Chandler Carruth [Wed, 13 Nov 2013 02:49:38 +0000 (02:49 +0000)]
Add another (perhaps better) video for Sean's talk. (Thanks Marshall!)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194549
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Wed, 13 Nov 2013 02:48:20 +0000 (02:48 +0000)]
Fix a null pointer dereference when copying a null polymorphic pointer.
This bug only bit the C++98 build bots because all of the actual uses
really do move. ;] But not *quite* ready to do the whole C++11 switch
yet, so clean it up. Also add a unit test that catches this immediately.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194548
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Wed, 13 Nov 2013 02:39:07 +0000 (02:39 +0000)]
R600: Fix selection failure on EXTLOAD
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194547
91177308-0d34-0410-b5e6-
96231b3b80d8
Juergen Ributzka [Wed, 13 Nov 2013 01:57:54 +0000 (01:57 +0000)]
SelectionDAG: Teach the legalizer to split SETCC if VSELECT needs splitting too.
This patch reapplies r193676 with an additional fix for the Hexagon backend. The
SystemZ backend has already been fixed by r194148.
The Type Legalizer recognizes that VSELECT needs to be split, because the type
is to wide for the given target. The same does not always apply to SETCC,
because less space is required to encode the result of a comparison. As a result
VSELECT is split and SETCC is unrolled into scalar comparisons.
This commit fixes the issue by checking for VSELECT-SETCC patterns in the DAG
Combiner. If a matching pattern is found, then the result mask of SETCC is
promoted to the expected vector mask type for the given target. Now the type
legalizer will split both VSELECT and SETCC.
This allows the following X86 DAG Combine code to sucessfully detect the MIN/MAX
pattern. This fixes PR16695, PR17002, and <rdar://problem/
14594431>.
Reviewed by Nadav
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194542
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Wed, 13 Nov 2013 01:51:36 +0000 (01:51 +0000)]
Give folks a reference to some material on the fundamental design
pattern in use here. Addresses review feedback from Sean (thanks!) and
others.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194541
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Wed, 13 Nov 2013 01:12:08 +0000 (01:12 +0000)]
Introduce an AnalysisManager which is like a pass manager but with a lot
more smarts in it. This is where most of the interesting logic that used
to live in the implicit-scheduling-hackery of the old pass manager will
live.
Like the previous commits, note that this is a very early prototype!
I expect substantial changes before this is ready to use.
The core of the design is the following:
- We have an AnalysisManager which can be used across a series of
passes over a module.
- The code setting up a pass pipeline registers the analyses available
with the manager.
- Individual transform passes can check than an analysis manager
provides the analyses they require in order to fail-fast.
- There is *no* implicit registration or scheduling.
- Analysis passes are different from other passes: they produce an
analysis result that is cached and made available via the analysis
manager.
- Cached results are invalidated automatically by the pass managers.
- When a transform pass requests an analysis result, either the analysis
is run to produce the result or a cached result is provided.
There are a few aspects of this design that I *know* will change in
subsequent commits:
- Currently there is no "preservation" system, that needs to be added.
- All of the analysis management should move up to the analysis library.
- The analysis management needs to support at least SCC passes. Maybe
loop passes. Living in the analysis library will facilitate this.
- Need support for analyses which are *both* module and function passes.
- Need support for pro-actively running module analyses to have cached
results within a function pass manager.
- Need a clear design for "immutable" passes.
- Need support for requesting cached results when available and not
re-running the pass even if that would be necessary.
- Need more thorough testing of all of this infrastructure.
There are other aspects that I view as open questions I'm hoping to
resolve as I iterate a bit on the infrastructure, and especially as
I start writing actual passes against this.
- Should we have separate management layers for function, module, and
SCC analyses? I think "yes", but I'm not yet ready to switch the code.
Adding SCC support will likely resolve this definitively.
- How should the 'require' functionality work? Should *that* be the only
way to request results to ensure that passes always require things?
- How should preservation work?
