Simon Atanasyan [Mon, 14 Jul 2014 09:01:40 +0000 (09:01 +0000)]
[Mips] Delete the empty directory.
llvm-svn: 212928
Simon Atanasyan [Mon, 14 Jul 2014 08:54:40 +0000 (08:54 +0000)]
[Mips] Remove unused test input file.
llvm-svn: 212927
Evgeniy Stepanov [Mon, 14 Jul 2014 08:54:10 +0000 (08:54 +0000)]
[asan] Remove XFAIL:android from 2 newly passing tests.
Fixed in r212872.
llvm-svn: 212926
Tobias Grosser [Mon, 14 Jul 2014 08:32:01 +0000 (08:32 +0000)]
DeadCodeElimination: Fix liveout computation
We move back to a simple approach where the liveout is the last must-write
statement for a data-location plus all may-write statements. The previous
approach did not work out. We would have to consider per-data-access
dependences, instead of per-statement dependences to correct it. As this adds
complexity and it seems we would not gain anything over the simpler approach
that we implement in this commit, I moved us back to the old approach of
computing the liveout, but enhanced it to also add may-write accesses.
We also fix the test case and explain why we can not perform dead code
elimination in this case.
llvm-svn: 212925
Alexander Kornienko [Mon, 14 Jul 2014 07:37:05 +0000 (07:37 +0000)]
Suggest automated replacements of C-style casts with C++ casts.
Summary:
This patch implements a subset of possible replacements of C-style
casts with const_cast/static_cast/reinterpret_cast. This should cover a large
portion of cases in real code. Handling of a few more cases may be implemented
eventually.
Reviewers: sbenza, djasper
Reviewed By: djasper
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D4478
llvm-svn: 212924
Todd Fiala [Mon, 14 Jul 2014 06:24:44 +0000 (06:24 +0000)]
Modified gdb-remote tests to run with automatically-chosen ports.
Now that llgs supports communicating the 0-port choose-a-port
mechanism and can communicate that back to a caller via the
--named-pipe option (at parity with debugserver), we use this
mechanism to always start llgs and debugserver gdb-remote
protocol tests without needing to use some port arbitration
mechanism. This eliminates some potential intermittent failures vs. the
previous random port and collision-avoidance strategy used.
llvm-svn: 212923
Bill Wendling [Mon, 14 Jul 2014 06:22:36 +0000 (06:22 +0000)]
Support lowering of empty aggregates.
This crash was pretty common while compiling Rust for iOS (armv7). Reason -
SjLj preparation step was lowering aggregate arguments as ExtractValue +
InsertValue. ExtractValue has assertion which checks that there is some data in
value, which is not true in case of empty (no fields) structures. Rust uses
them quite extensively so this patch uses a 'select true, %val, undef'
instruction to lower the argument.
Patch by Valerii Hiora.
llvm-svn: 212922
NAKAMURA Takumi [Mon, 14 Jul 2014 05:07:07 +0000 (05:07 +0000)]
[CMake] LINK_COMPONENTS: Add also corresponding MCTargetDesc and TargetInfo as well, when target names or "nativecodegen" are specified.
llvm-svn: 212921
NAKAMURA Takumi [Mon, 14 Jul 2014 05:01:53 +0000 (05:01 +0000)]
[CMake] Update libdeps.
llvm-svn: 212920
NAKAMURA Takumi [Mon, 14 Jul 2014 04:59:27 +0000 (04:59 +0000)]
[CMake] Reorder libdeps by alphabetical order.
llvm-svn: 212919
NAKAMURA Takumi [Mon, 14 Jul 2014 02:52:19 +0000 (02:52 +0000)]
NVPTX/LLVMBuild.txt: Add "Scalar" to required_libraries. It is really referenced.
llvm-svn: 212918
NAKAMURA Takumi [Mon, 14 Jul 2014 02:52:08 +0000 (02:52 +0000)]
Object/LLVMBuild.txt: Sort required_libraries by alphabetical order.
llvm-svn: 212917
Richard Smith [Mon, 14 Jul 2014 02:28:44 +0000 (02:28 +0000)]
Superficial fix for PR20218: binding a function lvalue to a const reference to
a function pointer is neither better nor worse than binding a function lvalue
to a function rvalue reference. Don't get confused and think that both bindings
are binding to a function lvalue (which would make the lvalue form win); the
const reference is binding to an rvalue.
