This patch will address using the xscpsgndp instruction to copy floating point
scalar registers instead of the xxlor (specifically XXLORf) instruction that is
currently used. Additionally, this patch of utilizing xscpsgndp will apply to
P9, while pre-P9 will still use xxlor.
Patch by amyk
Differential Revision: https://reviews.llvm.org/D50004
llvm-svn: 340643
Previously we allowed the store to be Custom. But without knowing for sure that the Custom handling won't split the store, we shouldn't convert a volatile store. We also probably shouldn't be creating a store the requires custom handling after LegalizeOps. This could lead to an infinite loop if the custom handling was to insert a bitcast. Though I guess isStoreBitCastBeneficial could be used to block such a loop.
The test changes here are due to the volatile part of this. The stores in the test are all volatile and i32 stores are marked custom, So we are no longer converting them
This is related to D50491 where I was trying to allow some bitcasting of volatile loads
Differential Revision: https://reviews.llvm.org/D50578
llvm-svn: 340626
Summary:
Sometimes reading an output *.ll file it is not easy to understand why some callsites are not inlined. We can read output of inline remarks (option --pass-remarks-missed=inline) and try correlating its messages with the callsites.
An easier way proposed by this patch is to add to every callsite processed by Inliner an attribute with the latest message that describes the cause of not inlining this callsite. The attribute is called //inline-remark//. By default this feature is off. It can be switched on by the option //-inline-remark-attribute//.
For example in the provided test the result method //@test1// has two callsites //@bar// and inline remarks report different inlining missed reasons:
remark: <unknown>:0:0: bar not inlined into test1 because too costly to inline (cost=-5, threshold=-6)
remark: <unknown>:0:0: bar not inlined into test1 because it should never be inlined (cost=never): recursive
It is not clear which remark correspond to which callsite. With the inline remark attribute enabled we get the reasons attached to their callsites:
define void @test1() {
call void @bar(i1 true) #0
call void @bar(i1 false) #2
ret void
}
attributes #0 = { "inline-remark"="(cost=-5, threshold=-6)" }
..
attributes #2 = { "inline-remark"="(cost=never): recursive" }
Patch by: yrouban (Yevgeny Rouban)
Reviewers: xbolva00, tejohnson, apilipenko
Reviewed By: xbolva00, tejohnson
Subscribers: eraman, llvm-commits
Differential Revision: https://reviews.llvm.org/D50435
llvm-svn: 340618
Once the invariant_start is reached, we know that no instruction *after* it can modify the memory. So, if we can prove the location isn't read *between entry into the loop and the execution of the invariant_start*, we can execute the invariant_start before entering the loop.
Differential Revision: https://reviews.llvm.org/D51181
llvm-svn: 340617
The way that PhiValues is integrated with BasicAA it is possible for a pass
which uses BasicAA to pick up an instance of BasicAA that uses PhiValues without
intending to, and then delete values from a function in a way that causes
PhiValues to return dangling pointers to these deleted values. Fix this by
having a set of callback value handles to invalidate values when they're
deleted.
llvm-svn: 340613
When used in cross-DSO mode, CFI will generate calls to special functions rather than trap instructions. For example, instead of generating
if (!InlinedFastCheck(f))
abort();
call *f
CFI generates
if (!InlinedFastCheck(f))
__cfi_slowpath(CallSiteTypeId, f);
call *f
This patch teaches cfi-verify to recognize calls to __cfi_slowpath and abort and treat them as trap functions.
In addition to normal symbols, we also parse the dynamic relocations to handle cross-DSO calls in libraries.
We also extend cfi-verify to recognize other patterns that occur using cross-DSO. For example, some indirect calls are not guarded by a branch to a trap but instead follow a call to __cfi_slowpath. For example:
if (!InlinedFastCheck(f))
call *f
else {
__cfi_slowpath(CallSiteTypeId, f);
call *f
}
In this case, the second call to f is not marked as protected by the current code. We thus recognize if indirect calls directly follow a call to a function that will trap on CFI violations and treat them as protected.
