d93c668f00
Calls clobber FPSCR.
...
llvm-svn: 80956
2009-09-03 22:12:28 +00:00
9dee94f8fb
Make ImmutableMap/ImmutableSet quicker by only canonicalizing the tree after an
...
Add or Remove operation complete, and not while building the intermediate tree.
This trades a little bit more memory usage for less accesses to the FoldingSet. On a benchmark for the clang static analyzer, this shaves off another 13% of execution time when using field/array sensitivity.
llvm-svn: 80955
2009-09-03 22:07:30 +00:00
d8df76eaeb
Disable some parts of the profiling-tool-chain test, which is currently failing
...
on a self-hosted build (although it seems to work on non-self hosted). I'll work
with Andreas to figure this out.
llvm-svn: 80947
2009-09-03 21:09:53 +00:00
abf2bb683a
Remove dead greps.
...
llvm-svn: 80946
2009-09-03 20:59:02 +00:00
5c7d9ca53d
Overhaul the TwoAddressInstructionPass to simplify the logic, especially
...
for the complicated case where one register is tied to multiple destinations.
This avoids the extra scan of instruction operands that was introduced by
my recent change. I also pulled some code out into a separate
TryInstructionTransform method, added more comments, and renamed some
variables.
Besides all those changes, this takes care of a FIXME in the code regarding
an assumption about there being a single tied use of a register when
converting to a 3-address form. I'm not aware of cases where that assumption
is violated, but the code now only attempts to transform an instruction,
either by commuting its operands or by converting to a 3-address form,
for the simple case where there is a single pair of tied operands.
llvm-svn: 80945
2009-09-03 20:58:42 +00:00
ed8f32022e
Smallvectorize switchExitBlocks.
...
llvm-svn: 80942
2009-09-03 20:36:13 +00:00
794a40dd61
There is not any need to copy metadata while merging modules.
...
llvm-svn: 80941
2009-09-03 20:35:57 +00:00
d0d5e685da
Recognize more opportunities to use SSE min and max instructions,
...
swapping the operands if necessary.
llvm-svn: 80940
2009-09-03 20:34:31 +00:00
eadd21ea3c
Test cases for vector shifts changes r80935
...
Changed the old vector shift test to use FileCheck
llvm-svn: 80936
2009-09-03 19:57:35 +00:00
3e82117210
Fixed a few problems with vector shifts
...
- when transforming a vector shift of a non-immediate scalar shift amount, zero
extend the i32 shift amount to i64 since the vector shift reads 64 bits
- when transforming i16 vectors to use a vector shift, zero extend i16 shift amount
- improve the code quality in some cases when transforming vectors to use a vector shift
llvm-svn: 80935
2009-09-03 19:56:25 +00:00
319cc69f88
Add a -disable-16bit flag and associated support for experimenting with
...
disabling the use of 16-bit operations on x86. This doesn't yet work for
inline asms with 16-bit constraints, vectors with 16-bit elements,
trampoline code, and perhaps other obscurities, but it's enough to try
some experiments.
llvm-svn: 80930
2009-09-03 17:18:51 +00:00
7d91218c75
Removed the non-target independent AsmToken::Register enum constant
...
from MCAsmLexer.h in preparation of supporting other targets. Changed the
X86AsmParser code to reflect this by removing AsmLexer::LexPercent and looking
for AsmToken::Percent when parsing in places that used AsmToken::Register.
Then changed X86ATTAsmParser::ParseRegister to parse out registers as an
AsmToken::Percent followed by an AsmToken::Identifier.
llvm-svn: 80929
2009-09-03 17:15:07 +00:00
85dac4771f
Use WeakVH to hold dead mdnodes. Check use_empty() before deleting a node.
...
llvm-svn: 80928
2009-09-03 17:03:47 +00:00
728a81ab18
Make bugpoint use ParseIRFile instead of doing the same thing manually.
...
llvm-svn: 80927
2009-09-03 16:32:58 +00:00
4c1bdcf5d7
Add a verifyAnalysis to LoopInfo, LoopSimplify, and LCSSA form that verify
...
that these passes are properly preserved.
