Commit Graph

78626 Commits

Author SHA1 Message Date
Chad Rosier
d8bd26ee24 Make the MemCpyOptimizer a bit more aggressive. I can't think of a scenerio
where this would be bad as the backend shouldn't have a problem inlining small
memcpys.
rdar://10510150

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145865 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 22:37:00 +00:00
Jim Grosbach
713c70238c Tweak ADDrr fix. Bad check for explicit .w
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145863 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 22:27:04 +00:00
Jim Grosbach
6e507c645d Update tests for r145860. Add a few new ones.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145861 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 22:21:28 +00:00
Jim Grosbach
927b9df4c6 Thumb2 prefer ADD register encoding T2 to T3 when possible.
rdar://10529664


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145860 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 22:16:39 +00:00
Bill Wendling
9bd5d044b7 Move 'returns_twice' definition into alphabetical place.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145854 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 21:27:54 +00:00
Akira Hatanaka
d6bc5237d8 Add definitions of 64-bit extract and insert instrucions and make
PerformANDCombine and PerformOrCombine aware of them. Test cases are included
too.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145853 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 21:26:34 +00:00
Akira Hatanaka
cee46abc16 Split ExtIns into two base classes and have instructions EXT and INS derive from
them.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145852 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 21:14:28 +00:00
Jim Grosbach
da84786bee Thumb2 prefer encoding T3 to T4 for ADD/SUB immediate instructions.
rdar://10529348


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145851 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 21:06:26 +00:00
Akira Hatanaka
2bf08ec854 Have LowerJumpTable support Mips64. Modify 2010-07-20-Switch.ll to test N64 and
O32 with relocation-model=pic too.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145850 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 21:03:03 +00:00
Jim Grosbach
253ef7a779 ARM assembly parsing for the rest of the VMUL data type aliases.
Finish up rdar://10522016.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145846 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 20:29:59 +00:00
Jim Grosbach
422faab909 Fix previous commit. Oops.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145844 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 20:12:26 +00:00
Jim Grosbach
45755a77ec Tidy up. No functional change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145843 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 20:09:44 +00:00
Jim Grosbach
afb500ace1 ARM assmebler parsing for two-operand VMUL instructions.
Combined destination and first source operand for f32 variant of the VMUL
(by scalar) instruction.

rdar://10522016

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145842 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 19:55:46 +00:00
Anna Zaks
e15402f92b Change the Dominators recalculate() function to only rely on GraphTraits
This is a patch by Guoping Long!

As part of utilizing LLVM Dominator computation in Clang, made two changes to LLVM dominators tree implementation:

 - (1) Change the recalculate() template function to only rely on GraphTraits.
 - (2) Add a size() method to GraphTraits template class to query the number of nodes in the graph.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145837 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 19:17:04 +00:00
Hal Finkel
fef3f9aed3 Add test case - this input used to crash because of duplicate generation of SPILL_CRs
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145820 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 17:55:22 +00:00
Hal Finkel
3fd0018af1 enable PPC register scavenging by default (update tests and remove some FIXMEs)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145819 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 17:55:17 +00:00
Hal Finkel
9489487f98 don't include CR bit subregs in callee-saved list
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145818 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 17:55:12 +00:00
Hal Finkel
c4785181a1 remove wasted space for extra bit copies of CR2 subregs
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145817 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 17:55:06 +00:00
Hal Finkel
2e313caa36 add register pressure for CR regs
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145816 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 17:54:17 +00:00
Benjamin Kramer
4da7f90b17 Add a little heuristic to Value::isUsedInBasicBlock to speed it up for small basic blocks.
- Calling getUser in a loop is much more expensive than iterating over a few instructions.
- Use it instead of the open-coded loop in AddrModeMatcher.
- 5% speedup on ARMDisassembler.cpp Release builds.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145810 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 17:23:27 +00:00
NAKAMURA Takumi
27de2a54f3 test/CodeGen/X86/pointer-vector.ll: Add explicit -mtriple=i686-linux.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145805 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 07:54:57 +00:00
Craig Topper
1dc0fbc168 Remove some leftover remnants that once tried to create 64-bit MMX PALIGNR instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145804 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 07:27:14 +00:00
Craig Topper
beabc6cc6d Clean up and optimizations to the X86 shuffle lowering code. No functional change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145803 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 06:56:46 +00:00
Nadav Rotem
1608769abe Add support for vectors of pointers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145801 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-05 06:29:09 +00:00
Jakub Staszak
8e1b12ae68 Fix table of contents.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145793 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-04 20:44:25 +00:00
Jakub Staszak
b170e2de83 Add 'llvm.expect' intrinsic description.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145792 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-04 18:29:26 +00:00
Eric Christopher
309bedd7bc Add inline subprogram names to the name lookup table since they may
not get there any other way.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145789 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-04 06:02:38 +00:00
Bob Wilson
6ce2deacef Fix 80-column issues.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145783 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-04 00:52:23 +00:00
Anton Korobeynikov
0cb2a45cce Emit the ctors in the proper order on ARM/EABI.
Maybe some targets should use this as well.

