v2: document why we hardcode VCC for now.

This is a candidate for the mesa-stable branch.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 176099

Prevent producing real strange tablegen code by using
proper register sizes, alignments and hierarchy.

Also cleanup the unused definitions and add some comments.

v2: add SGPR 512 bit registers, stop registers from wrapping around,
    fix SGPR alignment

This is a candidate for the mesa-stable branch.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 176098

This is a candidate for the mesa-stable branch.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 176097

llvm-svn: 176096

llvm-svn: 176095

The PowerPC TLS relocation types were not previously added to the
necessary list in MCELFStreamer::fixSymbolsInTLSFixups().  Now they are!

llvm-svn: 176094

Michael van der Westhuizen: correction to the libcxx build instructions when built with libcxxrt on Linux.

llvm-svn: 176093

Before:
if (Intervals[i].getRange().getFirst() < Intervals[i - 1]
                                             .getRange().getLast()) {}

After:
if (Intervals[i].getRange().getFirst() <
    Intervals[i - 1].getRange().getLast()) {}

llvm-svn: 176092

llvm-svn: 176091

llvm-svn: 176090

[sanitizer] A low-level vector implementation to be used in leak checking code. Patch by Sergey Matveev

llvm-svn: 176089

Sometimes android linker is "arm-linux-androideabi-ld", and not just "ld".

llvm-svn: 176088

Before:
for (const aaaaaaaaaaaaaaaaaaaaa &
         aaaaaaaaa : aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa) {}

After:
for (const aaaaaaaaaaaaaaaaaaaaa &aaaaaaaaa :
     aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa) {}

llvm-svn: 176087

Empty lines followed by line comments are often used to highlight the
comment. Empty lines somewhere else are usually left over from manual or
automatic formatting and should probably be removed.

Before (clang-format would keep):
S s = {
  a,

  b
};

After:
S s = { a, b };

llvm-svn: 176086

llvm-svn: 176085

We might want to move towards doing this if the formatting can be
significantly improved, but we need to carefully evaluate the different
situations first.

Before (the string literal was split by clang-format here):
aaaaaaaaaaaaaaaaaaaa(aaaaaaaaaaaaaaaaaaaa, aaaaaa("aaa aaaaa aaa aaa aaaaa aaa "
                                                  "aaaaa aaa aaa aaaaaa"));

After:
aaaaaaaaaaaaaaaaaaaa(
    aaaaaaaaaaaaaaaaaaaa,
    aaaaaa("aaa aaaaa aaa aaa aaaaa aaa aaaaa aaa aaa aaaaaa"));

llvm-svn: 176084

Define CMake option CLANG_INCLUDE_TESTS *before* traversing into tests/ subdirectory. Otherwise, while configuring the build tree for the first time, Clang unit tests could avoid being added to 'check-clang' command, and thus avoid being built and executed.

llvm-svn: 176080

llvm-svn: 176079

[asan] fix the output for range accesses (memset, etc); improve the tests; more strict checking in memcmp

llvm-svn: 176078

llvm-svn: 176077

These are two related changes (one in llvm, one in clang).
LLVM: 
- rename address_safety => sanitize_address (the enum value is the same, so we preserve binary compatibility with old bitcode)
- rename thread_safety => sanitize_thread
- rename no_uninitialized_checks -> sanitize_memory

CLANG: 
- add __attribute__((no_sanitize_address)) as a synonym for __attribute__((no_address_safety_analysis))
- add __attribute__((no_sanitize_thread))
- add __attribute__((no_sanitize_memory))

for S in address thread memory
If -fsanitize=S is present and __attribute__((no_sanitize_S)) is not
set llvm attribute sanitize_S

llvm-svn: 176076

These are two related changes (one in llvm, one in clang).
LLVM: 
- rename address_safety => sanitize_address (the enum value is the same, so we preserve binary compatibility with old bitcode)
- rename thread_safety => sanitize_thread
- rename no_uninitialized_checks -> sanitize_memory

CLANG: 
- add __attribute__((no_sanitize_address)) as a synonym for __attribute__((no_address_safety_analysis))
- add __attribute__((no_sanitize_thread))
- add __attribute__((no_sanitize_memory))

for S in address thread memory
If -fsanitize=S is present and __attribute__((no_sanitize_S)) is not
set llvm attribute sanitize_S

llvm-svn: 176075

llvm-svn: 176074

Fix a bug where if two atoms had the same index in the override map,
the compare would return false. It now goes to the next check when
they are equal.

No test because it currently can't be tested. An upcomming patch will test it.

llvm-svn: 176073

Use a DenseMap instead of a std::map for AnalysisID -> Pass* maps. This reduces the pass-manager overhead from FPPassManager::runOnFunction() by about 10%.

llvm-svn: 176072

llvm-svn: 176071

No functionality change.

llvm-svn: 176070

Normally, we need to look through derived-to-base casts when creating
temporary object regions (added in r175854). However, if the temporary
is a pointer (rather than a struct/class instance), we need to /preserve/
the base casts that have been applied.

This also ensures that we really do create a new temporary region when
we need to: MaterializeTemporaryExpr and lvalue CXXDefaultArgExprs.

Fixes PR15342, although the test case doesn't include the crash because
I couldn't isolate it.

llvm-svn: 176069

With the new support for trivial copy constructors, we are not always
consistent about whether a CXXTempObjectRegion gets reused or created
from scratch, which affects whether qualifiers are preserved. However,
we probably don't care anyway.

This also switches to using the current PrintingPolicy for the type,
which means C++ types don't get a spurious 'struct' prefix anymore.

llvm-svn: 176068

SymbolFileDWARF code:

  - If a class is being uniqued to another copy of itself
    and the method lists don't match exactly, take a slow
    path and at least unique the methods that they have
    in common.

  - Sort name_to_die maps before querying them.  This
    would otherwise result in uniquing failures because
    looking up a name in a map that contains it would
    often fail.

  - Tolerate classes in other symbol files in the case
    of debugging with .o files rather than with a
    .dSYM.  We used to assume that the classes being
    uniqued were in the same symbol file, causing
    unpredictable results.

This will dramatically reduce the number of cases where
a function does not have a valid DeclContext.

<rdar://problem/12153915>

llvm-svn: 176067

arguments type is a simple type.
rdar://13290455

llvm-svn: 176066

llvm-svn: 176065

plugin.dynamic-loader.darwin-kernel.load-kexts setting, don't print
any messages about loading the kexts (which isn't being done) and
don't read the Mach-O headers out of memory (which can be slow and
they're not being used for anything at this point).

llvm-svn: 176064

Set the exception breakpoints more efficiently by specifying two module basenames as module filters for Apple vendor targets.

llvm-svn: 176063

Really don't call sysctl again when we already have the result.

llvm-svn: 176062

Remove the getCategory from TestDataFormatterObjC.py, since it was superceded by the .categories file, 
and didn't work anyway (getCategories currently has to be a method on the test class, not on the test.)
Add a "basic_process" category, and start to find some tests for simple process running sniff tests.

llvm-svn: 176061

- Put expensive checking after simple one

llvm-svn: 176060

llvm-svn: 176059

- Check whether SSE is available before lowering all 1s vector building with
  PCMPEQD, which is only available from SSE2

llvm-svn: 176058

This changes the RecordingMemoryManager in lli to use mapped memory rather than malloc to allocate memory for sections and uses a 'near' MemoryBlock to keep the allocations together.  This works around a problem in MCJIT where relocations are applied to a generated image immediately oupon generation, which isn't appropriate for the remote case.

llvm-svn: 176057