Robert Lytton [Mon, 6 Jan 2014 14:20:53 +0000 (14:20 +0000)]
XCore target: Lower RETURNADDR
Only handles a depth of zero (the same as FRAMEADDR)
llvm-svn: 198613
Robert Lytton [Mon, 6 Jan 2014 14:20:47 +0000 (14:20 +0000)]
XCore target: Optimise entsp / retsp selection
llvm-svn: 198612
Robert Lytton [Mon, 6 Jan 2014 14:20:41 +0000 (14:20 +0000)]
XCore target: Refactor LR handling
We also narrow the liveness of FP & LR during the prologue to
reflect the actual usage of the registers.
I have been unable to construct a test to prove the previous live
range was too large.
llvm-svn: 198611
Robert Lytton [Mon, 6 Jan 2014 14:20:37 +0000 (14:20 +0000)]
XCore target: Refactor the loading of constants into a register
This common functionality will be used to lower FRAME_TO_ARGS_OFFSET.
llvm-svn: 198610
Robert Lytton [Mon, 6 Jan 2014 14:20:32 +0000 (14:20 +0000)]
XCore target: fix handling of unsized global arrays in large code model
llvm-svn: 198609
Marshall Clow [Mon, 6 Jan 2014 14:00:09 +0000 (14:00 +0000)]
Rename several internal templates to get rid of ___ (triple underscores) or worse, four. No functionality change.
llvm-svn: 198608
Alp Toker [Mon, 6 Jan 2014 12:54:51 +0000 (12:54 +0000)]
Correct underlying integer type of enum TokenKind
This matches up the underlying type against the actual storage type 'unsigned
short' and lets us get rid of some casts while we're at it.
Effort is made to keep this building in pre-C++11 but as with other features
Token will be less efficiently packed in in legacy configurations.
llvm-svn: 198607
Alp Toker [Mon, 6 Jan 2014 12:54:41 +0000 (12:54 +0000)]
Simplify diagnostic tag type using the token kind formatter
As far as the parser is concerned the tag type is always a keyword.
llvm-svn: 198606
Alp Toker [Mon, 6 Jan 2014 12:54:32 +0000 (12:54 +0000)]
Don't use magic constants in the digraph diagnostic
llvm-svn: 198605
Alp Toker [Mon, 6 Jan 2014 12:54:18 +0000 (12:54 +0000)]
Support diagnostic formatting of keyword tokens
Implemented with a new getKeywordSpelling() accessor. Unlike getTokenName() the
result of this function is stable and may be used in diagnostic output.
Uses of this feature are split out into the subsequent commit.
llvm-svn: 198604
Alp Toker [Mon, 6 Jan 2014 12:54:07 +0000 (12:54 +0000)]
Rename getTokenSimpleSpelling() to getPunctuatorSpelling()
That's what it does, what the documentation says it does and what callers
expect it to do.
llvm-svn: 198603
Tim Northover [Mon, 6 Jan 2014 12:00:44 +0000 (12:00 +0000)]
ARM: keep special non-AEABIness of "-darwin-eabi" triples for now
Longer term, we want to move users to "*-*-*-macho" for embedded work, but for
now people are relying on the last thing we told them, which is unfortunately
"*-*-darwin-eabi".
rdar://problem/
15703934
llvm-svn: 198602
Alp Toker [Mon, 6 Jan 2014 11:31:18 +0000 (11:31 +0000)]
Highlight the previous underlying enum type when diagnosing a mismatch
enum-scoped.cpp:93:6: error: enumeration redeclared with different underlying type 'short' (was 'int')
enum Redeclare6 : short;
^
enum-scoped.cpp:92:6: note: previous declaration is here
enum Redeclare6 : int;
^ ~~~
The redeclaration source range is still missing but this is a step forward,
potentially edging towards a FixIt.
llvm-svn: 198601
Alp Toker [Mon, 6 Jan 2014 11:31:06 +0000 (11:31 +0000)]
Diagnose enum redeclarations properly
In all three checks, the note indicates a previous declaration and never a 'use'.
