oota-llvm.git
8 years ago[ELFYAML] Fix handling SHT_NOBITS sections by obj2yaml/yaml2obj tools
Simon Atanasyan [Fri, 3 Jul 2015 23:00:54 +0000 (23:00 +0000)]
[ELFYAML] Fix handling SHT_NOBITS sections by obj2yaml/yaml2obj tools

SHT_NOBITS sections do not have content in an object file. Now the yaml2obj
tool does not accept `Content` field for such sections, and the obj2yaml
tool does not attempt to read the section content from a file.

Restore r241350 and r241352.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241377 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoUse a continue to reduce indentation.
Rafael Espindola [Fri, 3 Jul 2015 22:02:28 +0000 (22:02 +0000)]
Use a continue to reduce indentation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241375 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoUse a continue to reduce indentation.
Rafael Espindola [Fri, 3 Jul 2015 21:57:41 +0000 (21:57 +0000)]
Use a continue to reduce indentation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241374 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoContext is allocated just a few lines above. Don't check if it is null.
Rafael Espindola [Fri, 3 Jul 2015 21:54:41 +0000 (21:54 +0000)]
Context is allocated just a few lines above. Don't check if it is null.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241373 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFix build with -DLLVM_USE_INTEL_JITEVENTS=ON -DLLVM_USE_OPROFILE=ON.
Rafael Espindola [Fri, 3 Jul 2015 21:47:00 +0000 (21:47 +0000)]
Fix build with -DLLVM_USE_INTEL_JITEVENTS=ON -DLLVM_USE_OPROFILE=ON.

Is anyone using those?

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241372 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRemove always-true comparison, NFC.
Filipe Cabecinhas [Fri, 3 Jul 2015 20:12:01 +0000 (20:12 +0000)]
Remove always-true comparison, NFC.

Summary:
Looking at r241279, I noticed that UpgradedIntrinsics only gets written
to in the following code:

    if (UpgradeIntrinsicFunction(&F, NewFn))
      UpgradedIntrinsics[&F] = NewFn;

Looking through UpgradeIntrinsicFunction, we always return false OR
NewFn will be set to a different function from our source.

This patch pulls the F != NewFn into UpgradeIntrinsicFunction as an
assert, and removes the check from callers of UpgradeIntrinsicFunction.

Reviewers: rafael, chandlerc

Subscribers: llvm-commits-list

Differential Revision: http://reviews.llvm.org/D10915

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241369 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Added 32-bit builds to fp<->int tests.
Simon Pilgrim [Fri, 3 Jul 2015 20:07:57 +0000 (20:07 +0000)]
[X86] Added 32-bit builds to fp<->int tests.

Ensure that i686 x87/SSE/SSE2 targets all build.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241368 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoDelete dead code. NFC.
Benjamin Kramer [Fri, 3 Jul 2015 19:10:09 +0000 (19:10 +0000)]
Delete dead code. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241367 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoReturn ErrorOr from getSymbolAddress.
Rafael Espindola [Fri, 3 Jul 2015 18:19:00 +0000 (18:19 +0000)]
Return ErrorOr from getSymbolAddress.

It can fail trying to get the section on ELF and COFF. This makes sure the
error is handled.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241366 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoReplace a few more MachO only uses of getSymbolAddress.
Rafael Espindola [Fri, 3 Jul 2015 18:02:36 +0000 (18:02 +0000)]
Replace a few more MachO only uses of getSymbolAddress.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241365 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoUse getValue instead of getAddress in a few MachO only cases.
Rafael Espindola [Fri, 3 Jul 2015 17:44:18 +0000 (17:44 +0000)]
Use getValue instead of getAddress in a few MachO only cases.

In MachO the value of the symbol is always the address, so we can use the
simpler function.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241364 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoThis reverts commit r241350 and r241352.
Rafael Espindola [Fri, 3 Jul 2015 14:54:02 +0000 (14:54 +0000)]
This reverts commit r241350 and r241352.

r241350 broke lld tests.
r241352 depends on r241350.

Original messages:
"[ELFYAML] Fix handling SHT_NOBITS sections by obj2yaml/yaml2obj tools"
"[ELFYAML] Make the Size field for .bss section optional"

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241354 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoDelete dead code.
Rafael Espindola [Fri, 3 Jul 2015 14:46:17 +0000 (14:46 +0000)]
Delete dead code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241353 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ELFYAML] Make the Size field for .bss section optional
Simon Atanasyan [Fri, 3 Jul 2015 14:19:06 +0000 (14:19 +0000)]
[ELFYAML] Make the Size field for .bss section optional

It's a common case to have a zero-size .bss section in an object file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241352 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ELFYAML] Fix handling SHT_NOBITS sections by obj2yaml/yaml2obj tools
Simon Atanasyan [Fri, 3 Jul 2015 14:07:06 +0000 (14:07 +0000)]
[ELFYAML] Fix handling SHT_NOBITS sections by obj2yaml/yaml2obj tools

SHT_NOBITS sections do not have content in an object file. Now yaml2obj
tool does not accept `Content` field for such sections, and obj2yaml
tool does not attempt to read the section content from a file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241350 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAvoid warning about unused variable when building without assertions.
Rafael Espindola [Fri, 3 Jul 2015 12:53:50 +0000 (12:53 +0000)]
Avoid warning about unused variable when building without assertions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241348 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAvoid a use after free.
Rafael Espindola [Fri, 3 Jul 2015 12:20:34 +0000 (12:20 +0000)]
Avoid a use after free.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241345 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoContinue to remove the notion that ELF has dynamic and static symbols.
Rafael Espindola [Fri, 3 Jul 2015 12:00:05 +0000 (12:00 +0000)]
Continue to remove the notion that ELF has dynamic and static symbols.

