Commit Graph

11688 Commits

Author SHA1 Message Date
Nate Begeman
2ea8ee7c76 Formalize the notion that AVX and SSE are non-overlapping extensions from the compiler's point of view. Per email discussion, we either want to always use VEX-prefixed instructions or never use them, and are taking "HasAVX" to mean "Always use VEX". Passing -mattr=-avx,+sse42 should serve to restore legacy SSE support when desirable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121439 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-10 00:26:57 +00:00
Rafael Espindola
1c952b9cc9 Initial support for the cfi directives. This is just enough to get
f:
        .cfi_startproc
        nop
        .cfi_endproc

assembled (on ELF).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121434 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-09 23:48:29 +00:00
Kevin Enderby
193c3acbe5 Add support for parsing ARM arithmetic instructions that update or don't update
the condition codes.  Where the ones that do have an 's' suffix and the ones
that don't don't have the suffix.  The trick is if MatchInstructionImpl() fails
we try again after adding a CCOut operand with the correct value and removing
the 's' if present.  Four simple test cases added for now, lots more to come.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121401 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-09 19:19:43 +00:00
Jim Grosbach
c6f9261711 ARM stm/ldm instructions require more than one register in the register list.
Otherwise, a plain str/ldr should be used instead. Make sure we account for
that in prologue/epilogue code generation.
rdar://8745460

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121391 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-09 18:31:13 +00:00
Bruno Cardoso Lopes
908b6ddad6 Add ROTR and ROTRV mips32 instructions. Patch by Akira Hatanaka
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121377 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-09 17:32:30 +00:00
Chris Lattner
8fdca6a873 enhance memcpyopt to zap memcpy's that have the same src/dst.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121362 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-09 07:45:45 +00:00
Chris Lattner
f7f35467a9 fix PR8753, eliminating a case where we'd infinitely make a
substitution because it doesn't actually change the IR.  Patch by
Jakub Staszak!



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121361 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-09 07:39:50 +00:00
Eric Christopher
d8c0536651 Rewrite the darwin tlv support to use a chain and return to copying
the output to the correct register. Fixes a hidden problem uncovered
by the last patch where we'd try to DAG combine our MVT::Other node
oddly.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121358 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-09 06:25:53 +00:00
Dan Gohman
d8e0c0438a Really check that the bits that will become zero are actually already zero
before eliminating the operation that zeros them. This fixes rdar://8739316.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121353 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-09 02:52:17 +00:00
Eric Christopher
8bce7cc3bf Remove extraneous copy from DAG conversion for darwin tls. This was
popping up at O0 when it wasn't folded and the fast allocator would
complain.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121330 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-09 00:27:58 +00:00
Kevin Enderby
76331754d4 Allow a slash, '/', as a prefix separator for X86. rdar://8741045
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121320 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-08 23:57:59 +00:00
Eric Christopher
7b5d456d5c Move this test to tlv* to make it easier to notice versus linux tls
support.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121316 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-08 23:33:23 +00:00
Jason W Kim
a0871e7927 ARM/MC/ELF TPsoft is now a proper pseudo inst.
Added test to check bl __aeabi_read_tp gets emitted properly for ELF/ASM
as well as ELF/OBJ (including fixup)

Also added support for ELF::R_ARM_TLS_IE32



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121312 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-08 23:14:44 +00:00
Evan Cheng
06d65f5156 Fix a bad prologue / epilogue codegen bug where the compiler would emit illegal
vpush instructions to save / restore VFP / NEON registers like this:
vpush {d8,d10,d11}
vpop {d8,d10,d11}

vpush and vpop do not allow gaps in the register list.
rdar://8728956


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121197 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 23:08:38 +00:00
Bruno Cardoso Lopes
ab8d53a56a Match a pattern generated by a dag combiner opt where:
(select (load (load tga0)) (load tga1)) => (load (select (load tga0) tga1))

