oota-llvm.git
9 years agoSLPVectorizer: improved scheduling algorithm.
Erik Eckstein [Fri, 1 Aug 2014 09:20:42 +0000 (09:20 +0000)]
SLPVectorizer: improved scheduling algorithm.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214494 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[mips][PR19612] Fix va_arg for big-endian mode.
Daniel Sanders [Fri, 1 Aug 2014 09:17:39 +0000 (09:17 +0000)]
[mips][PR19612] Fix va_arg for big-endian mode.

Summary:
Big-endian mode was not correctly adjusting the offset for types smaller
than an ABI slot.

Fixes PR19612

Reviewers: dsanders

Reviewed By: dsanders

Subscribers: sstankovic, llvm-commits

Differential Revision: http://reviews.llvm.org/D4556

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214493 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoSLP Vectorizer: added statistics counter
Erik Eckstein [Fri, 1 Aug 2014 08:14:28 +0000 (08:14 +0000)]
SLP Vectorizer: added statistics counter

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214487 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoSLP Vectorizer: improve canonicalize tree operands of commutitive binary operands.
Erik Eckstein [Fri, 1 Aug 2014 08:05:55 +0000 (08:05 +0000)]
SLP Vectorizer: improve canonicalize tree operands of commutitive binary operands.

This reverts r214338 (except the test file) and replaces it with a more general algorithm.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214485 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRevert of 214418:
Sylvestre Ledru [Fri, 1 Aug 2014 06:16:03 +0000 (06:16 +0000)]
Revert of 214418:
"Create a default symver on Linux like ELF OSes."

Fails the build under Debian with ld.gold:
/usr/bin/ld.gold: --default-symver: unknown option

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214482 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PowerPC] Generate unaligned vector loads using intrinsics instead of regular loads
Hal Finkel [Fri, 1 Aug 2014 05:20:41 +0000 (05:20 +0000)]
[PowerPC] Generate unaligned vector loads using intrinsics instead of regular loads

Altivec vector loads on PowerPC have an interesting property: They always load
from an aligned address (by rounding down the address actually provided if
necessary). In order to generate an actual unaligned load, you can generate two
load instructions, one with the original address, one offset by one vector
length, and use a special permutation to extract the bytes desired.

When this was originally implemented, I generated these two loads using regular
ISD::LOAD nodes, now marked as aligned. Unfortunately, there is a problem with
this:

The alignment of a load does not contribute to its identity, and SDNodes
are uniqued. So, imagine that we have some unaligned load, L1, that is not
aligned. The routine will create two loads, L1(aligned) and (L1+16)(aligned).
Further imagine that there had already existed a load (L1+16)(unaligned) with
the same chain operand as the load L1. When (L1+16)(aligned) is created as part
of the lowering of L1, this load *is* also the (L1+16)(unaligned) node, just
now marked as aligned (because the new alignment overwrites the old). But the
original users of (L1+16)(unaligned) now get the data intended for the
permutation yielding the data for L1, and (L1+16)(unaligned) no longer exists
to get its own permutation-based expansion. This was PR19991.

A second potential problem has to do with the MMOs on these loads, which can be
used by AA during instruction scheduling to break chain-based dependencies. If
the new "aligned" loads get the MMO from the original unaligned load, this does
not represent the fact that it will load data from below the original address.
Normally, this would not matter, but this load might be combined with another
load pair for a previous vector, and then the dependency on the otherwise-
ignored lower bytes can matter.

To fix both problems, instead of generating the necessary loads using regular
ISD::LOAD instructions, ppc_altivec_lvx intrinsics are used instead. These are
provided with MMOs with a conservative address range.

Unfortunately, I no longer have a failing test case (since PR19991 was
reported, other changes in CodeGen have forced this bug back into hiding it
again). Nevertheless, this should fix the underlying problem.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214481 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoThis patch implements transform for pattern "(A & ~B) ^ (~A) -> ~(A & B)".
Suyog Sarda [Fri, 1 Aug 2014 05:07:20 +0000 (05:07 +0000)]
This patch implements transform for pattern "(A & ~B) ^ (~A) -> ~(A & B)".

Differential Revision: http://reviews.llvm.org/D4653

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214479 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoThis patch implements transform for pattern "(A | B) & ((~A) ^ B) -> (A & B)".
Suyog Sarda [Fri, 1 Aug 2014 04:59:26 +0000 (04:59 +0000)]
This patch implements transform for pattern "(A | B) & ((~A) ^ B) -> (A & B)".

Differential Revision: http://reviews.llvm.org/D4628

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214478 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoThis patch implements transform for pattern "( A & (~B)) | (A ^ B) -> (A ^ B)"
Suyog Sarda [Fri, 1 Aug 2014 04:50:31 +0000 (04:50 +0000)]
This patch implements transform for pattern "( A & (~B)) | (A ^ B) -> (A ^ B)"

Differential Revision: http://reviews.llvm.org/D4652

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214477 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoThis patch implements transform for pattern "(A & B) | ((~A) ^ B) -> (~A ^ B)".
Suyog Sarda [Fri, 1 Aug 2014 04:41:43 +0000 (04:41 +0000)]
This patch implements transform for pattern "(A & B) | ((~A) ^ B) -> (~A ^ B)".
Patch Credit to Ankit Jain !

