Saleem Abdulrasool [Mon, 2 Jun 2014 01:17:49 +0000 (01:17 +0000)]
Support: add Windows ARM EH data structures
Introduce the support structures necessary to deal with the Windows ARM EH data.
These definitions are extremely aggressive about assertions to aid future use
for generation of the entries and subsequent decoding.
The names for the various fields are meant to reflect the names used by the
Visual Studio toolchain to aid communication.
Due to the complexity in reading a few of the values, there are a couple of
additional utility functions to decode the information.
In general, there are two ways to encode the unwinding information:
- packed, which places the data inline into the
_IMAGE_ARM_RUNTIME_FUNCTION_ENTRY structure.
- unpacked, which places the data into auxiliary structures placed into the
.xdata section.
The set of structures allow reading of data in either encoding, with the minor
caveat that epilogue scopes need to be decoded manually by constructing the
structure from the data returned by the RuntimeFunction structure.
These definitions are meant for read-only access at the current point as the
first use of them will be to decode the exception information.
llvm-svn: 209998
Alp Toker [Sun, 1 Jun 2014 19:13:44 +0000 (19:13 +0000)]
Revert "Fix the undefined-but-used odr-use marker (DR48)"
Wrong patch got committed (this one isn't ready for prime time).
This reverts commit r209996.
llvm-svn: 209997
Alp Toker [Sun, 1 Jun 2014 18:49:32 +0000 (18:49 +0000)]
Fix the undefined-but-used odr-use marker (DR48)
We should treat tentative definitions as undefined for the purpose of
ODR-use linkage checking.
This broke somewhere around r149731 when tests were disabled.
Note that test coverage for these diagnostics is generally lacking due to a
separate issue (PR19910: Don't suppress unused/undefined warnings when there
are errors).
llvm-svn: 209996
Alp Toker [Sun, 1 Jun 2014 18:28:36 +0000 (18:28 +0000)]
Fix typos
llvm-svn: 209995
Alp Toker [Sun, 1 Jun 2014 18:01:33 +0000 (18:01 +0000)]
Add initial CMake build system
This is not yet supported for production builds but can already produce working
binaries on OS X and Linux with clang and gcc.
The intention is to improve support to the point where it can integrate with
the LLVM runtime platform, cover all platforms, runtime/release build
configurations and run the tests.
Patch by Jack Howarth!
llvm-svn: 209994
Rafael Espindola [Sun, 1 Jun 2014 16:16:02 +0000 (16:16 +0000)]
Don't compare an error_code with nullptr.
llvm-svn: 209993
Faisal Vali [Sun, 1 Jun 2014 16:11:54 +0000 (16:11 +0000)]
Fix PR18498: Support explicit template arguments with variadic generic lambdas
http://llvm.org/bugs/show_bug.cgi?id=18498
This code was resulting in a crash:
auto L = [](auto ... v) { };
L.operator()<int>(3);
The reason is that the partially-substituted-pack is incorrectly retained within the current-instantiation-scope during template-argument-finalization, and because lambda's are local, there parent instantiation scopes are merged, which leads to the expansion-pattern being retained in the finalized specialization.
This patch ensures that once we have finalized deduction of a parameter-pack, we remove the partially-substituted-pack so that it doesn't cause CheckParameterPacksForExpansion to incorrectly inform the caller that it needs to retain the expansion pattern.
Thanks to Richard Smith for the review!
http://reviews.llvm.org/D2135
llvm-svn: 209992
Rafael Espindola [Sun, 1 Jun 2014 13:45:09 +0000 (13:45 +0000)]
Test fixes and improvements.
Remove redundant -fno-stack-protector run with openbsd.
Add -target to the -fstack-protector tests so they pass on openbsd.
Patch by Brad Smith.
llvm-svn: 209991
Christian Pirker [Sun, 1 Jun 2014 09:30:52 +0000 (09:30 +0000)]
ARMEB: Fix function return type f64
Reviewed at http://reviews.llvm.org/D3968
llvm-svn: 209990
Elena Demikhovsky [Sun, 1 Jun 2014 08:45:11 +0000 (08:45 +0000)]
Updates in IntelJITEventListener.cpp - by Arch Robison.