Before:
enum-scoped.cpp:92:6: note: previous use is here
enum Redeclare6 : int;
^
After:
enum-scoped.cpp:92:6: note: previous declaration is here
enum Redeclare6 : int;
^
llvm-svn: 198600
Alp Toker [Mon, 6 Jan 2014 11:30:41 +0000 (11:30 +0000)]
Use token kind instead of '%select{.|->}0' in diagnostic
llvm-svn: 198599
Alp Toker [Mon, 6 Jan 2014 11:30:15 +0000 (11:30 +0000)]
Apply some LLVM_READONLY / LLVM_READNONE on diagnostic functions
llvm-svn: 198598
Joey Gouly [Mon, 6 Jan 2014 11:26:18 +0000 (11:26 +0000)]
[OpenCL] Produce an error if an address space is used on the return
type of a function.
llvm-svn: 198597
Tobias Grosser [Mon, 6 Jan 2014 09:53:00 +0000 (09:53 +0000)]
www: Fix typo II
llvm-svn: 198596
Tobias Grosser [Mon, 6 Jan 2014 09:52:42 +0000 (09:52 +0000)]
www: Fix typo
llvm-svn: 198595
Tobias Grosser [Mon, 6 Jan 2014 09:51:38 +0000 (09:51 +0000)]
www: Add link to newly released islplot library
llvm-svn: 198594
Elena Demikhovsky [Mon, 6 Jan 2014 08:45:54 +0000 (08:45 +0000)]
AVX-512: added intrinsic vcvtpd2ps (with rounding mode and without)
llvm-svn: 198593
Venkatraman Govindaraju [Mon, 6 Jan 2014 08:24:44 +0000 (08:24 +0000)]
[Sparc] Explicitly cast -1 to unsigned to fix buildbot errors.
llvm-svn: 198592
Venkatraman Govindaraju [Mon, 6 Jan 2014 08:08:58 +0000 (08:08 +0000)]
[Sparc] Add initial implementation of disassembler for sparc
llvm-svn: 198591
David Majnemer [Mon, 6 Jan 2014 07:39:46 +0000 (07:39 +0000)]
MC: Fatally error if subtraction operand is bad
Instead of crashing, raise an error when a subtraction expression
involves an undefined symbol.
This fixes PR18375.
llvm-svn: 198590
Craig Topper [Mon, 6 Jan 2014 06:57:27 +0000 (06:57 +0000)]
The rest of r198588. Remove SegOvrBits from X86 TSFlags since they weren't being used.
llvm-svn: 198589
Craig Topper [Mon, 6 Jan 2014 06:51:58 +0000 (06:51 +0000)]
Remove SegOvrBits from X86 TSFlags since they weren't being used.
llvm-svn: 198588
Craig Topper [Mon, 6 Jan 2014 06:09:03 +0000 (06:09 +0000)]
Remove argument to fix build bot failure.
llvm-svn: 198587
Craig Topper [Mon, 6 Jan 2014 06:02:58 +0000 (06:02 +0000)]
Add OpSize16 bit, for instructions which need 0x66 prefix in 16-bit mode
The 0x66 prefix toggles between 16-bit and 32-bit addressing mode.
So in 32-bit mode it is used to switch to 16-bit addressing mode for the
following instruction, while in 16-bit mode it's the other way round — it's
used to switch to 32-bit mode instead.
Thus, emit the 0x66 prefix byte for OpSize only in 32-bit (and 64-bit) mode,
and introduce a new OpSize16 bit which is used in 16-bit mode instead.
This is just the basic infrastructure for that change; a subsequent patch
will add the new OpSize16 bit to the 32-bit instructions that need it.
Patch from David Woodhouse.
llvm-svn: 198586
Bill Wendling [Mon, 6 Jan 2014 06:00:00 +0000 (06:00 +0000)]
Remove unnecessary #includes.
llvm-svn: 198585
Craig Topper [Mon, 6 Jan 2014 04:55:54 +0000 (04:55 +0000)]
[x86] Add basic support for .code16
This is not really expected to work right yet. Mostly because we will
still emit the OpSize (0x66) prefix in all the wrong places, along with
a number of other corner cases. Those will all be fixed in the subsequent
commits.
