Jim Grosbach [Fri, 26 Mar 2010 18:41:09 +0000 (18:41 +0000)]
vldm/vstm can only do up to 16 double-word registers at a time.
Radar
7797856
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99630
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Fri, 26 Mar 2010 18:32:20 +0000 (18:32 +0000)]
Add N3RegFrm to represent "NEON 3 vector register format" instructions.
Examples are VABA (Vector Absolute Difference and Accumulate), VABAL (Vector
Absolute Difference and Accumulate Long), and VABD (Vector Absolute Difference).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99628
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Fri, 26 Mar 2010 16:26:03 +0000 (16:26 +0000)]
Do not sibcall if stack needs to be dynamically aligned.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99620
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Fri, 26 Mar 2010 02:13:13 +0000 (02:13 +0000)]
Allow trivial sibcall of vararg callee when no arguments are being passed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99598
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Fri, 26 Mar 2010 02:12:24 +0000 (02:12 +0000)]
LiveVariables should clear kill / dead markers first. This allows us to remove a hack in the scheduler.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99597
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Fri, 26 Mar 2010 01:07:59 +0000 (01:07 +0000)]
Add N2RegVShLFrm and N2RegVShRFrm formats so that the disassembler can easily
dispatch to the appropriate routines to handle the different interpretations of
the shift amount encoded in the imm6 field. The Vd, Vm fields are interpreted
the same between the two, though.
See, for example, A8.6.367 VQSHL, VQSHLU (immediate) for N2RegVShLFrm format and
A8.6.368 VQSHRN, VQSHRUN for N2RegVShRFrm format.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99590
91177308-0d34-0410-b5e6-
96231b3b80d8
Jeffrey Yasskin [Fri, 26 Mar 2010 00:59:12 +0000 (00:59 +0000)]
Avoid leaking argv and env arrays from lli.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99589
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 26 Mar 2010 00:33:27 +0000 (00:33 +0000)]
Ignore debug intrinsics in yet more places.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99580
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Fri, 26 Mar 2010 00:07:25 +0000 (00:07 +0000)]
Try trivial remat before the coalescer gives up on a vr / physreg coalescing for fear of tying up a physical register.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99575
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Fri, 26 Mar 2010 00:02:44 +0000 (00:02 +0000)]
Handle DEBUG_VALUE in this pass.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99573
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 25 Mar 2010 23:47:34 +0000 (23:47 +0000)]
switch the flag for using NEON for SP floating point to a subtarget 'feature'.
Re-commit. This time complete with testsuite updates.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99570
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 25 Mar 2010 23:34:05 +0000 (23:34 +0000)]
need to fix 'make check' tests first. revert for a moment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99569
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 25 Mar 2010 23:32:19 +0000 (23:32 +0000)]
switch the flag for using NEON for SP floating point to a subtarget 'feature'
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99568
91177308-0d34-0410-b5e6-
96231b3b80d8
Gabor Greif [Thu, 25 Mar 2010 23:25:28 +0000 (23:25 +0000)]
rename pred_const_iterator to const_pred_iterator for consistency's sake
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99567
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Thu, 25 Mar 2010 23:11:56 +0000 (23:11 +0000)]
Removed instruction class NI from ARMInstrFormats.td.