This patch updates IntelJITEventListener.cpp to account for revision 206654, which removed some methods from DILineInfo.
llvm-svn: 209989
Matt Arsenault [Sun, 1 Jun 2014 07:38:21 +0000 (07:38 +0000)]
R600: Set all float vector expands in the same place
llvm-svn: 209988
Simon Atanasyan [Sun, 1 Jun 2014 06:28:42 +0000 (06:28 +0000)]
[Mips] Implement .{ctors,dtors}.<priority> sections ordering.
Arrange .ctors/.dtors sections in the following order:
.ctors from crtbegin.o or crtbegin?.o
.ctors from regular object files
.ctors.* (sorted) from regular object files
.ctors from crtend.o or crtend?.o
This order is specific for MIPS traget. For example, on X86
the .ctors.* sections are merged into the .init_array section.
llvm-svn: 209987
Saleem Abdulrasool [Sun, 1 Jun 2014 04:07:07 +0000 (04:07 +0000)]
compiler-rt: prefer .rodata for non MachO targets
Place constants into .rdata if targeting ELF or COFF/PE. This should be
functionally identical, however, the data would be placed into a different
section. This is purely a cleanup change.
llvm-svn: 209986
Saleem Abdulrasool [Sun, 1 Jun 2014 04:07:03 +0000 (04:07 +0000)]
compiler-rt: whitespace and uniformity for arm
Make the whitespace a bit more uniform in the various assembly routines. This
also makes the assembly files a bit more uniform on the ARM side by explicitly
stating that it is using the unified syntax and that the contents of the code is
in the text section (or segment). No functional change.
llvm-svn: 209985
David Blaikie [Sun, 1 Jun 2014 03:38:13 +0000 (03:38 +0000)]
DebugInfo: Assert that DbgVariables have associated DIEs
This was previously committed in r209680 and reverted in r209683 after
it caused sanitizer builds to crash.
The issue seems to be that the DebugLoc associated with dbg.value IR
intrinsics isn't necessarily accurate. Instead, we duplicate the
DIVariables and add an InlinedAt field to them to record their
location.
We were using this InlinedAt field to compute the LexicalScope for the
variable, but not using it in the abstract DbgVariable construction and
mapping. This resulted in a formal parameter to the current concrete
function, correctly having no InlinedAt information, but incorrectly
having a DebugLoc that described an inlined location within the
function... thus an abstract DbgVariable was created for the variable,
but its DIE was never constructed (since the LexicalScope had no such
variable). This DbgVariable was silently ignored (by testing for a
non-null DIE on the abstract DbgVariable).
So, fix this by using the right scoping information when constructing
abstract DbgVariables.
In the long run, I suspect we want to undo the work that added this
second kind of location tracking and fix the places where the DebugLoc
propagation on the dbg.value intrinsic fails. This will shrink debug
info (by not duplicating DIVariables), make it more efficient (by not
having to construct new DIVariable metadata nodes to try to map back to
a single variable), and benefit all instructions.
But perhaps there are insurmountable issues with DebugLoc quality that
I'm unaware of... I just don't know how we can't /just keep the DebugLoc
from the dbg.declare to the dbg.values and never get this wrong/.
Some history context:
http://llvm.org/viewvc/llvm-project?view=revision&revision=135629
http://llvm.org/viewvc/llvm-project?view=revision&revision=137253
llvm-svn: 209984
Alp Toker [Sun, 1 Jun 2014 02:02:23 +0000 (02:02 +0000)]
Don't add examples to the check-clang dependencies unless requested
The tests were correctly getting skipped but the targets still mistakenly got
built.
llvm-svn: 209983
Alp Toker [Sat, 31 May 2014 21:26:28 +0000 (21:26 +0000)]
Fix typos
llvm-svn: 209982
Alp Toker [Sat, 31 May 2014 21:26:17 +0000 (21:26 +0000)]
ExecutionEngine: avoid NDEBUG in headers
llvm-svn: 209981
Alp Toker [Sat, 31 May 2014 21:26:09 +0000 (21:26 +0000)]
Update a couple of header inclusion guards
llvm-svn: 209980
Matt Arsenault [Sat, 31 May 2014 19:25:17 +0000 (19:25 +0000)]
R600/SI: Remove redundant patterns
These patterns are already handled in the instruction definition.