Patch from David Woodhouse.
llvm-svn: 198584
Saleem Abdulrasool [Mon, 6 Jan 2014 04:14:03 +0000 (04:14 +0000)]
Support: add reference for ARM EHABI
Add a reference to the ARM EHABI Specification as a follow up to SVN r198576
llvm-svn: 198583
Kevin Qin [Mon, 6 Jan 2014 02:26:10 +0000 (02:26 +0000)]
[AArch64 NEON] Fix invalid constant used in vselect condition.
There is a wrong assumption that the vector element type and the
type of each ConstantSDNode in the build_vector were the same.
However, when promoting the integer operand of a legally typed
build_vector, the operand type and the vector element type do not
need to be the same
(See method 'DAGTypeLegalizer::PromoteIntOp_BUILD_VECTOR' in
LegalizeIntegerTypes.cpp).
in AArch64 backend, the following dag sequence:
C0: i1 = Constant<0>
C1: i1 = Constant<-1>
V: v8i1 = BUILD_VECTOR C1, C1, C0, C0, C0, C0, C0, C0
is type-legalized into:
NewC0: i32 = Constant<0>
NewC1: i32 = Constant<1>
V: v8i8 = BUILD_VECTOR NewC1, NewC1, NewC0, NewC0, NewC0, NewC0, NewC0, NewC0
Forcing a getZeroExtend to VTBits to ensure that the new constant
is correctly.
llvm-svn: 198582
Tobias Grosser [Mon, 6 Jan 2014 01:37:13 +0000 (01:37 +0000)]
Temporarily reformat Polly to silence buildbots
We may revert this depending on how the current discussion on llvm-commits
ends.
llvm-svn: 198581
Venkatraman Govindaraju [Mon, 6 Jan 2014 01:22:54 +0000 (01:22 +0000)]
[Sparc] Add ELF Object Writer for Sparc.
llvm-svn: 198580
Bill Wendling [Mon, 6 Jan 2014 00:43:20 +0000 (00:43 +0000)]
Refactor function that checks that __builtin_returnaddress's argument is constant.
This moves the check up into the parent class so that all targets can use it
without having to copy (and keep in sync) the same error message.
llvm-svn: 198579
Bill Wendling [Mon, 6 Jan 2014 00:43:09 +0000 (00:43 +0000)]
Remove a failing test to get the buildbots back to green.
llvm-svn: 198578
Bill Wendling [Mon, 6 Jan 2014 00:43:04 +0000 (00:43 +0000)]
Try to fix s390x build bot.
llvm-svn: 198577
Saleem Abdulrasool [Mon, 6 Jan 2014 00:15:00 +0000 (00:15 +0000)]
ARM: move ARMUnwindOp.h into Support
Move the ARM EHABI unwind opcode definitions from the ARM MCTargetDesc into LLVM
Support. This enables sharing of the definitions across the ARM target code as
well as llvm-readobj. This will allow implementation of the unwind decoding in
llvm-readobj.
llvm-svn: 198576
Aaron Ballman [Sun, 5 Jan 2014 21:08:29 +0000 (21:08 +0000)]
This helper method isn't needed, and it's unsafe for it to use StringRef in the first place. Replaced the unsafe code with the proper accessor.
llvm-svn: 198569
Benjamin Kramer [Sun, 5 Jan 2014 20:26:05 +0000 (20:26 +0000)]
SPARC: Make helper function static.
llvm-svn: 198567
Craig Topper [Sun, 5 Jan 2014 19:40:56 +0000 (19:40 +0000)]
Fix ModR/M byte output for 16-bit addressing modes (PR18220)
Add some tests to validate correct register selection, including a fix
to an existing test which was requiring the *wrong* output.
