Author: djg
Date: Mon Aug 20 14:23:34 2007
New Revision: 41189
URL: http://llvm.org/viewvc/llvm-project?rev=41189view=rev
Log:
When Intrinsic::getName is constructing names for overloaded intrinsics,
use the ValueType name instead of the llvm type name, to match what the
verifier expects. For
Author: djg
Date: Mon Aug 20 14:28:38 2007
New Revision: 41191
URL: http://llvm.org/viewvc/llvm-project?rev=41191view=rev
Log:
Minor cleanups to reduce some spurious differences between different
scheduler implementations.
Modified:
llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp
Author: djg
Date: Thu Aug 16 16:57:19 2007
New Revision: 41128
URL: http://llvm.org/viewvc/llvm-project?rev=41128view=rev
Log:
Add MVT::fAny for overloading intrinsics on floating-point types.
Modified:
llvm/trunk/include/llvm/CodeGen/ValueTypes.h
Author: djg
Date: Thu Aug 16 17:06:45 2007
New Revision: 41130
URL: http://llvm.org/viewvc/llvm-project?rev=41130view=rev
Log:
Fix the verification for overloaded intrinsic types. Check that they are
what they're supposed to be before using them.
Modified:
llvm/trunk/lib/VMCore/Verifier.cpp
Author: djg
Date: Wed Aug 15 08:42:36 2007
New Revision: 41098
URL: http://llvm.org/viewvc/llvm-project?rev=41098view=rev
Log:
Convert another test to use the count script. This one didn't fit the
regex used to convert all the others because the first '|' was on a
separate line.
Modified:
Author: djg
Date: Wed Aug 15 08:45:35 2007
New Revision: 41099
URL: http://llvm.org/viewvc/llvm-project?rev=41099view=rev
Log:
Delete extraneous uses of wc -l.
Modified:
llvm/trunk/test/CodeGen/X86/2006-05-08-CoalesceSubRegClass.ll
llvm/trunk/test/CodeGen/X86/test-pic-1.ll
Modified:
Author: djg
Date: Wed Aug 15 08:49:33 2007
New Revision: 41100
URL: http://llvm.org/viewvc/llvm-project?rev=41100view=rev
Log:
Convert tests using grep -c ... | grep ... to use the count script.
Modified:
llvm/trunk/test/CFrontend/2007-06-15-AnnotateAttribute.c
Author: djg
Date: Wed Aug 15 08:55:47 2007
New Revision: 41101
URL: http://llvm.org/viewvc/llvm-project?rev=41101view=rev
Log:
This test used wc | grep ...; convert it to use the count script.
Modified:
llvm/trunk/test/CodeGen/ARM/align.ll
Modified: llvm/trunk/test/CodeGen/ARM/align.ll
URL:
Author: djg
Date: Wed Aug 15 09:03:58 2007
New Revision: 41102
URL: http://llvm.org/viewvc/llvm-project?rev=41102view=rev
Log:
svn propdel the svn:mime-type property of application/x-troff.
Modified:
llvm/trunk/test/C++Frontend/2003-08-24-Cleanup.cpp.tr (props changed)
Author: djg
Date: Mon Aug 13 15:03:06 2007
New Revision: 41049
URL: http://llvm.org/viewvc/llvm-project?rev=41049view=rev
Log:
When x86 addresses matching exceeds its recursion limit, check to
see if the base register is already occupied before assuming it can be
used. This fixes bogus code
Why not make the script do the grep too? Something like:
countgrep foo 2
*shrug*, that's more than I needed to solve a specific problem :-). And
it's nice to be able to support grep options like -v, -F, -i, etc. in a
simple way.
Dan
--
Dan Gohman, Cray Inc
Author: djg
Date: Fri Aug 10 09:59:38 2007
New Revision: 40987
URL: http://llvm.org/viewvc/llvm-project?rev=40987view=rev
Log:
Fix EXTRACT_ELEMENT, EXTRACT_SUBVECTOR, and EXTRACT_VECTOR_ELT to
use an intptr ValueType instead of i32 for the index operand in
getCopyToParts.
Added:
getelementptrs into integer operations, which I hear basicaa
wouldn't immediately appreciate, though there are ways to fix that.