llvm-svn: 209979
Alp Toker [Sat, 31 May 2014 16:32:22 +0000 (16:32 +0000)]
Preprocessor: make C++ operator names as macro identifiers a compatible extension
With recent changes, this is now a compatible language extension and can be
safely enabled with -ms-extensions instead of requiring the full
-ms-compatibility MSVC drop-in mode. As such we can now also emit an extension
warning under -Wmicrosoft to help users port their code.
llvm-svn: 209978
Adam Nemet [Sat, 31 May 2014 16:23:20 +0000 (16:23 +0000)]
[SelectionDAG] Force cycle detection in AssignTopologicalOrder before aborting
DAG cycle detection is only enabled with ENABLE_EXPENSIVE_CHECKS. However we
can run it just before we would crash in order to provide more informative
diagnostics.
Now in addition to the "Overran sorted position" message we also get the Node
printed if a cycle was detected.
Tested by building several configs: Debug+Assert, Debug+Assert+Check (this is
ENABLE_EXPENSIVE_CHECKS), Release+Assert and Release. Also tried that the
AssignTopologicalOrder assert produces the expected results.
llvm-svn: 209977
Adam Nemet [Sat, 31 May 2014 16:23:17 +0000 (16:23 +0000)]
[SelectionDAG] Pass DAG to checkForCycles
Pass the DAG down to checkForCycles from all callers where we have it. This
allows target-specific nodes to be printed properly.
Also print some missing newlines.
llvm-svn: 209976
Adam Nemet [Sat, 31 May 2014 16:23:13 +0000 (16:23 +0000)]
[SelectionDAG] Remove duplicate declaration
Prefer the decl in SelectionDAGNodes.h because it's used there and
SelectionDAG.h includes SelectionDAGNodes.h.
llvm-svn: 209975
Benjamin Kramer [Sat, 31 May 2014 15:06:33 +0000 (15:06 +0000)]
Fix test on platforms where size_t is not 'unsigned long'.
llvm-svn: 209974
Benjamin Kramer [Sat, 31 May 2014 15:01:54 +0000 (15:01 +0000)]
[Reassociate] Similar to "X + -X" -> "0", added code to handle "X + ~X" -> "-1".
Handle "X + ~X" -> "-1" in the function Value *Reassociate::OptimizeAdd(Instruction *I, SmallVectorImpl<ValueEntry> &Ops);
This patch implements:
TODO: We could handle "X + ~X" -> "-1" if we wanted, since "-X = ~X+1".
Patch by Rahul Jain!
Differential Revision: http://reviews.llvm.org/D3835
llvm-svn: 209973
Justin Bogner [Sat, 31 May 2014 07:59:03 +0000 (07:59 +0000)]
Fix XFAIL condition from r208840
The XFAILs in r208840 were too general. They were meant to only apply
to testing when use_system_lib was set.
llvm-svn: 209972
Matt Arsenault [Sat, 31 May 2014 06:47:42 +0000 (06:47 +0000)]
R600/SI: Fix [s|u]int_to_fp for i1
llvm-svn: 209971
Alp Toker [Sat, 31 May 2014 06:11:09 +0000 (06:11 +0000)]
Support C++11 scoped enums in NamedDecl::isCXXClassMember()
Resolves an old FIXME.
No callers depend on this giving the right answer so I haven't been able to
cook up a useful test case.
llvm-svn: 209970
Alp Toker [Sat, 31 May 2014 06:11:02 +0000 (06:11 +0000)]
Rename utility function templates
isExternCTemplate() and getLanguageLinkageTemplate() have nothing to do with
templates despite the dubious naming scheme.
llvm-svn: 209969
Matt Arsenault [Sat, 31 May 2014 05:18:52 +0000 (05:18 +0000)]
Fix typos in tablegen error messages
llvm-svn: 209968
Simon Atanasyan [Sat, 31 May 2014 04:51:07 +0000 (04:51 +0000)]
[yaml2obj] Add new command line option `-docnum`.
Input YAML file might contain multiple object file definitions.
New option `-docnum` allows to specify an ordinal number (starting from 1)
of definition used for an object file generation.
Patch reviewed by Sean Silva.
llvm-svn: 209967
Jordan Rose [Sat, 31 May 2014 04:12:14 +0000 (04:12 +0000)]
Format strings: check against an enum's underlying type.
