Michael J. Spencer [Tue, 18 Nov 2014 01:26:46 +0000 (01:26 +0000)]
Fix covered switch warning
llvm-svn: 222209
Michael J. Spencer [Tue, 18 Nov 2014 01:14:25 +0000 (01:14 +0000)]
Support ELF files of unknown type.
llvm-svn: 222208
Greg Clayton [Tue, 18 Nov 2014 00:43:17 +0000 (00:43 +0000)]
Have CommandObjectCommandsAddRegex inherit from IOHandlerDelegateMultiline so it will not immediately terminate after the first regular expression in "command regex <name>" commands.
Fixed the prompt to not include non-printable characters as it was hosing up the prompt when you ran "command regex foo" and entered multi-line editing mode.
Fixed error strings to include more complete descriptions when bad regular expressions are entered.
Removed the old IOHandlerLinesUpdated function as it is no longer needed (inheriting from IOHandlerDelegateMultiline takes care of what this function used to do).
llvm-svn: 222207
Greg Clayton [Tue, 18 Nov 2014 00:39:31 +0000 (00:39 +0000)]
Fixed a broken test suite test after recent editline merges.
The problem is that editline currently is trying to be smart when we paste things into a terminal window. It detects that input is pending and multi-line input is accepted as long as there is anything pending.
So if you open lldb and paste the text between the quotes:
"command regex carp
s/^$/help/
carp
"
We still still be stuck in the "command regex" multi-line input reader as it will have all three lines:
"s/^$/help/
carp
"
The true fix for this is something Kate Stone will soon work on:
- multi-line input readers must opt into this paste/pending input feature ("expr" will opt into it, all other commands won't)
- If we are in a multi-line input reader that requests this and stuff is pasted, then it will do what it does today
- if we start in a IOHandler that doesn't need/want pending input and text is pasted, and we transistion to a IOHandler that does want this functionality, then disable the pending input. Example text would be:
"frame variable
thread backtrace
expr
for (int i=0;i<10;++i)
(int)printf("i = %i\n", i);
frame select 0
"
When we push the expression multi-line reader we would disable the pending input because we had pending input _before_ we entered "expr".
If we did this first:
(lldb) expr
Then we pasted:
"void foo()
{
}
void bar()
{
}
"
Then we would allow the pending input to not look for an empty line to terminate the expression. We filed radar
19008425 to track fixing this issue.
llvm-svn: 222206
Duncan P. N. Exon Smith [Tue, 18 Nov 2014 00:37:17 +0000 (00:37 +0000)]
IR: Split MDNode into GenericMDNode and MDNodeFwdDecl
Split `MDNode` into two classes:
- `GenericMDNode`, which is uniquable (and for now, always starts
uniqued). Once `Metadata` is split from the `Value` hierarchy, this
class will lose the ability to RAUW itself.
- `MDNodeFwdDecl`, which is used for the "temporary" interface, is
never uniqued, and isn't managed by `LLVMContext` at all.
I've left most of the guts in `MDNode` for now, but I'll incrementally
move things to the right places (or delete the functionality, as
appropriate).
Part of PR21532.
llvm-svn: 222205
Justin Bogner [Tue, 18 Nov 2014 00:34:46 +0000 (00:34 +0000)]
InstrProf: Don't emit coverage for uninstantiated templates
We include unused functions and methods in -fcoverage-mapping so that
we can differentiate between uninstrumented and unused. This can cause
problems for uninstantiated templates though, since they may involve
an incomplete type that can't be mangled. This shows up in things like
libc++'s <unordered_map> and makes coverage unusable.
Avoid the issue by skipping uninstantiated methods of a templated
class.
llvm-svn: 222204
Manman Ren [Tue, 18 Nov 2014 00:33:22 +0000 (00:33 +0000)]
Revert r222040 because of bot failure.
http://lab.llvm.org:8080/green/job/clang-Rlto_master/298/
Hopefully, bot will be green.
llvm-svn: 222203
Nick Kledzik [Tue, 18 Nov 2014 00:30:29 +0000 (00:30 +0000)]
[mach-o] zero-fill sections have no file offset
In mach-o, sections of type S_ZEROFILL are special cased and to always have
their section.offset field be zero.
llvm-svn: 222202
Nick Kledzik [Tue, 18 Nov 2014 00:30:25 +0000 (00:30 +0000)]
[mach-o] extract atom alignment information from mach-o files
llvm-svn: 222201
Nick Kledzik [Tue, 18 Nov 2014 00:30:22 +0000 (00:30 +0000)]
[mach-o] enhance arm64 reloc parser to support local pointer relocations
The arm64 assembler almost always uses r_extern=1 relocations in which the
r_symbolnum field is the index of the symbol the relocation references. But
sometimes it will set r_extern=0 in which case the linker needs to read the
content of the reloction to determine the target.