The ELFObjectFile now just reasons about a section/index pair, removing
one of the users that force ELF.h to maintain the difference.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241344 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agollvm/test/CodeGen/ARM/fnattr-trap.ll: Add -mtriple, to appease targeting *-win32.
NAKAMURA Takumi [Fri, 3 Jul 2015 08:21:38 +0000 (08:21 +0000)]
llvm/test/CodeGen/ARM/fnattr-trap.ll: Add -mtriple, to appease targeting *-win32.

  LLVM ERROR: CPU: 'generic' does not support ARM mode execution!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241329 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agowhitespace tidyup. NFC.
Simon Pilgrim [Fri, 3 Jul 2015 08:02:12 +0000 (08:02 +0000)]
whitespace tidyup. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241326 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86][SSE] Sign extension for target vector sizes less than 128 bits (pt2)
Simon Pilgrim [Fri, 3 Jul 2015 08:01:36 +0000 (08:01 +0000)]
[X86][SSE] Sign extension for target vector sizes less than 128 bits (pt2)

Add support for v2i8/v2i16 to v2f64 by using a sign extension to v2i32 before conversion to v2f64.

Differential Revision: http://reviews.llvm.org/D10589

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241325 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoInitialize booleans CallsUnwindInit and CallsEHReturn with false instead of 0.
Yaron Keren [Fri, 3 Jul 2015 07:56:24 +0000 (07:56 +0000)]
Initialize booleans CallsUnwindInit and CallsEHReturn with false instead of 0.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241324 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86][SSE] Sign extension for target vector sizes less than 128 bits (pt1)
Simon Pilgrim [Fri, 3 Jul 2015 07:51:01 +0000 (07:51 +0000)]
[X86][SSE] Sign extension for target vector sizes less than 128 bits (pt1)

This patch adds support for sign extension for sub 128-bit vectors, such as to v2i32. It concatenates with UNDEF subvectors up to 128-bits, performs the sign extension (i.e. as v4i32) and then extracts the target subvector.

Patch 1/2 of D10589 - the second patch covers the conversion of v2i8/v2i16 to v2f64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241323 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFix an overly aggressive assertion in getCopyFromPartsVector.
Nadav Rotem [Thu, 2 Jul 2015 23:23:52 +0000 (23:23 +0000)]
Fix an overly aggressive assertion in getCopyFromPartsVector.

The assertion in getCopyFromPartsVector assumed that the vector 'part' must
match the type of argument (arguments are potentially split into multiple
parts). However, in some cases the targets return a 'part' of the right size
but with a different type. We already handle this case correctly later on
and generate a bitcast. This commit just makes sure that we are actually
checking the property that we care about.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241312 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoDIBuilder: Now that DICompileUnit is distinct, stop using temporary nodes
Adrian Prantl [Thu, 2 Jul 2015 22:32:52 +0000 (22:32 +0000)]
DIBuilder: Now that DICompileUnit is distinct, stop using temporary nodes
for the arrays.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241308 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoUse function attribute "trap-func-name" and remove TargetOptions::TrapFuncName.
Akira Hatanaka [Thu, 2 Jul 2015 22:13:27 +0000 (22:13 +0000)]
Use function attribute "trap-func-name" and remove TargetOptions::TrapFuncName.

This commit changes normal isel and fast isel to read the user-defined trap
function name from function attribute "trap-func-name" attached to llvm.trap or
llvm.debugtrap instead of from TargetOptions::TrapFuncName. This is needed to
use clang's command line option "-ftrap-function" for LTO and enable changing
the trap function name on a per-call-site basis.

Out-of-tree projects currently using TargetOptions::TrapFuncName to specify the
trap function name should attach attribute "trap-func-name" to the call sites
of llvm.trap and llvm.debugtrap instead.

rdar://problem/21225723

Differential Revision: http://reviews.llvm.org/D10832

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241305 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAdd functions for adding and testing string attributes to CallInst. NFC.
Akira Hatanaka [Thu, 2 Jul 2015 22:08:48 +0000 (22:08 +0000)]
Add functions for adding and testing string attributes to CallInst. NFC.

This change is needed later when I make changes to attach string function
attributes to llvm.trap and llvm.debugtrap.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241304 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[WebAssembly] Set the HasFloatingPointExceptions flag for WebAssembly.
Dan Gohman [Thu, 2 Jul 2015 21:36:25 +0000 (21:36 +0000)]
[WebAssembly] Set the HasFloatingPointExceptions flag for WebAssembly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241302 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoTry to fix the build of IntelJITEventListener.
Rafael Espindola [Thu, 2 Jul 2015 21:24:12 +0000 (21:24 +0000)]
Try to fix the build of IntelJITEventListener.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241301 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoReturn ErrorOr from SymbolRef::getName.
Rafael Espindola [Thu, 2 Jul 2015 20:55:21 +0000 (20:55 +0000)]
Return ErrorOr from SymbolRef::getName.

This function can really fail since the string table offset can be out of
bounds.

Using ErrorOr makes sure the error is checked.

Hopefully a lot of the boilerplate code in tools/* can go away once we have
a diagnostic manager in Object.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241297 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[PPC64LE] Remove implicit-subreg restriction from VSX swap removal
Bill Schmidt [Thu, 2 Jul 2015 19:01:22 +0000 (19:01 +0000)]
[PPC64LE] Remove implicit-subreg restriction from VSX swap removal

In r241285, I removed the SUBREG_TO_REG restriction from VSX swap
removal, determining that this was overly conservative.  We have
another form of the same restriction in that we check for the presence
of implicit subregs in vector operations.  As with SUBREG_TO_REG for
partial register conversions, an implicit subreg is safe in and of
itself, provided no other operation makes a lane-sensitive assumption
about the result.  This patch removes that restriction, by removing
the HasImplicitSubreg flag and all code that relies on it.

