Jakob Stoklund Olesen [Sat, 19 May 2012 05:25:50 +0000 (05:25 +0000)]
Run proper recursive dead code elimination during coalescing.
Dead copies cause problems because they are trivial to coalesce, but
removing them gived the live range a dangling end point. This patch
enables full dead code elimination which trims live ranges to their uses
so end points don't dangle.
DCE may erase multiple instructions. Put the pointers in an ErasedInstrs
set so we never risk visiting erased instructions in the work list.
There isn't supposed to be any dead copies entering RegisterCoalescer,
but they do slip by as evidenced by test/CodeGen/X86/coalescer-dce.ll.
llvm-svn: 157101
Jakob Stoklund Olesen [Sat, 19 May 2012 05:25:46 +0000 (05:25 +0000)]
Allow LiveRangeEdit to be created with a NULL parent.
The dead code elimination with callbacks is still useful.
llvm-svn: 157100
Douglas Gregor [Sat, 19 May 2012 04:41:25 +0000 (04:41 +0000)]
Move _LIBCPP_VISIBLE_INLINE from the out-of-line definitions of member
functions to the original declarations, so that Clang will actually
see them. Part of <rdar://problem/
11489333>.
llvm-svn: 157097
Sean Callanan [Sat, 19 May 2012 02:36:49 +0000 (02:36 +0000)]
Updated LLVM to take a disassembler fix that causes
the LOCK prefix to be printed explicitly when it's
the first prefix on the instruction.
llvm-svn: 157096
Eric Christopher [Sat, 19 May 2012 01:36:50 +0000 (01:36 +0000)]
Update API usage for llvm DIBuilder changes for rvalue reference
types and ensure we are actually creating the type.
rdar://
11479676
llvm-svn: 157095
Eric Christopher [Sat, 19 May 2012 01:36:37 +0000 (01:36 +0000)]
Actually support DW_TAG_rvalue_reference_type that we were trying
to generate out of the front end.
rdar://
11479676
llvm-svn: 157094
Eric Christopher [Sat, 19 May 2012 00:51:56 +0000 (00:51 +0000)]
Add support for the 'd' mips inline asm output modifier.
Patch by Jack Carter.
llvm-svn: 157093
Andrew Trick [Sat, 19 May 2012 00:48:25 +0000 (00:48 +0000)]
SCEV: Add MarkPendingLoopPredicates to avoid recursive isImpliedCond.
getUDivExpr attempts to simplify by checking for overflow.
isLoopEntryGuardedByCond then evaluates the loop predicate which
may lead to the same getUDivExpr causing endless recursion.
Fixes PR12868: clang 3.2 segmentation fault.
llvm-svn: 157092
Jason Molenda [Sat, 19 May 2012 00:32:19 +0000 (00:32 +0000)]
bump to lldb-150.
llvm-svn: 157090
Anna Zaks [Sat, 19 May 2012 00:22:11 +0000 (00:22 +0000)]
[analyzer] c++11: do not crash on namespace alias
llvm-svn: 157089
Anna Zaks [Sat, 19 May 2012 00:22:07 +0000 (00:22 +0000)]
[analyzer] For locations, use isGLValue() instead of isLValue().
llvm-svn: 157088
Greg Clayton [Sat, 19 May 2012 00:18:00 +0000 (00:18 +0000)]
Forgot to bump the local string buffer up in size after debugging to make sure long strings would be correctly read when the buffer is too small for the string.
llvm-svn: 157087
Kaelyn Uhrain [Fri, 18 May 2012 23:42:49 +0000 (23:42 +0000)]
Suggest adding 'typename' when it would make the compiler
accept the template argument expression as a type.
llvm-svn: 157085
Johnny Chen [Fri, 18 May 2012 23:30:57 +0000 (23:30 +0000)]
Add test cases where we start three worker threads, with a write watchpoint set. As soon as a watchpoint is hit in either of the worker thread,
we delete the watchpoint. The test succeeds when no more watchpoint hit event fires after the deletion of the watchpoint.