- Probably some other things I'm forgetting. =]
Look forward to more patches in shorter order now that this is in place.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Wed, 13 Nov 2013 01:12:01 +0000 (01:12 +0000)]
Update the docs to match the function name.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194537
91177308-0d34-0410-b5e6-
96231b3b80d8
Aaron Ballman [Wed, 13 Nov 2013 00:20:43 +0000 (00:20 +0000)]
Removing llvm::huge_vald and llvm::huge_vall because they are not currently used, and HUGE_VALD does not appear to be supported everywhere anyways.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194535
91177308-0d34-0410-b5e6-
96231b3b80d8
Aaron Ballman [Wed, 13 Nov 2013 00:15:44 +0000 (00:15 +0000)]
Replacing HUGE_VALF with llvm::huge_valf in order to work around a warning triggered in MSVC 12.
Patch reviewed by Reid Kleckner and Jim Grosbach.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194533
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Tue, 12 Nov 2013 23:27:08 +0000 (23:27 +0000)]
Remove always true flag.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194530
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Tue, 12 Nov 2013 22:58:39 +0000 (22:58 +0000)]
Cleanup the stackmap operand folding code and fix a corner case.
I still don't know how to refer to the fixed operands symbolically. I
plan to look into it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194529
91177308-0d34-0410-b5e6-
96231b3b80d8
Sebastian Pop [Tue, 12 Nov 2013 22:47:30 +0000 (22:47 +0000)]
improve dependence analysis testcases
print the name of the function on which the dependence analysis is performed
such that changes to the testcase are easier to review.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194528
91177308-0d34-0410-b5e6-
96231b3b80d8
Sebastian Pop [Tue, 12 Nov 2013 22:47:20 +0000 (22:47 +0000)]
delinearization of arrays
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194527
91177308-0d34-0410-b5e6-
96231b3b80d8
Sebastian Pop [Tue, 12 Nov 2013 22:47:05 +0000 (22:47 +0000)]
remove virtual methods in SCEVApplyRewriter and SCEVParameterRewriter
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194526
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Tue, 12 Nov 2013 22:38:59 +0000 (22:38 +0000)]
Fold (iszero(A&K1) | iszero(A&K2)) -> (A&(K1|K2)) != (K1|K2) if we know that K1 and K2 are 'one-hot' (only one bit is on).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194525
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Tue, 12 Nov 2013 22:37:16 +0000 (22:37 +0000)]
FoldBranchToCommonDest merges branches into a single branch with or/and of the condition. It has a heuristics for estimating when some of the dependencies are processed by out-of-order processors. This patch adds another rule to the heuristics that says that if the "BonusInstruction" that we speculatively execute is used by the condition of the second branch then it is okay to hoist it. This change exposes more opportunities for other passes to transform the code. It does not matter that much that we if-convert the code because the selectiondag builder splits or/and branches into multiple branches when profitable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194524
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 12 Nov 2013 22:16:18 +0000 (22:16 +0000)]
[mips] Fix a bug in function CC_MipsO32_FP64. The second double precision
argument was not being passed in $f14.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194522
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 12 Nov 2013 22:06:45 +0000 (22:06 +0000)]
[mips] Run test case with command line option -mattr=+fp64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194519
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Tue, 12 Nov 2013 21:47:44 +0000 (21:47 +0000)]
Add a FIXME for 32-bit q modifiers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194515
91177308-0d34-0410-b5e6-
96231b3b80d8
Justin Bogner [Tue, 12 Nov 2013 21:44:01 +0000 (21:44 +0000)]
Protect user-supplied runtime library functions in LTO
Add user-supplied C runtime and compiler-rt library functions to
llvm.compiler.used to protect them from premature optimization by
passes like -globalopt and -ipsccp. Calls to (seemingly unused)
runtime library functions can be added by -instcombine and instruction
lowering.
Patch by Duncan Exon Smith, thanks!
Fixes <rdar://problem/
14740087>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194514
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 12 Nov 2013 21:32:41 +0000 (21:32 +0000)]
ARM: diagnose invalid system LDM/STM
The system LDM and STM instructions can't usually writeback to the base
register. The one exception is when an LDM is actually an exception-return
(i.e. contains PC in the register list).
