oota-llvm.git
9 years agoR600/SI: Don't promote f32 select to i32
Matt Arsenault [Fri, 12 Dec 2014 02:30:29 +0000 (02:30 +0000)]
R600/SI: Don't promote f32 select to i32

This is nice for the instruction patterns, but it complicates
min / max matching. The select doesn't have the correct type and would
require looking through the bitcasts for the real float operands.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224092 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUpdate the modules build to match r223802.
Richard Smith [Fri, 12 Dec 2014 02:25:18 +0000 (02:25 +0000)]
Update the modules build to match r223802.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224091 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoBitcode: Add missing "Remove in 4.0" comments
Duncan P. N. Exon Smith [Fri, 12 Dec 2014 02:11:31 +0000 (02:11 +0000)]
Bitcode: Add missing "Remove in 4.0" comments

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224090 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDocument that PassManager::add() may delete the pass right away.
Matthias Braun [Fri, 12 Dec 2014 01:27:01 +0000 (01:27 +0000)]
Document that PassManager::add() may delete the pass right away.

Also remove redundant documentation:
- doxygen will copy documentation to overriden methods.
- Use \copydoc on PIMPL classes instead of replicating the text.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224089 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoTom Stellard is now the code owner for libclc.
Peter Collingbourne [Fri, 12 Dec 2014 01:11:48 +0000 (01:11 +0000)]
Tom Stellard is now the code owner for libclc.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224088 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoComment and minor code cleanup for GCStrategy (NFC)
Philip Reames [Fri, 12 Dec 2014 00:49:03 +0000 (00:49 +0000)]
Comment and minor code cleanup for GCStrategy (NFC)

Updating comments to reflect the current state of the world after my recent changes to ownership structure and generally better describe what a GCStrategy is and how it works.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224086 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd target hook for whether it is profitable to reduce load widths
Matt Arsenault [Fri, 12 Dec 2014 00:00:24 +0000 (00:00 +0000)]
Add target hook for whether it is profitable to reduce load widths

Add an option to disable optimization to shrink truncated larger type
loads to smaller type loads. On SI this prevents using scalar load
instructions in some cases, since there are no scalar extloads.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224084 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoremove function names from comments; NFC
Sanjay Patel [Thu, 11 Dec 2014 23:38:43 +0000 (23:38 +0000)]
remove function names from comments; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224080 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Handle physical registers in getOpRegClass
Matt Arsenault [Thu, 11 Dec 2014 23:37:34 +0000 (23:37 +0000)]
R600/SI: Handle physical registers in getOpRegClass

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224079 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Don't verify constant bus usage of flag ops
Matt Arsenault [Thu, 11 Dec 2014 23:37:32 +0000 (23:37 +0000)]
R600/SI: Don't verify constant bus usage of flag ops

This was checking if pseudo-operands like the source
modifiers were using the constant bus, which happens to work
because the values these all can be happen to be valid inline
immediates.

This fixes a later commit which starts checking the register class
of the operands.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224078 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoBitcode: Use unsigned char to record MDStrings
Duncan P. N. Exon Smith [Thu, 11 Dec 2014 23:34:30 +0000 (23:34 +0000)]
Bitcode: Use unsigned char to record MDStrings

`MDString`s can have arbitrary characters in them.  Prevent an assertion
that fired in `BitcodeWriter` because of sign extension by copying the
characters into the record as `unsigned char`s.

Based on a patch by Keno Fischer; fixes PR21882.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224077 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoreturn without temporary; NFC
Sanjay Patel [Thu, 11 Dec 2014 23:30:36 +0000 (23:30 +0000)]
return without temporary; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224076 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoEnable MachineVerifier in debug mode for X86, ARM, AArch64, Mips.
Matthias Braun [Thu, 11 Dec 2014 23:18:03 +0000 (23:18 +0000)]
Enable MachineVerifier in debug mode for X86, ARM, AArch64, Mips.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224075 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] Add a temporary testcase for PR21876/r223996.
Ahmed Bougacha [Thu, 11 Dec 2014 23:07:52 +0000 (23:07 +0000)]
[X86] Add a temporary testcase for PR21876/r223996.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224074 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoBitcode: Add METADATA_NODE and METADATA_VALUE
Duncan P. N. Exon Smith [Thu, 11 Dec 2014 23:02:24 +0000 (23:02 +0000)]
Bitcode: Add METADATA_NODE and METADATA_VALUE

This reflects the typelessness of `Metadata` in the bitcode format,
removing types from all metadata operands.

`METADATA_VALUE` represents a `ValueAsMetadata`, and always has two
fields: the type and the value.

