David Greene [Fri, 29 Jul 2011 19:07:16 +0000 (19:07 +0000)]
[AVX] Make ListInits Unique
Ensure ListInits are unique and only created once. This will be
important for AVX as lists will be used extensively to pass generic
patterns, prefix information and other things to lower-level
pattern-generation classes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136493
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:07:15 +0000 (19:07 +0000)]
[AVX] Make CodeInit Unique
Use a StringMap to ensure CodeInits are unique and created only
once.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136492
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:07:14 +0000 (19:07 +0000)]
[AVX] Make StringInit Unique
Use a StringMap to ensure the StringInits are unique. This is
especially important for AVX where we will have many smallish
strings representing instruction prefixes, suffixes and the like.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136491
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:07:12 +0000 (19:07 +0000)]
[AVX] Make IntInit Unique
Use a DenseMap to make sure only one IntInit of any value exists.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136490
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:07:11 +0000 (19:07 +0000)]
[AVX] Make BitsInit Unique
Make BitsInit a FoldingSetNode so we can unique it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136489
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:07:10 +0000 (19:07 +0000)]
[AVX] Unique BitInit
Keep only two copies of BitInit: one for true and one for false.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136488
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:07:09 +0000 (19:07 +0000)]
[AVX] Unique UnsetInit
Keep only one UnsetInit around.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136487
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:07:07 +0000 (19:07 +0000)]
[AVX] Create Inits Via Factory Method
Replace uses of new *Init with *Init::get. This hides the allocation
implementation so that we can unique Inits in various ways.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136486
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:07:05 +0000 (19:07 +0000)]
[AVX] Constify Inits
Make references to Inits const everywhere. This is the final step
before making them unique.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136485
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:07:02 +0000 (19:07 +0000)]
[AVX] Remove non-const Iterators
Remove all non-const iterators from Init classes. This is another
step toward constifying Inits and ultimately turning them into
FoldingSetNodes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136484
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:07:00 +0000 (19:07 +0000)]
[AVX] Remove Mutating Members from Inits
Get rid of all Init members that modify internal state. This is in
preparation for making references to Inits const.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136483
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:06:59 +0000 (19:06 +0000)]
Add ListInit::getValues
Add a getValues ListInit method to return the sequence of values in
the list.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136482
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Fri, 29 Jul 2011 19:06:58 +0000 (19:06 +0000)]
Add a std::string Wrapper for TableGen
Create a std::string wrapper for use as a DenseMap key. DenseMap is
not safe in generate with strings, so this wrapper indicates that only
strings guaranteed not to have certain values should be used in the
DenseMap.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136481
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Fri, 29 Jul 2011 19:00:35 +0000 (19:00 +0000)]
Clean up debug info after reassociation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136480
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Fri, 29 Jul 2011 18:47:24 +0000 (18:47 +0000)]
ARM assembly parsing and encoding for RFE instruction.
Fill in the missing fixed bits and the register operand bits of the instruction
encoding. Refactor the definition to make the mode explicit, which is
consistent with how loads and stores are normally represented and makes
parsing much easier. Add parsing aliases for pseudo-instruction variants.
