Reid Kleckner [Fri, 10 Apr 2015 16:26:42 +0000 (16:26 +0000)]
[WinEH] Try to make outlining invokes work a little better
WinEH currently turns invokes into calls. Long term, we will reconsider
this, but for now, make sure we remap the operands and clone the
successors of the new terminator.
llvm-svn: 234608
Adrian McCarthy [Fri, 10 Apr 2015 16:18:08 +0000 (16:18 +0000)]
Add logging plugin for Windows
llvm-svn: 234607
Benjamin Kramer [Fri, 10 Apr 2015 16:02:34 +0000 (16:02 +0000)]
[CallSite] Use the injected base class name. NFC.
llvm-svn: 234606
Tobias Grosser [Fri, 10 Apr 2015 15:41:14 +0000 (15:41 +0000)]
Drop ISL_INSTALL path
There is no need for an isl install path anymore. isl is now part of Polly.
llvm-svn: 234605
Timur Iskhodzhanov [Fri, 10 Apr 2015 15:31:16 +0000 (15:31 +0000)]
[ASan] Use a better name for a function parameter
llvm-svn: 234604
Hal Finkel [Fri, 10 Apr 2015 15:05:02 +0000 (15:05 +0000)]
[PowerPC] Prefetching should also consider depth > 1 loops
Iterating over loops from the LoopInfo instance only provides top-level loops.
We need to search the whole tree of loops to find the inner ones.
llvm-svn: 234603
Timur Iskhodzhanov [Fri, 10 Apr 2015 15:02:19 +0000 (15:02 +0000)]
Split Mprotect into MmapNoAccess and MprotectNoAccess to be more portable
On Windows, we have to know if a memory to be protected is mapped or not.
On POSIX, Mprotect was semantically different from mprotect most people know.
llvm-svn: 234602
Benjamin Kramer [Fri, 10 Apr 2015 14:50:08 +0000 (14:50 +0000)]
[CallSite] Make construction from Value* (or Instruction*) explicit.
CallSite roughly behaves as a common base CallInst and InvokeInst. Bring
the behavior closer to that model by making upcasts explicit. Downcasts
remain implicit and work as before.
Following dyn_cast as a mental model checking whether a Value *V isa
CallSite now looks like this:
if (auto CS = CallSite(V)) // think dyn_cast
instead of:
if (CallSite CS = V)
This is an extra token but I think it is slightly clearer. Making the
ctor explicit has the advantage of not accidentally creating nullptr
CallSites, e.g. when you pass a Value * to a function taking a CallSite
argument.
llvm-svn: 234601
Benjamin Kramer [Fri, 10 Apr 2015 14:49:31 +0000 (14:49 +0000)]
Don't rely on implicit CallSite construction.
llvm-svn: 234600
Rafael Espindola [Fri, 10 Apr 2015 14:37:39 +0000 (14:37 +0000)]
Try to make MSVC happy.
llvm-svn: 234599
Rafael Espindola [Fri, 10 Apr 2015 14:30:43 +0000 (14:30 +0000)]
Return std::unique_ptr to avoid a release and recreate.
llvm-svn: 234598
Rafael Espindola [Fri, 10 Apr 2015 14:11:52 +0000 (14:11 +0000)]
Use a std::unique_ptr to make it easier to see who owns the stream.
llvm-svn: 234597
Szabolcs Sipos [Fri, 10 Apr 2015 13:55:39 +0000 (13:55 +0000)]
[clang-tidy] Fix for llvm.org/PR23161
The misc-static-assert check will not warn on the followings:
assert("Some message" == NULL);
assert(NULL == "Some message");
llvm-svn: 234596
Toma Tabacu [Fri, 10 Apr 2015 13:28:16 +0000 (13:28 +0000)]
[mips] [IAS] Improve comments in MipsAsmParser::expandLoadImm. NFC.
llvm-svn: 234595
Chad Rosier [Fri, 10 Apr 2015 13:19:27 +0000 (13:19 +0000)]
[AArch64] Changes some SchedAlias to WriteRes for Cortex-A57.
