Evan Cheng [Thu, 28 May 2009 18:48:53 +0000 (18:48 +0000)]
Fix PSIGND encoding bug. Patch by Sean Callanan.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72534
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 28 May 2009 18:41:02 +0000 (18:41 +0000)]
Incorporate patch feedbacks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72533
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 28 May 2009 18:31:40 +0000 (18:31 +0000)]
NightlyTest: Write "sentdata.txt" files and run the -submit-aux script before
sending data to the server.
- Otherwise if the server connection fails the external script never runs.
Also, create content before initiating connection to try and decrease time we
are connected to llvm.org.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72532
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjiv Gupta [Thu, 28 May 2009 18:24:11 +0000 (18:24 +0000)]
Emit debug info for locals with proper scope.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72531
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 May 2009 18:18:59 +0000 (18:18 +0000)]
Temporarily revert r72514 (and dependent patch r72516). It was causing this
failure during llvm-gcc bootstrap:
Assertion failed: (!Tmp2.getNode() && "Can't legalize BR_CC with legal condition!"), function ExpandNode, file /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvmCore.roots/llvmCore~obj/src/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp, line 2923.
/Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvmgcc42.roots/llvmgcc42~obj/src/gcc/libgcc2.c:1727: internal compiler error: Abort trap
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://developer.apple.com/bugreporter> for instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72530
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjiv Gupta [Thu, 28 May 2009 17:32:56 +0000 (17:32 +0000)]
Mark the branch insns correctly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72529
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjiv Gupta [Thu, 28 May 2009 06:59:57 +0000 (06:59 +0000)]
Fixing problems that got exposed after the refactoring of LegalizeDAG done in 72447.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72521
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Thu, 28 May 2009 05:10:20 +0000 (05:10 +0000)]
Add a testcase which got fixed by recent legalization work.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72517
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Thu, 28 May 2009 04:49:34 +0000 (04:49 +0000)]
Remove a couple of useless functions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72516
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Thu, 28 May 2009 04:39:57 +0000 (04:39 +0000)]
Remove special cases for more opcodes.
This is basically the end of this series of patches for LegalizeDAG; the
remaining special cases can't be removed without more infrastructure
work. There's a FIXME for each relevant opcode near the beginning of
SelectionDAGLegalize::LegalizeOp.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72514
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Thu, 28 May 2009 04:31:08 +0000 (04:31 +0000)]
Return the operand rather than a null SDValue when the given SELECT_CC
is actually legal. Part of LegalizeDAG cleanups.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72513
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Thu, 28 May 2009 04:08:10 +0000 (04:08 +0000)]
Use Operands.data() instead of &Operands[0] where Operands is a potentially
empty SmallVector.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72512
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Thu, 28 May 2009 03:56:57 +0000 (03:56 +0000)]
Remove special case for SETCC opcode; add some comments explaining why
some special cases are necessary.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72511
91177308-0d34-0410-b5e6-
96231b3b80d8
Jeffrey Yasskin [Thu, 28 May 2009 03:16:17 +0000 (03:16 +0000)]
This patch brings the list of attributes in CPPBackend.cpp up to date with the
list in Attributes.h. It also reorders the CPPBackend list to match so that
it's easier to see that it's complete.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72510
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Thu, 28 May 2009 03:06:16 +0000 (03:06 +0000)]
Some minor cleanups.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72509
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 May 2009 02:04:00 +0000 (02:04 +0000)]
"The instructions MMX_PSADBWrm and MMX_PSADBWrr have opcode
0b11100000 (e0), but
the Intel manual (screenshot) says it should be
0b11110110 (f6). The existing
encoding causes a disassembly conflict with MMX_PAVGBrm, which really should be
0f e0."
