platform/upstream/llvm.git
11 years agoARM ABI: passing illegal vector types as varargs.
Manman Ren [Tue, 16 Oct 2012 19:18:39 +0000 (19:18 +0000)]
ARM ABI: passing illegal vector types as varargs.

We expand varargs in clang and the call site is handled in the back end, it is
hard to match exactly how illegal vectors are handled in the backend. Therefore,
we legalize the illegal vector types in clang:
if (Size <= 32), legalize to i32.
if (Size == 64), legalize to v2i32.
if (Size == 128), legalize to v4i32.
if (Size > 128), use indirect.

rdar://12439123

llvm-svn: 166043

11 years agoARM ABI: fix ABI alignment issues when passing legal vector types as varargs.
Manman Ren [Tue, 16 Oct 2012 19:01:37 +0000 (19:01 +0000)]
ARM ABI: fix ABI alignment issues when passing legal vector types as varargs.

We create an aligned temporary space and copy the content over from ap.cur to
the temporary space. This is necessary if the natural alignment of the type is
greater than the ABI alignment.

rdar://12439123

llvm-svn: 166040

11 years agoImplement GCC's -Wint-to-pointer-cast.
David Blaikie [Tue, 16 Oct 2012 18:53:14 +0000 (18:53 +0000)]
Implement GCC's -Wint-to-pointer-cast.

This implementation doesn't warn on anything that GCC doesn't warn on with the
exception of templates specializations (GCC doesn't warn, Clang does). The
specific skipped cases (boolean, constant expressions, enums) are open for
debate/adjustment if anyone wants to demonstrate that GCC is being overly
conservative here. The only really obvious false positive I found was in the
Clang regression suite's MPI test - apparently MPI uses specific flag values in
pointer constants. (eg: #define FOO (void*)~0)

llvm-svn: 166039

11 years agoIf a full path to a compiler is provided on the
Sean Callanan [Tue, 16 Oct 2012 18:22:04 +0000 (18:22 +0000)]
If a full path to a compiler is provided on the
command line to dotest.py, replace / with _ in
the logfile names that mention that compiler so
that we don't try to put log files in weird
places.

llvm-svn: 166038

11 years agoAnd now we can call the other 'get' method from this one and not duplicate the code.
Bill Wendling [Tue, 16 Oct 2012 18:20:09 +0000 (18:20 +0000)]
And now we can call the other 'get' method from this one and not duplicate the code.

llvm-svn: 166037

11 years agoSupport v8f32 to v8i8/vi816 conversion through custom lowering
Michael Liao [Tue, 16 Oct 2012 18:14:11 +0000 (18:14 +0000)]
Support v8f32 to v8i8/vi816 conversion through custom lowering

- Add custom FP_TO_SINT on v8i16 (and v8i8 which is legalized as v8i16 due to
  vector element-wise widening) to reduce DAG combiner and its overhead added
  in X86 backend.

llvm-svn: 166036

11 years agoUse the appropriate Attributes::get method to create an Attributes object.
Bill Wendling [Tue, 16 Oct 2012 18:06:06 +0000 (18:06 +0000)]
Use the appropriate Attributes::get method to create an Attributes object.

llvm-svn: 166035

11 years agoWhen using a symbol with attribute weakref, search for it first and
Joerg Sonnenberger [Tue, 16 Oct 2012 17:45:27 +0000 (17:45 +0000)]
When using a symbol with attribute weakref, search for it first and
don't try the normal GetOrCreateLLVM. The latter could drop the weak
atrtibute on the second reference, if there is no explicit declaration
of the aliasee.

llvm-svn: 166032

11 years agoSpeculative fix the mask constants to be of type uintptr_t. I don't know of any...
Owen Anderson [Tue, 16 Oct 2012 17:10:33 +0000 (17:10 +0000)]
Speculative fix the mask constants to be of type uintptr_t.  I don't know of any case where the old form was incorrect, but I'm more confident that such cases don't exist in this version.

llvm-svn: 166031

11 years agoMinor tweak to last patch along with a test case.
Fariborz Jahanian [Tue, 16 Oct 2012 17:08:11 +0000 (17:08 +0000)]
Minor tweak to last patch along with a test case.
// rdar://12491143

llvm-svn: 166030

11 years agoDo not call ReplaceCFAllocator() before __CFInitialize(), otherwise crashes are possi...
Alexander Potapenko [Tue, 16 Oct 2012 16:58:10 +0000 (16:58 +0000)]
Do not call ReplaceCFAllocator() before __CFInitialize(), otherwise crashes are possible on 10.8.
Fixes http://code.google.com/p/address-sanitizer/issues/detail?id=122

llvm-svn: 166029

11 years agoCommit a leftover file from r166026: Add MACOS_VERSION_MOUNTAIN_LION.
Alexander Potapenko [Tue, 16 Oct 2012 16:48:51 +0000 (16:48 +0000)]
Commit a leftover file from r166026: Add MACOS_VERSION_MOUNTAIN_LION.