Differential Revision: http://reviews.llvm.org/D4655

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214476 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Fix build warning
Tom Stellard [Fri, 1 Aug 2014 02:05:57 +0000 (02:05 +0000)]
R600/SI: Fix build warning

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214475 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64] Fix the immediate versions of the {s|u}{add|sub}.with.overflow...
Juergen Ributzka [Fri, 1 Aug 2014 01:25:55 +0000 (01:25 +0000)]
[FastISel][AArch64] Fix the immediate versions of the {s|u}{add|sub}.with.overflow intrinsics.

ADDS and SUBS cannot encode negative immediates or immediates larger than 12bit.
This fix checks if the immediate version can be used under this constraints and
if we can convert ADDS to SUBS or vice versa to support negative immediates.

Also update the test cases to test the immediate versions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214470 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PowerPC] Recognize consecutive memory accesses from intrinsics
Hal Finkel [Fri, 1 Aug 2014 01:02:01 +0000 (01:02 +0000)]
[PowerPC] Recognize consecutive memory accesses from intrinsics

When generating unaligned vector loads, we need to search for other loads or
stores nearby offset by one vector width. If we find one, then we know that we
can safely generate another aligned load at that address. Otherwise, we must
generate the next load using an offset of the vector width minus one byte (so
we don't read off the end of the allocation if the base unaligned address
happened to be aligned at runtime). We had previously done this using only
other vector loads and stores, but did not consider the PowerPC-specific vector
load/store intrinsics. Now we'll also consider vector intrinsics. By itself,
this change is a feature enhancement, but is a necessary step toward fixing the
underlying problem behind PR19991.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214469 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMS inline asm: Fix null SMLoc when 'ptr' is missing after dword & co
Reid Kleckner [Fri, 1 Aug 2014 00:59:22 +0000 (00:59 +0000)]
MS inline asm: Fix null SMLoc when 'ptr' is missing after dword & co

This improves the diagnostics from the regular assembler, but more
importantly it fixes an assertion when parsing inline assembly.  Test
landing in Clang.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214468 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Do abs/neg folding with ComplexPatterns
Tom Stellard [Fri, 1 Aug 2014 00:32:39 +0000 (00:32 +0000)]
R600/SI: Do abs/neg folding with ComplexPatterns

Abs/neg folding has moved out of foldOperands and into the instruction
selection phase using complex patterns.  As a consequence of this
change, we now prefer to select the 64-bit encoding for most
instructions and the modifier operands have been dropped from
integer VOP3 instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214467 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoTableGen: Allow AddedComplexity values to be negative
Tom Stellard [Fri, 1 Aug 2014 00:32:36 +0000 (00:32 +0000)]
TableGen: Allow AddedComplexity values to be negative

This is useful for cases when stand-alone patterns are preferred to the
patterns included in the instruction definitions.  Instead of requiring
that stand-alone patterns set a larger AddedComplexity value, which
can be confusing to new developers, the allows us to reduce the
complexity of the included patterns to achieve the same result.

There will be test cases for this added to the R600 backend in a
future commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214466 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Simplify and fix handling of VOP2 in SIInstrInfo::legalizeOperands
Tom Stellard [Fri, 1 Aug 2014 00:32:35 +0000 (00:32 +0000)]
R600/SI: Simplify and fix handling of VOP2 in SIInstrInfo::legalizeOperands

We were incorrectly assuming that all VOP2 instructions can read SGPRs
in Src0, but this is not true for instructions that read carry-in from
VCC.

The old logic has been replaced with new logic which checks the defined
register classes of the VOP2 instruction to determine whether or not to
legalize the operands.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214465 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Fold immediates when shrinking instructions
Tom Stellard [Fri, 1 Aug 2014 00:32:33 +0000 (00:32 +0000)]
R600/SI: Fold immediates when shrinking instructions

This will prevent us from using extra MOV instructions once we prefer
selecting 64-bit instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214464 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Fix incorrect commute operation in shrink instructions pass
Tom Stellard [Fri, 1 Aug 2014 00:32:28 +0000 (00:32 +0000)]
R600/SI: Fix incorrect commute operation in shrink instructions pass

We were commuting the instruction by still shrinking it using the
original opcode.

NOTE: This is a candidate for the 3.5 branch.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214463 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd support for the X86 secure guard extensions instructions in assembler (SGX).
Kevin Enderby [Thu, 31 Jul 2014 23:57:38 +0000 (23:57 +0000)]
Add support for the X86 secure guard extensions instructions in assembler (SGX).

This allows assembling the two new instructions, encls and enclu for the
SKX processor model.