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm
Author: djg
Date: Fri Aug 10 10:07:05 2007
New Revision: 40988
URL: http://llvm.org/viewvc/llvm-project?rev=40988view=rev
Log:
Add a test script for counting lines. This can be used instead of the
| grep foo | wc -l | grep 2 idiom used by many tests, so that, for
example, tests don't mistakenly
yet still require 128-bit memory operand alignment (on
subtargets where alignment is required).
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
this
type and the regular FP128 type? The current fpext/fptrunc instructions don't
seem to cover this case. Maybe fptrunc could be generalized to cover all
floating-point conversions for which there are values for the operand type
which cannot be exactly represented in the result type.
Dan
--
Dan
Author: djg
Date: Thu Aug 2 09:27:55 2007
New Revision: 40736
URL: http://llvm.org/viewvc/llvm-project?rev=40736view=rev
Log:
Mark the SSE and MMX load instructions that
X86InstrInfo::isReallyTriviallyReMaterializable knows how to handle
with the isReMaterializable flag so that it is given a
Author: djg
Date: Thu Aug 2 16:06:40 2007
New Revision: 40754
URL: http://llvm.org/viewvc/llvm-project?rev=40754view=rev
Log:
Fix pastos in vector arithmetic intrinsics.
Modified:
llvm/trunk/lib/Target/X86/X86InstrSSE.td
Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td
URL:
Author: djg
Date: Thu Aug 2 16:17:01 2007
New Revision: 40756
URL: http://llvm.org/viewvc/llvm-project?rev=40756view=rev
Log:
Fix the alignment requirements of several unpck and shuf instructions.
Generalize isPSHUFDMask and add a unary SHUFPD pattern so that SHUFPD's
memory operand alignment
Author: djg
Date: Thu Aug 2 16:21:54 2007
New Revision: 40757
URL: http://llvm.org/viewvc/llvm-project?rev=40757view=rev
Log:
More explicit keywords.
Modified:
llvm/trunk/lib/CodeGen/BranchFolding.cpp
llvm/trunk/lib/CodeGen/VirtRegMap.h
llvm/trunk/lib/Target/ARM/ARMISelLowering.h
where I'm working on
the attached patch, which moves the getelementptr logic out of LSR into
SCEV, and generalizes it to handle more pointer expressions.
Dan
--
Dan Gohman, Cray Inc.
Index: include/llvm/Analysis/ScalarEvolutionExpander.h
;
}
The original line there was within 80 cols ;-).
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Author: djg
Date: Wed Aug 1 09:42:30 2007
New Revision: 40672
URL: http://llvm.org/viewvc/llvm-project?rev=40672view=rev
Log:
Change a .size directive to use a tab instead of a space, for consistency.
Modified:
llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp
Modified:
Author: djg
Date: Wed Aug 1 10:32:29 2007
New Revision: 40673
URL: http://llvm.org/viewvc/llvm-project?rev=40673view=rev
Log:
More explicit keywords.
Modified:
llvm/trunk/include/llvm/Analysis/LoopPass.h
llvm/trunk/include/llvm/CodeGen/MachineConstantPool.h
!= E; ++I, ++Idx) {
+if ((*I) == Arg
+ Attr-paramHasAttr(Idx, ParamAttr::NoAlias))
+ return NoAlias;
+ }
+}
+ }
Same as above.
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing
Author: djg
Date: Tue Jul 31 12:22:27 2007
New Revision: 40641
URL: http://llvm.org/viewvc/llvm-project?rev=40641view=rev
Log:
Use SCEVExpander::InsertCastOfTo instead of calling new IntToPtrInst
directly, because the insert point used by the SCEVExpander may vary
from what LSR originally
Author: djg
Date: Mon Jul 30 09:51:13 2007
New Revision: 40588
URL: http://llvm.org/viewvc/llvm-project?rev=40588view=rev
Log:
Fix pastos in comments for doFinalization functions.
Modified:
llvm/trunk/include/llvm/PassManager.h
llvm/trunk/include/llvm/PassManagers.h
Author: djg
Date: Mon Jul 30 09:51:59 2007
New Revision: 40589
URL: http://llvm.org/viewvc/llvm-project?rev=40589view=rev
Log:
More explicit keywords.
Modified:
llvm/trunk/include/llvm/Target/TargetData.h
llvm/trunk/lib/ExecutionEngine/Interpreter/Interpreter.h
Author: djg
Date: Mon Jul 30 09:58:59 2007
New Revision: 40590
URL: http://llvm.org/viewvc/llvm-project?rev=40590view=rev
Log:
Fix the comment for getClosestTargetForJIT to reflect the fact that
it does not have a Module parameter.
