Commit Graph

183947 Commits

Author SHA1 Message Date
Eric Christopher 78b833bb53 Remove unused variable.
llvm-svn: 218844
2014-10-02 00:52:57 +00:00
Eric Christopher 94ec17fac9 Remove test directories with no tests.
llvm-svn: 218843
2014-10-02 00:42:30 +00:00
Justin Bogner 5eec02a399 InstrProf: Simplify counting a file's regions when writing coverage (NFC)
When writing a coverage mapping we iterate through the mapping regions
in order of FileID, but we were then repeatedly searching from the
beginning of the list to count the number of regions with a given
FileID.

It is simpler and more efficient to search forward from the current
iterator to find the number of regions.

llvm-svn: 218842
2014-10-02 00:31:00 +00:00
Kostya Serebryany 36aa17a4e3 [asan] make new_array_cookie_test more robust
llvm-svn: 218841
2014-10-01 23:24:54 +00:00
David Blaikie e50bd2b21c Reduce the PR20399 test case.
I couldn't get something /really/ obvious, and I imagine Richard Smith
might be able to provide some text explaining the sequence of steps
that's demonstrated by these files - but at least it's a bit simpler
now.

llvm-svn: 218840
2014-10-01 23:16:30 +00:00
Chandler Carruth 8a16802d46 [x86] Improve and correct how the new vector shuffle lowering was
matching and lowering 64-bit insertions.

The first problem was that we weren't looking through bitcasts to
discover that we *could* lower as insertions. Once fixed, we in turn
weren't looking through bitcasts to discover that we could fold a load
into the lowering. Once fixed, we weren't forming a SCALAR_TO_VECTOR
node around the inserted element and instead were passing a scalar to
a DAG node that expected a vector. It turns out there are some patterns
that will "lower" this into the correct asm, but the rest of the X86
backend is very unhappy with such antics.

This should fix a few more edge case regressions I've spotted going
through the regression test suite to enable the new vector shuffle
lowering.

llvm-svn: 218839
2014-10-01 23:14:28 +00:00
Bob Wilson 650cd8a380 PR21101: tablegen's FastISel emitter should filter out unused functions.
FastISel has a fixed set of virtual functions that are overridden by the
tablegen-generated code for each target. These functions are distinguished by
the kinds of operands, e.g., register + immediate = "ri". The FastISel emitter
has been blindly emitting functions with different combinations of operand
kinds, even for combinations that are completely unused by FastISel, e.g.,
"fastEmit_rrr". Change to filter out functions that will be irrelevant for
FastISel and do not bother generating the code for them. Also add explicit
"override" keywords for the virtual functions that are overridden.

llvm-svn: 218838
2014-10-01 22:44:01 +00:00
Jim Ingham 5346f884da Add SBExecutionContext.cpp to this file as well, as the comment
in source/API/CMakeLists.txt instructs...

llvm-svn: 218837
2014-10-01 22:26:52 +00:00
Lang Hames 24f0c24de9 [MCJIT] Don't crash in debugging output for sections that aren't emitted.
llvm-svn: 218836
2014-10-01 21:57:47 +00:00
Enrico Granata da0fd6d3ad Add documentation about the new SBExecutionContext for Python commands feature. This marks the end of rdar://18142728
llvm-svn: 218835
2014-10-01 21:53:32 +00:00
Enrico Granata 06be059ad9 Allow Python commands to optionally take an SBExecutionContext argument in case they need to handle 'where they want to act' separately from the notion of 'currently-selected entity' that is associated to the debugger. Do this in an (hopefully) non-breaking way by running an argcount check before passing in the new argument. Update the test case to also check for this new feature. www update to follow
llvm-svn: 218834
2014-10-01 21:47:29 +00:00
Todd Fiala ebcf42cdec thread state coordinator: added error callbacks, cleaned up tests.
ThreadStateCoordinator changes:
* Most commands that run in the queue now take an error handler that
  will be called with an error string if an error occurs during processing.
  Errors generally stop the operation in progress.  The errors are checked
  at time of execution.  This is intended to help flush out ptrace/waitpid/state management
  issues as quickly as possible.