Update the disassembler for the new representations. Add tests for parsing and
encoding.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136479
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Fri, 29 Jul 2011 18:28:52 +0000 (18:28 +0000)]
Third time's the charm for implementing tied operand decoding properly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136478
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Fri, 29 Jul 2011 18:26:59 +0000 (18:26 +0000)]
Don't look at $PWD in GetCurrentDirectory.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136477
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Fri, 29 Jul 2011 17:51:39 +0000 (17:51 +0000)]
ARM SRS and RFE instructions are not code-gen only.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136475
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Fri, 29 Jul 2011 17:42:17 +0000 (17:42 +0000)]
ARM range checking for mode on CPS instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136473
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Fri, 29 Jul 2011 17:39:27 +0000 (17:39 +0000)]
ARM update tests for CPS instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136472
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Fri, 29 Jul 2011 17:36:04 +0000 (17:36 +0000)]
Update FIXME.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136470
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Fri, 29 Jul 2011 17:33:29 +0000 (17:33 +0000)]
Tweak comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136468
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Fri, 29 Jul 2011 17:32:03 +0000 (17:32 +0000)]
Fix a case where, when trying to track tied operands, we'd accidentally overwrite our mapping.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136467
91177308-0d34-0410-b5e6-
96231b3b80d8
Douglas Gregor [Fri, 29 Jul 2011 15:41:39 +0000 (15:41 +0000)]
Switch the CMake edis build over to add_llvm_library_dependencies
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136463
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Fri, 29 Jul 2011 07:50:02 +0000 (07:50 +0000)]
Avoid undefined behaviour if somehow NUM_GRAPHS equals 2^32 (or
whatever the size of unsigned is), though this can't actually
occur for any integer value of NUM_NODES.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136460
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Fri, 29 Jul 2011 04:42:39 +0000 (04:42 +0000)]
Teach Path::GetCurrentDirectory to use $PWD, to support users who like to do
screwy things by setting PWD != getcwd(). For example, some developers I know
will use this to control the value in gcc's DW_AT_comp_dir value in debug
output. With this patch, that trick will now work on clang too.
The only other effect of this change is that the static analysis will now
respect $PWD when reporting the directory of the files in its HTML output. I
think that's fine.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136459
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Fri, 29 Jul 2011 03:49:23 +0000 (03:49 +0000)]
Fix a lot of typos, improve (but not necessarily fix) grammaros and reflow some
lines. No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136458
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Fri, 29 Jul 2011 03:05:32 +0000 (03:05 +0000)]
Misc optimizer+codegen work for 'cmpxchg' and 'atomicrmw'. They appear to be
working on x86 (at least for trivial testcases); other architectures will
need more work so that they actually emit the appropriate instructions for
orderings stricter than 'monotonic'. (As far as I can tell, the ARM, PPC,
Mips, and Alpha backends need such changes.)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136457
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakub Staszak [Fri, 29 Jul 2011 02:36:53 +0000 (02:36 +0000)]
Remove incEdgeWeight and decEdgeWeight. Set edge weight directly to avoid
rounding errors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136456
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Fri, 29 Jul 2011 02:12:59 +0000 (02:12 +0000)]
Fix the MCDisassembler dependencies. These were just absurdly wrong.
First off, only depend on the actual MC-ized disassemblers in the
targets, not all of the libraries those in turn depend on.
Second off, only depend on those MC-ized disassemblers for targets we're
building.
This should fix builds of fewer than all targets.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136455
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Fri, 29 Jul 2011 02:05:28 +0000 (02:05 +0000)]
Fix two tests that I crashed in the previous commits. The mask elts
on the second half must be reindexed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136454
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Fri, 29 Jul 2011 01:31:15 +0000 (01:31 +0000)]
Match VPERMIL masks more strictly and update the target specific mask
generation to always catch the weird cases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136453
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Fri, 29 Jul 2011 01:31:11 +0000 (01:31 +0000)]
Add DecodeShuffle shuffle support for VPERMIPD variantes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136452
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Fri, 29 Jul 2011 01:31:07 +0000 (01:31 +0000)]
Add v8i32 and v4i64 vpermil patterns
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136451
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Fri, 29 Jul 2011 01:31:04 +0000 (01:31 +0000)]
Fix a bug while generating target specific VPERMIL masks: skip
undef mask elements. This fixes PR10529.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136450
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Fri, 29 Jul 2011 01:31:02 +0000 (01:31 +0000)]
Enable usage of SSE4 extracts and inserts in their 128-bit AVX forms.
Also tidy up code a bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136449
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Fri, 29 Jul 2011 01:30:59 +0000 (01:30 +0000)]
Cleanup PALIGNR handling and remove the old palign pattern fragment.
Also make PALIGNR masks to don't match 256-bits, which isn't supported
It's also a step to solve PR10489
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136448
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 29 Jul 2011 01:15:29 +0000 (01:15 +0000)]
Use the pointer type size.