Fix several transformation passes that claimed to preserve LoopSimplify
form but weren't.
llvm-svn: 80926
2009-09-03 16:31:42 +00:00
a7326b5ba5
Remove some unnecessary -f options.
...
llvm-svn: 80924
2009-09-03 16:11:53 +00:00
3a0ce3e639
Move getUniqueExitBlocks from LoopBase to Loop, since they depend on
...
LoopSimplify form, which is currently only available on Loops (and
not MachineLoops). Also, move the code out of the header file.
llvm-svn: 80923
2009-09-03 16:10:48 +00:00
0054b460ff
Use IRReader.h in opt, to support reading of LLVM Assembly files directly.
...
llvm-svn: 80922
2009-09-03 16:00:08 +00:00
22571485b3
Change PHINode::hasConstantValue to have a DominatorTree argument
...
instead of a bool argument, and to do the dominator check itself.
This makes it eaiser to use when DominatorTree information is
available.
llvm-svn: 80920
2009-09-03 15:34:35 +00:00
96a26bd35a
Don't try to verify a LoopPass analysis if the loop has been deleted.
...
llvm-svn: 80919
2009-09-03 15:09:24 +00:00
237d9e5213
Remove references to expression "handles", which are no longer used.
...
llvm-svn: 80918
2009-09-03 15:00:26 +00:00
13d048fdff
CppBackend: avoid printing unnecessary whitespace.
...
llvm-svn: 80917
2009-09-03 14:58:24 +00:00
0edc7100ba
Keep track of how many memmove calls were turned into
...
memcpy calls.
llvm-svn: 80915
2009-09-03 13:37:16 +00:00
d194e13b8d
Fix build warning.
...
llvm-svn: 80912
2009-09-03 09:11:10 +00:00
7e86c3856b
Code Cleanup.
...
Removed inverted flag form MaximumSpanningTree, also do not handle so much
information to MaximumSpanningTree.
llvm-svn: 80911
2009-09-03 08:52:52 +00:00
213f8f4860
Filter out -fno-rtti from CXXFLAGS as well (in an expensive checks build).
...
llvm-svn: 80910
2009-09-03 08:41:19 +00:00
b68e921c25
Code Cleanup.
...
(See http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090831/086139.html )
llvm-svn: 80909
2009-09-03 08:41:05 +00:00
f563ac919b
Reapply profiling tests.
...
llvm-svn: 80908
2009-09-03 07:38:00 +00:00
0dc21e4713
Remove undefined behavior when loading optimal edge profile info.
...
llvm-svn: 80907
2009-09-03 07:37:42 +00:00
51a0712538
don't call getOffset() on jump tables, this fixes three failing olden benchmarks
...
with the new asmprinter.
llvm-svn: 80906
2009-09-03 07:36:42 +00:00
c7b00730ad
Implement support for X86II::MO_GOT_ABSOLUTE_ADDRESS. We get very
...
different formatting from the old asmprinter, but it should be
semantically the same. We used to get:
popl %eax
addl $_GLOBAL_OFFSET_TABLE_ + [.-.Lllvm$6.$piclabel], %eax
...
Now we get:
popl %eax
.Lpicbaseref6:
addl $(_GLOBAL_OFFSET_TABLE_ + (.Lpicbaseref6 - .Lllvm$6.$piclabel)), %eax
...
llvm-svn: 80905
2009-09-03 07:30:56 +00:00
1b38952c99
Reference to hidden symbols do not have to go through non-lazy pointer in non-pic mode. rdar://7187172.
...
llvm-svn: 80904
2009-09-03 07:04:02 +00:00
88214fbd12
Remove VISIBILITY_HIDDEN from this file.
...
llvm-svn: 80903
2009-09-03 06:43:15 +00:00
cdb6fd2c7c
merge all the basic linux/32 pic tests together into one test.
...
llvm-svn: 80902
2009-09-03 06:29:23 +00:00
4f101f98d1
rename test
...
llvm-svn: 80901
2009-09-03 06:16:49 +00:00
438380aa4d
use a darwin triple
...
llvm-svn: 80900
2009-09-03 06:15:11 +00:00
41bf56de79
TAI -> MAI
...
llvm-svn: 80899
2009-09-03 06:13:54 +00:00
17b3a93ad0
adjust expected lines.
