Commit Graph

8542 Commits

Author SHA1 Message Date
David Greene
f5aefbdb79 Change IndexedModeAction representation.
This changes the IndexedModeAction representation to remove the
limitation on the number of value types in MVT.  This limitation
prevents us from specifying AVX types.

Prior to this change IndexedModActions was represented as follows...

uint64_t IndexedModeActions[2][ISD::LAST_INDEXED_MODE];

the first dimension was used to represent loads, then stores.  This
imposed a limitation of 32 on the number of value types that could be
handled with this method.  The value type was used to shift the two bits
into and out of the approprate bits in the uint64_t.

With this change the array is now represented as ...

uint8_t IndexedModeActions[MVT::LAST_VALUETYPE][2][ISD::LAST_INDEXED_MODE];

Takes more space but removes the limitation on MVT::LAST_VALUETYPE.  The
first dimension is now the value_type for the reference.  The second
dimension is the load [0] vs. store[1].  The third dimension represents
the various modes for load store.  Accesses are now direct, no shifting
or masking.

There are other limitations that need to be removed, so that
MVT::LAST_VALUETYPE can be greater than 32.  This is merely the first
step towards that goal.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73104 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-09 00:19:10 +00:00
David Greene
63ec51d86b Oops, didn't mean to commit 73102 yet. Revert it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73103 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-08 23:54:42 +00:00
David Greene
9df4eb8cfe Change IndexedModeAction representation.
This changes the IndexedModeAction representation to remove the
limitation on the number of value types in MVT.  This limitation
prevents us from specifying AVX types.

Prior to this change IndexedModActions was represented as follows...

uint64_t IndexedModeActions[2][ISD::LAST_INDEXED_MODE];

the first dimension was used to represent loads, then stores.  This
imposed a limitation of 32 on the number of value types that could be
handled with this method.  The value type was used to shift the two bits
into and out of the approprate bits in the uint64_t.

With this change the array is now represented as ...

uint8_t IndexedModeActions[MVT::LAST_VALUETYPE][2][ISD::LAST_INDEXED_MODE];

Takes more space but removes the limitation on MVT::LAST_VALUETYPE.  The
first dimension is now the value_type for the reference.  The second
dimension is the load [0] vs. store[1].  The third dimension represents
the various modes for load store.  Accesses are now direct, no shifting
or masking.

There are other limitations that need to be removed, so that
MVT::LAST_VALUETYPE can be greater than 32.  This is merely the first
step towards that goal.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73102 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-08 23:53:19 +00:00
Anton Korobeynikov
0eebf653a7 The attached patches implement most of the ARM AAPCS-VFP hard float
ABI. The missing piece is support for putting "homogeneous aggregates"
into registers.

Patch by Sandeep Patel!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73095 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-08 22:53:56 +00:00
David Greene
e1b469170b Make IntInits and ListInits typed. This helps deduce types of !if and
other operators.  For the rare cases where a list type cannot be
deduced, provide a []<type> syntax, where <type> is the list element
type.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73078 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-08 20:23:18 +00:00
Bruno Cardoso Lopes
a029a27fae Simple ELF32/64 binary files can now be emitted for x86 and x86_64 without
relocation sections.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73038 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-07 21:22:38 +00:00
Nick Lewycky
a5f54a06b0 Create FunctionType::isValidArgumentType to go along with isValidReturnType.
Also create isValidElementType for ArrayType, PointerType, StructType and
VectorType.

Make LLParser use them. This closes up some holes like an assertion failure on:

  %x = type {label}

but largely doesn't change any semantics. The only thing we accept now which
we didn't before is vectors of opaque type such as "<4 x opaque>". The opaque
can be resolved to an int or float when linking.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73016 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-07 07:26:46 +00:00
Dan Gohman
86fbf2fe4c Rename UnknownValue to CouldNotCompute, since it holds an instance of
SCEVCouldNotCompute, and not SCEVUnknown.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72999 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-06 14:37:11 +00:00
Bruno Cardoso Lopes
f5b0c5a1c7 Remove elf specific info from ELFWriter.h to Elf.h. Code cleanup and more comments added
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72982 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-06 03:56:29 +00:00
Douglas Gregor
318de60701 More portability checks for CMake's config.h.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72975 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-05 23:46:34 +00:00
Devang Patel
578efa920a Add new function attribute - noimplicitfloat
Update code generator to use this attribute and remove NoImplicitFloat target option.
Update llc to set this attribute when -no-implicit-float command line option is used.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72959 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-05 21:57:13 +00:00
Nate Begeman
abc0199680 Adapt the x86 build_vector dagcombine to the current state of the legalizer.
build vectors with i64 elements will only appear on 32b x86 before legalize.
Since vector widening occurs during legalize, and produces i64 build_vector 
elements, the dag combiner is never run on these before legalize splits them
into 32b elements.

Teach the build_vector dag combine in x86 back end to recognize consecutive 
loads producing the low part of the vector.

Convert the two uses of TLI's consecutive load recognizer to pass LoadSDNodes
since that was required implicitly.

