Chen Li [Sat, 25 Jul 2015 03:21:06 +0000 (03:21 +0000)]
[LoopUnswitch] Improve loop unswitch pass to find trivial unswitch conditions more effectively
Summary:
This patch improves trivial loop unswitch.
The current trivial loop unswitch only checks if loop header's terminator contains a trivial unswitch condition. But if the loop header only has one reachable successor (due to intentionally or unintentionally missed code simplification), we should consider the successor as part of the loop header. Therefore, instead of stopping at loop header's terminator, we should keep traversing its successors within loop until reach a *real* conditional branch or switch (whose condition can not be constant folded). This change will enable a single -loop-unswitch pass to unswitch multiple trivial conditions (unswitch one trivial condition could open opportunity to unswitch another one in the same loop), while the old implementation can unswitch only one per pass.
Reviewers: reames, broune
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D11481
llvm-svn: 243203
Rui Ueyama [Sat, 25 Jul 2015 03:19:34 +0000 (03:19 +0000)]
COFF: ARM: Implement BRANCH24T relocation.
llvm-svn: 243202
Rui Ueyama [Sat, 25 Jul 2015 03:03:46 +0000 (03:03 +0000)]
COFF: ARM: Implement MOV32T relocation.
llvm-svn: 243201
Jason Molenda [Sat, 25 Jul 2015 02:39:42 +0000 (02:39 +0000)]
Add some initial logging for when lldb is searching for binaries,
dSYMs, or reading binaries out of memory to the 'Host' log channel.
There's more to be done here, both for Mac and for other platforms,
but the initial set of new loggings are useful enough to check in
at this point.
llvm-svn: 243200
Rui Ueyama [Sat, 25 Jul 2015 02:25:14 +0000 (02:25 +0000)]
COFF: ARM: Set correct entry point address.
llvm-svn: 243199
Juergen Ributzka [Sat, 25 Jul 2015 02:16:53 +0000 (02:16 +0000)]
[AArch64][FastISel] Always use an AND instruction when truncating to non-legal types.
When truncating to non-legal types (such as i16, i8 and i1) always use an AND
instruction to mask out the upper bits. This was only done when the source type
was an i64, but not when the source type was an i32.
This commit fixes this and adds the missing i32 truncate tests.
This fixes rdar://problem/
21990703.
llvm-svn: 243198
Rui Ueyama [Sat, 25 Jul 2015 01:44:32 +0000 (01:44 +0000)]
COFF: Handle base relocation as a tuple of relocation type and RVA. NFC.
On x64 and x86, we use only one base relocation type, so we handled
base relocations just as a list of RVAs. That doesn't work well for
ARM becuase we have to handle two types of base relocations on ARM.
This patch changes the type of base relocation from uint32_t to
{reltype, uint32_t} to make it easy to port this code to ARM.
llvm-svn: 243197
Davide Italiano [Sat, 25 Jul 2015 01:19:32 +0000 (01:19 +0000)]
[SemaTemplate] Detect instantiation of unparsed exceptions.
This fixes the clang crash reported in PR24000.
Differential Revision: http://reviews.llvm.org/D11341
llvm-svn: 243196
Rui Ueyama [Sat, 25 Jul 2015 01:16:06 +0000 (01:16 +0000)]
COFF: Split ImportThunkChunk into x86 and x64. NFC.
This change should make it easy to port this code to ARM.
llvm-svn: 243195
Rui Ueyama [Sat, 25 Jul 2015 01:16:04 +0000 (01:16 +0000)]
COFF: Do not align import thunks on 16-byte boundaries on x86.
Looks like MSVC linker aligns them only on x64.
llvm-svn: 243194
Jim Ingham [Sat, 25 Jul 2015 00:52:38 +0000 (00:52 +0000)]
Shorten the lldb timeout, we were seeing occasional failure because
lldb didn't wake up before the target function got a chance to complete.
<rdar://problem/
21990308>
llvm-svn: 243193
Eric Christopher [Sat, 25 Jul 2015 00:48:08 +0000 (00:48 +0000)]
Fix PPCMaterializeInt to check the size of the integer based on the
extension property we're requesting - zero or sign extended.
This fixes cases where we want to return a zero extended 32-bit -1
and not be sign extended for the entire register. Also updated the
already out of date comment with the current behavior.
llvm-svn: 243192
Eric Christopher [Sat, 25 Jul 2015 00:48:06 +0000 (00:48 +0000)]
PPCMaterializeInt should only take a ConstantInt so represent this in the prototype
and fix up all uses.
llvm-svn: 243191
Nick Kledzik [Sat, 25 Jul 2015 00:44:37 +0000 (00:44 +0000)]
Add initial CODE_OWNERS.TXT file
llvm-svn: 243190
Akira Hatanaka [Sat, 25 Jul 2015 00:33:57 +0000 (00:33 +0000)]
Fix test case.
