Jordan Rose [Thu, 21 Feb 2013 01:34:51 +0000 (01:34 +0000)]
[analyzer] Tighten up safety in the use of lazy bindings.
- When deciding if we can reuse a lazy binding, make sure to check if there
are additional bindings in the sub-region.
- When reading from a lazy binding, don't accidentally strip off casts or
base object regions. This slows down lazy binding reading a bit but is
necessary for type sanity when treating one class as another.
A bit of minor refactoring allowed these two checks to be unified in a nice
early-return-using helper function.
<rdar://problem/
13239840>
llvm-svn: 175703
Enrico Granata [Thu, 21 Feb 2013 01:29:04 +0000 (01:29 +0000)]
If you say
(lldb) frame variable
without first launching the inferior, you get:
error: invalid frame
this is misleading and should probably hint that there is no process. Adding this flag makes sure that we get:
error: invalid process
The difference between eFlagRequiresProcess and eFlagProcessMustBeLaunched is an open question.
llvm-svn: 175702
Ted Kremenek [Thu, 21 Feb 2013 01:29:01 +0000 (01:29 +0000)]
Strip 'llvm' from 'llvm::Optional' uses.
llvm-svn: 175701
Sean Callanan [Thu, 21 Feb 2013 01:04:23 +0000 (01:04 +0000)]
Fixed a bug where certain vector code didn't
work on i386. Now we let the JIT emit SSE/SSE2
instructions on i386.
<rdar://problem/
13240476>
llvm-svn: 175700
Richard Trieu [Thu, 21 Feb 2013 00:50:43 +0000 (00:50 +0000)]
Remove commented out code.
llvm-svn: 175699
Fariborz Jahanian [Thu, 21 Feb 2013 00:40:10 +0000 (00:40 +0000)]
objective-C arc IR-gen. Retaining of strong
arguments in function prologue is done
with objc_StoreStrong to pair it with
similar objc_StoreStrong for release in function
epilogue. This is done with -O0 only.
// rdar://
13145317
llvm-svn: 175698
Bill Schmidt [Thu, 21 Feb 2013 00:38:25 +0000 (00:38 +0000)]
PPCDAGToDAGISel::PostprocessISelDAG()
This patch implements the PPCDAGToDAGISel::PostprocessISelDAG virtual
method to perform post-selection peephole optimizations on the DAG
representation.
One optimization is implemented here: folds to clean up complex
addressing expressions for thread-local storage and medium code
model. It will also be useful for large code model sequences when
those are added later. I originally thought about doing this on the
MI representation prior to register assignment, but it's difficult to
do effective global dead code elimination at that point. DCE is
trivial on the DAG representation.
A typical example of a candidate code sequence in assembly:
addis 3, 2, globalvar@toc@ha
addi 3, 3, globalvar@toc@l
lwz 5, 0(3)
When the final instruction is a load or store with an immediate offset
of zero, the offset from the add-immediate can replace the zero,
provided the relocation information is carried along:
addis 3, 2, globalvar@toc@ha
lwz 5, globalvar@toc@l(3)
Since the addi can in general have multiple uses, we need to only
delete the instruction when the last use is removed.
llvm-svn: 175697
David Blaikie [Thu, 21 Feb 2013 00:27:28 +0000 (00:27 +0000)]
Provide a "None" value for convenience when using Optional<T>()
This implementation of NoneType/None does have some holes but I haven't
found one that doesn't - open to improvement.
llvm-svn: 175696
David Blaikie [Thu, 21 Feb 2013 00:26:17 +0000 (00:26 +0000)]
Mark unimplemented function with LLVM_DELETED_FUNCTION
llvm-svn: 175695
David Blaikie [Thu, 21 Feb 2013 00:26:14 +0000 (00:26 +0000)]
Doxycomment SVal's getAs/castAs.
Also document TypeLoc's operations similarly, since it's a good idea.
Post-commit CR feedback from Anna Zaks regarding r175594.
llvm-svn: 175694
Bill Schmidt [Thu, 21 Feb 2013 00:05:29 +0000 (00:05 +0000)]
Relocation enablement for PPC DAG postprocessing pass
llvm-svn: 175693
Chad Rosier [Wed, 20 Feb 2013 23:57:30 +0000 (23:57 +0000)]
Formatting.
llvm-svn: 175692
Jason Molenda [Wed, 20 Feb 2013 23:51:13 +0000 (23:51 +0000)]
*ahem* Enrico.
llvm-svn: 175691
Jakob Stoklund Olesen [Wed, 20 Feb 2013 23:51:10 +0000 (23:51 +0000)]
Don't allocate memory in LiveInterval::join().
