Commit Graph

11584 Commits

Author SHA1 Message Date
Kalle Raiskila
9363f739cd Handle lshr for i128 correctly on SPU also when
shiftamount > 7.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120288 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-29 14:44:28 +00:00
Kalle Raiskila
c2ebfd454c Enable PostRA scheduling for SPU.
This speeds up selected test cases with up to
5% - no slowdowns observed.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120286 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-29 10:30:25 +00:00
NAKAMURA Takumi
86c817e9d4 test: Check the feature 'loadable_module' with load modules in %llvmshlibdir.
%llvmshlibdir should be 'bin' on Cygming.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120282 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-29 07:58:32 +00:00
Bill Wendling
2f17bf2a44 Add more Thumb encodings.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120279 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-29 01:07:48 +00:00
Bill Wendling
5cbbf68e35 More Thumb encodings.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120278 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-29 01:00:43 +00:00
Bill Wendling
d19ac0c75a Add Thumb encodings for REV instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120277 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-29 00:42:50 +00:00
NAKAMURA Takumi
38d439fb13 test: Use $SharedLibDir for loadable modules. On Cygming, loadable modules are not in lib/ but bin.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120274 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-29 00:20:21 +00:00
NAKAMURA Takumi
b3ccc12831 test: Add the new feature 'loadable_module'.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120273 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-29 00:20:09 +00:00
Bill Wendling
849f2e381e Add more Thumb encodings.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120272 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-29 00:18:15 +00:00
Chris Lattner
d8f717911d fix PR8686, accepting a 'b' suffix at the end of all the setcc
instructions.  I choose to handle this with an asmparser hack,
though it could be handled by changing all the instruction definitions
to allow be "setneb" instead of "setne".  The asm parser hack is
better in this case, because we want the disassembler to produce
setne, not setneb.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120260 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-28 20:23:50 +00:00
Bob Wilson
b1dfa7a8e0 Add support for NEON VLD2-dup instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120236 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-28 06:51:26 +00:00
Rafael Espindola
5bf7c534cf Lower TLS_addr32 and TLS_addr64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120225 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-27 20:43:02 +00:00
Rafael Espindola
bfd2d26159 Implement the data16 prefix.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120224 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-27 20:29:45 +00:00
NAKAMURA Takumi
d420e7b76c CMake: lit(check.vcproj) can run with multiple configurations on Visual Studio.
Unittests need LLVM_BUILD_MODE to pick up each test.

Confirmed on CentOS5, Mingw, MSYS, and with possible configurations on VS8 and VS10.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120212 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-27 13:10:11 +00:00
Bob Wilson
2a0e97431e Add NEON VLD1-dup instructions (load 1 element to all lanes).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120194 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-27 06:35:16 +00:00
Daniel Dunbar
44fa0473ae macho-dump: Fix typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120185 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-27 04:00:06 +00:00
NAKAMURA Takumi
28b6727a35 test/site.exp.in: Add "emitir", for now, fixing up r120156. CMake depends on site.exp.in, though, "emitir" might be unused.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120174 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-26 08:30:15 +00:00
Duncan Sands
20a00c4f80 Remove explicit uses of -emit-llvm, the test infrastructure adds it
automatically.  Use -S with llvm-gcc rather than -c, so tests can
work when llvm-gcc is really dragonegg (which can output IR with -S
but not -c).  Yes, dragonegg supports objective-c++ (poorly though).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120164 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-25 21:48:20 +00:00
Duncan Sands
ce60e8423e Remove explicit uses of -emit-llvm, the test infrastructure adds it
automatically.  Use -S with llvm-gcc rather than -c, so tests can
work when llvm-gcc is really dragonegg (which can output IR with -S
but not -c).  Yes, dragonegg supports objective-c (poorly though).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120163 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-25 21:46:07 +00:00
Duncan Sands
cf99abbc4e Use -S rather than -c for the benefit of dragonegg.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120161 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-25 21:41:35 +00:00
Duncan Sands
71a39f4b6f Remove explicit uses of -emit-llvm, the test infrastructure adds it
automatically.  Use -S with llvm-gcc rather than -c, so tests can
work when llvm-gcc is really dragonegg (which can output IR with -S
but not -c).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120160 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-25 21:39:17 +00:00
Duncan Sands
a0c67bb1f9 Judging from the comment, the system assembler is supposed to assemble
the output of this test.  Since it was producing bitcode, that clearly
wasn't happening!  Have it produce target assembler and assemble that
instead.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120159 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-25 21:26:21 +00:00
Duncan Sands
bb8f59003c Remove explicit uses of -emit-llvm, the test infrastructure adds it
automatically.  Use -S with llvm-gcc rather than -c, so tests can
work when llvm-gcc is really dragonegg (which can output IR with -S
but not -c).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120158 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-25 21:24:35 +00:00
Duncan Sands
53910d9993 Dragonegg cannot output bitcode, only human readable IR, so use -S rather
than -c.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120157 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-25 21:21:59 +00:00
Duncan Sands
48f296dada Use LLVMCC_EMITIR_FLAG rather than hard-coding "-emit-llvm".
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120156 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-25 21:19:52 +00:00
Rafael Espindola
2ace1b68ac Use multiple 0x66 prefixes so that all nops up to 15 bytes are a single instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120147 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-25 17:14:16 +00:00
Rafael Espindola
6b8e4357ec Factor some code to parseSectionFlags and fix the default type of a section.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120145 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-25 15:32:56 +00:00
Nick Lewycky
081f80078d Treat a call of function pointer like a load of the pointer when considering
whether the pointer can be replaced with the global variable it is a copy of.
Fixes PR8680.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120126 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-24 22:04:20 +00:00
Rafael Espindola
25958730df Behave a bit more like gnu as and use the symbol (instead of the section)
for any relocation to a symbol defined in a tls section.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120121 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-24 21:57:39 +00:00
Rafael Espindola
1683fcc823 Relocate with the symbol if the relocation is of kind NTPOFF.
Patch by David Meyer, I added the test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120104 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-24 19:23:50 +00:00
Rafael Espindola
bf8209daf8 Fix and add tests for all cases in x86 and x86_64 where gnu as implicitly
sets the type of a symbol to STT_TLS.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120100 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-24 18:51:21 +00:00
Rafael Espindola
5c7106b2e3 Testcase for r120017.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120099 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-24 18:03:57 +00:00
Kalle Raiskila
702a4046a9 Allow for 'fcmp ogt' in SPU.
Fix by Visa Putkinen!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-24 11:42:17 +00:00
Rafael Espindola
97551276c5 If a symbol is used as tls, mark it as tls even if not declare as so. Probably
fixes PR8659.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120076 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-24 02:19:40 +00:00
Benjamin Kramer
c21a821e9f The srem -> urem transform is not safe for any divisor that's not a power of two.
E.g. -5 % 5 is 0 with srem and 1 with urem.