This is a follow-up to r243185, which made changes to pass subtarget
feature "+reserve-x18" instead of backend option "-aarch64-reserve-x18".
llvm-svn: 243189
Rui Ueyama [Sat, 25 Jul 2015 00:20:06 +0000 (00:20 +0000)]
COFF: Don't assume !is64() means i386.
In many places we assumed that is64() means AMD64 and i386 otherwise.
This assumption is not sound because Windows also supports ARM.
The linker doesn't support ARM yet, but this is a first step.
llvm-svn: 243188
Dawn Perchik [Sat, 25 Jul 2015 00:19:39 +0000 (00:19 +0000)]
Specify a language to use when parsing expressions.
This patch adds the option -l/--language to the expression command, for
use when setting the language options or choosing an alternate FE. If
not specified, the target.language setting is used.
Reviewed by: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D11447
llvm-svn: 243187
Akira Hatanaka [Sat, 25 Jul 2015 00:18:31 +0000 (00:18 +0000)]
[AArch64] Define subtarget feature "reserve-x18", which is used to decide
whether register x18 should be reserved.
This change is needed because we cannot use a backend option to set
cl::opt "aarch64-reserve-x18" when doing LTO.
Out-of-tree projects currently using cl::opt option "-aarch64-reserve-x18"
to reserve x18 should make changes to add subtarget feature "reserve-x18"
to the IR.
rdar://problem/
21529937
Differential Revision: http://reviews.llvm.org/D11463
llvm-svn: 243186
Akira Hatanaka [Sat, 25 Jul 2015 00:18:00 +0000 (00:18 +0000)]
[AArch64] Pass subtarget feature "+reserve-x18" instead of passing backend
option "-aarch64-reserve-x18".
This change is needed since backend options do not make it to the backend
when doing LTO and are not capable of changing the behavior of code-gen
passes on a per-function basis.
rdar://problem/
21529937
Differential Revision: http://reviews.llvm.org/D11462
llvm-svn: 243185
Michael J. Spencer [Sat, 25 Jul 2015 00:09:22 +0000 (00:09 +0000)]
[ELF2] Remove broken test.
The exact error the system returns for this case is varies between OS's and file systems.
llvm-svn: 243184
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 23:59:25 +0000 (23:59 +0000)]
DI/Verifier: Fix argument bitrot in DILocalVariable
Add a verifier check that `DILocalVariable`s of tag
`DW_TAG_arg_variable` always have a non-zero 'arg:' field, and those of
tag `DW_TAG_auto_variable` always have a zero 'arg:' field. These are
the only configurations that are properly understood by the backend.
(Also, fix the bad examples in LangRef and test/Assembler, and fix the
bug in Kaleidoscope Ch8.)
A large number of testcases seem to have bitrotted their way forward
from some ancient version of the debug info hierarchy that didn't have
`arg:` parameters. If you have out-of-tree testcases that start failing
in the verifier and you don't care enough to get the `arg:` right, you
may have some luck just calling:
sed -e 's/, arg: 0/, arg: 1/'
or some such, but I hand-updated the ones in tree.
llvm-svn: 243183
Rui Ueyama [Fri, 24 Jul 2015 23:51:14 +0000 (23:51 +0000)]
COFF: Implement Safe SEH support for x86.
An object file compatible with Safe SEH contains a .sxdata section.
The section contains a list of symbol table indices, each of which
is an exception handler function. A safe SEH-enabled executable
contains a list of exception handler RVAs. So, what the linker has
to do to support Safe SEH is basically to read the .sxdata section,
interpret the contents as a list of symbol indices, unique-fy and
sort their RVAs, and then emit that list to .rdata. This patch
implements that feature.
llvm-svn: 243182
Jim Ingham [Fri, 24 Jul 2015 23:40:32 +0000 (23:40 +0000)]
Make sure we resolve ~ in paths coming from the plist in a dSYM before adding them
to the path remappings.
Also don't add the paths to the path mapping when DebugSymbols tells up about files, since
we'll just do that again when we read in the dSYM.
