Commit Graph

60399 Commits

Author SHA1 Message Date
Evan Cheng
afff40a62d Teach PHI elimination to remove REG_SEQUENCE instructions and update references of the source operands with references of the destination with subreg indices. e.g.
%reg1029<def>, %reg1030<def> = VLD1q16 %reg1024<kill>, ...
%reg1031<def> = REG_SEQUENCE %reg1029<kill>, 5, %reg1030<kill>, 6
=>
%reg1031:5<def>, %reg1031:6<def> = VLD1q16 %reg1024<kill>, ...

PHI elimination now does more than phi elimination. It is really a de-SSA pass.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103039 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 20:26:52 +00:00
Duncan Sands
55cd6cb141 Use llvm.foo as the intrinsic, rather than llvm.dbg.value. Since the
values passed to llvm.dbg.value were not valid for the intrinsic, it
might have caused trouble one day if the verifier ever started checking
for valid debug info.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103038 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 20:09:25 +00:00
Bob Wilson
34414a634e Defer adding critical edges to the "toSplit" list until after checking for
indirect branches in all the predecessors.  This avoids unnecessarily
splitting edges in cases where load PRE is not possible anyway.
Thanks to Jakub Staszak for pointing this out.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103034 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 20:03:21 +00:00
Chris Lattner
cd465012b3 one more thing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103024 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 18:16:00 +00:00
Chris Lattner
860e0b4c95 update instructions for llvm-gcc4, the brave new world! PR7037
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103023 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 18:15:33 +00:00
Chris Lattner
d4ac35b350 "on the rare occasion the SPU BE produces illegal assembly - it tries to emit an add instruction of the form 'a reg, reg, imm'."
Patch by Kalle Raiskila!



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103021 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 17:58:46 +00:00
Daniel Dunbar
e9f0fb4179 MC/X86: Chris pointed that 'as' isn't consistent in accepting the long form of
instructions which have no direct register usage.

Darwin 'as' accepts:
  add $0, (%rax)
but rejects
  mov $0, (%rax)
for example.

Given that, only accept suffix matches which match exactly one form. We still
need to emit nice diagnostics for failures...

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103015 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 17:31:02 +00:00
Evan Cheng
28428cd6f3 Rename variables for consistency.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103013 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 17:12:26 +00:00
Daniel Dunbar
c918d6043b MC/X86: Add "support" for matching ATT style mnemonic prefixes.
- The idea is that when a match fails, we just try to match each of +'b', +'w',
   +'l'. If exactly one matches, we assume this is a mnemonic prefix and accept
   it. If all match, we assume it is width generic, and take the 'l' form.

 - This would be a horrible hack, if it weren't so simple. Therefore it is an
   elegant solution! Chris gets the credit for this particular elegant
   solution. :)

