1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-31 07:52:55 +01:00
Commit Graph

88805 Commits

Author SHA1 Message Date
Edwin Vane
f5f16d3262 Fix gcc/printf/ISO C++ warning
Remove the use of the 't' length modifier to avoid a gcc warning. Based
on usage, 32 bits of precision is good enough for printing a stack
offset for a stack trace.

't' length modifier isn't in C++03 but it *is* in C++11. Added a FIXME
to reintroduce once LLVM makes the switch to C++11.

Reviewer: gribozavr
llvm-svn: 173711
2013-01-28 19:34:42 +00:00
Bill Schmidt
28daa2bbf4 This patch addresses bug 15031.
The common code in the post-RA scheduler to break anti-dependencies on the
critical path contained a flaw.  In the reported case, an anti-dependency
between the overlapping registers %X4 and %R4 exists:

	%X29<def> = OR8 %X4, %X4
	%R4<def>, %X3<def,dead,tied3> = LBZU 1, %X3<kill,tied1>

The unpatched code breaks the dependency by replacing %R4 and its uses
with %R3, the first register on the available list.  However, %R3 and
%X3 overlap, so this creates two overlapping definitions on the same
instruction.

The fix is straightforward, preventing selection of a register that
overlaps any other defined register on the same instruction.

The test case is reduced from the bug report, and verifies that we no
longer produce "lbzu 3, 1(3)" when breaking this anti-dependency.

llvm-svn: 173706
2013-01-28 18:36:58 +00:00
Eric Christopher
a8f811de13 Fix comment.
llvm-svn: 173698
2013-01-28 17:33:26 +00:00
Renato Golin
7fe847b274 Vectorization Factor clarification
llvm-svn: 173691
2013-01-28 16:02:45 +00:00
Evgeniy Stepanov
b2af9f793f [msan] Mostly disable msan-handle-icmp-exact.
It is way too slow. Change the default option value to 0.
Always do exact shadow propagation for unsigned ICmp with constants, it is
cheap (under 1% cpu time) and required for correctness.

llvm-svn: 173682
2013-01-28 11:42:28 +00:00
Evgeniy Stepanov
3008b317bf Revert r173678.
Broken tests.

llvm-svn: 173679
2013-01-28 09:18:40 +00:00
Evgeniy Stepanov
01bbba2ef8 [msan] Make msan-handle-icmp-exact=0 by default.
50% slowdown on one of the specs.

llvm-svn: 173678
2013-01-28 09:15:15 +00:00
Craig Topper
8c275cea97 Fix 256-bit PALIGNR comment decoding to understand that it works on independent 256-bit lanes.
llvm-svn: 173674
2013-01-28 07:41:18 +00:00
Nadav Rotem
0229e2b96e A bugfix for tblgen, in the function ‘emitSourceFileHeader’.
When the first parameter (‘Desc’) is more than 80 characters long, it will result the header line that contains the description to be more
Than (4GB!) long. Not only it takes forever to produce, the output file cannot be open, since its ginormous.

Patch by Elior Malul.

llvm-svn: 173672
2013-01-28 07:35:33 +00:00
Craig Topper
e60d6ab3f8 Add missing break in 256-bit palignr comment printing. No test case yet because the comment itself is still wrong.
llvm-svn: 173669
2013-01-28 07:19:11 +00:00
Craig Topper
97391f52d3 Fix inconsistent usage of PALIGN and PALIGNR when referring to the same instruction.
llvm-svn: 173667
2013-01-28 06:48:25 +00:00
Michael Gottesman
998e1da974 Created ObjCARCUtil.cpp for functions which in my humble opinion are too large to static inline and place in a header file such as ObjCARC.h.
llvm-svn: 173666
2013-01-28 06:39:31 +00:00
Craig Topper
493bcfb933 Remove addToNoHelperNeeded function that was left unused after r173649. Fixes a -Wunused warning.
llvm-svn: 173664
2013-01-28 06:09:24 +00:00
Michael Gottesman
ad3bed037a Cleaned up includes in various ObjCARC files and removed some whitespace violations.
llvm-svn: 173663
2013-01-28 05:51:58 +00:00
Michael Gottesman
86d4759cc7 Refactor ObjCARCAliasAnalysis into its own file.
llvm-svn: 173662
2013-01-28 05:51:54 +00:00
Bill Wendling
fbc4fffb48 Mid-air collision. reapply r173656.
llvm-svn: 173661
2013-01-28 05:51:40 +00:00
Bill Wendling
29a48a09e5 Rewrite the removeAttr() method.
This now uses the AttributeSet object instead of the Attribute /
AttributeWithIndex objects. It's fairly simple now. It goes through all of the
subsets before the one we're modifying, adds them to the new set. It then adds
the modified subset (with the requested attributes removed). And then adds the
rest of the subsets.