Modified:
Author: djg
Date: Mon Jul 30 10:01:09 2007
New Revision: 40591
URL: http://llvm.org/viewvc/llvm-project?rev=40591view=rev
Log:
Remove a FIXME comment that wasn't removed when the code it accompanied
was removed.
Modified:
llvm/trunk/tools/opt/opt.cpp
Modified: llvm/trunk/tools/opt/opt.cpp
Author: djg
Date: Mon Jul 30 10:04:59 2007
New Revision: 40592
URL: http://llvm.org/viewvc/llvm-project?rev=40592view=rev
Log:
Fix the comments for the 'fast' parameter in addPassesToEmitFile.
Modified:
llvm/trunk/include/llvm/Target/TargetMachine.h
Modified:
Author: djg
Date: Mon Jul 30 10:06:25 2007
New Revision: 40593
URL: http://llvm.org/viewvc/llvm-project?rev=40593view=rev
Log:
Print a space between the comment character and the basic block name,
for prettiness.
Modified:
llvm/trunk/lib/CodeGen/AsmPrinter.cpp
Modified:
Author: djg
Date: Mon Jul 30 10:08:02 2007
New Revision: 40594
URL: http://llvm.org/viewvc/llvm-project?rev=40594view=rev
Log:
Use tabs more consistently in assembler pseudo-ops.
Modified:
llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp
llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp
Author: djg
Date: Mon Jul 30 14:09:17 2007
New Revision: 40598
URL: http://llvm.org/viewvc/llvm-project?rev=40598view=rev
Log:
Fix a bug in getCopyFromParts turned up in the testcase for PR1132.
Added:
llvm/trunk/test/CodeGen/Alpha/illegal-element-type.ll
Author: djg
Date: Fri Jul 27 12:16:43 2007
New Revision: 40555
URL: http://llvm.org/viewvc/llvm-project?rev=40555view=rev
Log:
Re-apply 40504, but with a fix for the segfault it caused in oggenc:
Make the alignedload and alignedstore patterns always require 16-byte
alignment. This way when they
Author: djg
Date: Thu Jul 26 11:14:06 2007
New Revision: 40534
URL: http://llvm.org/viewvc/llvm-project?rev=40534view=rev
Log:
Fix the alias analysis query in DAGCombiner to not add in two
offsets. The SrcValueOffset values are the real offsets from the
SrcValue base pointers.
Modified:
Author: djg
Date: Thu Jul 26 11:06:08 2007
New Revision: 40533
URL: http://llvm.org/viewvc/llvm-project?rev=40533view=rev
Log:
Move the GET_SIDE_EFFECT_INFO logic from isInstructionTriviallyDead
to Instruction::mayWriteToMemory, fixing a FIXME, and helping
various places that call
Author: djg
Date: Thu Jul 26 10:24:15 2007
New Revision: 40529
URL: http://llvm.org/viewvc/llvm-project?rev=40529view=rev
Log:
In the .loc directive, print the fields as debug fields, so they
don't get decorated as if for immediate fields for instructions.
Modified:
Author: djg
Date: Thu Jul 26 10:11:50 2007
New Revision: 40528
URL: http://llvm.org/viewvc/llvm-project?rev=40528view=rev
Log:
Fix a whitespace difference between CMPSSrr and CMPSDrr.
Modified:
llvm/trunk/lib/Target/X86/X86InstrSSE.td
Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td
URL:
Author: djg
Date: Thu Jul 26 10:11:00 2007
New Revision: 40527
URL: http://llvm.org/viewvc/llvm-project?rev=40527view=rev
Log:
Fix a pasto in a comment.
Modified:
llvm/trunk/utils/TableGen/IntrinsicEmitter.cpp
Modified: llvm/trunk/utils/TableGen/IntrinsicEmitter.cpp
URL:
Author: djg
Date: Thu Jul 26 10:29:35 2007
New Revision: 40531
URL: http://llvm.org/viewvc/llvm-project?rev=40531view=rev
Log:
Remove a bogus return statement, what appears to have been a pasto
from Relation::contradicts in Relation::incorporate.
Modified:
Author: djg
Date: Thu Jul 26 10:25:08 2007
New Revision: 40530
URL: http://llvm.org/viewvc/llvm-project?rev=40530view=rev
Log:
DummyInst's member functions don't need to be virtual.