We also ignore indirect calls in the PLT, since on AArch64 each entry contains an indirect call that should not be protected by CFI, and these are labeled incorrectly when debug information is not present.
Differential Revision: https://reviews.llvm.org/D49383
llvm-svn: 340612
This patch makes the DoesKMove argument non-optional, to force people
to think about it. Most cases where it is false are either code hoisting
or code sinking, where we pick one instruction from a set of
equal instructions among different code paths.
Reviewers: dberlin, nlopes, efriedma, davide
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D47475
llvm-svn: 340606
Per LLVM's CommandGuide, llvm-profdata show -text is supposed to produce
textual output that can be passed as input to further llvm-profdata
invocations. This previously didn't work for two reasons:
1) -text was not sufficient to enable the machine-readable text format output;
instead, -text was effectively ignored if -counts was not also specified. (With
this patch, -counts is instead ignored if -text is specified, because the
machine-readable text format always includes counts.)
2) When the input data was an IR-level profile, the :ir marker was missing from
the output, resulting in a text format output that would not be usable as
profiling data due to function hash mismatches.
Differential Revision: https://reviews.llvm.org/D51188
llvm-svn: 340592
Fix a set of related bugs:
* Considering two locations as equivalent when their lines are the same
but their scopes are different causes erroneous debug info that
attributes a commoned call to be attributed to one of the two calls it
was commoned from.
* The previous code to compute a new location's scope was inaccurate and
would use the inlinedAt that was the /parent/ of the inlinedAt that is
the nearest common one, and also used that parent scope instead of the
nearest common scope.
* Not generating new locations generally seemed like a lower quality
choice
There was some risk that generating more new locations could hurt object
size by making more fine grained line table entries, but it looks like
that was offset by the decrease in line table (& address & ranges) size
caused by more accurately computing the scope - which likely lead to
fewer range entries (more contiguous ranges) & reduced size that way.
All up with these changes I saw minor reductions (-1.21%, -1.77%) in
.rela.debug_ranges and .rela.debug_addr (in a fission, compressed debug
info build) as well as other minor size changes (generally reductinos)
across the board (-1.32% debug_info.dwo, -1.28% debug_loc.dwo). Measured
in an optimized (-O2) build of the clang binary.
If you are investigating a size regression in an optimized debug builds,
this is certainly a patch to look into - and I'd be happy to look into
any major regressions found & see what we can do to address them.
llvm-svn: 340583
Lower integer arguments smaller than i32.
Support both register and stack arguments.
Define setLocInfo function for setting LocInfo field in ArgLocs vector.
Patch by Petar Avramovic.
Differential Revision: https://reviews.llvm.org/D51031
llvm-svn: 340572
Summary:
Splats are fewer bytes than v128.consts, so use them when either could
apply.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D51179
llvm-svn: 340569
The variable index pattern is different than the constant index
cases as shown in D51125. We might want to splat regardless of
whether the scalar is loaded from memory or transferred from GPR.
llvm-svn: 340565
Summary:
I got "Use not jointly dominated by defs" when removePartialRedundancy
attempted to prune then re-extend a subrange whose only liveness was a
dead def at the copy being removed.
V2: Removed junk from test. Improved comment.
V3: Addressed minor review comments.
Subscribers: MatzeB, qcolombet, nhaehnle, llvm-commits
Differential Revision: https://reviews.llvm.org/D50914
Change-Id: I6f894e9f517f71e921e0c6d81d28c5f344db8dad
llvm-svn: 340549
These changes expand the FunctionAttr logic in order to mark functions as
WriteOnly when appropriate. This is done through an additional bool variable
and extended logic.
Reviewers: hfinkel, jdoerfert
Differential Revision: https://reviews.llvm.org/D48387
llvm-svn: 340537
When GVN sets the incoming value for a phi to undef because the incoming block
is unreachable it needs to also invalidate the cached info for that phi in
MemoryDependenceAnalysis, otherwise later queries will return stale information.