`METADATA_NODE` represents an `MDNode`, and unlike `METADATA_OLD_NODE`,
doesn't store types.  It stores operands at their ID+1 so that `0` can
reference `nullptr` operands.

Part of PR21532.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224073 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PowerPC] Better lowering for add/or of a FrameIndex
Hal Finkel [Thu, 11 Dec 2014 22:51:06 +0000 (22:51 +0000)]
[PowerPC] Better lowering for add/or of a FrameIndex

If we have an add (or an or that is really an add), where one operand is a
FrameIndex and the other operand is a small constant, we can combine the
lowering of the FrameIndex (which is lowered as an add of the FI and a zero
offset) with the constant operand.

Amusingly, this is an old potential improvement entry from
lib/Target/PowerPC/README.txt which had never been resolved. In short, we used
to lower:

        %X = alloca { i32, i32 }
        %Y = getelementptr {i32,i32}* %X, i32 0, i32 1
        ret i32* %Y

as:

        addi 3, 1, -8
        ori 3, 3, 4
        blr

and now we produce:

        addi 3, 1, -4
        blr

which is much more sensible.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224071 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoBitcode: Add `OLD_` prefix to metadata node records
Duncan P. N. Exon Smith [Thu, 11 Dec 2014 22:30:48 +0000 (22:30 +0000)]
Bitcode: Add `OLD_` prefix to metadata node records

I'm about to change these, so move the old ones out of the way.

Part of PR21532.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224070 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoTry to include operand counts for mismatches in tablegen.
Matt Arsenault [Thu, 11 Dec 2014 22:27:14 +0000 (22:27 +0000)]
Try to include operand counts for mismatches in tablegen.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224068 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Use unordered equal instructions
Matt Arsenault [Thu, 11 Dec 2014 22:15:43 +0000 (22:15 +0000)]
R600/SI: Use unordered equal instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224067 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Make more unordered comparisons legal
Matt Arsenault [Thu, 11 Dec 2014 22:15:39 +0000 (22:15 +0000)]
R600/SI: Make more unordered comparisons legal

This saves a second compare and an and / or by using
the unordered comparison instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224066 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Use unordered not equal instructions
Matt Arsenault [Thu, 11 Dec 2014 22:15:35 +0000 (22:15 +0000)]
R600/SI: Use unordered not equal instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224065 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[ASan] Change fake stack and local variables handling.
Alexey Samsonov [Thu, 11 Dec 2014 21:53:03 +0000 (21:53 +0000)]
[ASan] Change fake stack and local variables handling.

This commit changes the way we get fake stack from ASan runtime
(to find use-after-return errors) and the way we represent local
variables:
  - __asan_stack_malloc function now returns pointer to newly allocated
    fake stack frame, or NULL if frame cannot be allocated. It doesn't
    take pointer to real stack as an input argument, it is calculated
    inside the runtime.
  - __asan_stack_free function doesn't take pointer to real stack as
    an input argument. Now this function is never called if fake stack
    frame wasn't allocated.
  - __asan_init version is bumped to reflect changes in the ABI.
  - new flag "-asan-stack-dynamic-alloca" allows to store all the
    function local variables in a dynamic alloca, instead of the static
    one. It reduces the stack space usage in use-after-return mode
    (dynamic alloca will not be called if the local variables are stored
    in a fake stack), and improves the debug info quality for local
    variables (they will not be described relatively to %rbp/%rsp, which
    are assumed to be clobbered by function calls). This flag is turned
    off by default for now, but I plan to turn it on after more
    testing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224062 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoCodeGen: Stop using LeakDetector for MachineInstr
Duncan P. N. Exon Smith [Thu, 11 Dec 2014 21:51:37 +0000 (21:51 +0000)]
CodeGen: Stop using LeakDetector for MachineInstr

Since `MachineInstr` is required to have a trivial destructor, it cannot
remove itself from `LeakDetection`.  Remove the calls.

As it happens, this requirement is because `MachineFunction` allocates
all `MachineInstr`s in a custom allocator; when the `MachineFunction` is
destroyed they're dropped of the edge.  There's no benefit to detecting
leaks.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224061 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoIR: Store MDNodes in a separate LeakDetector container
Duncan P. N. Exon Smith [Thu, 11 Dec 2014 21:39:39 +0000 (21:39 +0000)]
IR: Store MDNodes in a separate LeakDetector container

This gives us better leak detection messages, like `Value` has.