Patch from David Woodhouse.
llvm-svn: 198566
Venkatraman Govindaraju [Sun, 5 Jan 2014 19:25:57 +0000 (19:25 +0000)]
ELF relocation types for sparc.
llvm-svn: 198565
Craig Topper [Sun, 5 Jan 2014 19:25:13 +0000 (19:25 +0000)]
Remove opcode from MOV32r0 that I accidentally left when I converted it to Pseudo. Remove FIXME as well.
llvm-svn: 198564
Saleem Abdulrasool [Sun, 5 Jan 2014 16:36:37 +0000 (16:36 +0000)]
ARM: style changes to LDRD, STRD definition
Fix indentation, name registers similar to ARM ARM.
No functionality change!
llvm-svn: 198563
Elena Demikhovsky [Sun, 5 Jan 2014 14:21:07 +0000 (14:21 +0000)]
AVX-512: changed property name from "neverHasSideEffects=1" to "hasSideEffects=0", added this property to VMOVSS/VMOVSD;
Optimized a truncate pattern.
llvm-svn: 198562
Simon Atanasyan [Sun, 5 Jan 2014 13:40:27 +0000 (13:40 +0000)]
[Mips] Add support for DT_MIPS_RLD_MAP and DT_MIPS_PLTGOT dynamic
section tags to the llvm-readobj.
llvm-svn: 198561
Simon Atanasyan [Sun, 5 Jan 2014 13:40:17 +0000 (13:40 +0000)]
[Mips] Rename the test case input file. No functional changes.
llvm-svn: 198560
Daniel Jasper [Sun, 5 Jan 2014 13:23:23 +0000 (13:23 +0000)]
clang-format: Spacing inside enum braces.
Before (in Google style):
enum ShortEnum {A, B, C};
After:
enum ShortEnum { A, B, C };
llvm-svn: 198559
Daniel Jasper [Sun, 5 Jan 2014 12:38:10 +0000 (12:38 +0000)]
clang-format: Allow formatting short enums on a single line.
Before:
enum ShortEnum {
A,
B,
C
};
After:
enum ShortEnum { A, B, C };
This seems to be the predominant choice in LLVM/Clang as well as in
Google style.
llvm-svn: 198558
Elena Demikhovsky [Sun, 5 Jan 2014 10:46:09 +0000 (10:46 +0000)]
AVX-512: Added more intrinsics for convert and min/max.
Removed vzeroupper from AVX-512 mode - our optimization gude does not recommend to insert vzeroupper at all.
llvm-svn: 198557
Chandler Carruth [Sun, 5 Jan 2014 10:38:52 +0000 (10:38 +0000)]
[PM] Add a definition for the static PassID in the CallGraphAnalysis.
Missed this when adding the skeleton analysis. Caught by a build break
in the next patch I'm working on when trying to use the analysis.
llvm-svn: 198556
Chandler Carruth [Sun, 5 Jan 2014 09:14:53 +0000 (09:14 +0000)]
Add in a unittest for the one-use pattern matcher.
llvm-svn: 198552
Craig Topper [Sun, 5 Jan 2014 07:16:04 +0000 (07:16 +0000)]
Add the other form of movq xmm,xmm for the disassembler.
llvm-svn: 198551
Craig Topper [Sun, 5 Jan 2014 06:55:48 +0000 (06:55 +0000)]
Use patterns to remove some duplicate instructions.
llvm-svn: 198550
Alp Toker [Sun, 5 Jan 2014 06:38:57 +0000 (06:38 +0000)]
Fix 'declartion' typos
llvm-svn: 198549
Alp Toker [Sun, 5 Jan 2014 06:38:18 +0000 (06:38 +0000)]
Pre-declare '::type_info' in MicrosoftMode only, not MicrosoftExt
It was previously enabled in both but should only have been part of the drop-in
quirks mode that is 'MicrosoftMode' given that it's only useful for
compatibility with the Microsoft headers/runtime.
llvm-svn: 198548
Craig Topper [Sun, 5 Jan 2014 05:46:38 +0000 (05:46 +0000)]
Fix encoding for PUSH64i16. Add In64BitMode Predicate. Remove disassembler hack.
