1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 11:13:28 +01:00
Commit Graph

123460 Commits

Author SHA1 Message Date
Tim Northover
cbca7246d2 MachO: improve load command tests slightly
llvm-svn: 251815
2015-11-02 18:33:35 +00:00
Tim Northover
bd66b4c598 WatchOS: update default CPU for triple after t2dsp -> dsp rename
llvm-svn: 251814
2015-11-02 18:21:07 +00:00
Teresa Johnson
1cf56803da Clang format a few prior patches (NFC)
I had clang formatted my earlier patches using the wrong style.
Reformatted with the LLVM style.

llvm-svn: 251812
2015-11-02 18:02:11 +00:00
Tim Northover
58717c5330 TvOS: add missing support for some libcalls.
llvm-svn: 251811
2015-11-02 18:00:00 +00:00
Artur Pilipenko
0dd1f670a9 Preserve load alignment and dereferenceable metadata during some transformations
Reviewed By: hfinkel

Differential Revision: http://reviews.llvm.org/D13953

llvm-svn: 251809
2015-11-02 17:53:51 +00:00
Matthias Braun
80fb3bb810 lit: Add '-a' option to display commands+output of all tests
The existing -v option only displays commands and outputs for failed
tests, the newly introduced -a displays it for all executed tests.

llvm-svn: 251806
2015-11-02 16:13:46 +00:00
Silviu Baranga
439f20d646 Add missing override statements in ScalarEvolution.h. NFC
llvm-svn: 251805
2015-11-02 15:29:49 +00:00
Pawel Bylica
51296e1676 Use static instead of anonymous namespace for helper functions. NFC.
llvm-svn: 251801
2015-11-02 14:57:24 +00:00
Silviu Baranga
ab8c77a47d [SCEV][LV] Add SCEV Predicates and use them to re-implement stride versioning
Summary:
SCEV Predicates represent conditions that typically cannot be derived from
static analysis, but can be used to reduce SCEV expressions to forms which are
usable for different optimizers.

ScalarEvolution now has the rewriteUsingPredicate method which can simplify a
SCEV expression using a SCEVPredicateSet. The normal workflow of a pass using
SCEVPredicates would be to hold a SCEVPredicateSet and every time assumptions
need to be made a new SCEV Predicate would be created and added to the set.
Each time after calling getSCEV, the user will call the rewriteUsingPredicate
method.

We add two types of predicates
SCEVPredicateSet - implements a set of predicates
SCEVEqualPredicate - tests for equality between two SCEV expressions

We use the SCEVEqualPredicate to re-implement stride versioning. Every time we
version a stride, we will add a SCEVEqualPredicate to the context.
Instead of adding specific stride checks, LoopVectorize now adds a more
generic SCEV check.

We only need to add support for this in the LoopVectorizer since this is the
only pass that will do stride versioning.

Reviewers: mzolotukhin, anemet, hfinkel, sanjoy

Subscribers: sanjoy, hfinkel, rengolin, jmolloy, llvm-commits

Differential Revision: http://reviews.llvm.org/D13595

llvm-svn: 251800
2015-11-02 14:41:02 +00:00
Nemanja Ivanovic
6e3343082c Fix for bootstrap bug introduced in r244921
This revision has introduced an issue that only affects bootstrapped compiler
when it is printing the ASM. It turns out that the new code path taken due to
legalizing a scalar_to_vector of i64 -> v2i64 exposes a missing check in a
micro optimization to change a load followed by a scalar_to_vector into a
load and splat instruction on PPC.

llvm-svn: 251798
2015-11-02 14:01:11 +00:00
Rafael Espindola
b1fed5af0a This doesn't need a object::Archive::child_iterator.
llvm-svn: 251796
2015-11-02 13:30:46 +00:00
Rafael Espindola
3adda9b6e9 Avoid implicitly constructing a Archive::child_iterator.
llvm-svn: 251794
2015-11-02 13:17:11 +00:00
James Molloy
bcb542b66c [PatternMatch] Switch to use ValueTracking::matchSelectPattern
Instead of rolling our own min/max matching code (which is notoriously
hard to get completely right), use ValueTracking's instead.

llvm-svn: 251785
2015-11-02 09:54:00 +00:00
Pawel Bylica
669b84345b [Support] Extend sys::path with user_cache_directory function.
Summary:
The new function sys::path::user_cache_directory tries to discover
a directory suitable for cache storage for current system user.

On Windows and Darwin it returns a path to system-specific user cache directory.

On Linux it follows XDG Base Directory Specification, what is:
- use non-empty $XDG_CACHE_HOME env var,
- use $HOME/.cache.