Using SchedAliases is convenient and works well for latency and resource
lookup for instructions. However, this creates an entry in
AArch64WriteLatencyTable with a WriteResourceID of 0, breaking any
SchedReadAdvance since the lookup will fail.
http://reviews.llvm.org/D8043
Patch by Dave Estes <cestes@codeaurora.org>!
llvm-svn: 234594
Chad Rosier [Fri, 10 Apr 2015 13:19:21 +0000 (13:19 +0000)]
[AArch64] Adjusts Cortex-A57 machine model to handle zero shift.
http://reviews.llvm.org/D8043
Patch by Dave Estes <cestes@codeaurora.org>!
llvm-svn: 234593
Rafael Espindola [Fri, 10 Apr 2015 13:14:31 +0000 (13:14 +0000)]
Return a pointer instead of having a pointer outparam and a bool return.
llvm-svn: 234592
Aaron Ballman [Fri, 10 Apr 2015 13:05:04 +0000 (13:05 +0000)]
Initializing an uninitialized data member; should be NFC.
llvm-svn: 234591
Rafael Espindola [Fri, 10 Apr 2015 12:54:53 +0000 (12:54 +0000)]
Return a pointer instead of having a pointer outparam and a bool return.
llvm-svn: 234590
Benjamin Kramer [Fri, 10 Apr 2015 12:46:44 +0000 (12:46 +0000)]
Microoptimize DenseMap::clear.
Cache NumEntries locally, it's only used in an assert and using the member
variable prevents the compiler from eliminating the tombstone check for types
with trivial destructors. No functionality change intended.
llvm-svn: 234589
Denis Protivensky [Fri, 10 Apr 2015 11:41:27 +0000 (11:41 +0000)]
[ARM] Implement PLT for dynamic pass
This includes implementation of PLT0 entry.
For testing, libfn.so binary is added since
there's no way to link shared objects with lld yet.
llvm-svn: 234588
Benjamin Kramer [Fri, 10 Apr 2015 11:37:55 +0000 (11:37 +0000)]
Reduce dyn_cast<> to isa<> or cast<> where possible. Clang edition.
No functional change intended.
llvm-svn: 234587
Benjamin Kramer [Fri, 10 Apr 2015 11:24:51 +0000 (11:24 +0000)]
Reduce dyn_cast<> to isa<> or cast<> where possible.
No functional change intended.
llvm-svn: 234586
Denis Protivensky [Fri, 10 Apr 2015 11:05:11 +0000 (11:05 +0000)]
[ARM] Rework GOT/PLT entry generation
Use consistent naming: commonly used generator methods
don't have 'Entry' suffices.
llvm-svn: 234585
Toma Tabacu [Fri, 10 Apr 2015 10:46:59 +0000 (10:46 +0000)]
[mips] [IAS] Make the mips-expansions-bad.s test more readable. NFC.
Move the check lines below the code lines and change the indentation from 8
spaces to 2 spaces.
llvm-svn: 234584
Alexey Bataev [Fri, 10 Apr 2015 10:43:45 +0000 (10:43 +0000)]
[OPENMP] Codegen for 'reduction' clause in 'parallel' directive.
Emit a code for reduction clause. Next code should be emitted for reductions:
static kmp_critical_name lock = { 0 };
void reduce_func(void *lhs[<n>], void *rhs[<n>]) {
...
*(Type<i> *)lhs[i] = RedOp<i>(*(Type<i> *)lhs[i], *(Type<i> *)rhs[i]);
...
}
... void *RedList[<n>] = {&<RHSExprs>[0], ..., &<RHSExprs>[<n> - 1]};
switch (__kmpc_reduce{_nowait}(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>)) {
case 1:
...
<LHSExprs>[i] = RedOp<i>(*<LHSExprs>[i], *<RHSExprs>[i]);
...
__kmpc_end_reduce{_nowait}(<loc>, <gtid>, &<lock>);
break;
case 2:
...
Atomic(<LHSExprs>[i] = RedOp<i>(*<LHSExprs>[i], *<RHSExprs>[i]));
...
break;
default:
;
}
Reduction variables are a kind of a private variables, they have private copies, but initial values are chosen in accordance with the reduction operation.
