Benjamin Kramer [Wed, 13 Jan 2010 12:45:23 +0000 (12:45 +0000)]
Introduce Twine::toStringRef, a variant of toVector which avoids the copy if the
twine can be represented as a single StringRef. Use the new methode to simplify
some twine users.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93317
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 13 Jan 2010 08:45:40 +0000 (08:45 +0000)]
Re-enable extension optimization pass.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93313
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 08:08:33 +0000 (08:08 +0000)]
upgrade and MC'ize a few uses of makeNameProper.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93310
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 08:04:24 +0000 (08:04 +0000)]
MC'ize this a bit and upgrade APIs
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93309
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 08:02:14 +0000 (08:02 +0000)]
add a fixme, ELF MCSection isn't quite right and weak unnamed globals are broken
on linux (even though they are pointless, they shouldn't ICE).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93308
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 13 Jan 2010 08:01:32 +0000 (08:01 +0000)]
For now, avoid issuing extract_subreg to reuse lower 8-bit, it's not safe in 32-bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93307
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 13 Jan 2010 07:59:13 +0000 (07:59 +0000)]
Add comment; refactor; avoid pulling in DT if it's not used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93306
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 07:56:59 +0000 (07:56 +0000)]
eliminate some uses of Mangler::makeNameProper.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93305
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 07:50:21 +0000 (07:50 +0000)]
don't add the \1 to the name.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93304
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 07:30:49 +0000 (07:30 +0000)]
remove uses of deprecated functions, this generates slightly
different BlockAddress labels, but nothing semantically important.
Add a FIXME that BlockAddress codegen is broken if the LLVM BB has
an empty name (e.g. strip was run).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93303
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 07:16:53 +0000 (07:16 +0000)]
use the new form of getNameWithPrefix, not makeNameProper.
Among other things, this would do very weird things if the
basic block name had (e.g.) a space in it on darwin:
makeNameProper would add quotes, then the mcsymbol would
escape the quotes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93302
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 07:12:06 +0000 (07:12 +0000)]
add new isSingleStringRef()/getSingleStringRef() methods to twine,
and use them to avoid a copy of a string in getNameWithPrefix in
the common case. It seems like Value::setName and other places
should use this as well?
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93301
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 07:01:09 +0000 (07:01 +0000)]
ugh, my last patch just sped up a method and changed all the clients
that I want to completely eliminate. Add fixme's so I remember this
in the future, and add the missing helper that they should be upgraded
to use instead.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93300
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 06:38:18 +0000 (06:38 +0000)]
change Mangler::makeNameProper to return its result in a SmallVector
instead of returning it in an std::string. Based on this change:
1. Change TargetLoweringObjectFileCOFF::getCOFFSection to take a StringRef
2. Change a bunch of targets to call makeNameProper with a smallstring,
making several of them *much* more efficient.
3. Rewrite Mangler::makeNameProper to not build names and then prepend
prefixes, not use temporary std::strings, and to avoid other crimes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93298
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 05:02:57 +0000 (05:02 +0000)]
my mistake, Mangler::makeNameProper wants to take a twine, not a stringref!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93296
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 04:55:33 +0000 (04:55 +0000)]
change makeNameProper to take a stringref instead of std::string.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93295
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 04:50:20 +0000 (04:50 +0000)]
give StringRef a const_iterator member.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93294
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 04:39:46 +0000 (04:39 +0000)]
fix assert in AsmPrinter::EmitGlobalConstantLargeInt to match reality.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93293
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 04:38:16 +0000 (04:38 +0000)]
reduce nesting and code duplication in AsmPrinter::EmitGlobalConstantLargeInt.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93292
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 04:34:19 +0000 (04:34 +0000)]
reduce indentation and add a fast-path to EmitGlobalConstant for 8-byte
integers on 64-bit systems.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93291
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 13 Jan 2010 04:29:19 +0000 (04:29 +0000)]
reduce indentation and use early exits in AsmPrinter::EmitConstantValueOnly
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93290
91177308-0d34-0410-b5e6-
96231b3b80d8
Victor Hernandez [Wed, 13 Jan 2010 03:18:30 +0000 (03:18 +0000)]
Revert 93270 pending investigation of how stray non-constant values end up in ValueEnumerator's ValueList during WriteConstants()
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93289
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 13 Jan 2010 01:51:43 +0000 (01:51 +0000)]
Disable opt-ext pass to unbreak the build for now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93286
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 13 Jan 2010 01:43:20 +0000 (01:43 +0000)]
Remove debug option I accidentally left in.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93285
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Wed, 13 Jan 2010 01:39:38 +0000 (01:39 +0000)]
Fix a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93284
91177308-0d34-0410-b5e6-
96231b3b80d8
Ted Kremenek [Wed, 13 Jan 2010 01:02:47 +0000 (01:02 +0000)]
Update CMake file.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93283
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 13 Jan 2010 00:43:06 +0000 (00:43 +0000)]
Remove the JustSP single-register regclass.
