OSDN Git Service
Bob Wilson [Tue, 21 Dec 2010 06:43:19 +0000 (06:43 +0000)]
Add ARM-specific DAG combining to cast i64 vector element load/stores to f64.
Type legalization splits up i64 values into pairs of i32 values, which leads
to poor quality code when inserting or extracting i64 vector elements.
If the vector element is loaded or stored, it can be treated as an f64 value
and loaded or stored directly from a VPR register. Use the pre-legalization
DAG combiner to cast those vector elements to f64 types so that the type
legalizer won't mess them up. Radar
8755338.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122319
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Tue, 21 Dec 2010 04:22:09 +0000 (04:22 +0000)]
Layout one section until no relaxations are done and then move to the next
section.
This helps because in practice sections form a dag with debug sections pointing
to text sections. Finishing up the text sections first makes the debug section
relaxation trivial.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122314
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Tue, 21 Dec 2010 03:50:43 +0000 (03:50 +0000)]
Arm and thumb call instructions are also in different orders.
Fixes rdar://
8782223
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122313
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Tue, 21 Dec 2010 03:31:05 +0000 (03:31 +0000)]
Grammar corrections and some formatting changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122312
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 21 Dec 2010 02:38:05 +0000 (02:38 +0000)]
rename MVT::Flag to MVT::Glue. "Flag" is a terrible name for
something that just glues two nodes together, even if it is
sometimes used for flags.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122310
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Tue, 21 Dec 2010 02:12:07 +0000 (02:12 +0000)]
If we're not using reg+reg offset we're using reg+imm, set the opcode
to be the one we want to use. bugpoint reduced testcase is a little large,
I'll see if I can simplify it down more.
Fixes part of rdar://
8782207
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122307
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 21 Dec 2010 02:07:03 +0000 (02:07 +0000)]
improve "cannot yet select" errors a trivial amount: now
they are just as useless, but at least a bit more gramatical
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122305
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Tue, 21 Dec 2010 01:57:15 +0000 (01:57 +0000)]
Fix a copy-pasto. When the tBR_JTr instruction was converted to using the
tPseudoInst class, its size was changed from "special" to "2 bytes". This is
incorrect because the jump table will no longer be taken into account when
calculating branch offsets.
<rdar://problem/
8782216>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122303
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Tue, 21 Dec 2010 01:54:40 +0000 (01:54 +0000)]
Comment cleanups.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122302
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Tue, 21 Dec 2010 01:50:21 +0000 (01:50 +0000)]
Add EdgeBundles to SplitKit.
Edge bundles is an annotation on the CFG that turns it into a bipartite directed
graph where each basic block is connected to an outgoing and an ingoing bundle.
These bundles are useful for identifying regions of the CFG for live range
splitting.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122301
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Tue, 21 Dec 2010 00:48:17 +0000 (00:48 +0000)]
Use IntEqClasses to compute connected components of live intervals.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122296
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Tue, 21 Dec 2010 00:04:46 +0000 (00:04 +0000)]
Add ADT/IntEqClasses.h as a light-weight implementation of EquivalenceClasses.h.
This implementation already exists as ConnectedVNInfoEqClasses in
LiveInterval.cpp, and it seems to be generally useful to have a light-weight way
of forming equivalence classes of small integers.