...
llvm-svn: 80898
2009-09-03 06:13:45 +00:00
cdeb8d147d
improve comments.
...
llvm-svn: 80897
2009-09-03 06:00:00 +00:00
1945453843
fix MCSymbol printing to exactly match the normal mangler rules so
...
we can diff .s files.
llvm-svn: 80894
2009-09-03 05:57:47 +00:00
7a53b96394
remove extraneous hack.
...
llvm-svn: 80893
2009-09-03 05:54:00 +00:00
c3a0aba120
Make these functions static and local.
...
llvm-svn: 80892
2009-09-03 05:47:34 +00:00
76628def06
Tweak comment.
...
llvm-svn: 80891
2009-09-03 05:47:22 +00:00
f4366a3998
Thread an MCAsmInfo pointer through the various MC printing APIs,
...
and fix a few things using << on MCSymbols to use ->print(). No
functionality change other than unbreaking my previous patch.
llvm-svn: 80890
2009-09-03 05:46:51 +00:00
1a67fe8862
just use dump()
...
llvm-svn: 80889
2009-09-03 05:39:09 +00:00
1800dff44e
inline insertion operators.
...
llvm-svn: 80888
2009-09-03 05:33:01 +00:00
27266f164f
In C++, code is not allowed to call main. In C it is, this
...
simplifylibcalls optimization is thus valid for C++ but not C.
It's not important enough to worry about for C++ apps, so just
remove it.
rdar://7191924
llvm-svn: 80887
2009-09-03 05:19:59 +00:00
954b9cd41f
merge globaladdress symbol processing stuff into other stuff. Now
...
all global variable operand flag processing stuff is shared between
different operand types.
llvm-svn: 80886
2009-09-03 05:06:07 +00:00
155deabbcb
Unbreak x86_64 build.
...
llvm-svn: 80885
2009-09-03 05:01:00 +00:00
6370d5606c
Split the "operand -> symbol" logic from the "get offset and other munging
...
from operand" logic. GlobalAddress still todo.
llvm-svn: 80884
2009-09-03 04:56:20 +00:00
5daf61910e
implement lowering support for constant pool index operands, this gets a bunch more
...
olden programs working.
llvm-svn: 80881
2009-09-03 04:44:53 +00:00
47455a79ae
X86JITInfo::getLazyResolverFunction() should not read cpu id to determine whether sse is available. Just use consult subtarget.
...
No functionality changes.
llvm-svn: 80880
2009-09-03 04:37:05 +00:00
231dd7169d
Set the 'cached digest' flag after computing the digest for an
...
ImutAVLTree. This was accidentally left out, and essentially caused
digest caching to be ignored in ImmutableMap and ImmutableSet (this
bug was detected from shark traces that showed ComputeDigest was in
the hot path in the clang static analyzer).
This reduces the running time of the clang static analyzer on an
example benchmark by ~32% for both RegionStore (field-sensitivty) and
BasicStore (without field-sensitivity).
llvm-svn: 80877
2009-09-03 04:21:34 +00:00
11c989cc47
update test for alignment value in hex
...
llvm-svn: 80876
2009-09-03 04:03:44 +00:00
f16a122213
output alignment value in hex so that we get:
...
.align 3, 0x90
instead of,
.align 3, 144
suggested by eric.
llvm-svn: 80875
2009-09-03 04:01:10 +00:00
620713c5b4
simplify this by using SmallString::str(), much nicer!
...
llvm-svn: 80874
2009-09-03 03:54:02 +00:00
0b3720b3c3
Fixed a test that ensures the LocalRewriter does not attempt to
...
avoid reloads by reusing clobbered registers.
This was causing issues in 256.bzip2 when compiled with PIC for
a while (starting at r78217), though the problem has since been masked.
llvm-svn: 80872
2009-09-03 02:52:02 +00:00
2d60e1ec0c
back out my recent commit (r80858), it seems to break self-hosting buildbot's stage 2 configure
...
llvm-svn: 80871
2009-09-03 02:02:59 +00:00
f7188325ef
Now Bitcode reader bug is fixed. Reapply 80839.