Add test case that the r_extern=0 relocation round trips.
llvm-svn: 222200
Nick Kledzik [Tue, 18 Nov 2014 00:30:20 +0000 (00:30 +0000)]
[mach-o] fix typos. No function change
llvm-svn: 222199
Nick Kledzik [Tue, 18 Nov 2014 00:30:17 +0000 (00:30 +0000)]
[mach-o] enhance arm64 reloc parser to support local pointer relocations
The arm64 assembler almost always uses r_extern=1 relocations in which the
r_symbolnum field is the index of the symbol the relocation references. But
sometimes it will set r_extern=0 in which case the linker needs to read the
content of the reloction to determine the target.
Add test case that the r_extern=0 relocation round trips.
llvm-svn: 222198
Nick Kledzik [Tue, 18 Nov 2014 00:30:14 +0000 (00:30 +0000)]
[mach-o] fix typos. No function change
llvm-svn: 222197
Manman Ren [Tue, 18 Nov 2014 00:29:32 +0000 (00:29 +0000)]
Debug Info: context field of a global variable is DIScopeRef.
Paired commit with r222195.
llvm-svn: 222196
Manman Ren [Tue, 18 Nov 2014 00:29:08 +0000 (00:29 +0000)]
Debug Info: In DIBuilder, the context field of a global variable is updated to
use DIScopeRef.
A paired commit at clang will follow to show cases where we will use an
identifer for the context of a global variable.
rdar://
18958417
llvm-svn: 222195
John McCall [Tue, 18 Nov 2014 00:19:01 +0000 (00:19 +0000)]
Fix an assertion when ending a function definition.
The bug is that ExprCleanupObjects isn't always empty
in a fresh evaluation context. New evaluation contexts just
track the current depth of the stack.
The assertion will misfire whenever we finish processing
a function body inside an expression that contained an earlier
block literal with non-trivial captures. That's actually
a lot less likely than you'd think, though, because it has
to be a real function declaration, not just another block.
Mixed block/lambda code would work, as would a template
instantiation or a local class definition.
The code works correctly if the assertion is disabled.
rdar://
16356628
llvm-svn: 222194
Alexander Kornienko [Mon, 17 Nov 2014 23:46:02 +0000 (23:46 +0000)]
Make DiagnosticsEngine::takeClient return std::unique_ptr<>
Summary:
Make DiagnosticsEngine::takeClient return std::unique_ptr<>. Updated
callers to store conditional ownership using a pair of pointer and unique_ptr
instead of a pointer + bool. Updated code that temporarily registers clients to
use the non-owning registration (+ removed extra calls to takeClient).
Reviewers: dblaikie
Reviewed By: dblaikie
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D6294
llvm-svn: 222193
Reid Kleckner [Mon, 17 Nov 2014 23:36:45 +0000 (23:36 +0000)]
Handle use of default member initializers before end of outermost class
Specifically, when we have this situation:
struct A {
template <typename T> struct B {
int m1 = sizeof(A);
};
B<int> m2;
};
We can't parse m1's initializer eagerly because we need A to be
complete. Therefore we wait until the end of A's class scope to parse
it. However, we can trigger instantiation of B before the end of A,
which will attempt to instantiate the field decls eagerly, and it would
build a bad field decl instantiation that said it had an initializer but
actually lacked one.
Fixed by deferring instantiation of default member initializers until
they are needed during constructor analysis. This addresses a long
standing FIXME in the code.
Fixes PR19195.
Reviewed By: rsmith
Differential Revision: http://reviews.llvm.org/D5690
llvm-svn: 222192
Duncan P. N. Exon Smith [Mon, 17 Nov 2014 23:28:21 +0000 (23:28 +0000)]
IR: Simplify uniquing for MDNode
Change uniquing from a `FoldingSet` to a `DenseSet` with custom
`DenseMapInfo`. Unfortunately, this doesn't save any memory, since
`DenseSet<T>` is a simple wrapper for `DenseMap<T, char>`, but I'll come
back to fix that later.
I used the name `GenericDenseMapInfo` to the custom `DenseMapInfo` since
I'll be splitting `MDNode` into two classes soon: `MDNodeFwdDecl` for
temporaries, and `GenericMDNode` for everything else.
I also added a non-debug-info reduced version of a type-uniquing test
that started failing on an earlier draft of this patch.
