oota-llvm.git
10 years ago[LPM] Make LoopSimplify no longer a LoopPass and instead both a utility
Chandler Carruth [Thu, 23 Jan 2014 11:23:19 +0000 (11:23 +0000)]
[LPM] Make LoopSimplify no longer a LoopPass and instead both a utility
function and a FunctionPass.

This has many benefits. The motivating use case was to be able to
compute function analysis passes *after* running LoopSimplify (to avoid
invalidating them) and then to run other passes which require
LoopSimplify. Specifically passes like unrolling and vectorization are
critical to wire up to BranchProbabilityInfo and BlockFrequencyInfo so
that they can be profile aware. For the LoopVectorize pass the only
things in the way are LoopSimplify and LCSSA. This fixes LoopSimplify
and LCSSA is next on my list.

There are also a bunch of other benefits of doing this:
- It is now very feasible to make more passes *preserve* LoopSimplify
  because they can simply run it after changing a loop. Because
  subsequence passes can assume LoopSimplify is preserved we can reduce
  the runs of this pass to the times when we actually mutate a loop
  structure.
- The new pass manager should be able to more easily support loop passes
  factored in this way.
- We can at long, long last observe that LoopSimplify is preserved
  across SCEV. This *halves* the number of times we run LoopSimplify!!!

Now, getting here wasn't trivial. First off, the interfaces used by
LoopSimplify are all over the map regarding how analysis are updated. We
end up with weird "pass" parameters as a consequence. I'll try to clean
at least some of this up later -- I'll have to have it all clean for the
new pass manager.

Next up I discovered a really frustrating bug. LoopUnroll *claims* to
preserve LoopSimplify. That's actually a lie. But the way the
LoopPassManager ends up running the passes, it always ran LoopSimplify
on the unrolled-into loop, rectifying this oversight before any
verification could kick in and point out that in fact nothing was
preserved. So I've added code to the unroller to *actually* simplify the
surrounding loop when it succeeds at unrolling.

The only functional change in the test suite is that we now catch a case
that was previously missed because SCEV and other loop transforms see
their containing loops as simplified and thus don't miss some
opportunities. One test case has been converted to check that we catch
this case rather than checking that we miss it but at least don't get
the wrong answer.

Note that I have #if-ed out all of the verification logic in
LoopSimplify! This is a temporary workaround while extracting these bits
from the LoopPassManager. Currently, there is no way to have a pass in
the LoopPassManager which preserves LoopSimplify along with one which
does not. The LPM will try to verify on each loop in the nest that
LoopSimplify holds but the now-Function-pass cannot distinguish what
loop is being verified and so must try to verify all of them. The inner
most loop is clearly no longer simplified as there is a pass which
didn't even *attempt* to preserve it. =/ Once I get LCSSA out (and maybe
LoopVectorize and some other fixes) I'll be able to re-enable this check
and catch any places where we are still failing to preserve
LoopSimplify. If this causes problems I can back this out and try to
commit *all* of this at once, but so far this seems to work and allow
much more incremental progress.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199884 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoProspective Makefile build fix
Alp Toker [Thu, 23 Jan 2014 11:14:00 +0000 (11:14 +0000)]
Prospective Makefile build fix

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199882 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRefactor lli-child-target to remove duplicated code
Alp Toker [Thu, 23 Jan 2014 11:04:42 +0000 (11:04 +0000)]
Refactor lli-child-target to remove duplicated code

Eliminate the copies LLVM's System mmap and cache invalidation code. These were
slowly drifting away from the original version, and moreover the copied code
was a dead end in terms of portability.

We now statically link to Support but in practice with stripping this adds next
to no weight to the resultant binary.

Also avoid installing lli-child-target to the user's $PATH. It's not meant to
be run directly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199881 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIStore into II_S[BHWD], II_S[WD][LR], and II_SAVE
Daniel Sanders [Thu, 23 Jan 2014 10:31:31 +0000 (10:31 +0000)]
[mips][sched] Split IIStore into II_S[BHWD], II_S[WD][LR], and II_SAVE

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199876 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a variable to track whether or not we've used a unique section,
Eric Christopher [Thu, 23 Jan 2014 06:47:25 +0000 (06:47 +0000)]
Add a variable to track whether or not we've used a unique section,
e.g. linkonce, to TargetMachine and set it when we've done so
for ELF targets currently. This involved making TargetMachine
non-const in a TLOF use and propagating that change around - I'm
open to other ideas.

This will be used in a future commit to handle emitting debug
information with ranges.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199871 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[AArch64]Add CHECK for two test cases testing scalar_to_vector committed in r199461.
Hao Liu [Thu, 23 Jan 2014 02:09:30 +0000 (02:09 +0000)]
[AArch64]Add CHECK for two test cases testing scalar_to_vector committed in r199461.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199861 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agofix some spell mistakes around 'ConcatVector' and 'ShuffleVector' in AArch64 backend.
Kevin Qin [Thu, 23 Jan 2014 01:35:13 +0000 (01:35 +0000)]
fix some spell mistakes around 'ConcatVector' and 'ShuffleVector' in AArch64 backend.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199858 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoX86Disassembler.cpp: Fix @param introduced in r199804. [-Wdocumentation]
NAKAMURA Takumi [Thu, 23 Jan 2014 00:37:25 +0000 (00:37 +0000)]
X86Disassembler.cpp: Fix @param introduced in r199804. [-Wdocumentation]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199855 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Mips] formatting through clang-format
Jack Carter [Wed, 22 Jan 2014 23:31:38 +0000 (23:31 +0000)]
[Mips] formatting through clang-format

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199853 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Mips] TargetStreamer Support for .set mips16.
Jack Carter [Wed, 22 Jan 2014 23:08:42 +0000 (23:08 +0000)]
[Mips] TargetStreamer Support for .set mips16.