llvm-svn: 198547
Craig Topper [Sun, 5 Jan 2014 05:10:07 +0000 (05:10 +0000)]
Remove no longer needed x86 disassembler hack.
llvm-svn: 198546
Craig Topper [Sun, 5 Jan 2014 04:55:55 +0000 (04:55 +0000)]
Mark x86 _alt instructions as AsmParserOnly so they will be omitted from disassembler without string matches.
llvm-svn: 198545
Craig Topper [Sun, 5 Jan 2014 04:32:42 +0000 (04:32 +0000)]
Use new ForceDisassemble flag on the 2-byte forms of INC/DEC for 32-bit mode and remove disassmbler table emitter hack.
llvm-svn: 198544
Craig Topper [Sun, 5 Jan 2014 04:17:28 +0000 (04:17 +0000)]
Add a new x86 specific instruction flag to force some isCodeGenOnly instructions to go through to the disassembler tables without resorting to string matches. Apply flag to all _REV instructions.
llvm-svn: 198543
Alp Toker [Sun, 5 Jan 2014 04:17:27 +0000 (04:17 +0000)]
Tweak the parse recovery in r198540
Cover a hypothetical case when we might not have reached the final argument
declaration for some reason during recovery, and split out for readability.
llvm-svn: 198542
Chandler Carruth [Sun, 5 Jan 2014 03:28:29 +0000 (03:28 +0000)]
Add support to the pattern match library for matching NSW and NUW
instructions. I needed this for a quick experiment I was making, and
while I've no idea if that will ever get committed, I didn't want to
throw away the pattern match code and for anyone else to have to write
it again. I've added unittests to make sure this works correctly.
In fun news, this also uncovered the IRBuilder bug. Doh!
llvm-svn: 198541
Alp Toker [Sun, 5 Jan 2014 03:27:57 +0000 (03:27 +0000)]
Fix bungled parse recovery in K&R function declarations
void knrNoSemi(i) int i { }
Adherents of The C Programming Language unfortunate enough to miss a semicolon
as above would be met with a cascade of errors spanning the remainder of the
TU.
This patch introduces a beautiful parse error recovery, complete with helpful
FixIt to restore sanity.
Before (output redacted for brevity):
error: 'error' diagnostics seen but not expected:
File declarators.c Line 119: declaration does not declare a parameter
File declarators.c Line 123: declaration does not declare a parameter
File declarators.c Line 127: parameter named 'func_E12' is missing
File declarators.c Line 127: expected ';' at end of declaration
File declarators.c Line 133: parameter named 'func_E13' is missing
File declarators.c Line 133: expected ';' at end of declaration
File declarators.c Line 139: parameter named 'func_E14' is missing
File declarators.c Line 139: expected ';' at end of declaration
File declarators.c Line 145: parameter named 'func_E15' is missing
File declarators.c Line 145: expected ';' at end of declaration
File declarators.c Line 150: expected function body after function declarator
File declarators.c Line 119: declaration of 'enum E11' will not be visible outside of this function
File declarators.c Line 123: declaration of 'enum E12' will not be visible outside of this function
File declarators.c Line 133: ISO C forbids forward references to 'enum' types
File declarators.c Line 133: declaration of 'enum E13' will not be visible outside of this function
File declarators.c Line 139: ISO C forbids forward references to 'enum' types
File declarators.c Line 139: declaration of 'enum E14' will not be visible outside of this function
File declarators.c Line 145: ISO C forbids forward references to 'enum' types
File declarators.c Line 145: declaration of 'enum E15' will not be visible outside of this function
...