Reviewers: chapuni, aaron.ballman, rafael

Subscribers: rafael, aaron.ballman, llvm-commits

Differential Revision: http://reviews.llvm.org/D13801

llvm-svn: 251784
2015-11-02 09:49:17 +00:00
Igor Breger
dd070c17bb AVX512: Implemented encoding and intrinsics for VBROADCASTI32x2 and VBROADCASTF32x2 instructions.
Differential Revision: http://reviews.llvm.org/D14216

llvm-svn: 251781
2015-11-02 07:39:36 +00:00
Craig Topper
0354af2e80 [X86] Remove assertions that check for valid scale values on scatter/gather intrinsics. Nothing upstream prevented illegal values from getting here.
llvm-svn: 251780
2015-11-02 07:24:40 +00:00
Craig Topper
0ba8e1b4bc [X86] Don't pass a scale value of 0 to scatter/gather intrinsics. This causes the code emitter to throw an assertion if we try to encode it. Need to add a check to fail isel for this, but for now avoid testing it.
llvm-svn: 251779
2015-11-02 07:24:37 +00:00
Craig Topper
150b1555c7 [X86] Fold 'if' followed by just an llvm_unreachable into an assert.
llvm-svn: 251778
2015-11-02 07:24:34 +00:00
Craig Topper
df102b96ca [X86] Use isa instead of dyn_cast in a bool context. NFC
llvm-svn: 251777
2015-11-02 07:24:32 +00:00
Craig Topper
cbb29b67b1 [X86] Remove some llvm_unreachables after switches that already have an unreachable in their default case.
llvm-svn: 251776
2015-11-02 07:24:30 +00:00
Craig Topper
055cdb4c83 [X86] Remove a 'break' after an llvm_unreachable.
llvm-svn: 251775
2015-11-02 07:24:27 +00:00
Craig Topper
490cb1be76 [X86] Use cast instead of dyn_cast and a null check marked unreachable.
llvm-svn: 251774
2015-11-02 07:24:25 +00:00
Craig Topper
4f55fa704b Fix a -Wpessimizing-move warning.
llvm-svn: 251773
2015-11-02 05:24:28 +00:00
Craig Topper
edd05abeaa [X86] Use MVT instead of EVT when the type is known to be simple. NFC
llvm-svn: 251772
2015-11-02 05:24:22 +00:00
Xinliang David Li
1ef7153d55 [PGO] Value profiling (index format) code cleanup and testing
1. Added a set of public interfaces in InstrProfRecord
    class to access (read/write) value profile data.
 2. Changed IndexedProfile reader and writer code to 
    use the newly defined interfaces and hide implementation
    details.
 3. Added a couple of unittests for value profiling:
   - Test new interfaces to get and set value profile data
   - Test value profile data merging with various scenarios.

 No functional change is expected. The new interfaces will also
 make it possible to change on-disk format of value prof data
 to be more compact (to be submitted). 

llvm-svn: 251771
2015-11-02 05:08:23 +00:00
Sanjoy Das
f81bfefe0b [SCEV] Fix PR25369
Have `getConstantEvolutionLoopExitValue` work correctly with multiple
entry loops.

As far as I can tell, `getConstantEvolutionLoopExitValue` never did the
right thing for multiple entry loops; and before r249712 it would
silently return an incorrect answer.  r249712 changed SCEV to fail an
assert on a multiple entry loop, and this change fixes the underlying
issue.

llvm-svn: 251770
2015-11-02 02:06:01 +00:00
NAKAMURA Takumi
5ce9110622 Untabify.
llvm-svn: 251769
2015-11-02 01:38:12 +00:00
Davide Italiano
acc570386b [LibraryInfo] Point to FreeBSD HEAD repo and not to a dolphin branch.
The latter might go away (anytime soon).

llvm-svn: 251765
2015-11-01 17:00:13 +00:00
Elena Demikhovsky
f42814b247 AVX-512: Optimized SIMD truncate operations for AVX512F set.
Optimized <8 x i32> to <8 x i16>
<4 x i64> to < 4 x i32>
<16 x i16> to <16 x i8>
All these oprtrations use now AVX512F set (KNL). Before this change it was implemented with AVX2 set.


