Roman Divacky [Tue, 7 Jun 2011 17:31:02 +0000 (17:31 +0000)]
Make EmitIntValue() work properly on big-endian targets.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132715
91177308-0d34-0410-b5e6-
96231b3b80d8
Stuart Hastings [Tue, 7 Jun 2011 15:23:11 +0000 (15:23 +0000)]
Tweak this test for ARM-hosted 'bot.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132711
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Tue, 7 Jun 2011 06:05:58 +0000 (06:05 +0000)]
Switched to DenseMap for allowed sets in PBQP. Reduces total LLC time by 15% on CINT2006 for x86-32.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132707
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Tue, 7 Jun 2011 05:23:58 +0000 (05:23 +0000)]
Move the legalizer tests to the X86 directory because the test uses the x86
codegen. Thanks Galina.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132706
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 7 Jun 2011 02:17:21 +0000 (02:17 +0000)]
Add test case for C++ exception handling and fix the following mistakes in MipsFrameLowering::emitPrologue:
- cfi directives are not inserted at the right location or in the right order.
- The source MachineLocation for the cfi directive that changes the cfa register
to $fp should be MachineLocation::VirtualFP.
- A PROLOG_LABEL that marks the beginning of cfi_offset directives for
callee-saved register is emitted even when no callee-saved registers are
saved.
- When a callee-saved double precision register is saved, two cfi_offset
directives, one for each of the paired single precision registers, should be
emitted.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132703
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Tue, 7 Jun 2011 00:38:12 +0000 (00:38 +0000)]
Fix a merge bug in preRAsched for handling physreg aliases.
I've been sitting on this long enough trying to find a test case. I
think the fix should go in now, but I'll keep working on the test case.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132701
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Tue, 7 Jun 2011 00:08:49 +0000 (00:08 +0000)]
Fix for setjmp/longjmp exception handling on ARM. setjmp clobbers CPSR.
rdar://problem/
9556069
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132699
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 6 Jun 2011 23:55:20 +0000 (23:55 +0000)]
Simplify local live range splitting's safeguard to fix PR10070.
When local live range splitting creates a live range with the same
number of instructions as the old range, mark it as RS_Local. When such
a range is seen again, require that it be split in a way that reduces
the number of instructions. That guarantees we are making progress while
still being able to perform 3 -> 2+3 splits as required by PR10070.
This also means that the PrevSlot map is no longer needed. This was also
used to estimate new spill weights, but that is no longer necessary
after slotIndexes::insertMachineInstrInMaps() got the extra Late
insertion argument.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132697
91177308-0d34-0410-b5e6-
96231b3b80d8
Stuart Hastings [Mon, 6 Jun 2011 23:15:58 +0000 (23:15 +0000)]
Followup to 132458, omit unnecessary stack copy when x87 input is a
load. rdar://problem/
6373334
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132696
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Mon, 6 Jun 2011 22:45:46 +0000 (22:45 +0000)]
Add -O1/2/3 to bugpoint, so when you conclude opt -O2 reproduces an issue, you can just run bugpoint -O2. :) My implementation isn't precisely equivalent to what opt does, but as far as I can tell, it's close enough.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132695
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Mon, 6 Jun 2011 22:13:27 +0000 (22:13 +0000)]
We only do always-inlining at -O1; make opt reflect that.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132693
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 6 Jun 2011 21:02:04 +0000 (21:02 +0000)]
Get allocation orders from RegisterClassInfo when possible.
Only target-dependent hints require callbacks. The RCI allocation order
has CSR aliases last according to their order of appearance in the
getCalleeSavedRegs list. This can depend on the calling convention.