The "real" bug in PR20218 is still present: we're getting the wrong answer from
template argument deduction, and that's what leads us to this weird overload
set.
llvm-svn: 212916
Andrea Di Biagio [Sun, 13 Jul 2014 21:02:14 +0000 (21:02 +0000)]
[DAGCombiner] Fix a crash caused by a missing check for legal type when trying to fold shuffles.
Verify that DAGCombiner does not crash when trying to fold a pair of shuffles
according to rule (added at r212539):
(shuffle (shuffle A, Undef, M0), Undef, M1) -> (shuffle A, Undef, M2)
The DAGCombiner avoids folding shuffles if the resulting shuffle dag node
is not legal for the target. That means, the resulting shuffle must have
legal type and legal mask.
Before, the DAGCombiner only called method
'TargetLowering::isShuffleMaskLegal' to check if it was "safe" to fold according
to the above-mentioned rule. However, this caused a crash in the x86 backend
since method 'isShuffleMaskLegal' always expects to be called on a
legal vector type.
llvm-svn: 212915
Saleem Abdulrasool [Sun, 13 Jul 2014 19:03:45 +0000 (19:03 +0000)]
MC: make MCWin64EHInstruction a POD-like struct
This is the first of a number of changes designed to generalise
MCWin64EHInstruction to support different target architectures. An ordered set
(vector) of these instructions is saved per frame to permit the emission of
information for Windows NT style unwinding. The only bit of information which
is actually target specific here is the Opcode for the unwinding bytecode. The
remainder of the information is simply generic information that is relevant to
the Windows NT unwinding model.
Remove the accessors for the fields, making them const and public instead. Sink
the knowledge of the alias'ed name into the single source and sink a single-use
check method into the use.
llvm-svn: 212914
Saleem Abdulrasool [Sun, 13 Jul 2014 19:03:40 +0000 (19:03 +0000)]
MC: make helper function be more const-correct
Introduce const-ness on parameters, they are used as read-only and should not be
modified. NFC.
llvm-svn: 212913
Saleem Abdulrasool [Sun, 13 Jul 2014 19:03:36 +0000 (19:03 +0000)]
MC: make DWARF and Windows unwinding handling more similar
Rename member variables and functions for the MCStreamer for DWARF-like
unwinding management. Rename the Windows ones as well and make the naming and
handling similar across the two. No functional change intended.
llvm-svn: 212912
Pavel Chupin [Sun, 13 Jul 2014 17:11:45 +0000 (17:11 +0000)]
Remove trailing spaces
llvm-svn: 212911
Simon Atanasyan [Sun, 13 Jul 2014 16:18:56 +0000 (16:18 +0000)]
Add forgotten `break` statement.
llvm-svn: 212910
Yi Kong [Sun, 13 Jul 2014 16:17:30 +0000 (16:17 +0000)]
Add test cases for AArch64 hints codegen
llvm-svn: 212909
Simon Atanasyan [Sun, 13 Jul 2014 15:28:54 +0000 (15:28 +0000)]
[Mips] Support SHT_MIPS_ABIFLAGS section type flag in the llvm-readobj,
obj2yaml and yaml2obj tools.
llvm-svn: 212908
NAKAMURA Takumi [Sun, 13 Jul 2014 13:47:37 +0000 (13:47 +0000)]
[CMake] Enable loadable modules, aka plugins, with BUILD_SHARED_LIBS on cygming.
Loadable modules could be enabled without BUILD_SHARED_LIBS with tweaks in future.
llvm-svn: 212907
NAKAMURA Takumi [Sun, 13 Jul 2014 13:40:23 +0000 (13:40 +0000)]
[CMake] Give explicit dependencies to a couple of modules, PrintFunctionNames and SampleAnalyzerPlugin, for Win32.DLL.
llvm-svn: 212906
NAKAMURA Takumi [Sun, 13 Jul 2014 13:38:55 +0000 (13:38 +0000)]
PrintFunctionNames/CMakeLists.txt: Fix a comment.
llvm-svn: 212905
NAKAMURA Takumi [Sun, 13 Jul 2014 13:36:48 +0000 (13:36 +0000)]
[CMake] Add LLVM_LINK_COMPONENTS to loadable modules, LLVMHello and BugpointPasses, on Win32.
llvm-svn: 212904
NAKAMURA Takumi [Sun, 13 Jul 2014 13:33:26 +0000 (13:33 +0000)]
[CMake] Introduce moddir for MODULE -- corresponding to LIBRARY_OUTPUT_DIRECTORY.