Patch by Sean Callanan!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72508
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 28 May 2009 00:35:15 +0000 (00:35 +0000)]
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
Dan Gohman [Wed, 27 May 2009 21:10:47 +0000 (21:10 +0000)]
Revert 72493 and replace it with a more conservative fix, for now: don't
rewrite the comparison if there is any implicit extension or truncation
on the induction variable. I'm planning for IVUsers to eventually take
over some of the work of this code, and for it to be generalized.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72496
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 20:39:00 +0000 (20:39 +0000)]
Ger rid of some dead code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72494
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Wed, 27 May 2009 20:00:18 +0000 (20:00 +0000)]
In ChangeCompareStride, when the stride to be reused is truncated to
a smaller type, promoted its offset back up to the type of the new
comparison. This fixes PR4222.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72493
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 27 May 2009 18:38:01 +0000 (18:38 +0000)]
Fix sfence jit encoding. Patch by Sean Callanan.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72488
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Wed, 27 May 2009 17:43:21 +0000 (17:43 +0000)]
This looks like it passes now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72485
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Wed, 27 May 2009 17:23:44 +0000 (17:23 +0000)]
Added support for fround, fextend and FP_TO_SINT
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72483
91177308-0d34-0410-b5e6-
96231b3b80d8
Douglas Gregor [Wed, 27 May 2009 16:52:17 +0000 (16:52 +0000)]
Minor fix for CMake build system
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72480
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Wed, 27 May 2009 15:49:33 +0000 (15:49 +0000)]
CMake: Set LIBS on llvm-config so we can query the system libraries
used by CMake with --ldflags.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72470
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 12:42:55 +0000 (12:42 +0000)]
Minor cleanups; add a better explanation for the issue with
BUILD_VECTOR.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72469
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 12:20:41 +0000 (12:20 +0000)]
Remove more special cases for opcodes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72468
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 07:58:35 +0000 (07:58 +0000)]
Remove special cases for more opcodes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72467
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 07:32:27 +0000 (07:32 +0000)]
Removing more special cases from LegalizeDAG.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72465
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 07:05:37 +0000 (07:05 +0000)]
Eliminate more special cases for opcodes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72464
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 03:33:44 +0000 (03:33 +0000)]
Remove more special cases from LegalizeDAG.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72456
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 02:21:29 +0000 (02:21 +0000)]
Remove unused argument.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72455
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 02:16:40 +0000 (02:16 +0000)]
Remove more opcode special cases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72454
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Wed, 27 May 2009 02:07:15 +0000 (02:07 +0000)]
Add braces around an array initializer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72453
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Wed, 27 May 2009 02:00:53 +0000 (02:00 +0000)]
Teach SCEVExpander to avoid creating over-indexed GEP indices when
possible. For example, it now emits
%p.2.ip.1 = getelementptr [3 x [3 x double]]* %p, i64 2, i64 %tmp, i64 1
instead of the equivalent but less obvious
%p.2.ip.1 = getelementptr [3 x [3 x double]]* %p, i64 0, i64 %tmp, i64 19
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72452
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Wed, 27 May 2009 01:48:27 +0000 (01:48 +0000)]
Teach BasicAliasAnalysis to understand constant gep indices that fall
beyond their associated static array type.
I believe that this fixes a legitimate bug, because BasicAliasAnalysis
already has code to check for this condition that works for non-constant
indices, however it was missing the case of constant indices. With this
change, it checks for both.
This fixes PR4267, and miscompiles of SPEC 188.ammp and 464.h264.href.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72451
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 01:25:56 +0000 (01:25 +0000)]
Start of refactoring LegalizeDAG so that we don't need specialized
handling for every single opcode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72447
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Wed, 27 May 2009 00:47:34 +0000 (00:47 +0000)]
Don't abuse the quirky behavior of LegalizeDAG for XINT_TO_FP and
FP_TO_XINT. Necessary for some cleanups I'm working on. Updated
from the previous version (r72431) to fix a bug and make some things a
bit clearer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72445
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Tue, 26 May 2009 23:19:19 +0000 (23:19 +0000)]
Testcase for (llvm-gcc-4.2) 72442 (PR 4242).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72443
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Tue, 26 May 2009 21:27:02 +0000 (21:27 +0000)]
Back out r72431, it is causing a number of compilation crashes with clang.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72436
91177308-0d34-0410-b5e6-
96231b3b80d8
Stefanus Du Toit [Tue, 26 May 2009 21:04:35 +0000 (21:04 +0000)]
Update CPU capabilities for AMD machines
- added processors k8-sse3, opteron-sse3, athlon64-sse3, amdfam10, and
barcelona with appropriate sse3/4a levels
- added FeatureSSE4A for amdfam10 processors
in X86Subtarget:
- added hasSSE4A
- updated AutoDetectSubtargetFeatures to detect SSE4A
- updated GetCurrentX86CPU to detect family 15 with sse3 as k8-sse3 and
family 10h as amdfam10
New processor names match those used by gcc.