llvm-svn: 166028

11 years agoAdd MACOS_VERSION_MOUNTAIN_LION.
Alexander Potapenko [Tue, 16 Oct 2012 16:24:49 +0000 (16:24 +0000)]
Add MACOS_VERSION_MOUNTAIN_LION.

llvm-svn: 166026

11 years agofixes an objc crash involving objc bool literal on
Fariborz Jahanian [Tue, 16 Oct 2012 16:21:20 +0000 (16:21 +0000)]
fixes an objc crash involving objc bool literal on
hopelessly poorly written code after spewing several
errors. // rdar://12491143

llvm-svn: 166025

11 years agoFix function parameter spelling in comments. Caught by -Wdocumentation.
Dmitri Gribenko [Tue, 16 Oct 2012 15:37:50 +0000 (15:37 +0000)]
Fix function parameter spelling in comments.  Caught by -Wdocumentation.

llvm-svn: 166024

11 years agoGNUstep runtime version default to 1.6, generate correct property attribute
David Chisnall [Tue, 16 Oct 2012 15:11:55 +0000 (15:11 +0000)]
GNUstep runtime version default to 1.6, generate correct property attribute
metadata.

llvm-svn: 166023

11 years agoThis patch addresses PR13949.
Bill Schmidt [Tue, 16 Oct 2012 13:30:53 +0000 (13:30 +0000)]
This patch addresses PR13949.

For the PowerPC 64-bit ELF Linux ABI, aggregates of size less than 8
bytes are to be passed in the low-order bits ("right-adjusted") of the
doubleword register or memory slot assigned to them.  A previous patch
addressed this for aggregates passed in registers.  However, small
aggregates passed in the overflow portion of the parameter save area are
still being passed left-adjusted.

The fix is made in PPCTargetLowering::LowerCall_Darwin_Or_64SVR4 on the
caller side, and in PPCTargetLowering::LowerFormalArguments_64SVR4 on
the callee side.  The main fix on the callee side simply extends
existing logic for 1- and 2-byte objects to 1- through 7-byte objects,
and correcting a constant left over from 32-bit code.  There is also a
fix to a bogus calculation of the offset to the following argument in
the parameter save area.

On the caller side, again a constant left over from 32-bit code is
fixed.  Additionally, some code for 1, 2, and 4-byte objects is
duplicated to handle the 3, 5, 6, and 7-byte objects for SVR4 only.  The
LowerCall_Darwin_Or_64SVR4 logic is getting fairly convoluted trying to
handle both ABIs, and I propose to separate this into two functions in a
future patch, at which time the duplication can be removed.

The patch adds a new test (structsinmem.ll) to demonstrate correct
passing of structures of all seven sizes.  Eight dummy parameters are
used to force these structures to be in the overflow portion of the
parameter save area.

As a side effect, this corrects the case when aggregates passed in
registers are saved into the first eight doublewords of the parameter
save area:  Previously they were stored left-justified, and now are
properly stored right-justified.  This requires changing the expected
output of existing test case structsinregs.ll.

llvm-svn: 166022

11 years agoisl scheduler: Do not fail when returning an empty band list
Tobias Grosser [Tue, 16 Oct 2012 07:29:19 +0000 (07:29 +0000)]
isl scheduler: Do not fail when returning an empty band list

The bug was within isl. To fix it, we simply update the isl version that
is used by Polly. We still have some changes within Polly to be able to
write a proper test case.

Reported-by: Sameer Sahasrabuddhe <Sameer.Sahasrabuddhe@amd.com>
llvm-svn: 166021

11 years agoisl-codegen: Support '<' and '>'
Tobias Grosser [Tue, 16 Oct 2012 07:29:13 +0000 (07:29 +0000)]
isl-codegen: Support '<' and '>'

Previously isl always generated '<=' or '>='. However, in many cases '<' or '>'
leads to simpler code. This commit updates isl and adds the relevant code
generation support to Polly.

llvm-svn: 166020

11 years agoTransform pattern:
Alexey Samsonov [Tue, 16 Oct 2012 07:22:28 +0000 (07:22 +0000)]
Transform pattern:
  if (CGM.getModuleDebugInfo())
    DebugInfo = CGM.getModuleDebugInfo()
into a call:
  maybeInitializeDebugInfo();

This is a simplification for a possible future fix of PR13942.

llvm-svn: 166019

11 years agoIssue:
Stepan Dyatkovskiy [Tue, 16 Oct 2012 07:16:47 +0000 (07:16 +0000)]
Issue:
Stack is formed improperly for long structures passed as byval arguments for
EABI mode.

If we took AAPCS reference, we can found the next statements:

A: "If the argument requires double-word alignment (8-byte), the NCRN (Next
Core Register Number) is rounded up to the next even register number." (5.5
Parameter Passing, Stage C, C.3).