Thanks to Akira for pointing that.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121163 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 19:00:20 +00:00
Rafael Espindola
545b77ef50 Fix absolute recording of differences of symbols in two sections. Reduced from ctor_dtor_count-2.cpp.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121152 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 17:12:32 +00:00
Rafael Espindola
3b3148f864 Fix relocations with weak definitions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121114 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 05:57:28 +00:00
NAKAMURA Takumi
85c3772253 Revert test/Archive/check_binary_output.ll". It fails on a buildbot.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121113 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 05:57:02 +00:00
Chris Lattner
1945d58025 reapply r121100 with a tweak to constant fold ConstExprs with TargetData
(if available) as we go so that we get simple constantexprs not insane ones.
This fixes the failure of clang/test/CodeGenCXX/virtual-base-ctor.cpp
that the previous iteration of this patch had.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121111 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 04:33:29 +00:00
Rafael Espindola
bf60dad984 Fix pcrel relocations that cross sections.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121107 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 03:50:14 +00:00
NAKAMURA Takumi
b9949b2970 test/Archive/check_binary_output.ll: Add a new test to check output of 'llvm-ar -p' is sane. Thanks to Danil Malyshev!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121106 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 03:35:20 +00:00
NAKAMURA Takumi
50004bedfa test/Other/close-stderr.ll: Require the feature 'shell'. It is not executable on Win32 but it is executable on MSYS-bash.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121105 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 02:43:58 +00:00
NAKAMURA Takumi
e7331eea2d test: Add the feature 'shell' on LLVM_ON_UNIX.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121104 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 02:43:51 +00:00
Eric Christopher
6a3e305326 Temporarily revert r121100 as it's causing clang to fail
CodeGenCXX/virtual-base-ctor.cpp.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121102 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 02:41:11 +00:00
Chris Lattner
fb431099c5 fix PR8710 - teach global opt that some constantexprs are too complex to
put in a global variable's initializer.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121100 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 01:59:32 +00:00
Michael J. Spencer
44d392a6af Test: Fix Support.Path and _all_ of the unittest death tests. GetTempPath defaults to \Windows\.
If I typed anything else it would just decline into cursing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121095 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 01:23:49 +00:00
Rafael Espindola
f10d2be573 Fix a crash reduced from gcc produced assembly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121085 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 01:09:54 +00:00
Owen Anderson
eb6779c5b9 Second attempt at converting Thumb2's LDRpci, including updating the gazillion places that need to know about it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121082 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 00:45:21 +00:00
Frits van Bommel
6033b346e2 Implement jump threading of 'indirectbr' by keeping track of whether we're looking for ConstantInt*s or BlockAddress*s.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121066 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 23:36:56 +00:00
Devang Patel
afeaae7a94 If dbg_declare() or dbg_value() is not lowered by isel then emit DEBUG message instead of creating DBG_VALUE for undefined value in reg0.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121059 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 22:39:26 +00:00
Wesley Peck
dc80380de8 Fixed reversed operands for IDIV and CMP instructions in MBlaze backend.
Use BRAD instead of BRD for indirect branches in MBlaze backend.

patch contributed by Jack Whitham!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121044 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 22:06:49 +00:00
Wesley Peck
1e8cdd599c Fix a 16-bit immediate value detection bug in the MBlaze delay slot filler.
Address more hazards in the MBlaze delay slot filler.

patch contributed by Jack Whitham!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121037 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 21:11:01 +00:00
Rafael Espindola
0bbe0b440e Second try at making direct object emission produce the same results
as llc + llvm-mc. This time ELF is not changed and I tested that llvm-gcc
bootstrap on darwin10 using darwin9's assembler and linker.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121006 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 17:27:56 +00:00
Rafael Espindola
6d86492f5e Revert previous two patches while I try to find out how to make both
linux and darwin assemblers happy :-(

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121004 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 15:35:15 +00:00
Rafael Espindola
7c00391248 Update test for the extra =.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121001 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 15:05:36 +00:00
Che-Liang Chiou
f964486771 ptx: add shift instructions
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120982 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 04:00:03 +00:00
Rafael Espindola
340a7a1c8f Don't use PadSectionToAlignment on windows.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120978 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 03:03:44 +00:00
Chris Lattner
cc10244d77 Fix PR8728, a miscompilation I recently introduced. When optimizing
memcpy's like:
  memcpy(A, B)
  memcpy(A, C)

we cannot delete the first memcpy as dead if A and C might be aliases.
If so, we actually get:

  memcpy(A, B)
  memcpy(A, A)

which is not correct to transform into:

  memcpy(A, A)

This patch was heavily influenced by Jakub Staszak's patch in PR8728, thanks
Jakub!