Rewrite value numbers directly in the 'Other' LiveInterval which is
moribund anyway. This avoids allocating the OtherAssignments vector.
llvm-svn: 175690
Jim Ingham [Wed, 20 Feb 2013 23:38:37 +0000 (23:38 +0000)]
Mention "return" -> "thread return"
llvm-svn: 175689
Lang Hames [Wed, 20 Feb 2013 23:36:57 +0000 (23:36 +0000)]
Kill of TransferDeadFlag - Dead copies and subreg-to-reg instructions should
just be turned into kills on the spot.
llvm-svn: 175688
Pedro Artigas [Wed, 20 Feb 2013 23:30:56 +0000 (23:30 +0000)]
as the allocator is reset zero out the number of bytes allocated, this was just
missed before but probably what was intended.
llvm-svn: 175687
Jack Carter [Wed, 20 Feb 2013 23:11:17 +0000 (23:11 +0000)]
ELF symbol table field st_other support,
excluding visibility bits.
Mips specific standalone assembler directive "set at".
This directive changes the general purpose register
that the assembler will use when given the symbolic
register name $at.
This does not include negative testing. That will come
in a future patch.
A side affect of this patch recognizes the different
GPR register names for temporaries between old abi
and new abi so a test case for that is included.
Contributer: Vladimir Medic
llvm-svn: 175686
Jakob Stoklund Olesen [Wed, 20 Feb 2013 23:08:26 +0000 (23:08 +0000)]
Copy single reaching defs directly into the LiveInterval.
When findReachingDefs() finds that only one value can reach the basic
block, just copy the work list of visited blocks directly into the live
interval.
Sort the block list and use a LiveRangeUpdater to make the bulk add
fast.
When multiple reaching defs are found, transfer the work list to the
updateSSA() work list as before. Also use LiveRangeUpdater in
updateLiveIns() following updateSSA().
This makes live interval analysis more than 3x faster on one huge test
case.
llvm-svn: 175685
Bill Wendling [Wed, 20 Feb 2013 23:04:11 +0000 (23:04 +0000)]
Add and remove the attribute from the correct slot.
The slot that we're adding/removing the attribute from may not be the same as
the attribute coming in. Make sure that they match up before we try to
add/remove them.
PR15313
llvm-svn: 175684
Bill Schmidt [Wed, 20 Feb 2013 22:43:03 +0000 (22:43 +0000)]
Stabilize vec_constants.ll
llvm-svn: 175683
Erik Verbruggen [Wed, 20 Feb 2013 22:33:46 +0000 (22:33 +0000)]
Fix accidental concatenation for "outputuntil" in the -debug-buffer-size option description.
llvm-svn: 175682
Jordan Rose [Wed, 20 Feb 2013 22:28:41 +0000 (22:28 +0000)]
Revert "intmax_t is long long on Darwin, not long."
'long' and 'long long' are different for the purposes of mangling.
This caused <rdar://problem/
13254874>.
This reverts commit
c2f994d31ec85e9af811af38eb1b28709aef0b2c.
llvm-svn: 175681
Jim Grosbach [Wed, 20 Feb 2013 22:25:15 +0000 (22:25 +0000)]
MCParser: Update method names per coding guidelines.
s/ParseMSInlineAsm/parseMSInlineAsm/
llvm-svn: 175680
David Blaikie [Wed, 20 Feb 2013 22:23:23 +0000 (22:23 +0000)]
Include llvm::Optional in clang/Basic/LLVM.h
Post-commit CR feedback from Jordan Rose regarding r175594.
llvm-svn: 175679
David Blaikie [Wed, 20 Feb 2013 22:23:03 +0000 (22:23 +0000)]
Remove redundant Optional type in favor of llvm::Optional
llvm-svn: 175678
David Blaikie [Wed, 20 Feb 2013 22:23:01 +0000 (22:23 +0000)]
Use op-> directly rather than via Optional<T>::getPointer.