...
Use CallbackVH, instead of WeakVH, to hold MDNode elements.
Use FoldingSetNode to unique MDNodes in a context.
Use CallbackVH hooks to update context's MDNodeSet appropriately.
llvm-svn: 80868
2009-09-03 01:39:20 +00:00
116b4a0cb3
Add new value for given index in MDValuePtrs.
...
llvm-svn: 80867
2009-09-03 01:38:02 +00:00
18d6959f39
Improve llvm::getHostTriple for some cases where the LLVM_HOSTTRIPLE is not
...
reliable.
llvm-svn: 80863
2009-09-03 01:10:13 +00:00
14dfba6d66
re-commit r66920 (which has been backed out in r66953) I may have more luck this time. I'll back out if needed...
...
llvm-svn: 80858
2009-09-03 00:18:58 +00:00
f400442927
Added opaque 32-, 48-, and 80-bit memory operand types to the X86
...
instruction tables to support segmented addressing (and other objects
of obscure type).
Modified the X86 assembly printers to handle these new operand types.
Added JMP and CALL instructions that use segmented addresses.
llvm-svn: 80857
2009-09-03 00:04:47 +00:00
fc39586b4f
Show derived host triple in --version.
...
llvm-svn: 80855
2009-09-02 23:52:38 +00:00
96995e8179
Improve support for cross-hosted builds of LLVM.
...
--build=triple and other configure options are passed
to the BuildTools/ sub-invocation more consistently
llvm-svn: 80854
2009-09-02 23:52:23 +00:00
ef7c1fd909
Unbreak my CMake build. Say you'll link again.
...
llvm-svn: 80842
2009-09-02 22:45:31 +00:00
c180029be4
Revert 80839 for now. It causes test failures.
...
llvm-svn: 80841
2009-09-02 21:49:26 +00:00
36309cd391
Use CallbackVH, instead of WeakVH, to hold MDNode elements.
...
Use FoldingSetNode to unique MDNodes in a context.
Use CallbackVH hooks to update context's MDNodeSet appropriately.
llvm-svn: 80839
2009-09-02 21:22:09 +00:00
f0da41c3e4
More missed vdup patterns
...
llvm-svn: 80838
2009-09-02 21:21:28 +00:00
c76bfb777e
Switch llc from ParseBitcodeFile to ParseIRFile. This lets llc
...
transparently read either LLVM Assembly or LLVM Bitcode files.
llvm-svn: 80829
2009-09-02 19:35:19 +00:00
a7e8799c72
Add a comment noting the memory ownership rules.
...
llvm-svn: 80827
2009-09-02 19:21:56 +00:00
b7cb99bf50
Add convenience functions for reading in LLVM IR that autodetect
...
and LLVM Assembly and LLVM Bitcode and automatically call the
corresponding reader.
llvm-svn: 80809
2009-09-02 17:54:06 +00:00
7bd37b47c8
switch from std::string to SmallString + raw_svector_ostream.
...
llvm-svn: 80807
2009-09-02 17:37:38 +00:00
74f4ca7cb2
split mcinst lowering stuff out to its own file.
...
llvm-svn: 80806
2009-09-02 17:35:12 +00:00
957b1316e6
Fix the syntax of add/sub/mul nsw/nuw and sdiv exact.
...
llvm-svn: 80805
2009-09-02 17:31:42 +00:00
ae58e72018
Add const qualifiers for isBitcodeWrapper, and add new functions
...
isRawBitcode and isBitcode to allow clients to test whether a given
memory buffer holds a bitcode image.
llvm-svn: 80804
2009-09-02 17:21:29 +00:00
77ac99dc31
Refactor common code from ParseAssemblyString and ParseAssemblyFile,
...
to expose a low-level interface for parsing from an existing MemoryBuffer.
llvm-svn: 80803
2009-09-02 17:18:19 +00:00
87d69e80b1
Add const qualifiers to dominates' arguments.
...
llvm-svn: 80801
2009-09-02 17:05:05 +00:00
69e2afe030
Removed temporarily because of breaking Darwin builds.
...