related to rdar://problem/
11320188
llvm-svn: 157084
Greg Clayton [Fri, 18 May 2012 23:20:01 +0000 (23:20 +0000)]
Found a quick way to improve the speed with which we can read object files from memory when they are in the shared cache: always read the symbol table strings from memory and let the process' memory cache do the work.
llvm-svn: 157083
Anna Zaks [Fri, 18 May 2012 22:47:43 +0000 (22:47 +0000)]
[analyzer] Fix a c++11 crash: xvalues can be locations (VisitMemberExpr)
llvm-svn: 157082
Anna Zaks [Fri, 18 May 2012 22:47:40 +0000 (22:47 +0000)]
[analyzer] Malloc checker: remove unnecessary comparisons.
llvm-svn: 157081
Dan Gohman [Fri, 18 May 2012 22:17:29 +0000 (22:17 +0000)]
Fix replacing all the users of objc weak runtime routines
when deleting them. rdar://
11434915.
llvm-svn: 157080
Jakob Stoklund Olesen [Fri, 18 May 2012 22:10:15 +0000 (22:10 +0000)]
Modernize naming convention for class members.
No functional change.
llvm-svn: 157079
Greg Clayton [Fri, 18 May 2012 21:47:20 +0000 (21:47 +0000)]
Warn when we detect a valid dSYM file that is empty. This can happen when a dSYM file is created from a binary with no debug info, that has been stripped, or when the .o files are not available when the dSYM is created.
llvm-svn: 157078
Filipe Cabecinhas [Fri, 18 May 2012 21:35:43 +0000 (21:35 +0000)]
Cleanup a custom thread-format so we don't fail TestLoadUnload when running the
test suite for two architectures (the full path to d.c would appear).
llvm-svn: 157077
Fariborz Jahanian [Fri, 18 May 2012 21:22:49 +0000 (21:22 +0000)]
objc: use "class extension" instead of "continuation class"
to match documentation. // rdar://
11309706
llvm-svn: 157074
Jakob Stoklund Olesen [Fri, 18 May 2012 21:09:40 +0000 (21:09 +0000)]
Move all work list processing to copyCoalesceWorkList().
This will make it possible to filter out erased instructions later.
llvm-svn: 157073
Nuno Lopes [Fri, 18 May 2012 21:02:10 +0000 (21:02 +0000)]
allow LazyValueInfo::getEdgeValue() to reason about multiple edges from the same switch instruction by doing union of ranges (which may still be conservative, but it's more aggressive than before)
llvm-svn: 157071
Enrico Granata [Fri, 18 May 2012 19:55:37 +0000 (19:55 +0000)]
Optimizations to the Python ObjC formatters - Benchmarking infrastructure
llvm-svn: 157066
Benjamin Kramer [Fri, 18 May 2012 19:32:16 +0000 (19:32 +0000)]
Lexer::ReadToEndOfLine: Only build the string if it's actually used and do so in a less malloc-intensive way.
llvm-svn: 157064
Jim Grosbach [Fri, 18 May 2012 19:12:01 +0000 (19:12 +0000)]
Refactor data-in-code annotations.
Use a dedicated MachO load command to annotate data-in-code regions.
This is the same format the linker produces for final executable images,
allowing consistency of representation and use of introspection tools
for both object and executable files.
Data-in-code regions are annotated via ".data_region"/".end_data_region"
directive pairs, with an optional region type.
data_region_directive := ".data_region" { region_type }
region_type := "jt8" | "jt16" | "jt32" | "jta32"
end_data_region_directive := ".end_data_region"
The previous handling of ARM-style "$d.*" labels was broken and has
been removed. Specifically, it didn't handle ARM vs. Thumb mode when
marking the end of the section.
rdar://
11459456
llvm-svn: 157062
Nick Kledzik [Fri, 18 May 2012 18:39:06 +0000 (18:39 +0000)]
fix warnings when compiling with -Wshadow
llvm-svn: 157061
Eric Christopher [Fri, 18 May 2012 18:24:15 +0000 (18:24 +0000)]
Remove duplicate code that we could just fallthrough to.
llvm-svn: 157060
Jakob Stoklund Olesen [Fri, 18 May 2012 18:21:48 +0000 (18:21 +0000)]
Simplify RegisterCoalescer::copyCoalesceInMBB().