Modified:
llvm/trunk/lib/VMCore/BasicBlock.cpp
Modified: llvm/trunk/lib/VMCore/BasicBlock.cpp
URL:
may not be executed the
same number of times as the loop header.
Can you identify a testcase that doesn't involve an infinite loop where
the PDF test rejects a loop that the one of the other tests accepts?
Dan
--
Dan Gohman [EMAIL PROTECTED]
___
llvm
Author: djg
Date: Wed Jul 25 19:31:09 2007
New Revision: 40504
URL: http://llvm.org/viewvc/llvm-project?rev=40504view=rev
Log:
Remove X86ISD::LOAD_PACK and X86ISD::LOAD_UA and associated code from the
x86 target, replacing them with the new alignment attributes on memory
references.
Modified:
Author: djg
Date: Wed Jul 25 14:33:14 2007
New Revision: 40487
URL: http://llvm.org/viewvc/llvm-project?rev=40487view=rev
Log:
Don't ignore the return value of AsmPrinter::doInitialization and
AsmPrinter::doFinalization.
Modified:
llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp
Author: djg
Date: Tue Jul 24 18:00:27 2007
New Revision: 40480
URL: http://llvm.org/viewvc/llvm-project?rev=40480view=rev
Log:
Add const to CanBeFoldedBy, CheckAndMask, and CheckOrMask.
Modified:
llvm/trunk/include/llvm/CodeGen/SelectionDAGISel.h
+// If we can show that there is no inner cycle in the loop then we know
+// that the execution count must equal the branch count.
Would it be too intrusive to ask ScalarEvolution to use a
PostDominanceFrontier for this?
Dan
--
Dan Gohman, Cray Inc
like %bb2 is the loop header, and %cond_next is the block that
contains the exit branch. The frontier sets for these two blocks are
different, so they're not control-equivalent, and that disqualifies the
loop for what ScalarEvolution is doing here.
Dan
--
Dan Gohman, Cray Inc
Author: djg
Date: Mon Jul 23 14:26:08 2007
New Revision: 40441
URL: http://llvm.org/viewvc/llvm-project?rev=40441view=rev
Log:
Delete the svn:executable property on these files, which aren't executable.
Modified:
llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp (props changed)
Author: djg
Date: Mon Jul 23 15:24:29 2007
New Revision: 40443
URL: http://llvm.org/viewvc/llvm-project?rev=40443view=rev
Log:
Fix some uses of dyn_cast to be uses of cast.
Modified:
llvm/trunk/lib/CodeGen/DwarfWriter.cpp
llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
memory operations, not just the atomic ones.
How about llvm.membarrier, or llvm.memory.barrier if you like
Reid's proposed convention?
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman
Author: djg
Date: Fri Jul 20 11:31:26 2007
New Revision: 40101
URL: http://llvm.org/viewvc/llvm-project?rev=40101view=rev
Log:
Update these regression tests to accomodate X86InstrSSE.td now using
movups/movaps
for everything.
Modified:
llvm/trunk/test/CodeGen/X86/sse-align-10.ll
Author: djg
Date: Fri Jul 20 11:34:21 2007
New Revision: 40102
URL: http://llvm.org/viewvc/llvm-project?rev=40102view=rev
Log:
Optimize alignment of loads and stores.
Added:
llvm/trunk/test/Transforms/InstCombine/loadstore-alignment.ll
Modified:
Author: djg
Date: Fri Jul 20 18:14:50 2007
New Revision: 40130
URL: http://llvm.org/viewvc/llvm-project?rev=40130view=rev
Log:
Simplify the logic for setVolatile.
Modified:
llvm/trunk/include/llvm/Instructions.h
Modified: llvm/trunk/include/llvm/Instructions.h
URL:
Author: djg
Date: Fri Jul 20 18:17:34 2007
New Revision: 40131
URL: http://llvm.org/viewvc/llvm-project?rev=40131view=rev
Log:
Don't assume that only Uses can be kills. Defs are marked as kills initially
when there are no uses. This fixes a dangling-pointer bug, where pointers to
deleted
instruction :).
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Author: djg
Date: Wed Jul 18 11:29:46 2007
New Revision: 40004
URL: http://llvm.org/viewvc/llvm-project?rev=40004view=rev
Log:
It's not necessary to do rounding for alloca operations when the requested
alignment is equal to the stack alignment.