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120974 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 01:48:06 +00:00
Evan Cheng
48575f6ea7 Making use of VFP / NEON floating point multiply-accumulate / subtraction is
difficult on current ARM implementations for a few reasons.
1. Even though a single vmla has latency that is one cycle shorter than a pair
   of vmul + vadd, a RAW hazard during the first (4? on Cortex-a8) can cause
   additional pipeline stall. So it's frequently better to single codegen
   vmul + vadd.
2. A vmla folowed by a vmul, vmadd, or vsub causes the second fp instruction to
   stall for 4 cycles. We need to schedule them apart.
3. A vmla followed vmla is a special case. Obvious issuing back to back RAW
   vmla + vmla is very bad. But this isn't ideal either:
     vmul
     vadd
     vmla
   Instead, we want to expand the second vmla:
     vmla
     vmul
     vadd
   Even with the 4 cycle vmul stall, the second sequence is still 2 cycles
   faster.

Up to now, isel simply avoid codegen'ing fp vmla / vmls. This works well enough
but it isn't the optimial solution. This patch attempts to make it possible to
use vmla / vmls in cases where it is profitable.

A. Add missing isel predicates which cause vmla to be codegen'ed.
B. Make sure the fmul in (fadd (fmul)) has a single use. We don't want to
   compute a fmul and a fmla.
C. Add additional isel checks for vmla, avoid cases where vmla is feeding into
   fp instructions (except for the #3 exceptional case).
D. Add ARM hazard recognizer to model the vmla / vmls hazards.
E. Add a special pre-regalloc case to expand vmla / vmls when it's likely the
   vmla / vmls will trigger one of the special hazards.

Work in progress, only A+B are enabled.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120960 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 22:04:16 +00:00
Frits van Bommel
a4805cf6ef Fix PR 4170 by having ExtractValueInst::getIndexedType() reject out-of-bounds indexing.
Also add asserts that the indices are valid in InsertValueInst::init(). ExtractValueInst already asserts when constructed with invalid indices.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120956 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 20:50:26 +00:00
Frits van Bommel
7ac40c3ffa Teach SimplifyCFG to turn
(indirectbr (select cond, blockaddress(@fn, BlockA),
                            blockaddress(@fn, BlockB)))
into
  (br cond, BlockA, BlockB).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120943 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 18:29:03 +00:00
Chris Lattner
9637d5b22e Teach X86ISelLowering that the second result of X86ISD::UMUL is a flags
result.  This allows us to compile:

void *test12(long count) {
      return new int[count];
}

into:

test12:
	movl	$4, %ecx
	movq	%rdi, %rax
	mulq	%rcx
	movq	$-1, %rdi
	cmovnoq	%rax, %rdi
	jmp	__Znam                  ## TAILCALL

instead of:

test12:
	movl	$4, %ecx
	movq	%rdi, %rax
	mulq	%rcx
	seto	%cl
	testb	%cl, %cl
	movq	$-1, %rdi
	cmoveq	%rax, %rdi
	jmp	__Znam

Of course it would be even better if the regalloc inverted the cmov to 'cmovoq',
which would eliminate the need for the 'movq %rdi, %rax'.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120936 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 07:49:54 +00:00
Chris Lattner
b20e0b1fdd it turns out that when ".with.overflow" intrinsics were added to the X86
backend that they were all implemented except umul.  This one fell back
to the default implementation that did a hi/lo multiply and compared the
top.  Fix this to check the overflow flag that the 'mul' instruction
sets, so we can avoid an explicit test.  Now we compile:

void *func(long count) {
      return new int[count];
}

into:

__Z4funcl:                              ## @_Z4funcl
	movl	$4, %ecx                ## encoding: [0xb9,0x04,0x00,0x00,0x00]
	movq	%rdi, %rax              ## encoding: [0x48,0x89,0xf8]
	mulq	%rcx                    ## encoding: [0x48,0xf7,0xe1]
	seto	%cl                     ## encoding: [0x0f,0x90,0xc1]
	testb	%cl, %cl                ## encoding: [0x84,0xc9]
	movq	$-1, %rdi               ## encoding: [0x48,0xc7,0xc7,0xff,0xff,0xff,0xff]
	cmoveq	%rax, %rdi              ## encoding: [0x48,0x0f,0x44,0xf8]
	jmp	__Znam                  ## TAILCALL

instead of:

__Z4funcl:                              ## @_Z4funcl
	movl	$4, %ecx                ## encoding: [0xb9,0x04,0x00,0x00,0x00]
	movq	%rdi, %rax              ## encoding: [0x48,0x89,0xf8]
	mulq	%rcx                    ## encoding: [0x48,0xf7,0xe1]
	testq	%rdx, %rdx              ## encoding: [0x48,0x85,0xd2]
	movq	$-1, %rdi               ## encoding: [0x48,0xc7,0xc7,0xff,0xff,0xff,0xff]
	cmoveq	%rax, %rdi              ## encoding: [0x48,0x0f,0x44,0xf8]
	jmp	__Znam                  ## TAILCALL

Other than the silly seto+test, this is using the o bit directly, so it's going in the right
direction.




git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120935 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 07:30:36 +00:00
Chris Lattner
777dd07394 fix the rest of the linux miscompares :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120933 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 02:08:07 +00:00
Chris Lattner
96908b17ae generalize the previous check to handle -1 on either side of the
select, inserting a not to compensate.  Add a missing isZero check
that I lost somehow.

This improves codegen of:

void *func(long count) {
      return new int[count];
}

from:

__Z4funcl:                              ## @_Z4funcl
	movl	$4, %ecx                ## encoding: [0xb9,0x04,0x00,0x00,0x00]
	movq	%rdi, %rax              ## encoding: [0x48,0x89,0xf8]
	mulq	%rcx                    ## encoding: [0x48,0xf7,0xe1]
	testq	%rdx, %rdx              ## encoding: [0x48,0x85,0xd2]
	movq	$-1, %rdi               ## encoding: [0x48,0xc7,0xc7,0xff,0xff,0xff,0xff]
	cmoveq	%rax, %rdi              ## encoding: [0x48,0x0f,0x44,0xf8]
	jmp	__Znam                  ## TAILCALL
                                        ## encoding: [0xeb,A]

to:

__Z4funcl:                              ## @_Z4funcl
	movl	$4, %ecx                ## encoding: [0xb9,0x04,0x00,0x00,0x00]
	movq	%rdi, %rax              ## encoding: [0x48,0x89,0xf8]
	mulq	%rcx                    ## encoding: [0x48,0xf7,0xe1]
	cmpq	$1, %rdx                ## encoding: [0x48,0x83,0xfa,0x01]
	sbbq	%rdi, %rdi              ## encoding: [0x48,0x19,0xff]
	notq	%rdi                    ## encoding: [0x48,0xf7,0xd7]
	orq	%rax, %rdi              ## encoding: [0x48,0x09,0xc7]
	jmp	__Znam                  ## TAILCALL
                                        ## encoding: [0xeb,A]



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120932 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 02:00:51 +00:00
Chris Lattner
c8c20d1486 relax this to handle linux defaulting to -static.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120930 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 01:31:13 +00:00
Chris Lattner
a2b5600e61 Improve an integer select optimization in two ways:
1. generalize 
    (select (x == 0), -1, 0) -> (sign_bit (x - 1))
to:
    (select (x == 0), -1, y) -> (sign_bit (x - 1)) | y

2. Handle the identical pattern that happens with !=:
   (select (x != 0), y, -1) -> (sign_bit (x - 1)) | y

cmov is often high latency and can't fold immediates or
memory operands.  For example for (x == 0) ? -1 : 1, before 
we got:

< 	testb	%sil, %sil
< 	movl	$-1, %ecx
< 	movl	$1, %eax
< 	cmovel	%ecx, %eax

now we get:

> 	cmpb	$1, %sil
> 	sbbl	%eax, %eax
> 	orl	$1, %eax




git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120929 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 01:23:24 +00:00
Chris Lattner
bced6a1b8f merge some tests into select.ll and make them more specific.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120928 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 01:13:58 +00:00
Chris Lattner
bbdabf411b rename test
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120927 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 01:02:23 +00:00
Chris Lattner
63d7c17ff1 remove two tests that aren't really testing anything.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120926 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 01:02:13 +00:00