This also has the side effect of papering over a problem where
`MachineInstr`s are added as garbage to the leak detector and then
deleted without being removed.  If `MDNode::getTemporary()` allocates an
`MDNodeFwdDecl` in the same spot, the leak detector asserts.  By
separating `MDNode`s into their own container we lose that assertion.

Since `MachineInstr` is required to have a trivial destructor, its usage
of `LeakDetector` at all is pretty suspect.  I'll be sending a patch
soon to strip that out.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224060 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[CodeGen] Add print and verify pass after each MachineFunctionPass by default
Matthias Braun [Thu, 11 Dec 2014 21:26:47 +0000 (21:26 +0000)]
[CodeGen] Add print and verify pass after each MachineFunctionPass by default

Previously print+verify passes were added in a very unsystematic way, which is
annoying when debugging as you miss intermediate steps and allows bugs to stay
unnotice when no verification is performed.

To make this change practical I added the possibility to explicitely disable
verification. I used this option on all places where no verification was
performed previously (because alot of places actually don't pass the
MachineVerifier).
In the long term these problems should be fixed properly and verification
enabled after each pass. I'll enable some more verification in subsequent
commits.

This is the 2nd attempt at this after realizing that PassManager::add() may
actually delete the pass.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224059 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoLeakDetector: Simplify code and fix comments, NFC
Duncan P. N. Exon Smith [Thu, 11 Dec 2014 21:23:43 +0000 (21:23 +0000)]
LeakDetector: Simplify code and fix comments, NFC

Rather than requiring overloads in the wrapper and the impl, just
overload the impl and use templates in the wrapper.  This makes it less
error prone to add more overloads (`void *` defeats any chance the
compiler has at noticing bugs, so the easier the better).

At the same time, correct the comment that was lying about not changing
functionality for `Value`.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224058 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAsmParser: Don't crash on an ill-formed MDNodeVector
David Majnemer [Thu, 11 Dec 2014 20:51:54 +0000 (20:51 +0000)]
AsmParser: Don't crash on an ill-formed MDNodeVector

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224056 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[InstCombine][X86] Improved folding of calls to Intrinsic::x86_sse4a_insertqi.
Andrea Di Biagio [Thu, 11 Dec 2014 20:44:59 +0000 (20:44 +0000)]
[InstCombine][X86] Improved folding of calls to Intrinsic::x86_sse4a_insertqi.

This patch teaches the instruction combiner how to fold a call to 'insertqi' if
the 'length field' (3rd operand) is set to zero, and if the sum between
field 'length' and 'bit index' (4th operand) is bigger than 64.

From the AMD64 Architecture Programmer's Manual:
1. If the sum of the bit index + length field is greater than 64, then the
   results are undefined;
2. A value of zero in the field length is defined as a length of 64.

This patch improves the existing combining logic for intrinsic 'insertqi'
adding extra checks to address both point 1. and point 2.

Differential Revision: http://reviews.llvm.org/D6583

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224054 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAsmParser: Don't crash on an ill-formed MDNodeVector
David Majnemer [Thu, 11 Dec 2014 20:44:09 +0000 (20:44 +0000)]
AsmParser: Don't crash on an ill-formed MDNodeVector

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224053 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRemove a convoluted way of calling close by moving the call to the only caller.
Rafael Espindola [Thu, 11 Dec 2014 20:12:55 +0000 (20:12 +0000)]
Remove a convoluted way of calling close by moving the call to the only caller.

As a bonus we can actually check the return value.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224046 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoThis reverts commit r224043 and r224042.
Rafael Espindola [Thu, 11 Dec 2014 20:03:57 +0000 (20:03 +0000)]
This reverts commit r224043 and r224042.

check-llvm was failing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224045 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoSilence static analyzer warnings in LLVMSupport.
Michael Ilseman [Thu, 11 Dec 2014 19:46:38 +0000 (19:46 +0000)]
Silence static analyzer warnings in LLVMSupport.

The static analyzer catches a few potential bugs in LLVMSupport. Add
in asserts to silence the warnings.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224044 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoEnable machineverifier in debug mode for X86, ARM, AArch64, Mips
Matthias Braun [Thu, 11 Dec 2014 19:42:09 +0000 (19:42 +0000)]
Enable machineverifier in debug mode for X86, ARM, AArch64, Mips

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224043 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[CodeGen] Add print and verify pass after each MachineFunctionPass by default
Matthias Braun [Thu, 11 Dec 2014 19:42:05 +0000 (19:42 +0000)]
[CodeGen] Add print and verify pass after each MachineFunctionPass by default

Previously print+verify passes were added in a very unsystematic way, which is
annoying when debugging as you miss intermediate steps and allows bugs to stay
unnotice when no verification is performed.

