Commit Graph

6390 Commits

Author SHA1 Message Date
Chris Lattner
beac75da37 implement rdar://6653118 - fastisel should fold loads where possible.
Since mem2reg isn't run at -O0, we get a ton of reloads from the stack,
for example, before, this code:

int foo(int x, int y, int z) {
  return x+y+z;
}

used to compile into:

_foo:                                   ## @foo
	subq	$12, %rsp
	movl	%edi, 8(%rsp)
	movl	%esi, 4(%rsp)
	movl	%edx, (%rsp)
	movl	8(%rsp), %edx
	movl	4(%rsp), %esi
	addl	%edx, %esi
	movl	(%rsp), %edx
	addl	%esi, %edx
	movl	%edx, %eax
	addq	$12, %rsp
	ret

Now we produce:

_foo:                                   ## @foo
	subq	$12, %rsp
	movl	%edi, 8(%rsp)
	movl	%esi, 4(%rsp)
	movl	%edx, (%rsp)
	movl	8(%rsp), %edx
	addl	4(%rsp), %edx    ## Folded load
	addl	(%rsp), %edx     ## Folded load
	movl	%edx, %eax
	addq	$12, %rsp
	ret

Fewer instructions and less register use = faster compiles.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113102 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-05 02:18:34 +00:00
Chris Lattner
17aa68055b zap dead code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113073 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 18:12:00 +00:00
Bruno Cardoso Lopes
2eb63dfa0a Remove the last bit of isShuffleMaskLegal checks and improve the comment regarding mmx shuffles
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113059 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 02:58:56 +00:00
Bruno Cardoso Lopes
828f6ae03c make explicit that we not handle several mmx shuffles
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113058 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 02:50:13 +00:00
Bruno Cardoso Lopes
aace0f295b Emit target specific nodes to handle palignr. Do not touch it for MMX versions yet.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113056 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 02:36:07 +00:00
Bruno Cardoso Lopes
c800c0d25f Emit target specific nodes to handle splats starting at zero indicies
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113055 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 02:02:14 +00:00
Bruno Cardoso Lopes
bbfc31012b Emit target specific nodes for isPSHUFHWMask and isPSHUFLWMask
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113050 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 01:36:45 +00:00
Bruno Cardoso Lopes
4c827f5ae1 Emit target specific nodes for isSHUFPMask
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113048 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 01:22:57 +00:00
Bruno Cardoso Lopes
d344f28b9d Previous isMOVLMask matching already emits targets nodes, remove check
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113047 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 00:50:08 +00:00
Bruno Cardoso Lopes
e09abcd3c4 One more check from the original isShuffleMaskLegal goes away
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113045 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 00:46:16 +00:00
Bruno Cardoso Lopes
b733996110 Remove a duplicated but useless check that i've inserted in the previous commit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113044 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 00:43:12 +00:00
Bruno Cardoso Lopes
a22c84571a Refactor some code and remove the extra checks for unpckl_undef and unpckh_undef
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113043 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-04 00:39:43 +00:00
Bruno Cardoso Lopes
43c05744b5 Remove check for unpckh mask
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113035 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 23:32:47 +00:00
Bruno Cardoso Lopes
ef3adb3243 Remove check for unpckl mask
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113034 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 23:31:50 +00:00
Bruno Cardoso Lopes
7256e22f77 Inline isShuffleMaskLegal into LowerVECTOR_SHUFFLE, so we can start
checking each standalone condition and decide whether emit target
specific nodes or remove the condition if it's already matched before.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113031 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 23:24:06 +00:00
Bruno Cardoso Lopes
e8f279cbd4 Reapply considered harmfull part of rr112934 and r112942.
"Use target specific nodes instead of relying in unpckl and
unpckh pattern fragments during isel time. Also place a
depth limit in getShuffleScalarElt.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113020 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 22:09:41 +00:00
Dale Johannesen
caa9ba228d Remove the rest of the nonexistent 64-bit AVX instructions.
Bruno, please review.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113014 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 21:23:00 +00:00
Bruno Cardoso Lopes
190d0a54c1 Reapply last harmless part of r112934, the pattern fragment to match X86Unpcklpd
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113009 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 20:44:26 +00:00
Bruno Cardoso Lopes
2a4460606e Reintroduce a simple function refactoring done in r112934, also without any functionality changes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113008 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 20:20:02 +00:00
Bruno Cardoso Lopes
be8b084d8a Reapply piecies of r112942 and r112934 which don't do
functional changes


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113007 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 20:10:35 +00:00
Bruno Cardoso Lopes
b3e0669b8e Reapply Fix comment
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113006 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 19:55:05 +00:00
Daniel Dunbar
3139422058 Revert r112934, "- Use specific nodes to match unpckl masks.", which introduced
some infinite loop and select failures.
 - Apologies for eager reverting, but its branch day.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113000 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 19:38:11 +00:00
Daniel Dunbar
78541f258c Revert r112938 "Fix comment", which depends on r112934, which introduced some
infinite loop and select failures.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112999 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 19:38:08 +00:00
Daniel Dunbar
a87ccce95b Revert r112942, "Use punpckh and unpckh family of nodes instead of using unpckh
mask pattern fragment", which depends on r112934, which introduced some infinite
loop and select failures.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112998 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 19:38:05 +00:00
Bruno Cardoso Lopes
95f1e2d6b5 AVX doesn't support mm operations neither its instrinsics.
The AVX versions of PALIGN and PABS* should only exist for
128-bit. Remove the unnecessary stuff.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112944 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 02:08:45 +00:00
Bruno Cardoso Lopes
4b0c9f3e73 Use punpckh and unpckh family of nodes instead of using unpckh mask pattern fragment
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112942 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 01:39:08 +00:00
Bruno Cardoso Lopes
01f0847ce8 Fix comment
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112938 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 01:28:51 +00:00
Bruno Cardoso Lopes
5e5342b0a8 - Use specific nodes to match unpckl masks.
- Teach getShuffleScalarElt how to handle more target
specific nodes, so the DAGCombine can make use of it.
- Add another hack to avoid the node update problem
during legalization. More description on the comments