Post-commit CR feedback from Jordan Rose regarding r175594.
llvm-svn: 175677
David Blaikie [Wed, 20 Feb 2013 22:22:59 +0000 (22:22 +0000)]
Rename SVal::isType functions to SVal::isKind
Post-commit CR feedback from Jordan Rose regarding r175594.
llvm-svn: 175676
Jim Grosbach [Wed, 20 Feb 2013 22:21:35 +0000 (22:21 +0000)]
MCParser: Update method names per coding guidelines.
s/AddDirectiveHandler/addDirectiveHandler/
s/ParseMSInlineAsm/parseMSInlineAsm/
s/ParseIdentifier/parseIdentifier/
s/ParseStringToEndOfStatement/parseStringToEndOfStatement/
s/ParseEscapedString/parseEscapedString/
s/EatToEndOfStatement/eatToEndOfStatement/
s/ParseExpression/parseExpression/
s/ParseParenExpression/parseParenExpression/
s/ParseAbsoluteExpression/parseAbsoluteExpression/
s/CheckForValidSection/checkForValidSection/
http://llvm.org/docs/CodingStandards.html#name-types-functions-variables-and-enumerators-properly
No functional change intended.
llvm-svn: 175675
Cameron Zwarich [Wed, 20 Feb 2013 22:10:02 +0000 (22:10 +0000)]
Only use LiveIntervals in TwoAddressInstructionPass, not a mix of Liveintervals
and SlotIndexes.
llvm-svn: 175674
Cameron Zwarich [Wed, 20 Feb 2013 22:10:00 +0000 (22:10 +0000)]
Find anchoring end points for repairIntervalsInRange and repairIndexesInRange
automatically.
llvm-svn: 175673
Cameron Zwarich [Wed, 20 Feb 2013 22:09:57 +0000 (22:09 +0000)]
Make repairIntervalsInRange() more robust. There are now no longer any liveness-
related failures when running 'make check' without LiveVariables with the
verifier enabled. Some of the remaining failures elsewhere may still be fallout
from incorrect updating of LiveIntervals or the few missing cases left in the
two-address pass.
llvm-svn: 175672
Andrew Kaylor [Wed, 20 Feb 2013 21:39:02 +0000 (21:39 +0000)]
Change ELF relocation model to static
llvm-svn: 175671
Arnold Schwaighofer [Wed, 20 Feb 2013 21:33:32 +0000 (21:33 +0000)]
DAGCombiner: Fold pointless truncate, bitcast, buildvector series
(2xi32) (truncate ((2xi64) bitcast (buildvector i32 a, i32 x, i32 b, i32 y)))
can be folded into a (2xi32) (buildvector i32 a, i32 b).
Such a DAG would cause uneccessary vdup instructions followed by vmovn
instructions.
We generate this code on ARM NEON for a setcc olt, 2xf64, 2xf64. For example, in
the vectorized version of the code below.
double A[N];
double B[N];
void test_double_compare_to_double() {
int i;
for(i=0;i<N;i++)
A[i] = (double)(A[i] < B[i]);
}
radar://
13191881
Fixes bug 15283.
llvm-svn: 175670
Daniel Malea [Wed, 20 Feb 2013 21:31:47 +0000 (21:31 +0000)]
Add logging of which test cases are run (through dosep) to help diagnose any hangs/crashes in the buildbots.
llvm-svn: 175669
Jim Grosbach [Wed, 20 Feb 2013 21:31:28 +0000 (21:31 +0000)]
R600: Update for name changes from r175667.
llvm-svn: 175668
Jim Grosbach [Wed, 20 Feb 2013 21:13:59 +0000 (21:13 +0000)]
Update TargetLowering ivars for name policy.
http://llvm.org/docs/CodingStandards.html#name-types-functions-variables-and-enumerators-properly
ivars should be camel-case and start with an upper-case letter. A few in
TargetLowering were starting with a lower-case letter.
No functional change intended.
llvm-svn: 175667
Daniel Malea [Wed, 20 Feb 2013 21:11:20 +0000 (21:11 +0000)]
Fix invalid TestBitfields case (thanks Filipe!), and xfail one case due to GCC compiler bug.
Turns out unnamed bit fields should not be accessed ever in C (C99 section 4.7.8) or C++
(C++11 section 9.6).
Unrelated to the above, this commit marks TestBitfields as expected-to-fail with GCC as 4.7
has problems with padding... the size of the struct with a uint32_t member is reported as 12
when GCC is used. Clang emits correct code.