Patch by Evgeniy Stepanov!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145781 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 23:49:37 +00:00
Venkatraman Govindaraju
80b1ae9292 Sparc CodeGen: Fix AnalyzeBranch for PR 10282. Removing addSuccessor() since
AnalyzeBranch doesn't change the successor, just the order.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145779 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 21:24:48 +00:00
Benjamin Kramer
a86bfc1071 Simplify code. No functionality change.
-3% on ARMDissasembler.cpp.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145773 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 16:18:22 +00:00
Benjamin Kramer
c00c05f8f5 Clear the new cache.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145771 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 15:19:55 +00:00
Benjamin Kramer
feb9b4bc3b Add a "seen blocks" cache to LVI to avoid a linear scan over the whole cache just to remove no blocks from the maps.
-15% on ARMDisassembler.cpp (Release build).  It's not that great to add another
layer of caching to the caching-heavy LVI but I don't see a better way.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145770 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 15:16:45 +00:00
Sanjoy Das
199ce33b3b Check for stack space more intelligently.
libgcc sets the stack limit field in TCB to 256 bytes above the actual
allocated stack limit.  This means if the function's stack frame needs
less than 256 bytes, we can just compare the stack pointer with the
stack limit.  This should result in lesser calls to __morestack.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145766 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 09:32:07 +00:00
Sanjoy Das
40f8222e1e Fix a bug in the x86-32 code generated for segmented stacks.
Currently LLVM pads the call to __morestack with a add and sub of 8
bytes to esp.  This isn't correct since __morestack expects the call
to be followed directly by a ret.

This commit also adjusts the relevant test-case.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145765 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 09:21:07 +00:00
Nick Lewycky
3604924799 Creating multiple JITs on X86 in multiple threads causes multiple writes (of
the same value) to this variable. This code could be refactored, but it doesn't
matter since the old JIT is going away. Add tsan annotations to ignore the
race.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145745 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 02:45:50 +00:00
Chad Rosier
9eff1e33f6 [arm-fast-isel] Unaligned stores of floats require special care.
rdar://10510150

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145742 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 02:21:57 +00:00
Pete Cooper
de2e27cc52 Fixed deadstoreelimination bug where negative indices were incorrectly causing the optimisation to occur
Turns out long long + unsigned long long is unsigned.  Doh!

Fixes http://llvm.org/bugs/show_bug.cgi?id=11455

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145731 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 00:04:30 +00:00
Chad Rosier
24fbf2bf16 Add support for constant folding the pow intrinsic.
rdar://10514247

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145730 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-03 00:00:03 +00:00
Jim Grosbach
587f5062b9 ARM NEON VEXT aliases for data type suffices.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145726 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 23:34:39 +00:00
Jim Grosbach
e40ab244c1 ARM VEXT tighten up operand classes a bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145722 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 22:57:57 +00:00
Jim Grosbach
84defb51ca ARM VST1 single lane assembly parsing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145718 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 22:34:51 +00:00
Akira Hatanaka
99f50fb3ee Test cases for 64-bit multiplication and division.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145717 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 22:31:36 +00:00
Akira Hatanaka
fa341d919f Fix test cases to use FileCheck.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145716 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 22:28:09 +00:00
Nick Lewycky
8a8d479214 Move global variables in TargetMachine into new TargetOptions class. As an API
change, now you need a TargetOptions object to create a TargetMachine. Clang
patch to follow.

One small functionality change in PTX. PTX had commented out the machine
verifier parts in their copy of printAndVerify. That now calls the version in
LLVMTargetMachine. Users of PTX who need verification disabled should rely on
not passing the command-line flag to enable it.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145714 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 22:16:29 +00:00
Jim Grosbach
c4f0b309ee ARM tests for VLD1 single lane w/ writeback.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145713 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 22:03:52 +00:00
Jim Grosbach
872eedbb3a ARM VLD1 single lane assembly parsing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145712 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 22:01:52 +00:00
Jim Grosbach
204aa64f30 ARM encoder method needs the physical register number, not the enum.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145711 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 22:01:25 +00:00
Dylan Noblesmith
a930f3d8a3 TargetMachine: document unnamed bool argument
Its meaning was slightly mysterious without looking at
subclasses.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145705 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 20:53:57 +00:00
Dylan Noblesmith
35af1d7e2c unittests: add ErrorStr to ExecutionEngine test
Makes failures more self-explanatory.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145704 91177308-0d34-0410-b5e6-96231b3b80d8
2011-12-02 20:53:53 +00:00