598f8ff9e5
Reprioritize tests for tail duplication to be aggressive about indirect
...
branches even when optimizing for code size. Unless we find evidence to the
contrary in the future, the special treatment for indirect branches does not
have a significant effect on code size, and performance still matters with -Os.
llvm-svn: 90147
2009-11-30 18:56:45 +00:00
89b660c774
Fix a crash when ivar type is a __strong SEL. Fallout from
...
recent change to make SEL a builtin type (fixes radar 7425510).
llvm-svn: 90145
2009-11-30 18:43:52 +00:00
505ddaa4dc
Remove isProfitableToDuplicateIndirectBranch target hook. It is profitable
...
for all the processors where I have tried it, and even when it might not help
performance, the cost is quite low. The opportunities for duplicating
indirect branches are limited by other factors so code size does not change
much due to tail duplicating indirect branches aggressively.
llvm-svn: 90144
2009-11-30 18:35:03 +00:00
7bd1048f26
Add more sophisticated test for destruction order of C++ temporaries, please
...
feel free to extend!
llvm-svn: 90143
2009-11-30 18:17:34 +00:00
c168a52627
Fix some more ARM unified syntax warnings.
...
llvm-svn: 90141
2009-11-30 17:47:19 +00:00
5f089128e9
Use StringRef in Attr constructors.
...
llvm-svn: 90140
2009-11-30 17:08:26 +00:00
693ba203a1
Eliminate another VISIBILITY_HIDDEN
...
llvm-svn: 90139
2009-11-30 16:08:24 +00:00
7da76a20c0
Fix odd declaration.
...
llvm-svn: 90138
2009-11-30 15:52:29 +00:00
9fc223a6b1
Adapt to the DOTGraphTraits changes in LLVM.
...
llvm-svn: 90137
2009-11-30 14:16:05 +00:00
9caf3801ca
Fix last DOTGraphTraits problems in CompilationGraph.
...
llvm-svn: 90136
2009-11-30 13:34:51 +00:00
0729c6e281
Remove forgotten ShortNames in Trie and CompilationGraph
...
llvm-svn: 90135
2009-11-30 13:14:13 +00:00
dd7f2e797f
Remove ShortNames from getNodeLabel in DOTGraphTraits
...
llvm-svn: 90134
2009-11-30 12:38:47 +00:00
90d334032a
Instantiate DefaultDOTGraphTraits
...
llvm-svn: 90133
2009-11-30 12:38:13 +00:00
8d6f6b74b1
Do not point edge heads to source labels
...
If no destination label is available, just point to the node itself
instead of pointing to some source label. Source and destination labels are
not related in any way.
llvm-svn: 90132
2009-11-30 12:37:39 +00:00
f444599ef0
Only print edgeSourceLabels if they are not empty
...
Graphviz can layout the graphs better if a node does not contain source
ports. Therefore only print the ports if the source ports are useful,
that means are not labeled with the empty string "".
This patch also simplifies graphs without any edgeSourceLabels e.g. the
dominance trees.
llvm-svn: 90131
2009-11-30 12:24:40 +00:00
f7495f497a
Small PostDominatorTree improvements
...
* Do not SEGFAULT if tree entryNode() is NULL
* Print function names in dotty printer
llvm-svn: 90130
2009-11-30 12:06:37 +00:00
5f0988ceb0
Remove ":" after BB name in -view-cfg-only
...
llvm-svn: 90129
2009-11-30 11:55:24 +00:00
bd74028095
clang-cc: Add code (currently disabled) for parsing arguments using clang -cc1 functionality instead of LLVM's CommandLine.
...
- It works and passes test suite, but I want to polish and test a wee bit more before making the switch.
llvm-svn: 90128
2009-11-30 08:42:10 +00:00
a74f8ff15c
Add CodeGenOptions::{SoftFloat,FloatABI}, and update the all the (far too many) places to use this instead of using the backend -soft-float and -float-abi= options.
...
llvm-svn: 90127
2009-11-30 08:42:00 +00:00
91dea8252a
clang -cc1: Fix initialization of PreprocessorOptions::TokenCache.
...
llvm-svn: 90126
2009-11-30 08:41:42 +00:00
05762b1344
Add TextDiagnosticBuffer::FlushDiagnostics, for forwarding the buffered diagnostics to a different diagnostics engine.
...
llvm-svn: 90125
2009-11-30 08:41:34 +00:00
8c317acd2b
clang -cc1: Fix -include-pch now that the test code is in clang-cc.
...
llvm-svn: 90124
2009-11-30 08:41:22 +00:00
599abb2d56
clang -cc1: Initialize LangOptions::{Optimize,NoInline}
...
llvm-svn: 90123
2009-11-30 08:41:13 +00:00
7c71b6b95b
Use '-FOO' 'BAR' instead of '-FOO=BAR' in tests.
...
llvm-svn: 90122
2009-11-30 08:41:04 +00:00
84bb793718
Always pass -fmessage-length using separate arguments.
...
llvm-svn: 90121
2009-11-30 08:40:54 +00:00
0429c5ef3a
clang -cc1: Initialize LangOptions::DollarIdents correctly.
