Sanjay Patel [Mon, 15 May 2017 19:16:49 +0000 (19:16 +0000)]
[InstSimplify] restrict icmp fold with 2 sdiv exact operands (PR32949)
These folds were introduced with https://reviews.llvm.org/rL127064 as part of solving:
https://bugs.llvm.org/show_bug.cgi?id=9343
As shown here:
http://rise4fun.com/Alive/C8
...however, the sdiv exact case needs a stronger predicate.
I opted for duplicated code instead of adding another fallthrough because I think that's
easier to read (and edit in case we need/want to restrict/loosen the predicates any more).
This should fix:
https://bugs.llvm.org/show_bug.cgi?id=32949
https://bugs.llvm.org/show_bug.cgi?id=32948
Davide Italiano [Mon, 15 May 2017 18:50:53 +0000 (18:50 +0000)]
[NewGVN] Fix verification of MemoryPhis in verifyMemoryCongruency().
verifyMemoryCongruency() filters out trivially dead MemoryDef(s),
as we find them immediately dead, before moving from TOP to a new
congruence class.
This fixes the same problem for PHI(s) skipping MemoryPhis if all
the operands are dead.
[libFuzzer] fix a warning from Wunreachable-code-loop-increment reported by Christian Holler. This also fixes a logical bug, which however does not affect the libFuzzer's ability too much (I wasn't able to create a differentiating test)
Kyle Butt [Mon, 15 May 2017 17:30:47 +0000 (17:30 +0000)]
CodeGen: BlockPlacement: Increase tail duplication size for O3.
At O3 we are more willing to increase size if we believe it will improve
performance. The current threshold for tail-duplication of 2 instructions is
conservative, and can be relaxed at O3.
Benchmark results:
llvm test-suite:
6% improvement in aha, due to duplication of loop latch
3% improvement in hexxagon
2% slowdown in lpbench. Seems related, but couldn't completely diagnose.
Internal google benchmark:
Produces 4% improvement on internal google protocol buffer serialization
benchmarks.
Florian Hahn [Mon, 15 May 2017 15:15:22 +0000 (15:15 +0000)]
[AArch64] Enable FeatureFuseAES on Cortex-A72.
This patch enables fusing dependent AESE/AESMC and AESD/AESIMC
instruction pairs on Cortex-A72, as recommended in the Software
Optimization Guide, section 4.10.
John Brawn [Mon, 15 May 2017 11:57:54 +0000 (11:57 +0000)]
[ARM] Mark LEApcrel instructions as isAsCheapAsAMove
Doing this means that if an LEApcrel is used in two places we will rematerialize
instead of generating two MOVs. This is particularly useful for printfs using
the same format string, where we want to generate an address into a register
that's going to get corrupted by the call.
John Brawn [Mon, 15 May 2017 11:50:21 +0000 (11:50 +0000)]
[ARM] Mark LEApcrel as not having side effects
Doing this lets us hoist it out of loops, and I've also marked it as
rematerializable the same as the thumb1 and thumb2 counterparts.
It looks like it being marked as such was just a mistake, as the commit that
made that change only mentions LEApcrelJT and in thumb1 and thumb2 only the
LEApcrelJT instructions were marked as having side-effects, so it looks like
the intent was to only mark LEApcrelJT as having side-effects but LEApcrel was
accidentally marked as such also.
George Rimar [Mon, 15 May 2017 11:45:28 +0000 (11:45 +0000)]
[DWARF] - Speedup handling of relocations in DWARFContextInMemory.
I am working on a speedup of building .gdb_index in LLD and
noticed that relocations that are proccessed in DWARFContextInMemory often uses
the same symbol in a row. This patch introduces caching to reduce the relocations
proccessing time.
For benchmark,
I took debug LLC binary objects configured with -ggnu-pubnames and linked it using LLD.
Link time without --gdb-index is about 4,45s.