I've added a test case that fails to optimize before this patch is
applied, and optimizes properly with the patch.  Test based on a
report from Anton Blanchard.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241290 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Statepoints] Make operator bool() explicit.
Sanjoy Das [Thu, 2 Jul 2015 18:15:18 +0000 (18:15 +0000)]
[Statepoints] Make operator bool() explicit.

(Addressing post-commit review.)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241288 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRemove a report_fatal_error that should be unreachable.
Rafael Espindola [Thu, 2 Jul 2015 17:16:07 +0000 (17:16 +0000)]
Remove a report_fatal_error that should be unreachable.

If we created a relocation iterator, we have a valid relocation section.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241286 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[PPC64LE] Teach swap optimization about the doubleword splat idiom
Bill Schmidt [Thu, 2 Jul 2015 17:03:06 +0000 (17:03 +0000)]
[PPC64LE] Teach swap optimization about the doubleword splat idiom

With a previous patch, the VSX swap optimization is able to recognize
the doubleword load-splat idiom that can be implemented using lxvdsx.
However, that does not cover a doubleword splat where the source is a
register.  We can implement this using xxspltd (a special form of
xxpermdi).  This patch teaches the swap optimization pass about this
idiom.

As a prerequisite, it also permits swap optimization to succeed for
all forms of SUBREG_TO_REG.  Previously we were conservative and only
allowed SUBREG_TO_REG when it copied a full register.  However, on
reflection any form of SUBREG_TO_REG is safe in and of itself, so long
as an unsafe operation is not performed on its result.  In particular,
a widening SUBREG_TO_REG often occurs as an input to a doubleword
splat idiom, particularly in auto-vectorized code.

The doubleword splat idiom is an XXPERMDI operation where both source
registers are identical, and the selection mask is either 0 (splat the
first element) or 3 (splat the second element).  To determine whether
the registers are identical, we use the existing mechanism for looking
through "copy-like" operations.  That mechanism has a side effect of
marking the XXPERMDI operation as using a physical register, which
would invalidate its presence in a swap-optimized region.  This is
correct for the form of XXPERMDI that performs a swap and hence would
be removed, but is not what we want for a doubleword-splat variety of
XXPERMDI.  Therefore we reset the physical-register flag on the
XXPERMDI when it represents a splat.

A simple test case is added to verify that we generate the splat and
that we also remove the xxswapd instructions that would otherwise be
associated with the load and store of another operand.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241285 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoConvert a member variable to a local one.
Rafael Espindola [Thu, 2 Jul 2015 16:59:57 +0000 (16:59 +0000)]
Convert a member variable to a local one.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241284 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoReworking the test part of r241149
Gabor Ballabas [Thu, 2 Jul 2015 16:53:23 +0000 (16:53 +0000)]
Reworking the test part of r241149

The test part of r241149 has been reverted in r241451, due to misplaced test cases.
This patch splits those test cases among the appropriate targets.

Differential Revision: http://reviews.llvm.org/D10897

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241283 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFix for PR23310: llvm-dis crashes when trying to upgrade an intrinsic.
Rafael Espindola [Thu, 2 Jul 2015 16:22:40 +0000 (16:22 +0000)]
Fix for PR23310: llvm-dis crashes when trying to upgrade an intrinsic.

When trying to upgrade @llvm.x86.sse2.psrl.dq while parsing a module,
BitcodeReader adds the function to its worklist twice, resulting in a
crash when accessing it the second time.

This patch replaces the worklist vector by a map.

Patch by Philip Pfaffe.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241281 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoHandle .dynsym a bit more like we handle .symtab.
Rafael Espindola [Thu, 2 Jul 2015 16:03:38 +0000 (16:03 +0000)]
Handle .dynsym a bit more like we handle .symtab.

They have the same format and we find them in the same way, no reason to handle
them differently.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241280 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRangify some loops.
Rafael Espindola [Thu, 2 Jul 2015 15:55:09 +0000 (15:55 +0000)]
Rangify some loops.

Patch by Philip Pfaffe!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241279 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoUse default member initializes. NFC.
Rafael Espindola [Thu, 2 Jul 2015 15:48:05 +0000 (15:48 +0000)]
Use default member initializes. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241277 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Support] Lazy load of dbghlp.dll on Windows
Leny Kholodov [Thu, 2 Jul 2015 14:34:57 +0000 (14:34 +0000)]
[Support] Lazy load of dbghlp.dll on Windows

This patch changes linkage with dbghlp.dll for clang from static (at load time)
to on demand (at the first use of required functions). Clang uses dbghlp.dll
only in minor use-cases. First of all in case of crash and in case of plugin load.
The dbghlp.dll library can be absent on system. In this case clang will fail
to load. With lazy load of dbghlp.dll clang can work even if dbghlp.dll
is not available.