Parts of this patch prepared by Filipe Cabecinhas!
llvm-svn: 175666
Daniel Malea [Wed, 20 Feb 2013 20:58:20 +0000 (20:58 +0000)]
Skip TestCompletion case that fails on linux (with gcc) because the input prompt is not displayed (llvm.org/pr14637)
llvm-svn: 175665
Matt Kopec [Wed, 20 Feb 2013 20:54:10 +0000 (20:54 +0000)]
Fix TestPublicAPIHeaders test on Linux.
-Change the build_dir variable name to lib_dir
-Set lib_dir to the correct location on Linux
-Set LD_EXTRAS to the actual lldb library
llvm-svn: 175664
Bill Schmidt [Wed, 20 Feb 2013 20:41:42 +0000 (20:41 +0000)]
Additional fixes for bug 15155.
This handles the cases where the 6-bit splat element is odd, converting
to a three-instruction sequence to add or subtract two splats. With this
fix, the XFAIL in test/CodeGen/PowerPC/vec_constants.ll is removed.
llvm-svn: 175663
Jim Ingham [Wed, 20 Feb 2013 20:35:38 +0000 (20:35 +0000)]
Handle the case where the runtime uses class_getMethodImplementation for both scalar and structure
return methods.
rdar://problem/
13238168
llvm-svn: 175662
Reid Kleckner [Wed, 20 Feb 2013 20:29:48 +0000 (20:29 +0000)]
[Sanitizer] Try to fix infinite loop in frame pointer unwinder
Using > instead of >= ensures that each frame deref makes progress and
that we eventually terminate.
Issue: https://code.google.com/p/address-sanitizer/issues/detail?id=162
llvm-svn: 175661
Richard Smith [Wed, 20 Feb 2013 20:19:27 +0000 (20:19 +0000)]
PR15311: Finish implementation of the suggested resolution of core issue 1488,
which allows grouping parens in an abstract-pack-declarator. This was already
mostly implemented, but missed some cases. Add an ExtWarn for use of this
extension until CWG ratifies it.
llvm-svn: 175660
Michael J. Spencer [Wed, 20 Feb 2013 20:18:10 +0000 (20:18 +0000)]
[objdump] Print the PT_INTERP and PT_DYNAMIC correcctly.
llvm-svn: 175659
Chad Rosier [Wed, 20 Feb 2013 20:15:55 +0000 (20:15 +0000)]
Update a comment that looks to have been accidentally deleted many moons ago.
llvm-svn: 175658
Michael J. Spencer [Wed, 20 Feb 2013 20:13:47 +0000 (20:13 +0000)]
[ELF][Writer] Add .interp section.
llvm-svn: 175657
Daniel Malea [Wed, 20 Feb 2013 20:12:11 +0000 (20:12 +0000)]
Fix a bug that causes dotest.py to exit with a passing exit-code, even if some tests fail
- occurs when multiple compilers/architectures are tested (via -C or -A flags)
llvm-svn: 175656
Alexey Samsonov [Wed, 20 Feb 2013 19:55:47 +0000 (19:55 +0000)]
[ASan] revert part of r175631 that looks like accidental commit
llvm-svn: 175655
Michael J. Spencer [Wed, 20 Feb 2013 19:46:12 +0000 (19:46 +0000)]
[ELF][Writer] Add dynamic table.
llvm-svn: 175654
Daniel Malea [Wed, 20 Feb 2013 19:41:49 +0000 (19:41 +0000)]
Fix a problem with the const_variables test case (another lldb bug remains)
- make the test case step past the location where 'index' is initialized, otherwise its value cannot be printed (verified gcc 4.6/4.7 and recent clang)
- "Couldn't materialize struct" error from LLDB still prevents this test case from passing
llvm-svn: 175653
Bill Wendling [Wed, 20 Feb 2013 19:30:01 +0000 (19:30 +0000)]
Attempt to clean up tests for non-X86 platforms.
llvm-svn: 175652
Dan Gohman [Wed, 20 Feb 2013 19:28:46 +0000 (19:28 +0000)]
Rewrite comments.
llvm-svn: 175651
Krzysztof Parzyszek [Wed, 20 Feb 2013 19:25:09 +0000 (19:25 +0000)]
Add comment in Memory.inc explaining r175646.
llvm-svn: 175650
Richard Smith [Wed, 20 Feb 2013 19:22:51 +0000 (19:22 +0000)]
Process and handle attributes on conditions and for loop variables. Process and
diagnose attributes on alias declarations, using directives, and attribute
declarations.