This way, AllocationOrder::next doesn't have to check for reserved
registers, and CSRs are always allocated last, even with weird calling
conventions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132690
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Mon, 6 Jun 2011 20:55:56 +0000 (20:55 +0000)]
Add methods to support the integer-promotion of vector types. Methods to
legalize SDNodes such as BUILD_VECTOR, EXTRACT_VECTOR_ELT, etc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132689
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Mon, 6 Jun 2011 20:23:00 +0000 (20:23 +0000)]
This directory was missing the dg.exp to cause the tests to run. Some time since
it was added, the test has regressed, so XFAIL it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132686
91177308-0d34-0410-b5e6-
96231b3b80d8
Stuart Hastings [Mon, 6 Jun 2011 20:03:22 +0000 (20:03 +0000)]
Test case for PR10085.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132682
91177308-0d34-0410-b5e6-
96231b3b80d8
Stuart Hastings [Mon, 6 Jun 2011 16:44:31 +0000 (16:44 +0000)]
Avoid FGETSIGN of 80-bit types. Fixes PR10085.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132681
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 6 Jun 2011 16:36:30 +0000 (16:36 +0000)]
Don't try to be clever, just preserve the target's allocation order.
The order of registers returned by getCalleeSavedRegs is used to lay out
the fixed stack slots for CSRs. Some targets like their CSRs used from
one end, and some targets want them used from the other end.
When computing an allocation order, simply preserve the relative
ordering of CSRs that the target specifies in its allocation order.
Reordering CSRs would break some targets, ARM in particular.
We still place volatiles before the CSRs, providing slightly better
results with different calling conventions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132680
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Mon, 6 Jun 2011 05:46:34 +0000 (05:46 +0000)]
PR10077: fix fast-isel of extractvalue of aggregate constants.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132676
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Sun, 5 Jun 2011 18:20:05 +0000 (18:20 +0000)]
Harden tests for windows path separators.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132671
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Sun, 5 Jun 2011 14:36:47 +0000 (14:36 +0000)]
Use path API for path concatenation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132668
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Sun, 5 Jun 2011 02:43:45 +0000 (02:43 +0000)]
Basic support for macros with explicit arguments.
We still don't handle
* default values
* :req
* :vararg
* \()
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132656
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Sun, 5 Jun 2011 01:20:06 +0000 (01:20 +0000)]
Produce an undefined reference to _GLOBAL_OFFSET_TABLE_ if we have a
VK_GOTOFF reloc. This matches as' behavior, but it is not clear why the linker
might need this, so I added a FIXME.
I could test this by duplicating test/MC/ELF/got.s, but it doesn't look
worthwhile.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132655
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Sat, 4 Jun 2011 23:34:40 +0000 (23:34 +0000)]
Fix a test that keeps breaking when allocation orders change.
Who said FileCheck couldn't handle arbitrarily complex conditions?
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132654
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Sat, 4 Jun 2011 20:58:08 +0000 (20:58 +0000)]
TypeLegalizer: Add support for passing of vector-promoted types in registers (copyFromParts/copyToParts).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132649
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Sat, 4 Jun 2011 20:32:01 +0000 (20:32 +0000)]
TypeLegalizer: Fix a bug in the promotion of elements of integer vectors.
(only happens when using the -promote-elements option).
The correct legalization order is to first try to promote element. Next, we try
to widen vectors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132648
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Sat, 4 Jun 2011 18:16:26 +0000 (18:16 +0000)]
Refactor parsing of variable names (ie., %foo and @foo) since they have the same
rules. Also refactor "read string until quote" into its own function. No
functionality change!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132645
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Sat, 4 Jun 2011 17:38:07 +0000 (17:38 +0000)]
Add support for @GOTPTOFF in i386 mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132643
91177308-0d34-0410-b5e6-
96231b3b80d8
Hans Wennborg [Sat, 4 Jun 2011 16:00:19 +0000 (16:00 +0000)]
Cast unused parameter to void.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132641
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Sat, 4 Jun 2011 09:42:04 +0000 (09:42 +0000)]
If the block that we're threading through is jumped to by an indirect branch,
then we don't want to set the destination in the indirect branch to the
destination. This is because the indirect branch needs its destinations to have
had their block addresses taken. This isn't so of the new critical edge that's
split during this process. If it turns out that the destination block has only
one predecessor, and that being a BB with an indirect branch, then it won't be
marked as 'used' and may be removed.