It was only being used by instructions with the t_addrmode_sp addressing mode,
and that is pattern matched in a way that guarantees SP is used. There is
never any register allocation done from this class.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93280
91177308-0d34-0410-b5e6-
96231b3b80d8
Jeffrey Yasskin [Wed, 13 Jan 2010 00:31:43 +0000 (00:31 +0000)]
Try to fix the ARM and PPC buildbots. The -mattr=vector-unaligned-mem
flag doesn't exist there, and this is an x86 test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93279
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 13 Jan 2010 00:30:23 +0000 (00:30 +0000)]
Add a quick pass to optimize sign / zero extension instructions. For targets where the pre-extension values are available in the subreg of the result of the extension, replace the uses of the pre-extension value with the result + extract_subreg.
For now, this pass is fairly conservative. It only perform the replacement when both the pre- and post- extension values are used in the block. It will miss cases where the post-extension values are live, but not used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93278
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Wed, 13 Jan 2010 00:00:24 +0000 (00:00 +0000)]
Further progration of metadata operands. The
dumper doesn't really do what I want yet, but
at least it doesn't crash now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93272
91177308-0d34-0410-b5e6-
96231b3b80d8
Victor Hernandez [Tue, 12 Jan 2010 23:37:59 +0000 (23:37 +0000)]
Make WriteConstants() more robust against stray values in ValueEnumerator's ValueList
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93270
91177308-0d34-0410-b5e6-
96231b3b80d8
Bob Wilson [Tue, 12 Jan 2010 22:18:56 +0000 (22:18 +0000)]
Fix a comment typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93261
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 12 Jan 2010 22:00:04 +0000 (22:00 +0000)]
use consistent tag kinds for ilist_traits
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93259
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 12 Jan 2010 20:41:47 +0000 (20:41 +0000)]
1) Use the new SimplifyInstructionsInBlock routine instead of the copy
in JT.
2) When cloning blocks for PHI or xor conditions, use
instsimplify to simplify the code as we go. This allows us to
squish common cases early in JT which opens up opportunities for
subsequent iterations, and allows it to completely simplify the
testcase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93253
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 12 Jan 2010 19:40:54 +0000 (19:40 +0000)]
add a helper function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93251
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Tue, 12 Jan 2010 18:57:56 +0000 (18:57 +0000)]
Use Twine, instead of StringRef, for consistency.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93249
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Tue, 12 Jan 2010 18:34:06 +0000 (18:34 +0000)]
Use ilist_tratis to autoinsert and remove NamedMDNode from MDSymbolTable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93247
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Tue, 12 Jan 2010 18:31:19 +0000 (18:31 +0000)]
Eliminate or_not_add and just use AddedComplexity so isel tries or_is_add patterns first.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93245
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Tue, 12 Jan 2010 18:29:23 +0000 (18:29 +0000)]
Add nounwind.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93244
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Tue, 12 Jan 2010 17:46:16 +0000 (17:46 +0000)]
Revert commit 93204, since it causes the assembler to barf
on x86-64 linux with messages like this:
Error: Incorrect register `%r14' used with `l' suffix
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93242
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Tue, 12 Jan 2010 08:30:46 +0000 (08:30 +0000)]
Fix typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93235
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Tue, 12 Jan 2010 08:21:07 +0000 (08:21 +0000)]
Tweak commit 91745, which changed target data for both Mingw and Cygwin,
to not touch Cygwin: the change caused llvm-gcc build failures due to
long double getting the wrong size. Patch by Aaron Gray.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93234
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 12 Jan 2010 04:52:47 +0000 (04:52 +0000)]
Make several tests less fragile.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93230
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 12 Jan 2010 04:42:54 +0000 (04:42 +0000)]
Reapply the MOV64r0 patch, with a fix: MOV64r0 clobbers EFLAGS.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93229
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 12 Jan 2010 04:32:35 +0000 (04:32 +0000)]
Update a partially obsolete comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93228
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 12 Jan 2010 04:30:26 +0000 (04:30 +0000)]