This allows us to be more careful when dealing with enums whose fixed
underlying type requires special handling in a format string, like
NSInteger.
A refinement of r163266 from a year and a half ago, which added the
special handling for NSInteger and friends in the first place.
<rdar://problem/
16616623>
llvm-svn: 209966
Hans Wennborg [Sat, 31 May 2014 04:05:57 +0000 (04:05 +0000)]
AST Dump: print the Inherited flag on attributes
Also move the attribute-specific dumping to after dumping this and
the Implicit flag.
Differential Revision: http://reviews.llvm.org/D3971
llvm-svn: 209965
Rafael Espindola [Sat, 31 May 2014 03:40:00 +0000 (03:40 +0000)]
really fix the windows build.
llvm-svn: 209964
Alp Toker [Sat, 31 May 2014 03:38:17 +0000 (03:38 +0000)]
Preprocessor: recover gracefully when C++ operator names are used as macro identifiers
This failure mode shows up occasionally when users try to include C headers in
C++ projects or when porting from Windows. We might as well recover in the way
the user expected, thus avoiding confusing diagnostic messages at point of use.
llvm-svn: 209963
Alp Toker [Sat, 31 May 2014 03:38:08 +0000 (03:38 +0000)]
Preprocessor: don't exit early in CheckMacroName()
The checks below can hypothetically apply to converted operator name
identifiers.
In practice there are no builtin macros etc. with those names so there's no
behavioural change to test.
llvm-svn: 209962
Rafael Espindola [Sat, 31 May 2014 03:26:57 +0000 (03:26 +0000)]
Fix windows build.
llvm-svn: 209961
Rafael Espindola [Sat, 31 May 2014 03:21:04 +0000 (03:21 +0000)]
There is no std::errc::success, remove the llvm one.
llvm-svn: 209960
Rafael Espindola [Sat, 31 May 2014 03:20:52 +0000 (03:20 +0000)]
There is no std::errc:success, remove the llvm one.
llvm-svn: 209959
Rafael Espindola [Sat, 31 May 2014 03:20:37 +0000 (03:20 +0000)]
There is no std::errc:success, remove the llvm one.
llvm-svn: 209958
Rafael Espindola [Sat, 31 May 2014 02:29:28 +0000 (02:29 +0000)]
Turn errc and windows_error into enum classes.
llvm-svn: 209957
Rafael Espindola [Sat, 31 May 2014 02:23:01 +0000 (02:23 +0000)]
Use make_error_code in preparation for making errc an enum class.
llvm-svn: 209956
Nikola Smiljanic [Sat, 31 May 2014 02:10:59 +0000 (02:10 +0000)]
PR12961 - Extend DR532 to cover C++98/03.
llvm-svn: 209955
Hans Wennborg [Sat, 31 May 2014 02:08:49 +0000 (02:08 +0000)]
Diagnose dll attribute on member of class that already has a dll attribute
Differential Revision: http://reviews.llvm.org/D3973
llvm-svn: 209954
Rafael Espindola [Sat, 31 May 2014 02:00:59 +0000 (02:00 +0000)]
Use error_code() instead of error_code::succes()
There is no std::error_code::success, so this removes much of the noise
in transitioning to std::error_code.
llvm-svn: 209953
Rafael Espindola [Sat, 31 May 2014 01:37:45 +0000 (01:37 +0000)]
Use error_code() instead of error_code::succes()
There is no std::error_code::success, so this removes much of the noise
in transitioning to std::error_code.
llvm-svn: 209952
Peter Collingbourne [Sat, 31 May 2014 01:36:02 +0000 (01:36 +0000)]
Fix the behavior of ExecuteAndWait with a non-zero timeout.
llvm-svn: 209951
Hans Wennborg [Sat, 31 May 2014 01:30:30 +0000 (01:30 +0000)]
Make Attr::Clone() also clone the Inherited, IsPackExpansion and Implicit flags
I was bitten by this when working with the dll attributes: when a dll
attribute was cloned from a class template declaration to its
specialization, the Inherited flag didn't get cloned.