PR10072
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132638
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Sat, 4 Jun 2011 07:49:55 +0000 (07:49 +0000)]
Drop a RecordKeeper reference that wasn't necessary.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132636
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sat, 4 Jun 2011 07:05:05 +0000 (07:05 +0000)]
Add a testcase to demonstrate the problem where phi translation is
ignored for clobbering partial-alias loads.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132633
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sat, 4 Jun 2011 06:50:18 +0000 (06:50 +0000)]
Reapply r131781, now that the GVN bug with partially-aliasing loads
is disabled.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132632
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sat, 4 Jun 2011 06:48:50 +0000 (06:48 +0000)]
Disable the main feature of 130180, the elimination of loads that are
redundant with partially-aliasing loads.
When computing what portion of a clobbering load value is needed,
it doesn't consider phi-translation which may have occurred
between the clobbing load and the redundant load.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132631
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sat, 4 Jun 2011 06:21:23 +0000 (06:21 +0000)]
Remove this test, which should have been reverted along with r131781.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132628
91177308-0d34-0410-b5e6-
96231b3b80d8
Roman Divacky [Sat, 4 Jun 2011 06:15:42 +0000 (06:15 +0000)]
Use processor name in the target triple. This fixes PowerPC64 to generate
64bit code for example.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132627
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sat, 4 Jun 2011 05:11:22 +0000 (05:11 +0000)]
Revert r131781 again. Apparently there is more going on here.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132625
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Sat, 4 Jun 2011 05:09:36 +0000 (05:09 +0000)]
Silence compiler warnings.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132624
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Sat, 4 Jun 2011 04:11:37 +0000 (04:11 +0000)]
Teach TableGen to evaluate DAG expressions as set operations.
A TableGen backend can define how certain classes can be expanded into
ordered sets of defs, typically by evaluating a specific field in the
record. The SetTheory class can then evaluate DAG expressions that refer
to these named sets.
A number of standard set and list operations are predefined, and the
backend can add more specialized operators if needed. The -print-sets
backend is used by SetTheory.td to provide examples.
This is intended to simplify how register classes are defined:
def GR32_NOSP : RegisterClass<"X86", [i32], 32, (sub GR32, ESP)>;
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132621
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Sat, 4 Jun 2011 02:07:10 +0000 (02:07 +0000)]
Fold assert-only-used variable into the assert.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132620
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Sat, 4 Jun 2011 01:30:52 +0000 (01:30 +0000)]
Missing include of climits in the new BranchProbability pass.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132616
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Sat, 4 Jun 2011 01:16:30 +0000 (01:16 +0000)]
New BranchProbabilityInfo analysis. Patch by Jakub Staszak!
BranchProbabilityInfo provides an interface for IR passes to query the
likelihood that control follows a CFG edge. This patch provides an
initial implementation of static branch predication that will populate
BranchProbabilityInfo for branches with no external profile
information using very simple heuristics. It currently isn't hooked up
to any external profile data, so static prediction does all the work.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132613
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sat, 4 Jun 2011 00:46:31 +0000 (00:46 +0000)]
Reapply r131781 (revert r131809), now that some BasicAA shortcomings
it exposed are fixed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132611
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sat, 4 Jun 2011 00:31:50 +0000 (00:31 +0000)]
Fix BasicAA's recursion detection so that it doesn't pessimize
queries in the case of a DAG, where a query reaches a node
visited earlier, but it's not on a cycle. This avoids
MayAlias results in cases where BasicAA is expected to
return MustAlias or PartialAlias in order to protect TBAA.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132609
91177308-0d34-0410-b5e6-
96231b3b80d8
Stuart Hastings [Fri, 3 Jun 2011 23:53:54 +0000 (23:53 +0000)]
Reapply 132424 with fixes. This fixes PR10068.
rdar://problem/
5993888
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132606
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 3 Jun 2011 22:45:21 +0000 (22:45 +0000)]
Fix some tests that depend on register allocation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132602
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 3 Jun 2011 22:45:18 +0000 (22:45 +0000)]
Blackfin always uses a reserved call frame.
Materializing the stack pointer update before a call requires a scratch
register that may not be available.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132601
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Fri, 3 Jun 2011 22:09:12 +0000 (22:09 +0000)]
Another possible bug. Stopgap until we can autogenerate tables and
constraint lengths.
Part of rdar://
9037836 and rdar://
9119939
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132598
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Fri, 3 Jun 2011 20:56:31 +0000 (20:56 +0000)]
Testcase for llvm-gcc commit r132591.
