Commit Graph

1488 Commits

Author SHA1 Message Date
Anton Korobeynikov
051cfd683f Fix some typos and use type-based isel for VZIP/VUZP/VTRN
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79625 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-21 12:41:42 +00:00
Anton Korobeynikov
1c8e581832 Add lowering of ARM 4-element shuffles to multiple instructios via perfectshuffle-generated table.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79624 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-21 12:41:24 +00:00
Anton Korobeynikov
62e84f177d Add nodes & dummy matchers for some v{zip,uzp,trn} instructions
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79622 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-21 12:40:50 +00:00
Anton Korobeynikov
8e6c2b9041 Expand EXTRACT_SUBVECTOR
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79621 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-21 12:40:35 +00:00
Anton Korobeynikov
5da894f5c4 Provide vext.{16,32}
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79620 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-21 12:40:21 +00:00
Anton Korobeynikov
d0ac234b1b Use masks not nodes for vector shuffle predicates. Provide set of 'legal' masks, so legalizer won't infinite cycle
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79619 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-21 12:40:07 +00:00
Bob Wilson
d4b4cf524b Remove Neon intrinsics for VZIP, VUZP, and VTRN. We will represent these as
vector shuffles.  Temporarily remove the tests for these operations until the
new implementation is working.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79579 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-21 00:01:42 +00:00
Evan Cheng
89d177f017 Fix an obvious copy-n-paste bug.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79535 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-20 17:01:04 +00:00
David Goodwin
5d598aaf3d Update Cortex-A8 instruction itineraries for integer instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79436 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-19 18:00:44 +00:00
Bob Wilson
de95c1b88b Add support for Neon VEXT (vector extract) shuffles.
This is derived from a patch by Anton Korzh.  I modified it to recognize
the VEXT shuffles during legalization and lower them to a target-specific
DAG node.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79428 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-19 17:03:43 +00:00
Chris Lattner
6c2f9e14fd eliminate AsmPrinter::SwitchToSection and just have clients
talk to the MCStreamer directly instead.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79405 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-19 05:49:37 +00:00
Jakob Stoklund Olesen
c0823fe7c6 Simplify RegScavenger::FindUnusedReg.
- Drop the Candidates argument and fix all callers. Now that RegScavenger
  tracks available registers accurately, there is no need to restict the
  search.
- Make sure that no aliases of the found register are in use. This was a potential bug.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79369 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-18 21:14:54 +00:00
Evan Cheng
51f39961c3 Fix revsh pattern.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79318 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-18 05:43:23 +00:00
Benjamin Kramer
9ae7d44d95 Fix use after free in Thumb2SizeReduction (PR4707). A MachineInstr was used after erasure.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79189 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-16 11:56:42 +00:00
Bill Wendling
af56634058 Reapply r79127. It was fixed by d0k.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79136 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-15 21:21:19 +00:00
Bill Wendling
f865ea85bd Revert r79127. It was causing compilation errors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79135 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-15 21:14:01 +00:00
Evan Cheng
088880cb19 Change allowsUnalignedMemoryAccesses to take type argument since some targets
support unaligned mem access only for certain types. (Should it be size
instead?)

ARM v7 supports unaligned access for i16 and i32, some v6 variants support it
as well.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79127 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-15 19:23:44 +00:00
Evan Cheng
bc9b754091 Turn on if-conversion for thumb2.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79084 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-15 07:59:10 +00:00
Evan Cheng
010b1b9e7b Do not use frame register to reference fixed stack objects if the function is frameless.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79067 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-15 02:05:35 +00:00
Evan Cheng
98a0104014 Leaf functions which do not save CSRs can be frameless even with -disable-fp-elim.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79039 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 20:48:13 +00:00
Anton Korobeynikov
72977a45a8 Allow targets to specify their choice of calling conventions per
libcall. Take advantage of this in the ARM backend to rectify broken
choice of CC when hard float is in effect. PIC16 may want to see if
it could be of use in MakePIC16Libcall, which works unchanged.