It is no longer necessary to separate VirtCopies, PhysCopies, and
ImpDefCopies. Implicitly defined copies are extremely rare after we
added the ProcessImplicitDefs pass, and physical register copies are not
joined any longer.
llvm-svn: 157059
Nuno Lopes [Fri, 18 May 2012 17:44:58 +0000 (17:44 +0000)]
add test case for bugfix in r157032
llvm-svn: 157058
Eric Christopher [Fri, 18 May 2012 17:39:35 +0000 (17:39 +0000)]
Add support for the mips 'x' inline asm modifier.
Patch by Jack Carter.
llvm-svn: 157057
Jakob Stoklund Olesen [Fri, 18 May 2012 17:18:58 +0000 (17:18 +0000)]
Remove support for PhysReg joining.
This has been disabled for a while, and it is not a feature we want to
support. Copies between physical and virtual registers are eliminated by
good hinting support in the register allocator. Joining virtual and
physical registers is really a form of register allocation, and the
coalescer is not properly equipped to do that. In particular, it cannot
backtrack coalescing decisions, and sometimes that would cause it to
create programs that were impossible to register allocate, by exhausting
a small register class.
It was also very difficult to keep track of the live ranges of aliasing
registers when extending the live range of a physreg. By disabling
physreg joining, we can let fixed physreg live ranges remain constant
throughout the register allocator super-pass.
One type of physreg joining remains: A virtual register that has a
single value which is a copy of a reserved register can be merged into
the reserved physreg. This always lowers register pressure, and since we
don't compute live ranges for reserved registers, there are no problems
with aliases.
llvm-svn: 157055
Chad Rosier [Fri, 18 May 2012 16:56:52 +0000 (16:56 +0000)]
Typo.
llvm-svn: 157054
Joel Jones [Fri, 18 May 2012 16:24:01 +0000 (16:24 +0000)]
FileCheck-ify, apropos of nothing
llvm-svn: 157051
Filipe Cabecinhas [Fri, 18 May 2012 13:21:05 +0000 (13:21 +0000)]
Fixed the test for the new process launch abbreviation.
llvm-svn: 157050
Howard Hinnant [Fri, 18 May 2012 13:06:21 +0000 (13:06 +0000)]
Protect __shared_weak_count::__get_deleter declaration with _LIBCPP_NO_RTTI. Fixes llvm.org/bugs/show_bug.cgi?id=12867
llvm-svn: 157049
Dmitry Vyukov [Fri, 18 May 2012 09:41:52 +0000 (09:41 +0000)]
tsan: check for overflow in malloc()
llvm-svn: 157048
Dmitry Vyukov [Fri, 18 May 2012 08:53:16 +0000 (08:53 +0000)]
tsan: fix potential NULL deref
llvm-svn: 157047
Stepan Dyatkovskiy [Fri, 18 May 2012 08:32:28 +0000 (08:32 +0000)]
Recommited reworked r156804:
SelectionDAGBuilder::Clusterify : main functinality was replaced with CRSBuilder::optimize, so big part of Clusterify's code was reduced.
llvm-svn: 157046
Craig Topper [Fri, 18 May 2012 07:07:36 +0000 (07:07 +0000)]
Simplify code a bit. No functional change intended.
llvm-svn: 157044
Craig Topper [Fri, 18 May 2012 06:42:06 +0000 (06:42 +0000)]
Simplify handling of v16i8 shuffles and fix a missed optimization.