Differential Revision: http://reviews.llvm.org/D10737

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241271 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Convert an instruction relaxation test to use objdump instead of readobj
Michael Kuperstein [Thu, 2 Jul 2015 14:27:35 +0000 (14:27 +0000)]
[X86] Convert an instruction relaxation test to use objdump instead of readobj

Patch by: david.l.kreitzer@intel.com

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241270 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMove error checking from once per relocation to once per relocation section.
Rafael Espindola [Thu, 2 Jul 2015 14:27:07 +0000 (14:27 +0000)]
Move error checking from once per relocation to once per relocation section.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241269 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRemove whitespace from start of line, NFC.
Yaron Keren [Thu, 2 Jul 2015 14:25:09 +0000 (14:25 +0000)]
Remove whitespace from start of line, NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241268 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoExpose getRel and getRela to reduce code duplication.
Rafael Espindola [Thu, 2 Jul 2015 14:21:38 +0000 (14:21 +0000)]
Expose getRel and getRela to reduce code duplication.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241266 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoDelete whitespace at start of line.
Yaron Keren [Thu, 2 Jul 2015 14:17:12 +0000 (14:17 +0000)]
Delete whitespace at start of line.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241265 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoImprove error message.
Rafael Espindola [Thu, 2 Jul 2015 11:48:48 +0000 (11:48 +0000)]
Improve error message.

Thanks to Sean Silva for the suggestion.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241255 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoReapply r240291: Fix shl folding in DAG combiner.
Pawel Bylica [Thu, 2 Jul 2015 11:44:54 +0000 (11:44 +0000)]
Reapply r240291: Fix shl folding in DAG combiner.

The code responsible for shl folding in the DAGCombiner was assuming incorrectly that all constants are less than 64 bits. This patch simply changes the way values are compared.

It has been reverted previously because of some problems with comparing APInt with raw uint64_t. That has been fixed/changed with r241204.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241254 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[GraphWriter] Don't wait on xdg-open when not on Apple.
Charlie Turner [Thu, 2 Jul 2015 09:32:07 +0000 (09:32 +0000)]
[GraphWriter] Don't wait on xdg-open when not on Apple.

By default, the GraphWriter code assumes that the generic file open
program (`open` on Apple, `xdg-open` on other systems) can wait on the
forked proces to complete. When the fork ends, the code would delete
the temporary dot files created, and return.

On GNU/Linux, the xdg-open program does not have a "wait for your fork
to complete before dying" option. So the behaviour was that xdg-open
would launch a process, quickly die itself, and then the GraphWriter
code would think its OK to quickly delete all the temporary files.
Once the temporary files were deleted, the dot viewers would get very
upset, and often give you weird errors.

This change only waits on the generic open program on Apple platforms.
Elsewhere, we don't wait on the process, and hence we don't try and
clean up the temporary files.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241250 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[NFC] Minor editorial fixes to the CodeGen docs.
Charlie Turner [Thu, 2 Jul 2015 09:32:01 +0000 (09:32 +0000)]
[NFC] Minor editorial fixes to the CodeGen docs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241249 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[NFC] Make the Statepoint class more like CallSite
Sanjoy Das [Thu, 2 Jul 2015 02:53:45 +0000 (02:53 +0000)]
[NFC] Make the Statepoint class more like CallSite

Summary: Rename some methods to make Statepoint look more like CallSite.

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D10756

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241235 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Statepoints][NFC] Constify accessors on Statepoint.
Sanjoy Das [Thu, 2 Jul 2015 02:53:41 +0000 (02:53 +0000)]
[Statepoints][NFC] Constify accessors on Statepoint.

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D10755

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241234 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Statepoints][NFC] Rename variables to llvm style.
Sanjoy Das [Thu, 2 Jul 2015 02:53:39 +0000 (02:53 +0000)]
[Statepoints][NFC] Rename variables to llvm style.

Summary:
(I don't think this change needs review, this was uploaded to
phabricator to provide context for later dependent changes.)

Differential Revision: http://reviews.llvm.org/D10630

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241233 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Statepoints][NFC] Add Statepoint::operator bool()
Sanjoy Das [Thu, 2 Jul 2015 02:53:36 +0000 (02:53 +0000)]
[Statepoints][NFC] Add Statepoint::operator bool()

Summary:
This allows the "if (Statepoint SP = Statepoint(I))" idiom.

(I don't think this change needs review, this was uploaded to
phabricator to provide context for later dependent changes.)

Differential Revision: http://reviews.llvm.org/D10629

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241232 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Statepoints][NFC] Add Statepoint::getGCResult.
Sanjoy Das [Thu, 2 Jul 2015 02:53:34 +0000 (02:53 +0000)]
[Statepoints][NFC] Add Statepoint::getGCResult.

Summary:
Introduce a simple accessor to get the gc_result hanging off of a
statepoint.

(I don't think this change needs review, this was uploaded to
phabricator to provide context for later dependent changes.)

Differential Revision: http://reviews.llvm.org/D10627

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241231 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[LazyCallGraph] Port test case from r240039 to LCG.
Sanjoy Das [Thu, 2 Jul 2015 02:03:58 +0000 (02:03 +0000)]
[LazyCallGraph] Port test case from r240039 to LCG.

Summary:
r240039 adds a test case to check that CallGraph does the right thing
with respect to non-leaf intrinsics like statepoint and patchpoint.
This ports the same test case to LazyCallGraph.  LazyCallGraph already
does the right thing with respect to escaping function pointers so there
is no need to change any code.

Reviewers: chandlerc

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D10582

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241226 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMake an X86 specific directory and put the recent X86 tti specific
Eric Christopher [Thu, 2 Jul 2015 01:36:31 +0000 (01:36 +0000)]
Make an X86 specific directory and put the recent X86 tti specific
inlining test into it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241223 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoImplement TargetTransformInfo::hasCompatibleFunctionAttributes for X86.
Eric Christopher [Thu, 2 Jul 2015 01:11:50 +0000 (01:11 +0000)]
Implement TargetTransformInfo::hasCompatibleFunctionAttributes for X86.

