Duncan Sands [Fri, 27 Mar 2009 09:41:56 +0000 (09:41 +0000)]
A Release-Asserts build makes it sound like assertions
are turned on, while in fact they are turned off. Name
this Release-NoAsserts instead.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67831
91177308-0d34-0410-b5e6-
96231b3b80d8
John Mosby [Fri, 27 Mar 2009 06:09:40 +0000 (06:09 +0000)]
Shrink wrapping in PEI: initial release. Finishing development, enable with --shrink-wrap.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67828
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Fri, 27 Mar 2009 02:45:14 +0000 (02:45 +0000)]
Add a __builtin___memset_chk test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67815
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Fri, 27 Mar 2009 01:13:37 +0000 (01:13 +0000)]
One more place to skip debug info.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67811
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Thu, 26 Mar 2009 23:48:52 +0000 (23:48 +0000)]
While hoisting an instruction, update alias info set tracker.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67798
91177308-0d34-0410-b5e6-
96231b3b80d8
Mike Stump [Thu, 26 Mar 2009 23:43:14 +0000 (23:43 +0000)]
Update to account for driver renaming.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67791
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 26 Mar 2009 23:06:32 +0000 (23:06 +0000)]
-no-implicit-float means explicit fp operations are legal.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67784
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 26 Mar 2009 23:03:32 +0000 (23:03 +0000)]
Add -march=x86.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67783
91177308-0d34-0410-b5e6-
96231b3b80d8
Mikhail Glushenkov [Thu, 26 Mar 2009 21:23:48 +0000 (21:23 +0000)]
Fix misc. small issues with debug visualization.
Detailed bug report:
http://llvm.org/bugs/show_bug.cgi?id=3873
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67768
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 26 Mar 2009 19:09:01 +0000 (19:09 +0000)]
tADDhirr is a thumb instruction. Do not allow this code to be reached in non-thumb mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67765
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Thu, 26 Mar 2009 18:53:38 +0000 (18:53 +0000)]
Don't assign a new stack slot if the pre-alloc splitter already assigned one.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67764
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 26 Mar 2009 16:17:51 +0000 (16:17 +0000)]
fix a few spelling errors and typos
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67758
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 26 Mar 2009 14:45:34 +0000 (14:45 +0000)]
Use 'bool' for FoundRC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67750
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 26 Mar 2009 06:17:54 +0000 (06:17 +0000)]
Add -f to RUN line.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67744
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 26 Mar 2009 06:14:09 +0000 (06:14 +0000)]
Pull transform from target-dependent code into target-independent code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67742
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 26 Mar 2009 05:51:18 +0000 (05:51 +0000)]
no need for eh info
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67740
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 26 Mar 2009 05:29:34 +0000 (05:29 +0000)]
fix warning in -asserts mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67739
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 26 Mar 2009 05:28:26 +0000 (05:28 +0000)]
fix some warnings in release-asserts mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67738
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 26 Mar 2009 05:28:14 +0000 (05:28 +0000)]
fix an apparently real bug exposed by a warning in -asserts mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67737
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 26 Mar 2009 05:25:59 +0000 (05:25 +0000)]
fix warning in -asserts build.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67736
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 26 Mar 2009 01:52:47 +0000 (01:52 +0000)]
Add testcase for r67728.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67729
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 26 Mar 2009 01:47:50 +0000 (01:47 +0000)]
Match this pattern so that we can generate simpler code:
%a = ...
%b = and i32 %a, 2
%c = srl i32 %b, 1
%d = br i32 %c,
into
%a = ...
%b = and %a, 2
%c = X86ISD::CMP %b, 0
%d = X86ISD::BRCOND %c ...
This applies only when the AND constant value has one bit set and the SRL
constant is equal to the log2 of the AND constant. The back-end is smart enough
to convert the result into a TEST/JMP sequence.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67728
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 26 Mar 2009 01:46:56 +0000 (01:46 +0000)]
Doxygen-ify comments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67727
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Thu, 26 Mar 2009 01:15:07 +0000 (01:15 +0000)]
Skip debug info one more place. (This one gets
called from llc, not opt, but it's an IR level
optimization nevertheless.)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67724
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Wed, 25 Mar 2009 23:57:48 +0000 (23:57 +0000)]
Before deleting a basic block, give other loop passes a chance cleanup analysis values, related to the instructions in the basic block.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67719
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Wed, 25 Mar 2009 23:28:33 +0000 (23:28 +0000)]
Modify getRegisterValueType() to allow for a register being in mutliple
register classes. Before, MVT::Other would be returned anytime a reg was
in multiple register classes. Now, MVT::Other is only returned if the types
for those register classes differ.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67714
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 25 Mar 2009 20:30:19 +0000 (20:30 +0000)]
Add a test case for PR3779: when to promote the function return value.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67702
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 25 Mar 2009 20:20:11 +0000 (20:20 +0000)]
Revert 67132. This is breaking some objective-c apps.