To make this change practical I added the possibility to explicitely disable
verification. I used this option on all places where no verification was
performed previously (because alot of places actually don't pass the
MachineVerifier).
In the long term these problems should be fixed properly and verification
enabled after each pass. I'll enable some more verification in subsequent
commits.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224042 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[CodeGen] Let MachineVerifierPass own its banner string
Matthias Braun [Thu, 11 Dec 2014 19:41:51 +0000 (19:41 +0000)]
[CodeGen] Let MachineVerifierPass own its banner string

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224041 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Renaming classes in preparation for replacement.
Colin LeMahieu [Thu, 11 Dec 2014 19:01:28 +0000 (19:01 +0000)]
[Hexagon] Renaming classes in preparation for replacement.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224036 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoARM: convert isTargetIOS checks to isTargetDarwin.
Tim Northover [Thu, 11 Dec 2014 18:49:37 +0000 (18:49 +0000)]
ARM: convert isTargetIOS checks to isTargetDarwin.

The distinction is mostly useful in the front-end. By the time we get here,
there are very few situations where we actually want different behaviour for
Darwin and IOS (in fact Darwin mostly just exists in a few tests). So this
should reduce any surprising weirdness for anyone using it.

No functional change on anything anyone actually cares about.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224035 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PowerPC] Implement BuildSDIVPow2, lower i64 pow2 sdiv using sradi
Hal Finkel [Thu, 11 Dec 2014 18:37:52 +0000 (18:37 +0000)]
[PowerPC] Implement BuildSDIVPow2, lower i64 pow2 sdiv using sradi

PPCISelDAGToDAG contained existing code to lower i32 sdiv by a power-of-2 using
srawi/addze, but did not implement the i64 case. DAGCombine now contains a
callback specifically designed for this purpose (BuildSDIVPow2), and part of
the logic has been moved to an implementation of that callback. Doing this
lowering using BuildSDIVPow2 likely does not matter, compared to handling
everything in PPCISelDAGToDAG, for the positive divisor case, but the negative
divisor case, which generates an additional negation, can potentially benefit
from additional folding from DAGCombine. Now, both the i32 and the i64 cases
have been implemented.

Fixes PR20732.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224033 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRemove dead code. NFC.
Rafael Espindola [Thu, 11 Dec 2014 17:17:26 +0000 (17:17 +0000)]
Remove dead code. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224029 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[AVX512] Add support for 512b variable bit shift intrinsics.
Cameron McInally [Thu, 11 Dec 2014 17:13:05 +0000 (17:13 +0000)]
[AVX512] Add support for 512b variable bit shift intrinsics.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224028 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Ading i64 <- i32, i32 sextw pattern.
Colin LeMahieu [Thu, 11 Dec 2014 17:08:21 +0000 (17:08 +0000)]
[Hexagon] Ading i64 <- i32, i32 sextw pattern.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224027 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Adding encoding information for sign extend word instruction.
Colin LeMahieu [Thu, 11 Dec 2014 16:43:06 +0000 (16:43 +0000)]
[Hexagon] Adding encoding information for sign extend word instruction.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224026 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoOnly seek once before writing the member offsets.
Rafael Espindola [Thu, 11 Dec 2014 16:34:00 +0000 (16:34 +0000)]
Only seek once before writing the member offsets.

This cuts down the number on system calls done by a static llvm-ar producing
lib/libclangSema.a from 9164 to 442.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224025 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agollvm-ar: close input files early.
Rafael Espindola [Thu, 11 Dec 2014 15:16:14 +0000 (15:16 +0000)]
llvm-ar: close input files early.

We already have them mapped into memory, so we can just close the file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224020 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAVX-512: Added all forms of COMPRESS instruction
Elena Demikhovsky [Thu, 11 Dec 2014 15:02:24 +0000 (15:02 +0000)]
AVX-512: Added all forms of COMPRESS instruction
+ intrinsics + tests

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224019 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[mips][microMIPS] Implement CodeGen support for LI16 instruction.
Jozef Kolek [Thu, 11 Dec 2014 13:56:23 +0000 (13:56 +0000)]
[mips][microMIPS] Implement CodeGen support for LI16 instruction.

Differential Revision: http://reviews.llvm.org/D5840

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224017 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoInstSimplify: Remove usesless %a parameter from tests
David Majnemer [Thu, 11 Dec 2014 12:56:17 +0000 (12:56 +0000)]
InstSimplify: Remove usesless %a parameter from tests

No functional change intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224016 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoThe inliner needs to fix up debug information for llvm.dbg.declare, not only for...
Michael Kuperstein [Thu, 11 Dec 2014 12:41:10 +0000 (12:41 +0000)]
The inliner needs to fix up debug information for llvm.dbg.declare, not only for llvm.dbg.value.

