llvm-svn: 232090

llvm-svn: 232089

Avoids the need for an assert-only variable, among other benefits.

llvm-svn: 232088

llvm-svn: 232087

llvm-svn: 232086

The permps and permd instructions have their operands swapped compared to the
intrinsic definition. Therefore, they do not fall into the INTR_TYPE_2OP
category.

I did not create a new category for those two, as they are the only one AFAICT
in that case.

<rdar://problem/20108262>

llvm-svn: 232085

template class instantiation

No functional changes.

llvm-svn: 232081

This reverts commit r231967 which reinstates r231957.

Now that IntervalMap uses explicitely aligned storage, it should be safe.

llvm-svn: 232080

Currently IntervalMap would assert when used with keys bigger than host
pointers. This patch uses the AlignedCharArrayUnion functionality to
overcome that limitation.

Differential Revision: http://reviews.llvm.org/D8268

llvm-svn: 232079

llvm-svn: 232078

Add -p and -r options to lldb-mi command -file-exec-file-and-symbols to support iOS debugging on macOS.

The patch adds 2 options which are not present in the GDB MI. Those have been described in MIExtensions.txt.

Patch from Chuck Ries.

llvm-svn: 232077

llvm-svn: 232076

Make ModuleCache::Get to return instantiated ModuleSP instance so already created in-memory instance can be returned instead of creating a new one.

http://reviews.llvm.org/D8270

llvm-svn: 232075

Differential Revision: http://reviews.llvm.org/D8288

llvm-svn: 232074

Differential Revision: http://reviews.llvm.org/D8176

llvm-svn: 232073

Differential Revision: http://reviews.llvm.org/D8175

llvm-svn: 232072

classes. Use a Triple instead and simplify a lot of the querying
logic to use lookups on the Triple.

llvm-svn: 232071

A nullptr exception object can be caught by any pointer type catch
handler.  However, it is not possible to express this in the exception
info for the MS ABI.  As a middle ground, allow such exception objects
to be caught with pointer-to-void catch handlers.

llvm-svn: 232069

Avoid a failing test case by fixing things so the compiler generates a line table entry for line the very start of the printf() before any values have been loaded into registers.

The issue was the previous code tried to stop on the following code in main.c:

21    // Stop here and set values
22    printf ("Val - %d Mine - %d, %d, %llu. Ptr - %d, %d, %llu\n", 
23            val, 
24            mine.first_val, mine.second_val, mine.third_val,
25            ptr->first_val, ptr->second_val, ptr->third_val); 

We we set a source regex breakpoint on "// Stop here and set values" we would set a breakpoint on line 22 as expected. 

The problem is the most recent clang compiler generates a line table like this


0x1000: main.c:23 // Loading of "val" into a register
0x1010: main.c:24 // Load mine.first_val, mine.second_val, mine.third_val values into registers or on the stack
0x1020: main.c:25 // Load ptr->first_val, ptr->second_val, ptr->third_val values into registers or on the stack
0x1030: main.c:22 // Call to printf

In this test, we run to line 22, then we use python to modify the value of "val" and then continue to another breakpoint and try to read the STDOUT from the printf to verify the values changed correctly.

With the above line table the value for "val" had already been loaded into a register so the string from printf would be incorrect.

Doing an easy fix for now by changing the code to:

21    // Stop here and set values
22  printf ("Val - %d Mine - %d, %d, %llu. Ptr - %d, %d, %llu\n", val, 
23          mine.first_val, mine.second_val, mine.third_val,
24          ptr->first_val, ptr->second_val, ptr->third_val); 


Now we get a line table entry for line 22 that is before any locals are read from the stack into registers.

I need to follow up with the compiler guys and see if we can get a fix for this as anyone setting file + line breeakpoints might be very surprised to have code from lines below the current line already have had their code run.

llvm-svn: 232068

* put most of the cross-compiling support into a function llvm_create_cross_target_internal.
* when CrossCompile is included it still generates a NATIVE target.
* llvm_create_cross_target function takes a target_name which should match a toolchain.
* llvm_create_cross_target can now be used to target more than one cross-compilation target.

llvm-svn: 232067

Update the doxygen configuration file and the Makefile build rules
to provide better output (simply use the default stylesheet and template
from the Doxygen distribution.)

This CL has upgrade doxygen.cfg.in to Doxygen 1.8.6.

llvm-svn: 232066

llvm-svn: 232065

Update the doxygen configuration file and Makefile build rules
to provide better output (simply use the default stylesheet and template
from the Doxygen distribution.)