Differential Revision: http://reviews.llvm.org/D8915
llvm-svn: 234583
Denis Protivensky [Fri, 10 Apr 2015 10:30:04 +0000 (10:30 +0000)]
[ARM] Replace dyn_cast with isa check
llvm-svn: 234582
Arnaud A. de Grandmaison [Fri, 10 Apr 2015 10:13:52 +0000 (10:13 +0000)]
Remove threshold for inserting lifetime markers for named temporaries
Now that TailRecursionElimination has been fixed with r222354, the
threshold on size for lifetime marker insertion can be removed. This
only affects named temporary though, as the patch for unnamed temporaries
is still in progress.
My previous commit (r222993) was not handling debuginfo correctly, but
this could only be seen with some asan tests. Basically, lifetime markers
are just instrumentation for the compiler's usage and should not affect
debug information; however, the cleanup infrastructure was assuming it
contained only destructors, i.e. actual code to be executed, and was
setting the breakpoint for the end of the function to the closing '}', and
not the return statement, in order to show some destructors have been
called when leaving the function. This is wrong when the cleanups are only
lifetime markers, and this is now fixed.
llvm-svn: 234581
Denis Protivensky [Fri, 10 Apr 2015 09:54:10 +0000 (09:54 +0000)]
[ARM] Improve veneer handling and introduce handlePlain method
Handle veneers only for call-like relocations.
llvm-svn: 234580
Dmitry Vyukov [Fri, 10 Apr 2015 09:45:22 +0000 (09:45 +0000)]
[sanitizer][MIPS] Fix warnings on MIPS
Fixed:
- stack frame size warning.
- msse3 flag unused warning.
- GoTsanRuntimeCheck dependency warning reported by cmake.
Change by Sagar Thakur
Reviewed in http://reviews.llvm.org/D8963
llvm-svn: 234579
Szabolcs Sipos [Fri, 10 Apr 2015 08:43:58 +0000 (08:43 +0000)]
Reverting test commit.
llvm-svn: 234578
Szabolcs Sipos [Fri, 10 Apr 2015 08:42:08 +0000 (08:42 +0000)]
Test commit.
llvm-svn: 234577
Denis Protivensky [Fri, 10 Apr 2015 07:53:18 +0000 (07:53 +0000)]
[ARM] Remove unused variable in dynamic pass
llvm-svn: 234576
Alexey Bataev [Fri, 10 Apr 2015 07:48:12 +0000 (07:48 +0000)]
[OPENMP] Fixed cleanup of OpenMP code.
llvm-svn: 234575
Nick Lewycky [Fri, 10 Apr 2015 07:43:19 +0000 (07:43 +0000)]
Fix typo in test. There is no %clangcxx, but there is %clangxx.
llvm-svn: 234574
Denis Protivensky [Fri, 10 Apr 2015 07:42:08 +0000 (07:42 +0000)]
[ARM] Add skeleton for dynamic relocation pass
llvm-svn: 234573
Alexey Bataev [Fri, 10 Apr 2015 06:33:45 +0000 (06:33 +0000)]
[OPENMP] Fixing troubles with lambdas in cleanups.
llvm-svn: 234572
Kostya Serebryany [Fri, 10 Apr 2015 06:32:29 +0000 (06:32 +0000)]
[lib/Fuzzer] Section: How good is my fuzzer?
llvm-svn: 234571
Kostya Serebryany [Fri, 10 Apr 2015 05:44:43 +0000 (05:44 +0000)]
[lib/Fuzzer] explain compatibility with AFL
llvm-svn: 234570
Alexey Bataev [Fri, 10 Apr 2015 05:32:31 +0000 (05:32 +0000)]
[OPENMP] Fixed incompatibility with MSVC, NFC.
llvm-svn: 234568
Jingyue Wu [Fri, 10 Apr 2015 05:03:50 +0000 (05:03 +0000)]
Divergence analysis for GPU programs
Summary:
Some optimizations such as jump threading and loop unswitching can negatively
affect performance when applied to divergent branches. The divergence analysis
added in this patch conservatively estimates which branches in a GPU program
can diverge. This information can then help LLVM to run certain optimizations
selectively.