Also fixes SDISel so it *does not* force promote return value if the function is not marked signext / zeroext.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67701
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 25 Mar 2009 17:45:18 +0000 (17:45 +0000)]
One more test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67694
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Wed, 25 Mar 2009 17:36:08 +0000 (17:36 +0000)]
When optimizing with debug info, don't keep the
stoppoint nodes around until Legalize; doing this
imposed an ordering on a sequence of loads that
came from different lines, interfering with scheduling.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67692
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 25 Mar 2009 16:49:31 +0000 (16:49 +0000)]
Add __builtin___memcpy_chk tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67691
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Wed, 25 Mar 2009 16:43:59 +0000 (16:43 +0000)]
These tests pass on linux.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67689
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 25 Mar 2009 16:29:33 +0000 (16:29 +0000)]
Add a __builtin_object_size test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67688
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 25 Mar 2009 06:36:36 +0000 (06:36 +0000)]
Fix a bug in our autoupgrade support: in an argument list to a function
call, we should treat "i64 zext" as the start of a constant expr, but
"i64 0 zext" as an argument with an obsolete attribute on it (this form
is already tested by test/Assembler/2007-07-30-AutoUpgradeZextSext.ll).
Make the autoupgrade logic more discerning to avoid treating "i64 zext"
as an old-style attribute, causing us to reject a valid constant expr.
This fixes PR3876.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67682
91177308-0d34-0410-b5e6-
96231b3b80d8
Gabor Greif [Wed, 25 Mar 2009 06:32:59 +0000 (06:32 +0000)]
do not rely on callee being operand 0
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67681
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Wed, 25 Mar 2009 03:52:06 +0000 (03:52 +0000)]
Do not ignore DW_TAG_class_type!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67675
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 25 Mar 2009 01:48:21 +0000 (01:48 +0000)]
CodeGen still defaults to non-verbose asm, but llc now overrides it and default to verbose.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67669
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 25 Mar 2009 01:47:28 +0000 (01:47 +0000)]
CodeGen still defaults to non-verbose asm, but llc now overrides it and default to verbose.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67668
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 25 Mar 2009 01:08:42 +0000 (01:08 +0000)]
Don't print global names twice with -asm-verbose.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67667
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Wed, 25 Mar 2009 00:52:11 +0000 (00:52 +0000)]
Revert r67655 and r67656, as they are breaking the build. I'm
not going to persue this further at this time.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67666
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 25 Mar 2009 00:28:58 +0000 (00:28 +0000)]
Fix PR3874 by restoring a condition I removed, but making it more
precise than it used to be.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67662
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Wed, 25 Mar 2009 00:28:40 +0000 (00:28 +0000)]
Do not ignore DW_TAG_class_type!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67661
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Wed, 25 Mar 2009 00:09:05 +0000 (00:09 +0000)]
LLVM's master Makefile.common is in LLVM_SRC_ROOT, not LLVM_OBJ_ROOT.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67658
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 24 Mar 2009 23:48:25 +0000 (23:48 +0000)]
oops, I intended to remove this, not comment it out. Thanks Duncan!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67657
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 24 Mar 2009 23:47:11 +0000 (23:47 +0000)]
Regenerate configure.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67656
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 24 Mar 2009 23:46:25 +0000 (23:46 +0000)]
Fix paths; AutoRegen.sh changes its current working directory to be the
autoconf directory, but these paths need to be relative to the main
source directory.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67655
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 24 Mar 2009 23:45:13 +0000 (23:45 +0000)]
Update for autoconf 2.6x;
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67654
91177308-0d34-0410-b5e6-
96231b3b80d8
Misha Brukman [Tue, 24 Mar 2009 21:36:09 +0000 (21:36 +0000)]
Converted a1.ll to unittests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67652
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Tue, 24 Mar 2009 21:16:39 +0000 (21:16 +0000)]
Mention explicitly that the Ada front-end will not bootstrap.