<rdar://problem/
21986208>
llvm-svn: 243181
Greg Clayton [Fri, 24 Jul 2015 23:38:01 +0000 (23:38 +0000)]
Use double-checked locking to avoid locking the Module mutex if we don't need to. This avoid a deadlock we were seeing in Xcode.
<rdar://problem/
21512067>
llvm-svn: 243180
Michael Kruse [Fri, 24 Jul 2015 23:30:31 +0000 (23:30 +0000)]
Normalize whitespace in makefiles
Tabs for rules and space for line continuations.
llvm-svn: 243179
Rui Ueyama [Fri, 24 Jul 2015 23:24:45 +0000 (23:24 +0000)]
COFF: Fix base relocation type for x86.
llvm-svn: 243178
Rui Ueyama [Fri, 24 Jul 2015 22:58:44 +0000 (22:58 +0000)]
COFF: Fix __ImageBase symbol relocation.
__ImageBase is a special symbol whose value is the image base address.
Previously, we handled __ImageBase symbol as an absolute symbol.
Absolute symbols point to specific locations in memory and the locations
never change even if an image is base-relocated. That means that we
don't have base relocation entries for absolute symbols.
This is not a case for __ImageBase. If an image is base-relocated, its
base address changes, and __ImageBase needs to be shifted as well.
So we have to have base relocations for __ImageBase. That means that
__ImageBase is not really an absolute symbol but a different kind of
symbol.
In this patch, I introduced a new type of symbol -- DefinedRelative.
DefinedRelative is similar to DefinedAbsolute, but it has not a VA but RVA
and is a subject of base relocation. Currently only __ImageBase is of
the new symbol type.
llvm-svn: 243176
Jason Molenda [Fri, 24 Jul 2015 22:42:03 +0000 (22:42 +0000)]
Log socket communications to LIBLLDB_LOG_COMMUNICATION instead of
the Host channel.
http://reviews.llvm.org/D11497
llvm-svn: 243175
Zachary Turner [Fri, 24 Jul 2015 22:32:22 +0000 (22:32 +0000)]
Some fixes to get LLDB compiling with MSVC 2015.
llvm-svn: 243174
Alex Lorenz [Fri, 24 Jul 2015 22:22:50 +0000 (22:22 +0000)]
MIR Serialization: Serialize MachineFrameInfo's callee saved information.
This commit serializes the callee saved information from the class
'MachineFrameInfo'. This commit extends the YAML mappings for the fixed and
the ordinary stack objects and adds an optional 'callee-saved-register'
attribute. This attribute is used to serialize the callee save information.
llvm-svn: 243173
Lawrence Hu [Fri, 24 Jul 2015 22:01:49 +0000 (22:01 +0000)]
Handle loop with negtive induction variable increment
This patch extend LoopReroll pass to hand the loops which
is similar to the following:
while (len > 1) {
sum4 += buf[len];
sum4 += buf[len-1];
len -= 2;
}
llvm-svn: 243171
Alexey Samsonov [Fri, 24 Jul 2015 22:01:07 +0000 (22:01 +0000)]
[CMake] Workaround for PR24222: don't fail if we can't target COMPILER_RT_TEST_TARGET_TRIPLE unless it was explicitly set.
Summary:
Although we assume that we can always target host triple, relax
this check to avoid failing at configure-time for cases when we are
not able to correctly infer/verify host triple for some reasons.
See http://llvm.org/bugs/show_bug.cgi?id=24222 for more details.
Reviewers: hans
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D11496
llvm-svn: 243170
Aaron Ballman [Fri, 24 Jul 2015 21:49:29 +0000 (21:49 +0000)]
Replacing some diagnostic string literals with tablegen equivalents because the wording is identical; NFC.
llvm-svn: 243169
Michael Kruse [Fri, 24 Jul 2015 21:40:22 +0000 (21:40 +0000)]
Add isl_ffs.c to file list in autoconf builds
In most configurations it doesn't contain any symbol, but add for
completeness.
llvm-svn: 243168
Pete Cooper [Fri, 24 Jul 2015 21:38:01 +0000 (21:38 +0000)]
Remove unnecessary null check. NFC.
Since both places which set this variable do so with dyn_cast, and not
dyn_cast_or_null, its impossible to get a nullptr here, so we can remove
the check.
llvm-svn: 243167
Siva Chandra [Fri, 24 Jul 2015 21:30:58 +0000 (21:30 +0000)]
Add option eTypeOptionHideEmptyAggregates.
Summary:
For certain data structures, when the synthetic child provider returns
zero children, a summary like "Empty instance of <typename>" could be
more appropriate than something like "size=0 {}". This new option helps
hide the trailing "{}".