Patch by Paul Redmond!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72434
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Tue, 26 May 2009 19:18:56 +0000 (19:18 +0000)]
Don't abuse the quirky behavior of LegalizeDAG for XINT_TO_FP and
FP_TO_XINT. Necessary for some cleanups I'm working on.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72431
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Tue, 26 May 2009 19:11:47 +0000 (19:11 +0000)]
For the new ar --plugin libLLVMgold.so to work we need libLTO linked with
pthreads, otherwise ar complains of unresolved references to pthread_mutex_*.
LTO doesn't actually use pthreads, but ManagedStatic does.
Fix this by linking in LIBS (that contains pthreads) for LTO and gold.
For now this links in more libs than needed (libffi for example), we can figure
out later how to link in those libs per-tool later.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72430
91177308-0d34-0410-b5e6-
96231b3b80d8
Jeffrey Yasskin [Tue, 26 May 2009 18:27:15 +0000 (18:27 +0000)]
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
Dan Gohman [Tue, 26 May 2009 17:44:05 +0000 (17:44 +0000)]
For the return type of SCEVUDivExpr, use the RHS' type instead of
that of the LHS. It doesn't matter for correctness, but the LHS
is more likely than the RHS to be a pointer type in exotic cases,
and it's more tidy to have it return the integer type.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72424
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 26 May 2009 17:42:32 +0000 (17:42 +0000)]
Give SCEVNaryExpr a doxygen comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72423
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 26 May 2009 17:41:16 +0000 (17:41 +0000)]
In cases where a pointer value is an operand of a multiplication or
division operation, don't attempt to use the operation's value as
the base of a getelementptr. This fixes PR4271.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72422
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Tue, 26 May 2009 08:55:52 +0000 (08:55 +0000)]
Delete a bunch of dead code from LegalizeDAG.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72414
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Tue, 26 May 2009 06:25:46 +0000 (06:25 +0000)]
Eliminate VarInfo::UsedBlocks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72411
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 25 May 2009 21:28:56 +0000 (21:28 +0000)]
make memdep use the getModRefInfo method for stores instead of the
low-level alias() method, allowing it to reason more aggressively
about pointers into constant memory. PR4189
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72403
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Mon, 25 May 2009 21:28:11 +0000 (21:28 +0000)]
Audit the type constructors. Previously it was possible to create [0 x void]
or use labels as members of structures for example. Also included a couple of
whitespace fixes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72402
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 25 May 2009 20:28:19 +0000 (20:28 +0000)]
add some late optimizations that GCC does. It thinks these are a win
even on Core2, not just AMD processors which was a surprise to me.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72396
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 25 May 2009 19:51:07 +0000 (19:51 +0000)]
fix typo
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72395
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 25 May 2009 16:34:44 +0000 (16:34 +0000)]
we should eventually add -march=atom and the new atom movbe instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72387
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Mon, 25 May 2009 06:29:56 +0000 (06:29 +0000)]
Fix the crash debugger to actually bisect globals once it's determined that it
can't just eliminate all global initializers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72378
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Mon, 25 May 2009 05:30:00 +0000 (05:30 +0000)]
Add a bisection step on the list of instructions before doing the linear
simplification. It's not clear to me whether this can replace the first of the
linear instruction simplification stages or not, so I left it in.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72377
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sun, 24 May 2009 23:45:28 +0000 (23:45 +0000)]
Various comment fixes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72376
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sun, 24 May 2009 23:25:42 +0000 (23:25 +0000)]
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
Eli Friedman [Sun, 24 May 2009 20:32:10 +0000 (20:32 +0000)]
Add a comment which should hopefully make the purpose of this method a
bit clearer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72374
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sun, 24 May 2009 20:29:11 +0000 (20:29 +0000)]
Minor improvement to FCOPYSIGN to use BIT_CONVERT in cases where the
corresponding integer type is legal.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72373
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sun, 24 May 2009 20:08:21 +0000 (20:08 +0000)]
Move Rewriter.clear() earlier, to avoid triggerring the AssertingVH by
one of the RecursivelyDeleteTriviallyDeadInstructions.