Test Plan: test/Analysis/DivergenceAnalysis/NVPTX/diverge.ll
Reviewers: resistor, hfinkel, eliben, meheff, jholewinski
Subscribers: broune, bjarke.roune, madhur13490, tstellarAMD, dberlin, echristo, jholewinski, llvm-commits
Differential Revision: http://reviews.llvm.org/D8576
llvm-svn: 234567
David Majnemer [Fri, 10 Apr 2015 04:56:17 +0000 (04:56 +0000)]
[WinEHPrepare] Don't rely on the order of IR
The IPToState table must be emitted after we have generated labels for
all functions in the table. Don't rely on the order of the list of
globals. Instead, utilize WinEHFuncInfo to tell us how many catch
handlers we expect to outline. Once we know we've visited all the catch
handlers, emit the cppxdata.
llvm-svn: 234566
David Majnemer [Fri, 10 Apr 2015 04:52:06 +0000 (04:52 +0000)]
[Sema] Don't assume that an initializer list has an initializer
Given something like 'int({}, 1)', we would try to emit a diagnostic
regarding the excess element in the scalar initializer. However, we
assumed that the initializer list had an element in it.
llvm-svn: 234565
Alexey Bataev [Fri, 10 Apr 2015 04:50:10 +0000 (04:50 +0000)]
[OPENMP] Refactoring of codegen for OpenMP directives.
Refactored API of OpenMPRuntime for compatibility with combined directives.
Differential Revision: http://reviews.llvm.org/D8859
llvm-svn: 234564
Nico Weber [Fri, 10 Apr 2015 04:33:03 +0000 (04:33 +0000)]
Revert r234532 for a bit, it very likely caused crbug.com/475768
llvm-svn: 234563
Oleksiy Vyalov [Fri, 10 Apr 2015 03:59:52 +0000 (03:59 +0000)]
Remove 'z' modifier from printf/sscanf operations in AdbClient - the modifier isn't supported by MS C++ compiler.
llvm-svn: 234562
Hal Finkel [Fri, 10 Apr 2015 03:39:00 +0000 (03:39 +0000)]
[PowerPC] Don't crash on PPC32 i64 fp_to_uint on modern cores
When we have an instruction for this (and, thus, don't generate a runtime
call), we need to custom type legalize this (in a trivial way, just as we do
for fp_to_sint).
Fixes PR23173.
llvm-svn: 234561
Ekaterina Romanova [Fri, 10 Apr 2015 02:39:45 +0000 (02:39 +0000)]
_mm256_blend_epi16 is being cast to __m256d instead of __m256i. Fixing this.
llvm-svn: 234560
Oleksiy Vyalov [Fri, 10 Apr 2015 02:31:37 +0000 (02:31 +0000)]
Wrap socket error handling with SetLastError and IsInterrupted internal functions which can properly treat Windows and POSIX errors.
http://reviews.llvm.org/D8939
llvm-svn: 234559
Rui Ueyama [Fri, 10 Apr 2015 02:18:23 +0000 (02:18 +0000)]
Remove redundant parentheses.
llvm-svn: 234558
Rui Ueyama [Fri, 10 Apr 2015 02:15:13 +0000 (02:15 +0000)]
Do not use default arguments for trivial functions.
llvm-svn: 234557
Rui Ueyama [Fri, 10 Apr 2015 02:06:28 +0000 (02:06 +0000)]
Remove unused return values.
llvm-svn: 234556
Richard Smith [Fri, 10 Apr 2015 02:02:24 +0000 (02:02 +0000)]
[modules] Remove unused MACRO_TABLE record.
llvm-svn: 234555
Enrico Granata [Fri, 10 Apr 2015 01:55:57 +0000 (01:55 +0000)]
Fix a problem where 'process launch' was not correctly re-quoting arguments for the inferior process when handing them down for the actual launch
This covers most of rdar://
20490076, but leaves one corner case still open - namely the case where we try to have arguments of the form foo\ bar (unquoted, but slashed) go through argdumper
llvm-svn: 234554
Davide Italiano [Fri, 10 Apr 2015 01:40:25 +0000 (01:40 +0000)]
Reapply r234378, with test fixed (by emaste).
Hopefully this time the build won't be broken.
llvm-svn: 234553
Rui Ueyama [Fri, 10 Apr 2015 00:19:41 +0000 (00:19 +0000)]
ELF: Don't use APPLY_RELOC macro.
In other ELF ports, we don't use the macro. This patch removes the
macro for consistency and readability.
llvm-svn: 234552
Rui Ueyama [Fri, 10 Apr 2015 00:11:54 +0000 (00:11 +0000)]
ELF: Move Hexagon linker helper function to Hexagon directory.
