(Recommit after making sure IsVerbose gets properly initialized in
DiagnosticInfoOptimizationBase. See previous commit that takes care of
this.)
OptimizationRemarkAnalysis directly takes the role of the report that is
generated by LAA.
Then we need the magic to be able to turn an LAA remark into an LV
remark. This is done via a new OptimizationRemark ctor.
llvm-svn: 282813
Also, make foldSelectExtConst() a member of InstCombiner, remove
unnecessary parameters from its interface, and group visitSelectInst
helpers together in the header file.
llvm-svn: 282796
OptimizationRemarkAnalysis directly takes the role of the report that is
generated by LAA.
Then we need the magic to be able to turn an LAA remark into an LV
remark. This is done via a new OptimizationRemark ctor.
llvm-svn: 282758
Summary:
Not tunned up heuristic, but with this small heuristic there is about
+0.10% improvement on SPEC 2006
Reviewers: tejohnson, mehdi_amini, eraman
Subscribers: mehdi_amini, llvm-commits
Differential Revision: https://reviews.llvm.org/D24940
llvm-svn: 282733
The last one remaining after which emitAnalysis can be removed is when
we convert the LAA's report to a vectorization report. This requires
converting LAA to the new interface first.
llvm-svn: 282726
Summary:
The patch fixes regression caused by two earlier patches D18777 and D18867.
Reviewers: reames, sanjoy
Differential Revision: http://reviews.llvm.org/D24280
From: Li Huang
llvm-svn: 282650
This commit enables more unrolling for SystemZ by implementing the
SystemZTargetTransformInfo::getUnrollingPreferences() method.
It has been found that it is better to only unroll moderately, so the
DefaultUnrollRuntimeCount has been moved into UnrollingPreferences in order
to set this to a lower value for SystemZ (4).
Reviewers: Evgeny Stupachenko, Ulrich Weigand.
https://reviews.llvm.org/D24451
llvm-svn: 282570
There is really no reason for these to be separate.
The vectorizer started this pretty bad tradition that the text of the
missed remarks is pretty meaningless, i.e. vectorization failed. There,
you have to query analysis to get the full picture.
I think we should just explain the reason for missing the optimization
in the missed remark when possible. Analysis remarks should provide
information that the pass gathers regardless whether the optimization is
passing or not.
llvm-svn: 282542
(Re-committed after moving the template specialization under the yaml
namespace. GCC was complaining about this.)
This allows various presentation of this data using an external tool.
This was first recommended here[1].
As an example, consider this module:
1 int foo();
2 int bar();
3
4 int baz() {
5 return foo() + bar();
6 }
The inliner generates these missed-optimization remarks today (the
hotness information is pulled from PGO):
remark: /tmp/s.c:5:10: foo will not be inlined into baz (hotness: 30)
remark: /tmp/s.c:5:18: bar will not be inlined into baz (hotness: 30)
Now with -pass-remarks-output=<yaml-file>, we generate this YAML file:
--- !Missed
Pass: inline
Name: NotInlined
DebugLoc: { File: /tmp/s.c, Line: 5, Column: 10 }
Function: baz
Hotness: 30
Args:
- Callee: foo
- String: will not be inlined into
- Caller: baz
...
--- !Missed
Pass: inline
Name: NotInlined
DebugLoc: { File: /tmp/s.c, Line: 5, Column: 18 }
Function: baz
Hotness: 30
Args:
- Callee: bar
- String: will not be inlined into
- Caller: baz
...
This is a summary of the high-level decisions:
* There is a new streaming interface to emit optimization remarks.
E.g. for the inliner remark above:
ORE.emit(DiagnosticInfoOptimizationRemarkMissed(
DEBUG_TYPE, "NotInlined", &I)
<< NV("Callee", Callee) << " will not be inlined into "
<< NV("Caller", CS.getCaller()) << setIsVerbose());
NV stands for named value and allows the YAML client to process a remark
using its name (NotInlined) and the named arguments (Callee and Caller)
without parsing the text of the message.
Subsequent patches will update ORE users to use the new streaming API.
* I am using YAML I/O for writing the YAML file. YAML I/O requires you
to specify reading and writing at once but reading is highly non-trivial
for some of the more complex LLVM types. Since it's not clear that we
(ever) want to use LLVM to parse this YAML file, the code supports and
asserts that we're writing only.
On the other hand, I did experiment that the class hierarchy starting at
DiagnosticInfoOptimizationBase can be mapped back from YAML generated
here (see D24479).
* The YAML stream is stored in the LLVM context.
* In the example, we can probably further specify the IR value used,
i.e. print "Function" rather than "Value".
* As before hotness is computed in the analysis pass instead of
DiganosticInfo. This avoids the layering problem since BFI is in
Analysis while DiagnosticInfo is in IR.
[1] https://reviews.llvm.org/D19678#419445
Differential Revision: https://reviews.llvm.org/D24587
llvm-svn: 282539
LLVM developers might be surprised to learn that there are blocks
without valid insertion points (catchswitch), so it seems worth calling
that out explicitly. Also add a FIXME about what we should really be
doing if we ever need to make optimized Windows EH code debuggable.
While I'm here, make auto usage more consistent with LLVM standards and
avoid an unecessary call to insertBefore.
llvm-svn: 282521
This allows various presentation of this data using an external tool.
This was first recommended here[1].
As an example, consider this module:
1 int foo();
2 int bar();
3
4 int baz() {
5 return foo() + bar();
6 }
The inliner generates these missed-optimization remarks today (the
hotness information is pulled from PGO):
remark: /tmp/s.c:5:10: foo will not be inlined into baz (hotness: 30)
remark: /tmp/s.c:5:18: bar will not be inlined into baz (hotness: 30)
Now with -pass-remarks-output=<yaml-file>, we generate this YAML file:
--- !Missed
Pass: inline
Name: NotInlined
DebugLoc: { File: /tmp/s.c, Line: 5, Column: 10 }
Function: baz
Hotness: 30
Args:
- Callee: foo
- String: will not be inlined into
- Caller: baz
...
--- !Missed
Pass: inline
Name: NotInlined
DebugLoc: { File: /tmp/s.c, Line: 5, Column: 18 }
Function: baz
Hotness: 30
Args:
- Callee: bar
- String: will not be inlined into
- Caller: baz
...
This is a summary of the high-level decisions:
* There is a new streaming interface to emit optimization remarks.
E.g. for the inliner remark above:
ORE.emit(DiagnosticInfoOptimizationRemarkMissed(
DEBUG_TYPE, "NotInlined", &I)
<< NV("Callee", Callee) << " will not be inlined into "
<< NV("Caller", CS.getCaller()) << setIsVerbose());
NV stands for named value and allows the YAML client to process a remark
using its name (NotInlined) and the named arguments (Callee and Caller)
without parsing the text of the message.
Subsequent patches will update ORE users to use the new streaming API.
* I am using YAML I/O for writing the YAML file. YAML I/O requires you
to specify reading and writing at once but reading is highly non-trivial
for some of the more complex LLVM types. Since it's not clear that we
(ever) want to use LLVM to parse this YAML file, the code supports and
asserts that we're writing only.
On the other hand, I did experiment that the class hierarchy starting at
DiagnosticInfoOptimizationBase can be mapped back from YAML generated
here (see D24479).
* The YAML stream is stored in the LLVM context.
* In the example, we can probably further specify the IR value used,
i.e. print "Function" rather than "Value".
* As before hotness is computed in the analysis pass instead of
DiganosticInfo. This avoids the layering problem since BFI is in
Analysis while DiagnosticInfo is in IR.
[1] https://reviews.llvm.org/D19678#419445
Differential Revision: https://reviews.llvm.org/D24587
llvm-svn: 282499
Summary:
We don't currently need this facility for CFI. Disabling individual hot methods proved
to be a better strategy in Chrome.
Also, the design of the feature is suboptimal, as pointed out by Peter Collingbourne.
Reviewers: pcc
Subscribers: kcc
Differential Revision: https://reviews.llvm.org/D24948
llvm-svn: 282461
Summary:
This patch improves thinlto importer
by importing 3x larger functions that are called from hot block.
I compared performance with the trunk on spec, and there
were about 2% on povray and 3.33% on milc. These results seems
to be consistant and match the results Teresa got with her simple
heuristic. Some benchmarks got slower but I think they are just
noisy (mcf, xalancbmki, omnetpp)- running the benchmarks again with
more iterations to confirm. Geomean of all benchmarks including the noisy ones
were about +0.02%.
I see much better improvement on google branch with Easwaran patch
for pgo callsite inlining (the inliner actually inline those big functions)
Over all I see +0.5% improvement, and I get +8.65% on povray.
So I guess we will see much bigger change when Easwaran patch will land
(it depends on new pass manager), but it is still worth putting this to trunk
before it.
Implementation details changes:
- Removed CallsiteCount.
- ProfileCount got replaced by Hotness
- hot-import-multiplier is set to 3.0 for now,
didn't have time to tune it up, but I see that we get most of the interesting
functions with 3, so there is no much performance difference with higher, and
binary size doesn't grow as much as with 10.0.
Reviewers: eraman, mehdi_amini, tejohnson
Subscribers: mehdi_amini, llvm-commits
Differential Revision: https://reviews.llvm.org/D24638
llvm-svn: 282437
This patch ensures that we actually scalarize instructions marked scalar after
vectorization. Previously, such instructions may have been vectorized instead.
Differential Revision: https://reviews.llvm.org/D23889
llvm-svn: 282418
Summary:
If coroutine has no suspend points, remove heap allocation and turn a coroutine into a normal function.
Also, if a pattern is detected that coroutine resumes or destroys itself prior to coro.suspend call, turn the suspend point into a simple jump to resume or cleanup label. This pattern occurs when coroutines are used to propagate errors in functions that return expected<T>.
Reviewers: majnemer
Subscribers: mehdi_amini, llvm-commits
Differential Revision: https://reviews.llvm.org/D24408
llvm-svn: 282414
The index of the new insertelement instruction was evaluated in the
wrong way, it was considered as the index of the inserted value instead
of index of the position, where the value should be inserted.
llvm-svn: 282401
This patch fixes PR30366.
Function foldUDivShl() worked under the assumption that one of the values
in input to the function was always an instance of llvm::Instruction.
However, function visitUDivOperand() (the only user of foldUDivShl) was
clearly violating that precondition; internally, visitUDivOperand() uses pattern
matches to check the operands of a udiv. Pattern matchers for binary operators
know how to handle both Instruction and ConstantExpr values.
This patch fixes the problem in foldUDivShl(). Now we use pattern matchers
instead of explicit casts to Instruction. The reduced test case from PR30366
has been added to test file InstCombine/udiv-simplify.ll.
Differential Revision: https://reviews.llvm.org/D24565
llvm-svn: 282398
Stop looking at users of UndefValue and ConstantPointerNull in the
objective C ARC optimizers. The other users aren't actually
interesting, since they're not pointing at a particular object. I
imagine these calls could be optimized through -instcombine... maybe
they already are?