llvm-svn: 157043
Greg Clayton [Fri, 18 May 2012 02:38:05 +0000 (02:38 +0000)]
<rdar://problem/
11386214>
<rdar://problem/
11455913>
"target symbol add" should flush the cached frames
"register write" should flush the thread state in case registers modifications change stack
llvm-svn: 157042
Jason Molenda [Fri, 18 May 2012 02:03:26 +0000 (02:03 +0000)]
bump to lldb 149.
llvm-svn: 157040
Rafael Espindola [Fri, 18 May 2012 01:47:00 +0000 (01:47 +0000)]
Centralize the handling of the "attribute declaration must precede definition"
warning. This also makes us warn on tags, which, ironically, is the only case
gcc warns on.
llvm-svn: 157039
Evan Cheng [Fri, 18 May 2012 01:33:51 +0000 (01:33 +0000)]
Teach two-address pass to update the "source" map so it doesn't perform a
non-profitable commute using outdated info. The test case would still fail
because of poor pre-RA schedule. That will be fixed by MI scheduler.
rdar://
11472010
llvm-svn: 157038
Anna Zaks [Fri, 18 May 2012 01:16:10 +0000 (01:16 +0000)]
[analyzer]Malloc: refactor and report use after free by memory
allocating functions.
llvm-svn: 157037
Johnny Chen [Fri, 18 May 2012 00:51:36 +0000 (00:51 +0000)]
rdar://problem/
11140741
For "process attach", make the success criterion as the inferior changes its state to eStateStopped.
Otherwise, mark it as a failure and say so.
llvm-svn: 157036
Danil Malyshev [Fri, 18 May 2012 00:30:58 +0000 (00:30 +0000)]
Temporarily disabled the MCJIT tests for Darwin, because the RuntimeDyldMachO has a problems with relocations for 32bit x86.
llvm-svn: 157035
Fariborz Jahanian [Fri, 18 May 2012 00:19:25 +0000 (00:19 +0000)]
Another test for r157025 <rdar://problem/
11460990>.
llvm-svn: 157034
Eric Christopher [Fri, 18 May 2012 00:16:22 +0000 (00:16 +0000)]
Clarify comment.
llvm-svn: 157033
Nuno Lopes [Fri, 18 May 2012 00:14:36 +0000 (00:14 +0000)]
fix corner case in ConstantRange::intersectWith().
this fixes the missed optimization I was seeing in the CorrelatedValuePropagation pass
llvm-svn: 157032
Kevin Enderby [Fri, 18 May 2012 00:13:56 +0000 (00:13 +0000)]
Fixed a bug in llvm-objdump when disassembling using -macho option for a binary
containing no symbols. Fixed the crash and fixed it not disassembling anything.
llvm-svn: 157031
Jakob Stoklund Olesen [Fri, 18 May 2012 00:07:14 +0000 (00:07 +0000)]
Remove a test that was only testing for physreg joining.
This is the same as the other tests: Clever tricks are required to make
the arguments and return value line up in a single-instruction function.
It rarely happens in real life.
We have plenty other examples of this behavior.
llvm-svn: 157030
Jim Ingham [Fri, 18 May 2012 00:05:52 +0000 (00:05 +0000)]
Make the debug output that comes as printf's from code called in the target for getting ObjC class names and ObjC method implementations only come out when doing verbose logging.
llvm-svn: 157029
Greg Clayton [Fri, 18 May 2012 00:04:38 +0000 (00:04 +0000)]
<rdar://problem/
11405850>
The "run" and "r" aliases were for gdb compatability, so make then do what GDB does by default: launch in a shell.
For those that don't want launching with a shell by default, add the following to your ~/.lldbinit file:
command unalias run
command unalias r
command alias r process launch --
command alias run process launch --
llvm-svn: 157028
Jakob Stoklund Olesen [Thu, 17 May 2012 23:44:19 +0000 (23:44 +0000)]
Remove -join-physregs from the test suite.
This option has been disabled for a while, and it is going away so I can
clean up the coalescer code.