Patch by Amjad Aboud

Differential Revision: http://reviews.llvm.org/D6525

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224015 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd newline missing in r224010.
Michael Kuperstein [Thu, 11 Dec 2014 11:30:20 +0000 (11:30 +0000)]
Add newline missing in r224010.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224011 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] When converting movs to pushes, don't assume MOVmi operand is an actual immediate
Michael Kuperstein [Thu, 11 Dec 2014 11:26:16 +0000 (11:26 +0000)]
[X86] When converting movs to pushes, don't assume MOVmi operand is an actual immediate

This should fix PR21878.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224010 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoBugfix in InlineSpiller::traceSiblingValue().
Patrik Hagglund [Thu, 11 Dec 2014 10:40:17 +0000 (10:40 +0000)]
Bugfix in InlineSpiller::traceSiblingValue().

Properly determine whether or not a phi was added by splitting.
Check against the current VNInfo of OrigLI instead of against the
OrigVNI argument.

Patch provided by Jonas Paulsson. Reviewed by Quentin Colombet.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224009 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAVX-512: Fixed a bug in lowering setcc for MVT::i1 type
Elena Demikhovsky [Thu, 11 Dec 2014 10:21:12 +0000 (10:21 +0000)]
AVX-512: Fixed a bug in lowering setcc for MVT::i1 type

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224008 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agotest commit (spelling correction)
Kumar Sukhani [Thu, 11 Dec 2014 08:33:36 +0000 (08:33 +0000)]
test commit (spelling correction)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224007 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRefactor creation of overflow result tuples in InstCombineCalls.
Erik Eckstein [Thu, 11 Dec 2014 08:02:30 +0000 (08:02 +0000)]
Refactor creation of overflow result tuples in InstCombineCalls.

Extract the creation of overflow result tuples in a separate function. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224006 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUse range-based for loops. NFC
Craig Topper [Thu, 11 Dec 2014 07:04:54 +0000 (07:04 +0000)]
Use range-based for loops. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224005 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoJust use the Module unique_ptr object directly in many places instead of separate...
Craig Topper [Thu, 11 Dec 2014 07:04:52 +0000 (07:04 +0000)]
Just use the Module unique_ptr object directly in many places instead of separate pointer that's kept in sync with it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224004 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUse unique_ptr to remove an explicit delete. Change return type to pass the unique_pt...
Craig Topper [Thu, 11 Dec 2014 07:04:46 +0000 (07:04 +0000)]
Use unique_ptr to remove an explicit delete. Change return type to pass the unique_ptr to caller.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224003 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoIR: Canonicalize metadata formatting, NFC
Duncan P. N. Exon Smith [Thu, 11 Dec 2014 06:32:29 +0000 (06:32 +0000)]
IR: Canonicalize metadata formatting, NFC

Canonicalize formatting of metadata to make it easier to upgrade via
scripts -- in particular, one line per metadata definition makes it more
`sed`-able.

This is preparation for changing the assembly syntax for metadata [1].

[1]: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20141208/248449.html

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224002 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoReverting commit 223981, because the test that I added (incorrect-variable-debugloc1...
Ekaterina Romanova [Thu, 11 Dec 2014 06:22:35 +0000 (06:22 +0000)]
Reverting commit 223981, because the test that I added (incorrect-variable-debugloc1.ll) failed for llvm-ppc64.

The test is failing for llvm-ppc64 because for this platform the location list is not being generated at all (most likely because of the bug in PPC code optimization or generation). I will file a bug agains PPC compiler, but meanwhile, until PPC bug is fixed, I will have to revert my change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224000 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMake MultiClass::DefPrototypes own their Records to fix memory leaks.
Craig Topper [Thu, 11 Dec 2014 05:25:33 +0000 (05:25 +0000)]
Make MultiClass::DefPrototypes own their Records to fix memory leaks.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223998 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoReplace std::map<K, V*> with std::map<K, std::unique_ptr<V>> to handle ownership...
Craig Topper [Thu, 11 Dec 2014 05:25:30 +0000 (05:25 +0000)]
Replace std::map<K, V*> with std::map<K, std::unique_ptr<V>> to handle ownership and deletion of the values.

Ideally we would store the MultiClasses by value directly in the maps, but I had some trouble with that before and this at least fixes the leak.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223997 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] Add back AVX2 VR256 PMOVX patterns.
Ahmed Bougacha [Thu, 11 Dec 2014 04:32:17 +0000 (04:32 +0000)]
[X86] Add back AVX2 VR256 PMOVX patterns.