Added:
Author: djg
Date: Wed Jul 18 15:23:34 2007
New Revision: 40015
URL: http://llvm.org/viewvc/llvm-project?rev=40015view=rev
Log:
Implement initial memory alignment awareness for SSE instructions. Vector loads
and stores that have a specified alignment of less than 16 bytes now use
instructions that
Author: djg
Date: Wed Jul 18 15:51:11 2007
New Revision: 40016
URL: http://llvm.org/viewvc/llvm-project?rev=40016view=rev
Log:
Add constructor overloads for LoadInst and StoreInst that insert at the
end of a BasicBlock and have an alignment parameter.
Modified:
Author: djg
Date: Mon Jul 16 08:37:30 2007
New Revision: 39920
URL: http://llvm.org/viewvc/llvm-project?rev=39920view=rev
Log:
Remove the trailing semicolon from function declarations in LLVM,
where it's interpreted as a comment, not part of the syntax.
Modified:
Author: djg
Date: Mon Jul 16 09:29:03 2007
New Revision: 39921
URL: http://llvm.org/viewvc/llvm-project?rev=39921view=rev
Log:
Fix comments about vectors to use the current wording.
Modified:
llvm/trunk/include/llvm/Constants.h
llvm/trunk/include/llvm/DerivedTypes.h
Author: djg
Date: Mon Jul 16 09:37:41 2007
New Revision: 39922
URL: http://llvm.org/viewvc/llvm-project?rev=39922view=rev
Log:
Fix the documentation for llvm.sqrt so that sqrt(0) has defined behavior,
which appears to be the intent.
Modified:
llvm/trunk/docs/LangRef.html
Modified:
Author: djg
Date: Mon Jul 16 10:26:22 2007
New Revision: 39923
URL: http://llvm.org/viewvc/llvm-project?rev=39923view=rev
Log:
Use ConstantFoldFP for folding all unary floating-point operations which may
have an error, and refector out the code for binary operators into
ConstantFoldBinaryFP and
proceed.
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Author: djg
Date: Fri Jul 13 15:03:40 2007
New Revision: 39823
URL: http://llvm.org/viewvc/llvm-project?rev=39823view=rev
Log:
Don't call SimplifyVBinOp for non-vector operations, following earlier review
feedback. This theoretically makes the common (scalar) case more efficient.
Modified:
Author: djg
Date: Fri Jul 13 15:14:11 2007
New Revision: 39827
URL: http://llvm.org/viewvc/llvm-project?rev=39827view=rev
Log:
Fix the comment for LegalizeOp to more accurately reflect what it does.
Modified:
llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
Modified:
Author: djg
Date: Fri Jul 13 15:16:50 2007
New Revision: 39828
URL: http://llvm.org/viewvc/llvm-project?rev=39828view=rev
Log:
Eliminate an unused parameter.
Modified:
llvm/trunk/utils/TableGen/CodeGenTarget.cpp
llvm/trunk/utils/TableGen/CodeGenTarget.h
Modified:
Author: djg
Date: Tue Jul 10 10:19:29 2007
New Revision: 38491
URL: http://llvm.org/viewvc/llvm-project?rev=38491view=rev
Log:
Fix a bug in the folding of binary operators to undef.
Thanks to Lauro for spotting this!
Modified:
llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
Modified:
Author: djg
Date: Tue Jul 10 10:34:29 2007
New Revision: 38492
URL: http://llvm.org/viewvc/llvm-project?rev=38492view=rev
Log:
Add a regression test for folding spill code into scalar min and max.
Added:
llvm/trunk/test/CodeGen/X86/scalar-min-max-fill-operand.ll
Added:
working on a patch to add support
for intrinsic functions with vector argument and result types that
can be overloaded on vector length.
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman
Author: djg
Date: Tue Jul 10 09:20:37 2007
New Revision: 38489
URL: http://llvm.org/viewvc/llvm-project?rev=38489view=rev
Log:
Fix the folding of undef in several binary operators to recognize
undef in either the left or right operand.