(There's already a test that "ldm sp!, {r0-r3, pc}^" works, which is why there
is no positive test).
rdar://problem/
15223374
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194512
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 12 Nov 2013 21:10:25 +0000 (21:10 +0000)]
[mips] Revert part of r194510 that was accidentally committed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194511
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 12 Nov 2013 21:03:57 +0000 (21:03 +0000)]
[mips] Fix and re-enable a test case that has been disabled for a long time.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194510
91177308-0d34-0410-b5e6-
96231b3b80d8
Peter Zotov [Tue, 12 Nov 2013 20:55:49 +0000 (20:55 +0000)]
[OCaml] Dynamically link LLVM on --enable-shared builds
This commit significantly speeds up both bytecode and native
builds of LLVM clients (from ~20 second to sub-second link time),
and allows to invoke LLVM functions from OCaml toplevel.
The behavior for --disable-shared builds is unchanged.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194509
91177308-0d34-0410-b5e6-
96231b3b80d8
Peter Zotov [Tue, 12 Nov 2013 20:55:42 +0000 (20:55 +0000)]
[OCaml] Fix a typo
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194508
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Tue, 12 Nov 2013 20:21:43 +0000 (20:21 +0000)]
Corruptly merge constants with explicit and implicit alignments.
Constant merge can merge a constant with implicit alignment with one that has
explicit alignment. Before this change it was assuming that the explicit
alignment was higher than the implicit one, causing the result to be under
aligned in some cases.
Fixes pr17815.
Patch by Chris Smowton!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194506
91177308-0d34-0410-b5e6-
96231b3b80d8
Weiming Zhao [Tue, 12 Nov 2013 19:57:43 +0000 (19:57 +0000)]
Export intrinsics:__builtin_arm_{dmb,dsb} to frontend
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194505
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Tue, 12 Nov 2013 19:13:08 +0000 (19:13 +0000)]
[AArch64] Implemented AdvSIMD scalar x indexed element format and AdvSIMD scalar
copy in MC layer. Added the MC layer tests. Fixed triple setting in test cases.
Patch by Ana Pazos <apazos@codeaurora.org>.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194501
91177308-0d34-0410-b5e6-
96231b3b80d8
Roman Divacky [Tue, 12 Nov 2013 19:04:45 +0000 (19:04 +0000)]
Expand rotate instructions on sparcv9 as well.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194500
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Tue, 12 Nov 2013 18:06:12 +0000 (18:06 +0000)]
Simplify operand folding when rematerializing a load.
We already know how to fold a reload from a frameindex without
analyzing the load instruction. Generalize this to handle any
frameindex load. This streamlines the logic for rematerializing loads
from stack arguments. As a side effect, it allows stackmaps to record
a stack argument location without spilling it.
Verified no effect on codegen for llvm test-suite.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194497
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Tue, 12 Nov 2013 18:06:09 +0000 (18:06 +0000)]
GraphViz CFGPrinter: wrap long lines.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194496
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Tue, 12 Nov 2013 18:06:06 +0000 (18:06 +0000)]
whitespace
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194495
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Tue, 12 Nov 2013 16:37:31 +0000 (16:37 +0000)]
Revert "Remove unused variable."
This reverts commit r194485.
The variable is unused in some macro instantiations, but not others. We should
probably fix clang to not warn on this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194486
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Tue, 12 Nov 2013 16:31:59 +0000 (16:31 +0000)]
Remove unused variable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194485
91177308-0d34-0410-b5e6-
96231b3b80d8
Vincent Lejeune [Tue, 12 Nov 2013 16:26:47 +0000 (16:26 +0000)]
R600: Reenable llvm.R600.load.input/interp.input for compatibility
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194484
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Tue, 12 Nov 2013 12:56:01 +0000 (12:56 +0000)]
[mips][msa] Enable inlinse assembly for MSA.
Like GCC, this re-uses the 'f' constraint and a new 'w' print-modifier:
asm ("ldi.w %w0, 1", "=f"(result));
Unlike GCC, the 'w' print-modifer is not _required_ to produce the intended
output. This is a consequence of differences in the internal handling of
the registers in each compiler. To be source-compatible between the
compilers, users must use the 'w' print-modifier.
MSA registers (including control registers) are supported in clobber lists.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194476
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Tue, 12 Nov 2013 12:24:36 +0000 (12:24 +0000)]
SimplifyCFG: Use existing constant folding logic when forming switch tables.