This is also exposed with a -h option for the command "type summary add".
Reviewers: granata.enrico
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D11473
llvm-svn: 243166
Hans Wennborg [Fri, 24 Jul 2015 21:24:48 +0000 (21:24 +0000)]
LibompMicroTests.cmake: Invoke 'test' instead of using '[ x -eq y ]'
I was getting this cmake error on Mac OS X:
CMake Error: Error in cmake code at
/tmp/openmp/runtime/cmake/LibompMicroTests.cmake:140:
Parse error. Function missing ending ")". Instead found bad character with text "[".
Perhaps invoking 'test' is less confusing for cmake.
Differential Revision: http://reviews.llvm.org/D11493
llvm-svn: 243165
Pete Cooper [Fri, 24 Jul 2015 21:13:43 +0000 (21:13 +0000)]
Use make_range(rbegin(), rend()) to allow foreach loops. NFC.
Instead of the pattern
for (auto I = x.rbegin(), E = x.end(); I != E; ++I)
we can use make_range to construct the reverse range and iterate using
that instead.
llvm-svn: 243163
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 21:11:06 +0000 (21:11 +0000)]
DI: Fix unit tests after r243160
These always empty fields are gone, so don't test that they're empty.
llvm-svn: 243162
Michael J. Spencer [Fri, 24 Jul 2015 21:03:07 +0000 (21:03 +0000)]
[ELF2] Add a new ELF linker based on the new PE/COFF linker.
Differential Revision: http://reviews.llvm.org/D11188
llvm-svn: 243161
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 20:56:36 +0000 (20:56 +0000)]
DI: Remove unnecessary DICompositeTypeBase
Remove unnecessary and confusing common base class for `DICompositeType`
and `DISubroutineType`.
While at a high-level `DISubroutineType` is a sort of composite of other
types, it has no shared code paths, and its fields are completely
disjoint. This relationship was left over from the old debug info
hierarchy.
llvm-svn: 243160
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 20:56:10 +0000 (20:56 +0000)]
DI: Simplify DebugInfoFinder::processType(), NFC
Handle `DISubroutineType` up-front rather than as part of a branch for
`DICompositeTypeBase`. The only shared code path was looking through
the base type, but `DISubroutineType` can never have a base type.
This also removes the last use of `DICompositeTypeBase`, since we can
strengthen the cast to `DICompositeType`.
llvm-svn: 243159
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 20:46:46 +0000 (20:46 +0000)]
DI: Remove dead code: getDICompositeType()
llvm-svn: 243158
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 20:45:26 +0000 (20:45 +0000)]
AsmPrinter: Use DICompositeType in updateAcceleratorTables(), NFC
`DISubroutineType` is impossible at this `dyn_cast` site, since we're
only dealing with named types and `DISubroutineType` cannot be named.
Strengthen the `dyn_cast` to `DICompositeType`.
llvm-svn: 243157
Alex Lorenz [Fri, 24 Jul 2015 20:35:40 +0000 (20:35 +0000)]
MIR Serialization: Serialize the simple virtual register allocation hints.
This commit serializes the virtual register allocations hints of type 0.
These hints specify the preferred physical registers for allocations.
llvm-svn: 243156
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 20:34:41 +0000 (20:34 +0000)]
DI: Make getOrCreateLimitedType() return DICompositeType, NFC
Change `getOrCreateLimitedType()` to return a `DICompositeType` and
remove the casts from its callers. Inside, I've strengthened a `cast`
from `DICompositeTypeBase`, but the casts in the callers already prove
that this is safe. There should be no functionality change here.
llvm-svn: 243155
Michael Kruse [Fri, 24 Jul 2015 20:33:22 +0000 (20:33 +0000)]
Make the lit configuration Python 3 compatible
by using the same techniques as LLVM's lit configuration.
llvm-svn: 243154
Yaron Keren [Fri, 24 Jul 2015 20:18:27 +0000 (20:18 +0000)]
Select the highest version of the mingw toolchain found using Generic_GCC::GCCVersion
similar to the way Generic_GCC::GCCInstallationDetector::ScanLibDirForGCCTriple works.
llvm-svn: 243153
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 20:16:36 +0000 (20:16 +0000)]
DI: Remove DIDerivedTypeBase
Remove an unnecessary (and confusing) common subclass for
`DIDerivedType` and `DICompositeType`. These classes aren't really
related, and even in the old debug info hierarchy, there was a
long-standing FIXME to separate them.
llvm-svn: 243152
Ismail Pazarbasi [Fri, 24 Jul 2015 20:00:49 +0000 (20:00 +0000)]
__tsan::ThreadCreate takes incorrect value for detached argument
Summary:
PTHREAD_CREATE_DETACHED has a different value on Mac OS X. Since both
PTHREAD_CREATE_JOINABLE and PTHREAD_CREATE_DETACHED are non-zero,
`__tsan::ThreadCreate` always creates detached threads.