These early returns will be required at some point in the future, with a
WIP patch that asserts when someone accesses a use-list on ConstantData.
llvm-svn: 282338
Assumptions on UndefValue and ConstantPointerNull aren't relevant to
other users. Ignore them entirely to avoid wasting cycles walking
through their (possibly extremely extensive (cross-module)) use-lists.
It wasn't clear how to add a specific test for this, and it'll be
covered anyway by an eventual patch that asserts when trying to access
the use-list of an instance of ConstantData.
llvm-svn: 282334
Return early from llvm::isSafeToDestroyConstant() whenever the value
`isa<ConstantData>()`. These constants are shared across the
LLVMContext. We never really want to delete them here, and walking
their use-lists can be very expensive.
(This is motivated by an eventual goal of removing use-lists entirely
from ConstantData.)
llvm-svn: 282320
If inserting more than one constant into a vector:
define <4 x float> @foo(<4 x float> %x) {
%ins1 = insertelement <4 x float> %x, float 1.0, i32 1
%ins2 = insertelement <4 x float> %ins1, float 2.0, i32 2
ret <4 x float> %ins2
}
InstCombine could reduce that to a shufflevector:
define <4 x float> @goo(<4 x float> %x) {
%shuf = shufflevector <4 x float> %x, <4 x float> <float undef, float 1.0, float 2.0, float undef>, <4 x i32><i32 0, i32 5, i32 6, i32 3>
ret <4 x float> %shuf
}
Also, InstCombine tries to convert shuffle instruction to single insertelement, if one of the vectors is a constant vector and only a single element from this constant should be used in shuffle, i.e.
shufflevector <4 x float> %v, <4 x float> <float undef, float 1.0, float
undef, float undef>, <4 x i32> <i32 0, i32 5, i32 undef, i32 undef> ->
insertelement <4 x float> %v, float 1.0, 1
Differential Revision: https://reviews.llvm.org/D24182
llvm-svn: 282237
We already have the udiv variant of this transform, so I think this is ok for
InstCombine too even though there is an increase in IR instructions. As the
tests and TODO comments show, the transform can lead to follow-on combines.
This should fix: https://llvm.org/bugs/show_bug.cgi?id=28672
Differential Revision: https://reviews.llvm.org/D24527
llvm-svn: 282209
and also the dependent r282175 "GVN-hoist: do not dereference null pointers"
It's causing compiler crashes building Harfbuzz (PR30499).
llvm-svn: 282199
To hoist stores past loads, we used to search for potential
conflicting loads on the hoisting path by following a MemorySSA
def-def link from the store to be hoisted to the previous
defining memory access, and from there we followed the def-use
chains to all the uses that occur on the hoisting path. The
problem is that the def-def link may point to a store that does
not alias with the store to be hoisted, and so the loads that are
walked may not alias with the store to be hoisted, and even as in
the testcase of PR30216, the loads that may alias with the store
to be hoisted are not visited.
The current patch visits all loads on the path from the store to
be hoisted to the hoisting position and uses the alias analysis
to ask whether the store may alias the load. I was not able to
use the MemorySSA functionality to ask for whether load and
store are clobbered: I'm not sure which function to call, so I
used a call to AA->isNoAlias().
Store past store is still working as before using a MemorySSA
query: I added an extra test to pr30216.ll to make sure store
past store does not regress.
Differential Revision: https://reviews.llvm.org/D24517
llvm-svn: 282168
Without this patch, GVN-hoist would think that a branch instruction is a scalar instruction
and would try to value number it. The patch filters out all such kind of irrelevant instructions.
A bit frustrating is that there is no easy way to discard all those very infrequent instructions,
a bit like isa<TerminatorInst> that stands for a large family of instructions. I'm thinking that
checking for those very infrequent other instructions would cost us more in compilation time
than just letting those instructions getting numbered, so I'm still thinking that a simpler check:
if (isa<TerminatorInst>(I))
return false;
is better than listing all the other less frequent instructions.
Differential Revision: https://reviews.llvm.org/D23929
llvm-svn: 282160
The additional fix is:
When adding debug information to a lowered phi node in mem2reg
check that we have a valid insertion point after the phi for adding
the debug information.
This change addresses the issue in pr30468 where a lowered phi was
added before a catchswitch and no debug information should be added
after the phi in this case.
Differential Revision: https://reviews.llvm.org/D24797
llvm-svn: 282155
For MIPS '#' is the start of comment line. Therefore we get assembler errors if # is used in the structure names.
Differential: D24334
Reviewed by: zhaoqin
llvm-svn: 282141
Currently, we give up on loop interchange if we encounter a flow dependency
anywhere in the loop list. Worse yet, we don't even track output dependencies.
This patch updates the dependency matrix computation to track flow and output
dependencies in the same way we track anti dependencies.
This improves an internal workload by 2.2x.
Note the loop interchange pass is off by default and it can be enabled with
'-mllvm -enable-loopinterchange'
Differential Revision: https://reviews.llvm.org/D24564
llvm-svn: 282101
If we identify an instruction as uniform after vectorization, we know that we
should only use the value corresponding to the first vector lane of each unroll
iteration. However, when scalarizing such instructions, we still produce values
for the other vector lanes. This patch prevents us from generating the unused
scalars.
Differential Revision: https://reviews.llvm.org/D24275
llvm-svn: 282087
Summary: Now that we have more precise debug info, we should change back to use maximum to get basic block weight.
Reviewers: dnovillo
Subscribers: andreadb, llvm-commits
Differential Revision: https://reviews.llvm.org/D24788
llvm-svn: 282084
Summary:
This is an NFC refactoring change as a precursor to the actual fix for rematerializing in
presence of phi.
https://reviews.llvm.org/D24399
Pasted from review:
findRematerializableChainToBasePointer changed to return the root of the
chain. instead of true or false.
move the PHI matching logic into the caller by inspecting the root return value.
This includes an assertion that the alternate root is in the liveset for the
call.
Tested with current RS4GC tests.
Reviewers: reames, sanjoy
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24780
llvm-svn: 282023
This is a follow-up to r281284. Global Variables now can have
!dbg attachements, so ASAN should clone these when generating a
sanitized copy of a global variable.
<rdar://problem/24899262>
llvm-svn: 281994
The routines llvm::ConvertDebugDeclareToDebugValue() always returned
a true value which was never checked at the call site; change the
function return type to void.
This NFC cleanup was approved in the review https://reviews.llvm.org/D23715
llvm-svn: 281964
SROA doesn't preserve the llvm.mem.parallel_loop_access metadata when it
transforms loads/stores. This patch fixes a couple occurences of this
issue.
(Partially addresses PR28981).
Differential Revision: https://reviews.llvm.org/D23549
llvm-svn: 281960
When looking at the scribus_1.3 example from https://llvm.org/bugs/show_bug.cgi?id=10584, I noticed that we were spending a large amount of time computing loop exits in LCSSA. This code appears to be written with the assumption that LoopExits are stored in the Loop and thus cheap to query. This is not true, so we should cache the result across the potentially long running loop which tends to visit a small handful of Loops.
On the particular example from 10584, this change drops the time spent in LCSSA computation by about 80%.
Differential Revision: https://reviews.llvm.org/D24509
llvm-svn: 281949
Summary: Callsites in the same basic block should share the same hotness. This patch checks for the hottest callsite in the same basic block, and use the hotness for all callsites in that basic block for early inline decisions. It also fixes the test to add "-S" so theat the "CHECK-NOT" is actually checking the content.
Reviewers: dnovillo
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24734
llvm-svn: 281927
Summary: It does not make sense to set equal weights for all unkown branches as we have static branch prediction available.
Reviewers: dnovillo
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24732
llvm-svn: 281912
Summary: The call target count profile is directly derived from LBR branch->target data. This is more reliable than instruction frequency profiles that could be moved across basic block boundaries. This patches uses call target count profile to annotate call instructions.
Reviewers: davidxl, dnovillo
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24410
llvm-svn: 281911
Summary:
This patch is adding the support for a shadow memory with
dynamically allocated address range.
The compiler-rt needs to export a symbol containing the shadow
memory range.
This is required to support ASAN on windows 64-bits.
Reviewers: kcc, rnk, vitalybuka
Subscribers: kubabrecka, dberris, llvm-commits, chrisha
Differential Revision: https://reviews.llvm.org/D23354
llvm-svn: 281908
When phi nodes are created in the -mem2reg phase, the @llvm.dbg.declare
entries are converted to @llvm.dbg.value entries at the place where the
store instructions existed. However no entry is created to describe
the resulting value of the phi node.
The effect of this is especially noticeable in for loops which have a
constant for the intial value; the loop control variable's location
would be described as the intial constant value in the loop body once
the -mem2reg optimization phase was run.
This change adds the creation of the @llvm.dbg.value entries to describe
variables whose location is the result of a phi node created in -mem2reg.
Also when the phi node is finally lowered to a machine instruction it
is important that the lowered "load" instruction is placed before the
associated DEBUG_VALUE entry describing the value loaded.
Differential Revision: https://reviews.llvm.org/D23715
llvm-svn: 281895
We were updating metadata but not IR flags. Because we pick an arbitrary instruction to be the CSE candidate, it comes down to luck (50% or less chance) if this results in broken codegen or not, which is why PR30373 which is actually not the fault of the commit it was bisected down to.
Fixes PR30373.
llvm-svn: 281889
Summary: Previously we reline on inst-combine to remove inlinable invoke instructions. This causes trouble because a few extra optimizations are schedule early that could introduce too much CFG change (e.g. simplifycfg removes too much control flow). This patch handles invoke instruction in-place during sample profile annotation, so that we do not rely on instcombine to remove those invoke instructions.
Reviewers: davidxl, dnovillo
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24409
llvm-svn: 281870
Simplified GEP cloning in vectorizeMemoryInstruction().
Added an assertion that checks consecutive GEP, which should have only one loop-variant operand.
Differential Revision: https://reviews.llvm.org/D24557
llvm-svn: 281851
Summary:
This fixes an issue when files are compiled with -flto=thin
at default -O0. We need to rename anonymous globals before attempting
to write the module summary because all values need names for
the summary. This was happening at -O1 and above, but not before
the early exit when constructing the pipeline for -O0.
Also add an internal -prepare-for-thinlto option to enable this
to be tested via opt.
Fixes PR30419.
Reviewers: mehdi_amini
Subscribers: probinson, llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D24701
llvm-svn: 281840
The ValueSymbolTable is used to detect name conflict and rename
instructions automatically. This is not needed when the value
names are automatically discarded by the LLVMContext.
No functional change intended, just saving a little bit of memory.
This is a recommit of r281806 after fixing the accessor to return
a pointer instead of a reference and updating all the call-sites.
llvm-svn: 281813
computeKnownBits() already works for integer vectors, so allow vector types when calling that from InstCombine.