With this, we can now compile a simple EH program.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136446
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 29 Jul 2011 01:11:33 +0000 (01:11 +0000)]
And now something that compiles...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136445
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 29 Jul 2011 01:11:14 +0000 (01:11 +0000)]
Make sure to sext or trunc the result from the register.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136444
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 29 Jul 2011 00:27:35 +0000 (00:27 +0000)]
Transfer implicit operands in NEONMoveFixPass.
Later passes /are/ using this information when running the register
scavenger.
This fixes the second problem in PR10520.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136440
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 29 Jul 2011 00:27:32 +0000 (00:27 +0000)]
Add -verify-arm-pseudo-expand.
This hidden llc option runs the machine code verifier after expanding
ARM pseudo-instructions, but before if-conversion.
The machine code verifier is much better at pointing out liveness errors
that can trip up the register scavenger.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136439
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Fri, 29 Jul 2011 00:18:19 +0000 (00:18 +0000)]
Make sure to correctly clear the exact/nuw/nsw flags off of shifts when they are combined together. <rdar://problem/
9859829>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136435
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Fri, 29 Jul 2011 00:14:25 +0000 (00:14 +0000)]
Rewrite the CMake build to use explicit dependencies between libraries,
specified in the same file that the library itself is created. This is
more idiomatic for CMake builds, and also allows us to correctly specify
dependencies that are missed due to bugs in the GenLibDeps perl script,
or change from compiler to compiler. On Linux, this returns CMake to
a place where it can relably rebuild several targets of LLVM.
I have tried not to change the dependencies from the ones in the current
auto-generated file. The only places I've really diverged are in places
where I was seeing link failures, and added a dependency. The goal of
this patch is not to start changing the dependencies, merely to move
them into the correct location, and an explicit form that we can control
and change when necessary.
This also removes a serialization point in the build because we don't
have to scan all the libraries before we begin building various tools.
We no longer have a step of the build that regenerates a file inside the
source tree. A few other associated cleanups fall out of this.
This isn't really finished yet though. After talking to dgregor he urged
switching to a single CMake macro to construct libraries with both
sources and dependencies in the arguments. Migrating from the two macros
to that style will be a follow-up patch.
Also, llvm-config is still generated with GenLibDeps.pl, which means it
still has slightly buggy dependencies. The internal CMake
'llvm-config-like' macro uses the correct explicitly specified
dependencies however. A future patch will switch llvm-config generation
(when using CMake) to be based on these deps as well.
This may well break Windows. I'm getting a machine set up now to dig
into any failures there. If anyone can chime in with problems they see
or ideas of how to solve them for Windows, much appreciated.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136433
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakub Staszak [Fri, 29 Jul 2011 00:05:35 +0000 (00:05 +0000)]
Remove extra semicolon.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136432
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Thu, 28 Jul 2011 23:56:20 +0000 (23:56 +0000)]
Enhance the fixed-length decoder emitter to support tied operands.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136431
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 23:44:58 +0000 (23:44 +0000)]
Visit the landingpad instruction.
This generates the correct SDNodes for the landingpad instruction. It makes an
assumption that the result of the landingpad instruction has at least two
values. And that the first value is a pointer to the exception object and the
second value is the "selector."
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136430
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 23:42:57 +0000 (23:42 +0000)]
Add the AddLandingPadInfo function.
AddLandingPadInfo takes a landingpad instruction and grabs all of the
information from it that it needs for EH table generation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136429
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakub Staszak [Thu, 28 Jul 2011 23:42:08 +0000 (23:42 +0000)]
Change LBH_TAKEN_WEIGHT to 124 (from 128). Right now, sum of
LBH_TAKEN_WEIGHT + LBH_NONTAKEN_WEIGHT = 128 which in _most_ cases reduce
number of rounding errors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136428
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 28 Jul 2011 23:22:41 +0000 (23:22 +0000)]
PLD and PLI are not predicable in ARM mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136427
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 28 Jul 2011 21:59:38 +0000 (21:59 +0000)]
CBZ/CBNZ are Thumb2 only. No need for ARM mode tests for them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136408
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 28 Jul 2011 21:57:55 +0000 (21:57 +0000)]
ARM assembly parsing and encoding for BLX (immediate).