It doesn't seem to be used anywhere.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99566
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 25 Mar 2010 23:11:16 +0000 (23:11 +0000)]
switch the use-vml[as] instructions flag to a subtarget 'feature'
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99565
91177308-0d34-0410-b5e6-
96231b3b80d8
Gabor Greif [Thu, 25 Mar 2010 23:06:16 +0000 (23:06 +0000)]
rename use_const_iterator to const_use_iterator for consistency's sake
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99564
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 22:49:09 +0000 (22:49 +0000)]
llvm-mc: Add a -mc-relax-all option, which relaxes every fixup. We always need
exactly two passes in that case, and don't ever need to recompute any layout,
so this is a nice baseline for relaxation performance.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99563
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Thu, 25 Mar 2010 21:49:12 +0000 (21:49 +0000)]
Add NVDupLnFrm and change NVDupLane class to use that format.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99557
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 25 Mar 2010 20:48:50 +0000 (20:48 +0000)]
ARM cortex-a8 doesn't do vmla/vmls well. disable them by default for that cpu
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99549
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Thu, 25 Mar 2010 20:39:04 +0000 (20:39 +0000)]
Add NVCVTFrm (NEON Convert with fractional bits immediate) and modify N2VImm to
expect a Format arg. N2VCvtD/N2VCvtQ are modified to use the NVCVTFrm format.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99548
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 25 Mar 2010 20:01:07 +0000 (20:01 +0000)]
Add nounwind.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99546
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 25 Mar 2010 19:46:11 +0000 (19:46 +0000)]
Code clean up.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99544
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 19:35:56 +0000 (19:35 +0000)]
MC: Stop restarting layout on every relaxation.
- Still O(N^2), just a faster form, and now its the MCAsmLayout's fault.
On the .s I am tuning against (combine.s from 403.gcc):
--
ddunbar@lordcrumb:MC$ diff stats-before.txt stats-after.txt
5,10c5,10
< 1728 assembler - Number of assembler layout and relaxation steps
< 7707 assembler - Number of emitted assembler fragments
< 120588 assembler - Number of emitted object file bytes
<
2233448 assembler - Number of evaluated fixups
< 1727 assembler - Number of relaxed instructions
<
6723845 mcexpr - Number of MCExpr evaluations
---
> 3 assembler - Number of assembler layout and relaxation steps
> 7707 assembler - Number of emitted assembler fragments
> 120588 assembler - Number of emitted object file bytes
> 14796 assembler - Number of evaluated fixups
> 1727 assembler - Number of relaxed instructions
> 67889 mcexpr - Number of MCExpr evaluations
--
Feel free to LOL at the -before numbers, if you like.
I am a little surprised we make more than 2 relaxation passes. It's pretty
trivial for us to do relaxation out-of-order if that would give a speedup.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99543
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 19:35:53 +0000 (19:35 +0000)]
Fix -Asserts warning, again.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99542
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 25 Mar 2010 18:52:04 +0000 (18:52 +0000)]
Tag SSE2 integer instructions as SSEPackedInt.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99540
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 25 Mar 2010 18:52:01 +0000 (18:52 +0000)]
Teach TableGen to understand X.Y notation in the TSFlagsFields strings.
Remove much horribleness from X86InstrFormats as a result. Similar
simplifications are probably possible for other targets.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99539
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 25 Mar 2010 18:49:10 +0000 (18:49 +0000)]
fix a valgrind error on copy-constructor-synthesis.cpp, which is caused when
the custom insertion hook deletes the instruction, then we try to set dead
flags on it. Neither the code that I added nor the code that was there
before was safe.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99538
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 25 Mar 2010 18:37:23 +0000 (18:37 +0000)]
Remove an unused option.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99537
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 18:16:42 +0000 (18:16 +0000)]
MC: Simplify main section layout process by moving alignment into LayoutSection.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99529
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 18:16:38 +0000 (18:16 +0000)]
MC: Sink Section address assignment into LayoutSection.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99528
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 25 Mar 2010 17:25:00 +0000 (17:25 +0000)]
Add a late SSEDomainFix pass that twiddles SSE instructions to avoid domain crossings.
On Nehalem and newer CPUs there is a 2 cycle latency penalty on using a register
in a different domain than where it was defined. Some instructions have
equvivalents for different domains, like por/orps/orpd.
The SSEDomainFix pass tries to minimize the number of domain crossings by
changing between equvivalent opcodes where possible.
This is a work in progress, in particular the pass doesn't do anything yet. SSE
instructions are tagged with their execution domain in TableGen using the last
two bits of TSFlags. Note that not all instructions are tagged correctly. Life
just isn't that simple.