I don't think the change to use m_APInt in computeKnownBits is strictly necessary because we do check for
ConstantVector later, but it's more efficient to handle the splat case without needing to loop on vector elements.
This should work with InstSimplify, but doesn't yet, so I made that a FIXME comment on the test for PR24942:
https://llvm.org/bugs/show_bug.cgi?id=24942
Differential Revision: https://reviews.llvm.org/D24677
llvm-svn: 281777
A follow-up patch will rename this pass and the source file accordingly,
but I figured the non-NFC change will be easier to spot in isolation.
Differential Revision: https://reviews.llvm.org/D24641
llvm-svn: 281744
These 2 helper functions were already using APInt internally, so just
change the API and caller to allow folds for splats. The scalar
regression tests look quite thorough, so I just added a couple of
tests to prove that vectors are handled too.
These folds should be grouped with the other cmp+shift folds though.
That can be an NFC follow-up.
llvm-svn: 281663
GlobalOpt is already dead-code-eliminating global definitions. With
this change it also takes care of declarations.
Hopefully this should make it now a strict superset of GlobalDCE.
This is important for LTO/ThinLTO as we don't want the linker to see
"undefined reference" when it processes the input files: it could
prevent proper internalization (or even load an extra file from a
static archive, changing the behavior of the program!).
llvm-svn: 281653
This pattern is matched in foldICmpBinOpEqualityWithConstant() and already works
with vectors too. I changed some comments over there to point out the current
location. The tests for this transform are currently in 'sub.ll'.
Note that the remaining folds in this block all require a sub too, so they should
get grouped with the other icmp(sub) patterns.
llvm-svn: 281627
Summary: The return value of `maybeInsertAsanInitAtFunctionEntry` is ignored.
Reviewers: rnk
Subscribers: llvm-commits, chrisha, dberris
Differential Revision: https://reviews.llvm.org/D24568
llvm-svn: 281620
This is a big glob of transforms that probably should work for vectors,
but currently they are disallowed because of ConstantInt guards.
llvm-svn: 281614
Summary:
Function __asan_default_options is called by __asan_init before the
shadow memory got initialized. Instrumenting that function may lead
to flaky execution.
As the __asan_default_options is provided by users, we cannot expect
them to add the appropriate function atttributes to avoid
instrumentation.
Reviewers: kcc, rnk
Subscribers: dberris, chrisha, llvm-commits
Differential Revision: https://reviews.llvm.org/D24566
llvm-svn: 281503
This patch moves the processing of pointer induction variables in
collectLoopUniforms from the consecutive pointer phase of the analysis to the
phi node phase. Previously, if a pointer induction variable was used by both a
scalarized non-memory instruction as well as a vectorized memory instruction,
we would incorrectly identify the pointer as uniform. Pointer induction
variables should be treated the same as other phi nodes. That is, they are
uniform if all users of the induction variable and induction variable update
are uniform.
Differential Revision: https://reviews.llvm.org/D24511
llvm-svn: 281485
The '-asan-use-private-alias’ option (disabled by default) option is currently only enabled for Linux and ELF, but it also works on Darwin and Mach-O. This option also fixes a known problem with LTO on Darwin (https://github.com/google/sanitizers/issues/647). This patch enables the support for Darwin (but still keeps it off by default) and adds the LTO test case.
Differential Revision: https://reviews.llvm.org/D24292
llvm-svn: 281470
ObjC library call with call return.
ARC contraction tries to replace uses of an argument passed to an
objective-c library call with the call return value. For example, in the
following IR, it replaces uses of argument %9 and uses of the values
discovered traversing the chain upwards (%7 and %8) with the call return
%10, if they are dominated by the call to @objc_autoreleaseReturnValue.
This transformation enables code-gen to tail-call the call to
@objc_autoreleaseReturnValue, which is necessary to enable auto release
return value optimization.
%7 = tail call i8* @objc_loadWeakRetained(i8** %6)
%8 = bitcast i8* %7 to %0*
%9 = bitcast %0* %8 to i8*
%10 = tail call i8* @objc_autoreleaseReturnValue(i8* %9)
ret %0* %8
Since r276727, llvm started removing redundant bitcasts and as a result
started feeding the following IR to ARC contraction:
%7 = tail call i8* @objc_loadWeakRetained(i8** %6)
%8 = bitcast i8* %7 to %0*
%9 = tail call i8* @objc_autoreleaseReturnValue(i8* %7)
ret %0* %8
ARC contraction no longer does the optimization described above since it
only traverses the chain upwards and fails to recognize that the
function return can be replaced by the call return. This commit changes
ARC contraction to traverse the chain downwards too and replace uses of
bitcasts with the call return.
rdar://problem/28011339
Differential Revision: https://reviews.llvm.org/D24523
llvm-svn: 281419
Teach SimplifyLibcalls that in can treat functions annotated with
apcs, aapcs or aapcs_vfp like normal C functions if they only take
and return integer or pointer values, and the target is not iOS.
Differential Revision: https://reviews.llvm.org/D24453
llvm-svn: 281322
This patch reverses the edge from DIGlobalVariable to GlobalVariable.
This will allow us to more easily preserve debug info metadata when
manipulating global variables.
Fixes PR30362. A program for upgrading test cases is attached to that
bug.
Differential Revision: http://reviews.llvm.org/D20147
llvm-svn: 281284
Trying to infer the 'returned' attribute if an argument is already
'returned' can lead to verification failure: inference might determine
that a different argument is passed through which would result in two
different arguments marked as 'returned'.
This fixes PR30350.
llvm-svn: 281221
This should *actually* fix PR30244. This cranks up the workaround for PR30188 so that we never sink loads or stores of allocas.
The idea is that these should be removed by SROA/Mem2Reg, and any movement of them may well confuse SROA or just cause unwanted code churn. It's not ideal that the midend should be crippled like this, but that unwanted churn can really cause significant regressions in important workloads (tsan).
llvm-svn: 281162
Exposed by PR30244, we will split a block currently if we think we can sink at least one instruction. However this isn't right - the reason we split predecessors is so that we can sink instructions that otherwise couldn't be sunk because it isn't safe to do so - stores, for example.
So, change the heuristic to only split if it thinks it can sink at least one non-speculatable instruction.
Should fix PR30244.
llvm-svn: 281160
Everything under foldICmpInstWithConstant() should now be working for
splat vectors via m_APInt matchers. Ie, I've removed all of the FIXMEs
that I added while cleaning that section up. Note that not all of the
associated FIXMEs in the regression tests are gone though, because some
of the tests require earlier folds that are still scalar-only.
llvm-svn: 281139
Summary:
Could be useful for comparison when we suspect that alloca was skipped
because of this.
Reviewers: eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24437
llvm-svn: 281126
This would create a bitcast use which fails the verifier: swifterror values may
only be used by loads, stores, and as function arguments.
rdar://28233244
llvm-svn: 281114
Summary:
If one of the uses of the value is a single edge PHINode, handle it.
Original:
%val = something
<suspend>
%p = PHINode [%val]
After Spill + Part13:
%val = something
%slot = gep val.spill.slot
store %val, %slot
<suspend>
%p = load %slot
Plus tiny fixes/changes:
* use correct index for coro.free in CoroCleanup
* fixup id parameter in coro.free to allow authoring coroutine in plain C with __builtins
Reviewers: majnemer
Subscribers: mehdi_amini, llvm-commits
Differential Revision: https://reviews.llvm.org/D24242
llvm-svn: 281020
Summary: The hoisted instruction is executed speculatively. It could affect the debugging experience as user would see gdb go into code that may not be expected to execute. It will also affect sample profile accuracy by assigning incorrect frequency to source within then/else branch.
Reviewers: davidxl, dblaikie, chandlerc, kcc, echristo
Subscribers: mehdi_amini, probinson, eric_niebler, andreadb, llvm-commits
Differential Revision: https://reviews.llvm.org/D24164
llvm-svn: 280995
The test case included in r280979 wasn't checking what it was supposed to be
checking for the predicated store case. Fixing the test revealed that the
multi-use case (when a pointer is used by both vectorized and scalarized memory
accesses) wasn't being handled properly. We can't skip over
non-consecutive-like pointers since they may have looked consecutive-like with
a different memory access.
llvm-svn: 280992
Previously, all consecutive pointers were marked uniform after vectorization.
However, if a consecutive pointer is used by a memory access that is eventually
scalarized, the pointer won't remain uniform after all. An example is
predicated stores. Even though a predicated store may be consecutive, it will
still be scalarized, making it's pointer operand non-uniform.
This patch updates the logic in collectLoopUniforms to consider the cases where
a memory access may be scalarized. If a memory access may be scalarized, its
pointer operand is not marked uniform. The determination of whether a given
memory instruction will be scalarized or not has been moved into a common
function that is used by the vectorizer, cost model, and legality analysis.
Differential Revision: https://reviews.llvm.org/D24271
llvm-svn: 280979
I introduced this potential bug by missing this diff in:
https://reviews.llvm.org/rL280873
...however, I'm not sure how to reach this code path with a regression test.
We may be able to remove this code and assume that the transform to a constant
is always handled by InstSimplify?
llvm-svn: 280964
Refactor replaceDominatedUsesWith to have a flag to control whether to replace uses in BB itself.
Summary: This is in preparation for LoopSink pass which calls replaceDominatedUsesWith to update after sinking.
llvm-svn: 280949
Summary:
C allows to jump over variables declaration so lifetime.start can be
avoid before variable usage. To avoid false-positives on such rare cases
we detect them and remove from lifetime analysis.
PR27453
PR28267
Reviewers: eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24321
llvm-svn: 280907
Summary:
When cloning blocks for prologue/epilogue we need to replicate the loop
structure from the original loop. It wasn't a problem for the innermost
loops, but it led to an incorrect loop info when we unrolled a loop with
a child loop - in this case created prologue-loop had a child loop, but
loop info didn't reflect that.
This fixes PR28888.
Reviewers: chandlerc, sanjoy, hfinkel
Subscribers: llvm-commits, silvas
Differential Revision: https://reviews.llvm.org/D24203
llvm-svn: 280901
Summary:
C allows to jump over variables declaration so lifetime.start can be
avoid before variable usage. To avoid false-positives on such rare cases
we detect them and remove from lifetime analysis.
PR27453
PR28267
Reviewers: eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24321
llvm-svn: 280880
We can't create metadata-valued PHIs; don't try to do so when sinking.
I created a test case for this using the @llvm.type.test intrinsic, because it
takes a metadata parameter and does not have severe side effects (thus
SimplifyCFG is willing to otherwise sink it).
Previously, running the test case would crash with:
Invalid use of metadata!
%.sink = select i1 %flag, metadata <...>, metadata <0x4e45dc0>
LLVM ERROR: Broken function found, compilation aborted!
llvm-svn: 280866
This is a revert of r280676 which was a revert of r280637;
ie, this is r280637 again. It was speculatively reverted to
help debug buildbot failures.
llvm-svn: 280861
Summary:
LSV replaces multiple adjacent loads with one vectorized load and a
bunch of extractelement instructions. This patch makes the
extractelement instructions' names match those of the original loads,
for (hopefully) improved readability.