Reviewers: kcc, samsonov, glider
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D10606
llvm-svn: 243151
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 19:57:19 +0000 (19:57 +0000)]
Verifier: Sink filename check into visitMDCompositeType(), NFC
We really only want to check this for unions and classes (all the other
tags have been ruled out), so simplify the check and move it to the
right place.
llvm-svn: 243150
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 19:52:18 +0000 (19:52 +0000)]
Verifier: Remove unnecessary references to DW_TAG_subroutine_type, NFC
Remove unnecessary references to `DW_TAG_subroutine_type` in
`visitDICompositeType()` and `visitDIDerivedTypeBase()`, since
`visitDISubroutineType()` doesn't call either of those (and shouldn't,
since subroutine types are really quite special).
llvm-svn: 243149
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 19:42:12 +0000 (19:42 +0000)]
DI: Clarify isUnsignedDIType(), NFC
Refactor `isUnsignedDIType()` to deal with `DICompositeType` explicitly.
Since `DW_TAG_subroutine_type` isn't handled here (the assertions about
tags rule it out), this allows strengthening the `dyn_cast` to
`DIDerivedType`.
Besides making the code clearer, this it removes a use of
`DIDerivedTypeBase`.
llvm-svn: 243148
Renato Golin [Fri, 24 Jul 2015 19:29:05 +0000 (19:29 +0000)]
[libunwind] Flip order of extern "C" and attribute(visibility)
GCC doesn't seems to like having the attribute before extern, but
Clang accepts it either way. This patch makes it compile on both.
llvm-svn: 243147
Pete Cooper [Fri, 24 Jul 2015 19:19:26 +0000 (19:19 +0000)]
Add const to some Type* parameters which didn't need to be mutable. NFC.
We were only getting the size of the type which doesn't need to modify
the type.
llvm-svn: 243146
Diego Novillo [Fri, 24 Jul 2015 19:18:32 +0000 (19:18 +0000)]
Remove unused variable. NFC.
llvm-svn: 243145
Yaron Keren [Fri, 24 Jul 2015 19:18:17 +0000 (19:18 +0000)]
Base the sys-root/mingw/include path on sysroot and not on /usr.
Thanks to Richard Smith for pointing this out!
llvm-svn: 243144
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 19:17:20 +0000 (19:17 +0000)]
DI: Strengthen some dyn_casts to DIDerivedType, NFC
The surrounding code proves in both cases that these must be
`DIDerivedType` if they're `DIDerivedTypeBase`, so strengthen the
`dyn_cast`s to the more specific type.
llvm-svn: 243143
Michael Kruse [Fri, 24 Jul 2015 19:09:27 +0000 (19:09 +0000)]
Compile fix; add missing ISL files
The last ISL update added two files that we must also add to Polly.
llvm-svn: 243142
Jingyue Wu [Fri, 24 Jul 2015 19:05:53 +0000 (19:05 +0000)]
Remove the user-count threshold when analyzing read attributes
Summary:
This threshold limited FunctionAttrs ability to prove arguments to be read-only.
In NVPTX, a specialized instruction ld.global.nc can be used to load memory
with non-coherent texture cache. We notice that in SHOC [1] benchmark, some
function arguments are not marked with readonly because FunctionAttrs reaches
a hardcoded threshold when analysis uses.
Removing this threshold won't cause significant regression in compilation time, because the worst-case time complexity of the algorithm is still O(# of instructions) for each parameter.
Patched by Xuetian Weng.
[1] https://github.com/vetter/shoc
Reviewers: nlewycky, jingyue, nicholas
Subscribers: nicholas, test, llvm-commits
Differential Revision: http://reviews.llvm.org/D11311
llvm-svn: 243141
Philip Reames [Fri, 24 Jul 2015 19:01:39 +0000 (19:01 +0000)]
[RewriteStatepointsForGC] Adjust naming scheme to be more stable
The names for instructions inserted were previous dependent on iteration order. By deriving the names from the original instructions, we can avoid instability in tests without resorting to ordered traversals. It also makes the IR mildly easier to read at large scale.
llvm-svn: 243140
Evgeniy Stepanov [Fri, 24 Jul 2015 19:00:16 +0000 (19:00 +0000)]
[asan] Only link liblog on Android when it exists.