Part of PR21532.
llvm-svn: 222191
Enrico Granata [Mon, 17 Nov 2014 23:14:11 +0000 (23:14 +0000)]
Fix a problem where the StringPrinter could be mistaking an empty string for a read error, and reporting spurious 'unable to read data' messages. rdar://
19007243
llvm-svn: 222190
Enrico Granata [Mon, 17 Nov 2014 23:06:20 +0000 (23:06 +0000)]
Add APIs on SBFunction and SBCompileUnit to inquire about the language type that the function/compile unit is defined in
llvm-svn: 222189
Reid Kleckner [Mon, 17 Nov 2014 22:55:59 +0000 (22:55 +0000)]
Revert "ADT: correctly report isMSVCEnvironment for windows itanium"
This reverts commit r222180.
llvm-svn: 222188
David Blaikie [Mon, 17 Nov 2014 22:55:41 +0000 (22:55 +0000)]
Revert "Improve memory ownership/management in TableGen by unique_ptrifying TreePattern's Tree member."
This reverts commit r222183.
Broke on the MSVC buildbots due to MSVC not producing default move
operations - I'd fix it immediately but just broke my build system a
bit, so backing out until I have a chance to get everything going again.
llvm-svn: 222187
Eric Christopher [Mon, 17 Nov 2014 22:55:13 +0000 (22:55 +0000)]
Fix override/virtual warnings.
llvm-svn: 222186
Zachary Turner [Mon, 17 Nov 2014 22:42:57 +0000 (22:42 +0000)]
Change HostThread::GetNativeThread() to return a derived reference.
Previously using HostThread::GetNativeThread() required an ugly
cast to most-derived type. This solves the issue by simply returning
the derived type directly.
llvm-svn: 222185
Oleksiy Vyalov [Mon, 17 Nov 2014 22:42:28 +0000 (22:42 +0000)]
Fix broken NativeProcessLinux.cpp after signature change of ResolveExecutable.
llvm-svn: 222184
David Blaikie [Mon, 17 Nov 2014 22:16:55 +0000 (22:16 +0000)]
Improve memory ownership/management in TableGen by unique_ptrifying TreePattern's Tree member.
The next step is to actually use unique_ptr in TreePatternNode's
Children vector. That will be more intrusive, and may not work,
depending on exactly how these things are handled (I have a bad
suspicion things are shared more than they should be, making this more
DAG than tree - but if it's really a tree, unique_ptr should suffice)
llvm-svn: 222183
Oleksiy Vyalov [Mon, 17 Nov 2014 22:16:42 +0000 (22:16 +0000)]
Fix broken Linux build after signature change of ResolveExecutable.
llvm-svn: 222182
Peter Collingbourne [Mon, 17 Nov 2014 22:16:15 +0000 (22:16 +0000)]
CMake: Use the new USES_TERMINAL option for test suite targets when available.
llvm-svn: 222181
Saleem Abdulrasool [Mon, 17 Nov 2014 22:13:26 +0000 (22:13 +0000)]
ADT: correctly report isMSVCEnvironment for windows itanium
The itanium environment on Windows uses MSVC and is a MSVC environment. Report
this correctly.
llvm-svn: 222180
Saleem Abdulrasool [Mon, 17 Nov 2014 22:11:07 +0000 (22:11 +0000)]
CodeGen: make a check stricter
When targeting Windows itanium (a MSVC environment), use itanium style
exceptions rather than SEH. Existing test cases already test this code path.
Applying this change ensures that tests wont break due to a parallel change in
LLVM (to correctly report isMSVCEnvironment).
llvm-svn: 222179
Zachary Turner [Mon, 17 Nov 2014 21:31:30 +0000 (21:31 +0000)]
Move the thread logic around to fit better into LLDB's process model.
Previously we were directly updating the thread list and stopping
and restarting the process every time threads were created. With
this patch, we queue up thread launches and thread exits, resolve
these all internally, and only update the threads when we get an
UpdateThreadList call. We now only update the private state on
an actual stop (i.e. breakpoint).
llvm-svn: 222178
Zachary Turner [Mon, 17 Nov 2014 21:31:18 +0000 (21:31 +0000)]
Disable Editline on Windows.
Editline does not work correctly on Windows. This goes back at
least to r208369, and as a result r210105 was submitted to disable
libedit at runtime on Windows.
More recently, r222163 was submitted which re-writes editline
entirely, but makes the situation even worse on Windows, to the
point that it doesn't even compile. While it would be easy to
fix the compilation failure, this patch simply stops compiling
Editline entirely on Windows, as the simple compilation fix would
still result in a broken use of select on Windows, and as such a
broken implementation of Editline.
Since Editline was already disabled to begin with on Windows, we
don't attempt to fix the compilation failure or the underlying
issues, and instead just disable it "even more".
llvm-svn: 222177
Zachary Turner [Mon, 17 Nov 2014 21:30:58 +0000 (21:30 +0000)]
Fix broken build after signature change of ResolveExecutable.
llvm-svn: 222176
Matt Arsenault [Mon, 17 Nov 2014 21:11:37 +0000 (21:11 +0000)]
R600/SI: Don't copy flags when extracting subreg
This was resulting in use of a register after a kill.