llvm-svn: 173660
2013-01-28 05:44:14 +00:00
Bill Wendling
bc011af4a0 Rewrite the addAttr() method.
This now uses the AttributeSet object instead of the Attribute /
AttributeWithIndex objects. It's fairly simple now. It goes through all of the
subsets before the one we're modifying, adds them to the new set. It then adds
the modified subset. And then adds the rest of the subsets.

llvm-svn: 173659
2013-01-28 05:23:28 +00:00
NAKAMURA Takumi
e02fd3448b Revert r173646, "Use proper type for the index."
Unfortunately, msvc miscompiles it. Investigating.

llvm-svn: 173656
2013-01-28 04:29:01 +00:00
Michael Gottesman
a9c5dbd2a3 Refactored out pass ObjCARCAPElim from ObjCARCOpts.cpp => ObjCARCAPElim.cpp.
llvm-svn: 173654
2013-01-28 04:12:07 +00:00
Michael Gottesman
a793d420e6 Fixed case insensitive issue.
llvm-svn: 173653
2013-01-28 03:35:20 +00:00
Michael Gottesman
bddf287573 Removed extraneous doxygen end module statement.
llvm-svn: 173652
2013-01-28 03:30:34 +00:00
Michael Gottesman
4dcab23d5a Extracted pass ObjCARCExpand from ObjCARC.cpp => ObjCARCExpand.cpp.
I also added the local header ObjCARC.h for common functions used by the
various passes.

llvm-svn: 173651
2013-01-28 03:28:38 +00:00
Edwin Vane
69f1a085e0 Improved svn repo searching for 'make update'
Use a simple recursive bash function to search for svn repos for the 'make
update' target thus including projects like clang-tools-extra.

Reviewers: bkramer, echristo
llvm-svn: 173650
2013-01-28 03:19:57 +00:00
Reed Kotler
99a57ba847 Make some code a little simpler.
llvm-svn: 173649
2013-01-28 02:46:49 +00:00
Michael Gottesman
3d8ed99b1f Extracted ObjCARC.cpp into its own library libLLVMObjCARCOpts in preparation for refactoring the ARC Optimizer.
llvm-svn: 173647
2013-01-28 01:35:51 +00:00
Bill Wendling
5534b9814d Use proper type for the index.
llvm-svn: 173646
2013-01-28 01:30:29 +00:00
Bill Wendling
d3dbdfbd8e Remove another use of AttributeWithIndex, using the AttributeSetImpl accessors instead.
llvm-svn: 173644
2013-01-28 01:11:42 +00:00
Bill Wendling
46c2463ec4 Remove another use of AttributeWithIndex, using the AttributeSetImpl accessors instead.
llvm-svn: 173642
2013-01-28 00:46:02 +00:00
Bill Wendling
7de8bf67ce Remove a use of AttributeWithIndex.
We want to remove AttributeWithIndex because it provides a non-encapsulated view
of the AttributeSetImpl object. Instead, use accessor methods and iterators.

Eventually, this code can be simplified because the Attribute object will hold
only one attribute instead of multiple attributes.

llvm-svn: 173641
2013-01-28 00:21:34 +00:00
Bill Wendling
ebbee6a7f6 Improve the debug output a bit.
llvm-svn: 173640
2013-01-27 23:53:56 +00:00
Bill Wendling
5783e1e4be Use proper return type for attribute index.
llvm-svn: 173639
2013-01-27 23:50:44 +00:00
Bill Wendling
e21ffcc86f Use proper return type for attribute index.
llvm-svn: 173638
2013-01-27 23:49:44 +00:00
Bill Wendling
dd604e3b22 Push the calculation of the 'Raw' attribute mask down into the implementation. It in turn uses the correct list for calculating the 'Raw' value.
llvm-svn: 173637
2013-01-27 23:41:29 +00:00
Bill Wendling
fefacc5936 Don't erase these methods. They're used during testing.
llvm-svn: 173636
2013-01-27 22:46:17 +00:00
Bill Wendling
89179c02f8 Add special 'get' methods to create an Attribute with an alignment. Also do some random cleanup. No functionality change.
llvm-svn: 173635
2013-01-27 22:43:04 +00:00
Richard Osborne
373bdcfeb5 [XCore] Add missing l2rus instructions.
These instructions are not targeted by the compiler but they are
needed for the MC layer.