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112934 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 01:24:00 +00:00
Jakob Stoklund Olesen
3061c4442e Don't call Predicate_* from X86 target.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112921 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-03 00:35:18 +00:00
Anton Korobeynikov
ace53f2fbc Properly emit __chkstk call instead of __alloca on non-mingw windows targets.
Patch by Cameron Esfahani!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112902 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 23:03:46 +00:00
Bruno Cardoso Lopes
ed5c711a6e Move insertps mask decoding to header file
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112896 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 22:43:39 +00:00
Anton Korobeynikov
c7c62bb3ca Revert win64 changes. They seem to be incomplete
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112885 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 22:31:32 +00:00
Anton Korobeynikov
2f4fad99ea Properly allocate win64 shadow reg area.
Patch by Jan Sjodin!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112875 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 22:16:28 +00:00
Bruno Cardoso Lopes
5594560766 Move decoding of insertps back to avoid unused warnings in x86 isel lowering, and fix movlhps/movhlps to decode 4 elements shuffles
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112869 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 21:51:11 +00:00
Dan Gohman
24bde5bce1 Don't narrow the load and store in a load+twiddle+store sequence unless
there are clearly no stores between the load and the store. This fixes
this miscompile reported as PR7833.

This breaks the test/CodeGen/X86/narrow_op-2.ll optimization, which is
safe, but awkward to prove safe. Move it to X86's README.txt.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112861 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 21:18:42 +00:00
Bruno Cardoso Lopes
6b1d0a3b36 Move x86 specific shuffle mask decoding to its own header, it's also going to be used elsewhere. Also trim trailing whitespaces
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112846 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 18:40:13 +00:00
Bruno Cardoso Lopes
3722f007b6 Replace unpckl_undef and unpckh_undef matching with target specific opcodes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112806 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 05:23:12 +00:00
Bruno Cardoso Lopes
dd69db858c Move condition out to prepare for more matching
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112805 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 04:20:26 +00:00
Bruno Cardoso Lopes
ad10fb2b56 Remove checking for isUNPCKL_v_undef_Mask, the specific node is already emitted for it
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112804 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 03:57:58 +00:00
Bruno Cardoso Lopes
d00bfe1f8d become more strict about when it's safe to use X86ISD::MOVLPS
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112799 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-02 02:35:51 +00:00
Bruno Cardoso Lopes
4783a3ee13 Revert r112689, avoid those kind of checks cause they mess up with mmx
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112760 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-01 22:59:03 +00:00
Bruno Cardoso Lopes
29c353b9c3 Using target specific nodes for shuffle nodes makes the mask
check more strict, breaking some cases not checked in the
testsuite, but also exposes some foldings not done before,
as this example:

  movaps  (%rdi), %xmm0
  movaps  (%rax), %xmm1
  movaps  %xmm0, %xmm2
  movss %xmm1, %xmm2
  shufps  $36, %xmm2, %xmm0

now is generated as:

  movaps  (%rdi), %xmm0
  movaps  %xmm0, %xmm1
  movlps  (%rax), %xmm1
  shufps  $36, %xmm1, %xmm0



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112753 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-01 22:33:20 +00:00
Bruno Cardoso Lopes
56098f5d26 Use movlps, movlpd, movss and movsd specific nodes instead of pattern matching with movlp pattern fragment
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112694 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-01 05:08:25 +00:00
Bruno Cardoso Lopes
9cfad89a68 minor change, simplify some logic
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112689 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-01 00:57:08 +00:00
Bruno Cardoso Lopes
e654b56eb1 Move some functions around so they can be used for some other to come function
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112687 91177308-0d34-0410-b5e6-96231b3b80d8
2010-09-01 00:51:36 +00:00
Bruno Cardoso Lopes
013bb3dee9 Use x86 specific MOVSLDUP node, add more patterns to match it and remove useless load nodes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112661 91177308-0d34-0410-b5e6-96231b3b80d8
2010-08-31 22:35:05 +00:00
Bruno Cardoso Lopes
5023ef281c Use x86 specific MOVSHDUP node and add more patterns to match it
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112657 91177308-0d34-0410-b5e6-96231b3b80d8
2010-08-31 22:22:11 +00:00
Jakob Stoklund Olesen
63b1dbaadd Make %EFLAGS unallocatable.
No CCR virtual registers should exist, and %EFLAGS is used in ways that can
surprise RegAllocFast.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112650 91177308-0d34-0410-b5e6-96231b3b80d8
2010-08-31 21:51:07 +00:00
Bruno Cardoso Lopes
7ff30bb1a5 Use MOVHLPS node instead of matching using movhlps and movhlps_undef pattern fragments
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112644 91177308-0d34-0410-b5e6-96231b3b80d8
2010-08-31 21:38:49 +00:00
Bruno Cardoso Lopes
f2db5b48d0 Use MOVLHPS and MOVHLPS x86 nodes whenever possible. Also remove some useless nodes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112642 91177308-0d34-0410-b5e6-96231b3b80d8
2010-08-31 21:15:21 +00:00