Both simpler and more powerful than the hand-rolled folding logic.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194475
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Tue, 12 Nov 2013 11:14:18 +0000 (11:14 +0000)]
[mips][msa] Fix buildbot failures caused by an unused variable when assertions are disabled.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194472
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Tue, 12 Nov 2013 10:45:18 +0000 (10:45 +0000)]
[mips][msa] Added support for matching bclr, and bclri from normal IR (i.e. not intrinsics)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194471
91177308-0d34-0410-b5e6-
96231b3b80d8
Bradley Smith [Tue, 12 Nov 2013 10:38:05 +0000 (10:38 +0000)]
[ARM] Add support for FP_HP_extension build attribute
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194470
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Tue, 12 Nov 2013 10:31:49 +0000 (10:31 +0000)]
[mips][msa] Added support for matching bset, bseti, bneg, and bnegi from normal IR (i.e. not intrinsics)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194469
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Tue, 12 Nov 2013 10:14:18 +0000 (10:14 +0000)]
[mips][msa] Change constant used in ori tests to avoid conflict with bseti (also xori to avoid bnegi)
Upcoming commit(s) are going to add support for bseti and bnegi. This would
cause some existing tests to (correctly) change behaviour and emit a different
instruction. This patch prevents this by changing the constant used in ori and
xori tests so that they will not be matchable by the bseti and bnegi patterns
when these instructions are matchable from normal IR.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194467
91177308-0d34-0410-b5e6-
96231b3b80d8
Robert Lytton [Tue, 12 Nov 2013 10:11:35 +0000 (10:11 +0000)]
XCore target: fix bug in aligning 'byval i8*' on the stack
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194466
91177308-0d34-0410-b5e6-
96231b3b80d8
Robert Lytton [Tue, 12 Nov 2013 10:11:30 +0000 (10:11 +0000)]
XCore target test for hidden declaration
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194465
91177308-0d34-0410-b5e6-
96231b3b80d8
Robert Lytton [Tue, 12 Nov 2013 10:11:26 +0000 (10:11 +0000)]
Add XCore support for ATOMIC_FENCE.
ATOMIC_FENCE is lowered to a compiler barrier which is codegen only. There
is no need to emit an instructions since the XCore provides sequential
consistency.
Original patch by Richard Osborne
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194464
91177308-0d34-0410-b5e6-
96231b3b80d8
Robert Lytton [Tue, 12 Nov 2013 10:11:05 +0000 (10:11 +0000)]
XCore target: return error for unsupported alignment
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194463
91177308-0d34-0410-b5e6-
96231b3b80d8
Wan Xiaofei [Tue, 12 Nov 2013 09:40:41 +0000 (09:40 +0000)]
Change data structure to memorize computed result in ScalarEvolution
Replace std::map with SmallVector to memorize the cached result since SCEV usually belongs to little Loop/BB
Linear scan on SmallVector is faster than std::map.
Code reviewer : Andrew Trick.
Test result : Pass Unit Test & LLVM Test Suite
401.bzip2 0.425721 0.419981 101.37%
403.gcc 24.53855 24.2667 101.12%
429.mcf 0.060847 0.059944 101.51%
433.milc 0.646009 0.636119 101.55%
444.namd 1.383928 1.370614 100.97%
445.gobmk 5.836575 5.800225 100.63%
450.soplex 1.911257 1.895963 100.81%
456.hmmer 1.039565 1.032534 100.68%
458.sjeng 0.897401 0.885567 101.34%
464.h264ref 3.645908 3.577991 101.90%
470.lbm 0.049456 0.048398 102.19%
471.omnetpp 5.638575 5.60435 100.61%
bitmnp01 0.045738 0.045291 100.99%
cjpegv2data 0.304359 0.302833 100.50%
idctrn01 0.046433 0.045763 101.46%
quake2 4.534416 4.4952 100.87%
quake 2.688566 2.659208 101.10%
xcsoar 12.42545 12.30385 100.99%
linpack 0.038739 0.03803 101.86%
matrix01 0.053564 0.0528 101.45%
nbench 0.402867 0.395803 101.78%
tblook01 0.021265 0.021015 101.19%
ttsprk01 0.066384 0.065566 101.25%
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194459
91177308-0d34-0410-b5e6-
96231b3b80d8
Shuxin Yang [Tue, 12 Nov 2013 08:33:03 +0000 (08:33 +0000)]
Correct a glitch in r194424 which may invalidate iterator.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194457
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Tue, 12 Nov 2013 08:05:30 +0000 (08:05 +0000)]
Add new FileCheck feature to 3.4 release notes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194456
91177308-0d34-0410-b5e6-
96231b3b80d8
Yuchen Wu [Tue, 12 Nov 2013 05:57:06 +0000 (05:57 +0000)]
Revert "Added basic unit test for llvm-cov."
