Commit Graph

49115 Commits

Author SHA1 Message Date
Bruno Cardoso Lopes
3b86598cfa Instead of always leaving the work to the generic legalizer when
there is no support for native 256-bit shuffles, be more smart in some
cases, for example, when you can extract specific 128-bit parts and use
regular 128-bit shuffles for them. Example:

For this shuffle:
  shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32>
                <i32 1, i32 0, i32 7, i32 6>

This was expanded to:
  vextractf128  $1, %ymm1, %xmm2
  vpextrq $0, %xmm2, %rax
  vmovd %rax, %xmm1
  vpextrq $1, %xmm2, %rax
  vmovd %rax, %xmm2
  vpunpcklqdq %xmm1, %xmm2, %xmm1
  vpextrq $0, %xmm0, %rax
  vmovd %rax, %xmm2
  vpextrq $1, %xmm0, %rax
  vmovd %rax, %xmm0
  vpunpcklqdq %xmm2, %xmm0, %xmm0
  vinsertf128 $1, %xmm1, %ymm0, %ymm0
  ret

Now we get:
  vshufpd $1, %xmm0, %xmm0, %xmm0
  vextractf128  $1, %ymm1, %xmm1
  vshufpd $1, %xmm1, %xmm1, %xmm1
  vinsertf128 $1, %xmm1, %ymm0, %ymm0

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137733 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 18:21:54 +00:00
Devang Patel
e26a135248 Remove unnecessary version check.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137728 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 17:41:41 +00:00
Jim Grosbach
d0d3f7e01f ARM .align NOP padding uses different encoding pre-ARMv6.
Patch by Kristof Beyls and James Malloy.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137723 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 17:06:20 +00:00
Nadav Rotem
6d1f5d951b Revert r137562 because it caused PR10674
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137719 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 14:34:29 +00:00
David Chisnall
7a817eae6c Add a mechanism for optimisation plugins to register passes that all front ends can use without needing to be aware of the plugin (or the plugin be aware of the front end).
Before 3.0, I'd like to add a mechanism for automatically loading a set of plugins from a config file.  API suggestions welcome...



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137717 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 13:58:41 +00:00
Bill Wendling
05d0023a95 A few places where we want to skip the landingpad instruction for insertion.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137712 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 04:52:55 +00:00
Akira Hatanaka
614051a1c5 Fix handling of double precision loads and stores when Mips1 is targeted.
Mips1 does not support double precision loads or stores, therefore two single
precision loads or stores must be used in place of these instructions. This 
patch treats double precision loads and stores as if they are legal
instructions until MCInstLowering, instead of generating the single precision
instructions during instruction selection or Prolog/Epilog code insertion.

Without the changes made in this patch, llc produces code that has the same 
problem described in r137484 or bails out when
MipsInstrInfo::storeRegToStackSlot or loadRegFromStackSlot is called before
register allocation.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137711 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 03:51:51 +00:00
Akira Hatanaka
8957481e6a Define function MipsMCInstLower::LowerOperand.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137707 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 02:21:03 +00:00
Akira Hatanaka
78d1b11aa5 Add parameter Offset to MipsMCInstLower::LowerSymbolOperand.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137706 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 02:15:03 +00:00
Eli Friedman
2199dfb0e6 Revert a bit of r137667; the logic in question can safely handle atomic load/store.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137702 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 01:28:22 +00:00
Eli Friedman
6437f2c2e2 After talking with Bill, it seems like the LandingPad handling here is likely
to be wrong (or at least somewhat suspect).  Leave a FIXME for Bill.




git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137694 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 00:41:37 +00:00
Eli Friedman
33cb445fb6 Minor comment fixes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137693 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-16 00:20:11 +00:00
Eli Friedman
8176388d65 Update SimplifyCFG for atomic operations.
This commit includes a mention of the landingpad instruction, but it's not
changing the behavior around it.  I think the current behavior is correct,
though.  Bill, can you double-check that?



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137691 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 23:59:28 +00:00
Eli Friedman
fd06b3cfa1 Add comments and test for atomic load/store and mem2reg.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137690 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 23:55:52 +00:00
Devang Patel
02e603f87a Refactor.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137689 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 23:47:24 +00:00
Owen Anderson
ef2865a8ea Specify a necessary fixed bit for VLD3DUP, and otherwise rearrange the Thumb2 NEON decoding hooks to bring us closer to correctness.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137686 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 23:38:54 +00:00
Bruno Cardoso Lopes
8400bfe9fa While I'm here, remove the "_alt" hacks to a series of INSERT_SUBREG and
also add the AVX versions of the 128-bit patterns