We can't reach those from zext, but other parts of the backend (the shuffle
lowering) generate 256-bit VZEXT nodes.

Fixes PR21876.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223996 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix LLVMContext to match what MDKind names that the LL parser permits. Fixes PR21799!
Nick Lewycky [Thu, 11 Dec 2014 02:10:28 +0000 (02:10 +0000)]
Fix LLVMContext to match what MDKind names that the LL parser permits. Fixes PR21799!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223995 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoGCStrategy should not own GCFunctionInfo
Philip Reames [Thu, 11 Dec 2014 01:47:23 +0000 (01:47 +0000)]
GCStrategy should not own GCFunctionInfo

This change moves the ownership and access of GCFunctionInfo (the object which describes the safepoints associated with a safepoint under GCRoot) to GCModuleInfo. Previously, this was owned by GCStrategy which was in turned owned by GCModuleInfo. This made GCStrategy module specific which is 'surprising' given it's name and other purposes.

There's a few more changes needed, but we're getting towards the point we can reuse GCStrategy for gc.statepoint as well.

p.s. The style of this code ends up being a mess. I was trying to move code around without otherwise changing much. Once I get the ownership structure rearranged, I will go through and fixup spacing, naming, comments etc.

Differential Revision: http://reviews.llvm.org/D6587

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223994 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoIR: Add 'invalid-' to test names for invalid assembly
Duncan P. N. Exon Smith [Thu, 11 Dec 2014 01:34:46 +0000 (01:34 +0000)]
IR: Add 'invalid-' to test names for invalid assembly

Take the opportunity to sort these by `metadata`.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223993 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoLiveInterval: Use range based for loops for subregister ranges.
Matthias Braun [Thu, 11 Dec 2014 00:59:06 +0000 (00:59 +0000)]
LiveInterval: Use range based for loops for subregister ranges.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223991 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoTry again at sorting entries in CODE_OWNERS
Justin Bogner [Thu, 11 Dec 2014 00:52:59 +0000 (00:52 +0000)]
Try again at sorting entries in CODE_OWNERS

I apparently fail at the alphabet, E is not after G, and G isn't even
close to C. Sorry for the noise.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223990 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix some incorrectly sorted entries in CODE_OWNERS
Justin Bogner [Thu, 11 Dec 2014 00:41:15 +0000 (00:41 +0000)]
Fix some incorrectly sorted entries in CODE_OWNERS

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223989 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd code owners for profiling and coverage
Justin Bogner [Thu, 11 Dec 2014 00:37:25 +0000 (00:37 +0000)]
Add code owners for profiling and coverage

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223988 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoARM: correctly expand LDR-lit based globals.
Tim Northover [Wed, 10 Dec 2014 23:40:50 +0000 (23:40 +0000)]
ARM: correctly expand LDR-lit based globals.

Quite a major error here: the expansions for the Pseudos with and without
folded load were mixed up. Fortunately it only affects ARM-mode, when not using
movw/movt, on Darwin. I'm guessing no-one actually uses that combination.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223986 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoA fix for PR21176.
Ekaterina Romanova [Wed, 10 Dec 2014 23:19:56 +0000 (23:19 +0000)]
A fix for PR21176.

DW_OP_const <const> doesn't describe a constant value, but a value at a constant address.
The proper way to describe a constant value is DW_OP_constu <const>, DW_OP_stack_value.

Added DW_OP_stack_value to the stack.

-This line, and those below, will be ignored--

M    lib/CodeGen/AsmPrinter/DwarfDebug.cpp
A    test/DebugInfo/incorrect-variable-debugloc1.ll

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223981 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoLiveInterval: Use more range based for loops for value numbers and segments.
Matthias Braun [Wed, 10 Dec 2014 23:07:54 +0000 (23:07 +0000)]
LiveInterval: Use more range based for loops for value numbers and segments.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223978 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix PR21694. r219517 added a use of SCEV divide in HowFarToZero computation. This...
Mark Heffernan [Wed, 10 Dec 2014 22:53:52 +0000 (22:53 +0000)]
Fix PR21694. r219517 added a use of SCEV divide in HowFarToZero computation.  This divide can produce incorrect results as we are using an unsigned divide for what should be a modular divide.  This change reverts back to a more conservative computation using trailing zeros.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223974 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Adding combine ri/ir instructions.
Colin LeMahieu [Wed, 10 Dec 2014 22:23:07 +0000 (22:23 +0000)]
[Hexagon] Adding combine ri/ir instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223971 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoConstantFold: Clean up X * undef code
David Majnemer [Wed, 10 Dec 2014 21:58:17 +0000 (21:58 +0000)]
ConstantFold: Clean up X * undef code