Link time with --gdb-index: a) Without patch: 19,16s b) With patch: 15,52s
That means time spent on --gdb-index in this configuration is
19,16s - 4,45s = 14,71s (without patch) vs 15,52s - 4,45s = 11,07s (with patch).
Ayman Musa [Mon, 15 May 2017 11:30:54 +0000 (11:30 +0000)]
[X86] Relocate code of replacement of subtarget unsupported masked memory intrinsics to run also on -O0 option.
Currently, when masked load, store, gather or scatter intrinsics are used, we check in CodeGenPrepare pass if the subtarget support these intrinsics, if not we replace them with scalar code - this is a functional transformation not an optimization (not optional).
CodeGenPrepare pass does not run when the optimization level is set to CodeGenOpt::None (-O0).
Functional transformation should run with all optimization levels, so here I created a new pass which runs on all optimization levels and does no more than this transformation.
Simon Pilgrim [Mon, 15 May 2017 10:47:48 +0000 (10:47 +0000)]
[NVPTX] Don't rely on default arguments to SelectionDAG::getMemIntrinsicNode. NFC.
NFC followup to D33147, this explicitly sets all the arguments (instead of relying on the defaults) to SelectionDAG::getMemIntrinsicNode to help identify -verify-machineinstrs issues.
Summary:
We were asserting in RegisterBankInfo if RBI.copyCost() returns
UINT_MAX. This is OK for RegBankSelect::Mode::Fast since we only
try one instruction mapping and can't recover from this, but for
RegBankSelect::Mode::Greedy we will be considering multiple
instruction mappings, so we can recover if we see a UNIT_MAX copy
cost.
The copy cost for one pair of register banks in the AMDGPU backend
will be UNIT_MAX, so this patch will prevent AMDGPU tests from
breaking.
MCObjectStreamer : fail with a diagnostic when emitting an out of range value.
We were previously silently emitting bogus data in release mode,
making it very hard to diagnose the error, or crashing with an
assert in debug mode. A proper diagnostic is now always emitted
when the value to be emitted is out of range.
Zvi Rackover [Sun, 14 May 2017 21:30:38 +0000 (21:30 +0000)]
[X86] Utilize SelectionDAG::getSelect(). NFC.
Replace SelectionDAG::getNode(ISD::SELECT, ...)
and SelectionDAG::getNode(ISD::VSELECT, ...)
with SelectionDAG::getSelect(...)
Saves a few lines of code and in some cases saves the need to explicitly
check the type of the desired node.
Craig Topper [Sun, 14 May 2017 19:38:11 +0000 (19:38 +0000)]
[X86] Add avx512vl command lines to the 128/256-bit vector-lzcnt tests so we can see what compare instructions are being used in the lookup table code.
I noticed the 512-bit lzcnts don't use the X86 specific lookup table code and instead use the EXPAND case in LegalizeDAG. I was toying around with fixing this and noticed it would require compare instructions that generate i1 masks and then converting from mask to vector. Then I noticed that we don't test which compares are used with avx512vl and no avx512cd.
Dimitry Andric [Sun, 14 May 2017 18:35:38 +0000 (18:35 +0000)]
Fix DynamicLibraryTest.cpp on FreeBSD and NetBSD
Summary:
After rL301562, on FreeBSD the DynamicLibrary unittests fail, because
the test uses getMainExecutable("DynamicLibraryTests", Ptr), and since
the path does not contain any slashes, retrieving the main executable
will not work.
Reimplement getMainExecutable() for FreeBSD and NetBSD using sysctl(3),
which is more reliable than fiddling with relative or absolute paths.
Also add retrieval of the original argv[] from the GoogleTest framework,
to use as a fallback for other OSes.
Shoaib Meenai [Sun, 14 May 2017 18:34:56 +0000 (18:34 +0000)]
[COFF] Gracefully handle empty .drectve sections
Running `llvm-readobj -coff-directives msvcrt.lib` resulted in this error:
Invalid data was encountered while parsing the file
This happened because some of the object files in the archive have empty
`.drectve` sections. These empty sections result in a `parse_failed` error being
returned from `COFFObjectFile::getSectionContents()`, which in turn caused
`llvm-readobj` to stop. With this change, `getSectionContents` now returns
success, and like before the resulting array is empty.