The tests that required physreg joining to be enabled were almost all of
the form "tiny function with interference between arguments and return
value". Such functions are usually inlined in the real world.
The problem exposed by phys_subreg_coalesce-3.ll is real, but fairly
rare.
llvm-svn: 157027
Sean Callanan [Thu, 17 May 2012 23:29:56 +0000 (23:29 +0000)]
I have updated Clang to include support for Objective-C
boxed expressions returning numbers and strings.
I also added boxed expressions to our testcases, and
enabled boxed expressions when libarclite is linked into
the inferior.
llvm-svn: 157026
Douglas Gregor [Thu, 17 May 2012 23:13:29 +0000 (23:13 +0000)]
A selector match between two Objective-C methods does *not* guarantee
that the methods have the same number of parameters, although we
certainly assumed this in many places. Objective-C can be insane
sometimes. Fixes <rdar://problem/
11460990>.
llvm-svn: 157025
Nuno Lopes [Thu, 17 May 2012 23:04:08 +0000 (23:04 +0000)]
minor simplification in the call to ConstantRange constructor
llvm-svn: 157024
Matt Beaumont-Gay [Thu, 17 May 2012 22:46:56 +0000 (22:46 +0000)]
Remove trailing periods which snuck into a few diagnostic strings.
llvm-svn: 157022
Douglas Gregor [Thu, 17 May 2012 22:39:14 +0000 (22:39 +0000)]
In the override search for Objective-C methods, protect against ASTs that have NULL interfaces behind a category, which can happen in invalid code. Fixes <rdar://problem/
11478173>, a recent regression
llvm-svn: 157021
Andrew Trick [Thu, 17 May 2012 22:37:09 +0000 (22:37 +0000)]
comments
llvm-svn: 157020
Kevin Enderby [Thu, 17 May 2012 22:18:01 +0000 (22:18 +0000)]
Fix the encoding of the armv7m (MClass) for MSR APSR writes which was missing
the 0b10 mask encoding bits. Make MSR APSR writes without a _<bits> qualifier
an alias for MSR APSR_nzcvq even though ARM as deprecated it use. Also add
support for suffixes (_nzcvq, _g, _nzcvqg) for APSR versions. Some FIXMEs in
the code for better error checking when versions shouldn't be used.
rdar://
11457025
llvm-svn: 157019
Benjamin Kramer [Thu, 17 May 2012 21:14:26 +0000 (21:14 +0000)]
Drop the question mark when complaining about a non-existing -Wno- flag.
We're certain that it doesn't exist.
llvm-svn: 157017
Danil Malyshev [Thu, 17 May 2012 21:07:47 +0000 (21:07 +0000)]
- Added ExecutionEngine/MCJIT tests
- Added HOST_ARCH to Makefile.config.in
The HOST_ARCH will be used by MCJIT tests filter, because MCJIT supported only x86 and ARM architectures now.
llvm-svn: 157015
Chad Rosier [Thu, 17 May 2012 20:53:37 +0000 (20:53 +0000)]
Add -mno-implicit-float to the M group.
llvm-svn: 157014
Han Ming Ong [Thu, 17 May 2012 20:41:48 +0000 (20:41 +0000)]
<rdar://
11477301>
Restore Xcode as a valid white list client using the XPC root launcher
llvm-svn: 157012
Bill Wendling [Thu, 17 May 2012 20:27:58 +0000 (20:27 +0000)]
Remove extraneous ';'.
llvm-svn: 157011
Jim Ingham [Thu, 17 May 2012 18:51:37 +0000 (18:51 +0000)]
Reduce the timeout value for the "get class name" and "po" functions to .1 second. 1 second (what they were before) is way too long.
llvm-svn: 157009
Jim Ingham [Thu, 17 May 2012 18:38:42 +0000 (18:38 +0000)]
If we notice that a module with a given file path is replaced by another with the same file
path on rerunning, evict the old module from the target module list, inform the breakpoints
about this so they can do something intelligent as well.
rdar://problem/
11273043
llvm-svn: 157008
Andrew Trick [Thu, 17 May 2012 18:35:13 +0000 (18:35 +0000)]
misched: trace ReadyQ.
llvm-svn: 157007
Andrew Trick [Thu, 17 May 2012 18:35:10 +0000 (18:35 +0000)]
misched: Added 3-level regpressure back-off.