Note the diffs are a bigger than what might think, but to fit the new
MRM_CF and MRM_D7 in things in the right places things had to be
renumbered and shuffled down causing a bit more diffs.

rdar://16228228

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214460 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoX86 MC: Don't crash on empty memory operand parens
Reid Kleckner [Thu, 31 Jul 2014 23:26:35 +0000 (23:26 +0000)]
X86 MC: Don't crash on empty memory operand parens

Instead, create an absolute memory operand.

Fixes PR20504.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214457 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoX86 MC: Reject invalid segment registers before a memory operand colon
Reid Kleckner [Thu, 31 Jul 2014 23:03:22 +0000 (23:03 +0000)]
X86 MC: Reject invalid segment registers before a memory operand colon

Previously we would execute unreachable during object emission.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214456 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoWhite space fix.
Louis Gerbarg [Thu, 31 Jul 2014 22:57:46 +0000 (22:57 +0000)]
White space fix.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214455 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMake classof in MemSDNode consistent with MemIntrinsicSDNode
Hal Finkel [Thu, 31 Jul 2014 22:31:33 +0000 (22:31 +0000)]
Make classof in MemSDNode consistent with MemIntrinsicSDNode

If INTRINSIC_W_CHAIN and INTRINSIC_VOID are MemIntrinsicSDNodes, and a
MemIntrinsicSDNode is a MemSDNode, then INTRINSIC_W_CHAIN and INTRINSIC_VOID
must be MemSDNodes too.

Noticed by inspection.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214452 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600: Modernize work item intrinsics test
Jan Vesely [Thu, 31 Jul 2014 22:11:03 +0000 (22:11 +0000)]
R600: Modernize work item intrinsics test

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214451 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMake sure no loads resulting from load->switch DAGCombine are marked invariant
Louis Gerbarg [Thu, 31 Jul 2014 21:45:05 +0000 (21:45 +0000)]
Make sure no loads resulting from load->switch DAGCombine are marked invariant

Currently when DAGCombine converts loads feeding a switch into a switch of
addresses feeding a load the new load inherits the isInvariant flag of the left
side. This is incorrect since invariant loads can be reordered in cases where it
is illegal to reoarder normal loads.

This patch adds an isInvariant parameter to getExtLoad() and updates all call
sites to pass in the data if they have it or false if they don't. It also
changes the DAGCombine to use that data to make the right decision when
creating the new load.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214449 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoImprove the remark generated for -Rpass-missed.
Tyler Nowicki [Thu, 31 Jul 2014 21:22:22 +0000 (21:22 +0000)]
Improve the remark generated for -Rpass-missed.

The current remark is ambiguous and makes it sounds like explicitly specifying vectorization will allow the loop to be vectorized. This is not the case. The improved remark directs the user to -Rpass-analysis=loop-vectorize to determine the cause of the pass-miss.

Reviewed by Arnold Schwaighofer`

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214445 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRevert "Remove MCObjectDisassembler.cpp as it is untested and unused." as it is appar...
Eric Christopher [Thu, 31 Jul 2014 21:18:38 +0000 (21:18 +0000)]
Revert "Remove MCObjectDisassembler.cpp as it is untested and unused." as it is apparently used, but the build didn't return errors weirdly.

This reverts commits 214437 and 214438.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214444 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoImprove the remark generated when a variable that is used outside the loop is not...
Tyler Nowicki [Thu, 31 Jul 2014 21:02:40 +0000 (21:02 +0000)]
Improve the remark generated when a variable that is used outside the loop is not a reduction or induction variable.

Reviewed by Arnold Schwaighofer

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214440 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoReplaces a few pointers with references in llvm-nm.cpp.
Rafael Espindola [Thu, 31 Jul 2014 21:00:10 +0000 (21:00 +0000)]
Replaces a few pointers with references in llvm-nm.cpp.

This opens the way for a few std::uinque_ptr cleanups.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214439 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFixing CMake problems with MCObjectDisassembler.cpp not existing.
Aaron Ballman [Thu, 31 Jul 2014 20:48:54 +0000 (20:48 +0000)]
Fixing CMake problems with MCObjectDisassembler.cpp not existing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214438 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRemove MCObjectDisassembler.cpp as it is untested and unused.
Eric Christopher [Thu, 31 Jul 2014 20:44:46 +0000 (20:44 +0000)]
Remove MCObjectDisassembler.cpp as it is untested and unused.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214437 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agomsbuild integration: remove duplicated lines and BOM from 2014 integration (PR20341)
Hans Wennborg [Thu, 31 Jul 2014 20:33:22 +0000 (20:33 +0000)]
msbuild integration: remove duplicated lines and BOM from 2014 integration (PR20341)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214435 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDWOHolder takes ownership of the argument constructor, use std::unique_ptr.
Rafael Espindola [Thu, 31 Jul 2014 20:26:42 +0000 (20:26 +0000)]
DWOHolder takes ownership of the argument constructor, use std::unique_ptr.

Thanks to David Blaikie for noticing it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214434 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUse a reference instead of a pointer.
Rafael Espindola [Thu, 31 Jul 2014 20:19:36 +0000 (20:19 +0000)]
Use a reference instead of a pointer.

