This was a bug in keeping track of the available domains when merging
domain values.

The wrong domain mask caused ExecutionDepsFix to try to move VANDPSYrr
to the integer domain which is only available in AVX2.

Also add an assertion to catch future attempts at emitting AVX2
instructions.

llvm-svn: 145096

directory.
Patch by Markus Trippelsdorf.

llvm-svn: 145095

reversed in the function's original ordering, and we happened to
encounter it while handling an outer unnatural CFG structure.

Thanks to the test case reduced from GCC's source by Benjamin Kramer.
This may also fix a crasher in gzip that Duncan reduced for me, but
I haven't yet gotten to testing that one.

llvm-svn: 145094

llvm-svn: 145093

llvm-svn: 145092

appear in non-constant initializers in C++.

llvm-svn: 145087

to 30% of memory. The size doubling was killing us and we ended up with up to
just under 50% of empty capacity. Cleaning this up saves us a ton of memory.

llvm-svn: 145086

having the enumeration take up 32 bits for the type and by putting it into the
bitfields that were already being used.

llvm-svn: 145084

The assert wasn't adding much value but slowed down Release+Asserts builds.

llvm-svn: 145082

llvm-svn: 145079

llvm-svn: 145078

This fixes a crasher in tools like Include-What-You-Use which examine such
arguments. Patch from Dean Sturtevant at Google!

llvm-svn: 145077

Contributed by: Marcello Maggioni <hayarms@gmail.com>

llvm-svn: 145076

llvm-svn: 145075

This fixes (or hides) the problem of symbols not being available in the
autoconf build.

llvm-svn: 145074

llvm-svn: 145073

llvm-svn: 145072

This replaces the old option -polly-use-pocc. Also call the passes uniformly
-polly-opt-pocc and -polly-opt-isl.

llvm-svn: 145071

We can reuse the information and avoid looping over all the bytes again.

llvm-svn: 145070

making sure we perfectly size our vector of symbols on the symbol table.

llvm-svn: 145069

Fixed an issue with the options for memory read where --count couldn't be used
with the --binary option when writing data to a file.

Also removed the GDB format option from the --binary version of memory read.

llvm-svn: 145067

add basic PPC register-pressure feedback; adjust the vaarg test to match the new register-allocation pattern

llvm-svn: 145065

More fixes to the X86InstComments for shuffle instructions. In particular add AVX flavors of many instructions and fix the destination operand for some of the existing AVX entries.

llvm-svn: 145063

updateTerminator code didn't correctly handle EH terminators in one very
specific case. AnalyzeBranch would find no terminator instruction, and
so the fallback in updateTerminator is to assume fallthrough. This is
correct, but the destination of the fallthrough was assumed to be the
first successor.

This is *almost always* true, but in certain cases the loop
transformations will cause the landing pad to be the first successor!
Instead of this brittle logic, actually look through the successors for
a non-landing-pad accessor, and to assert if more than one is found.

This will hopefully fix some (if not all) of the self host miscompiles
with block placement. Thanks to Benjamin Kramer for reporting, Nick
Lewycky for an initial stab at a reduction, and Duncan for endless
advice on EH (which I know nothing about) as well as reviewing the
actual fix.

llvm-svn: 145062

llvm-svn: 145061

dropping weights on the floor for invokes. This was impeding my writing
further test cases for invoke when interacting with probabilities and
block placement.

No test case as there doesn't appear to be a way to test this stuff. =/
Suggestions for a test case of course welcome. I hope to be able to add
test cases that indirectly cover this eventually by adding probabilities
to the exceptional edge and reordering blocks as a result.

llvm-svn: 145060

This was put in because in a certain version of DragonFlyBSD stat(2) lied about the
size of some files. This was fixed a long time ago so we can remove the workaround.

llvm-svn: 145059

llvm-svn: 145058

llvm-svn: 145057

before the clobber so that we copy the value if needed.

Fixes pr11415.

llvm-svn: 145056

Fix shuffle decoding logic to handle UNPCKLPS/UNPCKLPD on 256-bit vectors correctly. Add support for decoding UNPCKHPS/UNPCKHPD for AVX 128-bit and 256-bit forms.

llvm-svn: 145055

implement __has_feature(address_sanitizer); also use LangOpts.AddressSanitizer instead of CodeGenOpts.AddressSanitizer

llvm-svn: 145054

Add methods for querying minimum SSE version along with AVX. Simplifies all the places that had to check a version of SSE and AVX.

llvm-svn: 145053

llvm-svn: 145052

easily be used in the next run.

llvm-svn: 145051

something like "display/4i $pc" (or something like this). With LLDB we already
were showing 3 lines of source before and 3 lines of source after the current
source line when showing a stop context. We now improve this by allowing the
user to control the number of lines with the new "stop-line-count-before" and
"stop-line-count-after" settings. Also, there is a new setting for how many
disassembly lines to show: "stop-disassembly-count". This will control how many
source lines are shown when there is no source or when we have no source line
info. 

settings set stop-line-count-before 3
settings set stop-line-count-after 3
settings set stop-disassembly-count 4
settings set stop-disassembly-display no-source

The default values are set as shown above and allow 3 lines of source before 
and after (what we used to do) the current stop location, and will display 4 
lines of disassembly if the source is not available or if we have no debug
info. If both "stop-source-context-before" and "stop-source-context-after" are
set to zero, this will disable showing any source when stopped. The 
"stop-disassembly-display" setting is an enumeration that allows you to control
when to display disassembly. It has 3 possible values:

"never" - never show disassembly no matter what
"no-source" - only show disassembly when there is no source line info or the source files are missing
"always" - always show disassembly.

llvm-svn: 145050

-ftemplate-depth= and -fconstexpr-depth= in it.

llvm-svn: 145049

llvm-svn: 145048

llvm-svn: 145047

semantics and defaults as the corresponding g++ arguments. The historical g++
argument -ftemplate-depth-N is kept for compatibility, but modern g++ versions
no longer document that option.

Add -cc1 argument -fconstexpr-depth N to implement the corresponding
functionality.

The -ftemplate-depth=N part of this fixes PR9890.

llvm-svn: 145045