Patch by Sandeep!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79033 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 20:10:52 +00:00
Evan Cheng
e6c835f424 Add Thumb2 lsr hooks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79032 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 20:09:37 +00:00
Evan Cheng
59bc0604e5 80 col violation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79026 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 19:11:20 +00:00
Evan Cheng
bba9f5f378 Indentation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79022 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 19:01:37 +00:00
Evan Cheng
31b99dd760 Also shrink immediate branches; also more assembler workarounds.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79014 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 18:31:44 +00:00
Bob Wilson
22cac0d9b3 Now that all the legal Neon shuffles (or at least the ones that have been
implemented so far) are recognized during legalization, it is easy to fall
back to the default expansion for other shuffles.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78995 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 05:16:33 +00:00
Bob Wilson
c1d287b4b7 Create a new ARM-specific DAG node, VDUP, to represent a splat from a
scalar_to_vector.  Generate these VDUP nodes during legalization instead
of trying to recognize the pattern during selection.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78994 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 05:13:08 +00:00
Bob Wilson
0ce3710825 During legalization, change Neon vdup_lane operations from shuffles to
target-specific VDUPLANE nodes.  This allows the subreg handling for the
quad-register version to be done easily with Pats in the .td file, instead
of with custom code in ARMISelDAGToDAG.cpp.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78993 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 05:08:32 +00:00
Evan Cheng
a1efbbdbf3 Shrink ADR and LDR from constantpool late during constantpool island pass.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78970 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 00:32:16 +00:00
Evan Cheng
1135a232eb New entry.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78968 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-14 00:16:47 +00:00
Owen Anderson
1d0be15f89 Push LLVMContexts through the IntegerType APIs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78948 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 21:58:54 +00:00
Daniel Dunbar
b42dad4761 Revert 78892 and 78895, these break generating working executables on
x86_64-apple-darwin10.

--- Reverse-merging r78895 into '.':
U    test/CodeGen/PowerPC/2008-12-12-EH.ll
U    lib/Target/DarwinTargetAsmInfo.cpp
--- Reverse-merging r78892 into '.':
U    include/llvm/Target/DarwinTargetAsmInfo.h
U    lib/Target/X86/X86TargetAsmInfo.cpp
U    lib/Target/X86/X86TargetAsmInfo.h
U    lib/Target/ARM/ARMTargetAsmInfo.h
U    lib/Target/ARM/ARMTargetMachine.cpp
U    lib/Target/ARM/ARMTargetAsmInfo.cpp
U    lib/Target/PowerPC/PPCTargetAsmInfo.cpp
U    lib/Target/PowerPC/PPCTargetAsmInfo.h
U    lib/Target/PowerPC/PPCTargetMachine.cpp
G    lib/Target/DarwinTargetAsmInfo.cpp


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78919 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 17:03:38 +00:00
Jim Grosbach
f35d21617e Add missing defs of R2 and D1.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78918 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 16:59:44 +00:00
David Goodwin
6d3d9c3fc3 Finalize itineraries for cortex-a8 integer multiply
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78908 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 15:51:13 +00:00
Jim Grosbach
8db5cce021 Remove unnecessary newline
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78905 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 15:12:16 +00:00
Jim Grosbach
1add659b0a Correct comment wording
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78904 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 15:11:43 +00:00
Evan Cheng
48bd7e3bbc tPOP_RET now has predicate operands.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78898 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 06:05:07 +00:00
Bob Wilson
bfcbb507c2 Add a fixme message about canonicalizing floating-point vector types.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78897 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 06:01:30 +00:00
Bob Wilson
bab812b4b0 Revert r78852 for now. I want to do this differently, but I don't have time
to fix it tonight.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78896 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 05:58:56 +00:00
Evan Cheng
86e5f7b6f8 It's ok to spill a tGPR register as long as it's still allocated a low register.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78893 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 05:40:51 +00:00
Chris Lattner
b2d3169d96 fix a minor fixme. When building with SL and later tools, the ".eh" symbols
don't need to be exported from the .o files.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78892 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 05:30:22 +00:00
Bruno Cardoso Lopes
b808588a3a Change MCSectionELF to represent a section semantically instead of
syntactically as a string, very similiar to what Chris did with MachO.
The parsing support and validation is not introduced yet.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78890 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 05:07:35 +00:00
Bob Wilson
28865062c1 Add a comment to describe why vector shuffles are legalized to custom DAG nodes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78884 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 02:13:04 +00:00
Bob Wilson
d06791f6d0 Use cast<> instead of dyn_cast<> in places where the type is known.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78881 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 01:57:47 +00:00
Dan Gohman
cf20ac4fd1 Various AsmWriter output cleanups. Use WriteAsOperand instead of
PrintUnmangledNameSafely.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78878 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-13 01:36:44 +00:00
Bob Wilson
af385baa1d Recognize Neon VDUP shuffles during legalization instead of selection.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78852 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-12 22:54:19 +00:00
Bob Wilson
d8e1757eac Recognize Neon VREV shuffles during legalization instead of selection.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78850 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-12 22:31:50 +00:00
Dan Gohman
a9ad04191c This void is implicit in C++.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78848 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-12 22:10:57 +00:00
Bob Wilson
114a266c94 Generate Neon VTBL and VTBX instructions from the corresponding intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78835 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-12 20:51:55 +00:00
Evan Cheng
3aaccffbce PredCC is meant to be 2 bits wide, like PredCC1.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78829 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-12 18:35:50 +00:00