Changes in directory llvm/lib/Target/X86:
X86ISelLowering.cpp updated: 1.310 -> 1.311
X86ISelLowering.h updated: 1.81 -> 1.82
X86InstrSSE.td updated: 1.174 -> 1.175
---
Log message:
With SSE2, expand FCOPYSIGN to a series of SSE bitwise operations.
---
Diffs of the changes: (+93 -8)
X86ISel
Changes in directory llvm/lib/Transforms/Scalar:
InstructionCombining.cpp updated: 1.580 -> 1.581
---
Log message:
Implement InstCombine/vec_shuffle.ll:%test7, simplifying shuffles with
undef operands.
---
Diffs of the changes: (+23 -1)
InstructionCombining.cpp | 24
Changes in directory llvm/test/Regression/Transforms/InstCombine:
vec_shuffle.ll updated: 1.6 -> 1.7
---
Log message:
new test
---
Diffs of the changes: (+4 -0)
vec_shuffle.ll |4
1 files changed, 4 insertions(+)
Index: llvm/test/Regression/Transforms/InstCombine/vec_shuffle.ll
Changes in directory llvm/test/Regression/Transforms/InstCombine:
vec_shuffle.ll updated: 1.5 -> 1.6
---
Log message:
manually upgrade this testcase
---
Diffs of the changes: (+27 -27)
vec_shuffle.ll | 54 +++---
1 files changed, 27 inserti
Changes in directory llvm-test/External/SPEC/CFP2006:
Makefile updated: 1.3 -> 1.4
---
Log message:
Filter out Fortran tests if fortran compiler isn't available.
---
Diffs of the changes: (+22 -5)
Makefile | 27 ++-
1 files changed, 22 insertions(+), 5 deletions(-)
Changes in directory llvm-test:
configure updated: 1.42 -> 1.43
---
Log message:
Auto-generate configure.
---
Diffs of the changes: (+8 -2)
configure | 10 --
1 files changed, 8 insertions(+), 2 deletions(-)
Index: llvm-test/configure
diff -u llvm-test/configure:1.42 llvm-test/c
Changes in directory llvm-test/autoconf:
configure.ac updated: 1.40 -> 1.41
---
Log message:
Add an extra check for f95 sanity. It seems some versions of autoconf may end
up generate code that does not initialize $sanity to 0.
---
Diffs of the changes: (+6 -2)
configure.ac |8 ++--
On Jan 4, 2007, at 7:19 PM, Reid Spencer wrote:
> On Thu, 2007-01-04 at 21:03 -0600, Chris Lattner wrote:
>>
>> Changes in directory llvm/test/Regression/Transforms/InstCombine:
>>
>> add.ll updated: 1.33 -> 1.34
>> ---
>> Log message:
>>
>> llvm upgrade doesn't accept 'define'
>
> I would prefer
On Thu, 2007-01-04 at 21:03 -0600, Chris Lattner wrote:
>
> Changes in directory llvm/test/Regression/Transforms/InstCombine:
>
> add.ll updated: 1.33 -> 1.34
> ---
> Log message:
>
> llvm upgrade doesn't accept 'define'
I would prefer it if you not use llvm-upgrade for new test cases.
>
>
>
Changes in directory llvm/lib/Transforms/Scalar:
InstructionCombining.cpp updated: 1.579 -> 1.580
---
Log message:
fold things like a^b != c^a -> b != c. This implements
InstCombine/xor.ll:test27
---
Diffs of the changes: (+33 -12)
InstructionCombining.cpp | 45
Changes in directory llvm/test/Regression/Transforms/InstCombine:
xor.ll updated: 1.18 -> 1.19
---
Log message:
new testcase
---
Diffs of the changes: (+9 -0)
xor.ll |9 +
1 files changed, 9 insertions(+)
Index: llvm/test/Regression/Transforms/InstCombine/xor.ll
diff -u llvm/
Changes in directory llvm/test/Regression/Transforms/InstCombine:
add.ll updated: 1.33 -> 1.34
---
Log message:
llvm upgrade doesn't accept 'define'
---
Diffs of the changes: (+2 -1)
add.ll |3 ++-
1 files changed, 2 insertions(+), 1 deletion(-)
Index: llvm/test/Regression/Transforms
Changes in directory llvm/lib/Transforms/Scalar:
InstructionCombining.cpp updated: 1.578 -> 1.579
---
Log message:
Compile X + ~X to -1. This implements Instcombine/add.ll:test34
---
Diffs of the changes: (+7 -1)
InstructionCombining.cpp |8 +++-
1 files changed, 7 insertions(+),
Changes in directory llvm/test/Regression/Transforms/InstCombine:
add.ll updated: 1.32 -> 1.33
---
Log message:
new testcase
---
Diffs of the changes: (+6 -0)
add.ll |6 ++
1 files changed, 6 insertions(+)
Index: llvm/test/Regression/Transforms/InstCombine/add.ll
diff -u llvm/tes
Changes in directory llvm/lib/CodeGen/SelectionDAG:
SelectionDAGISel.cpp updated: 1.339 -> 1.340
---
Log message:
GEP subscript is interpreted as a signed value.