This reverts commit r194451.
Not sure why the tests are failing on the buildbot. They run fine on my
local machine. Could it possibly be because of the endianness of the
architectures? The GCNO and GCDA files are little-endian encoded, and
llvm-cov expects it to remain that way. Is this a safe assumption?
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194454
91177308-0d34-0410-b5e6-
96231b3b80d8
Yuchen Wu [Tue, 12 Nov 2013 04:59:08 +0000 (04:59 +0000)]
llvm-cov: Added call to update run/program counts.
Also updated test files that were generated from this change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194453
91177308-0d34-0410-b5e6-
96231b3b80d8
Yuchen Wu [Tue, 12 Nov 2013 04:52:53 +0000 (04:52 +0000)]
Added basic unit test for llvm-cov.
This test compares the output of llvm-cov against a coverage file
generated by gcov. Since the source file must be in the current
directory when reading GCNO files, the test will first cd into the
Inputs directory.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194451
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Tue, 12 Nov 2013 02:35:51 +0000 (02:35 +0000)]
R600/SI: Change formatting of printed registers.
Print the range of registers used with a single letter prefix.
This better matches what the shader compiler produces and
is overall less obnoxious than concatenating all of the
subregister names together.
Instead of SGPR0, it will print s0. Instead of SGPR0_SGPR1,
it will print s[0:1] and so on.
There doesn't appear to be a straightforward way
to get the actual register info in the InstPrinter,
so this parses the generated name to print with the
new syntax.
The required test changes are pretty nasty, and register
matching regexes are now worse. Since there isn't a way to
add to a variable in FileCheck, some of the tests now don't
check the exact number of registers used, but I don't think that
will be a real problem.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194443
91177308-0d34-0410-b5e6-
96231b3b80d8
Reed Kotler [Tue, 12 Nov 2013 02:27:12 +0000 (02:27 +0000)]
Change the default branch instruction to be the 16 bit variety for mips16.
This has no material effect at this time since we don't have a direct
object emitter for mips16 and the assembler can't tell them apart. I
place a comment "16 bit inst" for those so that I can tell them apart in the
output. The constant island pass has only been minimally changed to allow
this. More complete branch work is forthcoming but this is the first
step.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194442
91177308-0d34-0410-b5e6-
96231b3b80d8
Reid Kleckner [Tue, 12 Nov 2013 01:31:00 +0000 (01:31 +0000)]
Extract a bc attr parsing helper that returns Attribute::None on error
The parsing method still returns llvm::error_code for consistency with
other parsing methods. Minor cleanup, no functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194437
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Mon, 11 Nov 2013 23:31:02 +0000 (23:31 +0000)]
R600/SI: Add test that fails due to requiring i64 mul for pointers
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194433
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Mon, 11 Nov 2013 23:00:41 +0000 (23:00 +0000)]
Lower X86::MORESTACK_RET and X86::MORESTACK_RET_RESTORE_R10 in
X86AsmPrinter::EmitInstruction, rather than X86MCInstLower::Lower.
The aim is to improve the reusability of the X86MCInstLower class by making it
more function-like. The X86::MORESTACK_RET_RESTORE_R10 pseudo broke the
function model by emitting an extra instruction to the MCStreamer attached to
the AsmPrinter.
The patch should have no impact on generated code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194431
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Mon, 11 Nov 2013 22:40:25 +0000 (22:40 +0000)]
Fix the recently added anyregcc convention to handle spilled operands.