Add parsing support for BLX (immediate). Since the register operand version is
predicated and the label operand version is not, we have to use some special
handling to get the operand list right for matching.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136406
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Thu, 28 Jul 2011 21:54:31 +0000 (21:54 +0000)]
Enhance the fixed-length decoder emitter to support parsing scattered fields.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136405
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Thu, 28 Jul 2011 21:48:00 +0000 (21:48 +0000)]
LangRef and basic memory-representation/reading/writing for 'cmpxchg' and
'atomicrmw' instructions, which allow representing all the current atomic
rmw intrinsics.
The allowed operands for these instructions are heavily restricted at the
moment; we can probably loosen it a bit, but supporting general
first-class types (where it makes sense) might get a bit complicated,
given how SelectionDAG works.
As an initial cut, these operations do not support specifying an alignment,
but it would be possible to add if we think it's useful. Specifying an
alignment lower than the natural alignment would be essentially
impossible to support on anything other than x86, but specifying a greater
alignment would be possible. I can't think of any useful optimizations which
would use that information, but maybe someone else has ideas.
Optimizer/codegen support coming soon.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136404
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakub Staszak [Thu, 28 Jul 2011 21:46:58 +0000 (21:46 +0000)]
If run with -debug give more information about Cyclic Probability.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136403
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakub Staszak [Thu, 28 Jul 2011 21:45:07 +0000 (21:45 +0000)]
Heuristics are in descending priority now. If we use one of them, skip the rest.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136402
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 28 Jul 2011 21:38:51 +0000 (21:38 +0000)]
Handle REG_SEQUENCE with implicitly defined operands.
Code like that would only be produced by bugpoint, but we should still
handle it correctly.
When a register is defined by a REG_SEQUENCE of undefs, the register
itself is undef. Previously, we would create a register with uses but no
defs.
Fixes part of PR10520.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136401
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 28 Jul 2011 21:37:05 +0000 (21:37 +0000)]
Remove obsolete FIXME reference in comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136400
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 28 Jul 2011 21:34:26 +0000 (21:34 +0000)]
ARM assembly parsing and encoding for BFC and BFI.
Add parsing support that handles converting the lsb+width source into the
odd way we represent the instruction (an inverted bitfield mask).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136399
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakub Staszak [Thu, 28 Jul 2011 21:33:46 +0000 (21:33 +0000)]
Add InEdges (edges from header to the loop) in Loop Branch Heuristics, so
there is no frequency difference whether condition is in the header or in
the latch.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136398
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 21:25:33 +0000 (21:25 +0000)]
Use ArrayRef instead of requiring an std::vector.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136396
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 21:14:13 +0000 (21:14 +0000)]
The personality function should be a Function* and not just a Value*.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136392
91177308-0d34-0410-b5e6-
96231b3b80d8
Douglas Gregor [Thu, 28 Jul 2011 20:55:16 +0000 (20:55 +0000)]
Fix Clang attribute reader tblgen output for a corresponding fix on the Clang side
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136390
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 28 Jul 2011 20:48:23 +0000 (20:48 +0000)]
Reverse order of RS_Split live ranges under -compact-regions.
There are two conflicting strategies in play:
- Under high register pressure, we want to assign large live ranges
first. Smaller live ranges are easier to place afterwards.
- Live range splitting is guided by interference, so splitting should be
deferred until interference is as realistic as possible.
With the recent changes to the live range stages, and with compact
regions enabled, it is less traumatic to split a live range too early.
If some of the split products were too big, they can often be split
again.
By reversing the RS_Split order, we get this queue order:
1. Normal live ranges, large to small.
2. RS_Split live ranges, large to small.
The large-to-small order improves RAGreedy's puzzle solving skills under
high register pressure. It may cause a bit more iterated splitting, but
we handle that better now.
With this change, -compact-regions is mostly an improvement on SPEC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136388
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 20:48:05 +0000 (20:48 +0000)]
Initial code to convert ResumeInsts into calls to _Unwind_Resume.
