that patch, the inliner can handle functions that call va_start in
certain situations as well. Go ahead and enable importing of all var
args functions. Measurements on a large binary show that this increases
imports and binary size by an insignificant amount.
Reviewers: davidxl
Subscribers: mehdi_amini, inglorion, eraman, steven_wu, dexonsmith, llvm-commits
Differential Revision: https://reviews.llvm.org/D54607
llvm-svn: 348068
These tests should probably go under a separate test file because they
should fold with just -constprop, but they're similar to the scalar
tests already in here.
llvm-svn: 348045
Summary:
When mem2reg inserts phi nodes in blocks with unreachable predecessors,
it adds undef operands for those incoming edges. When there are
multiple such predecessors, the order is currently based on the address
of the BasicBlocks. This change fixes that by using the BBNumbers in
the sort/search predicates, as is done elsewhere in mem2reg to ensure
determinism.
Also adds a testcase with a bunch of unreachable preds, which
(nodeterministically) fails without the fix.
Reviewers: majnemer
Reviewed By: majnemer
Subscribers: mgrang, llvm-commits
Differential Revision: https://reviews.llvm.org/D55077
llvm-svn: 348024
Summary:
An additional fix for PR39774. Need to update the references for the
RedcutionRoot instruction when it is replaced during the vectorization
phase to avoid compiler crash on reduction vectorization.
Reviewers: RKSimon, spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D55017
llvm-svn: 347997
Adding a new reduction pattern match for vectorizing code similar
to TSVC s3111:
for (int i = 0; i < N; i++)
if (a[i] > b)
sum += a[i];
This patch adds support for fadd, fsub and fmull, as well as multiple
branches and different (but compatible) instructions (ex. add+sub) in
different branches.
The difference from the previous patch(https://reviews.llvm.org/D49168)
is as follows:
- Added check of fast-math property of fp-instruction to the
previous patch
- Fix/add some pattern for if-reduction.ll
Differential Revision: https://reviews.llvm.org/D54464
Patch by Takahiro Miyoshi <takahiro.miyoshi@linaro.org>
and Masakazu Ueno <masakazu.ueno@linaro.org>
llvm-svn: 347989
Terminator folding transform lacks MemorySSA update for memory Phis,
while they exist within MemorySSA analysis. They need exactly the same
type of updates as regular Phis. Failing to update them properly ends up
with inconsistent MemorySSA and manifests in various assertion failures.
This patch adds Memory Phi updates to this transform.
Thanks to @jonpa for finding this!
Differential Revision: https://reviews.llvm.org/D55050
Reviewed By: asbirlea
llvm-svn: 347979
r320789 suppressed moving the insertion point of SCEV expressions with
dev/rem operations to the loop header in non-loop-invariant situations.
This, and similar, hoisting is also unsafe in the loop-invariant case,
since there may be a guard against a zero denominator. This is an
adjustment to the fix of r320789 to suppress the movement even in the
loop-invariant case.
This fixes PR30806.
Differential Revision: https://reviews.llvm.org/D54713
llvm-svn: 347934
Also revert fix r347876
One of the buildbots was reporting a failure in some relevant tests that I can't
repro or explain at present, so reverting until I can isolate.
llvm-svn: 347911
This is an almost direct move of the functionality from InstCombine to
InstSimplify. There's no reason not to do this in InstSimplify because
we never create a new value with this transform.
(There's a question of whether any dominance-based transform belongs in
either of these passes, but that's a separate issue.)
I've changed 1 of the conditions for the fold (1 of the blocks for the
branch must be the block we started with) into an assert because I'm not
sure how that could ever be false.
We need 1 extra check to make sure that the instruction itself is in a
basic block because passes other than InstCombine may be using InstSimplify
as an analysis on values that are not wired up yet.
The 3-way compare changes show that InstCombine has some kind of
phase-ordering hole. Otherwise, we would have already gotten the intended
final result that we now show here.
llvm-svn: 347896
This commit caused a large compile-time slowdown in some cases when NDEBUG is
off due to the dominator tree verification it added. Fix this by only doing
dominator tree and loop info verification when something has been hoisted.
Differential Revision: https://reviews.llvm.org/D52827
llvm-svn: 347889
Summary:
When splitting musttail calls, the split blocks' original terminators
get removed; inform the DTU when this happens.
Also add a testcase that fails an assertion in the DTU without this fix.
Reviewers: fhahn, junbuml
Reviewed By: fhahn
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D55027
llvm-svn: 347872
TFE and LWE support requires extra result registers that are written in the
event of a failure in order to detect that failure case.
The specific use-case that initiated these changes is sparse texture support.
This means that if image intrinsics are used with either option turned on, the
programmer must ensure that the return type can contain all of the expected
results. This can result in redundant registers since the vector size must be a
power-of-2.
This change takes roughly 6 parts:
1. Modify the instruction defs in tablegen to add new instruction variants that
can accomodate the extra return values.
2. Updates to lowerImage in SIISelLowering.cpp to accomodate setting TFE or LWE
(where the bulk of the work for these instruction types is now done)
3. Extra verification code to catch cases where intrinsics have been used but
insufficient return registers are used.
4. Modification to the adjustWritemask optimisation to account for TFE/LWE being
enabled (requires extra registers to be maintained for error return value).
5. An extra pass to zero initialize the error value return - this is because if
the error does not occur, the register is not written and thus must be zeroed
before use. Also added a new (on by default) option to ensure ALL return values
are zero-initialized that is required for sparse texture support.
6. Disable the inst_combine optimization in the presence of tfe/lwe (later TODO
for this to re-enable and handle correctly).
There's an additional fix now to avoid a dmask=0
For an image intrinsic with tfe where all result channels except tfe
were unused, I was getting an image instruction with dmask=0 and only a
single vgpr result for tfe. That is incorrect because the hardware
assumes there is at least one vgpr result, plus the one for tfe.
Fixed by forcing dmask to 1, which gives the desired two vgpr result
with tfe in the second one.
The TFE or LWE result is returned from the intrinsics using an aggregate
type. Look in the test code provided to see how this works, but in essence IR
code to invoke the intrinsic looks as follows:
%v = call {<4 x float>,i32} @llvm.amdgcn.image.load.1d.v4f32i32.i32(i32 15,
i32 %s, <8 x i32> %rsrc, i32 1, i32 0)
%v.vec = extractvalue {<4 x float>, i32} %v, 0
%v.err = extractvalue {<4 x float>, i32} %v, 1
Differential revision: https://reviews.llvm.org/D48826
Change-Id: If222bc03642e76cf98059a6bef5d5bffeda38dda
llvm-svn: 347871
This reverts commits r347776 and r347778.
The first one, r347776, caused significant compile time regressions
for certain input files, see PR39836 for details.
llvm-svn: 347867
John Brawn