The configure line had --disable-bootstrap, but it was not
clear that this is essential.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67651
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Tue, 24 Mar 2009 20:33:17 +0000 (20:33 +0000)]
Fix PR3845: Avoid stale MachineInstruction pointer reference.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67649
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 24 Mar 2009 20:21:37 +0000 (20:21 +0000)]
Fix bash-isms.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67647
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 24 Mar 2009 20:20:28 +0000 (20:20 +0000)]
Set the svn:mime-type to text/x-sh, so that diffs work.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67646
91177308-0d34-0410-b5e6-
96231b3b80d8
Gabor Greif [Tue, 24 Mar 2009 19:28:39 +0000 (19:28 +0000)]
simplify logic and get rid of the assumption that operand 0 is the callee
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67642
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 24 Mar 2009 18:35:40 +0000 (18:35 +0000)]
canonicalize inttoptr and ptrtoint instructions which cast pointers
to/from integer types that are not intptr_t to convert to intptr_t
then do an integer conversion to the dest type. This exposes the
cast to the optimizer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67638
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Tue, 24 Mar 2009 18:16:17 +0000 (18:16 +0000)]
fix one more fp80 case (used only by Interpreter)
and streamline code here a bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67636
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 24 Mar 2009 18:15:30 +0000 (18:15 +0000)]
two changes:
1. Make instcombine always canonicalize trunc x to i1 into an icmp(x&1). This
exposes the AND to other instcombine xforms and is more of what the code
generator expects.
2. Rewrite the remaining trunc pattern match to use 'match', which
simplifies it a lot.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67635
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Tue, 24 Mar 2009 17:35:45 +0000 (17:35 +0000)]
Change JIT for different layout of fp80.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67629
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 24 Mar 2009 16:38:27 +0000 (16:38 +0000)]
Add a testcase for the scheduling heuristic introduced in r67586.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67622
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 24 Mar 2009 15:27:37 +0000 (15:27 +0000)]
more tidying: name the components of PhysReg in the case when
the target constraint specifies a specific physreg.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67618
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 24 Mar 2009 15:25:07 +0000 (15:25 +0000)]
Tidy a bit more.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67617
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 24 Mar 2009 15:22:11 +0000 (15:22 +0000)]
simplify this code a bit now that "allocation to a vreg class" can never
fail.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67616
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 24 Mar 2009 01:04:34 +0000 (01:04 +0000)]
I was convinced that it's ok to allow a second i8 return value
to be returned in DL. LLVM's multiple-return-value support is
not ABI-conforming; front-ends that wish to have code emitted
that conforms to an ABI are currently expected to make
arrangements for this on their own rather than assuming that
multiple-return-values will automatically do the right thing.
This commit doesn't fundamentally change this situation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67588
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 24 Mar 2009 00:50:07 +0000 (00:50 +0000)]
Minor compile-time optimization; don't bother checking
canClobberPhysRegDefs if the successor node doesn't
clobber any physical registers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67587
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 24 Mar 2009 00:49:12 +0000 (00:49 +0000)]
Add a pre-pass to the burr-list scheduler which makes adjustments to
help out the register pressure reduction heuristics in the case of
nodes with multiple uses. Currently this uses very conservative
heuristics, so it doesn't have a broad impact, but in cases where it
does help it can make a big difference.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67586
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Tue, 24 Mar 2009 00:17:40 +0000 (00:17 +0000)]
Do not emit comments unless -asm-verbose.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67580
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Mon, 23 Mar 2009 23:39:20 +0000 (23:39 +0000)]
Use a SmallPtrSet instead of std::set.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67578
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 23 Mar 2009 22:57:19 +0000 (22:57 +0000)]
Fix a bug in spill weight computation. If the alias is a super-register, and the super-register is in the register class we are trying to allocate. Then add the weight to all sub-registers of the super-register even if they are not aliases.
e.g. allocating for GR32, bh is not used, updating bl spill weight.
bl should get the same spill weight otherwise it will be choosen
as a spill candidate since spilling bh doesn't make ebx available.
This fix PR2866.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67574
91177308-0d34-0410-b5e6-
96231b3b80d8
Ted Kremenek [Mon, 23 Mar 2009 21:54:33 +0000 (21:54 +0000)]
Add sanity check in Clang TableGen backend to check if 'Component' is a string.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67565
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Mon, 23 Mar 2009 21:16:53 +0000 (21:16 +0000)]
Fix internal representation of fp80 to be the
same as a normal i80 {low64, high16} rather
than its own {high64, low16}. A depressing number
of places know about this; I think I got them all.
Bitcode readers and writers convert back to the old
form to avoid breaking compatibility.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67562
91177308-0d34-0410-b5e6-
96231b3b80d8
John Mosby [Mon, 23 Mar 2009 21:00:45 +0000 (21:00 +0000)]
README.txt: test commit w/blank line appended
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67560
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 20:20:43 +0000 (20:20 +0000)]
When unfolding a load during scheduling, the new operator node has
a data dependency on the load node, so it really needs a
data-dependence edge to the load node, even if the load previously
existed.