Fixes <rdar://
15432754> [JS] Assertion: "Folded a def to a non-store!"
The primary purpose of anyregcc is to prevent a patchpoint's call
arguments and return value from being spilled. They must be available
in a register, although the calling convention does not pin the
register. It's up to the front end to avoid using this convention for
calls with more arguments than allocatable registers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194428
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Mon, 11 Nov 2013 22:40:22 +0000 (22:40 +0000)]
Print new JavaScript calling conventions symbolically.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194427
91177308-0d34-0410-b5e6-
96231b3b80d8
Vincent Lejeune [Mon, 11 Nov 2013 22:10:24 +0000 (22:10 +0000)]
R600: Use function inputs to represent data stored in gpr
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194425
91177308-0d34-0410-b5e6-
96231b3b80d8
Shuxin Yang [Mon, 11 Nov 2013 22:00:23 +0000 (22:00 +0000)]
Fix PR17952.
The symptom is that an assertion is triggered. The assertion was added by
me to detect the situation when value is propagated from dead blocks.
(We can certainly get rid of assertion; it is safe to do so, because propagating
value from dead block to alive join node is certainly ok.)
The root cause of this bug is : edge-splitting is conducted on the fly,
the edge being split could be a dead edge, therefore the block that
split the critial edge needs to be flagged "dead" as well.
There are 3 ways to fix this bug:
1) Get rid of the assertion as I mentioned eariler
2) When an dead edge is split, flag the inserted block "dead".
3) proactively split the critical edges connecting dead and live blocks when
new dead blocks are revealed.
This fix go for 3) with additional 2 LOC.
Testing case was added by Rafael the other day.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194424
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Mon, 11 Nov 2013 21:49:03 +0000 (21:49 +0000)]
[mips] Partially revert r193641. Stack alignment should not be determined by
the floating point register mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194423
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Atanasyan [Mon, 11 Nov 2013 20:51:48 +0000 (20:51 +0000)]
Add support for DT_VERxxx and DT_MIPS_xxx .dynamic section entries to the
llvm-readobj.
The patch reviewed by Michael Spencer.
http://llvm-reviews.chandlerc.com/D2113
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194421
91177308-0d34-0410-b5e6-
96231b3b80d8
Bob Wilson [Mon, 11 Nov 2013 20:08:24 +0000 (20:08 +0000)]
Change libLTO back to linking with @executable_path instead of @rpath.
This partially reverts r187641 until ld64 adopts a change to link with an
rpath setting.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194418
91177308-0d34-0410-b5e6-
96231b3b80d8
Arnaud A. de Grandmaison [Mon, 11 Nov 2013 19:56:14 +0000 (19:56 +0000)]
CalcSpillWeights: allow overidding the spill weight normalizing function
This will enable the PBQP register allocator to provide its own normalizing function.
No functionnal change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194417
91177308-0d34-0410-b5e6-
96231b3b80d8
Artyom Skrobov [Mon, 11 Nov 2013 19:56:13 +0000 (19:56 +0000)]
[ARM] Add support for MVFR2 which is new in ARMv8
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194416
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Kaylor [Mon, 11 Nov 2013 19:55:10 +0000 (19:55 +0000)]
Fixing a problem with iterator validity in RuntimeDyldImpl::resolveExternalSymbols
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194415
91177308-0d34-0410-b5e6-
96231b3b80d8
Justin Holewinski [Mon, 11 Nov 2013 19:28:19 +0000 (19:28 +0000)]
[NVPTX] Properly handle bitcast ConstantExpr when checking for the alignment of function parameters
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194410
91177308-0d34-0410-b5e6-
96231b3b80d8
Justin Holewinski [Mon, 11 Nov 2013 19:28:16 +0000 (19:28 +0000)]
[NVPTX] Fix logic error in loading vector parameters of more than 4 components
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194409
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Mon, 11 Nov 2013 19:11:11 +0000 (19:11 +0000)]
[AArch64] The shift right/left and insert immediate builtins expect 3
source operands, a vector, an element to insert, and a shift amount.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194406
91177308-0d34-0410-b5e6-
96231b3b80d8
Arnaud A. de Grandmaison [Mon, 11 Nov 2013 19:04:45 +0000 (19:04 +0000)]
CalcSpillWeights: give a better describing name to calculateSpillWeights
Besides, this relates it more obviously to the VirtRegAuxInfo::calculateSpillWeightAndHint.