llvm-svn: 173634
2013-01-27 22:28:30 +00:00
Bill Wendling
c1f9afeb84 Privitize some the copy c'tor and assignment operator of uniquified objects.
llvm-svn: 173632
2013-01-27 21:38:03 +00:00
Bill Wendling
4f0a1bba14 Add some helpful comments.
llvm-svn: 173631
2013-01-27 21:32:11 +00:00
Richard Osborne
72f31f9da6 [XCore] Add missing l2r instructions.
These instructions are not targeted by the compiler but they are
needed for the MC layer.

llvm-svn: 173629
2013-01-27 21:26:02 +00:00
Bill Wendling
bc0a92c4bd s/AttrList/pImpl/g in AttributeSet. No functionality change.
llvm-svn: 173628
2013-01-27 21:23:46 +00:00
Bill Wendling
2b431a6fb5 Start using more of the AttrNode in the AttributeSetImpl class.
Also add some asserts.

llvm-svn: 173627
2013-01-27 21:20:06 +00:00
Benjamin Kramer
6107b802a9 Legalizer: Reword comment again, per Duncan's suggestion.
llvm-svn: 173625
2013-01-27 21:02:52 +00:00
Richard Osborne
70183ce54a [XCore] Add missing 1r instructions.
These instructions are not targeted by the compiler but they are
needed for the MC layer.

llvm-svn: 173624
2013-01-27 20:46:21 +00:00
Richard Osborne
be996a5261 [XCore] Add missing 0r instructions.
These instructions are not targeted by the compiler but they are
needed for the MC layer.

llvm-svn: 173623
2013-01-27 20:42:57 +00:00
Hal Finkel
115cffae92 BBVectorize: Better use of TTI->getShuffleCost
When flipping the pair of subvectors that form a vector, if the
vector length is 2, we can use the SK_Reverse shuffle kind to get
more-accurate cost information. Also we can use the SK_ExtractSubvector
shuffle kind to get accurate subvector extraction costs.

The current cost model implementations don't yet seem complex enough
for this to make a difference (thus, there are no test cases with this
commit), but it should help in future.

Depending on how the various targets optimize and combine shuffles in
practice, we might be able to get more-accurate costs by combining the
costs of multiple shuffle kinds. For example, the cost of flipping the
subvector pairs could be modeled as two extractions and two subvector
insertions. These changes, however, should probably be motivated
by specific test cases.

llvm-svn: 173621
2013-01-27 20:07:01 +00:00
Benjamin Kramer
1e8d008612 Legalizer: Add an assert and tweak a comment to clarify the assumptions this code makes.
llvm-svn: 173620
2013-01-27 15:04:43 +00:00
Bill Wendling
a298631a83 In the AttributeSetImpl c'tor, fill in the AttrNodes data structure with the attributes being passed in.
llvm-svn: 173618
2013-01-27 12:50:02 +00:00
NAKAMURA Takumi
aff27067cb [CMake][Lit][unittests] Cleanup in AddLLVM.cmake.
llvm-svn: 173617
2013-01-27 12:46:53 +00:00
NAKAMURA Takumi
311d4c23b3 [CMake][Lit][unittests] Deprecate CMAKE_BUILD_TYPE in each build directory for unittests.
For example,
cur) unittests/ADT/Release/ADTTests
new) unittests/ADT/ADTTests

RUNTIME_BUILD_MODE can be substituted to CMAKE_CFG_INTDIR.

With Make and Ninja, the tree is not built with multiple configurations.
Then, including the build type in target directory doesn't make sense.
See also "How can I build multiple modes without switching?"
http://www.cmake.org/Wiki/CMake_FAQ
CMAKE_CFG_INTDIR is set to "."

With multiple-configuration-aware build system, like Visual Studio, each unittest is built on appropriate directory, for example,
unittests/ADT/Release/ADTTests.exe
CMAKE_CFG_INTDIR is set to build system's variable, like "$(Configuration)" or "$(OutDir)".

Thus, "--param build_config" is also deprecated.

llvm-svn: 173616
2013-01-27 12:20:50 +00:00