The SSE execution domain issue is very similar to the ARM NEON/VFP pipeline
issue handled by NEONMoveFixPass. This pass may become target independent to
handle both.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99524
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Thu, 25 Mar 2010 17:01:27 +0000 (17:01 +0000)]
Added a new instruction class NVDupLane to be inherited by VDUPLND and VDUPLNQ,
instead of the current N2V. Format of NVDupLane instances are set to NEONFrm
currently.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99518
91177308-0d34-0410-b5e6-
96231b3b80d8
Bob Wilson [Thu, 25 Mar 2010 16:36:14 +0000 (16:36 +0000)]
Reapply Kevin's change 94440, now that Chris has fixed the limitation on
opcode values fitting in one byte (svn r99494).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99514
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 16:09:18 +0000 (16:09 +0000)]
Sketch a few Clang release notes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99512
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Thu, 25 Mar 2010 15:09:44 +0000 (15:09 +0000)]
Add comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99507
91177308-0d34-0410-b5e6-
96231b3b80d8
John Criswell [Thu, 25 Mar 2010 13:59:09 +0000 (13:59 +0000)]
Fixed spurious warning problem noticed by Daniel Dunbar. The configure script
now configures prerequisite projects individually but also ignores them in the
big project switch statement to avoid the incorrect warning.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99506
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 08:08:54 +0000 (08:08 +0000)]
MC/Mach-O: Switch to MCSectionData::getOrdinal.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99504
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 25 Mar 2010 07:16:57 +0000 (07:16 +0000)]
Scheduler assumes SDDbgValue nodes are in source order. That's true currently. But add an assertion to verify it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99501
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 07:10:11 +0000 (07:10 +0000)]
MC: Explicity track section and fragment ordinals.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99500
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 07:10:05 +0000 (07:10 +0000)]
Fix -Asserts warning.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99499
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 07:10:01 +0000 (07:10 +0000)]
lit: Add LitTestCase and lit.load_test_suite, for adapting lit based suites for
use with Python's unittest.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99498
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 25 Mar 2010 06:33:05 +0000 (06:33 +0000)]
Change tblgen to emit FOOISD opcode names as two
bytes instead of one byte. This is important because
we're running up to too many opcodes to fit in a byte
and it is aggrevated by FIRST_TARGET_MEMORY_OPCODE
making the numbering sparse. This just bites the
bullet and bloats out the table. In practice, this
increases the size of the x86 isel table from 74.5K
to 76K. I think we'll cope :)
This fixes rdar://
7791648
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99494
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Thu, 25 Mar 2010 06:26:14 +0000 (06:26 +0000)]
Fix typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99493
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 25 Mar 2010 06:23:34 +0000 (06:23 +0000)]
Fix evil TableGen bug in template parameters with defaults.
If a TableGen class has an initializer expression containing an X.Y subexpression,
AND X depends on template parameters,
AND those template parameters have defaults,
AND some parameters with defaults are beyond position 1,
THEN parts of the initializer expression are evaluated prematurely with the default values when the first explicit template parameter is substituted, before the remaining explicit template parameters have been substituted.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99492
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Thu, 25 Mar 2010 06:04:47 +0000 (06:04 +0000)]
Include isFunctionLocal while calculating folding node set profile for a MDNode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99490
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 25 Mar 2010 06:02:53 +0000 (06:02 +0000)]
Remove a fixme that doesn't make sense any more.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99489
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 25 Mar 2010 05:58:19 +0000 (05:58 +0000)]
fix PR6642, GVN forwarding from memset to load of the base of the memset.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99488
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 25 Mar 2010 05:50:26 +0000 (05:50 +0000)]
Make sure SDDbgValue.Invalid is initialized to false by all the constructors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99487
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 25 Mar 2010 05:44:01 +0000 (05:44 +0000)]
eliminate a bunch more parallels now that scheduling
handles dead implicit results more aggressively. More
to come, I think this is now just a data entry problem.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99486
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 25 Mar 2010 05:40:48 +0000 (05:40 +0000)]
Make the NDEBUG assertion stronger and more clear what is
happening.