B: "The alignment of an aggregate shall be the alignment of its most-aligned
component." (4.3 Composite Types, 4.3.1 Aggregates).

So if we have structure with doubles (9 double fields) and 3 Core unused
registers (r1, r2, r3): caller should use r2 and r3 registers only.
Currently r1,r2,r3 set is used, but it is invalid.

Callee VA routine should also use r2 and r3 regs only. All is ok here. This
behaviour is guessed by rounding up SP address with ADD+BFC operations.

Fix:
Main fix is in ARMTargetLowering::HandleByVal. If we detected AAPCS mode and
8 byte alignment, we waste odd registers then.

P.S.:
I also improved LDRB_POST_IMM regression test. Since ldrb instruction will
not generated by current regression test after this patch.

llvm-svn: 166018

11 years agoReapply r165661, Patch by Shuxin Yang <shuxin.llvm@gmail.com>.
NAKAMURA Takumi [Tue, 16 Oct 2012 06:28:34 +0000 (06:28 +0000)]
Reapply r165661, Patch by Shuxin Yang <shuxin.llvm@gmail.com>.

Original message:

The attached is the fix to radar://11663049. The optimization can be outlined by following rules:

   (select (x != c), e, c) -> select (x != c), e, x),
   (select (x == c), c, e) -> select (x == c), x, e)
where the <c> is an integer constant.

 The reason for this change is that : on x86, conditional-move-from-constant needs two instructions;
however, conditional-move-from-register need only one instruction.

  While the LowerSELECT() sounds to be the most convenient place for this optimization, it turns out to be a bad place. The reason is that by replacing the constant <c> with a symbolic value, it obscure some instruction-combining opportunities which would otherwise be very easy to spot. For that reason, I have to postpone the change to last instruction-combining phase.

  The change passes the test of "make check-all -C <build-root/test" and "make -C project/test-suite/SingleSource".

Original message since r165661:

My previous change has a bug: I negated the condition code of a CMOV, and go ahead creating a new CMOV using the *ORIGINAL* condition code.

llvm-svn: 166017

11 years agoCleanup whitespace.
Bill Wendling [Tue, 16 Oct 2012 06:10:45 +0000 (06:10 +0000)]
Cleanup whitespace.

llvm-svn: 166016

11 years agoFix a bug in the set(I,E)/reset(I,E) methods that I recently added. The boundary...
Owen Anderson [Tue, 16 Oct 2012 06:04:27 +0000 (06:04 +0000)]
Fix a bug in the set(I,E)/reset(I,E) methods that I recently added.  The boundary condition for checking if I and E were in the same word were incorrect, and, beyond that, the mask computation was not using a wide enough constant.

llvm-svn: 166015

11 years agoMove X86MCInstLower class definition into implementation file. It's not needed outside.
Craig Topper [Tue, 16 Oct 2012 06:01:50 +0000 (06:01 +0000)]
Move X86MCInstLower class definition into implementation file. It's not needed outside.

llvm-svn: 166014

11 years agoCleanup whitespace.
Bill Wendling [Tue, 16 Oct 2012 06:01:44 +0000 (06:01 +0000)]
Cleanup whitespace.

llvm-svn: 166013

11 years agoHave AttributesImpl defriend the Attributes class.
Bill Wendling [Tue, 16 Oct 2012 05:57:28 +0000 (05:57 +0000)]
Have AttributesImpl defriend the Attributes class.

llvm-svn: 166012

11 years agoHave AttrBuilder defriend the Attributes class.
Bill Wendling [Tue, 16 Oct 2012 05:55:09 +0000 (05:55 +0000)]
Have AttrBuilder defriend the Attributes class.

llvm-svn: 166011

11 years agoUse the Attributes::get method which takes an AttrVal value directly to simplify...
Bill Wendling [Tue, 16 Oct 2012 05:23:44 +0000 (05:23 +0000)]
Use the Attributes::get method which takes an AttrVal value directly to simplify the code a bit. No functionality change.

llvm-svn: 166010

11 years agoUse the Attributes::get method which takes an AttrVal value directly to simplify...
Bill Wendling [Tue, 16 Oct 2012 05:23:31 +0000 (05:23 +0000)]
Use the Attributes::get method which takes an AttrVal value directly to simplify the code a bit. No functionality change.

llvm-svn: 166009

11 years agoPut simple c'tors inline.
Bill Wendling [Tue, 16 Oct 2012 05:22:28 +0000 (05:22 +0000)]
Put simple c'tors inline.

llvm-svn: 166008

11 years agoPass in the context to the Attributes::get method.
Bill Wendling [Tue, 16 Oct 2012 05:20:51 +0000 (05:20 +0000)]
Pass in the context to the Attributes::get method.

llvm-svn: 166007

11 years ago[asan] fix lint
Kostya Serebryany [Tue, 16 Oct 2012 04:50:32 +0000 (04:50 +0000)]
[asan] fix lint

llvm-svn: 166006

11 years agoAdd a new ABI plugin method which specifies whether the architecture
Jason Molenda [Tue, 16 Oct 2012 02:39:21 +0000 (02:39 +0000)]
Add a new ABI plugin method which specifies whether the architecture
must push something on the stack for a function call or not.  In
x86, the stack pointer is decremented when the caller's pc is saved
on the stack.  In arm, the stack pointer and frame pointer don't
necessarily have to change for a function call, although most
functions need to use some stack space during their execution.