llvm-svn: 175649
Dan Gohman [Wed, 20 Feb 2013 19:15:01 +0000 (19:15 +0000)]
SIGQUIT is a "kill" signal, rather than an "int" signal, in this context.
llvm-svn: 175648
Andrew Kaylor [Wed, 20 Feb 2013 18:24:34 +0000 (18:24 +0000)]
Formatting, grammar
llvm-svn: 175647
Krzysztof Parzyszek [Wed, 20 Feb 2013 18:24:30 +0000 (18:24 +0000)]
On PowerPC, the cache-flush instructions dcbf and icbi are treated as
loads. On FreeBSD, add PROT_READ page protection flag before flushing
cache.
llvm-svn: 175646
Jakob Stoklund Olesen [Wed, 20 Feb 2013 18:18:15 +0000 (18:18 +0000)]
Use LiveRangeUpdater instead of mergeIntervalRanges.
Performance is the same, but LiveRangeUpdater has a more flexible
interface.
llvm-svn: 175645
Jakob Stoklund Olesen [Wed, 20 Feb 2013 18:18:12 +0000 (18:18 +0000)]
Add a LiveRangeUpdater class.
Adding new segments to large LiveIntervals can be expensive because the
LiveRange objects after the insertion point may need to be moved left or
right. This can cause quadratic behavior when adding a large number of
segments to a live range.
The LiveRangeUpdater class allows the LIveInterval to be in a temporary
invalid state while segments are being added. It maintains an internal
gap in the LiveInterval when it is shrinking, and it has a spill area
for new segments when the LiveInterval is growing.
The behavior is similar to the existing mergeIntervalRanges() function,
except it allocates less memory for the spill area, and the algorithm is
turned inside out so the loop is driven by the clients.
llvm-svn: 175644
Andrew Kaylor [Wed, 20 Feb 2013 18:09:21 +0000 (18:09 +0000)]
Adding support for absolute relocations. This occurs in ELF files when a relocation is given with no name and an undefined section. The relocation is applied with an address of zero.
llvm-svn: 175643
Michael Liao [Wed, 20 Feb 2013 18:04:21 +0000 (18:04 +0000)]
Fix PR15267
- When extloading from a vector with non-byte-addressable element, e.g.
<4 x i1>, the current logic breaks. Extend the current logic to
fix the case where the element type is not byte-addressable by loading
all bytes, bit-extracting/packing each element.
llvm-svn: 175642
Chad Rosier [Wed, 20 Feb 2013 18:03:44 +0000 (18:03 +0000)]
[ms-inline asm] Make the comment a bit more verbose.
llvm-svn: 175641
Jakob Stoklund Olesen [Wed, 20 Feb 2013 17:46:51 +0000 (17:46 +0000)]
Add a default constructor for LiveRange.
It is useful to be able to create temporary LiveRange object whose
members are filled in later.
llvm-svn: 175639
Shankar Easwaran [Wed, 20 Feb 2013 17:42:00 +0000 (17:42 +0000)]
rename fixupAddend to relocAddend
llvm-svn: 175638
Chad Rosier [Wed, 20 Feb 2013 17:39:31 +0000 (17:39 +0000)]
[ms-inline asm] Remove this test while I investigate why eax isn't being clobbered.
llvm-svn: 175637
Shankar Easwaran [Wed, 20 Feb 2013 17:38:37 +0000 (17:38 +0000)]
functionality to handle global atoms in Merge sections
llvm-svn: 175636
Bill Schmidt [Wed, 20 Feb 2013 17:31:41 +0000 (17:31 +0000)]
Fix bug 14779 for passing anonymous aggregates [patch by Kai Nacke].
The PPC backend doesn't handle these correctly. This patch uses logic
similar to that in the X86 and ARM backends to track these arguments
properly.
llvm-svn: 175635
Jyotsna Verma [Wed, 20 Feb 2013 16:13:27 +0000 (16:13 +0000)]
Hexagon: Move HexagonMCInst.h to MCTargetDesc/HexagonMCInst.h.
Add HexagonMCInst class which adds various Hexagon VLIW annotations.
In addition, this class also includes some APIs related to the
constant extenders.
llvm-svn: 175634
NAKAMURA Takumi [Wed, 20 Feb 2013 15:52:44 +0000 (15:52 +0000)]
Revert r175626, "ADT/Optional.h: Appease msvc."
Sorry, I didn't cover +Asserts, by accident. :(
llvm-svn: 175633
Bill Schmidt [Wed, 20 Feb 2013 15:50:31 +0000 (15:50 +0000)]
Fix PR15155: lost vadd/vsplat optimization.