This checks subtarget feature compatibility for inlining by verifying
that the callee is a strict subset of the caller's features. This includes
the cpu as part of the subtarget we can get via the incoming functions as
the backend takes CPUs as feature sets.

This allows us to inline things like:

int foo() { return baz(); }

int __attribute__((target("sse4.2"))) bar() {
  return foo();
}

so that generic code can be inlined into specialized functions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241221 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAdd a routine to TargetTransformInfo that will allow targets to look
Eric Christopher [Thu, 2 Jul 2015 01:11:47 +0000 (01:11 +0000)]
Add a routine to TargetTransformInfo that will allow targets to look
at the attributes on a function to determine whether or not to allow
inlining.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241220 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoWebAssembly: start instructions
JF Bastien [Wed, 1 Jul 2015 23:41:25 +0000 (23:41 +0000)]
WebAssembly: start instructions

Summary:
* Add 64-bit address space feature.
* Rename SIMD feature to SIMD128.
* Handle single-thread model with an IR pass (same way ARM does).
* Rename generic processor to MVP, to follow design's lead.
* Add bleeding-edge processors, with all features included.
* Fix a few DEBUG_TYPE to match other backends.

Test Plan: ninja check

Reviewers: sunfish

Subscribers: jfb, llvm-commits

Differential Revision: http://reviews.llvm.org/D10880

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241211 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TwoAddressInstructionPass] Try 3 Addr Conversion After Commuting.
Quentin Colombet [Wed, 1 Jul 2015 23:12:13 +0000 (23:12 +0000)]
[TwoAddressInstructionPass] Try 3 Addr Conversion After Commuting.

TwoAddressInstructionPass stops after a successful commuting but 3 Addr
conversion might be good for some cases.

Consider:

int foo(int a, int b) {
  return a + b;
}

Before this commit, we emit:

addl %esi, %edi
movl %edi, %eax
ret

After this commit, we try 3 Addr conversion:

leal (%rsi,%rdi), %eax
ret

Patch by Volkan Keles <vkeles@apple.com>!

Differential Revision: http://reviews.llvm.org/D10851

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241206 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoChange APInt comparison with uint64_t.
Pawel Bylica [Wed, 1 Jul 2015 22:56:43 +0000 (22:56 +0000)]
Change APInt comparison with uint64_t.

Summary:
This patch changes the way APInt is compared with a value of type uint64_t.
Before the uint64_t value was truncated to the size of APInt before comparison.
Now the comparison takes into account full 64-bit precision.

Test Plan: Unit tests added. No regressions. Self-hosted check-all done as well.

Reviewers: chandlerc, dexonsmith

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D10655

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241204 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoTest for specific output in lit test
Matthias Braun [Wed, 1 Jul 2015 22:34:59 +0000 (22:34 +0000)]
Test for specific output in lit test

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241200 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[LoopVectorize] Use ReplaceInstWithInst() helper where appropriate.
Alexey Samsonov [Wed, 1 Jul 2015 22:18:30 +0000 (22:18 +0000)]
[LoopVectorize] Use ReplaceInstWithInst() helper where appropriate.

This is mostly an NFC, which increases code readability (instead of
saving old terminator, generating new one in front of old, and deleting
old, we just call a function). However, it would additionaly copy
the debug location from old instruction to replacement, which
would help PR23837.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241197 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoPack MCSymbol::Flags in to the bitfield with other members. NFC.
Pete Cooper [Wed, 1 Jul 2015 21:57:51 +0000 (21:57 +0000)]
Pack MCSymbol::Flags in to the bitfield with other members.  NFC.

All file formats only needed 16-bits right now which is enough to fit
in to the padding with other fields.

This reduces the size of MCSymbol to 24-bytes on a 64-bit system.  The
layout is now

   0 | class llvm::MCSymbol
   0 |   class llvm::PointerIntPair SectionOrFragmentAndHasName
   0 |     intptr_t Value
     |   [sizeof=8, dsize=8, align=8
     |    nvsize=8, nvalign=8]

   8 |   unsigned int IsTemporary
   8 |   unsigned int IsRedefinable
   8 |   unsigned int IsUsed
   8 |   _Bool IsRegistered
   8 |   unsigned int IsExternal
   8 |   unsigned int IsPrivateExtern
   8 |   unsigned int Kind
   9 |   unsigned int IsUsedInReloc
   9 |   unsigned int SymbolContents
   9 |   unsigned int CommonAlignLog2
  10 |   uint32_t Flags
  12 |   uint32_t Index
  16 |   union
  16 |     uint64_t Offset
  16 |     uint64_t CommonSize
  16 |     const class llvm::MCExpr * Value
     |   [sizeof=8, dsize=8, align=8
     |    nvsize=8, nvalign=8]

     | [sizeof=24, dsize=24, align=8
     |  nvsize=24, nvalign=8]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241196 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[WebAssembly] Define separate Target instances for 32-bit and 64-bit.
Dan Gohman [Wed, 1 Jul 2015 21:42:34 +0000 (21:42 +0000)]
[WebAssembly] Define separate Target instances for 32-bit and 64-bit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241193 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[NVPTX] expand extload/truncstore for vectors of floats
Jingyue Wu [Wed, 1 Jul 2015 21:32:42 +0000 (21:32 +0000)]
[NVPTX] expand extload/truncstore for vectors of floats

Summary:
According to PTX ISA:

For convenience, ld, st, and cvt instructions permit source and destination data operands to be wider than the instruction-type size, so that narrow values may be loaded, stored, and converted using regular-width registers. For example, 8-bit or 16-bit values may be held directly in 32-bit or 64-bit registers when being loaded, stored, or converted to other types and sizes. The operand type checking rules are relaxed for bit-size and integer (signed and unsigned) instruction types; floating-point instruction types still require that the operand type-size matches exactly, unless the operand is of bit-size type.