And add a few comments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67554
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 23 Mar 2009 18:27:36 +0000 (18:27 +0000)]
Update test for pr3864.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67545
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 23 Mar 2009 18:24:37 +0000 (18:24 +0000)]
Fix PR3391 and PR3864. Reg allocator infinite looping.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67544
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 17:39:36 +0000 (17:39 +0000)]
Don't set SUnit::hasPhysRegDefs to true unless the defs are
actually have uses, which reflects the way it's used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67540
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 16:23:01 +0000 (16:23 +0000)]
Fix canClobberPhysRegDefs to check all SDNodes grouped together
in an SUnit, instead of just the first one. This fix is needed
by some upcoming scheduler changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67531
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 16:10:52 +0000 (16:10 +0000)]
Add a new bit to SUnit to record whether a node has implicit physreg
defs, regardless of whether they are actually used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67528
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 15:57:19 +0000 (15:57 +0000)]
Now that errs() is properly non-buffered, there's no need to
explicitly flush it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67526
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 15:54:02 +0000 (15:54 +0000)]
Clarify a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67525
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 15:50:52 +0000 (15:50 +0000)]
LoopVR is not CFGOnly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67524
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 15:49:37 +0000 (15:49 +0000)]
Enhance LiveValues to work on PHI operands.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67523
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 15:48:29 +0000 (15:48 +0000)]
Make getOperandNumForIncomingValue and getOperandNumForIncomingBlock
static member functions, and add getIncomingValueNumForOperand
and getIncomingBlockNumForOperand, which are the respective
inverses.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67522
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 15:40:10 +0000 (15:40 +0000)]
Correct some comments. Operand numbers start at 0.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67518
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 23 Mar 2009 08:01:15 +0000 (08:01 +0000)]
Model inline asm constraint which ties an input to an output register as machine operand TIED_TO constraint. This eliminated the need to pre-allocate registers for these. This also allows register allocator can eliminate the unneeded copies.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67512
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 23 Mar 2009 07:19:58 +0000 (07:19 +0000)]
Do not fold away subreg_to_reg if the source register has a sub-register index. That means the source register is taking a sub-register of a larger register. e.g. On x86
%RAX<def> = ...
%RAX<def> = SUBREG_TO_REG 0, %EAX:3<kill>, 3
The first def is defining RAX, not EAX so the top bits were not zero-extended.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67511
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 23 Mar 2009 06:46:20 +0000 (06:46 +0000)]
fix a bug Alexei Svitkine pointed out.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67510
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 23 Mar 2009 05:55:36 +0000 (05:55 +0000)]
When we restore signal handlers, restore them back to what they
were when we came around, not to their default handler. This
should fix PR3848
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67509
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 23 Mar 2009 05:42:29 +0000 (05:42 +0000)]
factorize signal registration, part of PR3848.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67508
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 05:02:44 +0000 (05:02 +0000)]
Fix a grammaro in a comment that Bill noticed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67507
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 23 Mar 2009 04:53:34 +0000 (04:53 +0000)]
tt'ify.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67505
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 23 Mar 2009 04:52:53 +0000 (04:52 +0000)]
VC++ 6.0 is not future work :)
Do not recommend llvm::OStream anymore. Use raw_ostream or MemoryBuffer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67504
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Mon, 23 Mar 2009 04:28:24 +0000 (04:28 +0000)]
Add comments explaining why there's only one register for
i8 return values.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67502
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 23 Mar 2009 03:45:56 +0000 (03:45 +0000)]
Missed this as part of r67451.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67500
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 22 Mar 2009 19:22:53 +0000 (19:22 +0000)]
Fix PR3860 by correcting a predicate.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67473
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Sun, 22 Mar 2009 11:33:16 +0000 (11:33 +0000)]
Add some explanations of how apint loads and stores
work.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67471
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Sun, 22 Mar 2009 08:56:15 +0000 (08:56 +0000)]
Really should pass -dylib to the linker...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67469
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Sun, 22 Mar 2009 08:28:45 +0000 (08:28 +0000)]
A dylib should be built as a dylib and not a bundle.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67468
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 22 Mar 2009 00:18:18 +0000 (00:18 +0000)]
add method to access a template argument.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67458
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Sat, 21 Mar 2009 21:27:31 +0000 (21:27 +0000)]
Factorize out a concept - no functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67454
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Sat, 21 Mar 2009 18:12:24 +0000 (18:12 +0000)]
Re-commit r67334 and r67349 with fix.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67451
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sat, 21 Mar 2009 06:55:54 +0000 (06:55 +0000)]
add a fastpath to ConstantExpr::getBitCast to handle the case when an obviously
unneeded bitcast is requested. This is common for frontends who just unconditionally
cast even if the target is often the right type already. THis prevents going into
getFoldedCast which switches on the opcode and does a bunch of other stuff before
doing the same opzn.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@67435
91177308-0d34-0410-b5e6-
96231b3b80d8