Introduce the basic strategy for register pressure scheduling.
1) Respect target limits at all times.
2) Indentify critical register classes (pressure sets).
Track pressure within the scheduled region.
Avoid increasing scheduled pressure for critical registers.
3) Avoid exceeding the max pressure of the region prior to scheduling.
Added logic for picking between the top and bottom ready Q's based on
regpressure heuristics.
Status: functional but needs to be asjusted to achieve good results.
llvm-svn: 157006
Andrew Trick [Thu, 17 May 2012 18:35:07 +0000 (18:35 +0000)]
comment
llvm-svn: 157005
Andrew Trick [Thu, 17 May 2012 18:35:05 +0000 (18:35 +0000)]
regpressure: Fix getMaxUpwardPressureDelta.
llvm-svn: 157004
Andrew Trick [Thu, 17 May 2012 18:35:03 +0000 (18:35 +0000)]
misched: fix liveness iterators
llvm-svn: 157003
Andrew Trick [Thu, 17 May 2012 18:35:00 +0000 (18:35 +0000)]
whitespace
llvm-svn: 157002
Jakob Stoklund Olesen [Thu, 17 May 2012 18:32:42 +0000 (18:32 +0000)]
Never clear <undef> flags on already joined copies.
RegisterCoalescer set <undef> flags on all operands of copy instructions
that are scheduled to be removed. This is so they won't affect
shrinkToUses() by introducing false register reads.
Make sure those <undef> flags are never cleared, or shrinkToUses() could
cause live intervals to end at instructions about to be deleted.
This would be a lot simpler if RegisterCoalescer could just erase joined
copies immediately instead of keeping all the to-be-deleted instructions
around.
This fixes PR12862. Unfortunately, bugpoint can't create a sane test
case for this. Like many other coalescer problems, this failure depends
of a very fragile series of events.
<rdar://problem/
11474428>
llvm-svn: 157001
Jakob Stoklund Olesen [Thu, 17 May 2012 18:32:40 +0000 (18:32 +0000)]
Fix a verifier bug.
Make sure useless (def-only) intervals also get verified.
llvm-svn: 157000
Argyrios Kyrtzidis [Thu, 17 May 2012 18:16:05 +0000 (18:16 +0000)]
[arcmt] Remove the "it is not safe to remove an unused 'autorelease' message" ARC
migration error.
This is more trouble that it is worth; autoreleasing a value without holding on it
is a valid use-case, we should not "punish" correct code for the minority of
broken/fragile programs that depend on the behavior of -autorelease.
rdar://9914061
llvm-svn: 156999
Bill Wendling [Thu, 17 May 2012 17:59:51 +0000 (17:59 +0000)]
Relax the requirement that the exception object must be an instruction. During
bugpoint-ing, it may turn into something else.
llvm-svn: 156998
Evandro Menezes [Thu, 17 May 2012 16:46:46 +0000 (16:46 +0000)]
[Hexagon] Clean up Hexagon ELF definition.
llvm-svn: 156996
Chris Lattner [Thu, 17 May 2012 15:55:41 +0000 (15:55 +0000)]
enhance the intrinsic info stuff to emit encodings that don't fit in 32-bits into a
separate side table, using the handy SequenceToOffsetTable class. This encodes all
these weird things into another 256 bytes, allowing all intrinsics to be encoded this way.
llvm-svn: 156995
Filipe Cabecinhas [Thu, 17 May 2012 15:48:02 +0000 (15:48 +0000)]
We shouldn't save g_dummy_target_sp. Other code will simply call Destroy() on it.