After:
declarators.c:103:24: error: expected ';' at end of declaration
void knrNoSemi(i) int i { }
^
;
Patch found in a sealed envelope dated 1978 with the message "Do not open until
January 2014"
llvm-svn: 198540
Alp Toker [Sun, 5 Jan 2014 03:27:11 +0000 (03:27 +0000)]
Parse: Token consumption modernization and loop de-nesting
Cleanup only.
llvm-svn: 198539
Chandler Carruth [Sun, 5 Jan 2014 03:22:33 +0000 (03:22 +0000)]
Fix a bug in IRBuilder that's been there for who knows how long. It
failed to correctly propagate the NUW and NSW flags to the constant
folder for two instructions. I've added a unittest to cover flag
propagation for the rest of the instructions and constant expressions.
llvm-svn: 198538
Bill Wendling [Sun, 5 Jan 2014 03:10:56 +0000 (03:10 +0000)]
Attempt to fix buildbots by XFAILing some architectures.
llvm-svn: 198537
Venkatraman Govindaraju [Sun, 5 Jan 2014 03:07:04 +0000 (03:07 +0000)]
Add lit.local.cfg for MC/Sparc
llvm-svn: 198536
Rui Ueyama [Sun, 5 Jan 2014 02:41:07 +0000 (02:41 +0000)]
Replace nested switch statements.
Differential Revision: http://llvm-reviews.chandlerc.com/D2501
llvm-svn: 198535
Chandler Carruth [Sun, 5 Jan 2014 02:23:11 +0000 (02:23 +0000)]
Use a shorter name for the IRBuilder member. This will help the tests
I'm adding next be a lot more readable.
llvm-svn: 198534
Venkatraman Govindaraju [Sun, 5 Jan 2014 02:13:48 +0000 (02:13 +0000)]
[Sparc] Add initial implementation of MC Code emitter for sparc.
llvm-svn: 198533
Chandler Carruth [Sun, 5 Jan 2014 02:07:20 +0000 (02:07 +0000)]
Simplify the PatternMatch unittest by giving it a module, function, and
basic block to hold instructions, and managing all of their lifetimes in
a fixture. This makes it easy to sink the expectations into the test
cases themselves which also makes things a bit more explicit and clearer
IMO.
llvm-svn: 198532
Bill Wendling [Sun, 5 Jan 2014 01:47:20 +0000 (01:47 +0000)]
Emit an error message if the value passed to __builtin_returnaddress isn't a constant
__builtin_returnaddress requires that the value passed into is be a constant.
However, at -O0 even a constant expression may not be converted to a constant.
Emit an error message intead of crashing.
llvm-svn: 198531
Craig Topper [Sun, 5 Jan 2014 01:35:51 +0000 (01:35 +0000)]
Mark the 64-bit x86 push/pop instructions as In64BitMode. Mark the corresponding 32-bit versions with the same encodings Not64BitMode. Remove hack from tablegen disassembler table emitter. Fix bad test.
llvm-svn: 198530
Craig Topper [Sun, 5 Jan 2014 01:34:12 +0000 (01:34 +0000)]
Don't use PrintFatalError(which calls exit) for 'Primary decode conflict'. Just skip emitting the table. This way the main function will delete the output file instead of it remaining empty and confusing dependency checks if build is invoked a second time.
llvm-svn: 198529
Nico Weber [Sun, 5 Jan 2014 00:37:45 +0000 (00:37 +0000)]
Add a FIXME.
llvm-svn: 198528
Alp Toker [Sat, 4 Jan 2014 22:47:48 +0000 (22:47 +0000)]
Add missed cleanup from r198456
All other uses of this macro in LLVM/clang have been moved to the function
definition so follow suite (and the usage advice) here too for consistency.
llvm-svn: 198516
Craig Topper [Sat, 4 Jan 2014 22:29:41 +0000 (22:29 +0000)]
Tag x86 move to/from debug/control registers with Not64BitMode/In64BitMode. Remove disassembler hack.
llvm-svn: 198515
Joerg Sonnenberger [Sat, 4 Jan 2014 17:43:00 +0000 (17:43 +0000)]
Switch to using C++ style casts.
llvm-svn: 198505
Alp Toker [Sat, 4 Jan 2014 17:00:45 +0000 (17:00 +0000)]
Revert "Fix PR18361: Invalidate LoopDispositions after LoopSimplify hoists things."
This commit was the source of crasher PR18384:
While deleting: label %for.cond127
An asserting value handle still pointed to this value!
