Commit Graph

96802 Commits

Author SHA1 Message Date
Andrew Trick
6a7770b7ae Enable MI Sched for x86.
This changes the SelectionDAG scheduling preference to source
order. Soon, the SelectionDAG scheduler can be bypassed saving
a nice chunk of compile time.

Performance differences that result from this change are often a
consequence of register coalescing. The register coalescer is far from
perfect. Bugs can be filed for deficiencies.

On x86 SandyBridge/Haswell, the source order schedule is often
preserved, particularly for small blocks.

Register pressure is generally improved over the SD scheduler's ILP
mode. However, we are still able to handle large blocks that require
latency hiding, unlike the SD scheduler's BURR mode. MI scheduler also
attempts to discover the critical path in single-block loops and
adjust heuristics accordingly.

The MI scheduler relies on the new machine model. This is currently
unimplemented for AVX, so we may not be generating the best code yet.

Unit tests are updated so they don't depend on SD scheduling heuristics.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192750 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 23:33:07 +00:00
Eric Christopher
c622824ccb Make sure we're not attempting to construct a subprogram DIE
twice and just look up the value. Fix the one case where
we were trying to create a subprogram DIE and we should already
have had one. Reflow formatting in collectDeadVariables while fixing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192749 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 23:31:38 +00:00
Eric Christopher
4159733c1a Add an assert that we have a scope that matters for methods
and remove a call to getNonCompileUnitScope as a method
shouldn't be in the compile unit scope.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192748 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 23:31:36 +00:00
Eric Christopher
ff64b7d7e3 Clean up, formatting, comments. No functional change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192747 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 23:31:31 +00:00
NAKAMURA Takumi
cdfdb5e92b Support/FileSystem.h: Remove a trailing comma in enum file_magic::Impl.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192745 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 23:01:44 +00:00
David Blaikie
e4df43e36d Use ASSERT_EQ rather than ASSERT_TRUE for better unit test failures.
Also minor using namespace move so it's not hard-up against the function
definition and outside the namespace as is usual.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192744 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 23:00:17 +00:00
Vincent Lejeune
484091a50a R600/SI: Remove some leftover MI dump call
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192743 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 22:48:51 +00:00
Rui Ueyama
b32b0376d4 Path: Recognize Windows compiled resource file.
Some background: One can pass compiled resource files (.res files) directly
to the linker on Windows. If a resource file is given, the linker will run
"cvtres" command in background to convert the resource file to a COFF file
to link it.

What I'm trying to do with this patch is to make the linker to recognize
the resource file by file magic, so that it can run cvtres command.

Differential Revision: http://llvm-reviews.chandlerc.com/D1943

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192742 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 22:45:38 +00:00
Andrew Kaylor
72be32c633 Separating ELF and MachO stub info functions for RuntimeDyld
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192737 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 21:32:56 +00:00
Eric Christopher
977ce007ad Reformat.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192735 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 21:22:12 +00:00
Chad Rosier
1824bd0ef8 [AArch64] Add support for NEON scalar signed saturating absolute value and
scalar signed saturating negate instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192733 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 21:18:44 +00:00
Andrew Kaylor
ab950f5f33 Fixing some host==target assumptions in RuntimeDyld
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192732 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 20:44:55 +00:00
Adrian Prantl
73df4b9f57 Remove some dead code. (DarwinGDBCompat was retired in r189903).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192731 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 20:26:37 +00:00
Manman Ren
05ac87f864 Struct byval: fix a copy-paste error for thumb2.
PR17309