For some reason this showed up as a problem in many tests
when moving the SIFixSGPRCopies pass closer to instruction
selection.
llvm-svn: 222175
Matt Arsenault [Mon, 17 Nov 2014 21:11:34 +0000 (21:11 +0000)]
R600/SI: Assume SIFixSGPRCopies makes changes
I'm not sure if this was breaking anything.
llvm-svn: 222174
Rafael Espindola [Mon, 17 Nov 2014 21:06:38 +0000 (21:06 +0000)]
Fix the autoconf build.
llvm-svn: 222173
Rafael Espindola [Mon, 17 Nov 2014 20:51:01 +0000 (20:51 +0000)]
Factor common code it Linker::init.
The TypeFinder was not being used in one of the constructors.
llvm-svn: 222172
Jason Molenda [Mon, 17 Nov 2014 20:10:15 +0000 (20:10 +0000)]
Small tweaks to make the editline sources match the lldb
source layout.
llvm-svn: 222171
Rafael Espindola [Mon, 17 Nov 2014 20:06:27 +0000 (20:06 +0000)]
Pass a reference to ValueEnumerator.
NFC. This will just make it easier to use std::unique_ptr in a caller.
llvm-svn: 222170
Sylvestre Ledru [Mon, 17 Nov 2014 19:41:49 +0000 (19:41 +0000)]
Missing comma in a string array initialization (CID 1254893)
Summary:
We have this error from a while (Wed Jun 15 18:02:42 2011
r133103)
Reviewers: rsmith
Reviewed By: rsmith
Differential Revision: http://reviews.llvm.org/D6296
llvm-svn: 222169
Juergen Ributzka [Mon, 17 Nov 2014 19:39:56 +0000 (19:39 +0000)]
[SimplifyCFG] Make the value type of the hole check bitmask a power-of-2.
When converting a switch to a lookup table we might have to generate a bitmaks
to encode and check for holes in the original switch statement.
The type of this mask depends on the number of switch statements, which can
result in illegal types for pretty much all architectures.
To avoid unnecessary type legalization and help FastISel this commit increases
the size of the bitmask to next power-of-2 value when necessary.
This fixes rdar://problem/
18984639.
llvm-svn: 222168
Greg Clayton [Mon, 17 Nov 2014 19:39:20 +0000 (19:39 +0000)]
Fixed more fallout from running the test suite remotely on iOS devices.
Fixed include:
- Change Platform::ResolveExecutable(...) to take a ModuleSpec instead of a FileSpec + ArchSpec to help resolve executables correctly when we have just a path + UUID (no arch).
- Add the ability to set the listener in SBLaunchInfo and SBAttachInfo in case you don't want to use the debugger as the default listener.
- Modified all places that use the SBLaunchInfo/SBAttachInfo and the internal ProcessLaunchInfo/ProcessAttachInfo to not take a listener as a parameter since it is in the launch/attach info now
- Load a module's sections by default when removing a module from a target. Since we create JIT modules for expressions and helper functions, we could end up with stale data in the section load list if a module was removed from the target as the section load list would still have entries for the unloaded module. Target now has the following functions to help unload all sections a single or multiple modules:
size_t
Target::UnloadModuleSections (const ModuleList &module_list);
size_t
Target::UnloadModuleSections (const lldb::ModuleSP &module_sp);
llvm-svn: 222167
Rafael Espindola [Mon, 17 Nov 2014 19:26:40 +0000 (19:26 +0000)]
Revert commits r222146 and r222137.
They were producing the wrong result if NumBits == BitsInWord. The old mask
produced -1, the new mask 0.
This should fix the 32 bit bots.
llvm-svn: 222166
Marshall Clow [Mon, 17 Nov 2014 19:16:57 +0000 (19:16 +0000)]
Fix the tests I broke with the last commit. Sorry for the noise
llvm-svn: 222165
Hans Wennborg [Mon, 17 Nov 2014 19:16:36 +0000 (19:16 +0000)]
clang-cl: Allow /Fo without an argument (PR21589)
When it's used without an argument, the default file name is
used. The same goes for /Fe.
Also, allow using /Fo, /Fa and /Fe with multiple inputs if they
don't have an argument.
llvm-svn: 222164
Kate Stone [Mon, 17 Nov 2014 19:06:59 +0000 (19:06 +0000)]
Complete rewrite of interactive editing support for single- and multi-line input.