TestBackticksWithoutATarget.BackticksWithNoTargetTestCase was calling
GetDummyTarget() when executing for x86_64. When performing session
tearDown, it would get destroyed (and everything would be invalid (arch,
etc).
Then the test would run for i386. The dummy target wasn't being
reinitialized and was invalid. lldb complained that 'current process state
is unsuitable for expression parsing'.
llvm-svn: 156994
Chad Rosier [Thu, 17 May 2012 15:45:13 +0000 (15:45 +0000)]
[driver] Reword the warning message for missing value to a joined argument.
llvm-svn: 156993
Dmitry Vyukov [Thu, 17 May 2012 15:00:27 +0000 (15:00 +0000)]
tsan: remove shutdown code
tsan runtime shutdown is problematic for 2 reasons:
1. others crash during shutdown
2. we have to override user exit status (don't know it and can't return from atexit handler)
llvm-svn: 156991
Dmitry Vyukov [Thu, 17 May 2012 14:17:51 +0000 (14:17 +0000)]
tsan: detect accesses to freed memory
http://codereview.appspot.com/6214052
llvm-svn: 156990
Tim Northover [Thu, 17 May 2012 13:12:13 +0000 (13:12 +0000)]
Remove incorrect pattern for ARM SMML instruction.
Patch by Meador Inge.
llvm-svn: 156989
Abramo Bagnara [Thu, 17 May 2012 12:44:05 +0000 (12:44 +0000)]
Added two missing const qualifiers.
llvm-svn: 156988
Benjamin Kramer [Thu, 17 May 2012 12:01:52 +0000 (12:01 +0000)]
CXXThisScopeRAII objects aren't free, don't compute one if it's unused.
llvm-svn: 156987
Manuel Klimek [Thu, 17 May 2012 09:32:05 +0000 (09:32 +0000)]
Fix compile error.
llvm-svn: 156986
Stepan Dyatkovskiy [Thu, 17 May 2012 08:56:30 +0000 (08:56 +0000)]
SelectionDAGBuilder: CaseBlock, CaseRanges and CaseCmp changed representation of Low and High from signed to unsigned. Since unsigned ints usually simpler, faster and allows to reduce some extra signed bit checks needed before <,>,<=,>= comparisons.
llvm-svn: 156985
Kostya Serebryany [Thu, 17 May 2012 08:49:14 +0000 (08:49 +0000)]
[tsan] grammar fixes
llvm-svn: 156984
Kostya Serebryany [Thu, 17 May 2012 08:33:14 +0000 (08:33 +0000)]
[tsan] fix dependency rules in Makefile.old
llvm-svn: 156983
Dmitry Vyukov [Thu, 17 May 2012 08:31:47 +0000 (08:31 +0000)]
tsan: ValgrindSlowdown() should be weak for some time
llvm-svn: 156982
Dmitry Vyukov [Thu, 17 May 2012 08:04:41 +0000 (08:04 +0000)]
tsan: add ValgrindSlowdown() "dynamic annotation"
llvm-svn: 156981
Chris Lattner [Thu, 17 May 2012 05:13:57 +0000 (05:13 +0000)]
Genericize the intrinsics descriptor decoding a bit to make room
for future expansion, no functionality change yet though.
llvm-svn: 156979
Chris Lattner [Thu, 17 May 2012 05:03:24 +0000 (05:03 +0000)]
finish encoding all of the interesting details of intrinsics. Now intrinsics
are only rejected because they can't be encoded into a 32-bit unit, not because
they contain an unencodable feature.
llvm-svn: 156978
Chris Lattner [Thu, 17 May 2012 04:30:58 +0000 (04:30 +0000)]
strengthen the intrinsic descriptor stuff to be able to handle sin, cos and other
intrinsics that use passed-in arguments.
llvm-svn: 156977
Chris Lattner [Thu, 17 May 2012 04:07:48 +0000 (04:07 +0000)]
simplify code generated by tblgen that is not necessary since we dropped
compatibility with LLVM 2.x bitcode files.
llvm-svn: 156976