No functionnal change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194404
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 11 Nov 2013 18:52:39 +0000 (18:52 +0000)]
Unify the adding of enumerators with the construction of the enumeration.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194401
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 11 Nov 2013 18:52:36 +0000 (18:52 +0000)]
Formatting.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194400
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 11 Nov 2013 18:52:33 +0000 (18:52 +0000)]
80-col.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194399
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 11 Nov 2013 18:52:31 +0000 (18:52 +0000)]
Just pass the DIComposite type by value instead of by pointer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194398
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Mon, 11 Nov 2013 18:04:07 +0000 (18:04 +0000)]
[AArch64] Add support for NEON scalar floating-point convert to fixed-point instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194394
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Mon, 11 Nov 2013 17:23:41 +0000 (17:23 +0000)]
Vector forms of SHL, SRA, and SRL can be constant folded using SimplifyVBinOp too
Reviewers: dsanders
Reviewed By: dsanders
CC: llvm-commits, nadav
Differential Revision: http://llvm-reviews.chandlerc.com/D1958
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194393
91177308-0d34-0410-b5e6-
96231b3b80d8
Matheus Almeida [Mon, 11 Nov 2013 16:46:20 +0000 (16:46 +0000)]
[mips][msa] CHECK-DAG-ize MSA 3r-a.ll test.
No functional changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194391
91177308-0d34-0410-b5e6-
96231b3b80d8
Matheus Almeida [Mon, 11 Nov 2013 16:38:55 +0000 (16:38 +0000)]
[mips][msa] CHECK-DAG-ize MSA 2rf_int_float.ll test.
No functional changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194390
91177308-0d34-0410-b5e6-
96231b3b80d8
Matheus Almeida [Mon, 11 Nov 2013 16:31:46 +0000 (16:31 +0000)]
[mips][msa] CHECK-DAG-ize MSA 2rf_float_int.ll test.
No functional changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194389
91177308-0d34-0410-b5e6-
96231b3b80d8
Renato Golin [Mon, 11 Nov 2013 16:27:35 +0000 (16:27 +0000)]
Move debug message in vectorizer
No functional change, just better reporting.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194388
91177308-0d34-0410-b5e6-
96231b3b80d8
Matheus Almeida [Mon, 11 Nov 2013 16:24:53 +0000 (16:24 +0000)]
[mips][msa] CHECK-DAG-ize MSA 2rf.ll test.
No functional changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194387
91177308-0d34-0410-b5e6-
96231b3b80d8
Matheus Almeida [Mon, 11 Nov 2013 16:16:53 +0000 (16:16 +0000)]
[mips][msa] CHECK-DAG-ize MSA 2r.ll test.
No functional changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194386
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Mon, 11 Nov 2013 15:37:52 +0000 (15:37 +0000)]
Add a testcase for pr17852.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194385
91177308-0d34-0410-b5e6-
96231b3b80d8
Hal Finkel [Mon, 11 Nov 2013 14:58:40 +0000 (14:58 +0000)]
Add PPC option for full register names in asm
On non-Darwin PPC systems, we currently strip off the register name prefix
prior to instruction printing. So instead of something like this:
mr r3, r4
we print this:
mr 3, 4
The first form is the default on Darwin, and is understood by binutils, but not
yet understood by our integrated assembler. Once our integrated-as understands
full register names as well, this temporary option will be replaced by tying
this functionality to the verbose-asm option. The numeric-only form is
compatible with legacy assemblers and tools, and is also gcc's default on most
PPC systems. On the other hand, it is harder to read, and there are some
analysis tools that expect full register names.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194384
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 11 Nov 2013 14:54:34 +0000 (14:54 +0000)]
Simplify code. No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194383
91177308-0d34-0410-b5e6-
96231b3b80d8
Peter Zotov [Mon, 11 Nov 2013 14:47:28 +0000 (14:47 +0000)]
[OCaml] Add missing Llvm_target functions
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194382