So, the ISA does not support load with extending/store with truncatation for floating numbers. This is reflected in setting the loadext/truncstore actions to expand in the code for floating numbers, but vectors of floating numbers are not taken care of.

As a result, loading a vector of floats followed by a fp_extend may be combined by DAGCombiner to a extload, and the extload may be lowered to NVPTXISD::LoadV2 with extending information. However, NVPTXISD::LoadV2 does not perform extending, and no extending instructions are inserted. Finally, PTX instructions with mismatched types are generated, like
ld.v2.f32 {%fd3, %fd4}, [%rd2]

This patch adds the correct actions for vectors of floats, so DAGCombiner would not create loads with extending, and correct code is generated.

Patched by Gang Hu.

Test Plan: Test case attached.

Reviewers: jingyue

Reviewed By: jingyue

Subscribers: llvm-commits, jholewinski

Differential Revision: http://reviews.llvm.org/D10876

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241191 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoEncode MCSymbol alignment as log2(align).
Pete Cooper [Wed, 1 Jul 2015 21:07:03 +0000 (21:07 +0000)]
Encode MCSymbol alignment as log2(align).

Given that alignments are always powers of 2, just encode it this way.

This matches how we encode alignment on IR GlobalValue's for example.

This compresses the CommonAlign member down to 5 bits which allows it
to pack better with the surrounding fields.

Reviewed by Duncan Exon Smith.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241189 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[WinEH] Use llvm.x86.seh.recoverfp in WinEHPrepare
Reid Kleckner [Wed, 1 Jul 2015 20:59:25 +0000 (20:59 +0000)]
[WinEH] Use llvm.x86.seh.recoverfp in WinEHPrepare

Don't pattern match for frontend outlined finally calls on non-x64
platforms. The 32-bit runtime uses a different funclet prototype.  Now,
the frontend is pre-outlining the finally bodies so that it ends up
doing most of the heavy lifting for variable capturing. We're just
outlining the callsite, and adapting the frameaddress(0) call to line up
the frame pointer recovery.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241186 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[NVPTX] Move NVPTXPeephole after NVPTXPrologEpilogPass
Jingyue Wu [Wed, 1 Jul 2015 20:08:06 +0000 (20:08 +0000)]
[NVPTX] Move NVPTXPeephole after NVPTXPrologEpilogPass

Summary:
Offset of frame index is calculated by NVPTXPrologEpilogPass. Before
that the correct offset of stack objects cannot be obtained, which
leads to wrong offset if there are more than 2 frame objects. This patch
move NVPTXPeephole after NVPTXPrologEpilogPass. Because the frame index
is already replaced by %VRFrame in NVPTXPrologEpilogPass, we check
VRFrame register instead, and try to remove the VRFrame if there
is no usage after NVPTXPeephole pass.

Patched by Xuetian Weng.

Test Plan:
Strengthened test/CodeGen/NVPTX/local-stack-frame.ll to check the
offset calculation based on SP and SPL.

Reviewers: jholewinski, jingyue

Reviewed By: jingyue

Subscribers: jholewinski, llvm-commits

Differential Revision: http://reviews.llvm.org/D10853

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241185 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[PPC64LE] Enable missing lxvdsx optimization, and related swap optimization
Bill Schmidt [Wed, 1 Jul 2015 19:40:07 +0000 (19:40 +0000)]
[PPC64LE] Enable missing lxvdsx optimization, and related swap optimization

When adding little-endian vector support for PowerPC last year, I
inadvertently disabled an optimization that recognizes a load-splat
idiom and generates the lxvdsx instruction.  This patch moves the
offending logic so lxvdsx is once again generated.

This pattern is frequently generated by the vectorizer for scalar
loads of an effective constant.  Previously the lxvdsx instruction was
wrongly listed as lane-sensitive for the VSX swap optimization (since
both doublewords are identical, swaps are safe).  This patch fixes
this as well, so that vectorized code using lxvdsx can now have swaps
removed from the computation.

There is an existing test (@test50) in test/CodeGen/PowerPC/vsx.ll
that checks for the missing optimization.  However, vsx.ll was only
being tested for POWER7 with big-endian code generation.  I've added
a little-endian RUN statement and expected LE code generation for all
the tests in vsx.ll to give us a bit better VSX coverage, including
what's needed for this patch.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241183 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoadd a cl::opt override for TargetLoweringBase's JumpIsExpensive
Sanjay Patel [Wed, 1 Jul 2015 18:10:20 +0000 (18:10 +0000)]
add a cl::opt override for TargetLoweringBase's JumpIsExpensive

This patch is not intended to change existing codegen behavior for any target.
It just exposes the JumpIsExpensive setting on the command-line to allow for
easier testing and emergency overrides.

Also, change the existing regression test to use FileCheck, explicitly specify
the jump-is-expensive option, and use more precise checks.