UNREACHABLE executed at llvm/lib/IR/Value.cpp:671!
Reverting to get the builders green, feel free to re-land after fixing up.
(Renato has a handy isolated repro if you need it.)
This reverts commit r198478.
llvm-svn: 198503
Alp Toker [Sat, 4 Jan 2014 15:58:28 +0000 (15:58 +0000)]
Revert "ToolingTest.cpp: Fix r158592, runToolOnCode.FindsNoTopLevelDeclOnEmptyCode on msvc. LangOpts.MicrosoftExt still appends "class type_info;"."
type_info has been made an implicitly predeclared type in r198497 and will no
longer appear as a user-declared type so we can remove this old hack.
This reverts commit r158595.
llvm-svn: 198502
Alp Toker [Sat, 4 Jan 2014 15:25:02 +0000 (15:25 +0000)]
Move MS predefined type_info out of InitializePredefinedMacros
Instead of keeping it in amongst the macros, build the declaration at Sema init
the same way we do with other predeclared and builtin types.
In practice this means the declaration is marked implicit and therefore won't
show up as an unwanted user-declared type in tooling which has been a
frequently reported issue (though I haven't been able to cook up a test).
llvm-svn: 198497
Alp Toker [Sat, 4 Jan 2014 13:47:14 +0000 (13:47 +0000)]
Only mark dump() function definitions 'used' in debug builds
This has the dual effect of (1) enabling more dead-stripping in release builds
and (2) ensuring that debug helper functions aren't stripped away in debug
builds, as they're intended to be called from the debugger.
Note that the attribute is applied to definitions rather than declarations in
headers going forward because it's now conditional on NDEBUG:
/// \brief Mark debug helper function definitions like dump() that should not be
/// stripped from debug builds.
Requires corresponding macro added in LLVM r198456.
llvm-svn: 198489
Venkatraman Govindaraju [Sat, 4 Jan 2014 11:30:13 +0000 (11:30 +0000)]
[Sparc] Add the initial implementation of an asm parser for sparc/sparcv9.
llvm-svn: 198484
Yaron Keren [Sat, 4 Jan 2014 09:27:39 +0000 (09:27 +0000)]
80 cols fixes.
llvm-svn: 198482
Yaron Keren [Sat, 4 Jan 2014 08:56:00 +0000 (08:56 +0000)]
Implement the functions: clz, clzl, clzll, ctz, ctzl, and ctzll
for libcxx when compiled with Visual C++ on Win32 and Win64.
clang and gcc (MinGW) compilers provide these implementations themselves.
llvm-svn: 198481
Venkatraman Govindaraju [Sat, 4 Jan 2014 07:17:21 +0000 (07:17 +0000)]
[SparcV9]: Implement RETURNADDR and FRAMEADDR lowering in SPARC64.
Fixes PR18356.
llvm-svn: 198480
Argyrios Kyrtzidis [Sat, 4 Jan 2014 06:27:45 +0000 (06:27 +0000)]
Add an additional check in test/Sema/ext_vector_casts.c
llvm-svn: 198479
Andrew Trick [Sat, 4 Jan 2014 05:52:49 +0000 (05:52 +0000)]
Fix PR18361: Invalidate LoopDispositions after LoopSimplify hoists things.
getSCEV for an ashr instruction creates an intermediate zext
expression when it truncates its operand.
The operand is initially inside the loop, so the narrow zext
expression has a non-loop-invariant loop disposition.
LoopSimplify then runs on an outer loop, hoists the ashr operand, and
properly invalidate the SCEVs that are mapped to value.
The SCEV expression for the ashr is now an AddRec with the hoisted
value as the now loop-invariant start value.
The LoopDisposition of this wide value was properly invalidated during
LoopSimplify.
However, if we later get the ashr SCEV again, we again try to create
the intermediate zext expression. We get the same SCEV that we did
earlier, and it is still cached because it was never mapped to a
Value. When we try to create a new AddRec we abort because we're using
the old non-loop-invariant LoopDisposition.