 - Next step to making this more robust is to have the X86 matcher generate the
   mnemonic prefix information. Ideally we would also compute up-front exactly
   which mnemonic to attempt to match, but this may require more custom code in
   the matcher than is really worth it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103012 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 16:12:42 +00:00
Duncan Sands
d29f5282e0 Fix a problem exposed by my previous commit and noticed by a release-asserts
buildbot: the debugging and non-debugging versions of getFunction were not
functionally equivalent: the non-debugging version wrongly assumed that if a
metadata operand was not metadata, then it had a non-null containing function.
This is not true, since the operand might be a global value, constant etc.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103008 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 14:25:42 +00:00
Duncan Sands
203f7cb68b Fix a variant of PR6112 found by thinking about it: when doing
RAUW of a global variable with a local variable in function F,
if function local metadata M in function G was using the global
then M would become function-local to both F and G, which is not
allowed.  See the testcase for an example.  Fixed by detecting
this situation and zapping the metadata operand when it occurs.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103007 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 12:43:36 +00:00
Gabor Greif
2f256f4561 fix operand indexes when outputting InvokeInsts
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103003 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 09:23:54 +00:00
Devang Patel
9c488376ef Set DW_AT_APPLE_omit_frame_ptr in endFunction() where MachineFunction is available all the time.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103001 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 06:15:30 +00:00
Dan Gohman
acd8cab843 Use the SCEVAddRecExpr::getPostIncExpr utility function instead
of doing the same thing manually.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102997 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 01:12:27 +00:00
Dan Gohman
c0ed0091dc Fix a copy+pasto.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102996 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 01:11:15 +00:00
Devang Patel
01c5ff6437 Do not ignore debug loc attached with llvm.dbg.declare while collecting debug info used by a module.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102995 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 01:05:02 +00:00
Evan Cheng
f1ced259b5 Instruction selection optimizations may have moved the def of a function argument out of the entry block. rdar://7937489
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102993 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 00:58:39 +00:00
Kevin Enderby
a0161cd6f8 Fix to r102952. The MOV64toSDrm record in X86Instr64bit.td needed the opcode
changed to 0x7E from 0x6E as well as the previous change of RPDI to S3SI.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102991 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 00:42:46 +00:00
Daniel Dunbar
4f83e73a6d MC/Matcher: Add support for over-riding the default MatchInstruction function
name (for example, to allow targets to interpose the actual MatchInstruction
function).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102987 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 00:33:13 +00:00
Daniel Dunbar
31e8e1d085 llvm-mc: Fix case were we would skip a line in the .s file after an instruction
match failure.

Also, fixes a few memory leak FIXMEs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102986 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 00:33:07 +00:00
Evan Cheng
ba609c88a5 Teach scheduler about REG_SEQUENCE.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102984 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 00:22:40 +00:00
Dan Gohman
d622b0b208 Add a polygen rule that reflects the fact that nsw and nuw can be
used together in either order.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102983 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 00:13:24 +00:00
Dan Gohman
2131e2a75f Re-enable isel kill flags, now that the local allocator is ignoring them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102981 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 00:12:15 +00:00
Jim Grosbach
6e62b4ef14 rdar://7937137 - dbg values not being handled in thumb1 version of
eliminateFrameIndex(), leading to llvm_unreachable() assertion failure.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102980 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-04 00:11:37 +00:00
Dan Gohman
7ae3ac831e Update one more 2.7 to 2.8.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102978 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 23:52:21 +00:00
Dan Gohman
b44f6c6b6a Minimally update ReleaseNotes.html for 2.8 development; uncomment and
update the big red warning at the top. Most of the old content remains
and awaits revision.

