Added test in r305213.

OK

Raphael , do you have commit access? should I commit this?

LGTM

Please move the =new out of the PP.AddPragmaHandler calls.
While at it, this code still (as the original) leaks the PragmaHandlers. These should be deleted after RemovePragmaHandler or instead, simply use std::unique_ptr.

reverted in r300497

Adding Mateusz and Reid.

Added attribute test.

Thanks, I'll make a test.

r299207

No intention of any major work intended... just moved the code (locally) to Release and regression tests were still passing.

Most other module flags are added at CodeGenModule::Release(). For consistency, could this code be in CodeGenModule::Release() as well?

Could use dumpDeclContext() to test?

OK.

If something is broken now we don't break it even more.
__float128 remain be fixed to be compatible, not only some poor developer would have to fix the missing headers bug one day, he will have to re-fix limits.h the right way and undo this "fix".
There is a problems with limits.h, fix limits.h. Don't make all headers that happens to be in the same directory as limits.h disappear.

These directories are "mostly" equivalent, with some headers existing is mings gcc dir but missing in clang resource dir and thus will break compilation.
For example, the most popular (37089 download this week) https://sourceforge.net/projects/mingw-w64 distribuion for Windows had elected to place the includes

omp.h, quadmath.h, openacc.h

at

c:\mingw32\lib\gcc\i686-w64-mingw32\5.4.0\include\

with the corresponding libraries at

c:\mingw32\lib\gcc\i686-w64-mingw32\5.4.0\

I had verified this still holds true for the latest i686-6.3.0-release-posix-dwarf-rt_v5-rev1.7z package.
Removing the include dir will make clang less useful for mingw users using these includes, since they will not be found anymore.
Furthermore, It makes no sense to make such intrusive wide-ranging change where the original problem was with limits.h is actually due to:

/* The system's limits.h may, in turn, try to #include_next GCC's limits.h.
 Avert this #include_next madness. */
#if defined __GNUC__ && !defined _GCC_LIMITS_H_
#define _GCC_LIMITS_H_
#endif

so the solution may be as simple as

#if defined __GNUC__ && !defined _GCC_LIMITS_H_ && !defined __MINGW32__
#define _GCC_LIMITS_H_
#endif

What about omp.h and openacc.h ? many programs are using OpenMP.

The gcc include dirs in mingw contains headers not available esewhere and thus can't be removed.
Notable examples,

Hiding these two include dirs removes many headers. Most has clang equivalents but not all of them.
For example quadmath.h is only there, and without the include path programs using it will fail to compile.

This code is actually used with Windows as well as Linux (with the exception of line 218), see the comment blocks above for detailed include dirs from all platforms from which it was derived.

LGTM, matches the code in libstdc++ stdarg,h.
You can remove the 'hack' comment in line 46, __GNUC_VA_LIST is just a standard include guard for the typedef.

In D28399#642458, @v.g.vassilev wrote:

@yaron.keren, it seems that http://llvm.org/pr30994 concerns friend function declarations. My current patch focuses on friend class templates. Perhaps we should open another review item for a fix of http://llvm.org/pr30994.

While at it, http://llvm.org/pr30994

LGTM after fixing the inline comment

PR28749 was created with trunk clang on Ubuntu x64.
It reproduced reliably but is very fragile, changing just about anything in the code makes the problem disappear.

I added a small reproducer for what is probably the same problem in

You mean
Some target platforms do not support -fsanitize=address.

result of r269402, LGTM

Ah, OK, I am using regular svn and thus did not counter this problem.
Could you detail this exactly, the problem happens when using git-svn with core.autocrlf config=true but not with svn?
LGTM with this addition.

AFAIK the eol type is decided by svn based the svn:eol-style property, so as long the files were checked in correctly and you do not override it specifically while svn checkout, it just works.
Did you encounter problems with eol ?

Would be nice to have CodeGenAction::TheModule redirect to CodeGeneratorImpl::M if possible, but that's for another patch. LGTM.

It certainly makes sense to redirect the module request to its owner instead of duplicating it in a local copy.

We have tried to keep one copy of DataLayout around
http://reviews.llvm.org/D11103
Can it share the Module->getDataLayout() ?

The instantiated does get a new collection of ParmVarDecls while getting
the pattern Type which points to the pattern ParmVarDecls. So the
ParmVarDecls pointed in the instantiated are not the same as those pointed
by its Type.

#define NOGDI
is also useful.

+1 all general-purpose Windows stuff should move to WindowsSupport.h. I'm surprised we got away without so far.
We have NOMINMAX at lib/Driver/MSVCToolChain.cpp before Windows.h which should probably switched to use WindowsSupport.h.

We don't support building with mingw.org.
mingw-w64 has its own versionhelpers.h with IsWindows8OrGreater(), at least from gcc 4.93, probably earlier.

LGTM with some tests cases.

http://bb.pgr.jp/grid

MSVC 2013 Update 5 accepts for (bool SkipUnwritten : {false, true}).
Possibly changed in one of the Updates?

Only abstracting the whole llc/opt drivers which is more trouble than this is worth.
Maybe a commment in llc that opt does something similar and vice versa.
LGTM.

Do you see any reasonable way to share the duplicated code between llc.cpp and opt.cpp?

There is another copy of getEffectiveDeclContext in MicrosoftMangle.cpp which pre-commit was identical to this one.
Should it get this commit as well to keep both identical?
If so, could both copies of getEffectiveDeclContext be merged?

findGccDir() can return llvm::ErrorOr<std::string> and then all Base assignments happen at the same if-elseif-else:

This always searches for something-gcc and then discards the result if sysroot was provided, which is a waste.
Move the searching to a helper function and then it can be done only if sysroot was not provided, as it is now.

You are right, running passes over again for all but trivial code will not result in the same module.
The CloneModule approach should work.

Won't the second run output will overwrite the first run output?
If so, what do you think about renaming the second output so it may be binary compared to the first one?

This makes sense.

We have been fixed such bugs before, usually when a data member is added to one of the Streamers, so LGTM.

That's a great idea. Maybe add a LIT test that uses this flag so breakages will be caught at the buildbots?

r253769

Visual C++ does not support hexadecimal floating point literals.
Since all of these are actually integers they should work without the ".0" and ".0f".

Hi Andy,

The formatting is wrong, aligned to the left, clang-format the new code.

Neat and efficient, LGTM

We did not have cfe-commits as subscriber so I'm adding it now (this is a clang commit), see if someone would like to further comment.

I see the usage pattern is CopyEnvironment, push_back something, use std::vector<const char *> with ExecuteAndWait. For this you must have a consecutive vector of char * in memory, so someone will have to manage their lifetime. It's better not to leave this to the caller or split responsibility so you could objectize CopyEnvironment to do everything required:

This code was my first thought too, but I had been commented on the structure of very similar code in this file:

Yes, I think this is a better default than /usr.

The code could now be merged (untested). I disconnected the third part on purpose based on comment from Joerg Sonnenberger on the previous code that having else after #endif is confusing.

clang can't generate Visual C++ debug info except line numbers yet, so building with VC is essential to development on Windows. While most people and our bots are still using VC 2013 going forward supporting VC 2015 (even with bugs) is worth a small patch.