This makes using a std::unique_ptr in the caller more convenient.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214433 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd documentation for lit's --show-unsupported flag
Eric Fiselier [Thu, 31 Jul 2014 20:11:13 +0000 (20:11 +0000)]
Add documentation for lit's --show-unsupported flag

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214431 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoClarify in PowerPC release notes that 32-bit PIC support is incomplete.
Bill Schmidt [Thu, 31 Jul 2014 20:04:51 +0000 (20:04 +0000)]
Clarify in PowerPC release notes that 32-bit PIC support is incomplete.

As requested, changing this wording slightly.

Thanks,
Bill

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214430 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDisable IsSub subregister assert. pr18663.
Will Schmidt [Thu, 31 Jul 2014 19:50:53 +0000 (19:50 +0000)]
Disable IsSub subregister assert.  pr18663.

This is a follow-up to the activity in the bug at
http://llvm.org/bugs/show_bug.cgi?id=18663 .  The underlying issue has
to do with how the KILL pseudo-instruction is handled.  I defer to
Hal/Jakob/Uli for additional details and background.

This will disable the (bad?) assert, add an associated fixme comment,
and add a pair of tests.

The code change and the pr18663-2.ll test are copied from the referenced
bug.  That test does not immediately fail in my environment, but I have
added the pr18663.ll test which does.

(Comment from Hal)
to provide everyone else with some context, this assert was not bad when
it was written. At that time, we only generated KILL pseudo instructions
around subregister copies. This logic, unfortunately, had its own problems.
In r199797, the relevant logic in MachineCopyPropagation was replaced to
generate KILLs for other kinds of copies too. This change in semantics broke
this now-problematic assumption in AggressiveAntiDepBreaker. The
AggressiveAntiDepBreaker really needs a proper cleanup to deal with the
change, but removing the assert (which just allows the function to return
false) is a safe conservative behavior, and should do for the time being.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214429 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRemove unused argument.
Rafael Espindola [Thu, 31 Jul 2014 19:32:04 +0000 (19:32 +0000)]
Remove unused argument.

Thanks to Justin Bogner for noticing it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214426 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMove MCObjectSymbolizer.h to MC/MCAnalysis.
Rafael Espindola [Thu, 31 Jul 2014 19:29:23 +0000 (19:29 +0000)]
Move MCObjectSymbolizer.h to MC/MCAnalysis.

The cpp file is already in lib/MC/MCAnalysis.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214424 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix ScalarEvolutionExpander when creating a PHI in a block with duplicate predecessors
Hal Finkel [Thu, 31 Jul 2014 19:13:38 +0000 (19:13 +0000)]
Fix ScalarEvolutionExpander when creating a PHI in a block with duplicate predecessors

It seems that when I fixed this, almost exactly a year ago, I did not quite do
it correctly. When we have duplicate block predecessors, we can indeed not have
different incoming values for the same block, but we *must* have duplicate
entries. So, instead of skipping the duplicates, we explicitly add the
duplicate incoming values.

Fixes PR20442.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214423 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoverify-uselistorder: Change the default -num-shuffles=5
Duncan P. N. Exon Smith [Thu, 31 Jul 2014 18:46:24 +0000 (18:46 +0000)]
verify-uselistorder: Change the default -num-shuffles=5

Change the default for `-num-shuffles` to 5 and better document the
algorithm in the header docs of `verify-uselistorder`.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214419 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoCreate a default symver on Linux like ELF OSes.
Eric Christopher [Thu, 31 Jul 2014 18:43:43 +0000 (18:43 +0000)]
Create a default symver on Linux like ELF OSes.

Patch by Adam Jackson.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214418 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUseListOrder: Handle self-users
Duncan P. N. Exon Smith [Thu, 31 Jul 2014 18:33:12 +0000 (18:33 +0000)]
UseListOrder: Handle self-users

Correctly sort self-users (such as PHI nodes).  I added a targeted test
in `test/Bitcode/use-list-order.ll` and the final missing RUN line to
tests in `test/Assembly`.

This is part of PR5680.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214417 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix loop end condition.
Eric Christopher [Thu, 31 Jul 2014 18:28:08 +0000 (18:28 +0000)]
Fix loop end condition.

Note: This code appears to be untested.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214416 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoWrong heading level for PowerPC changes in release notes
Bill Schmidt [Thu, 31 Jul 2014 15:20:30 +0000 (15:20 +0000)]
Wrong heading level for PowerPC changes in release notes

Oops.  Used the wrong heading level by mistake.

Thanks,
Bill

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214405 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRelease Notes: Overriding PPC64 and PPC64LE ABI defaults is not yet supported.
Bill Schmidt [Thu, 31 Jul 2014 15:17:33 +0000 (15:17 +0000)]
Release Notes: Overriding PPC64 and PPC64LE ABI defaults is not yet supported.

I wrongly included a description of a patch that came in after 3.5 branched
and has not been backported.