This should be the only code necessary for DWARF EH prepare.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136387
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakub Staszak [Thu, 28 Jul 2011 20:17:18 +0000 (20:17 +0000)]
Fix stupid mistake from commit 136381.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136384
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakub Staszak [Thu, 28 Jul 2011 20:09:31 +0000 (20:09 +0000)]
Speed up BlockFrequencyInfo a little bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136381
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 28 Jul 2011 19:46:12 +0000 (19:46 +0000)]
Add fixme.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136375
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 18:12:20 +0000 (18:12 +0000)]
Use version 402 for the GCDA files when compiling for Apple.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136369
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Thu, 28 Jul 2011 17:56:55 +0000 (17:56 +0000)]
Update comments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136367
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Thu, 28 Jul 2011 17:53:25 +0000 (17:53 +0000)]
Fill in some encoding information for STRD instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136366
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Thu, 28 Jul 2011 17:18:57 +0000 (17:18 +0000)]
Revert r136295. It broke nightly testers because some parts of codegen weren't aware of the changes to operand ordering. I hope to revive this sometime in the future, but it's not strictly necessary for now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136362
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 28 Jul 2011 16:33:54 +0000 (16:33 +0000)]
ARM parsing and encoding for ADR.
The label does not have a '#' prefix. Add parsing and encoding tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136360
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 28 Jul 2011 16:00:41 +0000 (16:00 +0000)]
Update ARM tests for parsing and encoding of WFE, WFI and YIELD.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136358
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Thu, 28 Jul 2011 14:38:46 +0000 (14:38 +0000)]
CR fix: The ANY_EXTEND can be removed because the input and putput type must be
identical.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136355
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Thu, 28 Jul 2011 14:37:53 +0000 (14:37 +0000)]
Use unsigned rather than uint16_t in case anyone feels like testing
more graphs, like all graphs with 5 nodes or less. With a 32 bit
unsigned type, the maximum is graphs with 6 nodes or less, but that
would take a while to test - 5 nodes or less already requires a few
seconds.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136354
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Thu, 28 Jul 2011 14:33:01 +0000 (14:33 +0000)]
Check an additional property specific to the way LLVM
iterates over SCC's.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136353
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Thu, 28 Jul 2011 14:17:11 +0000 (14:17 +0000)]
Add a unittest for the simply connected components (SCC) iterator class.
This computes every graph with 4 or fewer nodes, and checks that the SCC
class indeed returns exactly the simply connected components reachable
from the initial node.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136351
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Thu, 28 Jul 2011 12:21:47 +0000 (12:21 +0000)]
Due to changes coming from the new LLVM type system, you now get
bitcasts in this test rather than getelementptr instructions;
llvm-gcc produces two bitcasts, clang produces one.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136349
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Thu, 28 Jul 2011 08:36:22 +0000 (08:36 +0000)]
This file was moved from Support to ADT. Correct a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136344
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 07:44:07 +0000 (07:44 +0000)]
Some minor cleanups. No functionalitical change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136341
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 07:31:46 +0000 (07:31 +0000)]
Leverage some of the code that John wrote to manage the landing pads.
The new EH is more simple in many respects. Mainly, we don't have to worry about
the "llvm.eh.exception" and "llvm.eh.selector" calls being in weird places.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136339
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 07:26:41 +0000 (07:26 +0000)]
Don't add in the asked for size so that we don't copy too much from the old to new vectors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136338
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Thu, 28 Jul 2011 06:48:33 +0000 (06:48 +0000)]
In DenseMapInfo<pair<T, U>> tombstone key, use the tombstone for T and U instead
of the empty key for U. This shouldn't really matter because the tombstone key
for the pair was still distinct from every other key, but it is odd. Patch by
Michael Ilseman!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136336
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 02:40:13 +0000 (02:40 +0000)]
Automatically merge the landingpad clauses when we come across a callee's
landingpad.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136329
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Thu, 28 Jul 2011 02:33:52 +0000 (02:33 +0000)]
Explicitly declare a library dependency of LLVM*Desc to
LLVM*AsmPrinter.