Differential Revision: http://reviews.llvm.org/D14108

llvm-svn: 251764
2015-11-01 11:45:47 +00:00
Saleem Abdulrasool
5a2aae9d0b RuntimeDyld: add COFF i386 support
This adds support for COFF I386.  This is sufficient for code execution in a
32-bit JIT, though, imported symbols need to custom lowered for the redirection.

llvm-svn: 251761
2015-11-01 01:26:15 +00:00
Rafael Espindola
4411a45b5e Make a few definitions static. NFC.
llvm-svn: 251759
2015-11-01 00:14:59 +00:00
Rafael Espindola
38d8e625e7 Use Child instead of child_iterator in the archive writer.
We never need to pass end(). This will also remove some complication
once we start adding error checking.

llvm-svn: 251758
2015-11-01 00:10:37 +00:00
Davide Italiano
cd97667400 Simplify a check. NFC.
llvm-svn: 251757
2015-11-01 00:09:16 +00:00
Sanjoy Das
367ed95f05 [SCEV] Don't create SCEV expressions that break LCSSA
Prevent `createNodeFromSelectLikePHI` from creating SCEV expressions
that break LCSSA.

A better fix for the same issue is to teach SCEVExpander to not break
LCSSA by inserting PHI nodes at appropriate places.  That's planned for
the future.

Fixes PR25360.

llvm-svn: 251756
2015-10-31 23:21:40 +00:00
Sanjoy Das
76488e1c9e [SCEV] Use auto and range for; NFC
llvm-svn: 251755
2015-10-31 23:21:32 +00:00
Davide Italiano
390134bbdd [SimplifyLibCalls] Factor out other common code.
llvm-svn: 251754
2015-10-31 23:17:45 +00:00
Rafael Espindola
cb48e87760 This can take a const reference. NFC.
llvm-svn: 251753
2015-10-31 22:25:59 +00:00
Diego Novillo
b170ccb76f SamplePGO - Count sample records in embedded profiles when computing coverage.
The initial coverage checking code for sample records failed to count
records inside inlined profiles. This change fixes the oversight.

llvm-svn: 251752
2015-10-31 21:53:58 +00:00
Craig Topper
9f7fa4107c [X86] Replace getScalarType with getVectorElementType when the type is already known to be a vector. This should result in slightly less code. NFC
llvm-svn: 251751
2015-10-31 21:44:52 +00:00
Rafael Espindola
52f24bb108 Don't store a Child to the first regular member.
This is a bit ugly, but has a few advantages:
* Archive is now easy to copy since there is no Archive -> Child -> Archive
  loop.
* It makes it clear that we already checked for errors when finding the Child
  data.

llvm-svn: 251750
2015-10-31 21:44:42 +00:00
Rafael Espindola
d268dd1895 Delete dead code.
llvm-svn: 251749
2015-10-31 21:16:01 +00:00
Rafael Espindola
5f017c0eef Simplify handling of archive Symbol tables.
We only need to store a StringRef.

llvm-svn: 251748
2015-10-31 21:03:29 +00:00
Davide Italiano
572717843f [SimplifyLibCalls] Add test to ensure transform is not executed if fast-math
attribute is not present.

During my refactor in r251595 I changed the behavior of optimizeSqrt(),
skipping the transformation if the function wasn't marked with unsafe-fp-math
attribute. This fixed a bug, as confirmed by Sanjay (before the optimization
was silently executed anyway), although it wasn't my primary aim.
This commit adds a test to ensure the code doesn't break again.

Reported by: Marcello Maggioni
Discussed with: Sanjay Patel

llvm-svn: 251747
2015-10-31 20:59:32 +00:00
Rafael Espindola
f0f9bc5a37 Simplify the handling of the archive string table.
We only need to store a StringRef

llvm-svn: 251746
2015-10-31 20:06:13 +00:00
Craig Topper
bfd831d3f2 [X86] Convert to MVT instead of calling EVT functions since we already know the type is simple. NFC
llvm-svn: 251745
2015-10-31 18:14:17 +00:00
Craig Topper
057ba18026 [X86] Call getScalarSizeInBits() instead of getScalarType().getScalarSizeInBits(). NFC
llvm-svn: 251744
2015-10-31 18:14:15 +00:00
Craig Topper
403a7454ac [X86] Remove two const references to the return value of a constructor and just use normal object creation syntax. NFC
llvm-svn: 251743
2015-10-31 17:28:02 +00:00
Craig Topper
70916f79d8 [X86] Replace EVT with MVT in some more places. NFC
llvm-svn: 251742
2015-10-31 17:27:59 +00:00
Craig Topper
c1776ebb91 [X86] Fix indentation of case statements in switch. NFC
llvm-svn: 251741
2015-10-31 17:27:56 +00:00
Craig Topper
e9bbc5ba7a [X86] Reduce math for index calculation for inserting and extracting subvectors and elements by exploiting the fact that all supported vector types have a power 2 number of elements.
llvm-svn: 251740
2015-10-31 17:27:52 +00:00