IntEqClasses doesn't allow enumeration of the elements in a class.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122293
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 20 Dec 2010 23:53:19 +0000 (23:53 +0000)]
Speculatively revert the use of DenseMap in LazyValueInfo, which may be causing Linux self-host failures.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122291
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 20 Dec 2010 23:23:18 +0000 (23:23 +0000)]
Attempt to appease the DragonEgg buildbots.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122288
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 20 Dec 2010 22:28:03 +0000 (22:28 +0000)]
Revert r122114 (CallbackVH observing use-list changes) because it caused severe slowdowns on the Linux self-host configuration.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122279
91177308-0d34-0410-b5e6-
96231b3b80d8
Nate Begeman [Mon, 20 Dec 2010 22:04:24 +0000 (22:04 +0000)]
Implement feedback from Bruno on making pblendvb an x86-specific ISD node in addition to being an intrinsic, and convert
lowering to use it. Hopefully the pattern fragment is doing the right thing with XMM0, looks correct in testing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122277
91177308-0d34-0410-b5e6-
96231b3b80d8
Wesley Peck [Mon, 20 Dec 2010 21:54:50 +0000 (21:54 +0000)]
Update the target feature matrix to reflect some new features in the MBlaze backend.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122276
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 20 Dec 2010 21:30:54 +0000 (21:30 +0000)]
Convert one of LVI's primary maps to a DenseMap, now that we know are more assured of iterator stability.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122273
91177308-0d34-0410-b5e6-
96231b3b80d8
Wesley Peck [Mon, 20 Dec 2010 21:18:04 +0000 (21:18 +0000)]
Teach the MBlaze disassembler to disassemble special purpose registers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122269
91177308-0d34-0410-b5e6-
96231b3b80d8
Roman Divacky [Mon, 20 Dec 2010 21:14:39 +0000 (21:14 +0000)]
Set the value of absolute symbols.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122268
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Mon, 20 Dec 2010 21:14:18 +0000 (21:14 +0000)]
This is still used and issuing an annoying warning. Don't deprecate something in
the library unless *all* uses have been converted over to the new form.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122267
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Mon, 20 Dec 2010 21:07:42 +0000 (21:07 +0000)]
Oops, forgot to add the pass itself!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122265
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Mon, 20 Dec 2010 20:54:37 +0000 (20:54 +0000)]
Add a new convenience pass for testing InstructionSimplify. Previously
it could only be tested indirectly, via instcombine, gvn or some other
pass that makes use of InstructionSimplify, which means that testcases
had to be carefully contrived to dance around any other transformations
that that pass did.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122264
91177308-0d34-0410-b5e6-
96231b3b80d8
Roman Divacky [Mon, 20 Dec 2010 20:49:43 +0000 (20:49 +0000)]
Print all 64bits for st_value and st_size. Adjust tests accordingly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122263
91177308-0d34-0410-b5e6-
96231b3b80d8
Wesley Peck [Mon, 20 Dec 2010 20:43:24 +0000 (20:43 +0000)]
Teach the MBlaze asm parser how to parse special purpose register names.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122261
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Mon, 20 Dec 2010 20:10:50 +0000 (20:10 +0000)]
Cosmetic changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122259
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 20 Dec 2010 20:00:31 +0000 (20:00 +0000)]
Add a check missing from my last commit and avoid a potential overflow situation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122258
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 20 Dec 2010 19:33:41 +0000 (19:33 +0000)]
More LVI cleanups, including trying to simplify the process of maintaining the OverDefinedCache.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122256
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 20 Dec 2010 18:18:16 +0000 (18:18 +0000)]
Reuse the reference into the LVI cache throughout the solver subsystem. This is much easier to
verify as being safe thanks its recent de-recursivization.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122254
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 20 Dec 2010 16:21:59 +0000 (16:21 +0000)]
Reduce indentation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122249
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 20 Dec 2010 16:18:51 +0000 (16:18 +0000)]
Teach InstCombine to merge (icmp ult (X + CA), C1) | (icmp eq X, C2) into (icmp ult (X + CA), C1 + 1) if C2 + CA == C1.
InstCombine creates these so now we compile x == 23 || x == 24 || x == 25 to
%x.off = add i32 %x, -23
%1 = icmp ult i32 %x.off, 3
instead of
%x.off = add i32 %x, -23
%1 = icmp ult i32 %x.off, 2
%cmp3 = icmp eq i32 %x, 25
%ret2 = or i1 %1, %cmp3
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122248
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Mon, 20 Dec 2010 15:45:51 +0000 (15:45 +0000)]
Add header...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122247
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Mon, 20 Dec 2010 15:07:39 +0000 (15:07 +0000)]
X86/MC/Mach-O: Split out createX86MachObjectWriter().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122246
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Mon, 20 Dec 2010 14:47:04 +0000 (14:47 +0000)]
Have SimplifyBinOp dispatch Xor, Add and Sub to the corresponding methods
(they had just been forgotten before). Adding Xor causes "main" in the
existing testcase 2010-11-01-lshr-mask.ll to be hugely more simplified.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122245
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael J. Spencer [Mon, 20 Dec 2010 13:30:37 +0000 (13:30 +0000)]
Fix spelling.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122244
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael J. Spencer [Mon, 20 Dec 2010 13:30:28 +0000 (13:30 +0000)]
Support/PathV2: Add missing has_relative_path impl.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122243
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Mon, 20 Dec 2010 13:10:23 +0000 (13:10 +0000)]
There is no need for isAssociative to take the type as an argument anymore.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122242
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Mon, 20 Dec 2010 10:57:01 +0000 (10:57 +0000)]
Discourage people from using isWeakForLinker when they should be using
mayBeOverridden.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122241
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Mon, 20 Dec 2010 09:47:13 +0000 (09:47 +0000)]
New cmake option LLVM_APPEND_VC_REV for controlling when the VC
revision id is appended to the LLVM version string. Defaults to
OFF.