On Win32.DLL, it points not lib but bin.
LIBRARY_OUTPUT_DIRECTORY affects add_library(MODULE), especially Win32.DLL.
llvm-svn: 212903
NAKAMURA Takumi [Sun, 13 Jul 2014 13:28:18 +0000 (13:28 +0000)]
bugpoint/ToolRunner.cpp: ProcessFailure(): Close ErrorFD immediately, or it couldn't be reopened on Win32.
FIXME: We may have an option in openFileForWrite(), not to use ResultFD but to close it.
llvm-svn: 212902
David Majnemer [Sun, 13 Jul 2014 05:19:56 +0000 (05:19 +0000)]
MS ABI: Stick internal vftables in a comdat if they have RTTI data
Previously, we would have a private backing variable and an internal
alias pointing at it.
However, -fdata-sections only fires if a global variable has non-private
linkage. This means that an unreferenced vftable wouldn't get
discarded, bloating the object file.
Instead, stick the backing variable in a comdat even if the alias has
internal linkage. This will allow the linker to drop the vftable if it
is unused.
llvm-svn: 212901
David Majnemer [Sun, 13 Jul 2014 04:56:11 +0000 (04:56 +0000)]
IR: Allow comdats to be applied to globals with internal linkage
Our verifier check for checking if a global has local linkage was too
strict. Forbid private linkage but permit local linkage.
Object file formats permit this and forbidding it prevents elimination
of unused, internal, vftables under the MSVC ABI.
llvm-svn: 212900
David Majnemer [Sun, 13 Jul 2014 04:31:19 +0000 (04:31 +0000)]
MC: Let non-temporary COFF aliases be in symtab
MC was aping a binutils bug where aliases would default their linkage to
private instead of internal.
I've sent a patch to the binutils maintainers and they've recently
applied it to the GNU assembler sources.
This fixes PR20152.
Differential Revision: http://reviews.llvm.org/D4395
llvm-svn: 212899
Matt Arsenault [Sun, 13 Jul 2014 03:08:59 +0000 (03:08 +0000)]
Remove unused include
llvm-svn: 212898
Matt Arsenault [Sun, 13 Jul 2014 03:06:43 +0000 (03:06 +0000)]
R600: Use range for and fix missing consts.
llvm-svn: 212897
Matt Arsenault [Sun, 13 Jul 2014 03:06:39 +0000 (03:06 +0000)]
R600: Make ShaderType private
llvm-svn: 212896
Matt Arsenault [Sun, 13 Jul 2014 02:46:17 +0000 (02:46 +0000)]
R600: Run more tests with promote alloca disabled.
Re-run tests changed in r211110 to test both paths.
Also fix broken check line.
llvm-svn: 212895
Matt Arsenault [Sun, 13 Jul 2014 02:18:06 +0000 (02:18 +0000)]
R600: Run private-memory test with and without alloca promote
The unpromoted path still needs to be tested since we can't
always promote to using LDS.
llvm-svn: 212894
Matt Arsenault [Sun, 13 Jul 2014 02:08:26 +0000 (02:08 +0000)]
R600: Add option to disable promote alloca
This can make writing some tests harder, so add a flag
to disable it.
llvm-svn: 212893
Saleem Abdulrasool [Sat, 12 Jul 2014 23:40:53 +0000 (23:40 +0000)]
tests: use a more precise target for tests
llvm-svn: 212892
Saleem Abdulrasool [Sat, 12 Jul 2014 23:27:26 +0000 (23:27 +0000)]
Headers: add hint intrinsics to arm_acle.h
This adds the ARM ACLE hint intrinsic wrappers to arm_acle.h. These need to be
protected with a !defined(_MSC_VER) since MSVC (and thus clang in compatibility
mode) provide these wrappers as proper builtin intrinsics.
llvm-svn: 212891
Saleem Abdulrasool [Sat, 12 Jul 2014 23:27:22 +0000 (23:27 +0000)]
CodeGen: support hint intrinsics from ACLE on AArch64
This adds support for the ACLE hint intrinsics on AArch64 similar to ARM. This
is required to properly support ACLE on AArch64.
