Changes in directory llvm/lib/CodeGen:
IfConversion.cpp updated: 1.55 -> 1.56
---
Log message:
Properly remove duplicate instructions as result of diamond if-conversion.
Other bug fixes.
---
Diffs of the changes: (+176 -103)
IfConversion.cpp | 279 ++---
Changes in directory llvm/include/llvm/CodeGen:
MachineInstr.h updated: 1.222 -> 1.223
---
Log message:
Add keyword explicit.
---
Diffs of the changes: (+1 -1)
MachineInstr.h |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
Index: llvm/include/llvm/CodeGen/MachineInstr.h
diff -u
Changes in directory llvm/lib/Target/X86:
X86InstrX86-64.td updated: 1.16 -> 1.17
---
Log message:
Define the pushq instruction for x86-64.
---
Diffs of the changes: (+2 -0)
X86InstrX86-64.td |2 ++
1 files changed, 2 insertions(+)
Index: llvm/lib/Target/X86/X86InstrX86-64.td
diff -u
> Is this necessary? ARM is already doing this by marking trivially re-
> materializable instructions with let isReMaterializable = 1.
It's not necessary, but it seems cleaner :-).
In the ARM target, other than instructions that are always trivially
rematerialiable, there's only LDR, with a ReMat
> Nifty. Have you done any performance analysis of this? What cases
> is it beneficial for? If this helps LSR, for example, can you please
> add a testcase to verify this doesn't break in the future?
It doesn't actually help LSR much. At this point, a signextend isn't
much less opaque than an u
Changes in directory llvm/lib/CodeGen/SelectionDAG:
SelectionDAGPrinter.cpp updated: 1.46 -> 1.47
---
Log message:
Make chain dependencies blue, in addition to being dashed.
---
Diffs of the changes: (+1 -1)
SelectionDAGPrinter.cpp |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
On Jun 17, 2007, at 9:31 PM, Owen Anderson wrote:
> --- llvm/lib/Transforms/Scalar/GVNPRE.cpp:1.34Fri Jun 15 19:26:54
> 2007
> +++ llvm/lib/Transforms/Scalar/GVNPRE.cpp Sun Jun 17 23:30:44 2007
> @@ -329,7 +329,12 @@
> lhsValid = true;
> break;
>
On Jun 18, 2007, at 7:46 AM, Dan Gohman wrote:
>> Is this necessary? ARM is already doing this by marking trivially re-
>> materializable instructions with let isReMaterializable = 1.
>
> It's not necessary, but it seems cleaner :-).
>
> In the ARM target, other than instructions that are always
Changes in directory llvm/lib/Analysis:
BasicAliasAnalysis.cpp updated: 1.111 -> 1.112
---
Log message:
Make BasicAliasAnalysis correctly register itself. Patch by Devang
---
Diffs of the changes: (+2 -0)
BasicAliasAnalysis.cpp |2 ++
1 files changed, 2 insertions(+)
Index: llvm/lib/
We can eliminate LDRcp, that's fine.
However, I still don't like the separate hook and targetinstrinfo
bit. Dan, can you have just a single isTriviallyReMaterializable hook
that encompass all these?
Evan
On Jun 18, 2007, at 9:42 AM, Chris Lattner wrote:
>
> On Jun 18, 2007, at 7:46 AM, Dan
Changes in directory llvm/lib/Analysis:
ScalarEvolution.cpp updated: 1.119 -> 1.120
---
Log message:
In SCEVAddExpr::get, skip over any cast operands before looking for nested
add operands after constant operands. The recent change to recognize
sign-extend expressions caused this to be exposed
> In SCEVAddExpr::get, skip over any cast operands before looking for
> nested
> add operands after constant operands. The recent change to recognize
> sign-extend expressions caused this to be exposed more often.
Ok
> + // Now we know the first non-constant operand. Skip past any
> cast SC
> > + // Now we know the first non-constant operand. Skip past any
> > cast SCEVs.
> > + while (Idx < Ops.size() && Ops[Idx]->getSCEVType() < scAddExpr)
> > +++Idx;
>
> Can you please introduce a predicate for this? Comparing against
> scAddExpr is not obvious :)
I think if you read t
On Jun 18, 2007, at 1:35 PM, Dan Gohman wrote:
>>> + // Now we know the first non-constant operand. Skip past any
>>> cast SCEVs.