* Threads now must be known to the coordinator before stops can be reported,
  resumes can be requested, thread deaths can be reported, or deferred stop
  notifications can be made.  Failure to know the thread will cause the coordinator
  to call the error callback for the event being processed.  Threads are introduced
  to the system by the NotifyThreadCreate method.
  
* The NotifyThreadCreate method now takes the initial state of the thread being
  introduces to the system.  We no longer just assume the thread is running.
  
The test cases were cleaned up, too:
* A gtest test fixture is now used, which allows creating less verbose helper
  methods that setup common pieces of callback code for some method invocations.
  Net result: the tests are simpler to read and shorter to write.

llvm-svn: 218833
2014-10-01 21:40:45 +00:00
Eric Christopher f6ed33e7fa constify the TargetMachine argument used in the subtarget and
lowering constructors.

llvm-svn: 218832
2014-10-01 21:36:28 +00:00
Chandler Carruth b8a4ecb5d9 [cmake] Include the new file in the CMake lists. Without this every
CMake build of any part of LLVM with LLDB checked out fails immediately.
=[

We appear to not even have a build bot covering the CMake build of LLDB
which makes this truly terrible. That needs to be fixed immediately.

llvm-svn: 218831
2014-10-01 21:33:28 +00:00
Fariborz Jahanian 56b11ca8a5 Test case for my r218780 patch.
Suggested by Richard Smith.
rdar://18508589.

llvm-svn: 218830
2014-10-01 21:33:22 +00:00
Duncan P. N. Exon Smith 379e375761 DIBuilder: Remove duplicated comments, NFC
These comments already appear in the header, and some of them are
out-of-date anyway.

llvm-svn: 218829
2014-10-01 21:32:15 +00:00
Duncan P. N. Exon Smith 9affbbaac0 Revert "DIBuilder: Remove dead code"
This reverts commit r218820.  It turns out that Adrian has an
outstanding SROA patch that uses this.

I've updated it to forward to `createExpression()`.

llvm-svn: 218828
2014-10-01 21:32:12 +00:00
Alexey Samsonov 1b42bde30d [ASan] Don't use large stack buffer for local variable name in stack frame description
llvm-svn: 218827
2014-10-01 21:28:54 +00:00
Johannes Doerfert c3285a1ded Change the output of arc unit
arc unit will now show the number of consecutive tests with the same
  result instead of printing a "." for each one. Due to the number of
  tests the "dots" didn't fit in one line any more. Furthermore the
  result list is shortened, only non passing tests or tests taking
  longer than a time threshold (50ms) will be reported (both to the user
  and to phabricator).

llvm-svn: 218826
2014-10-01 21:26:24 +00:00
Joerg Sonnenberger 2960178a77 Fix trailing commas in AMD define.
llvm-svn: 218825
2014-10-01 21:22:17 +00:00
Joerg Sonnenberger e028e05a7e Add the various signature macros.
llvm-svn: 218824
2014-10-01 21:21:42 +00:00
Joerg Sonnenberger cf0740454d Rename bit_RDRAND to bit_RDRND to match GCC's version of this header.
llvm-svn: 218823
2014-10-01 21:21:16 +00:00
Sanjay Patel 9ebfbb969d Lower FNEG ( FABS (x) ) -> FNABS (x) [X86 codegen] PR20578
Negative FABS of either a scalar or vector should be handled the same way
on x86 with SSE/AVX: a single OR instruction of the FP operand with a
constant to light up the sign bit(s).

http://llvm.org/bugs/show_bug.cgi?id=20578

Differential Revision: http://reviews.llvm.org/D5201

llvm-svn: 218822
2014-10-01 21:20:06 +00:00
David Blaikie 7fad1b4374 Update test name to match changes made in r218783
Addressing post commit review feedback from Justin Bogner.

llvm-svn: 218821
2014-10-01 21:19:39 +00:00
Duncan P. N. Exon Smith 1ce4fd36bf DIBuilder: Remove dead code
I neglected to update `DIBuilder::createPieceExpression()` in r218797,
which I noticed while rebasing a patch for PR17891.  On closer
inspection, it looks like dead code.