Until now the VC revision id was always appended to the revision
string whenever cmake was invoked (either explicitly or implicitly
because a cmake source file changed). This was causing massive
recompilations because config.h are reconfigured with the new contents
of PACKAGE_VERSION.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122240
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Mon, 20 Dec 2010 09:47:08 +0000 (09:47 +0000)]
Update cmake library dependencies.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122239
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 08:36:38 +0000 (08:36 +0000)]
fix PR8807 by making transformConstExprCastCall aware of byval arguments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122238
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 08:25:06 +0000 (08:25 +0000)]
various cleanups for transformConstExprCastCall
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122237
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 08:10:40 +0000 (08:10 +0000)]
when eliding a byval copy due to inlining a readonly function, we have
to make sure that the reused alloca has sufficient alignment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122236
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 07:57:41 +0000 (07:57 +0000)]
pull byval processing out to its own helper function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122235
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 07:45:28 +0000 (07:45 +0000)]
fix PR8769, a miscompilation by inliner when inlining a function with a byval
argument. The generated alloca has to have at least the alignment of the
byval, if not, the client may be making assumptions that the new alloca won't
satisfy.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122234
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 07:39:57 +0000 (07:39 +0000)]
merge two tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122233
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 07:38:24 +0000 (07:38 +0000)]
filecheckize
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122232
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Mon, 20 Dec 2010 04:19:48 +0000 (04:19 +0000)]
MachineVerifier should count landing pad successors as basic blocks rather than
out-edges. Fixes PR8824.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122228
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Mon, 20 Dec 2010 03:15:20 +0000 (03:15 +0000)]
Teach MachineVerifier that early clobber defs begin at USE slots and other defs
begin at DEF slots. Fixes the second half of PR8813.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122225
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Mon, 20 Dec 2010 02:59:51 +0000 (02:59 +0000)]
Add a missing check from r122218.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122224
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 02:11:23 +0000 (02:11 +0000)]
temporarily disable this: PR8823.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122222
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 02:05:39 +0000 (02:05 +0000)]
implement type legalization promotion support for SMULO and UMULO, giving
ARM (and other 32-bit-only) targets support for i8 and i16 overflow
multiplies. The generated code isn't great, but this at least fixes
CodeGen/Generic/overflow.ll when running on ARM hosts.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122221
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 01:37:09 +0000 (01:37 +0000)]
now that addc/adde are gone, "ADDC" in the X86 backend uses EFLAGS results,
the same as setcc. Optimize ADDC(0,0,FLAGS) -> SET_CARRY(FLAGS). This is
a step towards finishing off PR5443. In the testcase in that bug we now get:
movq %rdi, %rax
addq %rsi, %rax
sbbq %rcx, %rcx
testb $1, %cl
setne %dl
ret
instead of:
movq %rdi, %rax
addq %rsi, %rax
movl $0, %ecx
adcq $0, %rcx
testq %rcx, %rcx
setne %dl
ret
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122219
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Mon, 20 Dec 2010 01:22:37 +0000 (01:22 +0000)]
Don't assume that an instruction ending a register's live range always reads
the register; it may be a dead def instead. Fixes PR8820.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122218
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 01:16:03 +0000 (01:16 +0000)]
We lower setb to sbb with the hope that the and will go away, when it
doesn't, match it back to setb.
On a 64-bit version of the testcase before we'd get:
movq %rdi, %rax
addq %rsi, %rax
sbbb %dl, %dl
andb $1, %dl
ret
now we get:
movq %rdi, %rax
addq %rsi, %rax
setb %dl
ret
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122217
91177308-0d34-0410-b5e6-
96231b3b80d8
Mon P Wang [Mon, 20 Dec 2010 01:06:23 +0000 (01:06 +0000)]
Test case for r122215 when InstCombine optimizes memset
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122216
91177308-0d34-0410-b5e6-
96231b3b80d8
Mon P Wang [Mon, 20 Dec 2010 01:05:30 +0000 (01:05 +0000)]
Avoid dropping the address space when InstCombine optimizes memset
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122215
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 01:03:27 +0000 (01:03 +0000)]
use for loop over types.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122214
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 00:59:46 +0000 (00:59 +0000)]
Change the X86 backend to stop using the evil ADDC/ADDE/SUBC/SUBE nodes (which
their carry depenedencies with MVT::Flag operands) and use clean and beautiful
EFLAGS dependences instead.
We do this by changing the modelling of SBB/ADC to have EFLAGS input and outputs
(which is what requires the previous scheduler change) and change X86 ISelLowering
to custom lower ADDC and friends down to X86ISD::ADD/ADC/SUB/SBB nodes.
