Changes in directory llvm/lib/CodeGen:
RegAllocBigBlock.cpp updated: 1.3 -> 1.4
---
Log message:
pull evan's fixes - should help the nightly tester (but there are still
some issues)
---
Diffs of the changes: (+63 -48)
RegAllocBigBlock.cpp | 111 ---
Hi Duraid,
> +#include
Please remove this. It's forbidden in the main libraries.
Thanks!
-bw
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Changes in directory llvm/lib/CodeGen:
RegAllocBigBlock.cpp updated: 1.4 -> 1.5
---
Log message:
revert evan's fixes (and my doofusness) since they had a huge code
quality hit. will look at this soon.
---
Diffs of the changes: (+31 -63)
RegAllocBigBlock.cpp | 94
Changes in directory llvm/lib/CodeGen:
RegAllocBigBlock.cpp updated: 1.5 -> 1.6
---
Log message:
ok, this much doesn't seem to bork anything
---
Diffs of the changes: (+48 -25)
RegAllocBigBlock.cpp | 73 +--
1 files changed, 48 insertions(
Changes in directory llvm/lib/CodeGen:
RegAllocBigBlock.cpp updated: 1.6 -> 1.7
---
Log message:
ok, this is something of a dirty hack, but it seems to work. (fixes e.g.
the SPASS miscompilation)
---
Diffs of the changes: (+16 -2)
RegAllocBigBlock.cpp | 18 --
1 files ch
Changes in directory llvm/lib/Transforms/Scalar:
IndVarSimplify.cpp updated: 1.123 -> 1.124
---
Log message:
Fix a bug.
---
Diffs of the changes: (+4 -1)
IndVarSimplify.cpp |5 -
1 files changed, 4 insertions(+), 1 deletion(-)
Index: llvm/lib/Transforms/Scalar/IndVarSimplify.cpp
Changes in directory llvm/lib/CodeGen/SelectionDAG:
LegalizeDAG.cpp updated: 1.501 -> 1.502
---
Log message:
Make the comment for ScalarizeVectorOp mention that it is only for use
with single-element vectors.
---
Diffs of the changes: (+6 -4)
LegalizeDAG.cpp | 10 ++
1 files chan
Changes in directory llvm/include/llvm/Support:
Debug.h updated: 1.15 -> 1.16
---
Log message:
Allow DOUT to be used outside of the llvm namespace.
---
Diffs of the changes: (+2 -2)
Debug.h |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
Index: llvm/include/llvm/Support/Deb
Changes in directory llvm-test:
Makefile.programs updated: 1.271 -> 1.272
---
Log message:
spass+oggenc work, therefore everything works(tm) ;)
---
Diffs of the changes: (+1 -1)
Makefile.programs |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
Index: llvm-test/Makefile.program
Changes in directory llvm/lib/CodeGen/SelectionDAG:
SelectionDAGISel.cpp updated: 1.469 -> 1.470
---
Log message:
Use getVectorTypeBreakdown in FunctionLoweringInfo::CreateRegForValue
to compute the number and type of registers needed for vector values
instead of computing it manually. This fix
>> Index: llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
>> diff -u llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp:1.409 llvm/
>> lib/CodeGen/SelectionDAG/SelectionDAG.cpp:1.410
>> --- llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp:1.409 Fri Jun 22
>> 09:59:07 2007
>> +++ llvm/lib/CodeGen/Se
Changes in directory llvm/lib/Target/X86:
X86TargetAsmInfo.cpp updated: 1.42 -> 1.43
---
Log message:
Remove a redundant newline in the asm output for ELF .rodata sections.
---
Diffs of the changes: (+1 -1)
X86TargetAsmInfo.cpp |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
In
Changes in directory llvm/include/llvm/CodeGen:
ValueTypes.h updated: 1.38 -> 1.39
---
Log message:
Document the encoding of MVT::ValueType.