---
Diffs of the changes: (+1 -1)
SelectionDAGISel.cpp |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
Index: llvm/li
Changes in directory llvm/docs:
LangRef.html updated: 1.186 -> 1.187
---
Log message:
Correct the documentation for function declarations. They can have dllimport
or extern_weak linkage as well as "externally visible".
---
Diffs of the changes: (+5 -3)
LangRef.html |8 +---
1 files
This patch causes us to compile functions like:
p1 (f2, l)
float f2; short l; {
printf("%d\n", l);
}
into:
define i32 %p1(double %f2, i32 %l) {
instead of:
define i32 %p1(float %f2, i16 %l) {
due to K&R promotion rules. This fixes a miscompilation of
SingleSource/UnitTests/2007-01-04-K
Changes in directory llvm-test/SingleSource/UnitTests:
2007-01-04-KNR-Args.c added (r1.1)
---
Log message:
new testcase
---
Diffs of the changes: (+19 -0)
2007-01-04-KNR-Args.c | 19 +++
1 files changed, 19 insertions(+)
Index: llvm-test/SingleSource/UnitTests/2007-01-0
Changes in directory llvm/test/Regression/CodeGen/PowerPC:
2007-01-04-ArgExtension.ll updated: 1.1 -> 1.2
---
Log message:
add missing flags
---
Diffs of the changes: (+2 -2)
2007-01-04-ArgExtension.ll |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
Index: llvm/test/Regress
Changes in directory llvm/lib/CodeGen/SelectionDAG:
SelectionDAGISel.cpp updated: 1.338 -> 1.339
---
Log message:
fix PowerPC/2007-01-04-ArgExtension.ll, a bug handling K&R prototypes with
the recent signless changes.
---
Diffs of the changes: (+6 -4)
SelectionDAGISel.cpp | 10 ++
Changes in directory llvm/test/Regression/CodeGen/PowerPC:
2007-01-04-ArgExtension.ll added (r1.1)
---
Log message:
new testcase
---
Diffs of the changes: (+10 -0)
2007-01-04-ArgExtension.ll | 10 ++
1 files changed, 10 insertions(+)
Index: llvm/test/Regression/CodeGen/PowerPC/
Changes in directory llvm/docs:
ProgrammersManual.html updated: 1.100 -> 1.101
---
Log message:
Fix broken links, reported by Baptiste Lepilleur
---
Diffs of the changes: (+4 -4)
ProgrammersManual.html |8
1 files changed, 4 insertions(+), 4 deletions(-)
Index: llvm/docs/Pro
Changes in directory llvm/lib/CodeGen/SelectionDAG:
LegalizeDAG.cpp updated: 1.451 -> 1.452
---
Log message:
Expand fcopysign to a series of bitwise of operations when it's profitable to
do so.
---
Diffs of the changes: (+46 -19)
LegalizeDAG.cpp | 65 ++
On Thu, 2007-01-04 at 10:58 -0800, Chris Lattner wrote:
> There is a typo in the function header (lossleslly) of
> Type::canLosslesslyBitCastTo:
>
>
> // canLosslesllyBitCastTo - Return true if this type can be converted to
> // 'Ty' without any reinterpretation of bits. For example, uint to i
On Thu, 2007-01-04 at 10:58 -0800, Chris Lattner wrote:
> There is a typo in the function header (lossleslly) of
> Type::canLosslesslyBitCastTo:
>
>
> // canLosslesllyBitCastTo - Return true if this type can be converted to
> // 'Ty' without any reinterpretation of bits. For example, uint to i
There is a typo in the function header (lossleslly) of
Type::canLosslesslyBitCastTo:
// canLosslesllyBitCastTo - Return true if this type can be converted to
// 'Ty' without any reinterpretation of bits. For example, uint to int.
//
>// At this point we have only various mismatches of the
Chris,
Yes, this looks fine.
On Thu, 2007-01-04 at 12:46 -0600, Chris Lattner wrote:
>
> Changes in directory llvm/tools/llvm-upgrade:
>
> UpgradeParser.y updated: 1.31 -> 1.32
> ---
> Log message:
>
> If we hit a parse error, emit something bad to the output stream. This
> ensures that
> l
On Jan 4, 2007, at 10:42 AM, Reid Spencer wrote:
> On Thu, 2007-01-04 at 09:54 -0800, Chris Lattner wrote:
>> Perhaps in addition to returning an error code, llvm-upgrade should
>> emit something like "parse failed" to the output stream, so that
>> later llvm-as's will die.