This patch updates .set mips16 support which
affects the ELF ABI and its flags. In addition the patch uses
a common interface for both the MipsTargetSteamer and
MipsObjectStreamer that the assembler uses for
both ELF and ASCII output for these directives.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199851 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert r162101 and replace it with a solution that works for targets where the pointe...
Owen Anderson [Wed, 22 Jan 2014 22:34:17 +0000 (22:34 +0000)]
Revert r162101 and replace it with a solution that works for targets where the pointer type is illegal.
This is a horrible bit of code.  We're calling a simplification routine *in the middle* of type legalization.  We tell the
simplification routine that it's running after legalization, but some of the types it will encounter will be illegal!  The
fix is only to invoke the simplification if the types in question were legal, so that none of its invariants will be violated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199847 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd CHECK-LABELs
Matt Arsenault [Wed, 22 Jan 2014 22:32:58 +0000 (22:32 +0000)]
Add CHECK-LABELs

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199846 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert "R600: Add work-around for the CF stack entry HW bug"
Tom Stellard [Wed, 22 Jan 2014 22:20:54 +0000 (22:20 +0000)]
Revert "R600: Add work-around for the CF stack entry HW bug"

This reverts commit 35b8331cad6eb512a2506adbc394201181da94ba.

The -debug-only flag for llc doesn't appear to be available in
all build configurations.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199845 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoTweak r199835 to use can_execute() instead of exists()
Alp Toker [Wed, 22 Jan 2014 22:17:51 +0000 (22:17 +0000)]
Tweak r199835 to use can_execute() instead of exists()

The execution code path crashes if it can't execute the binary so we might as
well take precautions here.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199844 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoProvide a dummy section to fix a crash with inline assembly in LTO.
Rafael Espindola [Wed, 22 Jan 2014 22:11:14 +0000 (22:11 +0000)]
Provide a dummy section to fix a crash with inline assembly in LTO.

Fixes pr18508.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199843 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Add work-around for the CF stack entry HW bug
Tom Stellard [Wed, 22 Jan 2014 21:55:46 +0000 (21:55 +0000)]
R600: Add work-around for the CF stack entry HW bug

The CF stack can be corrupted if you use CF_ALU_PUSH_BEFORE,
CF_ALU_ELSE_AFTER, CF_ALU_BREAK, or CF_ALU_CONTINUE when the number of
sub-entries on the stack is greater than or equal to the stack entry
size and sub-entries modulo 4 is either 0 or 3 (on cedar the bug is
present when number of sub-entries module 8 is either 7 or 0)

We choose to be conservative and always apply the work-around when the
number of sub-enries is greater than or equal to the stack entry size,
so that we can safely over-allocate the stack when we are unsure of the
stack allocation rules.

reviewed-by: Vincent Lejeune <vljn at ovi.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199842 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Add some missing CF instruction definitions to the .td files.
Tom Stellard [Wed, 22 Jan 2014 21:55:44 +0000 (21:55 +0000)]
R600: Add some missing CF instruction definitions to the .td files.

reviewed-by: Vincent Lejeune <vljn at ovi.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199841 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Refactor stack size calculation
Tom Stellard [Wed, 22 Jan 2014 21:55:43 +0000 (21:55 +0000)]
R600: Refactor stack size calculation

reviewed-by: Vincent Lejeune <vljn at ovi.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199840 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: CF_PUSH is the same on Evergreen and Cayman
Tom Stellard [Wed, 22 Jan 2014 21:55:41 +0000 (21:55 +0000)]
R600: CF_PUSH is the same on Evergreen and Cayman

reviewed-by: Vincent Lejeune <vljn at ovi.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199839 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Add wavefront size property to the subtargets v2
Tom Stellard [Wed, 22 Jan 2014 21:55:40 +0000 (21:55 +0000)]
R600: Add wavefront size property to the subtargets v2

v2:
  - Initialize wavefront size to 0

reviewed-by: Vincent Lejeune <vljn at ovi.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199838 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Add stack size to .AMDGPUcsdata section
Tom Stellard [Wed, 22 Jan 2014 21:55:35 +0000 (21:55 +0000)]
R600: Add stack size to .AMDGPUcsdata section

reviewed-by: Vincent Lejeune <vljn at ovi.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199837 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoHandle an addrspacecast case in memcpyopt
Matt Arsenault [Wed, 22 Jan 2014 21:53:19 +0000 (21:53 +0000)]
Handle an addrspacecast case in memcpyopt

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199836 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoEliminate inappropriate use of FindProgramByName() from lli
Alp Toker [Wed, 22 Jan 2014 21:52:35 +0000 (21:52 +0000)]
Eliminate inappropriate use of FindProgramByName() from lli

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199835 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoGet right cost for addrspacecast in cost model
Matt Arsenault [Wed, 22 Jan 2014 20:30:16 +0000 (20:30 +0000)]
Get right cost for addrspacecast in cost model

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199833 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix pr18515.
Rafael Espindola [Wed, 22 Jan 2014 20:20:52 +0000 (20:20 +0000)]
Fix pr18515.