---
Diffs of the changes: (+6 -0)
ValueTypes.h |6 ++
1 files changed, 6 insertions(+)
Index: llvm/include/llvm/CodeGen/ValueTypes.h
diff -
+
+ // If either operand is undef, the result is undef
+ if (N0.getOpcode() == ISD::UNDEF || N1.getOpcode() == ISD::UNDEF)
+return DAG.getNode(ISD::UNDEF, VT);
+
return SDOperand();
}
>>>
>>> This is not safe for sdiv/udiv. Safe xforms are:
>>>
>>>/
Changes in directory llvm/include/llvm/Target:
TargetLowering.h updated: 1.132 -> 1.133
---
Log message:
Rename ("shrinkify") MVT::isExtendedValueType to MVT::isExtendedVT.
---
Diffs of the changes: (+13 -13)
TargetLowering.h | 26 +-
1 files changed, 13 insertion
Changes in directory llvm/lib/VMCore:
ValueTypes.cpp updated: 1.20 -> 1.21
---
Log message:
Rename ("shrinkify") MVT::isExtendedValueType to MVT::isExtendedVT.
---
Diffs of the changes: (+2 -2)
ValueTypes.cpp |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
Index: llvm/lib/V
Changes in directory llvm/lib/CodeGen/SelectionDAG:
LegalizeDAG.cpp updated: 1.502 -> 1.503
SelectionDAG.cpp updated: 1.410 -> 1.411
---
Log message:
Rename ("shrinkify") MVT::isExtendedValueType to MVT::isExtendedVT.
---
Diffs of the changes: (+2 -2)
LegalizeDAG.cpp |2 +-
SelectionD
Changes in directory llvm/include/llvm/CodeGen:
ValueTypes.h updated: 1.39 -> 1.40
---
Log message:
Rename ("shrinkify") MVT::isExtendedValueType to MVT::isExtendedVT.
---
Diffs of the changes: (+6 -6)
ValueTypes.h | 12 ++--
1 files changed, 6 insertions(+), 6 deletions(-)
Ind
>> @@ -76,64 +73,55 @@
>> iPTR = 255
>
> is 255 a good sentinel for iPTR anymore? Would we gain anything by
> making SimpleTypeBits less than 8?
If someone ever wants to actually do the arbitrary bitwidth stuff in
codegen this will need to be revisited, but right now 8 leaves 24
On Jun 27, 2007, at 2:50 AM, Zhou Sheng wrote:
> Fix a bug.
Testcase plz! What does this do, it seems wrong.
-Chris
> ---
> Diffs of the changes: (+4 -1)
>
> IndVarSimplify.cpp |5 -
> 1 files changed, 4 insertions(+), 1 deletion(-)
>
>
> Index: llvm/lib/Transforms/Scalar/IndVarSimpli
Changes in directory llvm/lib/Transforms/Scalar:
GVNPRE.cpp updated: 1.60 -> 1.61
---
Log message:
Fold a lot of code into two cases: binary instructions and ternary instructions.
This saves many lines of code duplication. No functionality change.
---
Diffs of the changes: (+62 -269)
GVNP
On Jun 27, 2007, at 9:16 AM, Dan Gohman wrote:
>>> @@ -76,64 +73,55 @@
>>> iPTR = 255
>>
>> is 255 a good sentinel for iPTR anymore? Would we gain anything by
>> making SimpleTypeBits less than 8?
>
> If someone ever wants to actually do the arbitrary bitwidth stuff in
> codegen t
Changes in directory llvm/utils/TableGen:
RegisterInfoEmitter.cpp updated: 1.58 -> 1.59
---
Log message:
One additional field in TargetRegisterDesc.
---
Diffs of the changes: (+1 -1)
RegisterInfoEmitter.cpp |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
Index: llvm/utils/TableG
On Jun 27, 2007, at 7:59 AM, Dan Gohman wrote:
>>> Index: llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
>>> diff -u llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp:1.409 llvm/
>>> lib/CodeGen/SelectionDAG/SelectionDAG.cpp:1.410
>>> --- llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp:1.409Fri J
>> I think that undef udiv intmax -> 0, no? If not, plz update
>> instcombine as well.
>
> intmax udiv intmax -> 1.