Also addresses Frits van Bommel's comments.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120049 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-23 20:33:57 +00:00
Bob Wilson
626613d5e8 Recognize sign/zero-extended constant BUILD_VECTORs for VMULL operations.
We need to check if the individual vector elements are sign/zero-extended
values.  For now this only handles constants values.  Radar 8687140.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120034 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-23 19:38:38 +00:00
Benjamin Kramer
b70ebd2aa3 InstCombine: Reduce "X shift (A srem B)" to "X shift (A urem B)" iff B is positive.
This allows to transform the rem in "1 << ((int)x % 8);" to an and.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120028 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-23 18:52:42 +00:00
Duncan Sands
5057f38141 Exploit distributive laws (eg: And distributes over Or, Mul over Add, etc) in a
fairly systematic way in instcombine.  Some of these cases were already dealt
with, in which case I removed the existing code.  The case of Add has a bunch of
funky logic which covers some of this plus a few variants (considers shifts to be
a form of multiplication), which I didn't touch.  The simplification performed is:
A*B+A*C -> A*(B+C).  The improvement is to do this in cases that were not already
handled [such as A*B-A*C -> A*(B-C), which was reported on the mailing list], and
also to do it more often by not checking for "only one use" if "B+C" simplifies.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120024 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-23 14:23:47 +00:00
Kalle Raiskila
0cc5b1f60e Division by pow-of-2 is not cheap on SPU, do it with
shifts.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120022 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-23 13:27:59 +00:00
Rafael Espindola
d93ceeb125 Produce a relocation for pcrel absolute values. Based on a patch by David Meyer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120006 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-23 07:20:12 +00:00
Chris Lattner
79b967bd9f duncan's spider sense was right, I completely reversed the condition
on this instcombine xform.  This fixes a miscompilation of 403.gcc.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119988 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-23 02:42:04 +00:00
Chris Lattner
2e1a75d6f4 filecheckize
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119987 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-23 02:26:52 +00:00
Benjamin Kramer
93f8455d73 InstCombine: Implement X - A*-B -> X + A*B.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119984 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-22 20:31:27 +00:00
Evan Cheng
ab5c703fdb Fix epilogue codegen to avoid leaving the stack pointer in an invalid
state. Previously Thumb2 would restore sp from fp like this:
mov sp, r7
sub, sp, #4
If an interrupt is taken after the 'mov' but before the 'sub', callee-saved
registers might be clobbered by the interrupt handler. Instead, try
restoring directly from sp:
add sp, #4
Or, if necessary (with VLA, etc.) use a scratch register to compute sp and
then restore it:
sub.w r4, r7, #8
mov sp, r7
rdar://8465407


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119977 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-22 18:12:04 +00:00
Duncan Sands
a63395a30f If a GEP index simply advances by multiples of a type of zero size,
then replace the index with zero.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119974 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-22 16:32:50 +00:00
Kalle Raiskila
d87e571e62 Fix a bug with extractelement on SPU.
In the attached testcase, the element was
never extracted (missing rotate).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119973 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-22 16:28:26 +00:00
Benjamin Kramer
ce750f0332 Implement the "if (X == 6 || X == 4)" -> "if ((X|2) == 6)" optimization.
This currently only catches the most basic case, a two-case switch, but can be
extended later.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119964 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-22 09:45:38 +00:00
Wesley Peck
46a928b864 Implement branch analysis in the MBlaze backend.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119951 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-21 21:53:36 +00:00
Duncan Sands
e60d79faf7 Add a rather pointless InstructionSimplify transform, inspired by recent constant
folding improvements: if P points to a type of size zero, turn "gep P, N" into "P".
More generally, if a gep index type has size zero, instcombine could replace the
index with zero, but that is not done here.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119942 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-21 13:53:09 +00:00
Bill Wendling
af2b573614 Add encoding for ARM "trap" instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119938 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-21 11:05:29 +00:00