With the previous series of changes, this causes no changes in the testsuite, woo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122213
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 00:56:59 +0000 (00:56 +0000)]
update comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122212
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 00:55:43 +0000 (00:55 +0000)]
Fix a bug in the scheduler's handling of "unspillable" vregs.
Imagine we see:
EFLAGS = inst1
EFLAGS = inst2 FLAGS
gpr = inst3 EFLAGS
Previously, we would refuse to schedule inst2 because it clobbers
the EFLAGS of the predecessor. However, it also uses the EFLAGS
of the predecessor, so it is safe to emit. SDep edges ensure that
the right order happens already anyway.
This fixes 2 testsuite crashes with the X86 patch I'm going to
commit next.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122211
91177308-0d34-0410-b5e6-
96231b3b80d8
Mon P Wang [Mon, 20 Dec 2010 00:54:26 +0000 (00:54 +0000)]
Add comment for testcase for 122206
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122210
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 00:51:56 +0000 (00:51 +0000)]
the result of CheckForLiveRegDef is dead, remove it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122209
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 20 Dec 2010 00:50:16 +0000 (00:50 +0000)]
reduce indentation, no functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122208
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Mon, 20 Dec 2010 00:08:10 +0000 (00:08 +0000)]
Ignore debug values when performing MachineVerifier liveness checks. Fixes
PR8822.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122207
91177308-0d34-0410-b5e6-
96231b3b80d8
Mon P Wang [Sun, 19 Dec 2010 23:55:53 +0000 (23:55 +0000)]
Prevents PerformShuffleCombine from creating a node with an illegal type after legalize types
has run, e.g., prevent creating an i64 node from a v2i64 when i64 is not a legal type.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122206
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 19 Dec 2010 23:50:53 +0000 (23:50 +0000)]
Early clobber operands are allowed to be defined at use indices. This fixes one
half of PR8813.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122205
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 23:24:04 +0000 (23:24 +0000)]
fix an oversight caught by Frits!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122204
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Sun, 19 Dec 2010 22:41:43 +0000 (22:41 +0000)]
Update the Target Feature Matrix for Mips
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122203
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 19 Dec 2010 22:12:45 +0000 (22:12 +0000)]
Fix PR8815 by checking for an explicit clobber def tied to a use operand in
ConnectedVNInfoEqClasses::Classify().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122202
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 22:08:31 +0000 (22:08 +0000)]
improve the setcc -> setcc_carry optimization to happen more
consistently by moving it out of lowering into dag combine.
Add some missing patterns for matching away extended versions of setcc_c.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122201
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 19 Dec 2010 21:37:23 +0000 (21:37 +0000)]
Fix PR8811 by teaching MachineVerifier about optional defs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122199
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 19 Dec 2010 21:32:29 +0000 (21:32 +0000)]
StrongPHIElimination will never run before TwoAddressInstructionPass.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122197
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 21:23:48 +0000 (21:23 +0000)]
simplify some code to just reuse a setcc if we can instead of
going through the CSE maps to get it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122196
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Sun, 19 Dec 2010 20:57:14 +0000 (20:57 +0000)]
Use the new way of silencing this warning.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122195
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Sun, 19 Dec 2010 20:43:38 +0000 (20:43 +0000)]
Add missing standard headers. Patch by Joerg Sonnenberger!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122193
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Sun, 19 Dec 2010 20:42:43 +0000 (20:42 +0000)]
Add missing std:: prefixes to some calls. C++ doesn't require that <cfoo>
headers provide symbols outside namespace std and the LLVM coding standards
state that we should prefix all of them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122192
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 20:38:28 +0000 (20:38 +0000)]
teach MaskedValueIsZero how to analyze ADDE. This is
enough to teach it that ADDE(0,0) is known 0 except the
low bit, for example.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122191
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 20:24:28 +0000 (20:24 +0000)]
tidy up
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122190
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 20:19:20 +0000 (20:19 +0000)]
now that generic vector types aren't selected onto MMX operations,
we don't need -disable-mmx anymore.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122189
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 20:12:58 +0000 (20:12 +0000)]
now that generic vector types aren't selected onto MMX registers, these
tests don't need -disable-mmx.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122188
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 20:07:10 +0000 (20:07 +0000)]
reduce copy/paste programming with the power of for loops.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122187
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 20:03:11 +0000 (20:03 +0000)]
X86 supports i8/i16 overflow ops (except i8 multiplies), we should
generate them.