Improvements include:
* Use of libedit's wide character support, which is imperfect but a distinct improvement over ASCII-only
* Fallback for ASCII editing path
* Support for a "faint" prompt clearly distinguished from input
* Breaking lines and insert new lines in the middle of a batch by simply pressing return
* Joining lines with forward and backward character deletion
* Detection of paste to suppress automatic formatting and statement completion tests
* Correctly reformatting when lines grow or shrink to occupy different numbers of rows
* Saving multi-line history, and correctly preserving the "tip" of history during editing
* Displaying visible ^C and ^D indications when interrupting input or sending EOF
* Fledgling VI support for multi-line editing
* General correctness and reliability improvements
llvm-svn: 222163
Marshall Clow [Mon, 17 Nov 2014 19:05:50 +0000 (19:05 +0000)]
Implement LWG2400 - 'shared_ptr's get_deleter() should use addressof()', and add tests. Mark LWG2400 and LWG2404 as complete
llvm-svn: 222161
Enrico Granata [Mon, 17 Nov 2014 19:00:20 +0000 (19:00 +0000)]
Not all things callable have an im_self, so harden the test logic against that. getattr(,,None) is the closest to ?. we have in Python, so use that
llvm-svn: 222160
Marshall Clow [Mon, 17 Nov 2014 18:52:25 +0000 (18:52 +0000)]
Reworked mismatch tests to count the number of comparisons, and make sure we are conforming with LWG2404. We are
llvm-svn: 222159
Daniel Jasper [Mon, 17 Nov 2014 18:42:22 +0000 (18:42 +0000)]
clang-format: Fix more incorrect pointer detection.
Before:
Constructor() : a(a), b(c, d *e) {}
After:
Constructor() : a(a), b(c, d * e) {}
llvm-svn: 222158
Hans Wennborg [Mon, 17 Nov 2014 18:40:45 +0000 (18:40 +0000)]
test/Driver/cl-outputs.c: tweak some check prefixes; NFC.
These tests deal with the /o option; rename to better reflect that.
llvm-svn: 222157
Greg Clayton [Mon, 17 Nov 2014 18:40:27 +0000 (18:40 +0000)]
Fixes for remote test suite execution of the "lldb/test/lang" directory.
Fixes include:
- Add a new lldbtest.TestBase function named registerSharedLibrariesWithTarget. This function can be called using the shared libraries for your test suite either as shared library basename ("foo"), path basename ("libfoo.dylib") or full path ("/tmp/lldb/test/lang/c/carp/libfoo.dylib"). These shared libraries are then registered with the target so they will be downloaded when the test is run remotely.
- Changed a lot of tests over to use SBDebugger::CreateTarget(...) calls instead of using "file a.out" commands.
- Cleaned up some tests to add new locations for breakpoints that all compilers should be able to abide by. Some tests and constants being loaded into values of structs and some compilers like arm64 will often combine two constant data loads into a single source line so some breakpoint locations were not being set correctly. Adding lines like 'puts("")' allow us to lock onto a source line that will have code.
llvm-svn: 222156
Saleem Abdulrasool [Mon, 17 Nov 2014 18:40:15 +0000 (18:40 +0000)]
Basic: tweak comment
Add a missing surrounding brace for doxygen group. This messes with pair
jumping in vim and is annoying. NFC.
llvm-svn: 222155
Zachary Turner [Mon, 17 Nov 2014 18:38:22 +0000 (18:38 +0000)]
Fix buildSwigWrapperClasses.py after recent break.
A re-ordering of some enum values exposed a lingering bug where an
invalid key was indexing a dictionary.
llvm-svn: 222154
Greg Clayton [Mon, 17 Nov 2014 18:32:17 +0000 (18:32 +0000)]
Fix some issue with running the test suite remotely on iOS.
- Added a new "--apple-sdk" flag that can be specified on Darwin only so the correct cross compilers can be auto-selected without having to specify the "--compiler" flag.
- Set SDKROOT if needed
llvm-svn: 222153
Sylvestre Ledru [Mon, 17 Nov 2014 18:26:39 +0000 (18:26 +0000)]
Fix the indentation: Nesting level does not match indentation (CID 1254863)
llvm-svn: 222152
Aaron Ballman [Mon, 17 Nov 2014 18:17:19 +0000 (18:17 +0000)]
Fixing a use of stringstream to use an LLVM helper function. Drive-by fixing header include order. NFC.
llvm-svn: 222151
Zachary Turner [Mon, 17 Nov 2014 18:06:21 +0000 (18:06 +0000)]
Reformat lldb-mi using clang-format.
Courtesy of dawn@burble.org.
llvm-svn: 222150
Rafael Espindola [Mon, 17 Nov 2014 17:51:45 +0000 (17:51 +0000)]
Fix GraphTraits for "const CallGraphNode *" and "const CallGraph *"
The specializations were broken. For example,
void foo(const CallGraph *G) {
auto I = GraphTraits<const CallGraph *>::nodes_begin(G);
auto K = I++;
...
}
or
void bar(const CallGraphNode *N) {
auto I = GraphTraits<const CallGraphNode *>::nodes_begin(G);
auto K = I++;
....
}
would not compile.