>>> + while (Idx < Ops.size() && Ops[Idx]->getSCEVType() < scAddExpr)
>>> +++Idx;
>>
>> Can you please introduce a predicate for this? Comparing against
>> scA
Changes in directory llvm/lib/CodeGen/SelectionDAG:
LegalizeDAG.cpp updated: 1.497 -> 1.498
---
Log message:
make ComputeTopDownOrdering significantly faster and use less stack space
by making it non-recursive
---
Diffs of the changes: (+37 -32)
LegalizeDAG.cpp | 69 +
Changes in directory llvm/lib/VMCore:
PassManager.cpp updated: 1.157 -> 1.158
---
Log message:
Fix quotes in debug messages.
---
Diffs of the changes: (+6 -6)
PassManager.cpp | 12 ++--
1 files changed, 6 insertions(+), 6 deletions(-)
Index: llvm/lib/VMCore/PassManager.cpp
diff
> We can eliminate LDRcp, that's fine.
>
> However, I still don't like the separate hook and targetinstrinfo
> bit. Dan, can you have just a single isTriviallyReMaterializable hook
> that encompass all these?
What do you think about the attached patch? It's not tested yet, but it
shows what I
Changes in directory llvm/include/llvm:
ParameterAttributes.h updated: 1.13 -> 1.14
---
Log message:
add helper
---
Diffs of the changes: (+7 -0)
ParameterAttributes.h |7 +++
1 files changed, 7 insertions(+)
Index: llvm/include/llvm/ParameterAttributes.h
diff -u llvm/include/llv
Revision: 128564
Author: clattner
Date: 2007-06-18 14:51:50 -0700 (Mon, 18 Jun 2007)
Log Message:
---
Use new attribute helper, no functionality change.
Modified Paths:
--
apple-local/branches/llvm/gcc/llvm-types.cpp
Modified: apple-local/branches/llvm/gcc/llvm-type
Yes, it looks good. I can deal with LDRcp and friends later.
Thanks.
Evan
On Jun 18, 2007, at 2:35 PM, Dan Gohman wrote:
>> We can eliminate LDRcp, that's fine.
>>
>> However, I still don't like the separate hook and targetinstrinfo
>> bit. Dan, can you have just a single isTriviallyReMateriali
Changes in directory llvm/test/CFrontend:
2006-09-25-DebugFilename.c updated: 1.3 -> 1.4
2007-06-15-AnnotateAttribute.c updated: 1.1 -> 1.2
---
Log message:
Modify annotate intrinsic to take 2 additional args: file and line number.
Fix up Annotate attribute test case.
Fix DebugFilename.c test c
Changes in directory llvm/include/llvm:
Intrinsics.td updated: 1.56 -> 1.57
---
Log message:
Modify annotate intrinsic to take 2 additional args: file and line number.
Fix up Annotate attribute test case.
Fix DebugFilename.c test case to look for right file name.
---
Diffs of the changes: (+
Revision: 128567
Author: lattner
Date: 2007-06-18 15:30:01 -0700 (Mon, 18 Jun 2007)
Log Message:
---
Add file and line number to annotate attribute.
Modified Paths:
--
apple-local/branches/llvm/gcc/llvm-backend.cpp
apple-local/branches/llvm/gcc/llvm-convert.cpp
Changes in directory llvm/lib/CodeGen:
BranchFolding.cpp updated: 1.67 -> 1.68
MachineBasicBlock.cpp updated: 1.46 -> 1.47
---
Log message:
Move CorrectExtraCFGEdges() from BranchFolding.cpp to a MachineBasicBlock
method.
---
Diffs of the changes: (+64 -75)
BranchFolding.cpp | 79 ++-
Changes in directory llvm/include/llvm/CodeGen:
MachineBasicBlock.h updated: 1.65 -> 1.66
---
Log message:
Move CorrectExtraCFGEdges() from BranchFolding.cpp to a MachineBasicBlock
method.
---
Diffs of the changes: (+10 -0)
MachineBasicBlock.h | 10 ++
1 files changed, 10 inserti
Changes in directory llvm/lib/CodeGen:
IfConversion.cpp updated: 1.56 -> 1.57
---
Log message:
Fix some fragile code wrt CFG edge updating.