Use this information in the RegisterContextLLDB to detect invalid
unwind scenarios more accurately.

<rdar://problem/12348574>

llvm-svn: 166005

11 years agoFix filename in file header.
Craig Topper [Tue, 16 Oct 2012 02:21:30 +0000 (02:21 +0000)]
Fix filename in file header.

llvm-svn: 166004

11 years agoFix the cpu name and add -verify-machineinstrs.
Rafael Espindola [Tue, 16 Oct 2012 01:13:06 +0000 (01:13 +0000)]
Fix the cpu name and add -verify-machineinstrs.

llvm-svn: 166003

11 years ago[analyzer] ObjCContainersASTChecker: minor cleanup and an extra test case.
Jordan Rose [Tue, 16 Oct 2012 00:47:25 +0000 (00:47 +0000)]
[analyzer] ObjCContainersASTChecker: minor cleanup and an extra test case.

Follow-up to r165838, which fixed a potential crash.

llvm-svn: 166002

11 years agomisched: Added handleMove support for updating all kill flags, not just for allocatab...
Andrew Trick [Tue, 16 Oct 2012 00:22:51 +0000 (00:22 +0000)]
misched: Added handleMove support for updating all kill flags, not just for allocatable regs.

This is a medium term workaround until we have a more robust solution
in the form of a register liveness utility for postRA passes.

llvm-svn: 166001

11 years agoPatch from Matt Kopec <matt.kopec@intel.com> to fix the problem that if two breakpoin...
Jim Ingham [Tue, 16 Oct 2012 00:09:33 +0000 (00:09 +0000)]
Patch from Matt Kopec <matt.kopec@intel.com> to fix the problem that if two breakpoints were set on consecutive addresses, the continue from the
first breakpoint would skip the second.

llvm-svn: 166000

11 years agoRemove unused BitVectors from getAllocatableSet().
Jakob Stoklund Olesen [Tue, 16 Oct 2012 00:05:06 +0000 (00:05 +0000)]
Remove unused BitVectors from getAllocatableSet().

llvm-svn: 165999

11 years agoLTO also needs to initialize the TargetTransform infrastructure.
Nadav Rotem [Mon, 15 Oct 2012 22:50:02 +0000 (22:50 +0000)]
LTO also needs to initialize the TargetTransform infrastructure.

llvm-svn: 165997

11 years ago[analyzer] CmpRuns - cleaned up the API, added relative path getter
Anna Zaks [Mon, 15 Oct 2012 22:48:21 +0000 (22:48 +0000)]
[analyzer] CmpRuns - cleaned up the API, added relative path getter

llvm-svn: 165995

11 years ago[analyzer] Embed the analyzer version into the plist output.
Anna Zaks [Mon, 15 Oct 2012 22:48:19 +0000 (22:48 +0000)]
[analyzer] Embed the analyzer version into the plist output.

llvm-svn: 165994

11 years ago[analyzer] Enhance the error message.
Anna Zaks [Mon, 15 Oct 2012 22:48:17 +0000 (22:48 +0000)]
[analyzer] Enhance the error message.

llvm-svn: 165993

11 years ago[analyzer] Do not warn on direct ivar assignments within copy methods.
Anna Zaks [Mon, 15 Oct 2012 22:48:14 +0000 (22:48 +0000)]
[analyzer] Do not warn on direct ivar assignments within copy methods.

llvm-svn: 165992

11 years agoAdded "process plugin packet send" and "process plugin packet history" for GDB remote...
Greg Clayton [Mon, 15 Oct 2012 22:42:16 +0000 (22:42 +0000)]
Added "process plugin packet send" and "process plugin packet history" for GDB remote. "process plugin packet send" will send a packet and receive a response. "process plugin packet history" will dump the packet history buffer.

llvm-svn: 165991

11 years agoRemove RegisterClassInfo::isReserved() and isAllocatable().
Jakob Stoklund Olesen [Mon, 15 Oct 2012 22:41:03 +0000 (22:41 +0000)]
Remove RegisterClassInfo::isReserved() and isAllocatable().