Thanks,
Bill

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214404 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd PowerPC release notes for 3.5.
Bill Schmidt [Thu, 31 Jul 2014 14:38:17 +0000 (14:38 +0000)]
Add PowerPC release notes for 3.5.

Here's my take on 3.5 changes for PowerPC.  Others please feel free to add,
edit, delete as desired.

Thanks,
Bill

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214403 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFixing an -Woverloaded-virtual warnings by exposing the hidden virtual function as...
Aaron Ballman [Thu, 31 Jul 2014 12:58:50 +0000 (12:58 +0000)]
Fixing an -Woverloaded-virtual warnings by exposing the hidden virtual function as well. No functional changes intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214400 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFixing a -Wcast-qual warning in GCC. No functional changes.
Aaron Ballman [Thu, 31 Jul 2014 12:55:49 +0000 (12:55 +0000)]
Fixing a -Wcast-qual warning in GCC. No functional changes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214399 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[msan] Fix handling of array types.
Evgeniy Stepanov [Thu, 31 Jul 2014 11:02:27 +0000 (11:02 +0000)]
[msan] Fix handling of array types.

Switch array type shadow from a single integer to
an array of integers (i.e. make it per-element).
This simplifies instrumentation of extractvalue and fixes PR20493.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214398 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[asan] Support x86 REP MOVS asm instrumentation.
Evgeniy Stepanov [Thu, 31 Jul 2014 09:11:04 +0000 (09:11 +0000)]
[asan] Support x86 REP MOVS asm instrumentation.

Patch by Yuri Gorshenin.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214395 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMergeFunctions, tiny refactoring:
Stepan Dyatkovskiy [Thu, 31 Jul 2014 07:16:59 +0000 (07:16 +0000)]
MergeFunctions, tiny refactoring:
cmpOperation has been renamed to cmpOperations (multiple form).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214392 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64] Add basic bitcast support for conversion between float and int.
Juergen Ributzka [Thu, 31 Jul 2014 06:25:37 +0000 (06:25 +0000)]
[FastISel][AArch64] Add basic bitcast support for conversion between float and int.

Fixes <rdar://problem/17867078>.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214389 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64] Add sqrt intrinsic support.
Juergen Ributzka [Thu, 31 Jul 2014 06:25:33 +0000 (06:25 +0000)]
[FastISel][AArch64] Add sqrt intrinsic support.

Fixes <rdar://problem/17867067>.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214388 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoInstCombine: Correctly propagate NSW/NUW for x-(-A) -> x+A
David Majnemer [Thu, 31 Jul 2014 04:49:29 +0000 (04:49 +0000)]
InstCombine: Correctly propagate NSW/NUW for x-(-A) -> x+A

We can only propagate the nsw bits if both subtraction instructions are
marked with the appropriate bit.

N.B.  We only propagate the nsw bit in InstCombine because the nuw case
is already handled in InstSimplify.

This fixes PR20189.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214385 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoInstSimplify: Simplify (X - (0 - Y)) if the second sub is NUW
David Majnemer [Thu, 31 Jul 2014 04:49:18 +0000 (04:49 +0000)]
InstSimplify: Simplify (X - (0 - Y)) if the second sub is NUW

If the NUW bit is set for 0 - Y, we know that all values for Y other
than 0 would produce a poison value.  This allows us to replace (0 - Y)
with 0 in the expression (X - (0 - Y)) which will ultimately leave us
with X.

This partially fixes PR20189.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214384 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix some grammatical errors.
Richard Smith [Thu, 31 Jul 2014 04:25:36 +0000 (04:25 +0000)]
Fix some grammatical errors.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214383 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64] Update and enable patchpoint and stackmap intrinsic tests for...
Juergen Ributzka [Thu, 31 Jul 2014 04:10:43 +0000 (04:10 +0000)]
[FastISel][AArch64] Update and enable patchpoint and stackmap intrinsic tests for FastISel.

This commit updates the existing SelectionDAG tests for the stackmap and patchpoint
intrinsics and enables FastISel testing. It also splits up the tests into separate
files, due to different codegen between SelectionDAG and FastISel.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214382 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64] Add MachO large code model support for function calls.
Juergen Ributzka [Thu, 31 Jul 2014 04:10:40 +0000 (04:10 +0000)]
[FastISel][AArch64] Add MachO large code model support for function calls.

Currently the large code model for MachO uses the GOT to make function calls.
Emit the required adrp and ldr instructions to load the address from the GOT.

Related to <rdar://problem/17733076>.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214381 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoA std::unique_ptr case I missed in the previous patch.
Rafael Espindola [Thu, 31 Jul 2014 03:36:00 +0000 (03:36 +0000)]
A std::unique_ptr case I missed in the previous patch.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214379 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUse std::unique_ptr to make the ownership explicit.
Rafael Espindola [Thu, 31 Jul 2014 03:12:45 +0000 (03:12 +0000)]
Use std::unique_ptr to make the ownership explicit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214377 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDon't fail tablegen immediately after failing to set a value.
Pete Cooper [Thu, 31 Jul 2014 01:44:00 +0000 (01:44 +0000)]
Don't fail tablegen immediately after failing to set a value.