Enhance scheduling to set the DEAD flag on implicit defs
more aggressively. Before, we'd set an implicit def operand
to dead if it were present in the SDNode corresponding to
the machineinstr but had no use. Now we do it in this case
AND if the implicit def does not exist in the SDNode at all.
This exposes a couple of problems: one is the FIXME, which
causes a live intervals crash on CodeGen/X86/sibcall.ll.
The second is that it makes machinecse and licm more
aggressive (which is a good thing) but also exposes a case
where licm hoists a set0 and then it doesn't get resunk.
Talking to codegen folks about both these issues, but I need
this patch in in the meantime.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99485
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Thu, 25 Mar 2010 05:36:13 +0000 (05:36 +0000)]
Include isFunctionLocal while calculating folding node set provide for a MDNode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99484
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Thu, 25 Mar 2010 04:49:10 +0000 (04:49 +0000)]
Reapply r99451 with a fix to move the NoInline check to the cost functions
instead of InlineFunction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99483
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 25 Mar 2010 04:41:16 +0000 (04:41 +0000)]
reapply 99444/99445, which I speculatively reverted in
r99453.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99482
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 02:00:07 +0000 (02:00 +0000)]
MC: Route access to SectionData offset and file size through MCAsmLayout.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99474
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 02:00:02 +0000 (02:00 +0000)]
MC: Route access to Fragment offset and effective size through MCAsmLayout.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99473
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Thu, 25 Mar 2010 01:46:07 +0000 (01:46 +0000)]
Make sure this runs in 64-bit only, 32-bit won't produce the correct stores.
Fariborz please review and make sure this is what you meant.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99472
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 25 Mar 2010 01:38:16 +0000 (01:38 +0000)]
Change how dbg_value sdnodes are converted into machine instructions. Their placement should be determined by the relative order of incoming llvm instructions. The scheduler will now use the SDNode ordering information to determine where to insert them. A dbg_value instruction is inserted after the instruction with the last highest source order and before the instruction with the next highest source order. It will optimize the placement by inserting right after the instruction that produces the value if they have consecutive order numbers.
Here is a theoretical example that illustrates why the placement is important.
tmp1 =
store tmp1 -> x
...
tmp2 = add ...
...
call
...
store tmp2 -> x
Now mem2reg comes along:
tmp1 =
dbg_value (tmp1 -> x)
...
tmp2 = add ...
...
call
...
dbg_value (tmp2 -> x)
When the debugger examine the value of x after the add instruction but before the call, it should have the value of tmp1.
Furthermore, for dbg_value's that reference constants, they should not be emitted at the beginning of the block (since they do not have "producers").
This patch also cleans up how SDISel manages DbgValue nodes. It allow a SDNode to be referenced by multiple SDDbgValue nodes. When a SDNode is deleted, it uses the information to find the SDDbgValues and invalidate them. They are not deleted until the corresponding SelectionDAG is destroyed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99469
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 01:03:24 +0000 (01:03 +0000)]
MC: Eliminate MC{Fragment,{Section,Symbol}Data}::getAddress.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99467
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Thu, 25 Mar 2010 01:03:17 +0000 (01:03 +0000)]
MC: Fix refacto in MCExpr evaluation, I mistakenly replaced a fragment address with a symbol address.
- This fixes the integrated-as nightly test regressions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99466
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 25 Mar 2010 01:01:37 +0000 (01:01 +0000)]
Avoid being influenced by dbg_value instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99465
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Thu, 25 Mar 2010 00:59:51 +0000 (00:59 +0000)]
Fix unused parameter warning.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99463
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Thu, 25 Mar 2010 00:10:31 +0000 (00:10 +0000)]
Disable folding loads into tail call in 32-bit PIC mode. It can introduce illegal code like this:
addl $12, %esp
popl %esi
popl %edi
popl %ebx
popl %ebp
jmpl *__Block_deallocator-L1$pb(%esi) # TAILCALL
The problem is the global base register is assigned GR32 register class. TCRETURNmi needs the registers making up the address mode to have the GR32_TC register class.