My understanding (from reading just the llvm code) is that
* most ppc cpus have a "sync n" instruction and an msync alias that is "sync 0".
* "book e" cpus instead have a msync instruction and not the more
general "sync n"

This patch reflects that in the .td files, allowing a single codepath for
asm ond obj streamer and incidentelly fixes a crash when EmitRawText was
called on a obj streamer.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199832 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a testcase for r199430.
Quentin Colombet [Wed, 22 Jan 2014 20:11:50 +0000 (20:11 +0000)]
Add a testcase for r199430.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199831 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: MOVA is vector only
Tom Stellard [Wed, 22 Jan 2014 19:24:24 +0000 (19:24 +0000)]
R600: MOVA is vector only

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199827 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Take alignment into account when calculating the stack offset
Tom Stellard [Wed, 22 Jan 2014 19:24:23 +0000 (19:24 +0000)]
R600: Take alignment into account when calculating the stack offset

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199826 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Add support for global addresses with constant initializers
Tom Stellard [Wed, 22 Jan 2014 19:24:21 +0000 (19:24 +0000)]
R600: Add support for global addresses with constant initializers

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199825 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Begin private memory at the second GPR.
Tom Stellard [Wed, 22 Jan 2014 19:24:19 +0000 (19:24 +0000)]
R600: Begin private memory at the second GPR.

This way private memory does not over-write work group information
stored in GPRs 0 and 1.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199824 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Add support for i8 and i16 private loads/stores
Tom Stellard [Wed, 22 Jan 2014 19:24:14 +0000 (19:24 +0000)]
R600/SI: Add support for i8 and i16 private loads/stores

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199823 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoBug 18228 - Fix accepting bitcasts between vectors of pointers with a
Matt Arsenault [Wed, 22 Jan 2014 19:21:33 +0000 (19:21 +0000)]
Bug 18228 - Fix accepting bitcasts between vectors of pointers with a
different number of elements.

Bitcasts were passing with vectors of pointers with different number of
elements since the number of elements was checking
SrcTy->getVectorNumElements() == SrcTy->getVectorNumElements() which
isn't helpful. The addrspacecast was also wrong, but that case at least
is caught by the verifier. Refactor bitcast and addrspacecast handling
in castIsValid to be more readable and fix this problem.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199821 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix inline assembly that switches between ARM and Thumb modes
Greg Fitzgerald [Wed, 22 Jan 2014 18:32:35 +0000 (18:32 +0000)]
Fix inline assembly that switches between ARM and Thumb modes

This patch restores the ARM mode if the user's inline assembly
does not.  In the object streamer, it ensures that instructions
following the inline assembly are encoded correctly and that
correct mapping symbols are emitted.  For the asm streamer, it
emits a .arm or .thumb directive.

This patch does not ensure that the inline assembly contains
the ADR instruction to switch modes at runtime.

The problem we need to solve is code like this:

  int foo(int a, int b) {
    int r = a + b;
    asm volatile(
        ".align 2     \n"
        ".arm         \n"
        "add r0,r0,r0 \n"
    : : "r"(r));
    return r+1;
  }

If we compile this function in thumb mode then the inline assembly
will switch to arm mode. We need to make sure that we switch back to
thumb mode after emitting the inline assembly or we will incorrectly
encode the instructions that follow (i.e. the assembly instructions
for return r+1).

Based on patch by David Peixotto

Change-Id: Ib57f6d2d78a22afad5de8693fba6230ff56ba48b

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199818 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDon't open or fstat files twice in llvm-ar.
Rafael Espindola [Wed, 22 Jan 2014 16:43:45 +0000 (16:43 +0000)]
Don't open or fstat files twice in llvm-ar.

We still read/mmap them twice, but the fix for that is a bit more complex.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199815 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove param doxygen comment for non-existing parameter.
Benjamin Kramer [Wed, 22 Jan 2014 16:22:17 +0000 (16:22 +0000)]
Remove param doxygen comment for non-existing parameter.