>
> How about we make ll
Applied:
Index: llvm-convert.cpp
===
--- llvm-convert.cpp(revision 121795)
+++ llvm-convert.cpp(working copy)
@@ -571,7 +571,7 @@ Value *TreeToLLVM::Emit(tree exp, Value
case UNGT_EXPR: Result = EmitCompare(exp, 0, 0,
F
Changes in directory llvm/tools/llvm-upgrade:
UpgradeParser.y updated: 1.31 -> 1.32
---
Log message:
If we hit a parse error, emit something bad to the output stream. This ensures
that
llvm-upgrade < foo | llvm-as | llvm-dis
will fail if llvm-upgrade fails.
---
Diffs of the changes: (+2
On Jan 4, 2007, at 10:04 AM, Reid Spencer wrote:
> On Thu, 2007-01-04 at 09:56 -0800, Chris Lattner wrote:
>> On Jan 4, 2007, at 9:45 AM, Reid Spencer wrote:
>
>>
>> != and == are currently right (UNE and OEQ). It's islessgreater
>> (LTGT_EXPR) that is wrong.
>
> Okay, can you just fix it then?
On Thu, 2007-01-04 at 09:54 -0800, Chris Lattner wrote:
> Perhaps in addition to returning an error code, llvm-upgrade should
> emit something like "parse failed" to the output stream, so that
> later llvm-as's will die.
How about we make llvm-as produce at least a warning and a non-zero
resu
>> Perhaps in addition to returning an error code, llvm-upgrade should
>> emit something like "parse failed" to the output stream, so that
>> later llvm-as's will die.
>
> Actually, it looks like there was something seriously broken locally
> in my tree. Cleaning llvm-upgrade and rebuilding it see
On Jan 4, 2007, at 9:54 AM, Chris Lattner wrote:
>
> On Jan 4, 2007, at 9:49 AM, Chris Lattner wrote:
>
>>
>> llvm-upgrade gets a parse error on this testcase, so it's not being
>> tested at all. This appears to be an llvm-upgrade bug:
>>
>> $ llvm-upgrade 2002-03-21-MissedRaise.ll
>> llvm-upg
Chris,
Here is what llvm-upgrade produces for this test case:
%Hash = type { { uint, sbyte*, \2 }**, int( uint )*, int }*
%hash = type { { uint, sbyte*, \2 }**, int( uint )*, int }
%hash_el = type { uint, sbyte*, \2 }*
implementation
define %Hash "MakeHash"(int %size, int( uint )* %map) {
%r
This is another testcase which isn't being tested because llvm-
upgrade apparently doesn't support "target".
-Chris
On Dec 29, 2006, at 12:02 PM, Reid Spencer wrote:
>
>
> Changes in directory llvm/test/Regression/Transforms/SimplifyCFG:
>
> 2006-12-08-Ptr-ICmp-Branch.ll updated: 1.1 -> 1.2
>
On Thu, 2007-01-04 at 09:56 -0800, Chris Lattner wrote:
> On Jan 4, 2007, at 9:45 AM, Reid Spencer wrote:
>
> != and == are currently right (UNE and OEQ). It's islessgreater
> (LTGT_EXPR) that is wrong.
Okay, can you just fix it then? I don't have a head version of your
repository from which
On Jan 4, 2007, at 9:45 AM, Reid Spencer wrote:
> On Thu, 2007-01-04 at 09:37 -0800, Chris Lattner wrote:
>> On Dec 22, 2006, at 10:08 PM, Reid Spencer wrote:
>>
>>> All,
>>>
>>> As the SetCondInst instruction has been replaced with the new
>>> FCmpInst
>>> and ICmpInst instrucitons in LLVM, th
Changes in directory llvm/docs/CommandGuide:
llvmc.pod updated: 1.12 -> 1.13
---
Log message:
Document this tool as experimental and list its deficiencies.
---
Diffs of the changes: (+28 -3)
llvmc.pod | 31 ---
1 files changed, 28 insertions(+), 3 deletions(-)
On Jan 4, 2007, at 9:49 AM, Chris Lattner wrote:
>
> llvm-upgrade gets a parse error on this testcase, so it's not being
> tested at all. This appears to be an llvm-upgrade bug:
>
> $ llvm-upgrade 2002-03-21-MissedRaise.ll
> llvm-upgrade: parse error
> 2002-03-21-MissedRaise.ll:7: while readi
Changes in directory llvm/include/llvm:
DerivedTypes.h updated: 1.75 -> 1.76
---
Log message:
remove the 'protected' workaround now that we don't care about gcc 2.95 anymore.
Reid already did this for FunctionType, this just cleans the rest up.