Patch by Speziale Ettore!
llvm-svn: 222149
Zachary Turner [Mon, 17 Nov 2014 17:46:43 +0000 (17:46 +0000)]
[ProcessWindows] Create a TargetThreadWindows class.
This creates a TargetThreadWindows class and updates the thread
list of the Process with the main thread. Additionally, we
fill out a few more overrides of Process base class methods. We
do not yet update the thread list as threads are created and/or
destroyed, and we do not yet propagate stop reasons to threads as
their states change.
llvm-svn: 222148
Zachary Turner [Mon, 17 Nov 2014 17:46:27 +0000 (17:46 +0000)]
[ProcessWindows] Implement read / write process memory.
llvm-svn: 222147
Rafael Espindola [Mon, 17 Nov 2014 17:43:27 +0000 (17:43 +0000)]
Avoid undefined behavior by masking the shift amount.
Should hopefully fix the mips bots.
llvm-svn: 222146
Alexander Kornienko [Mon, 17 Nov 2014 17:32:32 +0000 (17:32 +0000)]
[clang-tidy] Move the missing namespace comment warnings to the closing brace
Summary:
The google-readability-namespace-comments/llvm-namespace-comment
warnings are quite confusing when they appear at the beginning of a long
namespace and the closing brace is not in sight.
For convenience added notes pointing to the start of the namespace.
Reviewers: klimek
Reviewed By: klimek
Subscribers: curdeius, cfe-commits
Differential Revision: http://reviews.llvm.org/D6251
llvm-svn: 222145
Chad Rosier [Mon, 17 Nov 2014 16:34:47 +0000 (16:34 +0000)]
[Reassociate] Update test cases due to r222142.
llvm-svn: 222144
Marshall Clow [Mon, 17 Nov 2014 16:34:44 +0000 (16:34 +0000)]
Fix a warning in the test; no functionality change
llvm-svn: 222143
Chad Rosier [Mon, 17 Nov 2014 16:33:50 +0000 (16:33 +0000)]
[Reassociate] As the expression tree is rewritten make sure the operands are
emitted in canonical form.
llvm-svn: 222142
Alexey Volkov [Mon, 17 Nov 2014 16:17:51 +0000 (16:17 +0000)]
[X86] Use ADD/SUB instead of INC/DEC for Haswell and Broadwell CPUs
Differential Revision: http://reviews.llvm.org/D5934
llvm-svn: 222141
Timur Iskhodzhanov [Mon, 17 Nov 2014 15:53:50 +0000 (15:53 +0000)]
Minor readability tweaks to VFTableBuilder.cpp
llvm-svn: 222140
Chad Rosier [Mon, 17 Nov 2014 15:52:51 +0000 (15:52 +0000)]
[Reassociate] Canonicalize constants to RHS operand.
Fix a thinko where the RHS was already a constant.
llvm-svn: 222139
Marshall Clow [Mon, 17 Nov 2014 15:50:08 +0000 (15:50 +0000)]
Implement void_t from N3911. Add a private version for use in the library before C++1z. Update the 1z status page, marking a bunch of issues that don't require library changes as complete (2129, 2212, 2230, 2233, 2325, 2365, 2376)
llvm-svn: 222138
Rafael Espindola [Mon, 17 Nov 2014 15:46:21 +0000 (15:46 +0000)]
Use a more canonical way of computing a mask with N trailing 1s. NFC.
llvm-svn: 222137
Ed Maste [Mon, 17 Nov 2014 15:40:18 +0000 (15:40 +0000)]
Fix Darwin and FreeBSD OS type detection
Obtained in part from http://reviews.llvm.org/D6290
llvm-svn: 222136
Ed Maste [Mon, 17 Nov 2014 15:37:59 +0000 (15:37 +0000)]
Whitespace cleanup and remove non-canonical headers
llvm-svn: 222135
Ed Maste [Mon, 17 Nov 2014 15:27:09 +0000 (15:27 +0000)]
Add decorator for intermittently failing test on FreeBSD
This test has intermittently failed on FreeBSD for quite some time when
run as part of the full test suite. It generally passes when run by
itself. Mark as expected failure for now to reduce buildbot noise.
llvm.org/pr15039 test fails intermittently on FreeBSD
llvm-svn: 222134
Timur Iskhodzhanov [Mon, 17 Nov 2014 15:11:05 +0000 (15:11 +0000)]
Add a couple more examples illustrating why we need vtordisps and how they work
llvm-svn: 222133
Marshall Clow [Mon, 17 Nov 2014 15:04:46 +0000 (15:04 +0000)]
Add tests to ensure that reference_wrapper<T> is trivially copyable. This was added to C++1z with the adoption of N4277, but libc++ already implemented it as a conforming extension. No code changes were needed, just more tests.
llvm-svn: 222132
Samuel Benzaquen [Mon, 17 Nov 2014 14:55:49 +0000 (14:55 +0000)]
Remove VariadicOperatorMatcherInterface as it is redundant with logic from DynTypedMatcher.