Because no one except Hexagon uses the header, we don't need to maintain
the header in the common directory. Also de-template the function for
readability.
llvm-svn: 234551
Ahmed Bougacha [Fri, 10 Apr 2015 00:08:48 +0000 (00:08 +0000)]
[AArch64] Promote f16 operations to f32.
For the most common ones (such as fadd), we already did the promotion.
Do the same thing for all the others.
Currently, we'll just crash/assert on all these operations, as
there's no hardware or libcall support whatsoever.
f16 (half) is specified as an interchange - not arithmetic - format,
and is expected to be promoted to single-precision for arithmetic
operations.
While there, teach the legalizer about promoting some of the (mostly
floating-point) operations that we never needed before.
Differential Revision: http://reviews.llvm.org/D8648
See related discussion on the thread for: http://reviews.llvm.org/D8755
llvm-svn: 234550
Ilia K [Fri, 10 Apr 2015 00:04:59 +0000 (00:04 +0000)]
Skip lldb-server tests according to bug 23181
llvm-svn: 234549
Rui Ueyama [Fri, 10 Apr 2015 00:04:44 +0000 (00:04 +0000)]
Remove dead code.
llvm-svn: 234548
Nemanja Ivanovic [Thu, 9 Apr 2015 23:58:16 +0000 (23:58 +0000)]
Add Clang support for remaining integer divide and permute instructions from ISA 2.06
This patch corresponds to review:
http://reviews.llvm.org/D8398
It adds some builtin functions to access the extended divide and bit permute instructions.
llvm-svn: 234547
Nemanja Ivanovic [Thu, 9 Apr 2015 23:54:37 +0000 (23:54 +0000)]
Add LLVM support for remaining integer divide and permute instructions from ISA 2.06
This is the patch corresponding to review:
http://reviews.llvm.org/D8406
It adds some missing instructions from ISA 2.06 to the PPC back end.
llvm-svn: 234546
Fariborz Jahanian [Thu, 9 Apr 2015 23:39:53 +0000 (23:39 +0000)]
[Objective-C Sema] It is permissable to bridge cast to 'id'
of a CFType bridged to some unknown Objective-C type.
rdar://
20113785
llvm-svn: 234545
Benjamin Kramer [Thu, 9 Apr 2015 22:54:53 +0000 (22:54 +0000)]
Update test case for r234543.
llvm-svn: 234544
Benjamin Kramer [Thu, 9 Apr 2015 22:50:07 +0000 (22:50 +0000)]
[CodeGen] Do a more principled fix for PR231653, always use the inner type.
We were still using the MaterializeTemporaryExpr's type to check if the
transform is legal. Always use the inner Expr type.
llvm-svn: 234543
Peter Collingbourne [Thu, 9 Apr 2015 22:42:01 +0000 (22:42 +0000)]
Revert r234477, "Differential Revision: reviews.llvm.org/D7249"
Should unbreak fuzzer buildbot.
llvm-svn: 234542
Rui Ueyama [Thu, 9 Apr 2015 22:04:45 +0000 (22:04 +0000)]
Fix undefined behavior.
Having std:move(mb) and mb->getBuffer() in the same argument list is not safe
because the order of evaluation is not defined.
llvm-svn: 234541
Rui Ueyama [Thu, 9 Apr 2015 21:55:47 +0000 (21:55 +0000)]
ELF: Move CreateELF() from its own file to ELFReader.h.
CreateELF.h was included only by ELFReader.h, and it was used only
by ELFReader class. By making the function a member of the class,
we can remove template parameters.
llvm-svn: 234540
Enrico Granata [Thu, 9 Apr 2015 21:52:19 +0000 (21:52 +0000)]
Don't hardcode the name of the plugin - or it will get out of date, and this code will be broken
Thanks Greg!
llvm-svn: 234539
Rafael Espindola [Thu, 9 Apr 2015 21:50:11 +0000 (21:50 +0000)]
Propagate usage of std:unique_ptr a bit. NFC.
llvm-svn: 234538
Enrico Granata [Thu, 9 Apr 2015 21:33:57 +0000 (21:33 +0000)]
The dyld shared cache class table is not present in the iOS simulator, so do not actually warn to people when running under the simulator
rdar://
20403987
llvm-svn: 234537
Rafael Espindola [Thu, 9 Apr 2015 21:06:11 +0000 (21:06 +0000)]
Update for LLVM api changes.
llvm-svn: 234536
Rafael Espindola [Thu, 9 Apr 2015 21:06:08 +0000 (21:06 +0000)]
Simplify use of formatted_raw_ostream.
formatted_raw_ostream is a wrapper over another stream to add column and line
number tracking.