>
> It seems like folding undef/X to undef isn't safe either though, with
> the way it sounds like undef is intended to work. This code:
>
> %x = udiv i32 undef, %intmax
> %y = u
Changes in directory llvm/lib/Transforms/Scalar:
GVNPRE.cpp updated: 1.61 -> 1.62
---
Log message:
Use cached information that has already been computed to make clean() simpler
and faster. This is a small speedup on most cases.
---
Diffs of the changes: (+31 -52)
GVNPRE.cpp | 83 ++
Changes in directory llvm/lib/Support:
ConstantRange.cpp (r1.16) removed
---
Log message:
This has moved to lib/Analysis.
---
Diffs of the changes: (+0 -0)
0 files changed
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.
This is an initial implementation of ARM JIT support. So far, it hasn't
generated code machine yet. Next patch must provide such feature. The
tar.gzfile must be unpacked under lib/Target/ARM directory. Please,
send me any
feedback.
Best regards, Raul.
--
Raul Fernandes Herbster
Embedded and Perv
Changes in directory llvm/lib/CodeGen/SelectionDAG:
SelectionDAGISel.cpp updated: 1.470 -> 1.471
---
Log message:
Partial fix for PR1502: http://llvm.org/PR1502 : If a EH register is needed in
a successor of landing pad, add it as livein to all the blocks in the paths
between the landing pad
Changes in directory llvm/lib/CodeGen:
LiveIntervalAnalysis.cpp updated: 1.249 -> 1.250
---
Log message:
If a livein is not used in the block. It's live through.
---
Diffs of the changes: (+8 -5)
LiveIntervalAnalysis.cpp | 13 -
1 files changed, 8 insertions(+), 5 deletions(-)
Changes in directory llvm-www:
GSoCProjects.html updated: 1.2 -> 1.3
---
Log message:
Add link to the llvm-qemu project home. Btw, should we "add" this page
somewhere?
---
Diffs of the changes: (+2 -1)
GSoCProjects.html |3 ++-
1 files changed, 2 insertions(+), 1 deletion(-)
Index:
Changes in directory llvm-www:
GSoCProjects.html updated: 1.3 -> 1.4
---
Log message:
Make it to be real link
---
Diffs of the changes: (+5 -2)
GSoCProjects.html |7 +--
1 files changed, 5 insertions(+), 2 deletions(-)
Index: llvm-www/GSoCProjects.html
diff -u llvm-www/GSoCProjec
>
> Add link to the llvm-qemu project home. Btw, should we "add" this
> page somewhere?
GSoCProjects.html should be linked somewhere. Any suggestions?
-Chris
> ---
> Diffs of the changes: (+2 -1)
>
> GSoCProjects.html |3 ++-
> 1 files changed, 2 insertions(+), 1 deletion(-)
>
>
> Index
>>> I think that undef udiv intmax -> 0, no? If not, plz update
>>> instcombine as well.
>>
>> intmax udiv intmax -> 1.
>>
>> It seems like folding undef/X to undef isn't safe either though, with
>> the way it sounds like undef is intended to work. This code:
>>
>> %x = udiv i32 undef, %intmax
>
Changes in directory llvm/include/llvm/Analysis:
Dominators.h updated: 1.106 -> 1.107
ET-Forest.h (r1.10) removed
---
Log message:
Remove ETForest.
---
Diffs of the changes: (+1 -168)
Dominators.h | 169 ---
1 files changed, 1 insert
Changes in directory llvm/lib/VMCore:
Dominators.cpp updated: 1.118 -> 1.119
---
Log message:
Remove ETForest.
---
Diffs of the changes: (+0 -551)
Dominators.cpp | 551 -
1 files changed, 551 deletions(-)
Index: llvm/lib/VMCore/Dom
Changes in directory llvm/tools/llvm-nm:
llvm-nm.cpp updated: 1.36 -> 1.37
---
Log message:
llvm-nm must print the alias symbols.
---
Diffs of the changes: (+13 -5)
llvm-nm.cpp | 18 +-
1 files changed, 13 insertions(+), 5 deletions(-)
Index: llvm/tools/llvm-nm/llvm-nm
Something did today broke the compilation of the attached testcase.