Part of rdar://
9037836 and rdar://
9119939
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132592
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Fri, 3 Jun 2011 20:44:52 +0000 (20:44 +0000)]
Fix an off by one error.
Part of rdar://
9037836 and rdar://
9119939
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132590
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 3 Jun 2011 20:43:58 +0000 (20:43 +0000)]
Remove testcase. It's problematic for the buildbots.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132589
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 3 Jun 2011 20:34:53 +0000 (20:34 +0000)]
Switch AllocationOrder to using RegisterClassInfo instead of a BitVector
of reserved registers.
Use RegisterClassInfo in RABasic as well. This slightly changes som
allocation orders because RegisterClassInfo puts CSR aliases last.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132581
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 3 Jun 2011 20:34:50 +0000 (20:34 +0000)]
Preserve the original ordering when a CSR has multiple aliases.
Previously, these aliases would be ordered alphabetically. (BH, BL)
Print out the computed allocation orders.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132580
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 3 Jun 2011 20:17:36 +0000 (20:17 +0000)]
When merging MustAlias and PartialAlias, chose PartialAlias instead
of conservatively choosing MayAlias.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132579
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Fri, 3 Jun 2011 19:46:19 +0000 (19:46 +0000)]
Use IRBuilder, preserve line numbers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132578
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 3 Jun 2011 19:21:05 +0000 (19:21 +0000)]
Accomodate front-ends which use private instead of internal here.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132575
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Fri, 3 Jun 2011 17:24:37 +0000 (17:24 +0000)]
Make the Uv constraint a memory operand. This doesn't solve the
addressing mode problem mentioned in r132559.
Backend part of rdar://
9037836 and part of rdar://
9119939
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132561
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Fri, 3 Jun 2011 17:21:23 +0000 (17:21 +0000)]
Add a TODO about memory operands.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132559
91177308-0d34-0410-b5e6-
96231b3b80d8
Hans Wennborg [Fri, 3 Jun 2011 17:15:37 +0000 (17:15 +0000)]
Test commit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132558
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Fri, 3 Jun 2011 17:04:51 +0000 (17:04 +0000)]
A typedef's context is not the same as type's context. It is the context of typedef decl itself. Use extra parameter to communicate this to DIBuilder.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132556
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Fri, 3 Jun 2011 17:02:19 +0000 (17:02 +0000)]
Revert name change from r132533. Lower case naming was intended per style guidelines.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132555
91177308-0d34-0410-b5e6-
96231b3b80d8
Roman Divacky [Fri, 3 Jun 2011 15:47:49 +0000 (15:47 +0000)]
Fix wrong usages of CTR/MCTR where CTR8/MCTR8 was meant.
- Check for MTCTR8 in addition to MTCTR when looking up a hazard.
- When lowering an indirect call use CTR8 when targeting 64bit.
- Introduce BCTR8 that uses CTR8 and use it on 64bit when expanding ISD::BRIND.
The last change fixes PR8487. With those changes, we are able to compile a
running "ls" and "sh" on FreeBSD/PowerPC64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132552
91177308-0d34-0410-b5e6-
96231b3b80d8
Zhongxing Xu [Fri, 3 Jun 2011 08:29:51 +0000 (08:29 +0000)]
singed int causes signed extension, which contradicts the intention to pick up
integers with high 32 bits being zero.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132538
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Fri, 3 Jun 2011 08:25:39 +0000 (08:25 +0000)]
Rework the logic to not rely on undefined behaviour (1LL << 64). Also simplify.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132537
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Fri, 3 Jun 2011 06:27:15 +0000 (06:27 +0000)]
Bail on unswitching a switch statement for a case with a critical edge. We name
which edge to split by pred/succ pair, which means that we can end up splitting
the wrong edge (by case value) in the switch statement entirely. Fixes PR10031!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132535
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Fri, 3 Jun 2011 05:09:12 +0000 (05:09 +0000)]
Whitespace and other cleanup. Functionallity unchanged.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132533
91177308-0d34-0410-b5e6-
96231b3b80d8
Charles Davis [Fri, 3 Jun 2011 05:09:12 +0000 (05:09 +0000)]
Treat Win64 EH as a DWARF EH scheme. For GCC-style exceptions, the layout of
the handler's data area is similar to a DWARF-format LSDA. (It is, in fact,
a 32-bit pointer to the personality routine followed by the DWARF LSDA.)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132532
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Fri, 3 Jun 2011 02:20:48 +0000 (02:20 +0000)]
Corrections and additional information for "Creating and LLVM Project"
documentation. This should now reflect the current state of LLVM Makefiles.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132527
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Fri, 3 Jun 2011 02:16:53 +0000 (02:16 +0000)]
whitespace
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132526
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Fri, 3 Jun 2011 01:13:19 +0000 (01:13 +0000)]
Add ARM fast-isel support for materializing the address of a global in cases where the global uses an indirect symbol.