llvm-svn: 212890
Matt Arsenault [Sat, 12 Jul 2014 23:16:26 +0000 (23:16 +0000)]
Try to fix MSVC warning.
llvm-svn: 212889
Matt Arsenault [Sat, 12 Jul 2014 23:09:02 +0000 (23:09 +0000)]
Try to fix MSVC build
llvm-svn: 212888
Yi Kong [Sat, 12 Jul 2014 22:48:13 +0000 (22:48 +0000)]
Improve comments of ARM ACLE header file and tests
Include section number in ARM ACLE specification for easier navigation.
llvm-svn: 212887
Matt Arsenault [Sat, 12 Jul 2014 22:19:49 +0000 (22:19 +0000)]
Try to fix MSVC build
llvm-svn: 212886
Matt Arsenault [Sat, 12 Jul 2014 21:59:52 +0000 (21:59 +0000)]
Templatify DominanceFrontier.
Theoretically this should now work for MachineBasicBlocks.
llvm-svn: 212885
Ed Maste [Sat, 12 Jul 2014 21:46:39 +0000 (21:46 +0000)]
Fix lldb-gdbserver build
s_listen_thread had the wrong type.
llvm-svn: 212884
Saleem Abdulrasool [Sat, 12 Jul 2014 21:20:49 +0000 (21:20 +0000)]
AArch64: add support for llvm.aarch64.hint intrinsic
This adds a llvm.aarch64.hint intrinsic to mirror the llvm.arm.hint in order to
support the various hint intrinsic functions in the ACLE.
Add an optional pattern field that permits the subclass to specify the pattern
that matches the selection. The intrinsic pattern is set as mayLoad, mayStore,
so overload the value for the definition of the hint instruction.
llvm-svn: 212883
Saleem Abdulrasool [Sat, 12 Jul 2014 20:49:13 +0000 (20:49 +0000)]
MC: remove use of unnecessary variable
Due to the fact that the windows unwinding has the concept of chained frames, we
maintain a current frame info pointer that is adjusted on any push and pop of a
unwinding context. This just removes an unnecessary variable that was used to
mirror the DWARF unwinding code.
llvm-svn: 212882
Saleem Abdulrasool [Sat, 12 Jul 2014 20:49:09 +0000 (20:49 +0000)]
MC: rename MCW64UnwindInfo to MCWinFrameInfo
This structure contains information related to the call frame used to generate
unwinding information. Rename this to reflect the future use to represent the
shared state between various architectures for WinCFI information.
llvm-svn: 212881
Simon Atanasyan [Sat, 12 Jul 2014 18:25:08 +0000 (18:25 +0000)]
[ELFYAML] Group ELF section type flags to target specific blocks.
Recognize only flags which correspond to the current target.
llvm-svn: 212880
Todd Fiala [Sat, 12 Jul 2014 17:34:24 +0000 (17:34 +0000)]
llgs: modify to accept --native-regs flag.
This is the last flag sent by lldb-platform that was not accepted
by llgs and is accepted by debugserver.
Conditionalized out a bit more code in prep for Windows
support one day based on _WIN32 define.
Updated the lldb-gdbserver usage string to represent
recent updates to command line arguments (and some older
ones like --attach pid).
Implements https://github.com/tfiala/lldb/issues/37
llvm-svn: 212879
Ed Maste [Sat, 12 Jul 2014 15:41:03 +0000 (15:41 +0000)]
Add FreeBSD failure decorator for TestCallStopAndContinue
The testrun now completes successfully on my FreeBSD 11.0-CURRENT
laptop. There are some intermittent failures on the FreeBSD buildbot
still, which should be addressed in later commits.
llvm.org/pr20274
llvm-svn: 212878
Ed Maste [Sat, 12 Jul 2014 15:21:55 +0000 (15:21 +0000)]
Add FreeBSD decorator for TestMultipleDebuggers
llvm.org/pr20282
llvm-svn: 212877
Alexander Kornienko [Sat, 12 Jul 2014 13:16:58 +0000 (13:16 +0000)]
Fixed compilation errors in tests, enforce checking for errors in check_clang_tidy_fix.sh.