---
Diffs of the changes: (+39 -74)
IfConversion.cpp | 113 ++-
1 files changed, 39 insertions(+),
Changes in directory llvm/docs:
LangRef.html updated: 1.255 -> 1.256
---
Log message:
Update docs for llvm.var.annotation intrinsic.
---
Diffs of the changes: (+5 -4)
LangRef.html |9 +
1 files changed, 5 insertions(+), 4 deletions(-)
Index: llvm/docs/LangRef.html
diff -u llv
Changes in directory llvm/lib/Target/X86:
X86ISelLowering.cpp updated: 1.411 -> 1.412
---
Log message:
Look for VECTOR_SHUFFLE that's identity operation on either LHS or RHS. This
can happen before DAGCombiner catches it.
---
Diffs of the changes: (+19 -2)
X86ISelLowering.cpp | 21 ++
Changes in directory llvm/test/CodeGen/X86:
vec_shuffle-11.ll added (r1.1)
---
Log message:
New test case: identity operation of RHS / LHS of a VECTOR_SHUFFLE.
---
Diffs of the changes: (+11 -0)
vec_shuffle-11.ll | 11 +++
1 files changed, 11 insertions(+)
Index: llvm/test/CodeG
Changes in directory llvm/test/CFrontend:
2007-06-18-SextAttrAggregate.c added (r1.1)
---
Log message:
testcase for PR1513: http://llvm.org/PR1513
---
Diffs of the changes: (+11 -0)
2007-06-18-SextAttrAggregate.c | 11 +++
1 files changed, 11 insertions(+)
Index: llvm/test/CFr
Changes in directory llvm/include/llvm/CodeGen:
CallingConvLower.h updated: 1.4 -> 1.5
---
Log message:
add isvararg tracking to CCState
---
Diffs of the changes: (+3 -1)
CallingConvLower.h |4 +++-
1 files changed, 3 insertions(+), 1 deletion(-)
Index: llvm/include/llvm/CodeGen/Cal
Changes in directory llvm/lib/CodeGen/SelectionDAG:
CallingConvLower.cpp updated: 1.4 -> 1.5
---
Log message:
add isVarArg to CCState
---
Diffs of the changes: (+3 -2)
CallingConvLower.cpp |5 +++--
1 files changed, 3 insertions(+), 2 deletions(-)
Index: llvm/lib/CodeGen/SelectionDAG
Changes in directory llvm/lib/Target:
TargetCallingConv.td updated: 1.5 -> 1.6
---
Log message:
If a function is vararg, never pass inreg arguments in registers. Thanks to
Anton for half of this patch.
---
Diffs of the changes: (+2 -0)
TargetCallingConv.td |2 ++
1 files changed, 2
Changes in directory llvm/lib/Target/PowerPC:
PPCISelLowering.cpp updated: 1.272 -> 1.273
---
Log message:
If a function is vararg, never pass inreg arguments in registers. Thanks to
Anton for half of this patch.
---
Diffs of the changes: (+2 -1)
PPCISelLowering.cpp |3 ++-
1 files
Changes in directory llvm/lib/Target/X86:
X86CallingConv.td updated: 1.13 -> 1.14
X86ISelLowering.cpp updated: 1.412 -> 1.413
---
Log message:
If a function is vararg, never pass inreg arguments in registers. Thanks to
Anton for half of this patch.
---
Diffs of the changes: (+18 -14)
X8
Changes in directory llvm/lib/Target/IA64:
README updated: 1.5 -> 1.6
---
Log message:
If a function is vararg, never pass inreg arguments in registers. Thanks to
Anton for half of this patch.
---
Diffs of the changes: (+2 -0)
README |2 ++
1 files changed, 2 insertions(+)
Index:
Revision: 128583
Author: clattner
Date: 2007-06-18 17:15:05 -0700 (Mon, 18 Jun 2007)
Log Message:
---
Merge parameter attribute handling into the same loop that
translates the types for the arguments. This fixes PR1513 and
CFrontend/2007-06-18-SextAttrAggregate.c
Modified Paths:
--
Changes in directory llvm/include/llvm/Target:
TargetInstrInfo.h updated: 1.130 -> 1.131
---
Log message:
Replace CanBeDuplicated() with a M_NOT_DUPLICABLE bit.