Differential Revision: https://reviews.llvm.org/D51099
llvm-svn: 340529
This version of the patch fixes cleaning up ssa_copy intrinsics, so it does not
crash for instructions in blocks that have been marked unreachable.
This patch updates IPSCCP to use PredicateInfo to propagate
facts to true branches predicated by EQ and to false branches
predicated by NE.
As a follow up, we should be able to extend it to also propagate additional
facts about nonnull.
Reviewers: davide, mssimpso, dberlin, efriedma
Reviewed By: davide, dberlin
Differential Revision: https://reviews.llvm.org/D45330
llvm-svn: 340525
Most users won't have to worry about this as all of the
'getOrInsertFunction' functions on Module will default to the program
address space.
An overload has been added to Function::Create to abstract away the
details for most callers.
This is based on https://reviews.llvm.org/D37054 but without the changes to
make passing a Module to Function::Create() mandatory. I have also added
some more tests and fixed the LLParser to accept call instructions for
types in the program address space.
Reviewed By: bjope
Differential Revision: https://reviews.llvm.org/D47541
llvm-svn: 340519
subtarget features for indirect calls and indirect branches.
This is in preparation for enabling *only* the call retpolines when
using speculative load hardening.
I've continued to use subtarget features for now as they continue to
seem the best fit given the lack of other retpoline like constructs so
far.
The LLVM side is pretty simple. I'd like to eventually get rid of the
old feature, but not sure what backwards compatibility issues that will
cause.
This does remove the "implies" from requesting an external thunk. This
always seemed somewhat questionable and is now clearly not desirable --
you specify a thunk the same way no matter which set of things are
getting retpolines.
I really want to keep this nicely isolated from end users and just an
LLVM implementation detail, so I've moved the `-mretpoline` flag in
Clang to no longer rely on a specific subtarget feature by that name and
instead to be directly handled. In some ways this is simpler, but in
order to preserve existing behavior I've had to add some fallback code
so that users who relied on merely passing -mretpoline-external-thunk
continue to get the same behavior. We should eventually remove this
I suspect (we have never tested that it works!) but I've not done that
in this patch.
Differential Revision: https://reviews.llvm.org/D51150
llvm-svn: 340515
Aligning section contents is not required, but only
recommended, by the specification. Microsoft's documentation says
(https://docs.microsoft.com/en-us/windows/desktop/debug/pe-format#section-table-section-headers):
"For object files, the value should be aligned on a 4-byte boundary
for best performance."
However, according to my measurements, aligning section contents has
a neutral to negative effect on performance.
I measured the median run time of 100 links of Chromium's
base_unittests on Linux with lld-link and on Windows with link.exe with
both aligned and unaligned sections. On Linux I didn't see a measurable
performance difference, and on Windows the link was slightly faster
with unaligned sections (presumably because on Windows the bottleneck
is I/O).
Also, the sections created by cl.exe are unaligned, so we should expect
tools to broadly accept unaligned sections.
Differential Revision: https://reviews.llvm.org/D51149
llvm-svn: 340514
This patch's test case relies on debug prints which isn't generally an
OK way to test stuff in LLVM and fails whenever asserts aren't enabled.
I've send a heads-up to the commit and detailed comments on the review.
llvm-svn: 340513
In lib/CodeGen/LiveDebugVariables.cpp, it uses std::prev(MBBI) to
get DebugValue's SlotIndex. However, the previous instruction may be
also a debug instruction. It could not use a debug instruction to query
SlotIndex in mi2iMap.
Scan all debug instructions and use the first debug instruction to query
SlotIndex for following debug instructions. Only handle DBG_VALUE in
handleDebugValue().
Differential Revision: https://reviews.llvm.org/D50621
llvm-svn: 340508
The format is the same as in ELF: a sequence of ULEB128-encoded
symbol indexes.