With r242975, liblog is only used on certain older platforms.
llvm-svn: 243139
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 18:58:32 +0000 (18:58 +0000)]
DI: Strengthen block-byref cast to DIDerivedType, NFC
This code is visiting the members of a block-byref, and we know those
are all `DIDerivedType`. Strengthen the cast.
llvm-svn: 243138
Evgeniy Stepanov [Fri, 24 Jul 2015 18:57:44 +0000 (18:57 +0000)]
[asan] Fix android system property access on aarch64.
__system_property_get is gone from the NDK headers and libs in 64-bit
toolchains, but it is still exported by platform libc.so.
llvm-svn: 243137
Pete Cooper [Fri, 24 Jul 2015 18:55:49 +0000 (18:55 +0000)]
Use foreach loops for StructType::elements(). NFC.
We had a few places where we did
for (unsigned i = 0, e = STy->getNumElements(); i != e; ++i) {
but those could instead do
for (auto *EltTy : STy->elements()) {
llvm-svn: 243136
Pete Cooper [Fri, 24 Jul 2015 18:29:09 +0000 (18:29 +0000)]
Add const to a bunch of Type* in DataLayout. NFC.
Almost all methods in DataLayout took mutable pointers but didn't need to.
These were only accessing constant methods of the types, or using the Type*
to key a map. Neither of these needs a mutable pointer.
llvm-svn: 243135
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 18:17:17 +0000 (18:17 +0000)]
DI: Only DICompositeType has getElements(), NFC
There is an assertion inside `DICompositeTypeBase::getElements()` that
`this` is not a `DISubroutineType`, leaving only `DICompositeType`.
Make that clear at the call sites.
llvm-svn: 243134
Chih-Hung Hsieh [Fri, 24 Jul 2015 18:12:54 +0000 (18:12 +0000)]
Correct x86_64 Android fp128 mangled name
These changes are for Android x86_64 targets to be compatible with current Android g++.
https://llvm.org/bugs/show_bug.cgi?id=23897
Use 'g' and 'Cg' for "long double" and "long double _Complex" mangled type names.
Differential Revision: http://reviews.llvm.org/D11466
llvm-svn: 243133
Tom Stellard [Fri, 24 Jul 2015 18:07:14 +0000 (18:07 +0000)]
Fix double implementation of log
We need to use M_LOG2E instead of M_LOG2E_F.
llvm-svn: 243132
Tom Stellard [Fri, 24 Jul 2015 18:07:12 +0000 (18:07 +0000)]
Implement accurate log2 function
Use the implementation was ported from the AMD builtin library rather
than LLVM Intrinsics.
This has been tested with piglit, OpenCV, and the ocl conformance tests.
llvm-svn: 243131
Tom Stellard [Fri, 24 Jul 2015 18:07:06 +0000 (18:07 +0000)]
Use llvm intrinsics for native_log and native_log2
llvm-svn: 243130
Duncan P. N. Exon Smith [Fri, 24 Jul 2015 18:05:58 +0000 (18:05 +0000)]
DI: Make StaticDataMemberCache type-safe, NFC
Change `StaticDataMemberCache` to store references to `DIDerivedType`
directly, and remove now-unnecessary casts from the accessors.
llvm-svn: 243129
Alex Lorenz [Fri, 24 Jul 2015 17:44:49 +0000 (17:44 +0000)]
MIR Parser: Run the machine verifier after initializing machine functions.
llvm-svn: 243128
Lang Hames [Fri, 24 Jul 2015 17:40:04 +0000 (17:40 +0000)]
[RuntimeDyld] MachO: Add support for ARM scattered vanilla relocations.
llvm-svn: 243126
Chih-Hung Hsieh [Fri, 24 Jul 2015 17:39:13 +0000 (17:39 +0000)]
Test commit.
llvm-svn: 243125
Alex Lorenz [Fri, 24 Jul 2015 17:36:55 +0000 (17:36 +0000)]
MIR Tests: Add liveins and successors to make tests pass with machine verifier.
This commit adds the liveins and successors properties to machine basic blocks
in some of the MIR tests to ensure that the tests will pass when the MIR parser
will run the machine verifier after initializing a machine function.
llvm-svn: 243124
Alex Lorenz [Fri, 24 Jul 2015 17:31:55 +0000 (17:31 +0000)]
MIR Tests: Make the basic block successor test an X86 specific test.