rdar://
9431157
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132522
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Fri, 3 Jun 2011 00:48:58 +0000 (00:48 +0000)]
Basic PassManager diagnostics.
Added asserts whenever attempting to use a potentially
uninitialized pass. This helps people trying to develop a new pass and
people trying to understand the bug reports filed by the former people.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132520
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Fri, 3 Jun 2011 00:44:32 +0000 (00:44 +0000)]
whitespace
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132519
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Thu, 2 Jun 2011 23:57:27 +0000 (23:57 +0000)]
Test case pasto (failed when run with IR verifier).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132516
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 2 Jun 2011 23:41:40 +0000 (23:41 +0000)]
Avoid calling TRI->getAllocatableSet in RAFast.
When compiling a program with lots of small functions like
483.xalancbmk, this makes RAFast 11% faster.
Add some comments to clarify the difference between unallocatable and
reserved registers. It's quite subtle.
The fast register allocator depends on EFLAGS' not being allocatable on
x86. That way it can completely avoid tracking liveness, and it won't
mind when there are multiple uses of a single def.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132514
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Thu, 2 Jun 2011 23:16:42 +0000 (23:16 +0000)]
Have LowerOperandForConstraint handle multiple character constraints.
Part of rdar://
9119939
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132510
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 2 Jun 2011 23:07:24 +0000 (23:07 +0000)]
Flag unallocatable register classes instead of giving them empty
allocation orders.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132509
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 2 Jun 2011 23:07:20 +0000 (23:07 +0000)]
Make it possible to have unallocatable register classes.
Some register classes are only used for instruction operand constraints.
They should never be used for virtual registers. Previously, those
register classes were given an empty allocation order, but now you can
say 'let isAllocatable=0' in the register class definition.
TableGen calculates if a register is part of any allocatable register
class, and makes that information available in TargetRegisterDesc::inAllocatableClass.
The goal here is to eliminate use cases for overriding allocation_order_*
methods.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132508
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Thu, 2 Jun 2011 22:46:58 +0000 (22:46 +0000)]
Preserve line number information while converting Invoke into a Call.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132505
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 2 Jun 2011 22:26:15 +0000 (22:26 +0000)]
This should have been a C++ testcase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132504
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 2 Jun 2011 22:22:43 +0000 (22:22 +0000)]
Just use a SmallVector.
I was confused whether new uint8_t[] would zero-initialize the returned
array, and it seems that so is gcc-4.0.
This should fix the test failures on darwin 9.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132500
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 2 Jun 2011 22:12:42 +0000 (22:12 +0000)]
Testcase for r132493.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132495
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 2 Jun 2011 22:11:49 +0000 (22:11 +0000)]
Update for r132493 change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132494
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Thu, 2 Jun 2011 21:31:00 +0000 (21:31 +0000)]
Remove dead code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132488
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Thu, 2 Jun 2011 21:26:52 +0000 (21:26 +0000)]
Update DBG_VALUEs while breaking anti dependencies.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132487
91177308-0d34-0410-b5e6-
96231b3b80d8
Tanya Lattner [Thu, 2 Jun 2011 21:25:24 +0000 (21:25 +0000)]
Fix encoding for VEXTdf.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132486
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Thu, 2 Jun 2011 21:24:42 +0000 (21:24 +0000)]
PR10067: Add missing safety check to call return transformation in MemCpyOpt::processStore. If something accesses the dest of the "copy" between the call and the copy, the performCallSlotOptzn transformation is not valid.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132485
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Thu, 2 Jun 2011 20:07:12 +0000 (20:07 +0000)]
During post RA scheduling, do not try to chase reg defs. to preserve DBG_VALUEs. This approach has several downsides, for example, it does not work when dbg value is a constant integer, it does not work if reg is defined more than once, it places end of debug value range markers in the wrong place. It even causes misleading incorrect debug info when duplicate DBG_VALUE instructions point to same reg def.