Differential Revision: https://reviews.llvm.org/D51047
llvm-svn: 340499
If we have a min/max pair we can do a better job of counting sign bits if we look at them together. This is similar to what is done in the SelectionDAG version of computeNumSignBits for ISD::SMAX/SMIN.
Differential Revision: https://reviews.llvm.org/D51112
llvm-svn: 340480
Previously we asumed a vector reduction add is part of a loop and one of the input is a phi. But the code in SelectionDAGBuilder that sets vector reduction flag handles more cases than that. It just requires that the use chain ends in a horizontal reduction. And there are no other uses. This means it can handle unrolled reduction loops.
If the initial value of the reduction was 0, an unrolled loop would begin with a vector reduction add that has two sad inputs. Previously we would only transform one side of the add, but for this case we need to transform both sides.
I've created a lambda to reuse some of the code for both sides. And fixed the variables names to remove reference to "phi".
Differential Revision: https://reviews.llvm.org/D50817
llvm-svn: 340478
Summary:
This CL adds support for arbitrary BUILD_VECTORS, i.e. not splats and
not consts. This is the last feature needed to properly lower v2i64
multiplies without a i64x2.mul instruction (which is not in the spec),
so i64x2.mul is removed as well.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D51082
Remove unnecessary condition and fix whitespace
llvm-svn: 340472
This solves the motivating case from:
https://bugs.llvm.org/show_bug.cgi?id=38527
If we are legalizing an FP vector op that maps to 1 of the LLVM intrinsics that mimic libm calls,
but we're going to end up with scalar libcalls for that vector type anyway, then we should unroll
the vector op into scalars before widening. This avoids libcalls because we've lost the knowledge
that some of the scalar elements are undef.
Differential Revision: https://reviews.llvm.org/D50791
llvm-svn: 340469
The inline sequence is very long (about 70 bytes on Thumb1), so it's
not really a good idea to inline it, especially when optimizing for
size.
Differential Revision: https://reviews.llvm.org/D47917
llvm-svn: 340458
Instead of asserting that the function doesn't have any unreachable
code, just ignore it for the purpose of computing liveness.
Differential Revision: https://reviews.llvm.org/D51070
llvm-svn: 340456
Summary:
When we don't actually have stack-allocated variables but need SP only
to support EH, we don't need to write SP back in the epilog, because we
don't bump down the stack pointer.
Reviewers: dschuff
Subscribers: jgravelle-google, sbc100, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D51114
llvm-svn: 340454
On Windows, movw+movt pairs with relocations are handled with a single
relocation that covers them both. Therefore we can't inject anything
between these instructions, otherwise the relocation (which in LLVM
only is treated as the movw instruction's relocation, while the movt
instruction's relocation is dropped) will end up bogus.
These instructions are bundled up until right before the constant
islands pass, making this effectively the only place that can split
them apart.
Differential Revision: https://reviews.llvm.org/D51032
llvm-svn: 340451
This avoids a potential infinite loop setting and unsetting bits in the
mask.
Reduced from a failure on the polly-aosp bot.
Differential Revision: https://reviews.llvm.org/D51066
llvm-svn: 340446
Summary:
Add MemorySSA as a dependency to LoopSimplifyCFG and preserve it.
Disabled by default until all passes preserve MemorySSA.
Reviewers: bogner, chandlerc
Subscribers: sanjoy, jlebar, Prazek, george.burgess.iv, llvm-commits
Differential Revision: https://reviews.llvm.org/D50911
llvm-svn: 340445
Summary:
Add MemorySSA as a depency to LoopInstInstSimplify and preserve it.
Disabled by default until all passes preserve MemorySSA.
Reviewers: chandlerc
Subscribers: sanjoy, jlebar, Prazek, george.burgess.iv, llvm-commits
Differential Revision: https://reviews.llvm.org/D50906
llvm-svn: 340444