It is used only for asm printing.
This patch moves the its creation down to where we know we are printing
assembly. This has the following advantages:
* Simpler lifetime management: std::unique_ptr
* We don't compute column and line number of object files :-)
llvm-svn: 234535
Rui Ueyama [Thu, 9 Apr 2015 20:43:38 +0000 (20:43 +0000)]
Update design.rst to remove a mention about round-trip tests.
llvm-svn: 234534
Reid Kleckner [Thu, 9 Apr 2015 20:37:38 +0000 (20:37 +0000)]
Use Linux sysroot in Driver test case from r234521
We don't actually need a real Mac sysroot to make the test pass, just a
linker. This makes the test pass in environments where no ld is on
PATH.
llvm-svn: 234533
Reid Kleckner [Thu, 9 Apr 2015 20:37:24 +0000 (20:37 +0000)]
[SEH] Outline finally blocks using the new variable capture support
WinEHPrepare was going to have to pattern match the control flow merge
and split that the old lowering used, and that wasn't really feasible.
Now we can teach WinEHPrepare to pattern match this, which is much
simpler:
%fp = call i8* @llvm.frameaddress(i32 0)
call void @func(iN [01], i8* %fp)
This prototype happens to match the prototype used by the Win64 SEH
personality function, so this is really simple.
llvm-svn: 234532
Chaoren Lin [Thu, 9 Apr 2015 20:21:42 +0000 (20:21 +0000)]
Fix printf format specifier to silence warning.
llvm-svn: 234531
Ahmed Bougacha [Thu, 9 Apr 2015 20:04:47 +0000 (20:04 +0000)]
[CodeGen] Combine concat_vector of trunc'd scalar to scalar_to_vector.
We already do:
concat_vectors(scalar, undef) -> scalar_to_vector(scalar)
When the scalar is legal.
When it's not, but is a truncated legal scalar, we can also do:
concat_vectors(trunc(scalar), undef) -> scalar_to_vector(scalar)
Which is equivalent, since the upper lanes are undef anyway.
While there, teach the combine to look at more than 2 operands.
Differential Revision: http://reviews.llvm.org/D8883
llvm-svn: 234530
Juergen Ributzka [Thu, 9 Apr 2015 20:00:46 +0000 (20:00 +0000)]
[AArch64][FastISel] Fix integer extend optimization.
The integer extend optimization tries to fold the extend into the load
instruction. This requires us to identify if the extend has already been
emitted or not and act accordingly on it.
The check that was originally performed for this was not sufficient. Besides
checking the ValueMap for a mapped register we also need to check if the
virtual register has already an associated machine instruction that defines it.
This fixes rdar://problem/
20470788.
llvm-svn: 234529
David Majnemer [Thu, 9 Apr 2015 19:53:25 +0000 (19:53 +0000)]
[Sema] Diagnose references to unbound arrays in function definitions
A [*] is only allowed in a declaration for a function, not in its
definition. We didn't correctly recurse on reference types while
looking for it, causing us to crash in CodeGen instead of rejecting it.
llvm-svn: 234528
Siva Chandra [Thu, 9 Apr 2015 19:51:56 +0000 (19:51 +0000)]
[ItaniumABILanguageRuntime] Remove an unneccesary 'typename' keyword.
Summary: This will get the windows bots going.
Test Plan: Build LLDB on Windows.
Reviewers: zturner
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D8934
llvm-svn: 234527
Kaelyn Takata [Thu, 9 Apr 2015 19:43:04 +0000 (19:43 +0000)]
Properly implement warn_unused_result checking for classes/structs.
The previous implementation would copy the attribute from the class to
functions that have the class as their return type when the functions
are first declared. This proved to have two flaws:
1) if the class is forward-declared without the attribute and a
function or method with the class as a its return type is declared,
and afterward the class is defined with warn_unused_result, the
function or method would never inherit the attribute, and
2) the check simply failed for functions and methods that are part of
a template instantiation, regardless of whether the class with
warn_unused_result is part of a specific instantiation or part of
the template itself (presumably because those function/method
declaration does not hit the same code path as a non-template one
and so never inherits the attribute).