Add a comment explaining why the cache needs to be cleared.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72372
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sun, 24 May 2009 19:36:09 +0000 (19:36 +0000)]
Instead of clearing the rewriter, don't attempt to rewrite dead phi nodes.
Also fix 80 column violation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72371
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sun, 24 May 2009 19:11:38 +0000 (19:11 +0000)]
When rewriting the loop exit test with the canonical induction variable,
leave the original comparison in place if it has other uses, since the
other uses won't be dominated by the new comparison instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72369
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sun, 24 May 2009 19:02:45 +0000 (19:02 +0000)]
Fix this code for hosts where std::vector doesn't have .data().
Use &Ops[0] instead, which is safe since Ops will never be empty here.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72368
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sun, 24 May 2009 18:09:01 +0000 (18:09 +0000)]
When replacing a floating-point comparison with an integer
comparison, use takeName to give the integer comparison a name.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72367
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Sun, 24 May 2009 18:06:31 +0000 (18:06 +0000)]
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
Dan Gohman [Sun, 24 May 2009 18:02:35 +0000 (18:02 +0000)]
When the low bits of one operand of an add are zero, that number
of low bits of the other operand are preserved in the output.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72365
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sun, 24 May 2009 14:23:16 +0000 (14:23 +0000)]
The rewriter may hold references to instructions that are deleted because they are trivially dead.
Fix by clearing the rewriter cache before deleting the trivially dead
instructions.
Also make InsertedExpressions use an AssertingVH to catch these
bugs easier.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72364
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sun, 24 May 2009 10:21:20 +0000 (10:21 +0000)]
Rewrite ISD::FCOPYSIGN lowering to never use i64. Not really ideal, but
it's late, and I don't have any better ideas at the moment. Fixes PR4257.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72363
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sun, 24 May 2009 09:40:47 +0000 (09:40 +0000)]
Make bugpoint emit a bugpoint-reduced-blocks.bc, because bugpoint itself
can crash during instruction simplification (for example if it creates a
broken module).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72362
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sun, 24 May 2009 09:31:04 +0000 (09:31 +0000)]
Add -disable-global-remove option to bugpoint.
Sometimes when bugpointing a crash the bugpoint-reduced-simplified.bc reproduces
a totally different bug than the original one ("GV doesn't have initializer").
Although its useful to report that bug too, I need a way to reduce the original
bug, hence I introduced -disable-global-remove.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72361
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sun, 24 May 2009 09:13:13 +0000 (09:13 +0000)]
Update for CMakeLists; untested, so tell me if there are issues.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72360
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sun, 24 May 2009 08:42:01 +0000 (08:42 +0000)]
Remove checks of getTypeAction from LegalizeOp; we already assert that
all results and all operands are legal, so this change shouldn't affect
behavior at all.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72359
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sun, 24 May 2009 02:46:31 +0000 (02:46 +0000)]
Disable type legalization in LegalizeDAG.