Reviewers: asbirlea, tstellarAMD
Subscribers: arsenm, mzolotukhin
Differential Revision: https://reviews.llvm.org/D23748
llvm-svn: 280818
This fixes a similar issue to the one already fixed by r280804
(revieved in D24256). Revision 280804 fixed the problem with unsafe dyn_casts
in the extrq/extrqi combining logic. However, it turns out that even the
insertq/insertqi logic was affected by the same problem.
llvm-svn: 280807
This patch fixes an assertion failure caused by unsafe dynamic casts on the
constant operands of sse4a intrinsic calls to extrq/extrqi
The combine logic that simplifies sse4a extrq/extrqi intrinsic calls currently
checks if the input operands are constants. Internally, that logic relies on
dyn_casts of values returned by calls to method Constant::getAggregateElement.
However, method getAggregateElemet may return nullptr if the constant element
cannot be retrieved. So, all the dyn_casts can potentially fail. This is what
happens for example if a constexpr value is passed in input to an extrq/extrqi
intrinsic call.
This patch fixes the problem by using a dyn_cast_or_null (instead of a simple
dyn_cast) on the result of each call to Constant::getAggregateElement.
Added reproducible test cases to x86-sse4a.ll.
Differential Revision: https://reviews.llvm.org/D24256
llvm-svn: 280804
This reverts commit r280796, as it broke the AArch64 bots for no reason.
The tests were passing and we should try to keep them passing, so a proper
review should make that happen.
llvm-svn: 280802
In failure cases it's not guaranteed that the PHI we're inspecting is actually in the successor block! In this case we need to bail out early, and never query getIncomingValueForBlock() as that will cause an assert.
llvm-svn: 280794
I should have realised this the first time around, but if we're avoiding sinking stores where the operands come from allocas so they don't create selects, we also have to do the same for loads because SROA will be just as defective looking at loads of selected addresses as stores.
Fixes PR30188 (again).
llvm-svn: 280792
PR30292 showed a case where our PHI checking wasn't correct. We were checking that all values were used by the same PHI before deciding to sink, but we weren't checking that the incoming values for that PHI were what we expected. As a result, we had to bail out after block splitting which caused us to never reach a steady state in SimplifyCFG.
Fixes PR30292.
llvm-svn: 280790
Currently the pass updates branch weights in the IR if the function has
any PGO info (entry frequency is set). However we could still have
regions of the CFG that does not have branch weights collected (e.g. a
cold region). In this case we'd use static estimates. Since static
estimates for branches are determined independently, they are
inconsistent. Updating them can "randomly" inflate block frequencies.
I've run into this in a completely cold loop of h264ref from
SPEC. -Rpass-with-hotness showed the loop to be completely cold during
inlining (before JT) but completely hot during vectorization (after JT).
The new testcase demonstrate the problem. We check array elements
against 1, 2 and 3 in a loop. The check against 3 is the loop-exiting
check. The block names should be self-explanatory.
In this example, jump threading incorrectly updates the weight of the
loop-exiting branch to 0, drastically inflating the frequency of the
loop (in the range of billions).
There is no run-time profile info for edges inside the loop, so branch
probabilities are estimated. These are the resulting branch and block
frequencies for the loop body:
check_1 (16)
(8) / |
eq_1 | (8)
\ |
check_2 (16)
(8) / |
eq_2 | (8)
\ |
check_3 (16)
(1) / |
(loop exit) | (15)
|
(back edge)
First we thread eq_1 -> check_2 to check_3. Frequencies are updated to
remove the frequency of eq_1 from check_2 and then from the false edge
leaving check_2. Changed frequencies are highlighted with * *:
check_1 (16)
(8) / |
eq_1~ | (8)
/ |
/ check_2 (*8*)
/ (8) / |
\ eq_2 | (*0*)
\ \ |
` --- check_3 (16)
(1) / |
(loop exit) | (15)
|
(back edge)
Next we thread eq_1 -> check_3 and eq_2 -> check_3 to check_1 as new
back edges. Frequencies are updated to remove the frequency of eq_1 and
eq_3 from check_3 and then the false edge leaving check_3 (changed
frequencies are highlighted with * *):
check_1 (16)
(8) / |
eq_1~ | (8)
/ |
/ check_2 (*8*)
/ (8) / |
/-- eq_2~ | (*0*)
(back edge) |
check_3 (*0*)
(*0*) / |
(loop exit) | (*0*)
|
(back edge)
As a result, the loop exit edge ends up with 0 frequency which in turn makes
the loop header to have maximum frequency.
There are a few potential problems here:
1. The profile data seems odd. There is a single profile sample of the
loop being entered. On the other hand, there are no weights inside the
loop.
2. Based on static estimation we shouldn't set edges to "extreme"
values, i.e. extremely likely or unlikely.
3. We shouldn't create profile metadata that is calculated from static
estimation. I am not sure what policy is but it seems to make sense to
treat profile metadata as something that is known to originate from
profiling. Estimated probabilities should only be reflected in BPI/BFI.
Any one of these would probably fix the immediate problem. I went for 3
because I think it's a good policy to have and added a FIXME about 2.
Differential Revision: https://reviews.llvm.org/D24118
llvm-svn: 280713
Summary:
Move early uses of spilled variables after CoroBegin.
For example, if a parameter had address taken, we may end up with the code
like:
define @f(i32 %n) {
%n.addr = alloca i32
store %n, %n.addr
...
call @coro.begin
This patch fixes the problem by moving uses of spilled variables after CoroBegin.
Reviewers: majnemer
Subscribers: mehdi_amini, llvm-commits
Differential Revision: https://reviews.llvm.org/D24234
llvm-svn: 280678
Summary:
A frontend may designate a particular suspend to be final, by setting the second argument of the coro.suspend intrinsic to true. Such a suspend point has two properties:
* it is possible to check whether a suspended coroutine is at the final suspend point via coro.done intrinsic;
* a resumption of a coroutine stopped at the final suspend point leads to undefined behavior. The only possible action for a coroutine at a final suspend point is destroying it via coro.destroy intrinsic.
This patch adds final suspend handling logic to CoroEarly and CoroSplit passes.
Now, the final suspend point example from docs\Coroutines.rst compiles and produces expected result (see test/Transform/Coroutines/ex5.ll).
Reviewers: majnemer
Subscribers: mehdi_amini, llvm-commits
Differential Revision: https://reviews.llvm.org/D24068
llvm-svn: 280646
The transform in question:
icmp (and (trunc W), C2), C1 -> icmp (and W, C2'), C1'
...is still not enabled for vectors, thus no functional change intended.
It's not clear to me if this is a good transform for vectors or even
scalars in general. Changing that behavior may be a follow-on patch.
llvm-svn: 280627
memcpy with ld/st.
When InstCombine replaces a memcpy with loads+stores it does not copy over the
llvm.mem.parallel_loop_access from the memcpy instruction. This patch fixes
that.
Differential Revision: https://reviews.llvm.org/D23499
llvm-svn: 280617
Summary:
The inliner may need to determine where a given funclet unwinds to,
and this determination may depend on other funclets throughout the
funclet tree. The code that performs this walk in getUnwindDestToken
memoizes results to avoid redundant computations. In the case that
a funclet's unwind destination is derived from its ancestor, there's
code to walk back down the tree from the ancestor updating the memo
map of its descendants to record the unwind destination. This change
fixes that code to account for the case that some descendant has a
different unwind destination, which can happen if that unwind dest
is a descendant of the EHPad being queried and thus didn't determine
its unwind destination.
Also update test inline-funclets.ll, which is supposed to cover such
scenarios, to include a case that fails an assertion without this fix
but passes with it.
Fixes PR29151.
Reviewers: majnemer
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24117
llvm-svn: 280610
Inheriting from std::iterator uses more boiler-plate than manual
typedefs. Avoid that in both ilist_iterator and
MachineInstrBundleIterator.
This has the side effect of removing ilist_iterator from certain ADL
lookups in namespace std; calls to std::next need to be qualified by
"std::" that didn't have to before. The one case of this in-tree was
operating on a temporary, so I used the more compact operator++.
llvm-svn: 280570
The motivating case occurs with SSE/AVX scalar intrinsics, so this is a first step towards
shrinking that to a single shufflevector.
Note that the transform is intentionally limited to shuffles that are equivalent to vector
selects to avoid creating arbitrary shuffle masks that may not lower well.
This should solve PR29126:
https://llvm.org/bugs/show_bug.cgi?id=29126
Differential Revision: https://reviews.llvm.org/D23886
llvm-svn: 280504
For uniform instructions, we're only required to generate a scalar value for
the first vector lane of each unroll iteration. Thus, if we have a reverse
interleaved group, computing the member index off the scalar GEP corresponding
to the last vector lane of its pointer operand technically makes the GEP
non-uniform. We should compute the member index off the first scalar GEP
instead.
I've added the updated member index computation to the existing reverse
interleaved group test.
llvm-svn: 280497
We're sinking stores, which is a good thing, but in the process creating selects for the store address operand, which SROA/Mem2Reg can't look through, which caused serious regressions.
The real fix is in SROA, which I'll be looking into.
llvm-svn: 280470
r280425 | dehao | 2016-09-01 16:15:50 -0700 (Thu, 01 Sep 2016) | 9 lines
Refactor LICM pass in preparation for LoopSink pass.
Summary: LoopSink pass uses some common function in LICM. This patch refactor the LICM code to make it usable by LoopSink pass (https://reviews.llvm.org/D22778).
r280429 | dehao | 2016-09-01 16:31:25 -0700 (Thu, 01 Sep 2016) | 9 lines
Refactor LICM to expose canSinkOrHoistInst to LoopSink pass.
Summary: LoopSink pass shares the same canSinkOrHoistInst functionality with LICM pass. This patch exposes this function in preparation of https://reviews.llvm.org/D22778
llvm-svn: 280453
Summary: LoopSink pass shares the same canSinkOrHoistInst functionality with LICM pass. This patch exposes this function in preparation of https://reviews.llvm.org/D22778
Reviewers: chandlerc, davidxl, danielcdh
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24171
llvm-svn: 280429
Summary: This is in preparation for LoopSink pass which calls replaceDominatedUsesWith to update after sinking.
Reviewers: chandlerc, davidxl, danielcdh
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24170
llvm-svn: 280427
Summary: LoopSink pass uses some common function in LICM. This patch refactor the LICM code to make it usable by LoopSink pass (https://reviews.llvm.org/D22778).