Instead allow the variable to be declared, but don't attach an initializer. This allows more than a single error to be emitted before we exit.

Test case to follow soon in another patch.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214375 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd a better error message when failing to assign one tablegen value to another
Pete Cooper [Thu, 31 Jul 2014 01:43:57 +0000 (01:43 +0000)]
Add a better error message when failing to assign one tablegen value to another

This is currently for assigning from one bit init to another.  It can easily be extended to other types.

Test to follow soon in another patch.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214374 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix bit initializer which was one bit too long, but worked so long as we silently...
Pete Cooper [Thu, 31 Jul 2014 01:43:54 +0000 (01:43 +0000)]
Fix bit initializer which was one bit too long, but worked so long as we silently dropped the leading 0

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214373 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix bit initializer which was one bit too long, but worked so long as we silently...
Pete Cooper [Thu, 31 Jul 2014 01:43:51 +0000 (01:43 +0000)]
Fix bit initializer which was one bit too long, but worked so long as we silently dropped the leading 0

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214372 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDelete dead code.
Rafael Espindola [Thu, 31 Jul 2014 01:14:09 +0000 (01:14 +0000)]
Delete dead code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214370 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUseListOrder: Don't give constant IDs to GlobalValues
Duncan P. N. Exon Smith [Thu, 31 Jul 2014 00:13:28 +0000 (00:13 +0000)]
UseListOrder: Don't give constant IDs to GlobalValues

Since initializers of GlobalValues are being assigned IDs before
GlobalValues themselves, explicitly exclude GlobalValues from the
constant pool.  Added targeted test in `test/Bitcode/use-list-order.ll`
and added two more RUN lines in `test/Assembly`.

This is part of PR5680.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214368 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel] Fix the patchpoint intrinsic lowering in FastISel for large target addresses.
Juergen Ributzka [Thu, 31 Jul 2014 00:11:16 +0000 (00:11 +0000)]
[FastISel] Fix the patchpoint intrinsic lowering in FastISel for large target addresses.

This fixes a mistake where I accidentially dropped the upper 32bit of a
64bit pointer during FastISel lowering of the patchpoint intrinsic.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214367 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64 and X86] Don't emit stores for UNDEF arguments during function...
Juergen Ributzka [Thu, 31 Jul 2014 00:11:11 +0000 (00:11 +0000)]
[FastISel][AArch64 and X86] Don't emit stores for UNDEF arguments during function call lowering.

UNDEF arguments are not ment to be touched - especially for the webkit_js
calling convention. This fix reproduces the already existing behavior of
SelectionDAG in FastISel.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214366 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoverify-uselistorder: Add RUN lines to cases in test/Assembly
Duncan P. N. Exon Smith [Thu, 31 Jul 2014 00:10:27 +0000 (00:10 +0000)]
verify-uselistorder: Add RUN lines to cases in test/Assembly

Add RUN line for `verify-uselistorder` to every test in `test/Assembly`,
unless it's a negative check (assembler rejects it) or verification
fails.

There are three files that verification fails on (so I've left out the
RUN lines):

  - 2002-08-22-DominanceProblem.ll
  - ConstantExprFold.ll
  - ConstantExprFoldCast.ll

This is part of PR5680.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214365 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoX86 asm parser: Avoid duplicating the list of aliased instructions
Reid Kleckner [Thu, 31 Jul 2014 00:07:33 +0000 (00:07 +0000)]
X86 asm parser: Avoid duplicating the list of aliased instructions

No functional change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214364 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd mtpid/mfpid for BookE.
Joerg Sonnenberger [Wed, 30 Jul 2014 23:59:11 +0000 (23:59 +0000)]
Add mtpid/mfpid for BookE.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214363 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAttempt at fixing the windows dll build.
Rafael Espindola [Wed, 30 Jul 2014 23:39:30 +0000 (23:39 +0000)]
Attempt at fixing the windows dll build.

It looks like only direct (i.e., explicitly listed) dependencies are
scanned.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214361 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agollvm-profdata: Add a test for mismatched numbers of counters
Justin Bogner [Wed, 30 Jul 2014 23:36:06 +0000 (23:36 +0000)]
llvm-profdata: Add a test for mismatched numbers of counters

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214360 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agollvm-profdata: Use consistent file suffixes in tests
Justin Bogner [Wed, 30 Jul 2014 23:02:01 +0000 (23:02 +0000)]
llvm-profdata: Use consistent file suffixes in tests

In some places we've been using different suffixes for the different
file formats involved in instrprof, but in others we've just
ambiguously used .profdata. Update the test files to indicate the
types of file more obviously.

No functional change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214357 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUse "weak alias" instead of "alias weak"
Rafael Espindola [Wed, 30 Jul 2014 22:51:54 +0000 (22:51 +0000)]
Use "weak alias" instead of "alias weak"

Before this patch we had

@a = weak global ...
but
@b = alias weak ...