If there are any downstream users of this, you should transition to the
more general `createExpression()`.  Or, we can add this back, but then
it should just forward to `createExpression()`.

llvm-svn: 218820
2014-10-01 21:14:20 +00:00
Alexey Samsonov 0470e24780 [ASan] Make stack-buffer-overflow reports more robust
Summary:
Fix the function that gets stack frame description by address in
thread stack, so that it clearly indicates failures. Make this error non-fatal,
and print as much information as we can in this case. Make all errors in
ParseFrameDescription non-fatal.

Test Plan: check-asan testsuite

Reviewers: kcc

Reviewed By: kcc

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D5554

llvm-svn: 218819
2014-10-01 21:13:00 +00:00
Chandler Carruth f795e4805e [x86] Merge the remaining test cases into vector-blend.ll and remove all
the ISA-specific test files.

llvm-svn: 218818
2014-10-01 21:07:07 +00:00
Eric Christopher eb6e3bbf47 Now that the optimization level is adjusting the feature string
before we hit the subtarget, remove the constructor parameter.

llvm-svn: 218817
2014-10-01 21:05:35 +00:00
Chandler Carruth b21e033b2a [x86] Expand the ISA coverage of our blend test in preparation for
merging ISA-specific testing into this file.

llvm-svn: 218816
2014-10-01 21:03:21 +00:00
Argyrios Kyrtzidis 0b9f5507c8 Adds 'override' to overriding methods. NFC.
llvm-svn: 218815
2014-10-01 21:00:44 +00:00
Chandler Carruth a406d6c386 [x86] Merge the interesting test cases from blend-msb.ll into
vector-blend.ll and remove the former.

llvm-svn: 218814
2014-10-01 20:56:57 +00:00
Chandler Carruth 19a5366481 [x86] Move the AVX blend test to a generic name. I'm going to fold other
blend tests into this one.

llvm-svn: 218813
2014-10-01 20:52:55 +00:00
Enrico Granata d1fd3ce42e Add an accessor to PyCallable that allows one to determine the count of arguments that a Python function allows, and whether varargs/kwargs are also accepted by the same function
llvm-svn: 218812
2014-10-01 20:51:50 +00:00
Chandler Carruth b4b09c7a6c [x86] Remove a test that wasn't doing anything really. We have plenty of
better tests for zext of vectors at this point.

llvm-svn: 218811
2014-10-01 20:50:58 +00:00
Chandler Carruth ab5ddea2cb [x86] Add a 32-bit run to the sext test, and remove a sad vec_sext.ll
test file.

This old test had a bunch of functions that were never even checked. =/
The only thing it really did was to make sure that we did something
reasonable in 32-bit mode with SSE4.1. Adding another run line to the
main vector-sext.ll test seems a better way to do that.

llvm-svn: 218810
2014-10-01 20:49:54 +00:00
Fariborz Jahanian 77a835bf56 Objective-C Modernizer. Patch to remove dangling space
before the semicolon wahen modernizing to use 
NS_ENUM/NS_OPTIONS macros. rdar://18498539

llvm-svn: 218809
2014-10-01 20:46:32 +00:00
Enrico Granata e85e84a769 Add a new SBExecutionContext class that wraps an ExecutionContextRef. This class is a convenient way at the API level to package a target,process,thread and frame all together - or just a subset of those
llvm-svn: 218808
2014-10-01 20:43:45 +00:00
Chandler Carruth bbbdb9f0ee [x86] Teach both sext and zext vector tests to cover a nice wide range
of architectures: SSE2, SSSE3, SSE4.1, AVX, and AVX2.