Reviewers: chandlerc, davidxl, danielcdh
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24168
llvm-svn: 280425
We can now maintain scalar values in VectorLoopValueMap. Thus, we no longer
have to create temporary vectors with insertelement instructions when handling
pointer induction variables. This case was mistakenly missed from r279649 when
refactoring the other scalarization code.
llvm-svn: 280405
This patch moves the allocation of VectorParts for PHI nodes into the actual
PHI widening code. Previously, we allocated these VectorParts in
vectorizeBlockInLoop, and passed them by reference to widenPHIInstruction. Upon
returning, we would then map the VectorParts in VectorLoopValueMap. This
behavior is problematic for the cases where we only want to generate a scalar
version of a PHI node. For example, if in the future we only generate a scalar
version of an induction variable, we would end up inserting an empty vector
entry into the map once we return to vectorizeBlockInLoop. We now no longer
need to pass VectorParts to the various PHI widening functions, and we can keep
VectorParts allocation as close as possible to the point at which they are
actually mapped in VectorLoopValueMap.
llvm-svn: 280390
Previous change broke the C API for creating an EarlyCSE pass w/
MemorySSA by adding a bool parameter to control whether MemorySSA was
used or not. This broke the OCaml bindings. Instead, change the old C
API entry point back and add a new one to request an EarlyCSE pass with
MemorySSA.
llvm-svn: 280379
While removing a scalar shackle from an icmp fold, I noticed that I couldn't find any tests to trigger
this code path.
The 'and' shrinking transform should be handled by InstCombiner::foldCastedBitwiseLogic()
or eliminated with InstSimplify. The icmp narrowing is part of InstCombiner::foldICmpWithCastAndCast().
Differential Revision: https://reviews.llvm.org/D24031
llvm-svn: 280370
This was a real restriction in the original version of SinkIfThenCodeToEnd. Now it's been rewritten, the restriction can be lifted.
As part of this, we handle a very common and useful case where one of the incoming branches is actually conditional. Consider:
if (a)
x(1);
else if (b)
x(2);
This produces the following CFG:
[if]
/ \
[x(1)] [if]
| | \
| | \
| [x(2)] |
\ | /
[ end ]
[end] has two unconditional predecessor arcs and one conditional. The conditional refers to the implicit empty 'else' arc. This same pattern can also be caused by an empty default block in a switch.
We can't sink the call to x() down to end because no call to x() happens on the third incoming arc (assume that x() has sideeffects for the sake of argument; if something is safe to speculate we could indeed sink nevertheless but this cannot happen in the general case and causes many extra selects).
We are now able to detect this case and split off the unconditional arcs to a common successor:
[if]
/ \
[x(1)] [if]
| | \
| | \
| [x(2)] |
\ / |
[sink.split] |
\ /
[ end ]
Now we can sink the call to x() into %sink.split. This can cause significant code simplification in many testcases.
llvm-svn: 280364
r279460 rewrote this function to be able to handle more than two incoming edges and took pains to ensure this didn't regress anything.
This time we change the logic for determining if an instruction should be sunk. Previously we used a single pass greedy algorithm - sink instructions until one requires more than one PHI node or we run out of instructions to sink.
This had the problem that sinking instructions that had non-identical but trivially the same operands needed extra logic so we sunk them aggressively. For example:
%a = load i32* %b %d = load i32* %b
%c = gep i32* %a, i32 0 %e = gep i32* %d, i32 1
Sinking %c and %e would naively require two PHI merges as %a != %d. But the loads are obviously equivalent (and maybe can't be hoisted because there is no common predecessor).
This is why we implemented the fairly complex function areValuesTriviallySame(), to look through trivial differences like this. However it's just not clever enough.
Instead, throw areValuesTriviallySame away, use pointer equality to check equivalence of operands and switch to a two-stage algorithm.
In the "scan" stage, we look at every sinkable instruction in isolation from end of block to front. If it's sinkable, we keep track of all operands that required PHI merging.
In the "sink" stage, we iteratively sink the last non-terminator in the source blocks. But when calculating how many PHIs are actually required to be inserted (to work out if we should stop or not) we remove any values that have already been sunk from the set of PHI-merges required, which allows us to be more aggressive.
This turns an algorithm with potentially recursive lookahead (looking through GEPs, casts, loads and any other instruction potentially not CSE'd) to two linear scans.
llvm-svn: 280351
We iterate over the result from SafeToMergeTerminators, so make it a SmallSetVector instead of a SmallPtrSet.
Should fix stage3 convergence builds.
llvm-svn: 280342
A very important case is not handled here: multiple arcs to a single block with a PHI. Consider:
a:
%1 = icmp %b, 1
br %1, label %c, label %e
c:
%2 = icmp %b, 2
br %2, label %d, label %e
d:
br %e
e:
phi [0, %a], [1, %c], [2, %d]
FoldValueComparisonIntoPredecessors will refuse to fold this, as it doesn't know how to deal with two arcs to a common destination with different PHI values. The answer is obvious - just split all conflicting arcs.
llvm-svn: 280338
-fprofile-dir=path allows the user to specify where .gcda files should be
emitted when the program is run. In particular, this is the first flag that
causes the .gcno and .o files to have different paths, LLVM is extended to
support this. -fprofile-dir= does not change the file name in the .gcno (and
thus where lcov looks for the source) but it does change the name in the .gcda
(and thus where the runtime library writes the .gcda file). It's different from
a GCOV_PREFIX because a user can observe that the GCOV_PREFIX_STRIP will strip
paths off of -fprofile-dir= but not off of a supplied GCOV_PREFIX.
To implement this we split -coverage-file into -coverage-data-file and
-coverage-notes-file to specify the two different names. The !llvm.gcov
metadata node grows from a 2-element form {string coverage-file, node dbg.cu}
to 3-elements, {string coverage-notes-file, string coverage-data-file, node
dbg.cu}. In the 3-element form, the file name is already "mangled" with
.gcno/.gcda suffixes, while the 2-element form left that to the middle end
pass.
llvm-svn: 280306
This is prep work before changing the callers to also use APInt which will
allow folds for splat vectors. Currently, the callers have ConstantInt
guards in place, so no functional change intended with this commit.
llvm-svn: 280282
Summary:
Current implementation of LI verifier isn't ideal and fails to detect
some cases when LI is incorrect. For instance, it checks that all
recorded loops are in a correct form, but it has no way to check if
there are no more other (unrecorded in LI) loops in the function. This
patch adds a way to detect such bugs.
Reviewers: chandlerc, sanjoy, hfinkel
Subscribers: llvm-commits, silvas, mzolotukhin
Differential Revision: https://reviews.llvm.org/D23437
llvm-svn: 280280
Summary:
Use MemorySSA, if requested, to do less conservative memory dependency
checking.
This change doesn't enable the MemorySSA enhanced EarlyCSE in the
default pipelines, so should be NFC.
Reviewers: dberlin, sanjoy, reames, majnemer
Subscribers: mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D19821
llvm-svn: 280279
Passing the types/opcode check still doesn't guarantee we'll actually vectorize.
Therefore, just make it clear we're attempting to vectorize.
llvm-svn: 280263
This is a first step towards supporting deopt value lowering and reporting entirely with the register allocator. I hope to build on this in the near future to support live-on-return semantics, but I have a use case which allows me to test and investigate code quality with just the live-in semantics so I've chosen to start there. For those curious, my use cases is our implementation of the "__llvm_deoptimize" function we bind to @llvm.deoptimize. I'm choosing not to hard code that fact in the patch and instead make it configurable via function attributes.
The basic approach here is modelled on what is done for the "Live In" values on stackmaps and patchpoints. (A secondary goal here is to remove one of the last barriers to merging the pseudo instructions.) We start by adding the operands directly to the STATEPOINT SDNode. Once we've lowered to MI, we extend the remat logic used by the register allocator to fold virtual register uses into StackMap::Indirect entries as needed. This does rely on the fact that the register allocator rematerializes. If it didn't along some code path, we could end up with more vregs than physical registers and fail to allocate.
Today, we *only* fold in the register allocator. This can create some weird effects when combined with arguments passed on the stack because we don't fold them appropriately. I have an idea how to fix that, but it needs this patch in place to work on that effectively. (There's some weird interaction with the scheduler as well, more investigation needed.)
My near term plan is to land this patch off-by-default, experiment in my local tree to identify any correctness issues and then start fixing codegen problems one by one as I find them. Once I have the live-in lowering fully working (both correctness and code quality), I'm hoping to move on to the live-on-return semantics. Note: I don't have any *known* miscompiles with this patch enabled, but I'm pretty sure I'll find at least a couple. Thus, the "experimental" tag and the fact it's off by default.
Differential Revision: https://reviews.llvm.org/D24000
llvm-svn: 280250
We check that a sinking candidate is used by only one PHI node during our legality checks. However for instructions that are used by other sinking candidates our heuristic is less conservative. This can result in a candidate actually being illegal when we come to sink it because of how we sunk a predecessor. Do the used-by-only-one-PHI checks again during sinking to ensure we don't crash.
llvm-svn: 280228
We're sinking stores, which is a good thing, but in the process creating selects for the store address operand, which SROA/Mem2Reg can't look through, which caused serious regressions.
The real fix is in SROA, which I'll be looking into.
llvm-svn: 280219
A very important case is not handled here: multiple arcs to a single block with a PHI. Consider:
a:
%1 = icmp %b, 1
br %1, label %c, label %e
c:
%2 = icmp %b, 2
br %2, label %d, label %e
d:
br %e
e:
phi [0, %a], [1, %c], [2, %d]
FoldValueComparisonIntoPredecessors will refuse to fold this, as it doesn't know how to deal with two arcs to a common destination with different PHI values. The answer is obvious - just split all conflicting arcs.
llvm-svn: 280218
This was a real restriction in the original version of SinkIfThenCodeToEnd. Now it's been rewritten, the restriction can be lifted.
As part of this, we handle a very common and useful case where one of the incoming branches is actually conditional. Consider:
if (a)
x(1);
else if (b)
x(2);
This produces the following CFG:
[if]
/ \
[x(1)] [if]
| | \
| | \
| [x(2)] |
\ | /
[ end ]
[end] has two unconditional predecessor arcs and one conditional. The conditional refers to the implicit empty 'else' arc. This same pattern can also be caused by an empty default block in a switch.
We can't sink the call to x() down to end because no call to x() happens on the third incoming arc (assume that x() has sideeffects for the sake of argument; if something is safe to speculate we could indeed sink nevertheless but this cannot happen in the general case and causes many extra selects).
We are now able to detect this case and split off the unconditional arcs to a common successor:
[if]
/ \
[x(1)] [if]
| | \
| | \
| [x(2)] |
\ / |
[sink.split] |
\ /
[ end ]
Now we can sink the call to x() into %sink.split. This can cause significant code simplification in many testcases.
llvm-svn: 280217
r279460 rewrote this function to be able to handle more than two incoming edges and took pains to ensure this didn't regress anything.
This time we change the logic for determining if an instruction should be sunk. Previously we used a single pass greedy algorithm - sink instructions until one requires more than one PHI node or we run out of instructions to sink.