Add a testcase for the transform.

Old:
	subl	$28, %esp
	movl	32(%esp), %eax
	movl	4(%eax), %ecx
	movl	%ecx, 4(%esp)
	movl	(%eax), %eax
	movl	%eax, (%esp)
	movaps	(%esp), %xmm0
	pmovzxwd	%xmm0, %xmm0
	movl	36(%esp), %eax
	movaps	%xmm0, (%eax)
	addl	$28, %esp
	ret

New:
	movl	4(%esp), %eax
	pmovzxwd	(%eax), %xmm0
	movl	8(%esp), %eax
	movaps	%xmm0, (%eax)
	ret




git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72957 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-05 21:37:30 +00:00
Dan Gohman
1d09de3eca Move SCEVExpander::getOrInsertCanonicalInductionVariable out of line.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72949 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-05 16:35:53 +00:00
Dan Gohman
ae3a0be92e Split the Add, Sub, and Mul instruction opcodes into separate
integer and floating-point opcodes, introducing
FAdd, FSub, and FMul.

For now, the AsmParser, BitcodeReader, and IRBuilder all preserve
backwards compatability, and the Core LLVM APIs preserve backwards
compatibility for IR producers. Most front-ends won't need to change
immediately.

This implements the first step of the plan outlined here:
http://nondot.org/sabre/LLVMNotes/IntegerOverflow.txt


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72897 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-04 22:49:04 +00:00
Devang Patel
d18e31ae17 Add new function attribute - noredzone.
Update code generator to use this attribute and remove DisableRedZone target option.
Update llc to set this attribute when -disable-red-zone command line option is used.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72894 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-04 22:05:33 +00:00
Douglas Gregor
ef23d711bc Properly detect malloc_zone_statistics in CMake build system
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72873 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-04 17:22:52 +00:00
Torok Edwin
6c8db34165 Fix comments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72858 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-04 08:18:25 +00:00
Torok Edwin
e8ebb0fe1b Add support for outputting ANSI colors to raw_fd_ostream.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72854 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-04 07:09:50 +00:00
Bruno Cardoso Lopes
186c670e15 Use uint8_t and int32_t in {JIT,Machine}CodeEmiters
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72821 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-04 00:15:51 +00:00
Daniel Dunbar
e0f0e0be1c Change LTO to run the global opt pass twice.
- This matches llvm-ld.

It took a bit of archeology to figure out what the right thing to do was
(whether this was intentionally added or intentionally removed). My final
conclusion is that Chris added this intentionally here:
  http://llvm.org/viewvc/llvm-project?view=rev&revision=16913
but the changes weren't propogated to llvm-ld until here:
  http://llvm.org/viewvc/llvm-project?view=rev&revision=34058
which was after lto.cpp had been cloned off (of llvm-ld), here:
  http://llvm.org/viewvc/llvm-project?view=rev&revision=29494

From the commit message, it looks like the motivation for running global opt
again is because we ran it prior to inlining. Based on that I updated the
comment and also only run the pass if we actually ran the inliner.

Chris, please review.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72811 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-03 21:51:32 +00:00
Eli Friedman
a8ba2c25e9 PR4317: Handle splits where the new block is unreachable correctly in
DominatorTreeBase::Split.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72810 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-03 21:42:06 +00:00
Daniel Dunbar
006a034828 Add createStandardLTOPasses to StandardPasses.h, and move lto and llvm-ld over.
- I know it sounds crazy, but I think all the pass lists are now coalesced into
   StandardPasses.h.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72805 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-03 21:06:14 +00:00
Daniel Dunbar
e98c4d0b10 Remove some silly code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72788 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-03 18:13:05 +00:00
Daniel Dunbar
04d5f44882 Add StandardPasses.h which embeds the standard compilation passes shared by
clang/llvm-gcc.
 - Implemented as inline functions for linking simplicity.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72786 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-03 17:52:39 +00:00
Bruno Cardoso Lopes
bae049cd9e Revert 72650
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72783 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-03 16:55:02 +00:00
Dan Gohman
b3ae97cffe Change TargetFolder's TD member from a reference to a
pointer, now that ConstantFoldConstantExpression can
accept a null TargetData pointer.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72744 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-03 00:13:48 +00:00
Dan Gohman
7714285efd Change ConstantFoldConstantExpression to accept a null
TargetData pointer. The only thing it's used for are
calls to ConstantFoldCompareInstOperands and
ConstantFoldInstOperands, which both already accept a
null TargetData pointer. This makes
ConstantFoldConstantExpression easier to use in clients
where TargetData is optional.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72741 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-02 21:48:15 +00:00
Dan Gohman
014d53595c Delete an unnecessary forward declaration.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72740 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-02 21:37:43 +00:00
Torok Edwin
23e1e727bd fix build on VC++2k8.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72736 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-02 20:32:59 +00:00
Devang Patel
8900bcc1ff Simplify assertion message to avoid confusion.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72731 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-02 17:26:29 +00:00
Lang Hames
f41538d1b5 Update to in-place spilling framework. Includes live interval scaling and trivial rewriter.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72729 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-02 16:53:25 +00:00
Dale Johannesen
874ae251c3 Revert 72707 and 72709, for the moment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72712 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-02 03:12:52 +00:00
Dale Johannesen
4150d83abe Make the implicit inputs and outputs of target-independent
ADDC/ADDE use MVT::i1 (later, whatever it gets legalized to)
instead of MVT::Flag.  Remove CARRY_FALSE in favor of 0; adjust
all target-independent code to use this format.