Reviewers: sbenza, djasper
Reviewed By: djasper
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D4480
llvm-svn: 212876
Owen Anderson [Sat, 12 Jul 2014 07:12:47 +0000 (07:12 +0000)]
Fix an issue with the MergeBasicBlockIntoOnlyPred() helper function where it did
not properly handle the case where the predecessor block was the entry block to
the function. The only in-tree client of this is JumpThreading, which worked
around the issue in its own code. This patch moves the solution into the helper
so that JumpThreading (and other clients) do not have to replicate the same fix
everywhere.
llvm-svn: 212875
Hal Finkel [Sat, 12 Jul 2014 04:51:04 +0000 (04:51 +0000)]
Add nonnull in CodeGen for __attribute__((returns_nonnull))
As a follow-up to r212835, also add the LLVM nonnull function attribute when
__attribute__((returns_nonnull)) is provided.
llvm-svn: 212874
Todd Fiala [Sat, 12 Jul 2014 01:12:44 +0000 (01:12 +0000)]
llgs: implement --setsid.
The --setsid (-S) option changes the session id for the lldb-gdbserver process.
This is used by tools such as lldb-platform and allows the user to prevent
llgs from being in the same session as a calling terminal session.
This will prevents terminal group control signals from affecting
lldb-gdbserver.
See also:
https://github.com/tfiala/lldb/issues/38
llvm-svn: 212873
Alexey Samsonov [Sat, 12 Jul 2014 00:42:52 +0000 (00:42 +0000)]
[ASan] Collect unmangled names of global variables in Clang to print them in error reports.
Currently ASan instrumentation pass creates a string with global name
for each instrumented global (to include global names in the error report). Global
name is already mangled at this point, and we may not be able to demangle it
at runtime (e.g. there is no __cxa_demangle on Android).
Instead, create a string with fully qualified global name in Clang, and pass it
to ASan instrumentation pass in llvm.asan.globals metadata. If there is no metadata
for some global, ASan will use the original algorithm.
This fixes https://code.google.com/p/address-sanitizer/issues/detail?id=264.
llvm-svn: 212872
Reid Kleckner [Sat, 12 Jul 2014 00:40:59 +0000 (00:40 +0000)]
Require an x86 registered target for this test by splitting it out
This is a shot in the dark to fix the hexagon bot, so I'm not 100% sure
this is the issue.
llvm-svn: 212871
Matt Arsenault [Sat, 12 Jul 2014 00:36:19 +0000 (00:36 +0000)]
R600: Add missing tests for some intrinsics
llvm-svn: 212870
Reid Kleckner [Sat, 12 Jul 2014 00:29:05 +0000 (00:29 +0000)]
clang-cl: Make all x86 CPU feature flags available, such as -msse3
Ideally, we would use the /arch cl.exe flag for this stuff.
Unfortunately, MSVC supports only 5 /arch flag values, which isn't
nearly enough to cover all the CPU features that LLVM cares about. At
the very least, we need to know about SSE3 and SSE4.1 in addition to
SSE, SSE2, AVX, and AVX2.
In the future we should add the relevant /arch mappings in addition to
these gcc-style -m flags.
llvm-svn: 212869
Duncan P. N. Exon Smith [Sat, 12 Jul 2014 00:26:00 +0000 (00:26 +0000)]
BFI: Add constructor for Weight
llvm-svn: 212868
Greg Clayton [Sat, 12 Jul 2014 00:24:33 +0000 (00:24 +0000)]
lldb needs to support DW_op_piece masks for values in subregister and also to be able to piece together a value that is spread across multiple registers.
Patch from Adrian Prantl.
<rdar://problem/
16040521>
llvm-svn: 212867
Duncan P. N. Exon Smith [Sat, 12 Jul 2014 00:21:30 +0000 (00:21 +0000)]
BFI: Clean up BlockMass
Implementation is small now -- the interesting logic was moved to
`BranchProbability` a while ago. Move it into `bfi_detail` and get rid
of the related TODOs.
I was originally planning to define it within `BlockFrequencyInfoImpl`
(or `BFIIBase`), but it seems cleaner in a namespace. Besides,
`isPodLike` needs to be specialized before `BlockMass` can be used in
some of the other data structures, and there isn't a clear way to do
that.
llvm-svn: 212866
Reid Kleckner [Sat, 12 Jul 2014 00:18:58 +0000 (00:18 +0000)]
Option: Propagate flags from groups to options in each group
This should make it easy to set a flag for a whole group of clang driver
options.
llvm-svn: 212865
Lang Hames [Sat, 12 Jul 2014 00:16:47 +0000 (00:16 +0000)]
[RuntimeDyld] Fix stub size and offset for AArch64 in RuntimeDyldMachO.h.