This had the problem that sinking instructions that had non-identical but trivially the same operands needed extra logic so we sunk them aggressively. For example:
%a = load i32* %b %d = load i32* %b
%c = gep i32* %a, i32 0 %e = gep i32* %d, i32 1
Sinking %c and %e would naively require two PHI merges as %a != %d. But the loads are obviously equivalent (and maybe can't be hoisted because there is no common predecessor).
This is why we implemented the fairly complex function areValuesTriviallySame(), to look through trivial differences like this. However it's just not clever enough.
Instead, throw areValuesTriviallySame away, use pointer equality to check equivalence of operands and switch to a two-stage algorithm.
In the "scan" stage, we look at every sinkable instruction in isolation from end of block to front. If it's sinkable, we keep track of all operands that required PHI merging.
In the "sink" stage, we iteratively sink the last non-terminator in the source blocks. But when calculating how many PHIs are actually required to be inserted (to work out if we should stop or not) we remove any values that have already been sunk from the set of PHI-merges required, which allows us to be more aggressive.
This turns an algorithm with potentially recursive lookahead (looking through GEPs, casts, loads and any other instruction potentially not CSE'd) to two linear scans.
llvm-svn: 280216
This was deliberately disabled during my rewrite of SinkIfThenToEnd to keep behaviour
at least vaguely consistent with the previous version and keep it as close to NFC as
I could.
There's no real reason not to merge sideeffect calls though, so let's do it! Small fixup
along the way to ensure we don't create indirect calls.
Should fix PR28964.
llvm-svn: 280215
Summary:
1) CoroEarly now lowers llvm.coro.promise intrinsic that allows to obtain
a coroutine promise pointer from a coroutine frame and vice versa.
2) CoroFrame now interprets Promise argument of llvm.coro.begin to
place CoroutinPromise alloca at a deterministic offset from the coroutine frame.
Now, the coroutine promise example from docs\Coroutines.rst compiles and produces expected result (see test/Transform/Coroutines/ex4.ll).
Reviewers: majnemer
Subscribers: llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D23993
llvm-svn: 280184
It's much less code and easier to read if we don't duplicate
everything between the 'Inside' and not 'Inside' cases.
As noted with the FIXME, the goal is to make this vector-friendly
in a follow-up patch.
llvm-svn: 280183
Summary:
LSV was using two vector sets (heads and tails) to track pairs of adjiacent position to vectorize.
A recent optimization is trying to obtain the longest chain to vectorize and assumes the positions
in heads(H) and tails(T) match, which is not the case is there are multiple tails for the same head.
e.g.:
i1: store a[0]
i2: store a[1]
i3: store a[1]
Leads to:
H: i1
T: i2 i3
Instead of:
H: i1 i1
T: i2 i3
So the positions for instructions that follow i3 will have different indexes in H/T.
This patch resolves PR29148.
This issue also surfaced the fact that if the chain is too long, and TLI
returns a "not-fast" answer, the whole chain will be abandoned for
vectorization, even though a smaller one would be beneficial.
Added a testcase and FIXME for this.
Reviewers: tstellarAMD, arsenm, jlebar
Subscribers: mzolotukhin, wdng, llvm-commits
Differential Revision: https://reviews.llvm.org/D24057
llvm-svn: 280179
We don't need to limit predication to blocks that have a single incoming
edge, we just need to use the right mask.
This fixes PR30172.
Differential Revision: https://reviews.llvm.org/D24009
llvm-svn: 280148
Reverse iterators to doubly-linked lists can be simpler (and cheaper)
than std::reverse_iterator. Make it so.
In particular, change ilist<T>::reverse_iterator so that it is *never*
invalidated unless the node it references is deleted. This matches the
guarantees of ilist<T>::iterator.
(Note: MachineBasicBlock::iterator is *not* an ilist iterator, but a
MachineInstrBundleIterator<MachineInstr>. This commit does not change
MachineBasicBlock::reverse_iterator, but it does update
MachineBasicBlock::reverse_instr_iterator. See note at end of commit
message for details on bundle iterators.)
Given the list (with the Sentinel showing twice for simplicity):
[Sentinel] <-> A <-> B <-> [Sentinel]
the following is now true:
1. begin() represents A.
2. begin() holds the pointer for A.
3. end() represents [Sentinel].
4. end() holds the poitner for [Sentinel].
5. rbegin() represents B.
6. rbegin() holds the pointer for B.
7. rend() represents [Sentinel].
8. rend() holds the pointer for [Sentinel].
The changes are #6 and #8. Here are some properties from the old
scheme (which used std::reverse_iterator):
- rbegin() held the pointer for [Sentinel] and rend() held the pointer
for A;
- operator*() cost two dereferences instead of one;
- converting from a valid iterator to its valid reverse_iterator
involved a confusing increment; and
- "RI++->erase()" left RI invalid. The unintuitive replacement was
"RI->erase(), RE = end()".
With vector-like data structures these properties are hard to avoid
(since past-the-beginning is not a valid pointer), and don't impose a
real cost (since there's still only one dereference, and all iterators
are invalidated on erase). But with lists, this was a poor design.
Specifically, the following code (which obviously works with normal
iterators) now works with ilist::reverse_iterator as well:
for (auto RI = L.rbegin(), RE = L.rend(); RI != RE;)
fooThatMightRemoveArgFromList(*RI++);
Converting between iterator and reverse_iterator for the same node uses
the getReverse() function.
reverse_iterator iterator::getReverse();
iterator reverse_iterator::getReverse();
Why doesn't iterator <=> reverse_iterator conversion use constructors?
In order to catch and update old code, reverse_iterator does not even
have an explicit conversion from iterator. It wouldn't be safe because
there would be no reasonable way to catch all the bugs from the changed
semantic (see the changes at call sites that are part of this patch).
Old code used this API:
std::reverse_iterator::reverse_iterator(iterator);
iterator std::reverse_iterator::base();
Here's how to update from old code to new (that incorporates the
semantic change), assuming I is an ilist<>::iterator and RI is an
ilist<>::reverse_iterator:
[Old] ==> [New]
reverse_iterator(I) (--I).getReverse()
reverse_iterator(I) ++I.getReverse()
--reverse_iterator(I) I.getReverse()
reverse_iterator(++I) I.getReverse()
RI.base() (--RI).getReverse()
RI.base() ++RI.getReverse()
--RI.base() RI.getReverse()
(++RI).base() RI.getReverse()
delete &*RI, RE = end() delete &*RI++
RI->erase(), RE = end() RI++->erase()
=======================================
Note: bundle iterators are out of scope
=======================================
MachineBasicBlock::iterator, also known as
MachineInstrBundleIterator<MachineInstr>, is a wrapper to represent
MachineInstr bundles. The idea is that each operator++ takes you to the
beginning of the next bundle. Implementing a sane reverse iterator for
this is harder than ilist. Here are the options:
- Use std::reverse_iterator<MBB::i>. Store a handle to the beginning of
the next bundle. A call to operator*() runs a loop (usually
operator--() will be called 1 time, for unbundled instructions).
Increment/decrement just works. This is the status quo.
- Store a handle to the final node in the bundle. A call to operator*()
still runs a loop, but it iterates one time fewer (usually
operator--() will be called 0 times, for unbundled instructions).
Increment/decrement just works.
- Make the ilist_sentinel<MachineInstr> *always* store that it's the
sentinel (instead of just in asserts mode). Then the bundle iterator
can sniff the sentinel bit in operator++().
I initially tried implementing the end() option as part of this commit,
but updating iterator/reverse_iterator conversion call sites was
error-prone. I have a WIP series of patches that implements the final
option.
llvm-svn: 280032
Summary:
Fix a couple issues limiting the application of indirect call promotion
in ThinLTO mode:
- Invoke indirect call promotion before globalopt, since it may
eliminate imported functions which appear unreferenced.
- Invoke indirect call promotion with InLTO=true so that the PGOFuncName
metadata is used to get the name for locals which would have been
renamed during promotion.
Reviewers: davidxl, mehdi_amini
Subscribers: Prazek, llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D24004
llvm-svn: 280024
After r279649 when getting a vector value from VectorLoopValueMap, we create an
insertelement sequence on-demand if the value has been scalarized instead of
vectorized. We previously inserted this insertelement sequence before the
value's first vector user. However, this insert location is problematic if that
user is the phi node of a first-order recurrence. With this patch, we move the
insertelement sequence after the last scalar instruction we created when
scalarizing the value. Thus, the value's vector definition in the new loop will
immediately follow its scalar definitions. This should fix PR30183.
Reference: https://llvm.org/bugs/show_bug.cgi?id=30183
llvm-svn: 280001
Summary:
Calling __asan_poison_stack_memory and __asan_unpoison_stack_memory for small
variables is too expensive.
Code is disabled by default and can be enabled by -asan-experimental-poisoning.
PR27453
Reviewers: eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23947
llvm-svn: 279984
Summary: No functional changes, just refactoring to make D23947 simpler.
Reviewers: eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23954
llvm-svn: 279982
Summary:
While walking the use chain for identifying rematerializable values in RS4GC,
add the case where the current value and base value are the same PHI nodes.
This will aid rematerialization of geps and casts instead of relocating.
Reviewers: sanjoy, reames, igor
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23920
llvm-svn: 279975
Summary:
[Coroutines] Part 9: Add cleanup subfunction.
This patch completes coroutine heap allocation elision. Now, the heap elision example from docs\Coroutines.rst compiles and produces expected result (see test/Transform/Coroutines/ex3.ll)
Intrinsic Changes:
* coro.free gets a token parameter tying it to coro.id to allow reliably discovering all coro.frees associated with a particular coroutine.
* coro.id gets an extra parameter that points back to a coroutine function. This allows to check whether a coro.id describes the enclosing function or it belongs to a different function that was later inlined.
CoroSplit now creates three subfunctions:
# f$resume - resume logic
# f$destroy - cleanup logic, followed by a deallocation code
# f$cleanup - just the cleanup code
CoroElide pass during devirtualization replaces coro.destroy with either f$destroy or f$cleanup depending whether heap elision is performed or not.
Other fixes, improvements:
* Fixed buglet in Shape::buildFrame that was not creating coro.save properly if coroutine has more than one suspend point.
* Switched to using variable width suspend index field (no longer limited to 32 bit index field can be as little as i1 or as large as i<whatever-size_t-is>)
Reviewers: majnemer
Subscribers: llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D23844
llvm-svn: 279971
Without invalidating the entries in the MD cache we would try to access instructions
that were removed in previous iterations of hoisting.
Differential Revision: https://reviews.llvm.org/D23927
llvm-svn: 279907
Summary:
This is obviously an interesting case because it may motivate code
restructuring or LTO.
Reporting this requires instantiation of ORE in the loop where the call
sites are first gathered. I've checked compile-time
overhead *with* -Rpass-with-hotness and the worst slow-down was 6% in
mcf and quickly tailing off. As before without -Rpass-with-hotness
there is no overhead.