Instead, use simpler approach and let DBG_VALUE follow its predecessor instruction. After live debug value analysis pass, all DBG_VALUE instruction are placed at the right place. Thanks Jakob for the hint!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132483
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Thu, 2 Jun 2011 20:02:48 +0000 (20:02 +0000)]
Add test for PR10068.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132482
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Thu, 2 Jun 2011 19:57:47 +0000 (19:57 +0000)]
Revert 132424 to fix PR10068.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132479
91177308-0d34-0410-b5e6-
96231b3b80d8
Stuart Hastings [Thu, 2 Jun 2011 19:26:49 +0000 (19:26 +0000)]
Andy pointed out a dumb omission in this test case. Thanks Andy!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132477
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Thu, 2 Jun 2011 19:26:37 +0000 (19:26 +0000)]
Add a new parse hint for multi-letter constraints in inline asm.
Testcase will come when we use it.
Part of rdar://
9119939
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132476
91177308-0d34-0410-b5e6-
96231b3b80d8
Stuart Hastings [Thu, 2 Jun 2011 18:44:05 +0000 (18:44 +0000)]
Jakob pointed out a dumb omission in this test case. Thanks Jakob!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132472
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 2 Jun 2011 18:35:30 +0000 (18:35 +0000)]
Use RegisterClassInfo::getOrder in RAFast.
This saves two virtual function calls and an Allocatable BitVector test,
making RAFast run 2% faster.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132471
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 2 Jun 2011 17:14:04 +0000 (17:14 +0000)]
.cfi directive register parsing flexibility.
Parsing a register name/number for .cfi directives can't assume that a
register name starts with a '%' token. Be more flexible and check for a
register number instead. Still unlikely to be perfect, but it allows us
to parse both plain identifiers as register names and integers as register
numbers, which is what we're wanting to support at this point.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132466
91177308-0d34-0410-b5e6-
96231b3b80d8
Stuart Hastings [Thu, 2 Jun 2011 15:57:11 +0000 (15:57 +0000)]
Omit unnecessary stack copy when x87 input is a load.
rdar://problem/
6373334
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132458
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Thu, 2 Jun 2011 12:07:44 +0000 (12:07 +0000)]
Start with a zeroed CSRNum map.
Found by valgrind.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132457
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 2 Jun 2011 05:43:49 +0000 (05:43 +0000)]
Initialize members to fix problem found by valgrind.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132456
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 2 Jun 2011 05:43:46 +0000 (05:43 +0000)]
Use TRI::has{Sub,Super}ClassEq() where possible.
No functional change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132455
91177308-0d34-0410-b5e6-
96231b3b80d8
Stuart Hastings [Thu, 2 Jun 2011 05:05:39 +0000 (05:05 +0000)]
Tweak testcase for ARM bot. rdar://problem/
5993888
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132454
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Thu, 2 Jun 2011 02:34:55 +0000 (02:34 +0000)]
Don't hardcode the %reg format in the streamer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132451
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 2 Jun 2011 02:19:35 +0000 (02:19 +0000)]
Add a RegisterClassInfo class that lazily caches information about
register classes.
It provides information for each register class that cannot be
determined statically, like:
- The number of allocatable registers in a class after filtering out the
reserved and invalid registers.
- The preferred allocation order with registers that overlap callee-saved
registers last.
- The last callee-saved register that overlaps a given physical register.
This information usually doesn't change between functions, so it is
reused for compiling multiple functions when possible. The many
possible combinations of reserved and callee saves registers makes it
unfeasible to compute this information statically in TableGen.
Use RegisterClassInfo to count available registers in various heuristics
in SimpleRegisterCoalescing, making the pass run 4% faster.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132450
91177308-0d34-0410-b5e6-
96231b3b80d8