During lowering of a BUILD_VECTOR, we look for opportunities to use a
vector splat. When the splatted value fits in 5 signed bits, a single
splat does the job. When it doesn't fit in 5 bits but does fit in 6,
and is an even value, we can splat on half the value and add the result
to itself.
This last optimization hasn't been working recently because of improved
constant folding. To circumvent this, create a pseudo VADD_SPLAT that
can be expanded during instruction selection.
llvm-svn: 175632
Alexander Potapenko [Wed, 20 Feb 2013 15:46:02 +0000 (15:46 +0000)]
[ASan] Delete asan/dynamic dir and temporarily move the interposers declarations to asan_intercepted_functions.h
Now that we have only one dependency on asan_intercepted_functions.h, we can unite that code with the interceptors declarations in asan_interceptors.cc and get rid of asan_intercepted_functions.h
llvm-svn: 175631
Manuel Klimek [Wed, 20 Feb 2013 15:32:58 +0000 (15:32 +0000)]
Fix regression in string literal alignment.
Now correctly indents (again):
a = a + "a"
"a"
"a";
llvm-svn: 175630
Alexander Potapenko [Wed, 20 Feb 2013 15:27:58 +0000 (15:27 +0000)]
[ASan] Make isoc99_*scanf intercepted on Linux only. Those are unavailable on Darwin.
llvm-svn: 175629
Manuel Klimek [Wed, 20 Feb 2013 15:25:48 +0000 (15:25 +0000)]
Fixes bug in string literal alignment.
We now indent the following correctly:
1. some + "literal" /* comment */
"literal";
2. breaking string literals after which we have another string literal.
llvm-svn: 175628
NAKAMURA Takumi [Wed, 20 Feb 2013 15:12:32 +0000 (15:12 +0000)]
Whitespace.
llvm-svn: 175627
NAKAMURA Takumi [Wed, 20 Feb 2013 15:12:24 +0000 (15:12 +0000)]
ADT/Optional.h: Appease msvc.
llvm-svn: 175626
Kostya Serebryany [Wed, 20 Feb 2013 15:10:22 +0000 (15:10 +0000)]
[asan] move preinit_test to Linux since on Mac __asan_init is called from ld-preload-ed runtime
llvm-svn: 175625
Edwin Vane [Wed, 20 Feb 2013 14:35:53 +0000 (14:35 +0000)]
Tweaks and fixes to cpp11-migrate generated tests
* Fixed a comment typo
* Changed 'autogen' to 'generated-tests'
* Made the clean target not fail if the 'generated-tests' directory
doesn't already exist.
llvm-svn: 175624
Kostya Serebryany [Wed, 20 Feb 2013 14:28:08 +0000 (14:28 +0000)]
[asan] on linux, run __asan_init from .preinit_array (even earlier than before)
llvm-svn: 175623
Alexey Samsonov [Wed, 20 Feb 2013 13:54:32 +0000 (13:54 +0000)]
[Sanitizer] use raw syscall instead of _exit() function on Linux
llvm-svn: 175622
Benjamin Kramer [Wed, 20 Feb 2013 13:00:06 +0000 (13:00 +0000)]
Move the SplatByte helper to APInt and generalize it a bit.
llvm-svn: 175621
Daniel Jasper [Wed, 20 Feb 2013 12:56:39 +0000 (12:56 +0000)]
Don't remove blank lines within unwrapped lines.
If the code author decides to put empty lines anywhere into the code we
should treat them equally, i.e. reduce them to the configured
MaxEmptyLinesToKeep.
With this change, we e.g. keep the newline in:
SomeType ST = {
// First value
a,
// Second value
b
};
llvm-svn: 175620
Elena Demikhovsky [Wed, 20 Feb 2013 12:42:54 +0000 (12:42 +0000)]
I optimized the following patterns:
sext <4 x i1> to <4 x i64>
sext <4 x i8> to <4 x i64>
sext <4 x i16> to <4 x i64>
I'm running Combine on SIGN_EXTEND_IN_REG and revert SEXT patterns:
(sext_in_reg (v4i64 anyext (v4i32 x )), ExtraVT) -> (v4i64 sext (v4i32 sext_in_reg (v4i32 x , ExtraVT)))
The sext_in_reg (v4i32 x) may be lowered to shl+sar operations.