Because this could be a pretty noisy diagnostics, it is currently
qualified as 'verbose'. As of this patch, 'verbose' diagnostics are
only emitted with -Rpass-with-hotness, i.e. when the output is expected
to be filtered.
Reviewers: eraman, chandlerc, davidxl, hfinkel
Subscribers: tejohnson, Prazek, davide, llvm-commits
Differential Revision: https://reviews.llvm.org/D23415
llvm-svn: 279860
Like other recent changes near here, the goal is to allow vector types for
all of these folds. Splitting things up makes it easier to incrementally
enhance the code and easier to read.
llvm-svn: 279851
Summary: Dead store elimination gets very expensive when large numbers of instructions need to be analyzed. This patch limits the number of instructions analyzed per store to the value of the memdep-block-scan-limit parameter (which defaults to 100). This resulted in no observed difference in performance of the generated code, and no change in the statistics for the dead store elimination pass, but improved compilation time on some files by more than an order of magnitude.
Reviewers: dexonsmith, bruno, george.burgess.iv, dberlin, reames, davidxl
Subscribers: davide, chandlerc, dberlin, davidxl, eraman, tejohnson, mbodart, llvm-commits
Differential Revision: https://reviews.llvm.org/D15537
llvm-svn: 279833
We can't mark ORE (a function pass) preserved as required by the loop
passes because that is how we ensure that the required passes like
LazyBFI are all available any time ORE is used. See the new comments in
the patch.
Instead we use it directly just like the inliner does in D22694.
As expected there is some additional overhead after removing the caching
provided by analysis passes. The worst case, I measured was
LNT/CINT2006_ref/401.bzip2 which regresses by 12%. As before, this only
affects -Rpass-with-hotness and not default compilation.
llvm-svn: 279829
Removing the redundant 'CmpRHSV' local variable exposes a bug in the caller
foldICmpShrConstant() - it was sending in the div constant instead of the
cmp constant. But I have not been able to expose this in a regression test
yet - the affected folds all appear to be handled before we ever reach this
code. I'll keep trying to find a case as I make changes to allow vector folds
in both functions.
llvm-svn: 279828
Summary:
This fixes pr29105. The reason is that lifetime marks creates new
aliasing pointers the original ones, but before this patch aliases
were not checked in performMemCpyToMemSetOptzn.
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23846
llvm-svn: 279769
when unroll runtime iteration loop.
In llvm::UnrollRuntimeLoopRemainder, if the loop to be unrolled is the inner
loop inside a loop nest, the scalar evolution needs to be dropped for its
parent loop which is done by ScalarEvolution::forgetLoop. However, we can
postpone forgetLoop to the end of UnrollRuntimeLoopRemainder so TripCountSC
expansion can still reuse existing value.
Differential Revision: https://reviews.llvm.org/D23572
llvm-svn: 279748
It is invalid to hoist stores or loads if they are not executed on all paths
from the hoisting point to the exit of the function. In the testcase, there are
paths in the loop that do not execute the stores or the loads, and so hoisting
them within the loop is unsafe.
The problem is that the current implementation of hoistingFromAllPaths is
incomplete: it walks all blocks dominated by the hoisting point, and does not
return false when the loop contains a path on which the hoisted ld/st is
not executed.
Differential Revision: https://reviews.llvm.org/D23843
llvm-svn: 279732
There was no logic in foldICmpDivConstant, so no need for a separate function.
The code is directly copy/pasted, so further cleanups to follow.
llvm-svn: 279685
This patch unifies the data structures we use for mapping instructions from the
original loop to their corresponding instructions in the new loop. Previously,
we maintained two distinct maps for this purpose: WidenMap and ScalarIVMap.
WidenMap maintained the vector values each instruction from the old loop was
represented with, and ScalarIVMap maintained the scalar values each scalarized
induction variable was represented with. With this patch, all values created
for the new loop are maintained in VectorLoopValueMap.
The change allows for several simplifications. Previously, when an instruction
was scalarized, we had to insert the scalar values into vectors in order to
maintain the mapping in WidenMap. Then, if a user of the scalarized value was
also scalar, we had to extract the scalar values from the temporary vector we
created. We now aovid these unnecessary scalar-to-vector-to-scalar conversions.
If a scalarized value is used by a scalar instruction, the scalar value is used
directly. However, if the scalarized value is needed by a vector instruction,
we generate the needed insertelement instructions on-demand.
A common idiom in several locations in the code (including the scalarization
code), is to first get the vector values an instruction from the original loop
maps to, and then extract a particular scalar value. This patch adds
getScalarValue for this purpose along side getVectorValue as an interface into
VectorLoopValueMap. These functions work together to return the requested
values if they're available or to produce them if they're not.
The mapping has also be made less permissive. Entries can be added to
VectorLoopValue map with the new initVector and initScalar functions.
getVectorValue has been modified to return a constant reference to the mapped
entries.
There's no real functional change with this patch; however, in some cases we
will generate slightly different code. For example, instead of an insertelement
sequence following the definition of an instruction, it will now precede the
first use of that instruction. This can be seen in the test case changes.
Differential Revision: https://reviews.llvm.org/D23169
llvm-svn: 279649
I'm not sure if the `!isa<CallInst>(Inst) &&
!isa<TerminatorInst>(Inst))` bit is correct either, but this fixes the
case we know is broken.
llvm-svn: 279647
I deleted a fold from InstCombine at:
https://reviews.llvm.org/rL279568
because it (like any InstCombine to a constant?) should always happen in InstSimplify,
however, it's not obvious what the assumptions are in the remaining code.
Add a comment and assert to make it clearer.
Differential Revision: https://reviews.llvm.org/D23819
llvm-svn: 279626
div/rem instructions in basic blocks that require predication currently prevent
vectorization. This patch extends the existing mechanism for predicating stores
to handle other instructions and leverages it to predicate divs and rems.
Differential Revision: https://reviews.llvm.org/D22918
llvm-svn: 279620
manager, including both plumbing and logic to handle function pass
updates.
There are three fundamentally tied changes here:
1) Plumbing *some* mechanism for updating the CGSCC pass manager as the
CG changes while passes are running.
2) Changing the CGSCC pass manager infrastructure to have support for
the underlying graph to mutate mid-pass run.
3) Actually updating the CG after function passes run.
I can separate them if necessary, but I think its really useful to have
them together as the needs of #3 drove #2, and that in turn drove #1.
The plumbing technique is to extend the "run" method signature with
extra arguments. We provide the call graph that intrinsically is
available as it is the basis of the pass manager's IR units, and an
output parameter that records the results of updating the call graph
during an SCC passes's run. Note that "...UpdateResult" isn't a *great*
name here... suggestions very welcome.
I tried a pretty frustrating number of different data structures and such
for the innards of the update result. Every other one failed for one
reason or another. Sometimes I just couldn't keep the layers of
complexity right in my head. The thing that really worked was to just
directly provide access to the underlying structures used to walk the
call graph so that their updates could be informed by the *particular*
nature of the change to the graph.
The technique for how to make the pass management infrastructure cope
with mutating graphs was also something that took a really, really large
number of iterations to get to a place where I was happy. Here are some
of the considerations that drove the design:
- We operate at three levels within the infrastructure: RefSCC, SCC, and
Node. In each case, we are working bottom up and so we want to
continue to iterate on the "lowest" node as the graph changes. Look at
how we iterate over nodes in an SCC running function passes as those
function passes mutate the CG. We continue to iterate on the "lowest"
SCC, which is the one that continues to contain the function just
processed.
- The call graph structure re-uses SCCs (and RefSCCs) during mutation
events for the *highest* entry in the resulting new subgraph, not the
lowest. This means that it is necessary to continually update the
current SCC or RefSCC as it shifts. This is really surprising and
subtle, and took a long time for me to work out. I actually tried
changing the call graph to provide the opposite behavior, and it
breaks *EVERYTHING*. The graph update algorithms are really deeply
tied to this particualr pattern.
- When SCCs or RefSCCs are split apart and refined and we continually
re-pin our processing to the bottom one in the subgraph, we need to
enqueue the newly formed SCCs and RefSCCs for subsequent processing.
Queuing them presents a few challenges:
1) SCCs and RefSCCs use wildly different iteration strategies at
a high level. We end up needing to converge them on worklist
approaches that can be extended in order to be able to handle the
mutations.
2) The order of the enqueuing need to remain bottom-up post-order so
that we don't get surprising order of visitation for things like
the inliner.
3) We need the worklists to have set semantics so we don't duplicate
things endlessly. We don't need a *persistent* set though because
we always keep processing the bottom node!!!! This is super, super
surprising to me and took a long time to convince myself this is
correct, but I'm pretty sure it is... Once we sink down to the
bottom node, we can't re-split out the same node in any way, and
the postorder of the current queue is fixed and unchanging.
4) We need to make sure that the "current" SCC or RefSCC actually gets
enqueued here such that we re-visit it because we continue
processing a *new*, *bottom* SCC/RefSCC.
- We also need the ability to *skip* SCCs and RefSCCs that get merged
into a larger component. We even need the ability to skip *nodes* from
an SCC that are no longer part of that SCC.
This led to the design you see in the patch which uses SetVector-based
worklists. The RefSCC worklist is always empty until an update occurs
and is just used to handle those RefSCCs created by updates as the
others don't even exist yet and are formed on-demand during the
bottom-up walk. The SCC worklist is pre-populated from the RefSCC, and
we push new SCCs onto it and blacklist existing SCCs on it to get the
desired processing.
We then *directly* update these when updating the call graph as I was
never able to find a satisfactory abstraction around the update
strategy.
Finally, we need to compute the updates for function passes. This is
mostly used as an initial customer of all the update mechanisms to drive
their design to at least cover some real set of use cases. There are
a bunch of interesting things that came out of doing this:
- It is really nice to do this a function at a time because that
function is likely hot in the cache. This means we want even the
function pass adaptor to support online updates to the call graph!
- To update the call graph after arbitrary function pass mutations is
quite hard. We have to build a fairly comprehensive set of
data structures and then process them. Fortunately, some of this code
is related to the code for building the cal graph in the first place.
Unfortunately, very little of it makes any sense to share because the
nature of what we're doing is so very different. I've factored out the
one part that made sense at least.
- We need to transfer these updates into the various structures for the
CGSCC pass manager. Once those were more sanely worked out, this
became relatively easier. But some of those needs necessitated changes
to the LazyCallGraph interface to make it significantly easier to
extract the changed SCCs from an update operation.
- We also need to update the CGSCC analysis manager as the shape of the
graph changes. When an SCC is merged away we need to clear analyses
associated with it from the analysis manager which we didn't have
support for in the analysis manager infrsatructure. New SCCs are easy!
But then we have the case that the original SCC has its shape changed
but remains in the call graph. There we need to *invalidate* the
analyses associated with it.