<rdar://problem/
17648000>
llvm-svn: 212864
Greg Clayton [Sat, 12 Jul 2014 00:11:34 +0000 (00:11 +0000)]
Allow generic ARM cores to match any more specific ARM architecture.
<rdar://problem/
15932248>
llvm-svn: 212863
Reid Kleckner [Sat, 12 Jul 2014 00:06:46 +0000 (00:06 +0000)]
Avoid a warning from MSVC on "*/" in this code by inserting a space
llvm-svn: 212862
Duncan P. N. Exon Smith [Fri, 11 Jul 2014 23:56:50 +0000 (23:56 +0000)]
BFI: Mark the end of namespaces
llvm-svn: 212861
Reid Kleckner [Fri, 11 Jul 2014 23:54:29 +0000 (23:54 +0000)]
Form a CallExpr from __noop without parens
MSVC accepts __noop without any trailing parens and treats it like a
literal zero. We don't treat __noop as an integer literal, but now at
least we can parse a naked __noop expression.
Reviewers: rsmith
Differential Revision: http://reviews.llvm.org/D4476
llvm-svn: 212860
Lang Hames [Fri, 11 Jul 2014 23:52:07 +0000 (23:52 +0000)]
[RuntimeDyld] Add GOT support for AArch64 to RuntimeDyldMachO.
Test cases to follow once RuntimeDyldChecker supports introspection of stubs.
Fixes <rdar://problem/
17648000>
llvm-svn: 212859
Argyrios Kyrtzidis [Fri, 11 Jul 2014 23:47:48 +0000 (23:47 +0000)]
[Driver] clang::driver::getARMCPUForMArch() moved to llvm::Triple::getARMCPUForArch().
Depends on llvm r212846.
Suggested by Eric Christopher.
llvm-svn: 212858
Alexey Samsonov [Fri, 11 Jul 2014 23:34:26 +0000 (23:34 +0000)]
[ASan] Improve ODR-violation error reports.
Demangle names of involved globals. Print a more consistent summary line.
llvm-svn: 212857
Greg Clayton [Fri, 11 Jul 2014 23:15:11 +0000 (23:15 +0000)]
Remove assert now that we have a 'i' character that might come through as well as the 'q' character on the interrupt pipe.
<rdar://problem/
15840749>
llvm-svn: 212856
Juergen Ributzka [Fri, 11 Jul 2014 23:10:08 +0000 (23:10 +0000)]
Revert "[FastISel][X86] Implement the FastLowerIntrinsicCall hook."
This reverts commit r212851, because it broke the memset lowering.
llvm-svn: 212855
Todd Fiala [Fri, 11 Jul 2014 22:50:13 +0000 (22:50 +0000)]
Added llgs --named-pipe support and program_name-version_number printout support.
Added a unit test to test debugserver and llgs compliance on --named-pipe support.
Modified llgs to implement --named-pipe support. (Note: need to revisit with
new generic pipe support).
llvm-svn: 212854
Greg Clayton [Fri, 11 Jul 2014 22:43:15 +0000 (22:43 +0000)]
Don't crash when a SBType is handed out through the API and later used after the module that owns the type is deleted.
The fix adds a std::weak_ptr<Module> into the TypeImpl and fills in the weak pointer when possible. It also checks to make sure the module is still alive prior to using it which should make our API safer to use.
<rdar://problem/
15455145>
llvm-svn: 212853
Greg Clayton [Fri, 11 Jul 2014 22:41:30 +0000 (22:41 +0000)]
Don't use "lldb." global variables in LLDB commands.
llvm-svn: 212852
Juergen Ributzka [Fri, 11 Jul 2014 22:37:43 +0000 (22:37 +0000)]
[FastISel][X86] Implement the FastLowerIntrinsicCall hook.
Rename X86VisitIntrinsicCall -> FastLowerIntrinsicCall, which effectively
implements the target hook.
llvm-svn: 212851
Alexey Samsonov [Fri, 11 Jul 2014 22:36:02 +0000 (22:36 +0000)]
[ASan] Introduce a struct representing the layout of metadata entry in llvm.asan.globals.