The *proper* fix is for X86DAGToDAGISel::getGlobalBaseReg() to return a copy from the global base register of the machine function rather than returning the register itself. But that has the potential of causing it to be coalesced to a more restrictive register class: GR32_TC. It can introduce additional copies and spills. For something as important the PIC base, it's not worth it especially since this is not an issue on 64-bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99455
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Thu, 25 Mar 2010 00:03:04 +0000 (00:03 +0000)]
Docuemntation corrections from John Myers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99454
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 24 Mar 2010 23:41:19 +0000 (23:41 +0000)]
revert 99444/99445. This doesn't cause the failure of
2006-07-19-stwbrx-crash.ll for me, but it's the only likely
patch in the blame list of several bots. Lets see if this
fixes it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99453
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Wed, 24 Mar 2010 23:35:21 +0000 (23:35 +0000)]
Temporarily revert this, it's causing an issue with an internal project.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99451
91177308-0d34-0410-b5e6-
96231b3b80d8
Bob Wilson [Wed, 24 Mar 2010 23:26:29 +0000 (23:26 +0000)]
Speculatively revert this to see if it fixes buildbot failures.
--- Reverse-merging r99440 into '.':
U test/MC/AsmParser/X86/x86_32-bit_cat.s
U test/MC/AsmParser/X86/x86_32-encoding.s
U include/llvm/IntrinsicsX86.td
U include/llvm/CodeGen/SelectionDAGNodes.h
U lib/Target/X86/X86InstrSSE.td
U lib/Target/X86/X86ISelLowering.h
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99450
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 24 Mar 2010 23:07:47 +0000 (23:07 +0000)]
add a convenient TargetInstrDesc::getNumImplicitUses/Defs method.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99446
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 24 Mar 2010 22:47:12 +0000 (22:47 +0000)]
remove dead argument.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99445
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 24 Mar 2010 22:45:47 +0000 (22:45 +0000)]
split EmitNode in half to reduce indentation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99444
91177308-0d34-0410-b5e6-
96231b3b80d8
Kevin Enderby [Wed, 24 Mar 2010 22:33:33 +0000 (22:33 +0000)]
Added the Advanced Encryption Standard (AES) Instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99440
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Wed, 24 Mar 2010 22:31:46 +0000 (22:31 +0000)]
Make the use of the vmla and vmls VFP instructions controllable via cmd line.
Preliminary testing shows significant performance wins by not using these
instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99436
91177308-0d34-0410-b5e6-
96231b3b80d8
Kevin Enderby [Wed, 24 Mar 2010 22:28:42 +0000 (22:28 +0000)]
Fixed the SS42AI template for the SSE 4.2 instructions with TA prefix so it does
not get an "Unknown immediate size" assert failure when used. All instructions
of this form have an 8-bit immediate. Also added a test case of an example
instruction that is of this form.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99435
91177308-0d34-0410-b5e6-
96231b3b80d8
Nate Begeman [Wed, 24 Mar 2010 22:19:06 +0000 (22:19 +0000)]
Per chris's request, add some comments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99434
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Wed, 24 Mar 2010 21:30:35 +0000 (21:30 +0000)]
Use SP filename directly instead of SP's context's filename.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99429
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Wed, 24 Mar 2010 21:25:07 +0000 (21:25 +0000)]
Trivial formating change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99428
91177308-0d34-0410-b5e6-
96231b3b80d8
Nate Begeman [Wed, 24 Mar 2010 20:49:50 +0000 (20:49 +0000)]
BUILD_VECTOR was missing out on some prime opportunities to use SSE 4.1 inserts.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99423
91177308-0d34-0410-b5e6-
96231b3b80d8
Bob Wilson [Wed, 24 Mar 2010 20:25:25 +0000 (20:25 +0000)]
Revert Edwin's change that is breaking MultiSource/Applications/ClamAV/clamscan.