The "sar" does not exist on 64-bit operation, so lowering sext_in_reg (v4i64 x) has no vector solution.
I also added a cost of this operations to the AVX costs table.
llvm-svn: 175619
Kostya Serebryany [Wed, 20 Feb 2013 12:37:22 +0000 (12:37 +0000)]
[asan] add a lit test for invoke-with-no-return instrumentation
llvm-svn: 175618
Kostya Serebryany [Wed, 20 Feb 2013 12:35:15 +0000 (12:35 +0000)]
[asan] instrument invoke insns with noreturn attribute (as well as call insns)
llvm-svn: 175617
Logan Chien [Wed, 20 Feb 2013 12:21:33 +0000 (12:21 +0000)]
Fix thumbv5e frame lowering assertion failure.
It is possible that frame pointer is not found in the
callee saved info, thus FramePtrSpillFI may be incorrect
if we don't check the result of hasFP(MF).
Besides, if we enable the stack coloring algorithm, there
will be an assertion to ensure the slot is live. But in
the test case, %var1 is not live in the prologue of the
function, and we will get the assertion failure.
Note: There is similar code in ARMFrameLowering.cpp.
llvm-svn: 175616
Evgeniy Stepanov [Wed, 20 Feb 2013 11:06:07 +0000 (11:06 +0000)]
[sanitizer] Fix lint.
llvm-svn: 175615
Kostya Serebryany [Wed, 20 Feb 2013 10:43:05 +0000 (10:43 +0000)]
[asan] add test throw_call_test (broken with -static-libstdc++)
llvm-svn: 175614
Manuel Klimek [Wed, 20 Feb 2013 10:15:13 +0000 (10:15 +0000)]
Implements breaking of string literals if they stick out.
An alternative strategy to calculating the break on demand when hitting
a token that would need to be broken would be to put all possible breaks
inside the token into the optimizer.
Currently only supports breaking at spaces; more break points to come.
llvm-svn: 175613
Bill Wendling [Wed, 20 Feb 2013 08:01:34 +0000 (08:01 +0000)]
Remove target-specific features.
llvm-svn: 175610
Bill Wendling [Wed, 20 Feb 2013 07:48:23 +0000 (07:48 +0000)]
Use the attribute group reference instead of the attribute directly.
llvm-svn: 175609
David Blaikie [Wed, 20 Feb 2013 07:39:20 +0000 (07:39 +0000)]
Fully qualify llvm::next to avoid ambiguity when building as C++11.
llvm-svn: 175608
David Blaikie [Wed, 20 Feb 2013 07:39:18 +0000 (07:39 +0000)]
Fix the (clang -Werror) build by removing an unused member variable.
llvm-svn: 175607
Bill Wendling [Wed, 20 Feb 2013 07:22:19 +0000 (07:22 +0000)]
Modify the tests to use attribute group references instead of listing the
function attributes.
llvm-svn: 175606
Bill Wendling [Wed, 20 Feb 2013 07:21:42 +0000 (07:21 +0000)]
Modify the LLVM assembly output so that it uses references to represent function attributes.
This makes the LLVM assembly look better. E.g.:
define void @foo() #0 { ret void }
attributes #0 = { nounwind noinline ssp }
llvm-svn: 175605
Cameron Zwarich [Wed, 20 Feb 2013 06:46:48 +0000 (06:46 +0000)]
Add support to the two-address pass for updating LiveIntervals in many of the
common transformations. This includes updating repairIntervalsInRange() to
handle more cases.
llvm-svn: 175604
Cameron Zwarich [Wed, 20 Feb 2013 06:46:46 +0000 (06:46 +0000)]
Move the computation of the IsEarlyClobber flag into its own loop, since the
correct value is needed in every iteration of the loop for updating
LiveIntervals.
llvm-svn: 175603
Cameron Zwarich [Wed, 20 Feb 2013 06:46:44 +0000 (06:46 +0000)]
Modify LiveInterval::addRange() to match the comment about what it returns.
llvm-svn: 175602
Cameron Zwarich [Wed, 20 Feb 2013 06:46:41 +0000 (06:46 +0000)]
Add SlotIndexes::repairIndexesInRange(), which repairs SlotIndexes after adding
and removing instructions. The implementation seems more complicated than it
needs to be, but I couldn't find something simpler that dealt with all of the
corner cases.
Also add a call to repairIndexesInRange() from repairIntervalsInRange().
llvm-svn: 175601