aa598b3be6
misched: Recompute priority queue when DFSResults are updated.
...
This was found by MSVC10's STL debug mode on a test from the test suite. Sadly
std::is_heap isn't standard so there is no way to assert this without writing
our own heap verify, which looks like overkill to me.
llvm-svn: 168885
2012-11-29 14:36:26 +00:00
7ad7e83031
[msan] Basic handling of inline asm.
...
llvm-svn: 168884
2012-11-29 14:32:03 +00:00
0ac49bf846
Teach the legalizer how to handle operands for VSELECT nodes
...
If we need to split the operand of a VSELECT, it must be the mask operand. We
split the entire VSELECT operand with EXTRACT_SUBVECTOR.
llvm-svn: 168883
2012-11-29 14:26:28 +00:00
bc45119b44
Allow targets to prefer TypeSplitVector over TypePromoteInteger when computing the legalization method for vectors
...
For some targets, it is desirable to prefer scalarizing <N x i1> instead of promoting to a larger legal type, such as <N x i32>.
llvm-svn: 168882
2012-11-29 14:26:24 +00:00
857d9d2a59
[msan] Propagate shadow through (x<0) and (x>=0) comparisons.
...
This is a special case of signed relational comparison where result
only depends on the sign of x.
llvm-svn: 168881
2012-11-29 14:25:47 +00:00
eeb8b7c391
[msan] Fix shadow & origin store & load alignment.
...
This change ensures that shadow memory accesses have the same alignment
as corresponding app memory accesses.
llvm-svn: 168880
2012-11-29 14:05:53 +00:00
62ba611828
[msan] Optimize getOriginPtr.
...
Rewrite getOriginPtr in a way that lets subsequent optimizations factor out
the common part of Shadow and Origin address calculation. Improves perf by
up to 5%.
llvm-svn: 168879
2012-11-29 13:43:05 +00:00
da0072b676
[msan] Fix a few compilation warnings.
...
llvm-svn: 168878
2012-11-29 13:12:03 +00:00
c3b579e5ac
[msan] Add a test for r168873.
...
llvm-svn: 168877
2012-11-29 13:11:09 +00:00
62b5db9361
[msan] Transform memcpy and memset to library calls.
...
This was already done for memmove, where it is required for correctness.
This change improves performance by avoiding copyingthe same memory twice.
Also, the library functions are given __msan_ prefix to prevent instcombine
pass from converting them back to intrinsics.
llvm-svn: 168876
2012-11-29 12:49:04 +00:00
eace43bff7
I changed hasAVX() to hasFp256() and hasAVX2() to hasInt256() in X86IselLowering.cpp.
...
The logic was not changed, only names.
llvm-svn: 168875
2012-11-29 12:44:59 +00:00
4f9240d7e3
[msan] Update tests (broken in r168873).
...
llvm-svn: 168874
2012-11-29 12:43:56 +00:00
1d2da65bf8
[msan] Make sure that report callbacks do not get merged.
...
llvm-svn: 168873
2012-11-29 12:30:18 +00:00
dde9dcc24b
[asan] disable BuiltinLongJmpTest on Android
...
llvm-svn: 168872
2012-11-29 12:18:48 +00:00
194d3685ec
Documentation: use correct highlighter
...
llvm-svn: 168871
2012-11-29 12:00:32 +00:00
c2b2b75bd5
ASTTests/StmtPrinterTest/StmtPrinter.TestMSIntegerLiteral: Remove i128 stuff. Conditioning-out in macro argument was not accepted on MS cl.exe.
...
llvm-svn: 168867
2012-11-29 10:22:40 +00:00
d4bd7b73e3
Initial commit of MemorySanitizer.
...
Compiler pass only.
llvm-svn: 168866
2012-11-29 09:57:20 +00:00
b2df62570e
ASTTests/StmtPrinterTest/StmtPrinter.TestMSIntegerLiteral: Suppress i128 according to r168856, for now.
...
I think "i128", that I conditioned out, could be completely removed.
MS Compiler doesn't accept i128. We can assume no one would use i128.
llvm-svn: 168865
2012-11-29 09:57:11 +00:00
4b929dae93
[asan/tsan] initialize the asan/tsan callbacks in runOnFunction as opposed to doInitialization. This is required to allow the upcoming changes in PassManager behavior
...
llvm-svn: 168864
2012-11-29 09:54:21 +00:00
e73a480970
[asan] enable BuiltinLongJmpTest
...
llvm-svn: 168863
2012-11-29 09:02:14 +00:00
6732d350ef
[-cxx-abi microsoft] Also spill the argument-back-references context when mangling templates
...
llvm-svn: 168862
2012-11-29 08:58:47 +00:00
633bf93fb8
[asan] when checking the noreturn attribute on the call, also check it on the callee
...
llvm-svn: 168861
2012-11-29 08:57:20 +00:00
99507e6c10
ubsan: Don't assume that Clang provides __int128 unless it advertises that it does.
...
llvm-svn: 168857
2012-11-29 05:59:30 +00:00
e6a56db2e6
Reject uses of __int128 on platforms that don't support it. Also move the ugly
...
'getPointerWidth(0) >= 64' test to be a method on TargetInfo, ready to be
properly cleaned up.
llvm-svn: 168856
2012-11-29 05:41:51 +00:00
6be9b25cb6
Fix sentence construction-o.
...
llvm-svn: 168855
2012-11-29 05:29:23 +00:00
4eb7ee566a
Cleaned up a couple of comments.
...
llvm-svn: 168854
2012-11-29 04:30:52 +00:00
e1f363c724
Remove 'noreturn' attribute from friend declaration. This attribute will be
...
inherited from the previous out-of-class declaration, and attributes on friend
function declarations are ill-formed in C++11.
llvm-svn: 168853
2012-11-29 04:30:50 +00:00
bdb55e0c59
Use MCPhysReg for RegisterClassInfo allocation orders.