Clear out the API changes section, and start it up again with a
mention of the add->fadd transition.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102977 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 23:51:05 +00:00
Jakob Stoklund Olesen
a50fba933c Remove preexisting kill flags in RegAllocLocal, just like LiveVariables does.
This should make it possible to start producing kill flags in isel without
breaking stuff.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102976 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 23:49:20 +00:00
Dan Gohman
1fdc614bee Factor out FastISel's code for materializing constants and other values
in registers into a separate function to de-couple it from the
top-down-specific logic in getRegForValue.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102975 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 23:36:34 +00:00
Dale Johannesen
08673d2950 Implement builtin_return_address(x) and builtin_frame_address(x)
on PPC for x!=0.  7624113.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102972 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 22:59:34 +00:00
Dan Gohman
1eaac53261 Remove the API compatibility layer which converted add, sub, and mul
to fadd, fsub, and fmul, when used with a floating-point type. LLVM
has supported the new instructions since 2.6, so it's time to get
on board.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102971 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 22:44:19 +00:00
Jakob Stoklund Olesen
b98c549d13 Check that subregisters don't have independent values in RemoveCopyByCommutingDef().
This fixes PR6941.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102970 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 22:40:32 +00:00
Dan Gohman
d4d01154ef Fix tests to use fadd, fsub, and fmul, instead of add, sub, and mul,
when the type is floating-point.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102969 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 22:36:46 +00:00
Eric Christopher
b4c14aaa50 Reword a comment slightly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102966 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 22:18:49 +00:00
Dan Gohman
deff621abd Use getConstant instead of getIntegerSCEV. The two are basically the
same, now that getConstant has overloads consistent with ConstantInt::get.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102965 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 22:09:21 +00:00
Bill Wendling
cec9c50924 Revert r102948.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102964 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 21:51:21 +00:00
Bob Wilson
857a894781 Remove special case for llvmCore_Embedded project. This is no longer needed
since llvmCore is built as part of llvmgcc42 now.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102962 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 21:41:24 +00:00
Bob Wilson
ad98f795d0 Print basic block numbers in live interval debug output. Since the rest of the
debug output is showing machine instructions, the IR-level basic block names
aren't very meaningful, and because multiple machine basic blocks may be
derived from one IR-level BB, they're also not unique.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102960 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 21:38:11 +00:00
Nick Lewycky
a75586f37e Fix function prototype mismatch in LLVMUnionType(). Fixes PR7019.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102959 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 21:36:51 +00:00
Kevin Enderby
9d0838fba8 Changed llvm-mc to use the same suffixes with floating point compare
instructions as the Mac OS X darwin assembler.  Some of which like 'fcoml'
assembled to different opcodes.  While some of the suffixes were just different.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102958 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 21:31:40 +00:00
Kevin Enderby
eb612347f4 Fixed the encoding of two of the X86 movq instuctions. The Move quadword from
mm to mm/m64 and the Move quadword from xmm2/mem64 to xmm1 had the incorrect
encodings.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102952 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 21:03:31 +00:00
Kevin Enderby
3c979b06c0 Fixed the encoding of the x86 push instructions. Using a 32-bit immediate value
caused the a pushl instruction to be incorrectly encoding using only two bytes
of immediate, causing the following 2 instruction bytes to be part of the 32-bit
immediate value.  Also fixed the one byte form of push to be used when the
immediate would fit in a signed extended byte.  Lastly changed the names to not
include the 32 of PUSH32 since they actually push the size of the stack pointer.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102951 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 20:45:05 +00:00
Bill Wendling
5f84c0293f Testcase for r102947.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102948 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 20:39:35 +00:00
Dan Gohman
f16c6803d7 Silence warnings about -1 being converted to an unsigned value.
Also, pass true for isSigned even when creating constants for unsigned
comparisons, because the point is to create an all-ones constant,
rather than UINT64_MAX, even for integers wider than 64 bits.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102946 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 20:23:47 +00:00
Eric Christopher
0b12348ddf Add an option, defaulting to off, to disable the sse domain crossing opts.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102941 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 19:54:02 +00:00
Devang Patel
86b6f80a4c Check for side effects before splitting loop.
Patch by Jakub Staszak!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102928 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 18:06:58 +00:00
Dan Gohman
34c3e36e63 Use isTrueWhenEqual and isFalseWhenEqual instead of assuming that
SimplifyICmpOperands will simplify such cases to EQ or NE. This makes
the correcntess of the code independent on SimplifyICmpOperands doing
certain simplifications.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102927 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 18:00:24 +00:00
Dan Gohman
a189bae771 In ScalarEvolution::print, don't bother printing out the SCEVs for
comparison instructions, since they aren't interesting, despite having
integer result types.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102925 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 17:03:23 +00:00
Dan Gohman
3abb69c07f In SimplifyICmpOperands, avoid needlessly swapping the operands in the
case where both are addrecs in unrelated loops.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102924 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 17:00:11 +00:00
Dan Gohman
03557dc0ad Factor out the new <= and >= analysis code into SimplifyICmpOperands.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102922 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 16:35:17 +00:00
Dan Gohman
7c24ff12c4 Update this comment to note its relationship with the "happens-before"
concept in the proposed memory model changes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102911 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-03 14:59:34 +00:00