GenLibDeps.pl fails to detect vtable references. As this is the only
referenced symbol from LLVM*Desc to LLVM*AsmPrinter on optimized
builds, the algorithm that creates the list of libraries to be linked
into tools doesn't know about the dependency and sometimes places the
libraries on the wrong order, yielding error messages like this:
../../lib/libLLVMARMDesc.a(ARMMCTargetDesc.cpp.o): In function
`llvm::ARMInstPrinter::ARMInstPrinter(llvm::MCAsmInfo const&)':
ARMMCTargetDesc.cpp:(.text._ZN4llvm14ARMInstPrinterC1ERKNS_9MCAsmInfoE
[llvm::ARMInstPrinter::ARMInstPrinter(llvm::MCAsmInfo
const&)]+0x2a): undefined reference to `vtable for
llvm::ARMInstPrinter'
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136328
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Thu, 28 Jul 2011 02:33:33 +0000 (02:33 +0000)]
Updated cmake library dependencies.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136327
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 02:27:12 +0000 (02:27 +0000)]
Make sure that the landingpad instruction takes a Constant* as the clause's value.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136326
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 02:15:52 +0000 (02:15 +0000)]
Add a couple of convenience functions:
* InvokeInst: Get the landingpad instruction associated with this invoke.
* LandingPadInst: A method to reserve extra space for clauses.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136325
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Thu, 28 Jul 2011 01:26:53 +0000 (01:26 +0000)]
Invert the subvector insertion to be more likely to be taken as a COPY
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136324
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Thu, 28 Jul 2011 01:26:50 +0000 (01:26 +0000)]
Add patterns to generate copies for extract_subvector instead of
using vextractf128. This will reduce the number of issued instruction
for several avx codes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136323
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Thu, 28 Jul 2011 01:26:46 +0000 (01:26 +0000)]
movd/movq write zeros in the high 128-bit part of the vector. Use
them to match 256-bit scalar_to_vector+zext.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136322
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Thu, 28 Jul 2011 01:26:43 +0000 (01:26 +0000)]
Add a few patterns to match allzeros without having to use the fp unit.
Take advantage that the 128-bit vpxor zeros the higher part and use it.
This also fixes PR10491
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136321
91177308-0d34-0410-b5e6-
96231b3b80d8
Bruno Cardoso Lopes [Thu, 28 Jul 2011 01:26:39 +0000 (01:26 +0000)]
Add SINT_TO_FP and FP_TO_SINT support for v8i32 types. Also move
a convert pattern close to the instruction definition.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136320
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Thu, 28 Jul 2011 01:20:19 +0000 (01:20 +0000)]
Fix a use after free. An instruction can't be both an intrinsic call and a fence.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136319
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 28 Jul 2011 00:38:23 +0000 (00:38 +0000)]
Initial stab at getting inlining working with the EH rewrite.
This takes the new 'resume' instruction and turns it into a direct jump to the
caller's landing pad code. The caller's landingpad instruction is merged with
the landingpad instructions of the callee. This is a bit rough and makes some
assumptions in how the code works. But it passes a simple test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136313
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Thu, 28 Jul 2011 00:37:03 +0000 (00:37 +0000)]
ARM parsing and encoding tests.
UXTAB, UXTAB16, UXTAH, UXTB, UXTB16, and UXTH.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136312
91177308-0d34-0410-b5e6-
96231b3b80d8
Argyrios Kyrtzidis [Thu, 28 Jul 2011 00:29:20 +0000 (00:29 +0000)]
Add an optional 'bool makeAbsolute' in llvm::sys::fs::unique_file function.
If true and 'model' parameter is not an absolute path, a temp directory will be prepended.
Make it true by default to match current behaviour.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136310
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Wed, 27 Jul 2011 23:36:57 +0000 (23:36 +0000)]
Refactor and improve the encodings/decodings for addrmode3 loads, and make the writeback operand always the first.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136295
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Wed, 27 Jul 2011 23:22:03 +0000 (23:22 +0000)]
Emit an error is asm parser parsed X86_64 only registers, e.g. %rax, %sil.
This can happen in cases where TableGen generated asm matcher cannot check
whether a register operand is in the right register class. e.g. mem operands.
rdar://
8204588
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@136292
91177308-0d34-0410-b5e6-
96231b3b80d8