This leaves around 4000 lines of dead code; I'll clean that up
in subsequent commits.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72358
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Sun, 24 May 2009 02:46:06 +0000 (02:46 +0000)]
Remove border around table used for layout.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72357
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sun, 24 May 2009 02:03:36 +0000 (02:03 +0000)]
Make the PPC backend use a legal type for the operands to the BUILD_VECTOR
nodes it generates.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72356
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sat, 23 May 2009 23:03:28 +0000 (23:03 +0000)]
Fix a bug in the expansion of EXTRACT_SUBVECTOR in
ExpandExtractFromVectorThroughStack.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72351
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sat, 23 May 2009 22:44:52 +0000 (22:44 +0000)]
Make the X86 backend mark EXTRACT_SUBVECTOR as Expand, at least for the
moment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72350
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sat, 23 May 2009 22:37:25 +0000 (22:37 +0000)]
Add a proper implementation of EXTRACT_SUBVECTOR legalization that
doesn't split legal vector operands. This is necessary because the
type legalization (and therefore, vector splitting) code will be going
away soon.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72349
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sat, 23 May 2009 20:40:47 +0000 (20:40 +0000)]
Use latin symbols. I was told not everybody has a UTF-8 capable editor.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72343
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sat, 23 May 2009 20:21:09 +0000 (20:21 +0000)]
Write family name last, as other entries in this file.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72340
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sat, 23 May 2009 20:19:52 +0000 (20:19 +0000)]
Add myself.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72339
91177308-0d34-0410-b5e6-
96231b3b80d8
Anton Korobeynikov [Sat, 23 May 2009 19:51:43 +0000 (19:51 +0000)]
Add ARMv7 architecture, Cortex processors and different FPU modes handling.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72337
91177308-0d34-0410-b5e6-
96231b3b80d8
Anton Korobeynikov [Sat, 23 May 2009 19:51:20 +0000 (19:51 +0000)]
Emit ARM Build Attributes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72336
91177308-0d34-0410-b5e6-
96231b3b80d8
Anton Korobeynikov [Sat, 23 May 2009 19:50:50 +0000 (19:50 +0000)]
Propagate CPU string out of SubtargetFeatures
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72335
91177308-0d34-0410-b5e6-
96231b3b80d8
Jay Foad [Sat, 23 May 2009 17:57:59 +0000 (17:57 +0000)]
Work around a page size issue on Cygwin.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72332
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sat, 23 May 2009 17:29:48 +0000 (17:29 +0000)]
Fix PR4254.
The DAGCombiner created a negative shiftamount, stored in an
unsigned variable. Later the optimizer eliminated the shift entirely as being
undefined.
Example: (srl (shl X, 56) 48). ShiftAmt is
4294967288.
Fix it by checking that the shiftamount is positive, and storing in a signed
variable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72331
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sat, 23 May 2009 16:23:59 +0000 (16:23 +0000)]
stat64/open64/lseek64 for the interpreter
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72329
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Sat, 23 May 2009 14:06:57 +0000 (14:06 +0000)]
available_externall linkage is not local, this was confusing the codegenerator,
and it wasn't generating calls through @PLT for these functions.
hasLocalLinkage() is now false for available_externally,
I attempted to fix the inliner and dce to handle available_externally properly.
It passed make check.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72328
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sat, 23 May 2009 13:15:11 +0000 (13:15 +0000)]
Fix test to account for legalization changes; I think this ends up
running an extra DAGCombine pass which improves the code a bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72326
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sat, 23 May 2009 12:35:30 +0000 (12:35 +0000)]
Add a new step to legalization to legalize vector math operations. This
will allow simplifying LegalizeDAG to eliminate type legalization. (I
have a patch to do that, but it's not quite finished; I'll commit it
once it's finished and I've fixed any review comments for this patch.)
See the comment at the beginning of
lib/CodeGen/SelectionDAG/LegalizeVectorOps.cpp for more details on the
motivation for this patch.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72325
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Sat, 23 May 2009 09:59:16 +0000 (09:59 +0000)]
Make the x86 backend custom-lower UINT_TO_FP and FP_TO_UINT on 32-bit
systems instead of attempting to promote them to a 64-bit SINT_TO_FP or
FP_TO_SINT. This is in preparation for removing the type legalization
code from LegalizeDAG: once type legalization is gone from LegalizeDAG,
it won't be able to handle the i64 operand/result correctly.