Summary:
The generic variadic matcher is faster (one less virtual function call
per match) and doesn't require template instantiations which reduces
compile time and binary size.
Registry.cpp.o generates ~14% less symbols and compiles ~7.5% faster.
The change also speeds up our clang-tidy benchmark by ~2%.
Reviewers: klimek
Subscribers: klimek, cfe-commits
Differential Revision: http://reviews.llvm.org/D6278
llvm-svn: 222131
Alexander Kornienko [Mon, 17 Nov 2014 14:46:28 +0000 (14:46 +0000)]
A small correction for r221884. No functional changes.
llvm-svn: 222130
Renato Golin [Mon, 17 Nov 2014 14:08:57 +0000 (14:08 +0000)]
Fix ARM triple parsing
The triple parser should only accept existing architecture names
when the triple starts with armv, armebv, thumbv or thumbebv.
Patch by Gabor Ballabas.
llvm-svn: 222129
Daniel Jasper [Mon, 17 Nov 2014 13:55:04 +0000 (13:55 +0000)]
clang-format: Fix regression introduced in r221609.
Before:
void f() { f(a, c *d); }
After:
void f() { f(a, c * d); }
llvm-svn: 222128
Jason Molenda [Mon, 17 Nov 2014 11:43:37 +0000 (11:43 +0000)]
Add a little sketch of a program that can extract unwind
information from the compact unwind section used on darwin
for exception handling, and dump that information..
The UNWIND_X86_64_MODE_RBP_FRAME and UNWIND_X86_64_MODE_DWARF
entries look to be handled correctly.
UNWIND_X86_64_MODE_STACK_IMMD and UNWIND_X86_64_MODE_STACK_IND
are still a work in progress.
Only x86_64 is supported right now. Given that this is an
experiment in parsing the section contents, I don't expect to
add other architectures; they are trivial variations on this
arch. There exists a real dumper included in the Xcode tools,
unwinddump.
llvm-svn: 222127
David Majnemer [Mon, 17 Nov 2014 11:27:45 +0000 (11:27 +0000)]
ScalarEvolution: Construct SCEVDivision's Derived type instead of itself
SCEVDivision::divide constructed an object of SCEVDivision<Derived>
instead of Derived. divide would call visit which would cast the
SCEVDivision<Derived> to type Derived. As it happens,
SCEVDivision<Derived> and Derived currently have the same layout but
this is fragile and grounds for UB.
Instead, just construct Derived. No functional change intended.
llvm-svn: 222126
Oliver Stannard [Mon, 17 Nov 2014 11:18:10 +0000 (11:18 +0000)]
[Thumb1] Re-write emitThumbRegPlusImmediate
This was motivated by a bug which caused code like this to be
miscompiled:
declare void @take_ptr(i8*)
define void @test() {
%addr1.32 = alloca i8
%addr2.32 = alloca i32, i32 1028
call void @take_ptr(i8* %addr1)
ret void
}
This was emitting the following assembly to get the value of %addr1:
add r0, sp, #1020
add r0, r0, #8
However, "add r0, r0, #8" is not a valid Thumb1 instruction, and this
could not be assembled. The generated object file contained this,
resulting in r0 holding SP+8 rather tha SP+1028:
add r0, sp, #1020
add r0, sp, #8
This function looked like it could have caused miscompilations for
other combinations of registers and offsets (though I don't think it is
currently called with these), and the heuristic it used did not match
the emitted code in all cases.
llvm-svn: 222125
David Majnemer [Mon, 17 Nov 2014 11:17:17 +0000 (11:17 +0000)]
Object, COFF: Tighten the object file parser
We were a little lax in a few areas:
- We pretended that import libraries were like any old COFF file, they
are not. In fact, they aren't really COFF files at all, we should
probably grow some specialized functionality to handle them smarter.
- Our symbol iterators were more than happy to attempt to go past the
end of the symbol table if you had a symbol with a bad list of
auxiliary symbols.
llvm-svn: 222124
Oliver Stannard [Mon, 17 Nov 2014 10:49:31 +0000 (10:49 +0000)]
Fix optimisations of SELECT_CC which assumed result is boolean
Some optimisations in DAGCombiner cause miscompilations for targets that use
TargetLowering::UndefinedBooleanContent, because they assume that the results
of a SELECT_CC node are boolean values, and can be safely ANDed, ORed and
XORed. These optimisations are only valid for targets that use
ZeroOrOneBooleanContent or ZeroOrNegativeOneBooleanContent.