Unfortunately, this exposses the absolute horror of the code we generate
for many of these patterns. Anyone wanting to familiarize themselves
with the x86 backend and improve performance could do a lot of good
sitting down and making these test cases not look so terrible. While the
new vector shuffle code I'm working on well help some, it won't fix all
of the crimes here.

llvm-svn: 218807
2014-10-01 20:41:36 +00:00
Adrian Prantl e6579cd9a6 Update testcase to new intrinsic format
llvm-svn: 218806
2014-10-01 20:40:12 +00:00
Eric Christopher 36448af7f5 Rework the PPC TargetMachine so that the non-function specific
overrides happen at TargetMachine creation and not on every
subtarget creation.

llvm-svn: 218805
2014-10-01 20:38:26 +00:00
Eric Christopher 12f4a78581 constify TargetMachine parameter for X86TargetLowering.
llvm-svn: 218804
2014-10-01 20:38:22 +00:00
Sanjay Patel 7b2cd9ad86 Make the sqrt intrinsic return undef for a negative input.
As discussed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20140609/220598.html

And again here:
http://lists.cs.uiuc.edu/pipermail/llvmdev/2014-September/077168.html

The sqrt of a negative number when using the llvm intrinsic is undefined. 
We should return undef rather than 0.0 to match the definition in the LLVM IR lang ref.

This change should not affect any code that isn't using "no-nans-fp-math"; 
ie, no-nans is a requirement for generating the llvm intrinsic in place of a sqrt function call.

Unfortunately, the behavior introduced by this patch will not match current gcc, xlc, icc, and 
possibly other compilers. The current clang/llvm behavior of returning 0.0 doesn't either. 
We knowingly approve of this difference with the other compilers in an attempt to flag code 
that is invoking undefined behavior.

A front-end warning should also try to convince the user that the program will fail:
http://llvm.org/bugs/show_bug.cgi?id=21093

Differential Revision: http://reviews.llvm.org/D5527

llvm-svn: 218803
2014-10-01 20:36:33 +00:00
Chandler Carruth b5c9e04b51 [x86] Sort the ISA-specific RUN lines for vector-sext.ll to go from
oldest to newest. This makes more sense to me and is more consistent
with other tests.

llvm-svn: 218802
2014-10-01 20:32:44 +00:00
Tim Northover 6a1ef73140 ARM: yes it can (as of r218789)
llvm-svn: 218801
2014-10-01 20:31:58 +00:00
Chandler Carruth c66ea0fc12 [x86] Rename avx-{s,z}ext.ll to vector-{s,z}ext.ll.
These tests are far and away the best sext and zext tests we have for
vectors. I'm going to merge the other similar tests into them and expand
the ISA coverage.

llvm-svn: 218800
2014-10-01 20:30:30 +00:00
Chandler Carruth 011088fc46 [x86] Cleanup and re-generate the checks for avx-zext.ll using the new
script.

llvm-svn: 218799
2014-10-01 20:27:16 +00:00
Duncan P. N. Exon Smith f3dc429ac7 DIBuilder: Encapsulate DIExpression's element type
Update for corresponding LLVM API change for
`DIBuilder::createExpression()`.

llvm-svn: 218798
2014-10-01 20:26:18 +00:00
Duncan P. N. Exon Smith 611afb229c DIBuilder: Encapsulate DIExpression's element type
`DIExpression`'s elements are 64-bit integers that are stored as
`ConstantInt`.  The accessors already encapsulate the storage.  This
commit updates the `DIBuilder` API to also encapsulate that.

llvm-svn: 218797
2014-10-01 20:26:08 +00:00
Nick Kledzik 22c9073ada Add MachOLinkingContext parameter to MachOFileNode constructor.
No functionality change.  This removes a down-cast from LinkingContext to
MachOLinkingContext.

Also, remove const from LinkingContext::createImplicitFiles() to remove
the need for another const cast.  Seems reasonable for createImplicitFiles()
to need to modify the context (MachOLinkingContext does).

llvm-svn: 218796
2014-10-01 20:24:30 +00:00
Chandler Carruth fbba2fa8d9 [x86] Generate the FileCheck assertions for avx-blend.ll with my new
script to make them nice and predictable. This will ease updating them
for the new vector shuffle lowering and seeing the delta if any.

llvm-svn: 218795
2014-10-01 20:19:45 +00:00