This isn't quite ideal, but I don't think any other operation for any
target ends up in this situation, so treating this case specially seems
reasonable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72324
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Sat, 23 May 2009 02:37:24 +0000 (02:37 +0000)]
CMake: Use libdl only when available. Fixes build on FreeBSD.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72311
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Fri, 22 May 2009 23:10:53 +0000 (23:10 +0000)]
Fix bug in FoldFCmp_IntToFP_Cst. If inttofp is a uintofp, use unsigned instead of signed integer constant.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72300
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 22 May 2009 21:47:08 +0000 (21:47 +0000)]
Add a note mentioning that uses of the return value of an invoke
must be dominated by the normal label.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72285
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Fri, 22 May 2009 20:55:15 +0000 (20:55 +0000)]
CMake: Use libpthread in tblgen when needed. Updated list of source
files for PIC16 target.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72277
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Fri, 22 May 2009 20:36:31 +0000 (20:36 +0000)]
Add a new codegen pass that normalizes dwarf exception handling
code in preparation for code generation. The main thing it does
is handle the case when eh.exception calls (and, in a future
patch, eh.selector calls) are far away from landing pads. Right
now in practice you only find eh.exception calls close to landing
pads: either in a landing pad (the common case) or in a landing
pad successor, due to loop passes shifting them about. However
future exception handling improvements will result in calls far
from landing pads:
(1) Inlining of rewinds. Consider the following case:
In function @f:
...
invoke @g to label %normal unwind label %unwinds
...
unwinds:
%ex = call i8* @llvm.eh.exception()
...
In function @g:
...
invoke @something to label %continue unwind label %handler
...
handler:
%ex = call i8* @llvm.eh.exception()
... perform cleanups ...
"rethrow exception"
Now inline @g into @f. Currently this is turned into:
In function @f:
...
invoke @something to label %continue unwind label %handler
...
handler:
%ex = call i8* @llvm.eh.exception()
... perform cleanups ...
invoke "rethrow exception" to label %normal unwind label %unwinds
unwinds:
%ex = call i8* @llvm.eh.exception()
...
However we would like to simplify invoke of "rethrow exception" into
a branch to the %unwinds label. Then %unwinds is no longer a landing
pad, and the eh.exception call there is then far away from any landing
pads.
(2) Using the unwind instruction for cleanups.
It would be nice to have codegen handle the following case:
invoke @something to label %continue unwind label %run_cleanups
...
handler:
... perform cleanups ...
unwind
This requires turning "unwind" into a library call, which
necessarily takes a pointer to the exception as an argument
(this patch also does this unwind lowering). But that means
you are using eh.exception again far from a landing pad.
(3) Bugpoint simplifications. When bugpoint is simplifying
exception handling code it often generates eh.exception calls
far from a landing pad, which then causes codegen to assert.
Bugpoint then latches on to this assertion and loses sight
of the original problem.
Note that it is currently rare for this pass to actually do
anything. And in fact it normally shouldn't do anything at
all given the code coming out of llvm-gcc! But it does fire
a few times in the testsuite. As far as I can see this is
almost always due to the LoopStrengthReduce codegen pass
introducing pointless loop preheader blocks which are landing
pads and only contain a branch to another block. This other
block contains an eh.exception call. So probably by tweaking
LoopStrengthReduce a bit this can be avoided.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72276
91177308-0d34-0410-b5e6-
96231b3b80d8
Bob Wilson [Fri, 22 May 2009 17:38:41 +0000 (17:38 +0000)]
Only 64-bit targets support TImode libcalls. Disable the TImode shift libcalls
for ARM. This fixes rdar://
6908807.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72269
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 22 May 2009 16:47:11 +0000 (16:47 +0000)]
Teach IndVarSimplify's FixUsesBeforeDefs to handle InvokeInsts by
assuming that the use of the value is in a block dominated by the
"normal" destination. LangRef.html and other documentation sources
don't explicitly guarantee this, but it seems to be assumed in
other places in LLVM at least.
This fixes an assertion failure on the included testcase, which
is derived from the Ada testsuite.
FixUsesBeforeDefs is a temporary measure which I'm looking to
replace with a more capable solution.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72266
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjiv Gupta [Fri, 22 May 2009 13:58:45 +0000 (13:58 +0000)]
Emit debug information for globals (which include automatic variables as well because on PIC16 they are emitted as globals by the frontend).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72262
91177308-0d34-0410-b5e6-
96231b3b80d8