I don't have a solution for this other than to clear LoopDisposition
when LoopSimplify hoists things.
I think the long-term strategy should be to perform LoopSimplify on
all loops before computing SCEV and before running any loop opts on
individual loops. It's possible we may want to rerun LoopSimplify on
individual loops, but it should rarely do anything, so rarely require
invalidating SCEV.
llvm-svn: 198478
Andrew Trick [Sat, 4 Jan 2014 05:52:45 +0000 (05:52 +0000)]
whitespace
llvm-svn: 198477
Ted Kremenek [Sat, 4 Jan 2014 05:52:11 +0000 (05:52 +0000)]
[analyzer] Remove IdempotentOperations checker.
This checker has not been updated to work with interprocedural analysis,
and actually contains both logical correctness issues but also
memory bugs. We can resuscitate it from version control once there
is focused interest in making it a real viable checker again.
llvm-svn: 198476
Craig Topper [Sat, 4 Jan 2014 05:09:27 +0000 (05:09 +0000)]
Remove JMP64pcrel32 (jmpq ). There are no tests for it. I'm pretty sure it won't be emitted correctly since it was set to NoImm. And I can't prove that gas accepts 'jmpq' with an immediate either. Remove the special case for it from the disassembler table generator.
llvm-svn: 198475
Argyrios Kyrtzidis [Sat, 4 Jan 2014 03:31:22 +0000 (03:31 +0000)]
[Sema] When checking if a bitcast is appropriate between vector types, take into
consideration the num-of-elements*width-of-element width.
Disallow casts when such width is not equal between the vector types otherwise
we may end up with an invalid LLVM bitcast.
rdar://
15722308.
llvm-svn: 198474
Richard Trieu [Sat, 4 Jan 2014 01:57:42 +0000 (01:57 +0000)]
Ignore qualified templated functions for -Winfinite-recursion. This treats
functions like Foo<5>::run() the same way as run<5>() for this warning.
llvm-svn: 198470
Jason Molenda [Sat, 4 Jan 2014 01:37:52 +0000 (01:37 +0000)]
Use Address::SetLoadAddress() instead of SectionLoadList::ResolveLoadAddress().
The former will set the Address object's offset to the load address value if
it is not present in any section; the latter will only set the Address object
if the load addr is contained in one of its sections.
<rdar://problem/
15135987>
llvm-svn: 198469
Joey Gouly [Sat, 4 Jan 2014 01:22:05 +0000 (01:22 +0000)]
[MachO] Add some missing NListTypes.
llvm-svn: 198468
Reid Kleckner [Fri, 3 Jan 2014 23:51:09 +0000 (23:51 +0000)]
Fix MSVC warning about missing return in DataLayout
llvm-svn: 198465
Reid Kleckner [Fri, 3 Jan 2014 23:42:00 +0000 (23:42 +0000)]
[ms-cxxabi] Improve vbtable name mangling accuracy
Summary:
This makes us more compatible with MSVC 2012+ and fixes PR17748 where we
would give two tables the same name.
Rather than doing a fresh depth-first traversal of the inheritance graph
for every record's vbtables, now we memoize vbtable paths for each
record. By doing memoization, we end up considering virtual bases of
subobjects that come later in the depth-first traversal. Where
previously we would have ignored a virtual base that we'd already seen,
we now consider it for name mangling purposes without emitting a
duplicate vbtable for it.
Reviewers: majnemer
CC: cfe-commits
Differential Revision: http://llvm-reviews.chandlerc.com/D2509
llvm-svn: 198462
Adrian Prantl [Fri, 3 Jan 2014 23:34:30 +0000 (23:34 +0000)]
Debug info: Ensure that the last stop point in a function is still within
the lexical block formed by the compound statement that is the function
body.
rdar://problem/
15010825
llvm-svn: 198461
Joey Gouly [Fri, 3 Jan 2014 23:12:02 +0000 (23:12 +0000)]
[MachO] Begin to add some MachO specific File/Atoms, and add the start of
normalizedToAtoms.
llvm-svn: 198459