This commit moves and transforms the generic test
'CodeGen/MIR/successor-basic-blocks.mir' into an X86 specific test
'CodeGen/MIR/X86/successor-basic-blocks.mir'. This change is required in order
to enable the machine verifier for the MIR parser, as the machine verifier
verifies that the machine basic blocks contain instructions that actually
determine the machine basic block successors.
llvm-svn: 243123
Igor Breger [Fri, 24 Jul 2015 17:24:15 +0000 (17:24 +0000)]
AVX-512: Implemented encoding , DAG lowering and intrinsics for Integer Truncate with/without saturation
Added tests for DAG lowering ,encoding and intrinsic
Differential Revision: http://reviews.llvm.org/D11218
llvm-svn: 243122
Chandler Carruth [Fri, 24 Jul 2015 17:23:09 +0000 (17:23 +0000)]
Update for r243115 which changed the DataLayout API on TargetMachine but
didn't update the gold-plugin.
llvm-svn: 243121
Argyrios Kyrtzidis [Fri, 24 Jul 2015 17:18:25 +0000 (17:18 +0000)]
[test] Attempt to fix failing test in non-darwin bots.
llvm-svn: 243120
Argyrios Kyrtzidis [Fri, 24 Jul 2015 17:00:19 +0000 (17:00 +0000)]
[code-completion] Strip outer nullability annotations when completing method implementations.
The outer nullability is transferred from the declaration to the implementation so including them is redundant.
The inner ones are not transferred so they are kept to match the exact types. When we transfer the inner ones as well
adding them in the implementation will become redundant and we should strip those as well.
rdar://
21737451
llvm-svn: 243119
Greg Clayton [Fri, 24 Jul 2015 16:55:00 +0000 (16:55 +0000)]
Bind to the loopback when we are expecting a connection from 127.0.0.1 so we don't set off firewall protections.
<rdar://problem/
17897318>
llvm-svn: 243118
Hans Wennborg [Fri, 24 Jul 2015 16:16:09 +0000 (16:16 +0000)]
test-release.sh: Defer test errors until the end
This makes the script run to the end and produce tarballs even on test
failures, and then highlights any errors afterwards.
(I first tried just storing the errors in a global variable, but that
didn't work as the "test_llvmCore" function invocation is actually
running as a sub-shell.)
Differential Revision: http://reviews.llvm.org/D11478
llvm-svn: 243116
Mehdi Amini [Fri, 24 Jul 2015 16:04:29 +0000 (16:04 +0000)]
LLVM API Change: the Module always owns the DataLayout
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 243115
Mehdi Amini [Fri, 24 Jul 2015 16:04:22 +0000 (16:04 +0000)]
Remove access to the DataLayout in the TargetMachine
Summary:
Replace getDataLayout() with a createDataLayout() method to make
explicit that it is intended to create a DataLayout only and not
accessing it for other purpose.
This change is the last of a series of commits dedicated to have a
single DataLayout during compilation by using always the one owned
by the module.
Reviewers: echristo
Subscribers: jholewinski, llvm-commits, rafael, yaron.keren
Differential Revision: http://reviews.llvm.org/D11103
(cherry picked from commit
5609fc56bca971e5a7efeaa6ca4676638eaec5ea)
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 243114
Sanjay Patel [Fri, 24 Jul 2015 16:02:14 +0000 (16:02 +0000)]
fix wrong comment; NFC
llvm-svn: 243113
Aaron Ballman [Fri, 24 Jul 2015 15:47:32 +0000 (15:47 +0000)]
Correcting some comments. (1) the function is named throughUsingDecl() and not throughUsingDeclaration(). (2) Testing shows that this does work for shadowed variables as well as shadowed functions. I could not find an example where this matcher was failing. NFC.
llvm-svn: 243111
Tobias Grosser [Fri, 24 Jul 2015 13:12:17 +0000 (13:12 +0000)]
Update isl to isl-0.15-86-g595055e
Besides a couple of cleanups and refactorings in isl, this change set fixes a
couple of bugs in isl, that can cause issues during code generation.
llvm-svn: 243110
Aaron Ballman [Fri, 24 Jul 2015 12:47:27 +0000 (12:47 +0000)]
Silence a "not all control paths return a value" warning from MSVC. Should also silence a -Wreturn-type warning. NFC intended.
llvm-svn: 243109
Aaron Ballman [Fri, 24 Jul 2015 12:44:02 +0000 (12:44 +0000)]
Visual Studio is more permissive than other compilers; amends r243107 to fix a compile error.