Differential Revision: http://reviews.llvm.org/D3972
llvm-svn: 209950
Rafael Espindola [Sat, 31 May 2014 01:26:30 +0000 (01:26 +0000)]
Use error_code() instead of error_code::succes()
There is no std::error_code::success, so this removes much of the noise
in transitioning to std::error_code.
llvm-svn: 209949
Rafael Espindola [Sat, 31 May 2014 01:22:21 +0000 (01:22 +0000)]
Use error_code() instead of error_code::succes()
There is no std::error_code::success, so this removes much of the noise
in transitioning to std::error_code.
llvm-svn: 209948
Filipe Cabecinhas [Sat, 31 May 2014 00:52:23 +0000 (00:52 +0000)]
Make blend tests more specific
Following the lead set by r209324, I'm making these tests match the whole
instruction, so we can be sure we're lowering them correctly.
llvm-svn: 209947
Alexey Samsonov [Sat, 31 May 2014 00:33:05 +0000 (00:33 +0000)]
[ASan] Behave the same for functions w/o sanitize_address attribute and blacklisted functions
llvm-svn: 209946
Alexey Samsonov [Sat, 31 May 2014 00:26:03 +0000 (00:26 +0000)]
[TSan] Fixup a test case after r209939
llvm-svn: 209945
Jonathan Roelofs [Sat, 31 May 2014 00:25:59 +0000 (00:25 +0000)]
Add EH test case checking that handlers in noexcept functions can still unwind
llvm-svn: 209944
Jonathan Roelofs [Sat, 31 May 2014 00:24:58 +0000 (00:24 +0000)]
Add EH test cases corresponding to C++ ABI # 15.3.3
llvm-svn: 209943
Michael J. Spencer [Sat, 31 May 2014 00:22:12 +0000 (00:22 +0000)]
[CodeGen] Don't cast and use SizeTy instead of Int32Ty when constructing {extract,insert} vector element instructions.
llvm-svn: 209942
Richard Smith [Sat, 31 May 2014 00:18:32 +0000 (00:18 +0000)]
Refactor: use the DeclarationNameInfo form of BuildMemberReferenceExpr
consistently, rather than sometimes performing lookup and calling the
LookupResult form. No functionality change intended.
llvm-svn: 209941
Alexey Samsonov [Sat, 31 May 2014 00:12:20 +0000 (00:12 +0000)]
[TSan] Add a test case for r209939
llvm-svn: 209940
Alexey Samsonov [Sat, 31 May 2014 00:11:37 +0000 (00:11 +0000)]
[TSan] Behave the same for functions w/o sanitize_thread attribute and blacklisted functions
llvm-svn: 209939
Rafael Espindola [Sat, 31 May 2014 00:10:47 +0000 (00:10 +0000)]
delete dead code.
llvm-svn: 209938
Eric Christopher [Sat, 31 May 2014 00:07:32 +0000 (00:07 +0000)]
Have the TLOF creation take a Triple rather than needing a subtarget.
llvm-svn: 209937
Alp Toker [Sat, 31 May 2014 00:02:27 +0000 (00:02 +0000)]
cc1as: invert return bool to indicate failure
This simplifies code flow and matches the convention used in surrounding code.
No functional change.
llvm-svn: 209936
Alp Toker [Sat, 31 May 2014 00:02:21 +0000 (00:02 +0000)]
cc1as: fix a potential leak and unremoved output file in error conditions
llvm-svn: 209935
Andrea Di Biagio [Fri, 30 May 2014 23:17:53 +0000 (23:17 +0000)]
[X86] Add two combine rules to simplify dag nodes introduced during type legalization when promoting nodes with illegal vector type.
This patch teaches the backend how to simplify/canonicalize dag node
sequences normally introduced by the backend when promoting certain dag nodes
with illegal vector type.
This patch adds two new combine rules:
1) fold (shuffle (bitcast (BINOP A, B)), Undef, <Mask>) ->
(shuffle (BINOP (bitcast A), (bitcast B)), Undef, <Mask>)
2) fold (BINOP (shuffle (A, Undef, <Mask>)), (shuffle (B, Undef, <Mask>))) ->
(shuffle (BINOP A, B), Undef, <Mask>).
Both rules are only triggered on the type-legalized DAG.
In particular, rule 1. is a target specific combine rule that attempts
to sink a bitconvert into the operands of a binary operation.