The patch changes aliases to look more like global variables.

Looking at some really old code suggests that the reason was that the old
bison based parser had a reduction for alias linkages and another one for
global variable linkages. Putting the alias first avoided the reduce/reduce
conflict.

The days of the old .ll parser are long gone. The new one parses just "linkage"
and a later check is responsible for deciding if a linkage is valid in a
given context.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214355 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRefactor TLBIVAX and add tlbsx.
Joerg Sonnenberger [Wed, 30 Jul 2014 22:51:15 +0000 (22:51 +0000)]
Refactor TLBIVAX and add tlbsx.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214354 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoX86 asm parser: Use a loop to disambiguate suffixes instead of copy paste
Reid Kleckner [Wed, 30 Jul 2014 22:23:11 +0000 (22:23 +0000)]
X86 asm parser: Use a loop to disambiguate suffixes instead of copy paste

This works towards making the Intel syntax asm matcher use a completely
different disambiguation strategy.

No functional change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214352 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64] Add select folding support for the XALU intrinsics.
Juergen Ributzka [Wed, 30 Jul 2014 22:04:37 +0000 (22:04 +0000)]
[FastISel][AArch64] Add select folding support for the XALU intrinsics.

This improves the code generation for the XALU intrinsics when the
condition is feeding a select instruction.

This also updates and enables the XALU unit tests for FastISel.

This fixes <rdar://problem/17831117>.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214350 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64] Add branch folding support for the XALU intrinsics.
Juergen Ributzka [Wed, 30 Jul 2014 22:04:34 +0000 (22:04 +0000)]
[FastISel][AArch64] Add branch folding support for the XALU intrinsics.

This improves the code generation for the XALU intrinsics when the
condition is feeding a branch instruction.

This is related to <rdar://problem/17831117>.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214349 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64] Add {s|u}{add|sub|mul}.with.overflow intrinsic support.
Juergen Ributzka [Wed, 30 Jul 2014 22:04:31 +0000 (22:04 +0000)]
[FastISel][AArch64] Add {s|u}{add|sub|mul}.with.overflow intrinsic support.

This commit adds support for the {s|u}{add|sub|mul}.with.overflow intrinsics.
The unit tests for FastISel will be enabled in a later commit, once there is
also branch and select folding support.

This is related to <rdar://problem/17831117>.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214348 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel] Move the helper function isCommutativeIntrinsic into FastISel base class.
Juergen Ributzka [Wed, 30 Jul 2014 22:04:28 +0000 (22:04 +0000)]
[FastISel] Move the helper function isCommutativeIntrinsic into FastISel base class.

Move the helper function isCommutativeIntrinsic into the FastISel base class,
so it can be used by more than just one backend.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214347 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64] Create helper functions to create the various multiplies on AArch64.
Juergen Ributzka [Wed, 30 Jul 2014 22:04:25 +0000 (22:04 +0000)]
[FastISel][AArch64] Create helper functions to create the various multiplies on AArch64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214346 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[FastISel][AArch64] Add support for shift-immediate.
Juergen Ributzka [Wed, 30 Jul 2014 22:04:22 +0000 (22:04 +0000)]
[FastISel][AArch64] Add support for shift-immediate.

Currently the shift-immediate versions are not supported by tblgen and
hopefully this can be later removed, once the required support has been
added to tblgen.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214345 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoInstCombine: Simplify (A ^ B) or/and (A ^ B ^ C)
David Majnemer [Wed, 30 Jul 2014 21:26:37 +0000 (21:26 +0000)]
InstCombine: Simplify (A ^ B) or/and (A ^ B ^ C)

While we can already transform A | (A ^ B) into A | B, things get bad
once we have (A ^ B) | (A ^ B ^ Cst) because reassociation will morph
this into (A ^ B) | ((A ^ Cst) ^ B).  Our existing patterns fail once
this happens.

To fix this, we add a new pattern which looks through the tree of xor
binary operators to see that, in fact, there exists a redundant xor
operation.

What follows bellow is a correctness proof of the transform using CVC3.

$ cat t.cvc
A, B, C : BITVECTOR(64);

QUERY BVXOR(A, B) | BVXOR(BVXOR(B, C), A) = BVXOR(A, B) | C;
QUERY BVXOR(BVXOR(A, C), B) | BVXOR(A, B) = BVXOR(A, B) | C;

QUERY BVXOR(A, B) & BVXOR(BVXOR(B, C), A) = BVXOR(A, B) & ~C;
QUERY BVXOR(BVXOR(A, C), B) & BVXOR(A, B) = BVXOR(A, B) & ~C;

$ cvc3 < t.cvc
Valid.
Valid.
Valid.
Valid.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214342 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd rfdi and rfmci from the e500/e500mc ISA.
Joerg Sonnenberger [Wed, 30 Jul 2014 21:09:03 +0000 (21:09 +0000)]
Add rfdi and rfmci from the e500/e500mc ISA.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214339 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoSLP Vectorizer: Canonicalize tree operands of commutitive binary operands.
Chad Rosier [Wed, 30 Jul 2014 21:07:56 +0000 (21:07 +0000)]
SLP Vectorizer: Canonicalize tree operands of commutitive binary operands.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214338 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoSimplifyCFG: Avoid miscompilations due to removed lifetime intrinsics.
Rafael Espindola [Wed, 30 Jul 2014 21:04:00 +0000 (21:04 +0000)]
SimplifyCFG: Avoid miscompilations due to removed lifetime intrinsics.