Zachary Turner [Sun, 14 May 2017 17:11:05 +0000 (17:11 +0000)]
[StringExtras] Add llvm::to_integer.
This is a very thin wrapper around StringRef::getAsInteger.
It serves three purposes.
1) It allows a cleaner syntax when you have something other than
a StringRef - for example, a std::string or an llvm::SmallString.
Previously, in this case you would have to write something like:
StringRef(MyStr).getAsInteger(0, Result)
by explicitly constructing a temporary StringRef. This can be
done implicitly however with the new function by just writing:
to_integer(MyStr, ...).
2) Correcting the travesty that is getAsInteger's return value.
This function returns true on success, and false on failure.
While this may cause confusion with people familiar with the
getAsInteger API, there seems to be widespread agreement that
the return semantics of getAsInteger was a mistake.
3) It allows the Radix to be deduced as a default argument by
putting it last in the parameter list. Most uses of getAsInteger
pass 0 for the first argument. With this syntax it can just be
omitted.
Vivek Pandya [Sat, 13 May 2017 08:55:43 +0000 (08:55 +0000)]
Simplify MIR Output used for Codegen Testing
- MIRYamlMapping: Default value provided for fields which have optional
mappings. Implemented == operators for required classes. When a field's value is
same as default value specified YAML IO class will not print it.
- MIRPrinter: Above mentioned behaviour is not on by default. If -simplify-mir
option not specified, then make yaml::Output to print fields with default values
too.
Craig Topper [Sat, 13 May 2017 06:56:04 +0000 (06:56 +0000)]
[InstCombine] Prevent InstCombine from triggering an extra iteration if something changed in the initial Worklist creation
Summary:
If the Worklist build causes an IR change this change flag currently factors into the flag for running another iteration of the iteration loop. But only changes during processing should trigger another loop.
This patch captures the worklist creation change flag into the outside the loop flag currently used for DbgDeclares and only sends that flag up to the caller. Rerunning the loop only depends on IC.run() now.
This uses the debug output of InstCombine to determine if one or two iterations run. I couldn't think of a better way to detect it since the second spurious iteration shoudn't make any visible changes. Just wasted computation.
I can do a pre-commit of the test case with the CHECK-NOT as a CHECK if this is an ok way to check this.
This is a subset of D31678 as I'm still not sure how to verify the analysis behavior for that.
Davide Italiano [Sat, 13 May 2017 02:45:47 +0000 (02:45 +0000)]
[NewGVN] XFAIL a flaky test until I find out what's going on.
I bet the change is correct but this test seems to expose some underlying
problem that manifest only on some buildbots, and I'm not able to reproduce
locally. Unfortunately I can't debug right now but I don't want to annoy
people with spurious failures, so I'll XFAIL until I can take a look (over
the weekend).
Justin Bogner [Sat, 13 May 2017 00:12:52 +0000 (00:12 +0000)]
AA: Use generic intrinsics for tests instead of target specific ones
Update a few tests to use llvm.masked.load/store instead of arm neon
vector loads and stores, and move the tests that are actually specific
to those arm intrinsics to their own files. This lets us mark the
tests that use target specific intrinsics as requiring those targets.
[GISel]: Fix undefined behavior while accessing DefaultAction map
We end up dereferencing the end iterator here when the Aspect doesn't exist in the DefaultAction map.
Change the API to return Optional<LLT> and return None when not found.
Also update the callers to handle the None case
Craig Topper [Fri, 12 May 2017 21:45:44 +0000 (21:45 +0000)]
[APInt] In udivrem, remember the bit width in a local variable so we don't reread it from the LHS which might be aliased with Quotient or Remainder.
This helped the compiler generate better code for the single word case. It was able to remember that the bit width was still a single word when it created the Remainder APInt and not create code for it possibly being multiword.
projects / llvm / log