Most targets will still produce a Flag-setting target-dependent
version when selection is done.  X86 is converted to use i32
instead, which means TableGen needs to produce different code
in xxxGenDAGISel.inc.  This keys off the new supportsHasI1 bit
in xxxInstrInfo, currently set only for X86; in principle this
is temporary and should go away when all other targets have
been converted.  All relevant X86 instruction patterns are
modified to represent setting and using EFLAGS explicitly.  The
same can be done on other targets.

The immediate behavior change is that an ADC/ADD pair are no
longer tightly coupled in the X86 scheduler; they can be
separated by instructions that don't clobber the flags (MOV).
I will soon add some peephole optimizations based on using
other instructions that set the flags to feed into ADC.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72707 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-01 23:27:20 +00:00
Mike Stump
6330084a66 Dcoument that the opauque value used to construct QNaNs is truncated
as necessary.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72702 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-01 21:02:35 +00:00
Dan Gohman
55b42513e1 Fix the name of the function in this comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72666 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-31 16:18:03 +00:00
Bruno Cardoso Lopes
af90a1cd26 Use uint8_t and int32_t in {JIT,Machine}CodeEmiters
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72650 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-30 23:50:33 +00:00
Bruno Cardoso Lopes
cc5fc60468 Add AMD64 elf machine type
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72635 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-30 21:05:47 +00:00
Bruno Cardoso Lopes
a3f99f9033 First patch in the direction of splitting MachineCodeEmitter in two subclasses:
JITCodeEmitter and ObjectCodeEmitter. No functional changes yet. Patch by Aaron Gray



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72631 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-30 20:51:52 +00:00
Mike Stump
7d36d39e63 Add some documentation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72622 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-30 15:19:29 +00:00
Nick Lewycky
7a0370f66a Give embedded metadata its own type instead of relying on EmptyStructTy.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72610 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-30 05:06:04 +00:00
Mike Stump
c5ca713b80 Add support for letting the client choose different flavors of NaNs. Testcase to be
added in clang.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72606 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-30 03:49:43 +00:00
Bill Wendling
4d4eab219a Untabify.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72603 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-30 00:58:37 +00:00
Evan Cheng
abdbc57abb Default isNarrowingProfitable to false.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72561 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-29 05:31:40 +00:00
Evan Cheng
8b944d39b3 Added optimization that narrow load / op / store and the 'op' is a bit twiddling instruction and its second operand is an immediate. If bits that are touched by 'op' can be done with a narrower instruction, reduce the width of the load and store as well. This happens a lot with bitfield manipulation code.
e.g.
orl     $65536, 8(%rax)
=>
orb     $1, 10(%rax)

Since narrowing is not always a win, e.g. i32 -> i16 is a loss on x86, dag combiner consults with the target before performing the optimization.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72507 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-28 00:35:15 +00:00
Jeffrey Yasskin
493a3d015c LiveVariables::VarInfo contains an AliveBlocks BitVector, which has as many
entries as there are basic blocks in the function.  LiveVariables::getVarInfo
creates a VarInfo struct for every register in the function, leading to
quadratic space use.  This patch changes the BitVector to a SparseBitVector,
which doesn't help the worst-case memory use but does reduce the actual use in
very long functions with short-lived variables.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72426 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-26 18:27:15 +00:00
Dan Gohman
d9480d0f68 Give SCEVNaryExpr a doxygen comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72423 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-26 17:42:32 +00:00
Evan Cheng
aed4a430f4 Eliminate VarInfo::UsedBlocks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72411 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-26 06:25:46 +00:00
Dan Gohman
6c0866ca83 Various comment fixes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72376 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-24 23:45:28 +00:00
Dan Gohman
d594e6f034 Change ScalarEvolution::getSCEVAtScope to always return the original value
in the case where a loop exit value cannot be computed, instead of only in
some cases while using SCEVCouldNotCompute in others. This simplifies
getSCEVAtScope's callers.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72375 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-24 23:25:42 +00:00
Dan Gohman
453aa4fbf1 Generalize SCEVExpander::visitAddRecExpr's GEP persuit, and avoid
sending SCEVUnknowns to expandAddToGEP. This avoids the need for
expandAddToGEP to bend the rules and peek into SCEVUnknown
expressions.

Factor out the code for testing whether a SCEV can be factored by
a constant for use in a GEP index. This allows it to handle
SCEVAddRecExprs, by recursing.

As a result, SCEVExpander can now put more things in GEP indices,
so it emits fewer explicit mul instructions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72366 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-24 18:06:31 +00:00