Fix a typo in a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93227
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 12 Jan 2010 02:07:50 +0000 (02:07 +0000)]
tidy up
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93222
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 12 Jan 2010 02:07:17 +0000 (02:07 +0000)]
Teach jump threading to duplicate small blocks when the branch
condition is a xor with a phi node. This eliminates nonsense
like this from 176.gcc in several places:
LBB166_84:
testl %eax, %eax
- setne %al
- xorb %cl, %al
- notb %al
- testb $1, %al
- je LBB166_85
+ je LBB166_69
+ jmp LBB166_85
This is rdar://
7391699
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93221
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Tue, 12 Jan 2010 02:01:53 +0000 (02:01 +0000)]
Add MO_Metadata as an operand kind. Not used yet.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93220
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Tue, 12 Jan 2010 00:09:37 +0000 (00:09 +0000)]
Add TargetInstrInfo::isCoalescableInstr. It returns true if the specified
instruction is copy like where the source and destination registers can
overlap. This is to be used by the coalescable to coalesce the source and
destination registers of instructions like X86::MOVSX64rr32. Apparently
some crazy people believe the coalescer is too simple.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93210
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 11 Jan 2010 23:41:09 +0000 (23:41 +0000)]
some cleanup, and make it obvious that ProcessJumpOnPHI only works
on branches by renaming it and checking for a branch at the call site.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93208
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 11 Jan 2010 23:18:33 +0000 (23:18 +0000)]
disable this testcase, PR5997
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93206
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 11 Jan 2010 22:59:27 +0000 (22:59 +0000)]
Add manual ISD::OR fastisel selection routines. TableGen is no longer autogen them after 93152 and 93191.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93204
91177308-0d34-0410-b5e6-
96231b3b80d8
Victor Hernandez [Mon, 11 Jan 2010 22:53:48 +0000 (22:53 +0000)]
Add documentation for llvm.dbg.value intrinsic
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93203
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 11 Jan 2010 22:49:40 +0000 (22:49 +0000)]
reenable the piece that turns trunc(zext(x)) -> x even if zext has multiple uses,
codegen has no apparent problem with the trunc version of this, because it turns
into a simple subreg idiom
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93202
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 11 Jan 2010 22:45:25 +0000 (22:45 +0000)]
Disable folding sext(trunc(x)) -> x (and other similar cast/cast cases) when the
trunc has multiple uses. Codegen is not able to coalesce the subreg case
correctly and so this leads to higher register pressure and spilling (see PR5997).
This speeds up 256.bzip2 from 8.60 -> 8.04s on my machine, ~7%.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93200
91177308-0d34-0410-b5e6-
96231b3b80d8
Victor Hernandez [Mon, 11 Jan 2010 22:31:58 +0000 (22:31 +0000)]
Consolidate ConvertGlobalValIDToValue, ConvertGlobalOrMetadataValIDToValue, and ConvertValIDToValue into a more powerful ConvertValIDToValue() that does all three's work
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93197
91177308-0d34-0410-b5e6-
96231b3b80d8
Tobias Grosser [Mon, 11 Jan 2010 22:22:46 +0000 (22:22 +0000)]
Remove trailing white spaces in post dominators header file
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93195
91177308-0d34-0410-b5e6-
96231b3b80d8
Tobias Grosser [Mon, 11 Jan 2010 22:22:32 +0000 (22:22 +0000)]
Add getNode() to post dominators.
Implement the same interface as already available for dominators.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93194
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 11 Jan 2010 22:03:29 +0000 (22:03 +0000)]
Extend r93152 to work on OR r, r. If the source set bits are known not to overlap, then select as an ADD instead.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93191
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 11 Jan 2010 21:58:19 +0000 (21:58 +0000)]
reduce this to a sensible testcase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93189
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Mon, 11 Jan 2010 21:50:35 +0000 (21:50 +0000)]
Shorten up this testcase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93187
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 11 Jan 2010 21:13:41 +0000 (21:13 +0000)]
Revert 93158. It's breaking quite a few x86_64 tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93185
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 11 Jan 2010 21:02:33 +0000 (21:02 +0000)]
Avoid adding PHI arguments for a predecessor that has gone away when a BRCOND was constant folded.