The new approach is to instead modify the two places where a function or
method call is checked for the warn_unused_result attribute on the decl
by extending the checks to also look for the attribute on the decl's
return type.
Additionally, the check for return types that have the warn_unused_result
now excludes pointers and references to such types, as such return types do
not necessarily imply a transfer of ownership for the underlying object
being referred to by the return value. This does not change the behavior
of functions that are directly given the warn_unused_result attribute.
llvm-svn: 234526
Eric Christopher [Thu, 9 Apr 2015 19:20:37 +0000 (19:20 +0000)]
Remove duplicated code and consolidate initializers.
llvm-svn: 234525
Siva Chandra [Thu, 9 Apr 2015 19:13:54 +0000 (19:13 +0000)]
[RenderScriptRuntime] Fix build after r234522.
Test Plan: Build LLDB
Reviewers: domipheus
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D8933
llvm-svn: 234524
Siva Chandra [Thu, 9 Apr 2015 18:49:42 +0000 (18:49 +0000)]
[TestBitfields] Correct the attribute name to "aligned" in the testcase.
Summary:
This makes all parts of the test pass with Clang and GCC. They are
enabled with this patch.
Test Plan: dotest.py -C <clang|gcc> -p TestBitfields
Reviewers: vharron, clayborg
Reviewed By: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D8874
llvm-svn: 234523
Siva Chandra [Thu, 9 Apr 2015 18:48:34 +0000 (18:48 +0000)]
[IRForTarget] Strenghten handling of alternate mangling.
Summary:
This fixes an issue with GCC generated binaries wherein an expression
with method invocations on std::string variables was failing. Such use
cases are tested in TestSTL (albeit, in a test marked with
@unittest2.expectedFailure because of other reasons).
The reason for this particular failure with GCC is that the generated
DWARF for std::basic_string<...> is incomplete, which makes clang not
to use the alternate mangling scheme. GCC correctly generates the name
of basic_string<...>:
DW_AT_name "basic_string<char, std::char_traits<char>, std::allocator<char> >"
It also lists the template parameters of basic_string correctly:
DW_TAG_template_type_parameter
DW_AT_name "_CharT"
DW_AT_type <0x0000009c>
DW_TAG_template_type_parameter
DW_AT_name "_Traits"
DW_AT_type <0x00000609>
DW_TAG_template_type_parameter
DW_AT_name "_Alloc"
DW_AT_type <0x000007fb>
However, it does not list the template parameters of std::char_traits<>.
This makes Clang feel (while parsing the expression) that the string
variable is not actually a basic_string instance, and consequently does
not use the alternate mangling scheme.
Test Plan:
dotest.py -C gcc -p TestSTL
-- See it go past the "for" loop expression successfully.
Reviewers: clayborg, spyffe
Reviewed By: clayborg, spyffe
Subscribers: tberghammer, zturner, lldb-commits
Differential Revision: http://reviews.llvm.org/D8846
llvm-svn: 234522
Hans Wennborg [Thu, 9 Apr 2015 18:47:01 +0000 (18:47 +0000)]
[ASan] Don't link against libc++abi when not using libc++
This is a follow-up to r233860 which added -lc++abi when using ASan
on Mac, and broke Chromium's ASan build which doesn't use libc++.
llvm-svn: 234521
Fariborz Jahanian [Thu, 9 Apr 2015 18:36:50 +0000 (18:36 +0000)]
[Objective-C modern translation]. Patch to fix type of
objc_msgSend's first argument to "Class" because
objc_getClass is passed. rdar://
20477025
llvm-svn: 234520
Rafael Espindola [Thu, 9 Apr 2015 18:32:58 +0000 (18:32 +0000)]
clang-format bits of code to make a followup patch easy to read.
llvm-svn: 234519
Rafael Espindola [Thu, 9 Apr 2015 18:29:32 +0000 (18:29 +0000)]
Revert "Refactoring and enhancement to FMA combine."