Found by -Wdocumentation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199814 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoPass the computed magic to createBinary and createObjectFile if available.
Rafael Espindola [Wed, 22 Jan 2014 16:04:52 +0000 (16:04 +0000)]
Pass the computed magic to createBinary and createObjectFile if available.

identify_magic is not free, so we should avoid calling it twice. The argument
also makes it cheap for createBinary to just forward to createObjectFile.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199813 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Silence unused diReg variable warning in non-asserting builds
David Woodhouse [Wed, 22 Jan 2014 15:31:32 +0000 (15:31 +0000)]
[x86] Silence unused diReg variable warning in non-asserting builds

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199812 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Fix uninitialized variable warning in translate{Src,Dst}Index
David Woodhouse [Wed, 22 Jan 2014 15:31:29 +0000 (15:31 +0000)]
[x86] Fix uninitialized variable warning in translate{Src,Dst}Index

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199811 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Remove now-unused isSrcOp() and isDstOp() from X86AsmParser
David Woodhouse [Wed, 22 Jan 2014 15:08:58 +0000 (15:08 +0000)]
[x86] Remove now-unused isSrcOp() and isDstOp() from X86AsmParser

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199810 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Allow segment and address-size overrides for INS[BWLQ] (PR9385)
David Woodhouse [Wed, 22 Jan 2014 15:08:55 +0000 (15:08 +0000)]
[x86] Allow segment and address-size overrides for INS[BWLQ] (PR9385)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199809 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Allow segment and address-size overrides for OUTS[BWLQ] (PR9385)
David Woodhouse [Wed, 22 Jan 2014 15:08:49 +0000 (15:08 +0000)]
[x86] Allow segment and address-size overrides for OUTS[BWLQ] (PR9385)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199808 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Allow segment and address-size overrides for MOVS[BWLQ] (PR9385)
David Woodhouse [Wed, 22 Jan 2014 15:08:42 +0000 (15:08 +0000)]
[x86] Allow segment and address-size overrides for MOVS[BWLQ] (PR9385)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199807 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago]x86] Allow segment and address-size overrides for CMPS[BWLQ] (PR9385)
David Woodhouse [Wed, 22 Jan 2014 15:08:36 +0000 (15:08 +0000)]
]x86] Allow segment and address-size overrides for CMPS[BWLQ] (PR9385)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199806 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Allow address-size overrides for SCAS{8,16,32,64} (PR9385)
David Woodhouse [Wed, 22 Jan 2014 15:08:27 +0000 (15:08 +0000)]
[x86] Allow address-size overrides for SCAS{8,16,32,64} (PR9385)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199805 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Allow address-size overrides for STOS[BWLQ] (PR9385)
David Woodhouse [Wed, 22 Jan 2014 15:08:21 +0000 (15:08 +0000)]
[x86] Allow address-size overrides for STOS[BWLQ] (PR9385)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199804 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Allow segment and address-size overrides for LODS[BWLQ] (PR9385)
David Woodhouse [Wed, 22 Jan 2014 15:08:08 +0000 (15:08 +0000)]
[x86] Allow segment and address-size overrides for LODS[BWLQ] (PR9385)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199803 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoLoop strength reduce: fix function name.
Tim Northover [Wed, 22 Jan 2014 13:27:00 +0000 (13:27 +0000)]
Loop strength reduce: fix function name.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199801 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAVX512: combining setcc and zext is wrong on AVX512
Elena Demikhovsky [Wed, 22 Jan 2014 12:26:19 +0000 (12:26 +0000)]
AVX512: combining setcc and zext is wrong on AVX512
because vector compare instruction puts result in mask register.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199798 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMachineCopyPropagation has special logic for removing COPY instructions. It will...
James Molloy [Wed, 22 Jan 2014 09:12:27 +0000 (09:12 +0000)]
MachineCopyPropagation has special logic for removing COPY instructions. It will remove plain COPYs using eraseFromParent(), but if the COPY has imp-defs/imp-uses it will convert it to a KILL, to keep the imp-def around.

This actually totally breaks and causes the machine verifier to cry in several cases, one of which being:

%RAX<def> = COPY %RCX<kill>
%ECX<def> = COPY %EAX<kill>, %RAX<imp-use,kill>

These subregister copies are together identified as noops, so are both removed. However, the second one as it has an imp-use gets converted into a kill:

%ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill>

As the original COPY has been removed, the verifier goes into tears at the use of undefined EAX and RAX.

There are several hacky solutions to this hacky problem (which is all to do with imp-use/def weirdnesses), but the least hacky I've come up with is to *always* remove COPYs by converting to KILLs. KILLs are no-ops to the code generator so the generated code doesn't change (which is why they were partially used in the first place), but using them also keeps the def/use and imp-def/imp-use chains alive:

%RAX<def> = KILL %RCX<kill>
%ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill>

The patch passes all test cases including the ones that check the removal of MOVs in this circumstance, along with an extra test I added to check subregister behaviour (which made the machine verifier fall over before my patch).

The patch also adds some DEBUG() statements because the file hadn't got any.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199797 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd unused result attr to the casting templates
Alp Toker [Wed, 22 Jan 2014 07:28:49 +0000 (07:28 +0000)]
Add unused result attr to the casting templates

This helped catch a couple of bugs locally.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199793 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[AArch64 NEON] Try to generate CONCAT_VECTOR when lowering BUILD_VECTOR or SHUFFLE_VE...
Kevin Qin [Wed, 22 Jan 2014 06:11:03 +0000 (06:11 +0000)]
[AArch64 NEON] Try to generate CONCAT_VECTOR when lowering BUILD_VECTOR or SHUFFLE_VECTOR.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199791 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoReformat a loop for basic hygeine. Self review.
Andrew Trick [Wed, 22 Jan 2014 03:38:55 +0000 (03:38 +0000)]
Reformat a loop for basic hygeine. Self review.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199788 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Sparc] Add support for inline assembly constraints which specify registers by their...
Venkatraman Govindaraju [Wed, 22 Jan 2014 03:18:42 +0000 (03:18 +0000)]
[Sparc] Add support for inline assembly constraints which specify registers by their aliases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199786 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoWhitespace.
NAKAMURA Takumi [Wed, 22 Jan 2014 03:12:43 +0000 (03:12 +0000)]
Whitespace.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199785 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix typo
Matt Arsenault [Wed, 22 Jan 2014 02:38:23 +0000 (02:38 +0000)]
Fix typo