Differential Revision: http://reviews.llvm.org/D10846

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241179 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoDisallow in-source builds (as we already do for the cmake build).
Jonathan Roelofs [Wed, 1 Jul 2015 18:09:21 +0000 (18:09 +0000)]
Disallow in-source builds (as we already do for the cmake build).

http://reviews.llvm.org/D10614

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241178 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRevert "[DWARF] Fix debug info generation for function static variables, typedefs...
David Blaikie [Wed, 1 Jul 2015 18:07:16 +0000 (18:07 +0000)]
Revert "[DWARF] Fix debug info generation for function static variables, typedefs, and records"

Caused PR24008

This reverts commit 37cb5f1c2db9f42d29f26b215585f56bb64ae4f5.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241176 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agofix formatting; NFC
Sanjay Patel [Wed, 1 Jul 2015 17:58:53 +0000 (17:58 +0000)]
fix formatting; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241175 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agofix typos in comment; NFC
Sanjay Patel [Wed, 1 Jul 2015 17:55:07 +0000 (17:55 +0000)]
fix typos in comment; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241174 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoLivePhysRegs: Add support to add pristine registers when populating with live-in...
Matthias Braun [Wed, 1 Jul 2015 17:17:17 +0000 (17:17 +0000)]
LivePhysRegs: Add support to add pristine registers when populating with live-in/live-out registers.

Differential Revision: http://reviews.llvm.org/D10139

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241172 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[SEH] Don't assert if the parent function lacks a personality
Reid Kleckner [Wed, 1 Jul 2015 16:45:47 +0000 (16:45 +0000)]
[SEH] Don't assert if the parent function lacks a personality

The EH code might have been deleted as unreachable and the personality
pruned while the filter is still present.  Currently I'm hitting this at
-O0 due to the clang bug PR24009.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241170 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AsmPrinter] Hide implementation details
Benjamin Kramer [Wed, 1 Jul 2015 16:18:16 +0000 (16:18 +0000)]
[AsmPrinter] Hide implementation details

NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241169 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AArch64] Implement add/adds/sub/subs/cmp/cmn with negative immediate aliases
Arnaud A. de Grandmaison [Wed, 1 Jul 2015 15:05:58 +0000 (15:05 +0000)]
[AArch64] Implement add/adds/sub/subs/cmp/cmn with negative immediate aliases

This patch teaches the AsmParser to accept add/adds/sub/subs/cmp/cmn
with a negative immediate operand and convert them as shown:

  add  Rd, Rn, -imm -> sub  Rd, Rn, imm
  sub  Rd, Rn, -imm -> add  Rd, Rn, imm
  adds Rd, Rn, -imm -> subs Rd, Rn, imm
  subs Rd, Rn, -imm -> adds Rd, Rn, imm
  cmp  Rn, -imm     -> cmn  Rn, imm
  cmn  Rn, -imm     -> cmp  Rn, imm

Those instructions are an alternate syntax available to assembly coders,
and are needed in order to support code already compiling with some other
assemblers (gas). They are documented in the "ARMv8 Instruction Set
Overview", in the "Arithmetic (immediate)" section. This makes llvm-mc
a programmer-friendly assembler !

This also fixes PR20978: "Assembly handling of adding negative numbers
not as smart as gas".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241166 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[SDAG] Give InstrEmitter hidden visibility
Benjamin Kramer [Wed, 1 Jul 2015 14:55:10 +0000 (14:55 +0000)]
[SDAG] Give InstrEmitter hidden visibility

NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241165 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[CodeGen] Reduce visibility of implementation details
Benjamin Kramer [Wed, 1 Jul 2015 14:47:39 +0000 (14:47 +0000)]
[CodeGen] Reduce visibility of implementation details

NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241164 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Sparc] Rearrange SparcInstrInfo, no change.
James Y Knight [Wed, 1 Jul 2015 14:38:07 +0000 (14:38 +0000)]
[Sparc] Rearrange SparcInstrInfo, no change.

Move some instructions into order of sections in the spec, as the rest
already were.

Differential Revision: http://reviews.llvm.org/D9102

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241163 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoTest committed in r241153 is more target-specific than I thought.
Michael Kuperstein [Wed, 1 Jul 2015 13:45:25 +0000 (13:45 +0000)]
Test committed in r241153 is more target-specific than I thought.
Moving the (original, x86-only) test to the X86 directory.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241162 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoExpand Phabricator docs slightly
Scott Douglass [Wed, 1 Jul 2015 13:41:18 +0000 (13:41 +0000)]
Expand Phabricator docs slightly

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241161 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAVX-512: Implemented missing encoding for FMA scalar instructions
Igor Breger [Wed, 1 Jul 2015 13:24:28 +0000 (13:24 +0000)]
AVX-512: Implemented missing encoding for FMA scalar instructions
Added tests for encoding

Differential Revision: http://reviews.llvm.org/D10865

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241159 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFix non-target-specific test not to use the x86 triple.
Michael Kuperstein [Wed, 1 Jul 2015 13:05:57 +0000 (13:05 +0000)]
Fix non-target-specific test not to use the x86 triple.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241158 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoReturn ErrorOr from getSection.
Rafael Espindola [Wed, 1 Jul 2015 12:56:27 +0000 (12:56 +0000)]
Return ErrorOr from getSection.

This also improves the logic of what is an error:

* getSection(uint_32): only return an error if the index is out of bounds. The
  index 0 corresponds to a perfectly valid entry.
* getSection(Elf_Sym): Returns null for symbols that normally don't have
  sections and error for out of bound indexes.

In many places this just moves the report_fatal_error up the stack, but those
can then be fixed in smaller patches.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241156 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[DWARF] Fix debug info generation for function static variables, typedefs, and records
Michael Kuperstein [Wed, 1 Jul 2015 12:33:11 +0000 (12:33 +0000)]
[DWARF] Fix debug info generation for function static variables, typedefs, and records

Function static variables, typedefs and records (class, struct or union) declared inside
a lexical scope were associated with the function as their parent scope, rather than the
lexical scope they are defined or declared in.