This reverts commit r234513. It was failing on the bots.
llvm-svn: 234518
Ilia K [Thu, 9 Apr 2015 18:18:10 +0000 (18:18 +0000)]
Fix Debugger::HandleProcessEvent in case when ProcessIOHandler doesn't exist
Summary:
Previously the Debugger::HandleProcessEvent hid a top IOHandler if the
process's IOHandler was inactive and later refreshed it. Usually the
IOHandler.Refresh() prints the (lldb) prompt. The problem was in case of
iOS remote platform when trying to execute 'command source' command.
On this platform the process's IOHandler is empty, therefore the
Debugger::HandleProcessEvent hid a top IOHandler and later refreshed it.
So that the (lldb) prompt was printed with a program output in mixed
order:
was:
```
longlonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglong
longlonglonglonglonglonglonglonglonglonglonglonglonglonglon(lldb)
glonglonglonglonglonglonglonglonglonglonglonglong string
```
now:
```
longlonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglong
longlonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglong
longlonglonglonglonglonglonglonglong string
```
Reviewers: zturner, jingham, clayborg
Reviewed By: clayborg
Subscribers: lldb-commits, jingham, zturner, clayborg
Differential Revision: http://reviews.llvm.org/D8929
llvm-svn: 234517
Zachary Turner [Thu, 9 Apr 2015 18:08:50 +0000 (18:08 +0000)]
[Python] Fix issue configuring sys.path during startup.
Previously, users on Windows had to manually specify PYTHONPATH
to point to the site-packages directory before running LLDB.
The reason for this was because sys.path was being initialized
with a path containing unescaped backslashes, causing escape
sequences to end up in the paths.
llvm-svn: 234516
Rafael Espindola [Thu, 9 Apr 2015 18:08:15 +0000 (18:08 +0000)]
Define a function with "... llvm::func...".
Using this instead of
namespace llvm {
func...
}
Has the advantage that the build fails with a compiler error if it gets out
of sync with the .h file.
llvm-svn: 234515
Robert Flack [Thu, 9 Apr 2015 18:07:58 +0000 (18:07 +0000)]
Fix @skipUnlessPlatform and @skipIfPlatform decorators to work on classes.
The decorators to skip if or unless certain platforms relied on calling
skipTest, which only worked if they were decorating individual test cases.
However for decorating entire classes it needs different behavior. This behavior
is already encapsulated in unittest2.skipIf and unittest2.skipUnless so this
modifies the platform decorators to use these unittest skip decorators.
Test Plan:
./dotest.py -t -p TestObjCMethods2.py
See skipped tests on linux, and see tests run on macosx.
Differential Revision: http://reviews.llvm.org/D8903
llvm-svn: 234514
Olivier Sallenave [Thu, 9 Apr 2015 17:55:26 +0000 (17:55 +0000)]
Refactoring and enhancement to FMA combine.
llvm-svn: 234513
Samuel Benzaquen [Thu, 9 Apr 2015 17:51:01 +0000 (17:51 +0000)]
[clang-tidy] Ignore expressions with incompatible deleters.
Summary:
Do not warn on .reset(.release()) expressions if the deleters are not
compatible.
Using plain assignment will probably not work.
Reviewers: klimek
Subscribers: curdeius, cfe-commits
Differential Revision: http://reviews.llvm.org/D8422
llvm-svn: 234512
Sean Callanan [Thu, 9 Apr 2015 17:42:48 +0000 (17:42 +0000)]
Install a diagnostic consumer into each new AST
context as the first thing we do. This prevents
crashes if some of the initial setup produces
messages or errors.
<rdar://problem/
20457882>
llvm-svn: 234511
Duncan P. N. Exon Smith [Thu, 9 Apr 2015 17:41:20 +0000 (17:41 +0000)]
IR: Preserve use-list order by default in bitcode
Pull the `-preserve-*-use-list-order` flags out of "experimental" mode,
and preserve use-list order by default when serializing to bitcode.
llvm-svn: 234510
Olivier Sallenave [Thu, 9 Apr 2015 17:38:50 +0000 (17:38 +0000)]
Added flag to disable isel instruction on PPC target. Using regular branches instead of isel is more efficient in some cases.
llvm-svn: 234509
Sean Callanan [Thu, 9 Apr 2015 17:26:21 +0000 (17:26 +0000)]
Added the RenderScript language runtime plugin
to the .xcodeproj.
llvm-svn: 234508