Modified:
Author: djg
Date: Tue Jul 10 13:20:44 2007
New Revision: 38503
URL: http://llvm.org/viewvc/llvm-project?rev=38503view=rev
Log:
Change the peep for EXTRACT_VECTOR_ELT of BUILD_PAIR to look for
the new CONCAT_VECTORS node type instead, as that's what legalize
uses now. And add a peep for
Author: djg
Date: Mon Jul 9 10:25:17 2007
New Revision: 38457
URL: http://llvm.org/viewvc/llvm-project?rev=38457view=rev
Log:
Move the APInt form of SCEVUnknown::getIntegerSCEV to SCEVConstant::get, and
use SCEVConstant::get instead of SCEVUnknown::get when constructing a SCEV
for a ConstantInt.
and
passes with the new one.
Thanks for your help,
Dan
On Fri, Jul 06, 2007 at 01:40:33PM -0700, Evan Cheng wrote:
This seems to have fixed llvm-test failures, but Mac OS X PPC
bootstrapping is still unhappy. :-(
Evan
--
Dan Gohman, Cray Inc.
Index: lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
Author: djg
Date: Mon Jul 9 15:49:44 2007
New Revision: 38468
URL: http://llvm.org/viewvc/llvm-project?rev=38468view=rev
Log:
Initialize the IndexedModeActions array with memset before
updating it with calls to setIndexedLoadAction/setIndexedStoreAction,
which only update a few bits at a time.
Author: djg
Date: Mon Jul 9 15:59:04 2007
New Revision: 38471
URL: http://llvm.org/viewvc/llvm-project?rev=38471view=rev
Log:
Change getCopyToParts and getCopyFromParts to always use target-endian
register ordering, for both physical and virtual registers. Update the PPC
target lowering for
Author: djg
Date: Mon Jul 9 17:18:38 2007
New Revision: 38473
URL: http://llvm.org/viewvc/llvm-project?rev=38473view=rev
Log:
Preserve volatililty and alignment information when lowering or
simplifying loads and stores.
Modified:
llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
Author: djg
Date: Mon Jul 9 19:05:58 2007
New Revision: 38478
URL: http://llvm.org/viewvc/llvm-project?rev=38478view=rev
Log:
Define non-intrinsic instructions for vector min, max, sqrt, rsqrt, and rcp,
in addition to the intrinsic forms. Add spill-folding entries for these new
instructions, and
-Int64ToFP.c
Dan
On Thu, Jul 05, 2007 at 06:48:31PM -0700, Evan Cheng wrote:
Hi Dan,
I've committed this:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-
Mon-20070702/051079.html
as a workaround. Please figure out a proper fix. Thanks!
Evan
--
Dan Gohman, Cray Inc.
Index
Author: djg
Date: Fri Jul 6 08:15:51 2007
New Revision: 37943
URL: http://llvm.org/viewvc/llvm-project?rev=37943view=rev
Log:
Add an explicit triple to the big-endian tests so that the assembly
output format is always consistent with what the greps are looking for.
Modified:
. On different hosts, we're getting different defaults.
I added explicit target triples to the test cases, which should fix this.
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman
Author: djg
Date: Fri Jul 6 08:27:33 2007
New Revision: 37944
URL: http://llvm.org/viewvc/llvm-project?rev=37944view=rev
Log:
Fix a comment so that the element numbering match the actual element
operand numbering.
Modified:
llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h
Modified:
Author: djg
Date: Fri Jul 6 08:59:28 2007
New Revision: 37946
URL: http://llvm.org/viewvc/llvm-project?rev=37946view=rev
Log:
Remove redundant declarations.
Modified:
llvm/trunk/include/llvm/Target/TargetOptions.h
Modified: llvm/trunk/include/llvm/Target/TargetOptions.h
URL:
are taking endianness into consideration?
Thanks,
Evan
On Jul 2, 2007, at 9:18 AM, Dan Gohman wrote:
Author: djg
Date: Mon Jul 2 11:18:06 2007
New Revision: 37843
URL: http://llvm.org/viewvc/llvm-project?rev=37843view=rev
Log:
Replace ExpandScalarFormalArgs and ExpandScalarCallArgs
Author: djg
Date: Thu Jul 5 15:12:34 2007
New Revision: 37921
URL: http://llvm.org/viewvc/llvm-project?rev=37921view=rev
Log:
Add a parameter to getCopyToParts and getCopyFromParts to specify whether
endian swapping should be done, and update the code to use it. This fixes
some register ordering
Author: djg
Date: Thu Jul 5 15:15:43 2007
New Revision: 37922
URL: http://llvm.org/viewvc/llvm-project?rev=37922view=rev
Log:
Make the debug string for ISD::MERGE_VALUES consistent with the others.