--- Reverse-merging r99400 into '.':
D test/CodeGen/Generic/2010-03-24-liveintervalleak.ll
U lib/CodeGen/LiveIntervalAnalysis.cpp
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99419
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 24 Mar 2010 20:19:04 +0000 (20:19 +0000)]
Move OptChkCall off LibCallOptimization into StrCpyOpt.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99418
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Wed, 24 Mar 2010 19:56:17 +0000 (19:56 +0000)]
Trim #includes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99416
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Wed, 24 Mar 2010 19:47:14 +0000 (19:47 +0000)]
Reverted r99326 which added NVdVmVCVTFrm, and later renamed to NVCVTFrm.
NVCVTFrm will later be used to describe "vcvt with fractional bits".
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99415
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Wed, 24 Mar 2010 19:38:02 +0000 (19:38 +0000)]
Fix minor style issues.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99414
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Wed, 24 Mar 2010 19:00:02 +0000 (19:00 +0000)]
It's not necessary to call raw_ostream::close explicitly on automatic
raw_ostream variables immediately before they go out of scope.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99413
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Wed, 24 Mar 2010 18:48:00 +0000 (18:48 +0000)]
Do not rely on getCompileUnit() to find source file information for a subprogram.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99410
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Wed, 24 Mar 2010 18:46:34 +0000 (18:46 +0000)]
Reverted r99376. The disassembler will deal with the 2-reg format of these two
N3VX instructions using special case code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99409
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Wed, 24 Mar 2010 16:15:14 +0000 (16:15 +0000)]
tweak the arm if conversion heuristic
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99402
91177308-0d34-0410-b5e6-
96231b3b80d8
Torok Edwin [Wed, 24 Mar 2010 13:50:36 +0000 (13:50 +0000)]
Fix memory leak in liveintervals: the destructor for VNInfos must be called,
otherwise the SmallVector it contains doesn't free its memory.
In most cases LiveIntervalAnalysis could get away by not calling the destructor,
because VNInfos are bumpptr-allocated, and smallvectors usually don't grow.
However when the SmallVector does grow it always leaks.
This is the valgrind shown leak from the original testcase:
==8206== 18,304 bytes in 151 blocks are definitely lost in loss record 164 of 164
==8206== at 0x4A079C7: operator new(unsigned long) (vg_replace_malloc.c:220)
==8206== by 0x4DB7A7E: llvm::SmallVectorBase::grow_pod(unsigned long, unsigned long) (in /home/edwin/clam/git/builds/defaul
t/libclamav/.libs/libclamav.so.6.1.0)
==8206== by 0x4F90382: llvm::VNInfo::addKill(llvm::SlotIndex) (in /home/edwin/clam/git/builds/default/libclamav/.libs/libcl
amav.so.6.1.0)
==8206== by 0x5126B5C: llvm::LiveIntervals::handleVirtualRegisterDef(llvm::MachineBasicBlock*, llvm::ilist_iterator<llvm::M
achineInstr>, llvm::SlotIndex, llvm::MachineOperand&, unsigned int, llvm::LiveInterval&) (in /home/edwin/clam/git/builds/defau
lt/libclamav/.libs/libclamav.so.6.1.0)
==8206== by 0x512725E: llvm::LiveIntervals::handleRegisterDef(llvm::MachineBasicBlock*, llvm::ilist_iterator<llvm::MachineI
nstr>, llvm::SlotIndex, llvm::MachineOperand&, unsigned int) (in /home/edwin/clam/git/builds/default/libclamav/.libs/libclamav
.so.6.1.0)
==8206== by 0x51278A8: llvm::LiveIntervals::computeIntervals() (in /home/edwin/clam/git/builds/default/libclamav/.libs/libc
lamav.so.6.1.0)
==8206== by 0x5127CB4: llvm::LiveIntervals::runOnMachineFunction(llvm::MachineFunction&) (in /home/edwin/clam/git/builds/de
fault/libclamav/.libs/libclamav.so.6.1.0)
==8206== by 0x4DAE935: llvm::FPPassManager::runOnFunction(llvm::Function&) (in /home/edwin/clam/git/builds/default/libclama