This fixes PR5980.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93184
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 11 Jan 2010 20:33:12 +0000 (20:33 +0000)]
Remove unused string functions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93183
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 11 Jan 2010 20:18:04 +0000 (20:18 +0000)]
Do not turn 8-bit OR to ADD since ADD8ri is not 3-addressfiable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93182
91177308-0d34-0410-b5e6-
96231b3b80d8
Mon P Wang [Mon, 11 Jan 2010 20:12:49 +0000 (20:12 +0000)]
Disable transformation of select of two loads to a select of address and then a load if the
loads are not in the default address space because the transformation discards src value info.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93180
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 11 Jan 2010 19:45:18 +0000 (19:45 +0000)]
Add StrInStrNoCase, a StringRef version of CStrInCStrNoCase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93174
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Mon, 11 Jan 2010 19:35:55 +0000 (19:35 +0000)]
Add top level section for named metadata.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93172
91177308-0d34-0410-b5e6-
96231b3b80d8
Jeffrey Yasskin [Mon, 11 Jan 2010 19:19:26 +0000 (19:19 +0000)]
Improve unclear bits and inaccuracies in structure and insertvalue
documentation.
Patch by Dustin Laurence!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93170
91177308-0d34-0410-b5e6-
96231b3b80d8
Jeffrey Yasskin [Mon, 11 Jan 2010 18:53:47 +0000 (18:53 +0000)]
Update the -tailcallopt description to match djg's improvements to the calling
convention.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93167
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Mon, 11 Jan 2010 18:52:33 +0000 (18:52 +0000)]
s/NextValueNo/NextMDValueNo while processing metadata.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93165
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 11 Jan 2010 18:44:35 +0000 (18:44 +0000)]
Turns out llvm-gcc still uses SplitString with a vector. Add it back until I
have a fix.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93163
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 11 Jan 2010 18:03:24 +0000 (18:03 +0000)]
Reimplement getToken and SplitString as "StringRef helper functions"
- getToken is modeled after StringRef::split but it can split on multiple
separator chars and skips leading seperators.
- SplitString is a StringRef::split variant for more than 2 elements with the
same behaviour as getToken.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93161
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 11 Jan 2010 17:58:34 +0000 (17:58 +0000)]
Use a 32-bit and with implicit zero-extension instead of a 64-bit and if it
has an immediate with at least 32 bits of leading zeros, to avoid needing to
materialize that immediate in a register first.
FileCheckize, tidy, and extend a testcase to cover this case.
This fixes rdar://
7527390.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93160
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 11 Jan 2010 17:37:57 +0000 (17:37 +0000)]
Re-instate MOV64r0 and MOV16r0, with adjustments to work with the
new AsmPrinter. This is perhaps less elegant than describing them
in terms of MOV32r0 and subreg operations, but it allows the
current register to rematerialize them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93158
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 11 Jan 2010 17:24:27 +0000 (17:24 +0000)]
Generalize this check to avoid depending on a specific register assignment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93157
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 11 Jan 2010 17:23:56 +0000 (17:23 +0000)]
Make this test less trivial, to avoid spurious failures.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93156
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 11 Jan 2010 17:21:05 +0000 (17:21 +0000)]
Pattern top-level operators don't need to be restricted to a
single user. The _su forms are intended for non-top-level nodes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93155
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 11 Jan 2010 17:14:46 +0000 (17:14 +0000)]
Reword this comment to reference a more fundamental issue.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93154
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 11 Jan 2010 17:03:47 +0000 (17:03 +0000)]
Select an OR with immediate as an ADD if the input bits are known zero. This allow the instruction to be 3address-fied if needed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93152
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Mon, 11 Jan 2010 16:29:42 +0000 (16:29 +0000)]
Implement a feature (-vector-unaligned-mem) to allow targets to
ignore alignment requirements for SIMD memory operands. This
is useful on architectures like the AMD 10h that do not trap on
unaligned references if a status bit is twiddled at startup time.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93151
91177308-0d34-0410-b5e6-
96231b3b80d8
Victor Hernandez [Mon, 11 Jan 2010 07:45:19 +0000 (07:45 +0000)]
Respond to Chris' review:
Make InsertDbgValueIntrinsic() and get Offset take and recieve a uint64_t.
Get constness correct for getVariable() and getValue().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93149
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 11 Jan 2010 06:55:24 +0000 (06:55 +0000)]
add one more bitfield optimization, allowing clang to generate
good code on PR4216:
_test_bitfield: ## @test_bitfield
orl $32962, %edi
movl $
4294941946, %eax
andq %rdi, %rax
ret
instead of:
_test_bitfield:
movl $
4294941696, %ecx
movl %edi, %eax
orl $194, %edi
orl $32768, %eax
andq $250, %rdi
andq %rax, %rcx
movq %rdi, %rax
orq %rcx, %rax
ret
Evan is looking into the remaining andq+imm -> andl optimization.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93147
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 11 Jan 2010 04:05:13 +0000 (04:05 +0000)]
Extend CanEvaluateZExtd to handle and/or/xor more aggressively in the
BitsToClear case. This allows it to promote expressions which have an
and/or/xor after the lshr, promoting cases like test2 (from PR4216)
and test3 (random extample extracted from a spec benchmark).