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192730 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 19:42:32 +00:00
Michael Liao
bfa7b1e695 Fix PR17546
- Type of index used in extract_vector_elt or insert_vector_elt supposes
  to be TLI.getVectorIdxTy() which is pointer type on most targets. It'd
  better to truncate (or zero-extend in case it's changed later) it to
  mask element type to guarantee they are matching instead of asserting
  that.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192722 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 17:51:58 +00:00
Michael Liao
dc8c044a9a Fix PR16807
- Lower signed division by constant powers-of-2 to target-independent
  DAG operators instead of target-dependent ones to support them better
  on targets where vector types are legal but shift operators on that
  types are illegal. E.g., on AVX, PSRAW is only available on <8 x i16>
  though <16 x i16> is a legal type.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192721 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 17:51:02 +00:00
Benjamin Kramer
d45b3c4653 LoopVectorize: Properly reflect PODness in comments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192717 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 16:19:54 +00:00
Pekka Jaaskelainen
d54946ac14 Guard the debug temp variable with NDEBUG to avoid warning/error with NDEBUG defined.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192709 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 14:40:46 +00:00
Pekka Jaaskelainen
862385112e Do not assert when trying to add a meta data operand with
MachineInstr::addOperand().



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192707 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 14:18:10 +00:00
Daniel Sanders
62e87cb241 [mips][msa] Added support for build_vector for v4f32 and v2f64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192699 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 13:14:41 +00:00
Anders Waldenborg
bfcd45cfe1 Revert "Add AllTargetsBindings sublibrary" as it breaks cmake build on (atleast) windows and darwin.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192697 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 13:04:27 +00:00
Anders Waldenborg
6480db4b3a Add AllTargetsBindings sublibrary instead of having static inlines in the llvm-c headers.
This new library will be linked in when using the "all-targets"
component and contains the LLVMInitializeAll* functions.

This means that those functions will exist as real symbols in
the shared library, and can therefore can be called from
bindings that are using ffi the shared library.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192690 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 12:08:59 +00:00
Richard Sandiford
f8f993b675 [SystemZ] Use A(G)SI when spilling the target of a constant addition
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192681 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 08:42:59 +00:00
David Majnemer
ecfa16cc21 docs: Remove incompatibility with Solaris shell
There doesn't seem to be a need in checking if a directory exists if we
will just rm -rf it once we affirm that it does.  Instead, just blindly
try to delete it.

This fixes PR17541.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192679 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 08:30:07 +00:00
Job Noorman
3edacfa970 Fix MSP430 calling convention to match MSPGCC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192678 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 08:19:39 +00:00
NAKAMURA Takumi
3734f33acc llvm/test/CodeGen/X86/break-avx-dep.ll: Relax an expression to be matched to also r[89], not only rXX.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192675 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 06:36:36 +00:00
Craig Topper
390ff499f0 Remove x86_sse42_crc32_64_8 intrinsic. It has no functional difference from x86_sse42_crc32_32_8 and was not mapped to a clang builtin. I'm not even sure why this form of the instruction is even called out explicitly in the docs. Also add AutoUpgrade support to convert it into the other intrinsic with appropriate trunc and zext.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192672 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 05:20:47 +00:00
Andrew Trick
51dee24ca6 Improve on r192635, ExeDepsFix for avx, and add a test case.
rdar:15221834 False AVX register dependencies cause 5x slowdown on
flops-5/6 and significant slowdown on several others.

This was blocking the switch to MI-Sched.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192669 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 03:39:43 +00:00
Chandler Carruth
cc3d76d982 Simplify formatting and sort these. No functionality changed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192668 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 02:03:44 +00:00
Akira Hatanaka
2ef99c5dff [mips] Define a pseudo instruction which writes to both the lower and higher
parts of the accumulators and gets expanded post-RA.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192667 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 01:48:30 +00:00
Akira Hatanaka
397f6da28c [mips] Use predicates to guard instructions using accumulator registers instead
of relying on AddedComplexity.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192665 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 01:21:37 +00:00
Akira Hatanaka
adb1297d49 [mips] Rename isel nodes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192663 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 01:12:50 +00:00
Akira Hatanaka
89fee2ff92 [mips] Transfer kill flag to the newly created operand.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192662 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 01:06:30 +00:00
Akira Hatanaka
b004913417 [mips] Set HI/LO registers' HWEncoding field.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192661 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 01:00:00 +00:00
Akira Hatanaka
b6ed641c71 [mips] Delete unnecessary code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192660 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-15 00:48:42 +00:00
Rafael Espindola
efc053bd67 Remove dead code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192642 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 23:40:11 +00:00
Michael Gottesman
611ee2ace9 Update comment list of GLOBALVAR modifiers in BitcodeWriter to include externally_initialized.
Thanks to Shuxin Yang for catching this.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192637 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 22:36:51 +00:00
Quentin Colombet
41d6f451a5 [X86][FastISel] During X86 fastisel, the address of indirect call was resolved
through bitcast, ptrtoint, and inttoptr instructions. This is valid
only if the related instructions are in that same basic block, otherwise
we may reference variables that were not live accross basic blocks
resulting in undefined virtual registers.