No functionality change.
llvm-svn: 212850
Juergen Ributzka [Fri, 11 Jul 2014 22:19:02 +0000 (22:19 +0000)]
[FastISel] Add target-independent patchpoint intrinsic support. WIP.
This implements the target-independent lowering for the patchpoint
intrinsic. Targets have to implement the FastLowerCall
hook to support this intrinsic.
Related to <rdar://problem/
17427052>
llvm-svn: 212849
Juergen Ributzka [Fri, 11 Jul 2014 22:01:42 +0000 (22:01 +0000)]
[FastISel] Add basic infrastructure to support a target-independent call lowering hook in FastISel. WIP
The infrastructure mimics the call lowering we have already in place for
SelectionDAG, but with limitations. For example structure return demotion and
non-simple types are not supported (yet).
Currently every backend has its own implementation and duplicated code for call
lowering. There is also no specified interface that could be called from
target-independent code. The target-hook is opt-in and doesn't affect current
implementations.
llvm-svn: 212848
Aditya Nandakumar [Fri, 11 Jul 2014 21:49:39 +0000 (21:49 +0000)]
When we sink an instruction, this can open up opportunity for the operands to be sunk - add them to the worklist
llvm-svn: 212847
Argyrios Kyrtzidis [Fri, 11 Jul 2014 21:44:54 +0000 (21:44 +0000)]
Move the API and implementation of clang::driver::getARMCPUForMArch() to llvm::Triple::getARMCPUForArch().
Suggested by Eric Christopher.
llvm-svn: 212846
Alp Toker [Fri, 11 Jul 2014 20:53:51 +0000 (20:53 +0000)]
Consolidate header inclusion diagnostics
Make argument orders match, unify diagnostic IDs and reword the message to be a
little less saccharine.
llvm-svn: 212845
Juergen Ributzka [Fri, 11 Jul 2014 20:50:47 +0000 (20:50 +0000)]
[FastISel] Make isInTailCallPosition independent of SelectionDAG.
Break out the arguemnts required from SelectionDAG, so that this function can
also be used by FastISel.
llvm-svn: 212844
Juergen Ributzka [Fri, 11 Jul 2014 20:42:12 +0000 (20:42 +0000)]
[FastISel] Breakout intrinsic lowering into a separate function and add a target-hook.
Create a separate helper function for target-independent intrinsic lowering. Also
add an target-hook that allows to directly call into a target-sepcific intrinsic
lowering method. Currently the implementation is opt-in and doesn't affect
existing target implementations.
llvm-svn: 212843
Kevin Enderby [Fri, 11 Jul 2014 20:30:00 +0000 (20:30 +0000)]
Add the "-s" flag to llvm-nm for Mach-O files that prints symbols only in
the specified section. This is same functionality as darwin’s nm(1) "-s" flag.
There is one FIXME in the code and I’m all ears to anyone that can help me
with that. This option takes exactly two strings and should be allowed
anywhere on the command line. Such that "llvm-nm -s __TEXT __text foo.o"
would work. But that does not as the CommandLine Library does not have a
way to make this work as far as I can tell. For now the "-s __TEXT __text"
has to be last on the command line.
llvm-svn: 212842
David Majnemer [Fri, 11 Jul 2014 20:28:10 +0000 (20:28 +0000)]
CodeGen: Don't emit a thread-wrapper if we can't touch the backing variable
OS X TLS has all accesses going through the thread-wrapper function and
gives the backing thread-local variable internal linkage. This means
that thread-wrappers must have WeakAnyLinkage so that references to the
internal thread-local variables do not get propagated to other code.
It also means that translation units which do not provide a definition
for the thread-local variable cannot attempt to emit a thread-wrapper
because the thread wrapper will attempt to reference the backing
variable.