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199784 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Sparc] Add support for inline assembly constraint 'I'.
Venkatraman Govindaraju [Wed, 22 Jan 2014 01:29:51 +0000 (01:29 +0000)]
[Sparc] Add support for inline assembly constraint 'I'.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199781 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoChange createObjectFile to return an ErrorOr.
Rafael Espindola [Wed, 22 Jan 2014 00:14:49 +0000 (00:14 +0000)]
Change createObjectFile to return an ErrorOr.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199776 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Sparc] Do not add PC to _GLOBAL_OFFSET_TABLE_ address to access GOT in absolute...
Venkatraman Govindaraju [Wed, 22 Jan 2014 00:13:18 +0000 (00:13 +0000)]
[Sparc] Do not add PC to _GLOBAL_OFFSET_TABLE_ address to access GOT in absolute code.
Fixes PR#18521

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199775 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[SROA] Fix a bug which could cause the common type finding to return
Chandler Carruth [Tue, 21 Jan 2014 23:16:05 +0000 (23:16 +0000)]
[SROA] Fix a bug which could cause the common type finding to return
inconsistent results for different orderings of alloca slices. The
fundamental issue is that it is just always a mistake to return early
from this function. There is no effective early exit to leverage. This
patch stops trynig to do so and simplifies the code a bit as
a consequence.

Original diagnosis and patch by James Molloy with some name tweaks by me
in part reflecting feedback from Duncan Smith on the mailing list.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199771 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoBe a bit more consistent about using ErrorOr when constructing Binary objects.
Rafael Espindola [Tue, 21 Jan 2014 23:06:54 +0000 (23:06 +0000)]
Be a bit more consistent about using ErrorOr when constructing Binary objects.

The constructors of classes deriving from Binary normally take an error_code
as an argument to the constructor. My original intent was to change them
to have a trivial constructor and move the initial parsing logic to a static
method returning an ErrorOr. I changed my mind because:

* A constructor with an error_code out parameter is extremely convenient from
  the implementation side. We can incrementally construct the object and give
  up when we find an error.
* It is very efficient when constructing on the stack or when there is no
  error. The only inefficient case is where heap allocating and an error is
  found (we have to free the memory).

The result is that this is a much smaller patch. It just standardizes the
create* helpers to return an ErrorOr.

Almost no functionality change: The only difference is that this found that
we were trying to read past the end of COFF import library but ignoring the
error.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199770 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoCodeGen: Stop treating vectors as aggregates
Duncan P. N. Exon Smith [Tue, 21 Jan 2014 22:46:46 +0000 (22:46 +0000)]
CodeGen: Stop treating vectors as aggregates

Fix a crash in SjLjEHPrepare::lowerIncomingArguments caused by treating
VectorType like an aggregate.  It's first-class!

<rdar://problem/15854596>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199768 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoTweak the spelling of the asserts requirement a bit more. This makes it
Chandler Carruth [Tue, 21 Jan 2014 22:39:19 +0000 (22:39 +0000)]
Tweak the spelling of the asserts requirement a bit more. This makes it
match the (reasonably prevelant) usage in Clang's test suite and so
seems more "canonical".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199767 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix PR18572 - llc crash during GenericScheduler::initPolicy().
Andrew Trick [Tue, 21 Jan 2014 21:27:37 +0000 (21:27 +0000)]
Fix PR18572 - llc crash during GenericScheduler::initPolicy().

Generalized the heuristic that looks at the (very rough) size of the
register file before enabling regpressure tracking.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199766 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoForgot to add testcase for r198590
David Majnemer [Tue, 21 Jan 2014 20:39:11 +0000 (20:39 +0000)]
Forgot to add testcase for r198590

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199765 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix pointer info on PPC byval stores
Hal Finkel [Tue, 21 Jan 2014 20:15:58 +0000 (20:15 +0000)]
Fix pointer info on PPC byval stores

For PPC64 SVR (and Darwin), the stores that take byval aggregate parameters
from registers into the stack frame had MachinePointerInfo objects with
incorrect offsets. These offsets are relative to the object itself, not to the
stack frame base.

This fixes self hosting on PPC64 when compiling with -enable-aa-sched-mi.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199763 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdding new LTO APIs to parse metadata nodes and extract linker options and
Yunzhong Gao [Tue, 21 Jan 2014 18:31:27 +0000 (18:31 +0000)]
Adding new LTO APIs to parse metadata nodes and extract linker options and
dependent libraries from a bitcode module.