Clients can use the equivalent functions in MRI.

llvm-svn: 165990

11 years agoAdd __builtin_setjmp/_longjmp supprt in X86 backend
Michael Liao [Mon, 15 Oct 2012 22:39:43 +0000 (22:39 +0000)]
Add __builtin_setjmp/_longjmp supprt in X86 backend

- Besides used in SjLj exception handling, __builtin_setjmp/__longjmp is also
  used as a light-weight replacement of setjmp/longjmp which are used to
  implementation continuation, user-level threading, and etc. The support added
  in this patch ONLY addresses this usage and is NOT intended to support SjLj
  exception handling as zero-cost DWARF exception handling is used by default
  in X86.

llvm-svn: 165989

11 years agoUn-revert r164907 and r164902 (+ follow-ups), 10.6 build fix to follow.
Daniel Dunbar [Mon, 15 Oct 2012 22:23:53 +0000 (22:23 +0000)]
Un-revert r164907 and r164902 (+ follow-ups), 10.6 build fix to follow.

llvm-svn: 165988

11 years agoplatform/clang_darwin: Drop an x86_64 slice into cc_kext_ios5.
Daniel Dunbar [Mon, 15 Oct 2012 22:23:34 +0000 (22:23 +0000)]
platform/clang_darwin: Drop an x86_64 slice into cc_kext_ios5.
 - This doesn't get used, but it is a simple workaround for PR14013.

llvm-svn: 165987

11 years agoUn-revert r164907 and r164902 (+ follow-ups), 10.6 build fix to follow.
Daniel Dunbar [Mon, 15 Oct 2012 22:23:32 +0000 (22:23 +0000)]
Un-revert r164907 and r164902 (+ follow-ups), 10.6 build fix to follow.

llvm-svn: 165986

11 years agoRemove LIS::isAllocatable() and isReserved() helpers.
Jakob Stoklund Olesen [Mon, 15 Oct 2012 22:14:34 +0000 (22:14 +0000)]
Remove LIS::isAllocatable() and isReserved() helpers.

All callers can simply use the corresponding MRI functions.

llvm-svn: 165985

11 years agoAdd range-based set()/reset() to BitVector. These allow fast setting/resetting of...
Owen Anderson [Mon, 15 Oct 2012 22:05:27 +0000 (22:05 +0000)]
Add range-based set()/reset() to BitVector.  These allow fast setting/resetting of ranges of bits, particularly useful when dealing with very large BitVector's.

llvm-svn: 165984

11 years agoSwitch most getReservedRegs() clients to the MRI equivalent.
Jakob Stoklund Olesen [Mon, 15 Oct 2012 21:57:41 +0000 (21:57 +0000)]
Switch most getReservedRegs() clients to the MRI equivalent.

Using the cached bit vector in MRI avoids comstantly allocating and
recomputing the reserved register bit vector.

llvm-svn: 165983

11 years agoFreeze the reserved registers as soon as isel is complete.
Jakob Stoklund Olesen [Mon, 15 Oct 2012 21:33:06 +0000 (21:33 +0000)]
Freeze the reserved registers as soon as isel is complete.

Also provide an MRI::getReservedRegs() function to access the frozen
register set, and isReserved() and isAllocatable() methods to test
individual registers.

The various implementations of TRI::getReservedRegs() are quite
complicated, and many passes need to look at the reserved register set.
This patch makes it possible for these passes to use the cached copy in
MRI, avoiding a lot of malloc traffic and repeated calculations.

llvm-svn: 165982

11 years agoARM: v1i64 and v2i64 VBSL intrinsic support.
Jim Grosbach [Mon, 15 Oct 2012 21:23:40 +0000 (21:23 +0000)]
ARM: v1i64 and v2i64 VBSL intrinsic support.

rdar://12502028

llvm-svn: 165981

11 years agoBuild with SDKROOT = macosx for "Debug" "DebugClang" and "Release" builds when buildi...
Greg Clayton [Mon, 15 Oct 2012 21:17:37 +0000 (21:17 +0000)]
Build with SDKROOT = macosx for "Debug" "DebugClang" and "Release" builds when building with Xcode.

llvm-svn: 165980

11 years agoFixed a logic error that showed up when compiling with a newer version of clang where:
Greg Clayton [Mon, 15 Oct 2012 21:16:43 +0000 (21:16 +0000)]
Fixed a logic error that showed up when compiling with a newer version of clang where:

lldb::BasicType
ClangASTContext::GetLLDBBasicTypeEnumeration (clang_type_t clang_type)

would return a bogus value.

llvm-svn: 165979

11 years agoAdd dependency on llvm-bcanalyzer from tests to the CMake build.
David Blaikie [Mon, 15 Oct 2012 21:11:46 +0000 (21:11 +0000)]
Add dependency on llvm-bcanalyzer from tests to the CMake build.

This fixes a CMake build break introduced by r165739.