llvm-svn: 243108
Aaron Ballman [Fri, 24 Jul 2015 12:35:41 +0000 (12:35 +0000)]
Add an AST matcher, isFinal(), for testing whether a method or class declaration are marked final.
llvm-svn: 243107
NAKAMURA Takumi [Fri, 24 Jul 2015 11:55:11 +0000 (11:55 +0000)]
llvm/test/tools/dsymutil/ARM/lit.local.cfg: Fix possibly typo, s/X86/ARM/.
llvm-svn: 243106
Yaron Keren [Fri, 24 Jul 2015 11:01:45 +0000 (11:01 +0000)]
Apparently some of the bots add .svn dirs inside the test/Driver/Inputs
directory structure. Try to make mingw toolchain resilient to such surprises.
llvm-svn: 243105
Ewan Crawford [Fri, 24 Jul 2015 10:01:11 +0000 (10:01 +0000)]
Initialize variable to prevent garbage values (RenderScriptRuntime)
The kernels_found variable is not initialized, so if it is not assigned true on line 823, then it will be a garbage value in the branch condition on line 828. This patch initializes the variable to false.
Patch by neilparikh.
Reviewers: domipheus
Differential Revision: http://reviews.llvm.org/D11323
llvm-svn: 243104
Luke Cheeseman [Fri, 24 Jul 2015 09:57:05 +0000 (09:57 +0000)]
[ARM] - Fix lowering of shufflevectors in AArch32
Some shufflevectors are currently being incorrectly lowered in the AArch32
backend as the existing checks for detecting the NEON operations from the
shufflevector instruction expects the shuffle mask and the vector operands to be
of the same length.
This is not always the case as the mask may be twice as long as the operand;
here only the lower half of the shufflemask gets checked, so provided the lower
half of the shufflemask looks like a vector transpose (or even is just all -1
for undef) then the intrinsics may get incorrectly lowered into a vector
transpose (VTRN) instruction.
This patch fixes this by accommodating for both cases and adds regression tests.
Differential Revision: http://reviews.llvm.org/D11407
llvm-svn: 243103
Pavel Labath [Fri, 24 Jul 2015 09:52:25 +0000 (09:52 +0000)]
Revert "Fix an issue where LLDB would run out of stack space ..."
This commit introduced an infinite recursion in
ValueObjectChild::CanUpdateWithInvalidExecutionContext (because FollowParentChain also considers
the current object), which broke nearly all the tests. Ignoring the current object removes the
recursion, but two tests still time out (TestDataFormatterLibcxxList.py and
TestValueObjectRecursion.py) for some reason. Reverting for now.
llvm-svn: 243102
Yaron Keren [Fri, 24 Jul 2015 09:31:57 +0000 (09:31 +0000)]
Try to appease clang buildbot by forcing libstdc++ in mingw.cpp test.
llvm-svn: 243101
Luke Cheeseman [Fri, 24 Jul 2015 09:31:48 +0000 (09:31 +0000)]
When lowering vector shifts a check is performed to see if the value to shift by
is an immediate, in this check the value is negated and stored in and int64_t.
The value can be -2^63 yet the result cannot be stored in an int64_t and this
gives some undefined behaviour causing failures. The negation is only necessary
when the values is within a certain range and so it should not need to negate
-2^63, this patch introduces this and also a regression test.
Differential Revision: http://reviews.llvm.org/D11408
llvm-svn: 243100
Tamas Berghammer [Fri, 24 Jul 2015 08:54:22 +0000 (08:54 +0000)]
Improve C++ function name handling and step-in avoid regerxp handling
If the function is a template then the return type is part of the
function name. This CL fixes the parsing of these function names in
the case when the return type contains ':'.
The name of free functions in C++ don't have context part. Fix the
logic geting the function name without arguments out from a full
function name to handle this case.
Change the handling of step-in-avoid-regexp to match the value against
the function name without it's arguments and return value. This is
required because the default regex ("^std::") would match any template
function returning an std object.
Fifferential revision: http://reviews.llvm.org/D11461
llvm-svn: 243099
Yaron Keren [Fri, 24 Jul 2015 08:50:15 +0000 (08:50 +0000)]
Add extensive tests for the mingw toolchain and remove trailing slash from Arch.
Address Richard Smith comments: remove the trailing seperator from the Arch
variable, implement six mingw_* trees under tools/clangtest/Driver/Inputs
and merge linux and Windows tests into a universal test that uses these trees.
llvm-svn: 243098