This CL has upgrade doxygen.cfg.in to Doxygen 1.8.6.

llvm-svn: 232064

llvm-svn: 232063

This CL adds --enable-doxygen-search to enable doxygen search engine
and --enable-doxygen-qt-help to enable the Qt help file generation.

llvm-svn: 232062

Suggested by Craig Topper in D8275.

This is a follow-on to r232052.

llvm-svn: 232061

llvm-svn: 232060

llvm-svn: 232059

LLDUnitTests target doesn't exist unless LLVM_INCLUDE_TESTS is On.

llvm-svn: 232058

llvm-svn: 232057

Suggested by Craig Topper in D8184.

This goes with r232047.

llvm-svn: 232056

In a static build the dependency is picked up implictly, but not in a shared
library build. This is needed for the new ObjC matchers that reference Selector.

llvm-svn: 232055

* There is no reason to require SDKROOT as an environment variable because we can derive it from xcrun
* Setting CMAKE_RANLIB makes our static archives usable

llvm-svn: 232053

This is very much like D8088 (checked in at r231792).

Now that we've replaced the vinsertf128 intrinsics,
do the same for their extract twins.

Differential Revision: http://reviews.llvm.org/D8275

llvm-svn: 232052

Add some matchers for Objective-C selectors and messages to
ASTMatchers.h. Minor mods to ASTMatchersTest.h to allow test files with
".m" extension in addition to ".cpp".  New tests added to
ASTMatchersTest.c.

Patch by Dean Sutherland.

llvm-svn: 232051

CloudABI does not expose a table on its own.

llvm-svn: 232050

Systems like FreeBSD's Capsicum and Nuxi CloudABI apply the concept of
capability-based security on the way processes can interact with the
filesystem API. It is no longer possible to interact with the VFS
through calls like open(), unlink(), rename(), etc. Instead, processes
are only allowed to interact with files and directories to which they
have been granted access. The *at() functions can be used for this
purpose.

This change adds a new config switch called
_LIBCPP_HAS_NO_GLOBAL_FILESYSTEM_NAMESPACE. If set, all functionality
that requires the global filesystem namespace will be disabled. More
concretely:

- fstream's open() function will be removed.
- cstdio will no longer pull in fopen(), rename(), etc.
- The test suite's get_temp_file_name() will be removed. This will cause
  all tests that use the global filesystem namespace to break, but will
  at least make all the other tests run (as get_temp_file_name will not
  build anyway).

It is important to mention that this change will make fstream rather
useless on those systems for now. Still, I'd rather not have fstream
disabled entirely, as it is of course possible to come up with an
extension for fstream that would allow access to local filesystem
namespaces (e.g., by adding an openat() member function).

Differential revision:	http://reviews.llvm.org/D8194
Reviewed by:		jroelofs (thanks!)

llvm-svn: 232049

Summary:
Add low-frame/high-frame options to -stack-list-arguments

All tests pass on OS X.

Reviewers: clayborg, abidh

Reviewed By: abidh

Subscribers: lldb-commits, clayborg, abidh

Differential Revision: http://reviews.llvm.org/D8282

llvm-svn: 232048

This is a convenience function to ease mask creation of ShuffleVectors
in AutoUpgrade and other places.

Differential Revision: http://reviews.llvm.org/D8184

llvm-svn: 232047

Part of the folding logic implemented by function 'PerformISDSETCCCombine'
only worked under the assumption that the condition code in input could have
been either SETNE or SETEQ.
Unfortunately that assumption was incorrect, and in some cases the algorithm
ended up incorrectly folding SETCC nodes.

The incorrect folding only affected SETCC dag nodes where:
 - one of the operands was a build_vector of all zeroes;
 - the other operand was a SIGN_EXTEND from a vector of MVT:i1 elements;
 - the condition code was neither SETNE nor SETEQ.

Example:
  (setcc (v4i32 (sign_extend v4i1:%A)), (v4i32 VectorOfAllZeroes), setge)

Before this patch, the entire dag node sequence from the example was
incorrectly folded to node %A.

With this patch, the dag node sequence is folded to a
  (xor %A, (v4i1 VectorOfAllOnes)).

Added test setcc-combine.ll.

Thanks to Greg Bedwell for spotting this issue.

llvm-svn: 232046