Thanks Jan Voung for the quick suggestion/fix.

llvm-svn: 165978

11 years agoFixed CXXNamedCast right paren location propagation to instantiations.
Abramo Bagnara [Mon, 15 Oct 2012 21:08:58 +0000 (21:08 +0000)]
Fixed CXXNamedCast right paren location propagation to instantiations.

llvm-svn: 165977

11 years agoFixed LabelDecl source range.
Abramo Bagnara [Mon, 15 Oct 2012 21:07:44 +0000 (21:07 +0000)]
Fixed LabelDecl source range.

llvm-svn: 165976

11 years agoFixed ClassTemplateSpecializationDecl source range.
Abramo Bagnara [Mon, 15 Oct 2012 21:06:42 +0000 (21:06 +0000)]
Fixed ClassTemplateSpecializationDecl source range.

llvm-svn: 165975

11 years agoFixed FunctionTypeLoc range for trailing return type.
Abramo Bagnara [Mon, 15 Oct 2012 21:05:46 +0000 (21:05 +0000)]
Fixed FunctionTypeLoc range for trailing return type.

llvm-svn: 165974

11 years agoFixes location of overriding declaration with no comment
Fariborz Jahanian [Mon, 15 Oct 2012 20:57:52 +0000 (20:57 +0000)]
Fixes location of overriding declaration with no comment
of their own.

llvm-svn: 165972

11 years agoPatch from Daniel Malea -- Updated build instructions for Linux.
Greg Clayton [Mon, 15 Oct 2012 20:39:39 +0000 (20:39 +0000)]
Patch from Daniel Malea -- Updated build instructions for Linux.

llvm-svn: 165964

11 years agoRevert r164907 and r164902 (+ follow-ups). They broke building on 10.6.
Nico Weber [Mon, 15 Oct 2012 20:37:13 +0000 (20:37 +0000)]
Revert r164907 and r164902 (+ follow-ups). They broke building on 10.6.

See PR14013.

llvm-svn: 165963

11 years agoRevert r164907 and r164902 (+ follow-ups). They broke building on 10.6.
Nico Weber [Mon, 15 Oct 2012 20:37:01 +0000 (20:37 +0000)]
Revert r164907 and r164902 (+ follow-ups). They broke building on 10.6.

See PR14013.

llvm-svn: 165962

11 years agoMove the Attributes::Builder outside of the Attributes class and into its own class...
Bill Wendling [Mon, 15 Oct 2012 20:36:26 +0000 (20:36 +0000)]
Move the Attributes::Builder outside of the Attributes class and into its own class named AttrBuilder. No functionality change.

llvm-svn: 165961

11 years agoMove the Attributes::Builder outside of the Attributes class and into its own class...
Bill Wendling [Mon, 15 Oct 2012 20:35:56 +0000 (20:35 +0000)]
Move the Attributes::Builder outside of the Attributes class and into its own class named AttrBuilder. No functionality change.

llvm-svn: 165960

11 years agoCheck output of the misched unit tests
Andrew Trick [Mon, 15 Oct 2012 20:33:14 +0000 (20:33 +0000)]
Check output of the misched unit tests

llvm-svn: 165959

11 years agoAdd comments.
Bill Wendling [Mon, 15 Oct 2012 19:58:25 +0000 (19:58 +0000)]
Add comments.

llvm-svn: 165958

11 years ago[ms-inline asm] Rework the front-end to use the API introduced in r165946.
Chad Rosier [Mon, 15 Oct 2012 19:56:10 +0000 (19:56 +0000)]
[ms-inline asm]  Rework the front-end to use the API introduced in r165946.

-The front-end now builds a single assembly string and feeds it to the
 AsmParser.  The front-end iterates on a per statement basis by calling the
 ParseStatement() function.  Please note, the calling of ParseStatement() and
 and any notion of MCAsmParsedOperands will be sunk into the MC layer in the
 near future.  I plan to expose more basic APIs such as getClobbers, etc.

-The enumeration of the AsmString expressions have been reworked to use SMLocs
 rather than assembly Pieces, which were being parsed in the front-end.

-The test case, t8(), was modified due to r129223.  I'll have to find a way to
 work around things such as these.

Sorry for the large commit, but breaking this in multiple smaller commits proved
too irritating.

llvm-svn: 165957

11 years agoAdd a cpu to try to fix the atom builder.
Rafael Espindola [Mon, 15 Oct 2012 19:25:43 +0000 (19:25 +0000)]
Add a cpu to try to fix the atom builder.

llvm-svn: 165956

11 years ago[ms-inline asm] If we parsed a statement and the opcode is valid, then it's an instru...
Chad Rosier [Mon, 15 Oct 2012 19:08:18 +0000 (19:08 +0000)]
[ms-inline asm] If we parsed a statement and the opcode is valid, then it's an instruction.

llvm-svn: 165955

11 years agoAdd testcase for pr14088.
Rafael Espindola [Mon, 15 Oct 2012 19:00:10 +0000 (19:00 +0000)]
Add testcase for pr14088.