No functional change intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223970 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoConstantFold, InstSimplify: undef >>a x can be either -1 or 0, choose 0
David Majnemer [Wed, 10 Dec 2014 21:58:15 +0000 (21:58 +0000)]
ConstantFold, InstSimplify: undef >>a x can be either -1 or 0, choose 0

Zero is usually a nicer constant to have than -1.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223969 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoConstantFold: an undef shift amount results in undef
David Majnemer [Wed, 10 Dec 2014 21:38:05 +0000 (21:38 +0000)]
ConstantFold: an undef shift amount results in undef

X shifted by undef results in undef because the undef value can
represent values greater than the width of the operands.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223968 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Adding encodings for JR class instructions. Updating complier usages.
Colin LeMahieu [Wed, 10 Dec 2014 21:24:10 +0000 (21:24 +0000)]
[Hexagon] Adding encodings for JR class instructions.  Updating complier usages.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223967 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMove three methods only used by MCJIT to MCJIT.
Rafael Espindola [Wed, 10 Dec 2014 20:46:55 +0000 (20:46 +0000)]
Move three methods only used by MCJIT to MCJIT.

These methods are only used by MCJIT and are very specific to it. In fact, they
are also fairly specific to the fact that we have a dynamic linker of
relocatable objects.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223964 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[AArch64] MachO large code-model: Materialize FP constants in code.
Juergen Ributzka [Wed, 10 Dec 2014 19:43:32 +0000 (19:43 +0000)]
[AArch64] MachO large code-model: Materialize FP constants in code.

In the large code model we have to first get the address of the GOT entry, load
the address of the constant, and then load the constant itself.

To avoid these loads and the GOT entry alltogether this commit changes the way
how FP constants are materialized in the large code model. The constats are now
materialized in a GPR and then bitconverted/moved into the FPR.

Reviewed by Tim Northover

Fixes rdar://problem/16572564.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223941 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Use getTargetConstant in AdjustRegClass
Marek Olsak [Wed, 10 Dec 2014 19:25:31 +0000 (19:25 +0000)]
R600/SI: Use getTargetConstant in AdjustRegClass

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223940 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Adding JR class predicated call reg instructions.
Colin LeMahieu [Wed, 10 Dec 2014 18:24:16 +0000 (18:24 +0000)]
[Hexagon] Adding JR class predicated call reg instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223933 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdded documentation for MergeFunctions pass:
Stepan Dyatkovskiy [Wed, 10 Dec 2014 17:42:01 +0000 (17:42 +0000)]
Added documentation for MergeFunctions pass:
Pass looks for equivalent functions that are mergable and folds them.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223931 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMatch new shuffle codegen for MOVHPD patterns
Sanjay Patel [Wed, 10 Dec 2014 16:58:54 +0000 (16:58 +0000)]
Match new shuffle codegen for MOVHPD patterns

Add patterns to match SSE (shufpd) and AVX (vpermilpd) shuffle codegen
when storing the high element of a v2f64. The existing patterns were
only checking for an unpckh type of shuffle.

http://llvm.org/bugs/show_bug.cgi?id=21791

Differential Revision: http://reviews.llvm.org/D6586

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223929 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoSilencing a -Wsequence-point warning, and the resulting undefined behavior. NFC.
Aaron Ballman [Wed, 10 Dec 2014 14:14:54 +0000 (14:14 +0000)]
Silencing a -Wsequence-point warning, and the resulting undefined behavior. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223926 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoConstantFold: div undef, 0 should fold to undef, not zero
David Majnemer [Wed, 10 Dec 2014 09:14:55 +0000 (09:14 +0000)]
ConstantFold: div undef, 0 should fold to undef, not zero

Dividing by zero yields an undefined value.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223924 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoInstSimplify: [al]shr exact undef, %X -> undef
David Majnemer [Wed, 10 Dec 2014 09:14:52 +0000 (09:14 +0000)]
InstSimplify: [al]shr exact undef, %X -> undef

Exact shifts always keep the non-zero bits of their input.  This means
it keeps it's undef bits.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223923 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] Make a code path in EltsFromConsecutiveLoads work only on vectors it expects
Michael Kuperstein [Wed, 10 Dec 2014 08:46:12 +0000 (08:46 +0000)]
[X86] Make a code path in EltsFromConsecutiveLoads work only on vectors it expects

EltsFromConsecutiveLoads was apparently only ever called for 128-bit vectors, and assumed this implicitly. r223518 started calling it for AVX-sized vectors, causing the code path that had this assumption to crash.
This adds a check to make this path fire only for 128-bit vectors.