- We also need to invalidate analyses after we *finish* processing an
SCC. But the analyses we need to invalidate here are *only those for
the newly updated SCC*!!! Because we only continue processing the
bottom SCC, if we split SCCs apart the original one gets invalidated
once when its shape changes and is not processed farther so its
analyses will be correct. It is the bottom SCC which continues being
processed and needs to have the "normal" invalidation done based on
the preserved analyses set.
All of this is mostly background and context for the changes here.
Many thanks to all the reviewers who helped here. Especially Sanjoy who
caught several interesting bugs in the graph algorithms, David, Sean,
and others who all helped with feedback.
Differential Revision: http://reviews.llvm.org/D21464
llvm-svn: 279618
Summary:
This patch adds coroutine frame building algorithm. Now, simple coroutines such as ex0.ll and ex1.ll (first examples from docs\Coroutines.rst can be compiled).
Documentation and overview is here: http://llvm.org/docs/Coroutines.html.
Upstreaming sequence (rough plan)
1.Add documentation. (https://reviews.llvm.org/D22603)
2.Add coroutine intrinsics. (https://reviews.llvm.org/D22659)
...
7. Split coroutine into subfunctions. (https://reviews.llvm.org/D23461)
8. Coroutine Frame Building algorithm <= we are here
9. Add f.cleanup subfunction.
10+. The rest of the logic
Reviewers: majnemer
Subscribers: mehdi_amini, llvm-commits
Differential Revision: https://reviews.llvm.org/D23586
llvm-svn: 279609
Summary:
This is part of a serious of patches to evolve ADCE.cpp to support
removing of unnecessary control flow.
This patch adds the ability to compute control dependences using
the iterated dominance frontier. We extend the liveness propagation
to alternate between data and control dependences until convergences.
Modify the pass manager intergation to compute the post-dominator tree
needed for iterator dominance frontier.
We still force all terminators live for now until we add code to
handlinge removing control flow in a later patch.
No changes to effective behavior with this patch
Previous patches:
D23225 [ADCE] Modify data structures to support removing control flow
D23065 [ADCE] Refactor anticipating new functionality (NFC)
D23102 [ADCE] Refactoring for new functionality (NFC)
Reviewers: nadav, majnemer, mehdi_amini
Subscribers: twoh, freik, llvm-commits
Differential Revision: https://reviews.llvm.org/D23559
llvm-svn: 279594
Summary:
In clang commit r268509 we started to invoke loop-unroll pass from the
driver even under -Os. However, we happen to not initialize optsize
thresholds properly, which si fixed with this change.
r268509 led to some big compile time regressions, because we started to
unroll some loops that we didn't unroll before. With this change I hope
to recover most of the regressions. We still are slightly slower than
before, because we do some checks here and there in loop-unrolling
before we bail out, but at least the slowdown is not that huge now.
Reviewers: hfinkel, chandlerc
Subscribers: mzolotukhin, llvm-commits
Differential Revision: https://reviews.llvm.org/D23388
llvm-svn: 279585
There will only be 3 lines of code in foldICmpShrConst() when the cleanup is done,
so it doesn't make much sense to have a separate function for a single fold.
llvm-svn: 279575
AFAICT, these already worked in all cases for scalar types, and I enhanced
the code to work for vector types in:
https://reviews.llvm.org/rL279543
llvm-svn: 279568
The test case included with r279125 exposed an existing signed integer
overflow. Since getTreeCost can return INT_MAX, we can't sum this cost together
with other costs, such as getReductionCost.
This patch removes the possibility of assigning a cost of INT_MAX. Since we
were previously using INT_MAX as an indicator for "should not vectorize", we
now explicitly check this condition with "isTreeTinyAndNotFullyVectorizable"
before computing a cost.
This patch adds a run-line to the test case used for r279125 that ensures we
don't vectorize. Previously, this line would vectorize the test case by chance
due to undefined behavior in the cost calculation.
Differential Revision: https://reviews.llvm.org/D23723
llvm-svn: 279562
Summary: GVNHoist: Use the pass version of MemorySSA and preserve it.
Reviewers: sebpop, george.burgess.iv
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23782
llvm-svn: 279504
Given that we're not currently using blocker info, and whether or not we
will end up using it it is unclear, don't waste 8 (or 4) bytes of memory
per path node.
llvm-svn: 279493
[Recommitting now an unrelated assertion in SROA is sorted out]
The new version has several advantages:
1) IMSHO it's more readable and neater
2) It handles loads and stores properly
3) It can handle any number of incoming blocks rather than just two. I'll be taking advantage of this in a followup patch.
With this change we can now finally sink load-modify-store idioms such as:
if (a)
return *b += 3;
else
return *b += 4;
=>
%z = load i32, i32* %y
%.sink = select i1 %a, i32 5, i32 7
%b = add i32 %z, %.sink
store i32 %b, i32* %y
ret i32 %b
When this works for switches it'll be even more powerful.
Round 4. This time we should handle all instructions correctly, and not replace any operands that need to be constant with variables.
This was really hard to determine safely, so the helper function should be put into the Instruction API. I'll do that as a followup.
llvm-svn: 279460
Summary: We can allow sinking if the single user block has only one unique predecessor, regardless of the number of edges. Note that a switch statement with multiple cases can have the same destination.
Reviewers: mcrosier, majnemer, spatel, reames
Subscribers: reames, mcrosier, llvm-commits
Differential Revision: https://reviews.llvm.org/D23722
llvm-svn: 279448
The new version has several advantages:
1) IMSHO it's more readable and neater
2) It handles loads and stores properly
3) It can handle any number of incoming blocks rather than just two. I'll be taking advantage of this in a followup patch.
With this change we can now finally sink load-modify-store idioms such as:
if (a)
return *b += 3;
else
return *b += 4;
=>
%z = load i32, i32* %y
%.sink = select i1 %a, i32 5, i32 7
%b = add i32 %z, %.sink
store i32 %b, i32* %y
ret i32 %b
When this works for switches it'll be even more powerful.
Round 4. This time we should handle all instructions correctly, and not replace any operands that need to be constant with variables.
This was really hard to determine safely, so the helper function should be put into the Instruction API. I'll do that as a followup.
llvm-svn: 279443
This change needs to be reverted in order to revert -r278267 which cause performance regression on MultiSource/Benchmarks/TSVC/Symbolics-flt/Symbolics-flt from LNT and some other bechmarks.
See comments on https://reviews.llvm.org/D18777 for details.
llvm-svn: 279432
Summary: r279379 introduced crash on arm 32bit bot. I suspect this is alignment issue.
Reviewers: eugenis
Subscribers: llvm-commits, aemerson
Differential Revision: https://reviews.llvm.org/D23762
llvm-svn: 279413
The callers still have ConstantInt guards, so there is no functional change
intended from this change. But relaxing the callers will allow more folds
for vector types.
llvm-svn: 279396
Summary:
We can insert function call instead of multiple store operation.
Current default is blocks larger than 64 bytes.
Changes are hidden behind -asan-experimental-poisoning flag.
PR27453
Differential Revision: https://reviews.llvm.org/D23711
llvm-svn: 279383
Summary:
Callbacks are not being used yet.
PR27453
Reviewers: kcc, eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23634
llvm-svn: 279380
Summary: Reduce store size to avoid leading and trailing zeros.
Reviewers: kcc, eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23648
llvm-svn: 279379
Summary:
We are going to combine poisoning of red zones and scope poisoning.
PR27453
Reviewers: kcc, eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23623
llvm-svn: 279373
The test case included in r279125 exposed existing undefined behavior in the
SLP vectorizer that it did not introduce. This patch reapplies the original
patch, but modifies the test case to avoid hitting the undefined behavior. This
allows us to close PR28330 while keeping the UBSan bot happy. The undefined
behavior the original test uncovered will be addressed in a follow-on patch.
Reference: https://llvm.org/bugs/show_bug.cgi?id=28330
llvm-svn: 279370
This is a partial enablement (move the ConstantInt guard down) because there are many
different folds here and one of the later ones will require reworking 'isSignBitCheck'.
llvm-svn: 279339
Currently nodes_iterator may dereference to a NodeType* or a NodeType&. Make them all dereference to NodeType*, which is NodeRef later.
Differential Revision: https://reviews.llvm.org/D23704
Differential Revision: https://reviews.llvm.org/D23705
llvm-svn: 279326
Specifically, this is done near the end of "SimplifyICmpInst" using
computeKnownBits() as the broader solution. There are even vector
tests (yay!) for this in test/Transforms/InstSimplify/compare.ll.
I considered putting an assert here instead of just deleting, but
then we could assert every possible fold in InstSimplify in
InstCombine, so...less is more?
llvm-svn: 279300
The intended transform is:
// Simplify icmp eq (or (ptrtoint P), (ptrtoint Q)), 0
// -> and (icmp eq P, null), (icmp eq Q, null).
P and Q are both pointer types, but may have different types. We need
two calls to getNullValue() to make the icmps.
llvm-svn: 279271
CGSCC use a WeakVH to track call sites. RAUW a call within a function
can result in that WeakVH getting confused about whether or not the call
site is still around.
llvm-svn: 279268
Of course, we really need to refactor and fix all of the cmp predicates,
but this one is interesting because without it, we later perform an
information-losing transform of icmp (shl 1, Y), C, and we can't recover
the better fold.
llvm-svn: 279263
The new version has several advantages:
1) IMSHO it's more readable and neater
2) It handles loads and stores properly
3) It can handle any number of incoming blocks rather than just two. I'll be taking advantage of this in a followup patch.
With this change we can now finally sink load-modify-store idioms such as:
if (a)
return *b += 3;
else
return *b += 4;
=>
%z = load i32, i32* %y
%.sink = select i1 %a, i32 5, i32 7
%b = add i32 %z, %.sink
store i32 %b, i32* %y
ret i32 %b
When this works for switches it'll be even more powerful.
llvm-svn: 279229
Summary: Reduce store size to avoid leading and trailing zeros.
Reviewers: kcc, eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23648
llvm-svn: 279178
Clean up the existing code by:
1. Renaming variables
2. Adding local variables
3. Making it vector-safe
This is still guarded by a ConstantInt check, so no functional change is intended.
But this should be ready to go: if we move the ConstantInt check down, all of
these folds should do the right thing for vector types.
llvm-svn: 279150
We abort building vectorizable trees in some cases (e.g., if the maximum
recursion depth is reached, if the region size is too large, etc.). If this
happens for a reduction, we can be left with a root entry that needs to be
gathered. For these cases, we need make sure we actually set VectorizedValue to
the resulting vector.
This patch ensures we properly set VectorizedValue, and it also ensures the
insertelement sequence generated for the gathers is inserted at the correct
location.
Reference: https://llvm.org/bugs/show_bug.cgi?id=28330
Differential Revison: https://reviews.llvm.org/D23410
llvm-svn: 279125
It causes a regression on our internal benchmark. Introduce cvp-dont-process flag and set it off by default while investigating the regression.
llvm-svn: 279082
Adam Nemet