(See http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090831/086214.html )
llvm-svn: 80799
2009-09-02 16:47:24 +00:00
bc751989f1
Rearrange code to eliminate redundancy and avoid gotos.
...
llvm-svn: 80798
2009-09-02 16:35:35 +00:00
2baa6a38d2
plug another leak in LLParser::PerFunctionState::SetInstName()
...
llvm-svn: 80792
2009-09-02 15:02:57 +00:00
da6b0fa8ed
Changed profiling-tool-chain.ll test to use optimal-edge-profiling instead of
...
edge-profiling, this is more useful since the loading of the
optimal-edge-profiling is more complicated.
The edge-profiling is tested in edge-profiling.ll where only the
instrumentation is tested.
llvm-svn: 80791
2009-09-02 14:24:08 +00:00
2fcee70aeb
plug memory leak in LLParser::PerFunctionState::SetInstName() by deleting a value after replacing it
...
llvm-svn: 80790
2009-09-02 14:22:03 +00:00
ae866b0c66
Sort edges in MaximumSpanningTree more stable in case of equal weight.
...
(See http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090824/085890.html )
llvm-svn: 80789
2009-09-02 14:03:11 +00:00
964fa2bdac
Changed set of BlocksToInstrument to set of InsertedBlocks that do not have to
...
be instrumented.
llvm-svn: 80788
2009-09-02 13:59:05 +00:00
4469c164d0
Code cleanups and added comments.
...
llvm-svn: 80781
2009-09-02 12:38:39 +00:00
82131d6ad7
Opaque types didn't work if llvm_is_multithreaded().
...
AlwaysOpaqueTy is always NULL at this point, and it causes an assertion failure.
Fix it by using the just constructed tmp instead.
llvm-svn: 80780
2009-09-02 12:23:05 +00:00
18e03dd70c
Fix DbgStopPointInst->getFileName/getDirectory, broken by the MDNodification in
...
r80406, and readd a -print-dbginfo test.
llvm-svn: 80778
2009-09-02 11:13:56 +00:00
68c5f477fa
Retype from unsigned to CallingConv::ID accordingly. Approved by Bob Wilson.
...
llvm-svn: 80773
2009-09-02 08:44:58 +00:00
defbbc9105
Fix month.
...
llvm-svn: 80769
2009-09-02 06:34:22 +00:00
4916267c97
fix PR4815: some cases where DeleteDeadInstruction can delete
...
the instruction BBI points to.
llvm-svn: 80768
2009-09-02 06:31:02 +00:00
09a79dcfdf
clean up this code a bit.
...
llvm-svn: 80767
2009-09-02 06:15:37 +00:00
2dd09dbdf7
eliminate VISIBILITY_HIDDEN from Transforms/Scalar. PR4861
...
llvm-svn: 80766
2009-09-02 06:11:42 +00:00
d5f4fcceae
refactor select 'sched insertion' out to its own method.
...
llvm-svn: 80764
2009-09-02 05:57:00 +00:00
cc8c581a5b
Add support for modeling whether or not the processor has support for
...
conditional moves as a subtarget feature. This is the easy part of
PR4841.
llvm-svn: 80763
2009-09-02 05:53:04 +00:00
950e60ea97
fix PR4848 an infinite loop when indexing down through a recursive gep
...
and we get the original pointer type. This doesn't mean that we're
at the first pointer being indexed. Correct the predicate.
llvm-svn: 80762
2009-09-02 05:35:45 +00:00
64b5842986
fix PR4837, some bugs folding vector compares. These
...
return a vector of i1, not i1 itself.
llvm-svn: 80761
2009-09-02 05:12:37 +00:00
65fb597793
revert my patch, duncan points out what is wrong with my logic. Add
...
a comment so that I don't change this in the future :)
llvm-svn: 80760
2009-09-02 04:39:04 +00:00
8f23276431
one more try at making this simpler, hopefully it won't break everything :)
...
llvm-svn: 80759
2009-09-02 04:34:06 +00:00
5632d96176
Complicate Chris's simplification, avoiding complaints
...
about singular iterators when building with expensive
checks turned on.
llvm-svn: 80757
2009-09-02 03:48:41 +00:00
David Goodwin