Differential Revision: http://llvm-reviews.chandlerc.com/D2343

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199759 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDon't clobber CMAKE_REQUIRED_FLAGS, it ends up being used in
Chandler Carruth [Tue, 21 Jan 2014 18:09:19 +0000 (18:09 +0000)]
Don't clobber CMAKE_REQUIRED_FLAGS, it ends up being used in
C compilations as well and these flags don't make any sense there.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199756 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix VS2012 ID/version check.
Amara Emerson [Tue, 21 Jan 2014 16:41:07 +0000 (16:41 +0000)]
Fix VS2012 ID/version check.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199753 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRename these methods to match the style guide.
Rafael Espindola [Tue, 21 Jan 2014 16:09:45 +0000 (16:09 +0000)]
Rename these methods to match the style guide.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199751 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IILoad into II_L[BHWD], II_L[BHW]U, II_L[WD][LR], and II_RESTORE
Daniel Sanders [Tue, 21 Jan 2014 15:21:14 +0000 (15:21 +0000)]
[mips][sched] Split IILoad into II_L[BHWD], II_L[BHW]U, II_L[WD][LR], and II_RESTORE

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199749 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIFmoveC1 into II_M[FT]C1, II_M[FT]HC1, II_DM[FT]C1
Daniel Sanders [Tue, 21 Jan 2014 15:03:52 +0000 (15:03 +0000)]
[mips][sched] Split IIFmoveC1 into II_M[FT]C1, II_M[FT]HC1, II_DM[FT]C1

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199748 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIFStore into II_S[WD]C1, and II_S[WDU]XC1
Daniel Sanders [Tue, 21 Jan 2014 14:50:20 +0000 (14:50 +0000)]
[mips][sched] Split IIFStore into II_S[WD]C1, and II_S[WDU]XC1

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199747 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[NVPTX] Add missing patterns for div.approx with immediate denominator
Justin Holewinski [Tue, 21 Jan 2014 14:40:05 +0000 (14:40 +0000)]
[NVPTX] Add missing patterns for div.approx with immediate denominator

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199746 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIFLoad into II_L[WD]C1, and II_L[WDU]XC1
Daniel Sanders [Tue, 21 Jan 2014 13:59:56 +0000 (13:59 +0000)]
[mips][sched] Split IIFLoad into II_L[WD]C1, and II_L[WDU]XC1

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199743 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Removed IIFrecipFsqrtStep. No instructions use it.
Daniel Sanders [Tue, 21 Jan 2014 13:45:41 +0000 (13:45 +0000)]
[mips][sched] Removed IIFrecipFsqrtStep. No instructions use it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199742 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Renamed II_FsqrtSingle and II_FsqrtDouble to II_SQRT_S and II_SQRT_D...
Daniel Sanders [Tue, 21 Jan 2014 13:36:45 +0000 (13:36 +0000)]
[mips][sched] Renamed II_FsqrtSingle and II_FsqrtDouble to II_SQRT_S and II_SQRT_D respectively

No functional change

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199741 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Renamed II_FdivSingle and II_FdivDouble to II_DIV_S and II_DIV_D respec...
Daniel Sanders [Tue, 21 Jan 2014 13:22:08 +0000 (13:22 +0000)]
[mips][sched] Renamed II_FdivSingle and II_FdivDouble to II_DIV_S and II_DIV_D respectively

No functional change

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199738 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIFmulDouble into II_MUL_D, II_MADD_D, II_MSUB_D, II_NMADD_D...
Daniel Sanders [Tue, 21 Jan 2014 13:07:31 +0000 (13:07 +0000)]
[mips][sched] Split IIFmulDouble into II_MUL_D, II_MADD_D, II_MSUB_D, II_NMADD_D, and II_NMSUB_S

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199737 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIFmulSingle into II_MUL_S, II_MADD_S, II_MSUB_S, II_NMADD_S...
Daniel Sanders [Tue, 21 Jan 2014 12:51:44 +0000 (12:51 +0000)]
[mips][sched] Split IIFmulSingle into II_MUL_S, II_MADD_S, II_MSUB_S, II_NMADD_S, and II_NMSUB_S

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199734 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIFadd into II_ADD_[DS], II_SUB_[DS]
Daniel Sanders [Tue, 21 Jan 2014 12:38:07 +0000 (12:38 +0000)]
[mips][sched] Split IIFadd into II_ADD_[DS], II_SUB_[DS]

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199732 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIFcmp into II_C_CC_[SD]
Daniel Sanders [Tue, 21 Jan 2014 11:42:48 +0000 (11:42 +0000)]
[mips][sched] Split IIFcmp into II_C_CC_[SD]

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199728 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIFmove into II_C[FT]C1, II_MOV[FNTZ]_[SD], II_MOV_[SD]
Daniel Sanders [Tue, 21 Jan 2014 11:28:03 +0000 (11:28 +0000)]
[mips][sched] Split IIFmove into II_C[FT]C1, II_MOV[FNTZ]_[SD], II_MOV_[SD]

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199727 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIFcvt into II_(ROUND|TRUNC|CEIL|FLOOR|CVT), II_ABS, II_NEG
Daniel Sanders [Tue, 21 Jan 2014 10:56:23 +0000 (10:56 +0000)]
[mips][sched] Split IIFcvt into II_(ROUND|TRUNC|CEIL|FLOOR|CVT), II_ABS, II_NEG

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199722 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips][sched] Split IIslt into II_SLT_SLTU, II_SLTI_SLTIU
Daniel Sanders [Tue, 21 Jan 2014 10:42:13 +0000 (10:42 +0000)]
[mips][sched] Split IIslt into II_SLT_SLTU, II_SLTI_SLTIU

No functional change since the InstrItinData's have been duplicated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199719 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMIPS: mark intrinsics IntrNoMem so all patterns using them are consistent.
Tim Northover [Tue, 21 Jan 2014 10:41:16 +0000 (10:41 +0000)]
MIPS: mark intrinsics IntrNoMem so all patterns using them are consistent.