Differential Revision: http://reviews.llvm.org/D4109
llvm-svn: 212841
Brad Smith [Fri, 11 Jul 2014 20:27:39 +0000 (20:27 +0000)]
oops. correct the prefix.
llvm-svn: 212840
Reid Kleckner [Fri, 11 Jul 2014 20:22:55 +0000 (20:22 +0000)]
MS extension: Make __noop be the integer zero, not void
We still don't accept '__noop;', and we don't consider __noop to be the
integer literal zero. More work is needed.
llvm-svn: 212839
Brad Smith [Fri, 11 Jul 2014 20:12:08 +0000 (20:12 +0000)]
Handle SPARC float command line parameters for SPARCv9.
llvm-svn: 212838
Alp Toker [Fri, 11 Jul 2014 18:23:08 +0000 (18:23 +0000)]
Simplify the raw_svector_ostream tweak from r212816
The memcpy() and overlap helps didn't help much with timings, so clean up the change.
The difference at this point is that we now leave growth of the storage buffer
up to SmallVector's implementation:
- OS.reserve(OS.capacity() * 2);
+ OS.reserve(OS.size() + 64);
llvm-svn: 212837
Richard Smith [Fri, 11 Jul 2014 18:22:58 +0000 (18:22 +0000)]
[modules] When merging a class template, also merge the definition of its pattern.
llvm-svn: 212836
Hal Finkel [Fri, 11 Jul 2014 17:35:21 +0000 (17:35 +0000)]
Add nonnull in CodeGen for __attribute__((nonnull))
We now have an LLVM-level nonnull attribute that can be applied to function
parameters, and we emit it for reference types (as of r209723), but did not
emit it when an __attribute__((nonnull)) was provided. Now we will.
llvm-svn: 212835
Ulrich Weigand [Fri, 11 Jul 2014 17:34:44 +0000 (17:34 +0000)]
[MC] Constify MCELF::GetVisibility and MCELF::getOther
These two routines didn't take a "const MCSymbolData &SD"
like the other MCELF::Get routines for some reason ...
llvm-svn: 212834
Sylvestre Ledru [Fri, 11 Jul 2014 17:23:13 +0000 (17:23 +0000)]
Test for warning being silenced
Summary:
Added some tests to see if the new warning would be silenced with a flag.
Patch by Arthur Marble <arthur@info9.net> in the context of Debian Google Summer of code 2014.
Reviewers: sylvestre.ledru
Reviewed By: sylvestre.ledru
Differential Revision: http://reviews.llvm.org/D4475
llvm-svn: 212833
Ulrich Weigand [Fri, 11 Jul 2014 17:19:31 +0000 (17:19 +0000)]
[PowerPC] Fix invalid displacement created by LocalStackAlloc
This commit fixes a bug in PPCRegisterInfo::isFrameOffsetLegal that
could result in the LocalStackAlloc pass creating an MI instruction
out-of-range displacement:
%vreg17<def> = LD 33184, %vreg31; mem:LD8[%g](align=32)
%G8RC:%vreg17 G8RC_and_G8RC_NOX0:%vreg31
(In final assembler output the top bits are stripped off, resulting
in a negative offset loading from below the stack pointer.)
Common code expects the isFrameOffsetLegal routine to verify whether
adding a given offset to the offset already present in the instruction
results in a valid displacement. However, on PowerPC the routine
did not take the already present instruction offset into account.
This commit fixes isFrameOffsetLegal to add the instruction offset,
and updates a local caller (needsFrameBaseReg) to no longer add the
instruction offset itself before calling isFrameOffsetLegal.
Reviewed by Hal Finkel.
llvm-svn: 212832
Marek Olsak [Fri, 11 Jul 2014 17:11:52 +0000 (17:11 +0000)]
R600/SI: Use i32 vectors for resources and samplers
This affects new intrinsics only.
What surprises me is that v32i8 still works.
llvm-svn: 212831
Marek Olsak [Fri, 11 Jul 2014 17:11:46 +0000 (17:11 +0000)]
R600/SI: add sample and image intrinsics exposing all instruction fields
We need the intrinsics with offsets, so why not just add them all.
The R128 parameter will also be useful for reducing SGPR usage.
GL_ARB_image_load_store also adds some image GLSL modifiers like "coherent",
so Mesa will probably translate those to slc, glc, etc.
When LLVM 3.5 is released, I'll switch Mesa to these new intrinsics.
llvm-svn: 212830
Marek Olsak [Fri, 11 Jul 2014 17:11:39 +0000 (17:11 +0000)]
R600/SI: fix shadow mapping for 1D and 2D array textures
It was conflicting with def TEX_SHADOW_ARRAY, which also handles them.
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 212829