The lifetime intrinsics need some work in order to make it clear which
optimizations are or are not valid.

For now dropping this optimization avoids a miscompilation.

Patch by Björn Steinbrink.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214336 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd BookE's tlbre, tlbwe and tlbivax instructions.
Joerg Sonnenberger [Wed, 30 Jul 2014 20:44:04 +0000 (20:44 +0000)]
Add BookE's tlbre, tlbwe and tlbivax instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214332 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agodocs: update the command guide documentation for llvm-profdata.
Alex Lorenz [Wed, 30 Jul 2014 20:30:11 +0000 (20:30 +0000)]
docs: update the command guide documentation for llvm-profdata.

Differential Revision: http://reviews.llvm.org/D4726

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214331 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix test case introduced in r214322
Louis Gerbarg [Wed, 30 Jul 2014 20:26:09 +0000 (20:26 +0000)]
Fix test case introduced in r214322

This patch adds an explicit triple to the test case introduced by r214322. This
should fix build failueres that are occuring on bots that are cross building.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214330 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoLangRef: add a note about the mangling-suppressing \01 prefix
Hans Wennborg [Wed, 30 Jul 2014 20:02:08 +0000 (20:02 +0000)]
LangRef: add a note about the mangling-suppressing \01 prefix

Someone asked about this on IRC the other day, and I couldn't
find the magic prefix documented anywhere.

Differential Revision: http://reviews.llvm.org/D4728

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214329 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRefactor duplicated code.
Rafael Espindola [Wed, 30 Jul 2014 19:42:16 +0000 (19:42 +0000)]
Refactor duplicated code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214328 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFixing a few -Woverloaded-virtual warnings by exposing the hidden virtual function...
Aaron Ballman [Wed, 30 Jul 2014 19:23:59 +0000 (19:23 +0000)]
Fixing a few -Woverloaded-virtual warnings by exposing the hidden virtual function as well. No functional changes intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214325 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRetain alignment requirements for load->selects modified by DAGCombine
Louis Gerbarg [Wed, 30 Jul 2014 18:24:41 +0000 (18:24 +0000)]
Retain alignment requirements for load->selects modified by DAGCombine

DAGCombine may choose to rewrite graphs where two loads feed a select into
graphs where a select of two addresses feed a load. While it sanity checks the
loads to make sure they are broadly equivalent it currently just uses the
alignment restriction of the left node. In cases where the right node has
stronger alignment requiresment this may lead to bad codegen, such as generating
an aligned load where an unaligned load is required. This patch makes the
combine generate a load with an alignment that is the same as whichever is more
restrictive of the two alignments.

Tests included.

rdar://17762530

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214322 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUseListOrder: Visit global values
Duncan P. N. Exon Smith [Wed, 30 Jul 2014 17:51:09 +0000 (17:51 +0000)]
UseListOrder: Visit global values

When predicting use-list order, we visit functions in reverse order
followed by `GlobalValue`s and write out use-lists at the first
opportunity.  In the reader, this will translate to *after* the last use
has been added.

For this to work, we actually need to descend into `GlobalValue`s.
Added a targeted test in `use-list-order.ll` and `RUN` lines to the
newly passing tests in `test/Bitcode`.

There are two remaining failures in `test/Bitcode`:

  - blockaddress.ll: I haven't thought through how to model the way
    block addresses change the order of use-lists (or how to work around
    it).

  - metadata-2.ll: There's an old-style `@llvm.used` global array here
    that I suspect the .ll parser isn't upgrading properly.  When it
    round-trips through bitcode, the .bc reader *does* upgrade it, so
    the extra variable (`i8* null`) has an extra use, and the shuffle
    vector doesn't match.

    I think the fix is to upgrade old-style global arrays (or reject
    them?) in the .ll parser.

This is part of PR5680.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214321 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix line endings before adding RUN lines, NFC
Duncan P. N. Exon Smith [Wed, 30 Jul 2014 17:49:00 +0000 (17:49 +0000)]
Fix line endings before adding RUN lines, NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214320 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRename llvm-uselistorder => verify-uselistorder
Duncan P. N. Exon Smith [Wed, 30 Jul 2014 17:11:27 +0000 (17:11 +0000)]
Rename llvm-uselistorder => verify-uselistorder

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214318 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agollvm-uselistorder: Improve the tool description
Duncan P. N. Exon Smith [Wed, 30 Jul 2014 16:59:19 +0000 (16:59 +0000)]
llvm-uselistorder: Improve the tool description

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214317 91177308-0d34-0410-b5e6-96231b3b80d8