---
Diffs of the changes: (+20 -16)
TargetInstrInfo.h | 36
1 files changed, 20 insertions(
Changes in directory llvm/lib/Target/ARM:
ARMInstrInfo.cpp updated: 1.37 -> 1.38
ARMInstrInfo.h updated: 1.14 -> 1.15
ARMInstrInfo.td updated: 1.110 -> 1.111
ARMInstrThumb.td updated: 1.30 -> 1.31
---
Log message:
Replace TargetInstrInfo::CanBeDuplicated() with a M_NOT_DUPLICABLE bit.
---
Diff
Changes in directory llvm/lib/Target:
Target.td updated: 1.103 -> 1.104
---
Log message:
Replace TargetInstrInfo::CanBeDuplicated() with a M_NOT_DUPLICABLE bit.
---
Diffs of the changes: (+1 -0)
Target.td |1 +
1 files changed, 1 insertion(+)
Index: llvm/lib/Target/Target.td
diff -u l
Changes in directory llvm/lib/CodeGen:
IfConversion.cpp updated: 1.57 -> 1.58
---
Log message:
Replace TargetInstrInfo::CanBeDuplicated() with a M_NOT_DUPLICABLE bit.
---
Diffs of the changes: (+2 -2)
IfConversion.cpp |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
Index: ll
Changes in directory llvm/utils/TableGen:
CodeGenInstruction.h updated: 1.28 -> 1.29
CodeGenTarget.cpp updated: 1.92 -> 1.93
InstrInfoEmitter.cpp updated: 1.60 -> 1.61
---
Log message:
Replace TargetInstrInfo::CanBeDuplicated() with a M_NOT_DUPLICABLE bit.
---
Diffs of the changes: (+3 -0)
Changes in directory llvm/lib/CodeGen:
LiveIntervalAnalysis.cpp updated: 1.247 -> 1.248
VirtRegMap.cpp updated: 1.112 -> 1.113
---
Log message:
Replace M_REMATERIALIZIBLE and the newly-added isOtherReMaterializableLoad
with a general target hook to identify rematerializable instructions. Some
i
Changes in directory llvm/lib/Target/ARM:
ARMInstrInfo.cpp updated: 1.38 -> 1.39
ARMInstrInfo.h updated: 1.15 -> 1.16
ARMInstrInfo.td updated: 1.111 -> 1.112
ARMInstrThumb.td updated: 1.31 -> 1.32
---
Log message:
Replace M_REMATERIALIZIBLE and the newly-added isOtherReMaterializableLoad
with a
Changes in directory llvm/utils/TableGen:
CodeGenInstruction.h updated: 1.29 -> 1.30
CodeGenTarget.cpp updated: 1.93 -> 1.94
InstrInfoEmitter.cpp updated: 1.61 -> 1.62
---
Log message:
Replace M_REMATERIALIZIBLE and the newly-added isOtherReMaterializableLoad
with a general target hook to ident
Changes in directory llvm/lib/Target/X86:
X86InstrFPStack.td updated: 1.9 -> 1.10
X86InstrInfo.cpp updated: 1.91 -> 1.92
X86InstrInfo.h updated: 1.66 -> 1.67
X86InstrInfo.td updated: 1.307 -> 1.308
X86InstrMMX.td updated: 1.32 -> 1.33
X86InstrSSE.td updated: 1.183 -> 1.184
---
Log message:
Repl
Changes in directory llvm/lib/Target:
Target.td updated: 1.104 -> 1.105
---
Log message:
Replace M_REMATERIALIZIBLE and the newly-added isOtherReMaterializableLoad
with a general target hook to identify rematerializable instructions. Some
instructions are only rematerializable with specific ope
Changes in directory llvm/include/llvm/Target:
TargetInstrInfo.h updated: 1.131 -> 1.132
---
Log message:
Replace M_REMATERIALIZIBLE and the newly-added isOtherReMaterializableLoad
with a general target hook to identify rematerializable instructions. Some
instructions are only rematerializable
Changes in directory llvm/lib/Transforms/Scalar:
GVNPRE.cpp updated: 1.37 -> 1.38
---
Log message:
Refactor GVNPRE to use a much smart method of uniquing value sets, and
centralize a lot of the value numbering information. No functionality change.