This is apparently a bit of a white lie (they can affect DSPControl for
overflow etc) but similar to how we currently handle floating-point operations.
When it becomes relevant the whole lot can be reviewed properly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199718 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoChecked return warning from coverity
Renato Golin [Tue, 21 Jan 2014 10:24:35 +0000 (10:24 +0000)]
Checked return warning from coverity

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199716 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix libstdc++4.7 test on Android.
Evgeniy Stepanov [Tue, 21 Jan 2014 09:00:30 +0000 (09:00 +0000)]
Fix libstdc++4.7 test on Android.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199714 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoUse ArrayRef to simplify some code.
Craig Topper [Tue, 21 Jan 2014 07:20:05 +0000 (07:20 +0000)]
Use ArrayRef to simplify some code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199712 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agotools: use 64-bit print specifier
Saleem Abdulrasool [Tue, 21 Jan 2014 04:31:29 +0000 (04:31 +0000)]
tools: use 64-bit print specifier

Try to repair the ARM Cortex-A15 buildbot by using a more appropriate conversion
specifier.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199711 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agotools: support decoding ARM EHABI opcodes in readobj
Saleem Abdulrasool [Tue, 21 Jan 2014 02:33:15 +0000 (02:33 +0000)]
tools: support decoding ARM EHABI opcodes in readobj

Add support to llvm-readobj to decode the actual opcodes.  The ARM EHABI opcodes
are a variable length instruction set that describe the operations required for
properly unwinding stack frames.

The primary motivation for this change is to ease the creation of tests for the
ARM EHABI object emission as well as the unwinding directive handling in the ARM
IAS.

Thanks to Logan Chien for an extra test case!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199708 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM IAS: add support for .unwind_raw directive
Saleem Abdulrasool [Tue, 21 Jan 2014 02:33:10 +0000 (02:33 +0000)]
ARM IAS: add support for .unwind_raw directive

This implements the unwind_raw directive for the ARM IAS.  The unwind_raw
directive takes the form of a stack offset value followed by one or more bytes
representing the opcodes to be emitted.  The opcode emitted will interpreted as
if it were assembled by the opcode assembler via the standard unwinding
directives.

Thanks to Logan Chien for an extra test!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199707 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM IAS: support .personalityindex
Saleem Abdulrasool [Tue, 21 Jan 2014 02:33:02 +0000 (02:33 +0000)]
ARM IAS: support .personalityindex

The .personalityindex directive is equivalent to the .personality directive with
the ARM EABI personality with the specific index (0, 1, 2).  Both of these
directives indicate personality routines, so enhance the personality directive
handling to take into account personalityindex.

Bonus fix: flush the UnwindContext at the beginning of a new function.

Thanks to Logan Chien for additional tests!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199706 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[AArch64 NEON] Fix a bug caused by undef lane when generating VEXT.
Kevin Qin [Tue, 21 Jan 2014 01:48:52 +0000 (01:48 +0000)]
[AArch64 NEON] Fix a bug caused by undef lane when generating VEXT.

It was commited as r199628 but reverted in r199628 as causing
regression test failed. It's because of old vervsion of patch
I used to commit. Sorry for mistake.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199704 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd operator!= for FoldingSetNodeID and FoldingSetNodeIDRef. Implementation in
Nick Lewycky [Tue, 21 Jan 2014 01:29:37 +0000 (01:29 +0000)]
Add operator!= for FoldingSetNodeID and FoldingSetNodeIDRef. Implementation in
the header forwards to operator== which is not in the header.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199702 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoTweak the MCExternalSymbolizer to not use the SymbolLookUp() call back
Kevin Enderby [Tue, 21 Jan 2014 00:23:17 +0000 (00:23 +0000)]
Tweak the MCExternalSymbolizer to not use the SymbolLookUp() call back
to not guess at a symbol name in some cases.

The problem is that in object files assembled starting at address 0, when
trying to symbolicate something that starts like this:

% cat x.s
_t1:
vpshufd $0x0, %xmm1, %xmm0

the symbolic disassembly can end up like this:

% otool -tV x.o
x.o:
(__TEXT,__text) section
_t1:
0000000000000000 vpshufd $_t1, %xmm1, %xmm0

Which is in this case produced incorrect symbolication.