llvm-svn: 165954

11 years agostructured document comment: patch to provide comment for overriding function
Fariborz Jahanian [Mon, 15 Oct 2012 18:58:50 +0000 (18:58 +0000)]
structured document comment: patch to provide comment for overriding function
template when comment is comming from overridden declaration.
// rdar://12378793

llvm-svn: 165953

11 years agomisched tests: add a triple to speculatively fix windows builders.
Andrew Trick [Mon, 15 Oct 2012 18:21:08 +0000 (18:21 +0000)]
misched tests: add a triple to speculatively fix windows builders.

llvm-svn: 165952

11 years agoMake sure we iterate over newly created instructions. Fixes pr13625. Testcase to
Rafael Espindola [Mon, 15 Oct 2012 18:21:07 +0000 (18:21 +0000)]
Make sure we iterate over newly created instructions. Fixes pr13625. Testcase to
follow in one sec.

llvm-svn: 165951

11 years agomisched: ILP scheduler for experimental heuristics.
Andrew Trick [Mon, 15 Oct 2012 18:02:27 +0000 (18:02 +0000)]
misched: ILP scheduler for experimental heuristics.

llvm-svn: 165950

11 years agoAdd an entry in CREDITS.TXT
Argyrios Kyrtzidis [Mon, 15 Oct 2012 17:34:53 +0000 (17:34 +0000)]
Add an entry in CREDITS.TXT

llvm-svn: 165949

11 years ago[ms-inline asm] Update the end loc for ParseIntelMemOperand.
Chad Rosier [Mon, 15 Oct 2012 17:26:38 +0000 (17:26 +0000)]
[ms-inline asm] Update the end loc for ParseIntelMemOperand.

llvm-svn: 165947

11 years ago[ms-inline asm] Add a few new APIs to the AsmParser class in support of MS-Style
Chad Rosier [Mon, 15 Oct 2012 17:19:13 +0000 (17:19 +0000)]
[ms-inline asm] Add a few new APIs to the AsmParser class in support of MS-Style
inline assembly.  For the time being, these will be called directly by clang.
However, in the near future I expect these to be sunk back into the MC layer
and more basic APIs (e.g., getClobbers(), getConstraints(), etc.) will be called
by clang.

llvm-svn: 165946

11 years ago[ms-inline asm] Use incoming argument rather than hard coding to false.
Chad Rosier [Mon, 15 Oct 2012 16:50:34 +0000 (16:50 +0000)]
[ms-inline asm] Use incoming argument rather than hard coding to false.

llvm-svn: 165945

11 years agoFix a typo in bitcode docs, from 165814.
Jan Wen Voung [Mon, 15 Oct 2012 16:47:58 +0000 (16:47 +0000)]
Fix a typo in bitcode docs, from 165814.

llvm-svn: 165944

11 years agoTeach TargetInfo to hold on to the TargetOptions with which it was
Douglas Gregor [Mon, 15 Oct 2012 16:45:32 +0000 (16:45 +0000)]
Teach TargetInfo to hold on to the TargetOptions with which it was
created.

llvm-svn: 165943

11 years agoResubmit the changes to llvm core to update the functions to support different pointe...
Micah Villmow [Mon, 15 Oct 2012 16:24:29 +0000 (16:24 +0000)]
Resubmit the changes to llvm core to update the functions to support different pointer sizes on a per address space basis.

llvm-svn: 165941

11 years agoPowerPC: add EmitTCEntry class for TOC creation
Adhemerval Zanella [Mon, 15 Oct 2012 15:43:14 +0000 (15:43 +0000)]
PowerPC: add EmitTCEntry class for TOC creation

This patch replaces the EmitRawText by a EmitTCEntry class (specialized for
each Streamer) in PowerPC64 TOC entry creation.

llvm-svn: 165940

11 years agoImplement internal_memmove.
Alexander Potapenko [Mon, 15 Oct 2012 15:34:41 +0000 (15:34 +0000)]
Implement internal_memmove.
Use internal_memmove() and internal_memcpy() in the memcpy() and memmove() wrappers
when building the dynamic runtime (OS X only), to work around a bug in resolver functions wrapping.
See also http://code.google.com/p/address-sanitizer/issues/detail?id=116

llvm-svn: 165939

11 years ago[asan] fix a test
Kostya Serebryany [Mon, 15 Oct 2012 14:30:30 +0000 (14:30 +0000)]
[asan] fix a test

llvm-svn: 165938

11 years ago[asan] make AddressSanitizer to be a FunctionPass instead of ModulePass. clang part...
Kostya Serebryany [Mon, 15 Oct 2012 14:22:56 +0000 (14:22 +0000)]
[asan] make AddressSanitizer to be a FunctionPass instead of ModulePass. clang part: for FunctionPass we need to run asan at a different point, otherwise it will run before inlining

llvm-svn: 165937

11 years ago[asan] make AddressSanitizer to be a FunctionPass instead of ModulePass. This will...
Kostya Serebryany [Mon, 15 Oct 2012 14:20:06 +0000 (14:20 +0000)]
[asan] make AddressSanitizer to be a FunctionPass instead of ModulePass. This will simplify chaining other FunctionPasses with asan. Also some minor cleanup

llvm-svn: 165936

11 years ago[asan] fix more tests
Kostya Serebryany [Mon, 15 Oct 2012 13:30:38 +0000 (13:30 +0000)]
[asan] fix more tests

llvm-svn: 165934

11 years agoOnly report first error when no compilation database is found.
Daniel Jasper [Mon, 15 Oct 2012 13:12:24 +0000 (13:12 +0000)]
Only report first error when no compilation database is found.