---
Diffs of the changes: (+451 -273)
GV
Changes in directory llvm/lib/Transforms/Scalar:
GVNPRE.cpp updated: 1.38 -> 1.39
---
Log message:
Remember to clear the maximal sets between functions.
Thanks to Nicholas for valgrinding this.
---
Diffs of the changes: (+3 -2)
GVNPRE.cpp |5 +++--
1 files changed, 3 insertions(+), 2
Revision: 128592
Author: clattner
Date: 2007-06-18 22:10:41 -0700 (Mon, 18 Jun 2007)
Log Message:
---
Fix bugs handling the ldmxcsr and stmxcsr builtins
Modified Paths:
--
apple-local/branches/llvm/gcc/config/i386/llvm-i386.cpp
Modified: apple-local/branches/llvm/gc
On Jun 18, 2007, at 6:25 PM, Evan Cheng wrote:
> Log message:
>
> Replace TargetInstrInfo::CanBeDuplicated() with a M_NOT_DUPLICABLE
> bit.
>
Thanks Evan!
-Chris
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman
> Replace M_REMATERIALIZIBLE and the newly-added
> isOtherReMaterializableLoad
> with a general target hook to identify rematerializable
> instructions. Some
> instructions are only rematerializable with specific operands, such
> as loads
...
> // Floating point constant loads.
> -let isReM
Changes in directory llvm/lib/Transforms/Scalar:
GVNPRE.cpp updated: 1.39 -> 1.40
---
Log message:
Be careful to erase values from all of the appropriate sets when they're not
needed anymore. This fixes a few more memory-related issues.
---
Diffs of the changes: (+14 -2)
GVNPRE.cpp | 1
Changes in directory llvm/test/Transforms/GVNPRE:
2007-06-18-EraseNumbering.ll added (r1.1)
---
Log message:
Add a new testcase for memory corruption issues.
---
Diffs of the changes: (+26 -0)
2007-06-18-EraseNumbering.ll | 26 ++
1 files changed, 26 insertions(+)
Changes in directory llvm/lib/Transforms/Scalar:
InstructionCombining.cpp updated: 1.772 -> 1.773
---
Log message:
silence a bogus warning Duraid ran into.
---
Diffs of the changes: (+1 -1)
InstructionCombining.cpp |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
Index: llvm/li
Changes in directory llvm/lib/Target/PowerPC:
PPCISelLowering.cpp updated: 1.273 -> 1.274
---
Log message:
describe an argument, hide it.
---
Diffs of the changes: (+3 -1)
PPCISelLowering.cpp |4 +++-
1 files changed, 3 insertions(+), 1 deletion(-)
Index: llvm/lib/Target/PowerPC/PPC
Changes in directory llvm/docs:
GettingStarted.html updated: 1.162 -> 1.163
---
Log message:
don't use binutils 2.17
---
Diffs of the changes: (+9 -3)
GettingStarted.html | 12 +---
1 files changed, 9 insertions(+), 3 deletions(-)
Index: llvm/docs/GettingStarted.html
diff -u ll
Changes in directory llvm/test/Transforms/GVNPRE:
2007-06-18-ConstantInPhi.ll added (r1.1)
---
Log message:
Testcase for instances where a constant only occurs as an operand to a phi node.
---
Diffs of the changes: (+22 -0)
2007-06-18-ConstantInPhi.ll | 22 ++
1 files
Changes in directory llvm/lib/CodeGen:
LLVMTargetMachine.cpp updated: 1.14 -> 1.15
---
Log message:
document and hide two options.
---
Diffs of the changes: (+5 -2)
LLVMTargetMachine.cpp |7 +--
1 files changed, 5 insertions(+), 2 deletions(-)
Index: llvm/lib/CodeGen/LLVMTargetMa
Changes in directory llvm/utils/TableGen:
DAGISelEmitter.cpp updated: 1.297 -> 1.298
---
Log message:
silence warning when assertions are disabled.
---
Diffs of the changes: (+1 -2)
DAGISelEmitter.cpp |3 +--
1 files changed, 1 insertion(+), 2 deletions(-)
Index: llvm/utils/TableGen/
61 matches
Mail list logo