Rule 2. is a target independet rule that attempts to move a shuffle
immediately after a binary operation.
llvm-svn: 209930
Filipe Cabecinhas [Fri, 30 May 2014 23:03:11 +0000 (23:03 +0000)]
Convert a vselect into a concat_vector if possible
Summary:
If both vector args to vselect are concat_vectors and the condition is
constant and picks half a vector from each argument, convert the vselect
into a concat_vectors.
Added a test.
The ConvertSelectToConcatVector is assuming it doesn't get vselects with
arguments of, for example, <undef, undef, true, true>. Those get taken
care of in the checks above its call.
Reviewers: nadav, delena, grosbach, hfinkel
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3916
llvm-svn: 209929
Nick Kledzik [Fri, 30 May 2014 22:51:04 +0000 (22:51 +0000)]
[mach-o] Add support for custom sections
llvm-svn: 209928
Eric Christopher [Fri, 30 May 2014 22:47:53 +0000 (22:47 +0000)]
isSVR4ABI() returned !isDarwin() so just move that to the else
block and remove the unreachable code.
llvm-svn: 209927
Eric Christopher [Fri, 30 May 2014 22:47:48 +0000 (22:47 +0000)]
Rename CreateTLOF->createTLOF to match the rest of the file and the
rest of the targets with a similar function name.
llvm-svn: 209926
Greg Fitzgerald [Fri, 30 May 2014 22:44:35 +0000 (22:44 +0000)]
Disable stack-overflow test on ARM
llvm-svn: 209925
Richard Smith [Fri, 30 May 2014 22:16:51 +0000 (22:16 +0000)]
Take PrintingPolicy::SuppressUnwrittenScope into account when printing the
qualified name of a NamedDecl. Patch by Volodymyr Sapsai!
llvm-svn: 209924
Filipe Cabecinhas [Fri, 30 May 2014 21:31:21 +0000 (21:31 +0000)]
Separate the check for blend shuffle_vector masks
Summary:
Separate the check for blend shuffle_vector masks into isBlendMask.
This function will also be used to check if a vector shuffle is legal. No
change in functionality was intended, but we ended up improving codegen on
two tests, which were being (more) optimized only if the resulting shuffle
was legal.
Reviewers: nadav, delena, andreadb
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3964
llvm-svn: 209923
Ben Langmuir [Fri, 30 May 2014 21:20:54 +0000 (21:20 +0000)]
Invalidate the file system cache entries for files that may rebuild
This reapplies r209910 with a fix for the assertion failures hit on the
buildbots.
original commit message:
I thought we could get away without this, but it means that the
FileEntry objects actually refer to the wrong files, since pcms are not
updated inplace, they are atomically renamed into place after compiling
a module.
So we are close to the original behaviour of invalidating the cache for
all modules being removed, but now we should only invalidate the ones
that depend on whichever module failed to load.
Unfortunately I haven't come up with a new test that didn't require
a race between parallel invocations of clang.
<rdar://problem/
17038180>
llvm-svn: 209922
Sanjay Patel [Fri, 30 May 2014 21:18:13 +0000 (21:18 +0000)]
fixed more typos
llvm-svn: 209921
Adrian Prantl [Fri, 30 May 2014 21:10:13 +0000 (21:10 +0000)]
Roll DbgVariable::setMInsn into the constructor. No functional changes.
llvm-svn: 209920
Alp Toker [Fri, 30 May 2014 21:07:40 +0000 (21:07 +0000)]
Remove unused struct from test
This presumably got inadvertently copied from dllimport-members.cpp.
llvm-svn: 209919
Sanjay Patel [Fri, 30 May 2014 21:07:25 +0000 (21:07 +0000)]
added link to CMake page
llvm-svn: 209918
Greg Clayton [Fri, 30 May 2014 21:06:57 +0000 (21:06 +0000)]
Don't use libc's "char *basename(char *)" or "char *dirname(char *)" as they are not thread safe.
I switched the lldb_private::FileSpec code over to use "llvm::StringRef llvm::sys::path::filename(llvm::StringRef)" for basename() and "llvm::StringRef llvm::sys::path::parent_path(llvm::StringRef)" for dirname().