Now we compile:
define zeroext i8 @X(i8 signext %a, i8 signext %b) nounwind ssp {
entry:
%0 = tail call %0 @llvm.sadd.with.overflow.i8(i8 %a, i8 %b)
%cmp = extractvalue %0 %0, 1
br i1 %cmp, label %if.then, label %if.end
into:
_X: ## @X
## BB#0: ## %entry
subl $12, %esp
movb 16(%esp), %al
addb 20(%esp), %al
jo LBB0_2
Before we were generating:
_X: ## @X
## BB#0: ## %entry
pushl %ebp
movl %esp, %ebp
subl $8, %esp
movb 12(%ebp), %al
testb %al, %al
setge %cl
movb 8(%ebp), %dl
testb %dl, %dl
setge %ah
cmpb %cl, %ah
sete %cl
addb %al, %dl
testb %dl, %dl
setge %al
cmpb %al, %ah
setne %al
andb %cl, %al
testb %al, %al
jne LBB0_2
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122186
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 20:01:13 +0000 (20:01 +0000)]
add a general coverage test for overflow intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122185
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 19:43:52 +0000 (19:43 +0000)]
move a transformation to a more logical place, simplifying it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122183
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 19:37:52 +0000 (19:37 +0000)]
recognize an unsigned add with overflow idiom into uadd.
This resolves a README entry and technically resolves PR4916,
but we still get poor code for the testcase in that PR because
GVN isn't CSE'ing uadd with add, filed as PR8817.
Previously we got:
_test7: ## @test7
addq %rsi, %rdi
cmpq %rdi, %rsi
movl $42, %eax
cmovaq %rsi, %rax
ret
Now we get:
_test7: ## @test7
addq %rsi, %rdi
movl $42, %eax
cmovbq %rsi, %rax
ret
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122182
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 19:35:32 +0000 (19:35 +0000)]
optimize uadd(x, cst) into a comparison when the normal
result is dead. This is required for my next patch to not
regress the testsuite.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122181
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 19:16:22 +0000 (19:16 +0000)]
add a version of IRBuilder::SetInsertPoint that takes an instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122180
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 18:38:44 +0000 (18:38 +0000)]
use IC.ReplaceInstUsesWith instead of a raw RAUW so that uses of
the old thing end up on the instcombine worklist. Not doing this
can cause an extra top-level iteration of instcombine, burning
compile time.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122179
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 18:35:09 +0000 (18:35 +0000)]
generalize the sadd creation code to not require that the
sadd formed is half the size of the original type. We can
now compile this into a sadd.i8:
unsigned char X(char a, char b) {
int res = a+b;
if ((unsigned )(res+128) > 255U)
abort();
return res;
}
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122178
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 18:22:06 +0000 (18:22 +0000)]
fix another miscompile in the llvm.sadd formation logic: it wasn't
checking to see if the high bits of the original add result were dead.
Inserting a smaller add and zexting back to that size is not good enough.
This is likely to be the fix for 8816.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122177
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 19 Dec 2010 18:03:27 +0000 (18:03 +0000)]
Remove some checks for StrongPHIElim. These checks make it impossible to use an
alternative register allocator that does not require LiveIntervals by specifying
it on the command-line for a target that has StrongPHIElimination enabled by
default.
These checks are pretty meaningless anyways, since StrongPHIElimination and
PHIElimination are never used at the same time.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122176
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 17:59:02 +0000 (17:59 +0000)]
fix a bug (possibly 8816) in the sadd forming xform: it isn't
profitable (or safe) to promote code when the add-with-constant
has other uses.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122175
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 17:52:50 +0000 (17:52 +0000)]
rework the code added in r122072 to pull it out to its own
helper function, clean up comments, and reduce indentation.
No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122174
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 05:57:25 +0000 (05:57 +0000)]
Enhance LICM to promote alias sets whose pointers themselves are stored,
which doesn't affect the memory address being promoted.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122172
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 05:51:54 +0000 (05:51 +0000)]
fix PR8602, a bug in an assertion: a volatile store *of* a pointer
does not make the alias set for that pointer volatile, just stores
*to* the pointer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122171
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 04:58:57 +0000 (04:58 +0000)]
fix PR8642: if a critical edge has a PHI value that can trap,
isel is *required* to split the edge. PHI values get evaluated
on the edge, not in their predecessor block.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122170
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 04:49:11 +0000 (04:49 +0000)]
remove dead header
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122169
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 19 Dec 2010 04:23:03 +0000 (04:23 +0000)]
revert r122164, I'm going to go with a different approach.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122168
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Sun, 19 Dec 2010 04:18:56 +0000 (04:18 +0000)]
Fixed version of 122160 (the previous one would fold undefined symbols).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122167
91177308-0d34-0410-b5e6-
96231b3b80d8