This fixes PR19238

Patch by: amjad.aboud@intel.com
Differential Revision: http://reviews.llvm.org/D9758

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241153 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Avoid over-relaxation of 8-bit immediates in integer arithmetic instructions.
Michael Kuperstein [Wed, 1 Jul 2015 10:54:42 +0000 (10:54 +0000)]
[X86] Avoid over-relaxation of 8-bit immediates in integer arithmetic instructions.

Only consider an instruction a candidate for relaxation if the last operand of the
instruction is an expression. We previously checked whether any operand is an expression,
which is useless, since for all instructions concerned, the only operand that may be
affected by relaxation is the last one.
In addition, this removes the check for having RIP as an argument, since it was
plain wrong - even when one of the arguments is RIP, relaxation may still be needed.

This fixes PR9807.

Patch by: david.l.kreitzer@intel.com
Differential Revision: http://reviews.llvm.org/D10766

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241152 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRevert part of r241149, "Fix PR23872: Integrated assembler error message when using...
NAKAMURA Takumi [Wed, 1 Jul 2015 10:28:09 +0000 (10:28 +0000)]
Revert part of r241149, "Fix PR23872: Integrated assembler error message when using .type directive with @ in AArch32 assembly."

The test should be split among targets. llvm/test/MC/ELF/ is assumed as X86.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241151 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[mips][microMIPS] Implement SLL and NOP instructions
Zoran Jovanovic [Wed, 1 Jul 2015 09:54:51 +0000 (09:54 +0000)]
[mips][microMIPS] Implement SLL and NOP instructions
http://reviews.llvm.org/D10474

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241150 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFix PR23872: Integrated assembler error message when using .type directive with ...
Gabor Ballabas [Wed, 1 Jul 2015 08:58:49 +0000 (08:58 +0000)]
Fix PR23872: Integrated assembler error message when using .type directive with @ in AArch32 assembly.

The AArch32 assembler parses the '@' as a comment symbol, so the error message shouldn't suggest
that '@<type>' is a valid replacement when assembling for AArch32 target.

Differential Revision: http://reviews.llvm.org/D10651

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241149 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[LoopUnroll] Use undef for phis with no value live
David Majnemer [Wed, 1 Jul 2015 05:38:07 +0000 (05:38 +0000)]
[LoopUnroll] Use undef for phis with no value live

We would create a phi node with a zero initialized operand instead of
undef in the case where no value was originally available.  This was
problematic for x86_mmx which has no null value.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241143 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[SCCP] Turn loads of null into undef instead of zero initialized values
David Majnemer [Wed, 1 Jul 2015 05:37:57 +0000 (05:37 +0000)]
[SCCP] Turn loads of null into undef instead of zero initialized values

Surprisingly, this is a correctness issue: the mmx type exists for
calling convention purposes, LLVM doesn't have a zero representation for
them.

This partially fixes PR23999.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241142 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[NaryReassociate] enhances nsw by leveraging @llvm.assume
Jingyue Wu [Wed, 1 Jul 2015 03:38:49 +0000 (03:38 +0000)]
[NaryReassociate] enhances nsw by leveraging @llvm.assume

Summary:
nsw are flaky and can often be removed by optimizations. This patch enhances
nsw by leveraging @llvm.assume in the IR. Specifically, NaryReassociate now
understands that

    assume(a + b >= 0) && assume(a >= 0) ==> a +nsw b

As a result, it can split more sext(a + b) into sext(a) + sext(b) for CSE.

Test Plan: nary-gep.ll

Reviewers: broune, meheff

Subscribers: jholewinski, llvm-commits

Differential Revision: http://reviews.llvm.org/D10822

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241139 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoGetting started docs: https, and check signature
JF Bastien [Wed, 1 Jul 2015 03:32:08 +0000 (03:32 +0000)]
Getting started docs: https, and check signature

Summary: Download should be over https, not insecure ftp at least for the signature and key files. The signature should also get verified.

Test Plan: None

Reviewers: chandlerc

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D10845

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241138 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[SanitizerCoverage] Don't add instrumentation to unreachable blocks.
Alexey Samsonov [Tue, 30 Jun 2015 23:11:45 +0000 (23:11 +0000)]
[SanitizerCoverage] Don't add instrumentation to unreachable blocks.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241127 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFix several typos in LangRef.rst related to loop unrolling metadata.
Mark Heffernan [Tue, 30 Jun 2015 22:48:51 +0000 (22:48 +0000)]
Fix several typos in LangRef.rst related to loop unrolling metadata.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241126 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[SEH] Add new intrinsics for recovering and restoring parent frames
Reid Kleckner [Tue, 30 Jun 2015 22:46:59 +0000 (22:46 +0000)]
[SEH] Add new intrinsics for recovering and restoring parent frames

The incoming EBP value established by the runtime is actually a pointer
to the end of the EH registration object, and not the true parent
function frame pointer. Clang doesn't need llvm.x86.seh.exceptioninfo
anymore because we know that the exception info pointer is at a fixed
offset from this incoming EBP.

The llvm.x86.seh.recoverfp intrinsic takes an EBP value provided by the
EH runtime and returns a pointer that is usable with llvm.framerecover.

The llvm.x86.seh.restoreframe intrinsic is inserted by the 32-bit
specific preparation pass in blocks targetted by the EH runtime. It
re-establishes any physical registers used by the parent function to
address the stack, such as the frame, base, and stack pointers.

Neither of these intrinsics correctly handle stack realignment prologues
yet, but it's possible to add that later.

Reviewers: majnemer

Differential Revision: http://reviews.llvm.org/D10848

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241125 91177308-0d34-0410-b5e6-96231b3b80d8