...
llvm-svn: 90120
2009-11-30 08:40:46 +00:00
2c11cd156e
Don't try to generate common globals for C++ files, instead of depending on the FE to set NoCommon, and simplify CodeGenOptions initialization.
...
llvm-svn: 90119
2009-11-30 08:40:34 +00:00
8e70505692
Eliminate CodeGenOptions::TimePasses.
...
llvm-svn: 90118
2009-11-30 08:39:52 +00:00
de18224d9c
Eliminate CodeGenOptions::SimplifyLibCalls.
...
llvm-svn: 90117
2009-11-30 08:39:32 +00:00
7348432ade
First pass at llvm.objectsize documentation.
...
llvm-svn: 90116
2009-11-30 08:03:53 +00:00
fe994279ad
Simplify.
...
llvm-svn: 90115
2009-11-30 07:18:20 +00:00
1c39f3c89d
Put CompilerInvocation testing code in clang-cc instead of clang for now, I can't bear to link all of clang into 'clang' yet. :)
...
- Still triggered by a magic leading -cc1 argument, as before.
llvm-svn: 90114
2009-11-30 07:18:13 +00:00
2d32947099
Revert r90107, fixing test/Transforms/GVN/2009-11-29-ReverseMap.ll and the
...
llvm-gcc build.
llvm-svn: 90113
2009-11-30 07:05:51 +00:00
8a29dd4c7f
Add a testcase for the current llvm-gcc build failure.
...
llvm-svn: 90112
2009-11-30 07:02:18 +00:00
1c2720320b
Remove the 'simple jit' tutorial as it wasn't really being maintained and its
...
material is covered by the Kaleidoscope tutorial.
llvm-svn: 90111
2009-11-30 04:23:17 +00:00
710bb87147
Fix PR5633 by making the preprocessor handle the case where we can
...
stat a file but where mmaping it fails. In this case, we emit an
error like:
t.c:1:10: fatal error: error opening file '../../foo.h'
instead of "cannot find file".
llvm-svn: 90110
2009-11-30 04:18:44 +00:00
031cb00246
Add test case for r90108
...
llvm-svn: 90109
2009-11-30 02:42:27 +00:00
32f8bb9ed4
Added support to allow clients to custom widen. For X86, custom widen vectors for
...
divide/remainder since these operations can trap by unroll them and adding undefs
for the resulting vector.
llvm-svn: 90108
2009-11-30 02:42:02 +00:00
4d252d20e8
reapply r90093 with an addition of keeping the forward
...
and reverse nonlocal memdep maps in synch, this should
fix 255.vortex.
llvm-svn: 90107
2009-11-30 02:26:29 +00:00
fef0c67d01
Fix this test on 64-bit systems which seem to use i64 for gep indices sometimes
...
while 32-bit gcc uses i32.
llvm-svn: 90106
2009-11-30 02:23:57 +00:00
03aa2f1262
Minor cleanup (no functionality change).
...
llvm-svn: 90105
2009-11-30 01:19:33 +00:00
95ef6c9560
Commit r90099 made LLVM simplify one of these constant expressions a little
...
more. Update the syntax we're checking for and filecheckize it too.
This will fix the selfhost buildbots but will 'break' the others (sigh) because
they're still linked against older LLVM which is emitting less optimized IR.
llvm-svn: 90104
2009-11-30 00:38:56 +00:00
0ac66427a9
clang -cc1: Use proper diagnostics for all parsing errors.
...
llvm-svn: 90100
2009-11-29 21:52:53 +00:00
e35e6f097d
Teach ConstantFolding to do a better job when folding gep(bitcast).
...
This permits the devirtualization of llvm.org/PR3100#c9 when compiled by clang.
llvm-svn: 90099
2009-11-29 21:40:55 +00:00
c76493a65d
Don't pass false (default) for isVolatile parameter to CreateLoad.
...
llvm-svn: 90098
2009-11-29 21:23:36 +00:00
3efc050ac4
Revert r90089 for now, it's breaking selfhost.
...
llvm-svn: 90097
2009-11-29 21:17:48 +00:00
0311ade94c
revert this patch for now, it causes failures of:
...
LLVM::Transforms/GVN/2009-02-17-LoadPRECrash.ll
LLVM::Transforms/GVN/2009-06-17-InvalidPRE.ll
llvm-svn: 90096
2009-11-29 21:14:59 +00:00
1dd0164bbe
Explicitly use setVolatile instead of extra IRBuilder argument.
...
llvm-svn: 90095
2009-11-29 21:11:47 +00:00
dacbe6bdcd
Don't pass false (default) for isVolatile parameter to CreateStore.
...
llvm-svn: 90094
2009-11-29 21:11:41 +00:00
52e7715b0b
Fix a really nasty caching bug I introduced in memdep. An entry
...
was being added to the Result vector, but not being put in the
cache. This means that if the cache was reused wholesale for a
later query that it would be missing this entry and we'd do an
incorrect load elimination.
Unfortunately, it's not really possible to write a useful
testcase for this, but this unbreaks 255.vortex.
llvm-svn: 90093
2009-11-29 21:09:36 +00:00
Bob Wilson