Modified:
llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
Modified:
Author: djg
Date: Thu Jul 5 15:39:35 2007
New Revision: 37924
URL: http://llvm.org/viewvc/llvm-project?rev=37924view=rev
Log:
Make MachineFunctionPass::runOnFunction non-virtual. Subclasses override
the virtual function runOnMachineFunction instead. And add access specifiers.
Modified:
Author: djg
Date: Thu Jul 5 15:40:15 2007
New Revision: 37925
URL: http://llvm.org/viewvc/llvm-project?rev=37925view=rev
Log:
Add explicit keywords.
Modified:
llvm/trunk/include/llvm/CodeGen/IntrinsicLowering.h
llvm/trunk/include/llvm/CodeGen/MachineModuleInfo.h
Author: djg
Date: Tue Jul 3 09:03:57 2007
New Revision: 37851
URL: http://llvm.org/viewvc/llvm-project?rev=37851view=rev
Log:
Fix several over-aggressive folds for undef nodes in dagcombine, to
follow the rules for undef used in instcombine.
Modified:
,
but it is a pessimization in some obscure cases.
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
isn't very valuable to me, and the easiest
way to fix the code was to revert the change.
Looking at it a little more, I think the instcombine folds are actually
wrong here. For example, it does X + undef - undef, but that's wrong if
X is a NaN.
Dan
--
Dan Gohman, Cray Inc
Author: djg
Date: Mon Jul 2 09:53:37 2007
New Revision: 37839
URL: http://llvm.org/viewvc/llvm-project?rev=37839view=rev
Log:
Add explicit keywords.
Modified:
llvm/trunk/include/llvm/Analysis/LoopPass.h
llvm/trunk/include/llvm/CallGraphSCCPass.h
Author: djg
Date: Mon Jul 2 10:06:26 2007
New Revision: 37840
URL: http://llvm.org/viewvc/llvm-project?rev=37840view=rev
Log:
Remove declarations for code no longer in the tree, to avoid confusion.
Removed:
llvm/trunk/include/llvm/CodeGen/InstrScheduling.h
Modified:
Author: djg
Date: Mon Jul 2 10:43:20 2007
New Revision: 37841
URL: http://llvm.org/viewvc/llvm-project?rev=37841view=rev
Log:
New test case. DAGCombiner should be able to fold -sin(-x)
in -enable-unsafe-fp-math mode.
Added:
llvm/trunk/test/CodeGen/X86/negative-sin.ll
Added:
Author: djg
Date: Mon Jul 2 11:18:06 2007
New Revision: 37843
URL: http://llvm.org/viewvc/llvm-project?rev=37843view=rev
Log:
Replace ExpandScalarFormalArgs and ExpandScalarCallArgs with the newly
refactored getCopyFromParts and getCopyToParts, which are more general.
This effectively adds
Author: djg
Date: Mon Jul 2 11:21:53 2007
New Revision: 37844
URL: http://llvm.org/viewvc/llvm-project?rev=37844view=rev
Log:
Vector results may be returned in XMM0 and XMM1, not just XMM0. With
the recent lowering changes, this allows types like 4 x double to
be returned, using two vector
Author: djg
Date: Mon Jul 2 11:23:47 2007
New Revision: 37845
URL: http://llvm.org/viewvc/llvm-project?rev=37845view=rev
Log:
Add a basic test-case for passing and returning 4 x double and
8 x float values on X86.
Added:
llvm/trunk/test/CodeGen/X86/illegal-vector-args-return.ll
Added:
Author: djg
Date: Mon Jul 2 10:48:56 2007
New Revision: 37842
URL: http://llvm.org/viewvc/llvm-project?rev=37842view=rev
Log:
Teach GetNegatedExpression to negate 0-B to B in UnsafeFPMath mode, and
visitFSUB to fold 0-B to -B in UnsafeFPMath mode. Also change visitFNEG
to use
different names to do essentially the same thing.
Dan
--
Dan Gohman, Cray Inc.
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Duncan pointed out that I confused myself. If something is undef, we
can choose to pick any specific value for the undef to pick the
cancellation.
Thanks Chris and Duncan for explaining this. I'll submit a fix for the
DAGCombiner changes accordingly.
Dan
--
Dan Gohman, Cray Inc
201 - 300 of 446 matches
Mail list logo