<rdar://problem/
16870083>
llvm-svn: 209917
Sanjay Patel [Fri, 30 May 2014 20:55:55 +0000 (20:55 +0000)]
fixed typo
llvm-svn: 209916
Richard Smith [Fri, 30 May 2014 19:44:17 +0000 (19:44 +0000)]
Add a FIXME for a bug Doug pointed out in code review.
llvm-svn: 209915
Matt Arsenault [Fri, 30 May 2014 18:34:43 +0000 (18:34 +0000)]
Make bitcast, extractelement, and insertelement considered cheap for speculation.
This helps more branches into selects. On R600,
vectors are cheap and anything that helps
remove branches is very good.
llvm-svn: 209914
DeLesley Hutchins [Fri, 30 May 2014 18:20:04 +0000 (18:20 +0000)]
Thread Safety Analysis: fix handling of string and char literals.
llvm-svn: 209913
Todd Fiala [Fri, 30 May 2014 17:59:47 +0000 (17:59 +0000)]
gdb-remote signal delivery test cleanup.
Learned that MacOSX only accepts signal delivery on a thread that is
already signal handling. Reworked the test exe to cause a SIGSEGV
and recover if either nothing intercepts the SIGSEGV handler, or
if a SIGUSR1 is inserted. The test uses the latter part to test
signal delivery on continue using the SIGUSR1.
I still don't have this working on MacOSX. I'm seeing the
signal get delivered to a different thread than the one I'm
specifying with $Hc{thread-id} + $C{signo}, or with
$vCont;C{signo}:{thread-id};c. I'll come back to this
after getting it working on the llgs branch on Linux x86_64.
llvm-svn: 209912
Ben Langmuir [Fri, 30 May 2014 17:54:59 +0000 (17:54 +0000)]
Revert "Invalidate the file system cache entries for files that may rebuild"
This reverts commit r209910, which is breaking some of the bots.
llvm-svn: 209911
Ben Langmuir [Fri, 30 May 2014 17:42:15 +0000 (17:42 +0000)]
Invalidate the file system cache entries for files that may rebuild
I thought we could get away without this, but it means that the
FileEntry objects actually refer to the wrong files, since pcms are not
updated inplace, they are atomically renamed into place after compiling
a module.
So we are close to the original behaviour of invalidating the cache for
all modules being removed, but now we should only invalidate the ones
that depend on whichever module failed to load.
Unfortunately I haven't come up with a new test that didn't require
a race between parallel invocations of clang.
<rdar://problem/
17038180>
llvm-svn: 209910
Nico Weber [Fri, 30 May 2014 17:27:21 +0000 (17:27 +0000)]
Let libc++abi compile with gcc.
There was a single problem in cxa_demangle.cpp, where gcc would complain
`error: changes meaning of 'String'` about the line `typedef String String;`.
According to 3.3.7p2, this diagnostic is allowed (but not required, so clang
does not have to report this).
As a fix, make string_pair a template and pass String as template parameter.
This fixes the error with gcc and also removes some repetition from the code.
No behavior change.
llvm-svn: 209909
Hans Wennborg [Fri, 30 May 2014 16:59:42 +0000 (16:59 +0000)]
Start adding support for dllimport/dllexport on classes (PR11170)
This implements the central part of support for dllimport/dllexport on
classes: allowing the attribute on class declarations, inheriting it
to class members, and forcing emission of exported members. It's based
on Nico Rieck's patch from http://reviews.llvm.org/D1099.
This patch doesn't propagate dllexport to bases that are template
specializations, which is an interesting problem. It also doesn't
look at the rules when redeclaring classes with different attributes,
I'd like to do that separately.
Differential Revision: http://reviews.llvm.org/D3877
llvm-svn: 209908
Logan Chien [Fri, 30 May 2014 16:48:56 +0000 (16:48 +0000)]
Fix MIPS exception personality encoding.
For MIPS, we have to encode the personality routine with
an indirect pointer to absptr; otherwise, some link warning
warning will be raised, and the program might crash in some
early MIPS Android device.
llvm-svn: 209907
Fariborz Jahanian [Fri, 30 May 2014 16:35:53 +0000 (16:35 +0000)]
Objective-C. Diagnose assigning a block pointer type to
an Objective-C object type other than 'id'.
// rdar://
16739120
llvm-svn: 209906
Tim Northover [Fri, 30 May 2014 16:01:29 +0000 (16:01 +0000)]
MachO: move test to ARM directory.