Review: http://llvm-reviews.chandlerc.com/D62
llvm-svn: 165933

11 years ago[asan] change the asan output slightly to avoid user confusion: a) add ':' after...
Kostya Serebryany [Mon, 15 Oct 2012 13:04:58 +0000 (13:04 +0000)]
[asan] change the asan output slightly to avoid user confusion: a) add ':' after 'AddressSanitizer' and b) changed 'crashed' to 'SEGV'

llvm-svn: 165932

11 years agoUpdate the memcpy rewriting to fully support widened int rewriting. This
Chandler Carruth [Mon, 15 Oct 2012 10:24:43 +0000 (10:24 +0000)]
Update the memcpy rewriting to fully support widened int rewriting. This
includes extracting ints for copying elsewhere and inserting ints when
copying into the alloca. This should fix the CanSROA assertion coming
out of Clang's regression test suite.

llvm-svn: 165931

11 years agoFollow-up fix to r165928: handle memset rewriting for widened integers,
Chandler Carruth [Mon, 15 Oct 2012 10:24:40 +0000 (10:24 +0000)]
Follow-up fix to r165928: handle memset rewriting for widened integers,
and generally clean up the memset handling. It had rotted a bit as the
other rewriting logic got polished more.

llvm-svn: 165930

11 years agoFixed PR13938: the ARM backend was crashing because it couldn't select a VDUPLANE...
Silviu Baranga [Mon, 15 Oct 2012 09:41:32 +0000 (09:41 +0000)]
Fixed PR13938: the ARM backend was crashing because it couldn't select a VDUPLANE node with the vector input size different from the output size. This was bacause the BUILD_VECTOR lowering code didn't check that the size of the input vector was correct for using VDUPLANE.

llvm-svn: 165929

11 years agoFirst major step toward addressing PR14059. This teaches SROA to handle
Chandler Carruth [Mon, 15 Oct 2012 08:40:30 +0000 (08:40 +0000)]
First major step toward addressing PR14059. This teaches SROA to handle
cases where we have partial integer loads and stores to an otherwise
promotable alloca to widen[1] those loads and stores to cover the entire
alloca and bitcast them into the appropriate type such that promotion
can proceed.

These partial loads and stores stem from an annoying confluence of ARM's
calling convention and ABI lowering and the FCA pre-splitting which
takes place in SROA. Clang lowers a { double, double } in-register
function argument as a [4 x i32] function argument to ensure it is
placed into integer 32-bit registers (a really unnerving implicit
contract between Clang and the ARM backend I would add). This results in
a FCA load of [4 x i32]* from the { double, double } alloca, and SROA
decomposes this into a sequence of i32 loads and stores. Inlining
proceeds, code gets folded, but at the end of the day, we still have i32
stores to the low and high halves of a double alloca. Widening these to
be i64 operations, and bitcasting them to double prior to loading or
storing allows promotion to proceed for these allocas.

I looked quite a bit changing the IR which Clang produces for this case
to be more friendly, but small changes seem unlikely to help. I think
the best representation we could use currently would be to pass 4 i32
arguments thereby avoiding any FCAs, but that would still require this
fix. It seems like it might eventually be nice to somehow encode the ABI
register selection choices outside of the parameter type system so that
the parameter can be a { double, double }, but the CC register
annotations indicate that this should be passed via 4 integer registers.

This patch does not address the second problem in PR14059, which is the
reverse: when a struct alloca is loaded as a *larger* single integer.

This patch also does not address some of the code quality issues with
the FCA-splitting. Those don't actually impede any optimizations really,
but they're on my list to clean up.

[1]: Pedantic footnote: for those concerned about memory model issues
here, this is safe. For the alloca to be promotable, it cannot escape or
have any use of its address that could allow these loads or stores to be
racing. Thus, widening is always safe.

llvm-svn: 165928

11 years agoHoist the canConvertValue predicate and the convertValue transform out
Chandler Carruth [Mon, 15 Oct 2012 08:40:22 +0000 (08:40 +0000)]
Hoist the canConvertValue predicate and the convertValue transform out
into static helper functions. They're really quite generic and are going
to be needed elsewhere shortly.

llvm-svn: 165927

11 years agoUse enum values instead of magic numbers for indexing into the attribute list.
Bill Wendling [Mon, 15 Oct 2012 07:31:59 +0000 (07:31 +0000)]
Use enum values instead of magic numbers for indexing into the attribute list.

llvm-svn: 165925