...
This saves a bit of memory.
llvm-svn: 168852
2012-11-29 03:34:17 +00:00
fd41aee2f8
Fix crash-on-invalid. <rdar://problem/12765391>.
...
llvm-svn: 168851
2012-11-29 03:13:49 +00:00
7afe1663e9
Add an MCPhysReg typedef to replace naked uint16_t.
...
Use this type for arrays of physical registers.
llvm-svn: 168850
2012-11-29 02:39:28 +00:00
01ab5d718b
Instruction::isAssociative() returns true for fmul/fadd if they are tagged "unsafe" mode.
...
Approved by: Eli and Michael.
llvm-svn: 168848
2012-11-29 01:47:31 +00:00
8bf22e5b52
The declaration of a special member can require overload resolution to be
...
performed, to determine whether that special member is deleted or constexpr.
That overload resolution process can in turn trigger the instantiation of a
template, which can do anything, including triggering the declaration of that
very same special member function. When this happens, do not try to recursively
declare the special member -- that's impossible. Instead, only try to realise
the truth. There is no special member.
llvm-svn: 168847
2012-11-29 01:34:07 +00:00
ad6e5b262a
Change SValBuilder::getConditionType() to return BoolTy in C++. Fixes <rdar://problem/12772656>.
...
llvm-svn: 168846
2012-11-29 01:03:10 +00:00
75a0333bf8
<rdar://problem/12445557>
...
Make stack frames fix up their line table entries when the target has source remappings. Also rearranged how the m_sc.target_sp was filled in so it can be used during the StackFrame::GetSymbolContext(...) function.
llvm-svn: 168845
2012-11-29 00:53:06 +00:00
ea09a44ceb
Don't return decorated (i.e., const or pointer)
...
versions of UnknownAnyTy for ObjectiveC value types.
<unknown type>* makes no sense and can cause the
parser to behave very oddly.
<rdar://problem/12518999>
llvm-svn: 168844
2012-11-29 00:50:56 +00:00
2317f30f4d
Correctly handle IntegralToBool casts in C++ in the static analyzer. Fixes <rdar://problem/12759044>.
...
llvm-svn: 168843
2012-11-29 00:50:20 +00:00
9ae53f8e3a
[driver] -mkernel implies -mstrict-align; don't add the redundant option.
...
rdar://12771737
llvm-svn: 168841
2012-11-29 00:42:06 +00:00
8c3f2764fb
Add an example of using the target.process.extra-startup-command to turn on debugserver logging since
...
we always forget how to do this...
llvm-svn: 168840
2012-11-29 00:41:12 +00:00
7730b9a47a
Tighten up how we acquire the underlying frame in the SBFrame methods. We were getting
...
the frame and then getting the run lock. Which means that our frame could have gotten
invalidated by stopping between the time we got the frame and assured the the target was
stopped. Now we get the run lock first, and THEN resolve the underlying frame object.
<rdar://problem/12621607>
llvm-svn: 168838
2012-11-29 00:26:19 +00:00
546e9e85f1
Avoid rewriting instructions twice.
...
This could cause miscompilations in targets where sub-register
composition is not always idempotent (ARM).
<rdar://problem/12758887>
llvm-svn: 168837
2012-11-29 00:26:11 +00:00
e812109868
Rename __lambda to __lambda_node (apply r164404 from libcxxabi)
...
- fixes gcc 4.6 build problems
- resolves open bugzilla http://llvm.org/bugs/show_bug.cgi?id=13889
llvm-svn: 168835
2012-11-29 00:05:50 +00:00
b8e5c79af1
Correct the label for the com.apple.debugserver.posix plist.
...
<rdar://problem/12769073>
llvm-svn: 168834
2012-11-29 00:01:49 +00:00
1af94eb075
Issue a fatal error if the line doesn't have a regular expression.
...
Also a couple not-user-visible changes; using empty() instead of size(), and
make inSection() not insert NULL Regex*'s into StringMap when doing a lookup.
llvm-svn: 168833
2012-11-29 00:01:38 +00:00
307d767177
When combining consecutive stores allow loads in between the stores, if the loads do not alias.
...
llvm-svn: 168832
2012-11-29 00:00:08 +00:00
16e126c70b
Linux buildbot fix: detect swig tool from PATH in shell script (before searching hardcoded directories)
...
llvm-svn: 168831
2012-11-28 23:49:11 +00:00
f3614fd8e2
When we delete a dead basic block, see if any of its successors are dead and
...
delete those as well.
llvm-svn: 168829
2012-11-28 23:23:48 +00:00
9bbc8da517
Add 'class' keyword to friend definition (to fix gcc 4.6 build)
...
llvm-svn: 168827
2012-11-28 23:20:22 +00:00
ddc016d4d7
Implement C++11 [dcl.attr.grammar] p4: If an attribute-specifier-seq appertains to a friend declaration, that declaration shall be a definition.
...
llvm-svn: 168826
2012-11-28 23:17:40 +00:00
998f0a3360
objective-C blocks: Make sure that identical logic is used
...
in deciding a copy/dispose field is needed in a byref structure
and when generating the copy/dispose helpers. In certain
cases, these fields were being added but no copy/dispose was
being generated. This was uncovered in ARC, but not in MRR.
// rdar://12759433
llvm-svn: 168825
2012-11-28 23:12:17 +00:00
a118c6a8c0
Per C++11 [except.spec]p2, rvalue references are not permitted in exception specifications.
...
llvm-svn: 168824
2012-11-28 22:52:42 +00:00
Benjamin Kramer