clang now compiles the code in PR4216 into:
_test_bitfield: ## @test_bitfield
movl %edi, %eax
orl $194, %eax
movl $
4294902010, %ecx
andq %rax, %rcx
orl $32768, %edi
andq $39936, %rdi
movq %rdi, %rax
orq %rcx, %rax
ret
instead of:
_test_bitfield: ## @test_bitfield
movl %edi, %eax
orl $194, %eax
movl $
4294902010, %ecx
andq %rax, %rcx
shrl $8, %edi
orl $128, %edi
shlq $8, %rdi
andq $39936, %rdi
movq %rdi, %rax
orq %rcx, %rax
ret
which is still not great, but is progress.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93145
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 11 Jan 2010 03:32:00 +0000 (03:32 +0000)]
Remove the dead TD argument to CanEvaluateZExtd, and add a
new BitsToClear result which allows us to start promoting
expressions that end with a lshr-by-constant. This is
conservatively correct and better than what we had before
(see testcases) but still needs to be extended further.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93144
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 11 Jan 2010 02:43:35 +0000 (02:43 +0000)]
improve comments, remove dead TD argument to CanEvaluateSExtd.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93143
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 20:30:41 +0000 (20:30 +0000)]
teach sext optimization to handle truncs from types that are not
the dest of the sext.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93128
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 20:25:54 +0000 (20:25 +0000)]
teach zext optimization how to deal with truncs that don't come from
the zext dest type. This allows us to handle test52/53 in cast.ll,
and allows llvm-gcc to generate much better code for PR4216 in -m64
mode:
_test_bitfield: ## @test_bitfield
orl $32962, %edi
movl %edi, %eax
andl $-25350, %eax
ret
This also fixes a bug handling vector extends, ensuring that the
mask produced is a vector constant, not an integer constant.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93127
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 20:21:42 +0000 (20:21 +0000)]
fix a buggy assertion, CreateIntegerCast should allow
integer vectors as well as just integers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93126
91177308-0d34-0410-b5e6-
96231b3b80d8
Mikhail Glushenkov [Sun, 10 Jan 2010 18:48:49 +0000 (18:48 +0000)]
Suppress a warning on gcc 4.4.
warning: suggest parentheses around ‘&&’ within ‘||’.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93121
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 07:57:20 +0000 (07:57 +0000)]
simplify CanEvaluateSExtd to return a bool now that we have a
simpler profitability predicate.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93111
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 07:42:21 +0000 (07:42 +0000)]
the NumCastsRemoved argument to CanEvaluateSExtd is dead, remove it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93110
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 07:40:50 +0000 (07:40 +0000)]
now that the cost model has changed, we can always consider
elimination of a sign extend to be a win, which simplifies
the client of CanEvaluateSExtd, and allows us to eliminate
more casts (examples taken from real code).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93109
91177308-0d34-0410-b5e6-
96231b3b80d8
Victor Hernandez [Sun, 10 Jan 2010 07:14:18 +0000 (07:14 +0000)]
Compute isFunctionLocal in MDNode ctor or via argument in new function getWhenValsUnresolved().
Document PFS argument to ParseValID() and ConvertGlobalOrMetadataValIDToValue().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93108
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 07:08:30 +0000 (07:08 +0000)]
change the preferred canonical form for a sign extension to be
lshr+ashr instead of trunc+sext. We want to avoid type
conversions whenever possible, it is easier to codegen expressions
without truncates and extensions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93107
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 06:59:55 +0000 (06:59 +0000)]
fix indentation of switch statements, no functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93106
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 06:50:04 +0000 (06:50 +0000)]
fix pasto that broke bootstrap.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93105
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 02:50:04 +0000 (02:50 +0000)]
simplify CanEvaluateZExtd now that we don't care about the number of
bits known clear in the result and don't care about the # casts
eliminated. TD is also dead but keeping it for now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93098
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 10 Jan 2010 02:39:31 +0000 (02:39 +0000)]
two changes:
1) don't try to optimize a sext or zext that is only used by a trunc, let
the trunc get optimized first. This avoids some pointless effort in
some common cases since instcombine scans down a block in the first pass.
2) Change the cost model for zext elimination to consider an 'and' cheaper
than a zext. This allows us to do it more aggressively, and for the next
patch to simplify the code quite a bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93097
91177308-0d34-0410-b5e6-
96231b3b80d8