Differential Revision: http://reviews.llvm.org/D6579

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223922 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoInstSimplify: div %X, 0 -> undef
David Majnemer [Wed, 10 Dec 2014 07:52:18 +0000 (07:52 +0000)]
InstSimplify: div %X, 0 -> undef

We already optimized rem %X, 0 to undef, we should do the same for div.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223919 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUse unique_ptr instead of DeleteContainerSeconds.
Craig Topper [Wed, 10 Dec 2014 06:18:57 +0000 (06:18 +0000)]
Use unique_ptr instead of DeleteContainerSeconds.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223918 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[LIT] Add support for `UNSUPPORTED` tag to `TestRunner.parseIntegratedTestScript`
Eric Fiselier [Wed, 10 Dec 2014 03:42:09 +0000 (03:42 +0000)]
[LIT] Add support for `UNSUPPORTED` tag to `TestRunner.parseIntegratedTestScript`

Summary:
This patch gives me just enough to leverage the existing functionality in `TestRunner` for use in `libc++` and `libc++abi` .

It does the following:
* Adds the `UNSUPPORTED` tag to `TestRunner.parseIntegratedTestScript`.
* Allows `parseIntegratedTestScript` to return an empty script if a script is not required by the caller.

Reviewers: ddunbar, EricWF

Reviewed By: EricWF

Subscribers: cfe-commits, llvm-commits

Differential Revision: http://reviews.llvm.org/D6589

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223915 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDataLayout: Provide nicer diagnostics for malformed strings
David Majnemer [Wed, 10 Dec 2014 02:36:41 +0000 (02:36 +0000)]
DataLayout: Provide nicer diagnostics for malformed strings

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223911 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAsmParser: Don't allow null bytes in BB labels
David Majnemer [Wed, 10 Dec 2014 02:10:35 +0000 (02:10 +0000)]
AsmParser: Don't allow null bytes in BB labels

Since Value objects can't have null bytes in their name, we shouldn't
allow them in the labels of basic blocks.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223907 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoIR: Move call to dropAllReferences() to MDNode subclasses
Duncan P. N. Exon Smith [Wed, 10 Dec 2014 01:45:04 +0000 (01:45 +0000)]
IR: Move call to dropAllReferences() to MDNode subclasses

Don't call `dropAllReferences()` from `MDNode::~MDNode()`, call it
directly from `~MDNodeFwdDecl()` and `~GenericMDNode()`.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223904 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDataLayout: Be more verbose when diagnosing problems in pointer specs
David Majnemer [Wed, 10 Dec 2014 01:38:28 +0000 (01:38 +0000)]
DataLayout: Be more verbose when diagnosing problems in pointer specs

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223903 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoI didn't intend to commit this with r223898
David Majnemer [Wed, 10 Dec 2014 01:17:48 +0000 (01:17 +0000)]
I didn't intend to commit this with r223898

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223899 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDataLayout: Move asserts over to report_fatal_error
David Majnemer [Wed, 10 Dec 2014 01:17:08 +0000 (01:17 +0000)]
DataLayout: Move asserts over to report_fatal_error

As indicated by the tests, it is possible to feed the AsmParser an
invalid datalayout string.  We should verify the result of parsing this
string regardless of whether or not we have assertions enabled.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223898 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMachineVerifier: Allow physreg use if just a subreg is defined.
Matthias Braun [Wed, 10 Dec 2014 01:13:13 +0000 (01:13 +0000)]
MachineVerifier: Allow physreg use if just a subreg is defined.

We can't mark partially undefined registers, so we have to allow reading
a register in the machine verifier if just parts of a register are
defined.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223896 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMachineVerifier: Allow LiveInterval segments to end at a partial write.
Matthias Braun [Wed, 10 Dec 2014 01:13:11 +0000 (01:13 +0000)]
MachineVerifier: Allow LiveInterval segments to end at a partial write.

In the subregister liveness tracking case we do not create implicit
reads on partial register writes anymore, still we need to produce a new
SSA value for partial writes so the live segment has to end.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223895 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoVirtRegMap: Improve block live-in info if subregister liveness is available.
Matthias Braun [Wed, 10 Dec 2014 01:13:08 +0000 (01:13 +0000)]
VirtRegMap: Improve block live-in info if subregister liveness is available.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223894 91177308-0d34-0410-b5e6-96231b3b80d8