Lauro
Error message:
[EMAIL PROTECTED]:~$ llvm/llvm/build/Debug/bin/llc <
/tmp/bugpoint-reduced-simplified.bc
llc: /home/laurov/llvm/llvm/include/llvm/Target/TargetLowering.h:125:
llvm::TargetRegisterClass*
llvm::TargetLowering:
Changes in directory llvm/test/Other:
2006-02-05-PassManager.ll updated: 1.2 -> 1.3
---
Log message:
Update. Now, -etforest is an invalid option.
---
Diffs of the changes: (+1 -1)
2006-02-05-PassManager.ll |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
Index: llvm/test/Other/2
Changes in directory llvm/lib/Target:
SubtargetFeature.cpp updated: 1.13 -> 1.14
---
Log message:
Set implied features based upon the CPU's feature list.
---
Diffs of the changes: (+7 -0)
SubtargetFeature.cpp |7 +++
1 files changed, 7 insertions(+)
Index: llvm/lib/Target/Subtarg
Hi Lauro,
This was caused by my fix for PR1529.
I'm currently considering the following patch, which fixes your attached
test case:
Index: lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
===
RCS file: /var/cvs/llvm/llvm/lib/CodeGen/Se
Changes in directory llvm/lib/Transforms/Scalar:
GVNPRE.cpp updated: 1.62 -> 1.63
---
Log message:
Make many sets a much more reasonable size. This decreases the time to optimize
Anton's testcase from 35.5s to 34.7s.
---
Diffs of the changes: (+65 -65)
GVNPRE.cpp | 130 ++
Changes in directory llvm/lib/Transforms/Scalar:
LoopUnswitch.cpp updated: 1.70 -> 1.71
---
Log message:
If a condition is not inside a loop then the condition is suitable
to loop unswitch candidate for the loop.
---
Diffs of the changes: (+7 -0)
LoopUnswitch.cpp |7 +++
1 files ch
Changes in directory llvm/lib/Transforms/Scalar:
LoopUnswitch.cpp updated: 1.71 -> 1.72
---
Log message:
Update LoopUnswitch pass to preserve DomiantorTree.
---
Diffs of the changes: (+55 -32)
LoopUnswitch.cpp | 87 ++-
1 files changed,
> If a condition is not inside a loop then the condition is suitable
> to loop unswitch candidate for the loop.
I don't understand, why reject unswitching on these values?
cases like this should be unswitched:
int foo(int arg) {
for (..)
if (arg)
...
}
-Chris
>
> ---
> Diffs of
On Jun 27, 2007, at 5:52 PM, Chris Lattner wrote:
>> If a condition is not inside a loop then the condition is suitable
>> to loop unswitch candidate for the loop.
>
> I don't understand, why reject unswitching on these values?
>
> cases like this should be unswitched:
>
> int foo(int arg) {
>
Chris,
Attached is the testcase, which will get:
opt:
/developer/home2/zsth/llvm-gcc-dev/HEAD/llvm/llvm/lib/VMCore/Constants.cpp:1559:
static llvm::Constant* llvm::ConstantExpr::getZExt(llvm::Constant*,
const llvm::Type*): Assertion `C->getType()->getPrimitiveSizeInBits() <
Ty->getPrimitiveS
Changes in directory llvm/lib/Transforms/Scalar:
LoopUnswitch.cpp updated: 1.72 -> 1.73
---
Log message:
- Undo previous check and allow loop switch for condtion that is not inside
loop.
- Avoid loop unswich for loop header branch.
- While cloning dominators fix typo and handle self dominating
Changes in directory llvm/include/llvm/Analysis:
Dominators.h updated: 1.107 -> 1.108
---
Log message:
Handle the case when block dominates itself.
---
Diffs of the changes: (+4 -1)
Dominators.h |5 -
1 files changed, 4 insertions(+), 1 deletion(-)
Index: llvm/include/llvm/Analys
Changes in directory llvm/include/llvm/Analysis:
Dominators.h updated: 1.108 -> 1.109
---
Log message:
Remove unnecessary comments.
---
Diffs of the changes: (+0 -3)
Dominators.h |3 ---
1 files changed, 3 deletions(-)
Index: llvm/include/llvm/Analysis/Dominators.h
diff -u llvm/inclu
50 matches
Mail list logo