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137685 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 23:36:51 +00:00
Bruno Cardoso Lopes
1deddbbd56 Reorder declarations of vmovmskp* and also put the necessary AVX
predicate and TB encoding fields. This fix the encoding for the
attached testcase. This fixes PR10625.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137684 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 23:36:45 +00:00
Devang Patel
3655a21b15 Continue to hoist uses of getCompileUnit() up. The goal is to get rid of uses of getCompileUnit().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137683 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 23:36:40 +00:00
Bill Wendling
5e38c473a4 In places where it's using "getFirstNonPHI", skip the landingpad instruction if necessary.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137679 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 23:19:54 +00:00
Jim Grosbach
19cb7f491f MCTargetAsmParser target match predicate support.
Allow a target assembly parser to do context sensitive constraint checking
on a potential instruction match. This will be used, for example, to handle
Thumb2 IT block parsing.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137675 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 23:03:29 +00:00
Devang Patel
6326a4238d Add a finalize() hook, that'll let DIBuilder construct compile unit lazily.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137673 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 23:00:00 +00:00
Bill Wendling
6456a62178 Don't sink the instruction to before a landingpad instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137672 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 22:53:05 +00:00
Devang Patel
d30243402b This is somewhat déjà-vu, but avoid using getCompileUnit() as much as possible.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137668 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 22:24:32 +00:00
Eli Friedman
3d30b435e2 Update inter-procedural optimizations for atomic load/store.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137667 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 22:16:46 +00:00
Eli Friedman
cc4a0435b7 Update instcombine for atomic load/store.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137664 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 22:09:40 +00:00
Devang Patel
d0b5a5ece4 Refactor. Variables are part of compile unit so let CompileUnit create new variable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137663 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 22:04:40 +00:00
Eli Friedman
c869b4397a Add some comments here because the lack of a check for volatile/atomic here is a bit unusual.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137662 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 21:56:39 +00:00
Bruno Cardoso Lopes
50b37c7920 Fix PR10656. It's only profitable to use 128-bit inserts and extracts
when AVX mode is one. Otherwise is just more work for the type
legalizer.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137661 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 21:45:54 +00:00
Devang Patel
7a328270be There is no need to maintain a set to keep track of variables that use location expressions. In such cases, AT_location attribute's value will be a label.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137659 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 21:43:21 +00:00
Devang Patel
c890b19226 Fix warning.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137658 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 21:35:16 +00:00
Devang Patel
ff9dd0ac55 Simplify. Let DbgVariable keep track of variable's DBG_VALUE machine instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137656 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 21:24:36 +00:00
Bill Wendling
9d6070f161 Duncan pointed out that the LandingPadInst might read memory. (It might also
write to memory.) Marking it as such makes some checks for immobility go away.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137655 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 21:14:31 +00:00
Eli Friedman
0455bb89cd Fix llvm::CloneModule to correctly clone globals. Patch per bug report by Simon Moll on llvmdev.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137654 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 21:05:06 +00:00
Eli Friedman
e5e771263a Fix predicates methods on Instruction to handle atomic load/store correctly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137652 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 21:00:18 +00:00
Eli Friedman
667ccf231b Misc analysis passes that need to be aware of atomic load/store.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137650 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 20:54:19 +00:00
Eli Friedman
97671565ff Atomic load/store support in LICM.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137648 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 20:52:09 +00:00
Owen Anderson
c537f3be0c Enforce the constraint that Rt must be even on LDRD/STRD instructions in ARM mode. Update tests to reflect this fact.
Patch by James Molloy.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137647 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 20:51:32 +00:00
Owen Anderson
92be8ab6a4 Remove dead classes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137643 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 20:11:11 +00:00
Bill Wendling
187b1924a4 The "landingpad" instruction will never be "trivially" dead.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137642 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 20:10:51 +00:00
Owen Anderson
5df7ef6cdb Fix incorrect encoding of UMAAL and friends. Patch by James Molloy.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137641 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 20:08:25 +00:00
Devang Patel
5a1a67cd3f Simplify mapping to variable from its abstract variable info.
When a variable is inlined multiple places, abstract variable keeps name, location, type etc.. info and all other concreate instances of the variable directly refers to abstract variable.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137637 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 19:01:20 +00:00
Owen Anderson
305e046e53 Fix decoding LDRSB and LDRSH in Thumb1 mode. Patch by James Molloy.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137636 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 19:00:06 +00:00
Owen Anderson
7a2e1770ea Fix problems decoding the to/from-lane NEON memory instructions, and add a comprehensive NEON decoding testcase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137635 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 18:44:44 +00:00
Devang Patel
a098c506a2 Refactor.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137632 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 18:40:16 +00:00
Devang Patel
59bc4093d5 Refactor.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137631 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 18:35:42 +00:00
Bill Wendling
06f20ebfcf Don't try to sink the landingpad instruction. It's immobile.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137629 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 18:23:40 +00:00
Bill Wendling
da005e7d20 The landingpad instruction isn't loop-invariant.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137628 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 18:22:49 +00:00
Bill Wendling
bc4fcbbc7b Mark the SCC as "might unwind" if we run into a 'resume' instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137627 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 18:22:00 +00:00
Bill Wendling
1178fb4701 Skip the insertion iterator past the landingpad instruction if there.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137626 91177308-0d34-0410-b5e6-96231b3b80d8
2011-08-15 18:21:07 +00:00