The bug was exposed when both SDISel and FastISel were used within the same
function, i.e., one basic block is issued with FastISel and another with SDISel,
as demonstrated with the testcase.

<rdar://problem/15192473>


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192636 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 22:32:09 +00:00
Andrew Trick
a6a9ac5aa1 Fix the ExecutionDepsFix pass to handle AVX instructions.
This pass is needed to break false dependencies. Without it, unlucky
register assignment can result in wild (5x) swings in
performance. This pass was trying to handle AVX but not getting it
right. AVX doesn't have partial register defs, it has unused register
reads in which the high bits of a source operand are copied into the
unused bits of the dest.

Fixing this requires conservative liveness analysis. This is awkard
because the pass already has its own pseudo-liveness. However, proper
liveness is expensive, and we would like to use a generic utility to
compute it. The fix only invokes liveness on-demand. It is rare to
detect a case that needs undef-read dependence breaking, but when it
happens, it can be needed many times within a very large block.

I think the existing heuristic which uses a register window of 16 is
too conservative for loop-carried false dependencies. If the loop is a
reduction. The out-of-order engine may be able to execute several loop
iterations in parallel. However, I'll leave this tuning exercise for
next time.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192635 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 22:19:03 +00:00
Andrew Trick
966772931e LiveRegUnits: Use *MBB for consistency and convenience.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192634 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 22:18:59 +00:00
Andrew Trick
ff09d7119d whitespace
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192633 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 22:18:56 +00:00
Nick Lewycky
a5a076341d Fix a typo, in a comment, in a test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192632 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 22:02:53 +00:00
Eric Christopher
9d0e94c16e Revert part of a fix from 2010, changes since then:
a) x86-64 TLS has been documented
b) the code path should use movq for the correct relocation
   to be generated.

I've also added a fixme for the test case that we should improve
the code generated, it should look something like is documented
in the tls abi document.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192631 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 21:52:26 +00:00
Eric Christopher
739eb9d82d Reformat this routine slightly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192630 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 21:52:23 +00:00
Eric Christopher
e108345b57 Remove some extraneous whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192629 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 21:52:18 +00:00
Andrew Trick
6c325c9133 LiveRegUnits::removeRegsInMask safety.
Clobbering is exclusive not inclusive on register units.
For liveness, we need to consider all the preserved registers.
e.g. A regmask that clobbers YMM0 may preserve XMM0.
Units are only clobbered when all super-registers are clobbered.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192623 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 20:45:19 +00:00
Andrew Trick
7c489ab365 Use a SparseSet in LiveRegUnits.
Some clients may add block live ins and may track liveness over a
large scope. This guarantees an efficient implementation in all cases
with no memory allocation/deallocation, independent of the number of
target registers. It could be slightly less convenient but is fine in
the expected case.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192622 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 20:45:17 +00:00
Andrew Trick
eb3b9f8ed9 Move LiveRegUnits implementation into .cpp. Comment and format.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192621 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 20:45:14 +00:00
Andrew Trick
a3e69a0467 Remove extra indentation in LiveRegUnits.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192620 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 20:45:11 +00:00
Andrew Trick
5601abb60d Convert LiveRegUnits methods to the current convention (it's new code).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192619 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-14 20:45:09 +00:00