This test specifies an ARM triple, so it needs ARM as a registered
target.
llvm-svn: 209905
DeLesley Hutchins [Fri, 30 May 2014 15:55:25 +0000 (15:55 +0000)]
Thread Safety Analysis: fix uninitialized member in TIL Literal.
llvm-svn: 209904
Rafael Espindola [Fri, 30 May 2014 15:54:32 +0000 (15:54 +0000)]
PR19753: Optimize comparisons with "ashr exact" of a constanst.
Patch by suyog sarda.
llvm-svn: 209903
Rafael Espindola [Fri, 30 May 2014 15:00:45 +0000 (15:00 +0000)]
[pr19636] Fix known bit computation in urem instruction with power of two.
Patch by Andrey Kuharev.
llvm-svn: 209902
Tim Northover [Fri, 30 May 2014 14:41:51 +0000 (14:41 +0000)]
SelectionDAG: skip barriers for unordered atomic operations
Unordered is strictly weaker than monotonic, so if the latter doesn't have any
barriers then the former certainly shouldn't.
rdar://problem/
16548260
llvm-svn: 209901
Dmitry Vyukov [Fri, 30 May 2014 14:27:31 +0000 (14:27 +0000)]
tsan: add a test from data-race-test suite:
https://code.google.com/p/data-race-test/source/browse/trunk/unittest/racecheck_unittest.cc
llvm-svn: 209900
Tim Northover [Fri, 30 May 2014 14:14:07 +0000 (14:14 +0000)]
AArch64: combine arm64 & aarch64 cases in switch
There shouldn't be any difference in behaviour here, at least not in
any configurations people care about and possibly not in any reachable
configurations.
llvm-svn: 209899
Dmitry Vyukov [Fri, 30 May 2014 14:08:51 +0000 (14:08 +0000)]
tsan: make positive tests more robust
Add a script that is used to deflake inherently flaky tsan tests.
It is invoked from lit tests as:
%deflake %run %t
The script runs the target program up to 10 times,
until it produces a tsan warning.
llvm-svn: 209898
Dmitry Vyukov [Fri, 30 May 2014 13:36:29 +0000 (13:36 +0000)]
tsan: optimize memory access functions
The optimization is two-fold:
First, the algorithm now uses SSE instructions to
handle all 4 shadow slots at once. This makes processing
faster.
Second, if shadow contains the same access, we do not
store the event into trace. This increases effective
trace size, that is, tsan can remember up to 10x more
previous memory accesses.
Perofrmance impact:
Before:
[ OK ] DISABLED_BENCH.Mop8Read (2461 ms)
[ OK ] DISABLED_BENCH.Mop8Write (1836 ms)
After:
[ OK ] DISABLED_BENCH.Mop8Read (1204 ms)
[ OK ] DISABLED_BENCH.Mop8Write (976 ms)
But this measures only fast-path.
On large real applications the speedup is ~20%.
Trace size impact:
On app1:
Memory accesses :
1163265870
Including same :
791312905 (68%)
on app2:
Memory accesses :
166875345
Including same :
150449689 (90%)
90% of filtered events means that trace size is effectively 10x larger.
llvm-svn: 209897
Daniel Jasper [Fri, 30 May 2014 13:31:14 +0000 (13:31 +0000)]
Revert "Remove redundant check discovered in post-commit review of r209505."
This breaks with MSVC.
With IsLateTemplateParsed, FunctionDecl::doesThisDeclarationHaveABody() returns true regardless of Body.
This reinstates what was fixed in r208985.
llvm-svn: 209896
Tim Northover [Fri, 30 May 2014 13:23:06 +0000 (13:23 +0000)]
ARM: use AAPCS-style prologues for embedded MachO.
Darwin prologues save their GPRs in two stages: a narrow push of r0-r7 & lr,
followed by a wide push of the remaining registers if there are any. AAPCS uses
a single push.w instruction.
It turns out that, on average, enough registers get pushed that code is smaller
in the AAPCS prologue, which is a nice property for M-class programmers. They
also have other options available for back-traces, so can hopefully deal with
the fact that FP & LR aren't adjacent in memory.
rdar://problem/
15909583
llvm-svn: 209895