This is a follow-up to D6210/r221693.
llvm-svn: 222123
Yaron Keren [Mon, 17 Nov 2014 09:29:33 +0000 (09:29 +0000)]
silence gcc 4.9.1 warning in /llvm/lib/Support/Windows/Path.inc:564:39:
warning: suggest parentheses around assignment used as truth value [-Wparentheses]
if (ec = widenPath(path, path_utf16))
llvm-svn: 222122
Erik Eckstein [Mon, 17 Nov 2014 09:13:57 +0000 (09:13 +0000)]
Optimize switch lookup tables with linear mapping.
This is a simple optimization for switch table lookup:
It computes the output value directly with an (optional) mul and add if there is a linear mapping between index and output.
Example:
int f1(int x) {
switch (x) {
case 0: return 10;
case 1: return 11;
case 2: return 12;
case 3: return 13;
}
return 0;
}
generates:
define i32 @f1(i32 %x) #0 {
entry:
%0 = icmp ult i32 %x, 4
br i1 %0, label %switch.lookup, label %return
switch.lookup:
%switch.offset = add i32 %x, 10
ret i32 %switch.offset
return:
ret i32 0
}
llvm-svn: 222121
Bob Wilson [Mon, 17 Nov 2014 08:00:45 +0000 (08:00 +0000)]
Fix CR/LF line endings in test case.
llvm-svn: 222120
Craig Topper [Mon, 17 Nov 2014 05:58:26 +0000 (05:58 +0000)]
Add missing semicolon from r222118.
llvm-svn: 222119
Craig Topper [Mon, 17 Nov 2014 05:50:14 +0000 (05:50 +0000)]
Move register class name strings to a single array in MCRegisterInfo to reduce static table size and number of relocation entries.
Indices into the table are stored in each MCRegisterClass instead of a pointer. A new method, getRegClassName, is added to MCRegisterInfo and TargetRegisterInfo to lookup the string in the table.
llvm-svn: 222118
Rafael Espindola [Mon, 17 Nov 2014 02:28:27 +0000 (02:28 +0000)]
Add back r222061 with a fix.
This adds back r222061, but now calls initializePAEvalPass from the correct
library to avoid link problems.
Original message:
Don't make assumptions about the name of private global variables.
Private variables are can be renamed, so it is not reliable to make
decisions on the name.
The name is also dropped by the assembler before getting to the
linker, so using the name causes a disconnect between how llvm makes a
decision (var name) and how the linker makes a decision (section it is
in).
This patch changes one case where we were looking at the variable name to use
the section instead.
Test tuning by Michael Gottesman.
llvm-svn: 222117
Rui Ueyama [Mon, 17 Nov 2014 02:04:54 +0000 (02:04 +0000)]
[PECOFF] Fix 32-bit delay-import table.
llvm-svn: 222116
Frederic Riss [Mon, 17 Nov 2014 01:34:15 +0000 (01:34 +0000)]
Implement MachODumper::printFileHeaders
Patch by Chilledheart.
Differential Revision: http://reviews.llvm.org/D6163
llvm-svn: 222115
Craig Topper [Mon, 17 Nov 2014 00:26:50 +0000 (00:26 +0000)]
Replace a couple asserts with static_asserts.
llvm-svn: 222114
Tobias Grosser [Sun, 16 Nov 2014 22:50:23 +0000 (22:50 +0000)]
Enable SCEV based code generation by default
SCEV based code generation allows Polly to detect and generate code for loops
that do not have an explicit induction variable, but only virtual induction
variables given by SCEV.
Being able to do so has two main benefits:
- We can detect more scops by default
- We require less canonicalization before Polly, which means we get closer
to our goal of not touching the IR before analyzing its properties.
Specifically, we do not need to run -polly-indvars to introduce explicit
canonical induction variables.
This switch became possible as both the isl code generation and -polly-parallel
are LNT error free with SCEV based code generation and the isl ast generator.
llvm-svn: 222113
Tobias Grosser [Sun, 16 Nov 2014 22:43:21 +0000 (22:43 +0000)]
Modify test cases to work with SCEV based code generation
This patch includes tests where we actually need to adjust the CHECK lines
for SCEV based code generation. Besides these adjustments we add explicit
calls to -polly-codegen-scev=[true|false] and make sure we test both cases.
llvm-svn: 222112
Tobias Grosser [Sun, 16 Nov 2014 21:43:28 +0000 (21:43 +0000)]
Make usage of scev based code generation explicit in tests
This is in preparation of using SCEV based codegen by default in polly
llvm-svn: 222111
Tobias Grosser [Sun, 16 Nov 2014 21:19:35 +0000 (21:19 +0000)]
Fix typo
llvm-svn: 222110
Craig Topper [Sun, 16 Nov 2014 21:17:18 +0000 (21:17 +0000)]
Convert some EVTs to MVTs where only a SimpleValueType is needed.
llvm-svn: 222109