But it is useful in some cases to use the SymbolLookUp() call back
to guess at some immediate values.  For example one like this
that does not have an external relocation entry:

% cat y.s
_t1:
movl $_d1, %eax
.data
_d1: .long 0

% clang -c -arch i386 y.s

% otool -tV y.o
y.o:
(__TEXT,__text) section
_t1:
0000000000000000 movl $_d1, %eax

% otool -rv y.o
y.o:
Relocation information (__TEXT,__text) 1 entries
address  pcrel length extern type    scattered symbolnum/value
00000001 False long   False  VANILLA False     2 (__DATA,__data)

So the change is based on it is not likely that an immediate Value
coming from an instruction field of a width of 1 byte, other than branches
and items with relocation, are not likely symbol addresses.

With the change the first case above simply becomes:

% otool -tV x.o
x.o:
(__TEXT,__text) section
_t1:
0000000000000000 vpshufd $0x0, %xmm1, %xmm0

and the second case continues to work as expected.

rdar://14863405

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199698 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoTo allow the X86 verbose assembly to print its informative comments
Kevin Enderby [Tue, 21 Jan 2014 00:18:51 +0000 (00:18 +0000)]
To allow the X86 verbose assembly to print its informative comments
when used with symbolic disassembly, add a check that the operand
is an immediate and has not been symbolicated to MCExpr operand.

I’m trying to enable the ‘C’ disassembly API option
LLVMDisassembler_Option_SetInstrComments for darwin’s
otool(1) that uses the llvm disassembler API.  The problem is
that the disassembler API can change an immediate operand to
an MCExpr operand if it symbolicates it with the call backs.
And if it does the code in llvm::EmitAnyX86InstComments()
will crash when it assumes these operands are immediates.

The fix for this is very straight forward to just protect the call
to getImm() with a check of isImm().  So if the immediate for
an instruction is symbolicated it simply doesn’t get the X86
verbose assembly comments:

% otool -tV test_asm.o
test_asm.o:
(__TEXT,__text) section
_t1:
0000000000000000 vpshufd $_t1, %xmm1, %xmm0
0000000000000005 retq
0000000000000006 nopw %cs:_t1(%rax,%rax)
_t2:
0000000000000010 vpshufd $-0x1, %xmm0, %xmm0     ## xmm0 = xmm0[3,3,3,3]
0000000000000015 retq
0000000000000016 nopw %cs:_t1(%rax,%rax)
_t3:
0000000000000020 vpshufd $_t1, %xmm1, %xmm0
0000000000000025 retq
0000000000000026 nopw %cs:_t1(%rax,%rax)
_t4:
0000000000000030 vpshufd $0x2d, %xmm0, %xmm0     ## xmm0 = xmm0[1,3,2,0]
0000000000000035 retq

The fact that the immediate $0x0 is being symbolicated at
all in this case is a different problem which my next patch
will address.

rdar://10989286

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199697 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoUpdate StackProtector when coloring merges stack slots
Hal Finkel [Mon, 20 Jan 2014 19:49:14 +0000 (19:49 +0000)]
Update StackProtector when coloring merges stack slots

StackProtector keeps a ValueMap of alloca instructions to layout kind tags for
use by PEI and other later passes. When stack coloring replaces one alloca with
a bitcast to another one, the key replacement in this map does not work.
Instead, provide an interface to manage this updating directly. This seems like
an improvement over the old behavior, where the layout map would not get
updated at all when the stack slots were merged. In practice, however, there is
likely no observable difference because PEI only did anything special with
'large array' kinds, and if one large array is merged with another, than the
replacement should already have been a large array.

This is an attempt to unbreak the clang-x86_64-darwin11-RA builder.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199684 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[X86] Teach how to combine a vselect into a movss/movsd
Andrea Di Biagio [Mon, 20 Jan 2014 19:35:22 +0000 (19:35 +0000)]
[X86] Teach how to combine a vselect into a movss/movsd

Add target specific rules for combining vselect dag nodes into movss/movsd
when possible.

If the vector type of the vselect dag node in input is either MVT::v4i13 or
MVT::v4f32, then try to fold according to rules:

  1) fold (vselect (build_vector (0, -1, -1, -1)), A, B) -> (movss A, B)
  2) fold (vselect (build_vector (-1, 0, 0, 0)), A, B) -> (movss B, A)

If the vector type of the vselect dag node in input is either MVT::v2i64 or
MVT::v2f64 (and we have SSE2), then try to fold according to rules:

  3) fold (vselect (build_vector (0, -1)), A, B) -> (movsd A, B)
  4) fold (vselect (build_vector (-1, 0)), A, B) -> (movsd B, A)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199683 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebug info: On ARM ensure that all __TEXT sections come before the
Adrian Prantl [Mon, 20 Jan 2014 19:15:59 +0000 (19:15 +0000)]
Debug info: On ARM ensure that all __TEXT sections come before the
optional DWARF sections, so compiling with -g does not result in
different code being generated for PC-relative loads.

This is reapplying a diet r197922 (__TEXT-only).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199681 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert "Debug info: On ARM ensure that the data sections come before the"
Adrian Prantl [Mon, 20 Jan 2014 19:15:55 +0000 (19:15 +0000)]
Revert "Debug info: On ARM ensure that the data sections come before the"
Cut back on the cargo cult. The order of __DATA sections doesn't affect
generated code.

This reverts commit r197922.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199680 91177308-0d34-0410-b5e6-96231b3b80d8