v/.libs/libclamav.so.6.1.0)
==8206== by 0x4DAEB10: llvm::FunctionPassManagerImpl::run(llvm::Function&) (in /home/edwin/clam/git/builds/default/libclama
v/.libs/libclamav.so.6.1.0)
==8206== by 0x4DAED3D: llvm::FunctionPassManager::run(llvm::Function&) (in /home/edwin/clam/git/builds/default/libclamav/.l
ibs/libclamav.so.6.1.0)
==8206== by 0x4D8BE8E: llvm::JIT::runJITOnFunctionUnlocked(llvm::Function*, llvm::MutexGuard const&) (in /home/edwin/clam/git/builds/default/libclamav/.libs/libclamav.so.6.1.0)
==8206== by 0x4D8CA72: llvm::JIT::getPointerToFunction(llvm::Function*) (in /home/edwin/clam/git/builds/default/libclamav/.libs/libclamav.so.6.1.0)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99400
91177308-0d34-0410-b5e6-
96231b3b80d8
Gabor Greif [Wed, 24 Mar 2010 13:21:49 +0000 (13:21 +0000)]
Finally land the InvokeInst operand reordering.
I have audited all getOperandNo calls now, fixing
hidden assumptions. CallSite related uglyness will
be eliminated successively.
Note this patch has a long and griveous history,
for all the back-and-forths have a look at
CallSite.h's log.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99399
91177308-0d34-0410-b5e6-
96231b3b80d8
Gabor Greif [Wed, 24 Mar 2010 11:58:07 +0000 (11:58 +0000)]
tighten a type and remove trailing whitespace, no functional changes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99398
91177308-0d34-0410-b5e6-
96231b3b80d8
Gabor Greif [Wed, 24 Mar 2010 10:29:52 +0000 (10:29 +0000)]
increase const goodness and remove pointless getUser() calls
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99395
91177308-0d34-0410-b5e6-
96231b3b80d8
Gabor Greif [Wed, 24 Mar 2010 10:12:54 +0000 (10:12 +0000)]
cache result of UI.getOperandNo() instead of calling it twice, it is cheaper this way
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99394
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Wed, 24 Mar 2010 09:05:14 +0000 (09:05 +0000)]
Fix comment typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99392
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Wed, 24 Mar 2010 03:43:40 +0000 (03:43 +0000)]
MC: Direct all {fragment,section,symbol} address access through the MCAsmLayout object.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99380
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 24 Mar 2010 01:50:28 +0000 (01:50 +0000)]
dbg_value may end a block.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99378
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Wed, 24 Mar 2010 01:29:25 +0000 (01:29 +0000)]
Mark VMOVDneon and VMOVQ as having the N2RegFrm form to help the disassembler.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99376
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 24 Mar 2010 01:02:12 +0000 (01:02 +0000)]
Switch INC8r to defining its pattern in terms of X86inc_flag
and defining the add pattern with Pat<>, eliminating a use of
parallel.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99375
91177308-0d34-0410-b5e6-
96231b3b80d8
Johnny Chen [Wed, 24 Mar 2010 00:57:50 +0000 (00:57 +0000)]
Renamed NVdVmImmFrm and NVdVmVCVTFrm to the more proper N2RegFrm and NVCVTFrm,
respectively, and add some more comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99373
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Wed, 24 Mar 2010 00:53:38 +0000 (00:53 +0000)]
Remove the ConvertActions table and associated code, which is unused.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99372
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 24 Mar 2010 00:53:27 +0000 (00:53 +0000)]
Add a method to get a StringMapEntry from a pointer to the string
data it contains (similar to GetStringMapEntryFromValue).
Patch by Greg Clayton!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99371
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 24 Mar 2010 00:49:29 +0000 (00:49 +0000)]
switch SDTBinaryArithWithFlags to be a multiple-result node as well.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99370
91177308-0d34-0410-b5e6-
96231b3b80d8