---
Diffs of the changes: (+0 -39)
DerivedTyp
llvm-upgrade gets a parse error on this testcase, so it's not being
tested at all. This appears to be an llvm-upgrade bug:
$ llvm-upgrade 2002-03-21-MissedRaise.ll
llvm-upgrade: parse error
2002-03-21-MissedRaise.ll:7: while reading token: '%Hash'
-Chris
On Dec 29, 2006, at 12:02 PM, Reid
Changes in directory llvm/test/Regression/Transforms/LevelRaise:
2002-02-11-ArrayShape.ll (r1.7) removed
---
Log message:
remove xfailed test that depends on obsolete argument to lli
---
Diffs of the changes: (+0 -0)
0 files changed
___
llvm-c
On Thu, 2007-01-04 at 09:37 -0800, Chris Lattner wrote:
> On Dec 22, 2006, at 10:08 PM, Reid Spencer wrote:
>
> > All,
> >
> > As the SetCondInst instruction has been replaced with the new FCmpInst
> > and ICmpInst instrucitons in LLVM, the attached patch compensates for
> > this in llvm-gcc. Sinc
On Dec 22, 2006, at 10:08 PM, Reid Spencer wrote:
> All,
>
> As the SetCondInst instruction has been replaced with the new FCmpInst
> and ICmpInst instrucitons in LLVM, the attached patch compensates for
> this in llvm-gcc. Since the Apple developers are unlikely to commit
> this
> for over a w
Combined patch [checked in.] Waiting for reviews.
-- Jim
rs6000.patch
Description: Binary data
smime.p7s
Description: S/MIME cryptographic signature
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/l
That is *such* a welcome change :)
Reid.
On Thu, 2007-01-04 at 11:41 -0400, Jim Laskey wrote:
> This is just a start. Next round will clean up #includes and macro
> usage.
>
> -- Jim
>
>
> ___
> llvm-commits mailing list
> llvm-commits@cs.uiuc.edu
Changes in directory llvm/test/Regression/Transforms/SimplifyCFG:
2006-12-08-Ptr-ICmp-Branch.ll updated: 1.3 -> 1.4
---
Log message:
Remove a manual renaming of a variable that was introduced before
llvm-upgrade could properly handle collapsed type planes.
---
Diffs of the changes: (+1 -1)
Changes in directory llvm/docs:
LangRef.html updated: 1.185 -> 1.186
---
Log message:
Clean up usage of "unsigned" and "signed" in the documentation to indicate
only that specific instructions regard their operands as signed and
unsigned not that the operands *are* signed or unsigned.
---
Di
This is just a start. Next round will clean up #includes and macro
usage.
-- Jim
rs6000.patch
Description: Binary data
smime.p7s
Description: S/MIME cryptographic signature
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.c
Changes in directory llvm/test/Regression/CodeGen/ARM:
ctors_dtors.ll added (r1.1)
---
Log message:
Add test for constructor and destructor sections.
---
Diffs of the changes: (+16 -0)
ctors_dtors.ll | 16
1 files changed, 16 insertions(+)
Index: llvm/test/Regression/
Changes in directory llvm/test/Regression/CodeGen/ARM:
extloadi1.ll added (r1.1)
---
Log message:
Add extload(i1) test.
---
Diffs of the changes: (+22 -0)
extloadi1.ll | 22 ++
1 files changed, 22 insertions(+)
Index: llvm/test/Regression/CodeGen/ARM/extloadi1.ll
di
Changes in directory llvm/test/Regression/CodeGen/ARM:
fcopysign.ll added (r1.1)
---
Log message:
Add test for FCOPYSIGN.
---
Diffs of the changes: (+21 -0)
fcopysign.ll | 21 +
1 files changed, 21 insertions(+)
Index: llvm/test/Regression/CodeGen/ARM/fcopysign.ll
d
Changes in directory llvm/lib/Target/ARM:
ARMISelDAGToDAG.cpp updated: 1.96 -> 1.97
---
Log message:
Expand SELECT (f32/f64) and FCOPYSIGN (f32/f64).
---
Diffs of the changes: (+5 -0)
ARMISelDAGToDAG.cpp |5 +
1 files changed, 5 insertions(+)
Index: llvm/lib/Target/ARM/ARMISelDAG
Changes in directory llvm/docs:
CodingStandards.html updated: 1.33 -> 1.34
---
Log message:
Removed reference to "cnull" because typical users shouldn't need to use
it.
---
Diffs of the changes: (+1 -7)
CodingStandards.html |8 +---
1 files changed, 1 insertion(+), 7 deletions(-)
On Jan 3, 2007, at 11:08 PM, Chris Lattner wrote:
>
>
> Changes in directory llvm/docs:
>
> GettingStarted.html updated: 1.150 -> 1.151
> ---
> Log message:
>
> ding dong llvm-gcc3 is dead
>
>
YAY THANK GOD
-bw
___
llvm-commits mailing list
56 matches
Mail list logo