91177308-0d34-0410-b5e6-
96231b3b80d8
Peter Zotov [Mon, 11 Nov 2013 14:47:20 +0000 (14:47 +0000)]
[OCaml] Accept context explicitly in Llvm_target functions
Llvm_target.intptr_type used to implicitly use global context. As
none of other functions in OCaml bindings do, it is changed to
accept context explicitly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194381
91177308-0d34-0410-b5e6-
96231b3b80d8
Peter Zotov [Mon, 11 Nov 2013 14:47:11 +0000 (14:47 +0000)]
[OCaml] Make Llvm_target.DataLayout.t automatically managed
This breaks the API by removing Llvm_target.DataLayout.dispose.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194380
91177308-0d34-0410-b5e6-
96231b3b80d8
Peter Zotov [Mon, 11 Nov 2013 14:47:01 +0000 (14:47 +0000)]
[llvm-c] Remove dead typedef
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194379
91177308-0d34-0410-b5e6-
96231b3b80d8
Sylvestre Ledru [Mon, 11 Nov 2013 14:27:56 +0000 (14:27 +0000)]
Update of the 'Code Review' page. The link is now called 'Create Diff' instead of 'Create Revision'
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194378
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Mon, 11 Nov 2013 13:53:52 +0000 (13:53 +0000)]
[autoconf] Prune "runtime" stuff in configure, corresponding to r191835.
config.status: executing runtime/Makefile commands
autoconf/install-sh: runtime/Makefile does not exist.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194376
91177308-0d34-0410-b5e6-
96231b3b80d8
Evgeniy Stepanov [Mon, 11 Nov 2013 13:37:10 +0000 (13:37 +0000)]
[msan] Propagate origin for insertvalue, extractvalue.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194374
91177308-0d34-0410-b5e6-
96231b3b80d8
Justin Holewinski [Mon, 11 Nov 2013 12:58:14 +0000 (12:58 +0000)]
[NVPTX] Blacklist TailDuplicate pass
This causes issues with virtual registers. We will likely need
to fix TailDuplicate in the future, or introduce a new version
that plays nicely with vregs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194373
91177308-0d34-0410-b5e6-
96231b3b80d8
Pete Cooper [Mon, 11 Nov 2013 05:14:42 +0000 (05:14 +0000)]
Don't universally enable initialiser lists on GCC. Thanks for catching this Chandler
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194365
91177308-0d34-0410-b5e6-
96231b3b80d8
Pete Cooper [Mon, 11 Nov 2013 03:58:00 +0000 (03:58 +0000)]
Add LLVM_HAS_INITIALIZER_LISTS for upcoming C++11 support. Use it in ArrayRef
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194362
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Mon, 11 Nov 2013 03:35:43 +0000 (03:35 +0000)]
AArch64: refactor vector list creation to be more uniform
Instructions taking a vector list (e.g. "ld2 {v0.2d, v1.d2}, [x0]") need a
special register-class to deal with the constraints, and C++ code to support
selection. However, that C++ code can be made reasonably uniform to simplify
the selection process. Hence this patch.
No functionality change, so no tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194361
91177308-0d34-0410-b5e6-
96231b3b80d8
Arnaud A. de Grandmaison [Sun, 10 Nov 2013 17:46:31 +0000 (17:46 +0000)]
CalculateSpillWeights does not need to be a pass
Based on discussions with Lang Hames and Jakob Stoklund Olesen at the hacker's lab, and in the light of upcoming work on the PBQP register allocator, it was though that CalcSpillWeights does not need to be a pass. This change will enable to customize / tune the spill weight computation depending on the allocator.
Update the documentation style while there.
No functionnal change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194356
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Sun, 10 Nov 2013 14:26:08 +0000 (14:26 +0000)]
Mark 36 tests as XFAIL:vg_leak in llvm/test/TableGen.
In historical reason, tblgen is not strictly required to be free from memory leaks.
For now, I mark them as XFAIL, they could be fixed, though.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194353
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Sun, 10 Nov 2013 14:25:44 +0000 (14:25 +0000)]
Remove 6 of XFAIL(s) in llvm/test/TableGen, since r193736. They have been XPASSing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194352
91177308-0d34-0410-b5e6-
96231b3b80d8