-u behaviour is apparently not portable between linkers (see cfe-commits
discussions for r204379 and r205012). I've moved the logic to IRGen,
where it should have been in the first place.
I don't have a Linux system to test this on, so it's possible this logic
*still* doesn't pull in the instrumented profiling runtime on Linux.
I'm in the process of getting tests going on the compiler-rt side
(llvm-commits "[PATCH] InstrProf: Add initial compiler-rt test"). Once
we have tests for the full flow there, the runtime logic should get a
whole lot less brittle.
<rdar://problem/16458307>
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@205023 91177308-0d34-0410-b5e6-96231b3b80d8
This follows the LLVM change to canonicalise the Windows target triple
spellings. Rather than treating each Windows environment as a single entity,
the environments are now modelled properly as an environment. This is a
mechanical change to convert the triple use to reflect that change.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@204978 91177308-0d34-0410-b5e6-96231b3b80d8
In gcc using -Ofast forces linking of crtfastmath.o.
In the current clang crtfastmath.o is only linked when -ffast-math/-funsafe-math-optimizations passed. It can lead to performance issues, when using only -Ofast without explicit -ffast-math (I faced with it).
My patch fixes inconsistency with gcc behaviour and also introduces few tests on it.
Patch by Zinovy Nis!
Differential Revision: http://llvm-reviews.chandlerc.com/D3114
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@204742 91177308-0d34-0410-b5e6-96231b3b80d8
These flags are deprecated since at least Clang 3.3. Users should instead
use -fsanitize= with appropriate values.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@204330 91177308-0d34-0410-b5e6-96231b3b80d8
This is because the PCH is tied to the module files, if one of the module files changes or gets removed
the build system should re-build the PCH file.
rdar://16321245
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@203885 91177308-0d34-0410-b5e6-96231b3b80d8
When enabled, always validate the system headers when loading a module.
The end result of this is that when these headers change, we will notice
and rebuild the module.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@203630 91177308-0d34-0410-b5e6-96231b3b80d8
Use the new getObjectFormat/setObjectFormat instead of Environment now that the
file format is a separate field.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@203161 91177308-0d34-0410-b5e6-96231b3b80d8
This is used to avoid conflicts with user modules with the same name from different workspaces.
rdar://16042513
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@202683 91177308-0d34-0410-b5e6-96231b3b80d8
Generating RTTI in the MS ABI is currently not supported, and the failures
are confusing to users, so let's disable it by default for now.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@202178 91177308-0d34-0410-b5e6-96231b3b80d8
-march which doesnt exist on sparc gcc to -mcpu. While here adjust a
few tests to not write an unused temporary file.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@202177 91177308-0d34-0410-b5e6-96231b3b80d8
The integrated assembler is a feature. This makes the new flags the default
option, and the previous versions aliases. Ideally, at some point the aliases
would be entirely removed.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@201963 91177308-0d34-0410-b5e6-96231b3b80d8
Forward the -no-integrated-as option to -cc1 rather than simply invoking the
appropriate tool. This is useful since this option has been overloaded to
permit disabling of parsing inline assembly at the MC layer.
This re-applies the previous version of the patch with a renaming of the driver
option to the public name rather than the internal name (-target vs -triple).
The actual failure is fixed separately of an overly aggressive negative pattern
match in the MIPS driver tests. It also fixes the incorrect test for targets
that have the integrated assembler disabled by default.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@201960 91177308-0d34-0410-b5e6-96231b3b80d8
Forward the -no-integrated-as option to -cc1 rather than simply invoking the
appropriate tool. This is useful since this option has been overloaded to
permit disabling of parsing inline assembly at the MC layer.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@201952 91177308-0d34-0410-b5e6-96231b3b80d8
Added two new options for -mfpu when targetting ARM:
* fpv4-sp-d16
* fp4-sp-d16
The first is the same spelling as gcc.
The lack of a leading `v' is correct, this is consistent with ARM's
documentation and gcc's spelling of the option.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@201846 91177308-0d34-0410-b5e6-96231b3b80d8
Clang itself only emits CodeView line tables, so it seems more
consistent to ask cl.exe for the same format.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@201721 91177308-0d34-0410-b5e6-96231b3b80d8
I'm holding this change to give maintainers of Darwin buildbots more time
to update their toolchains.
This reverts commit r201375.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@201520 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is not strictly correct nor accounts for all uses (shared
objects, for example), but it allows one to test the compiler-rt library
on GNU targets.
Using this patch to run the test-suite has already shown me problems
on ARM. Since this is a Darwin-only flag, nobody is using it, so it
shouldn't be a problem.
I will need extension to deal with the shared cases, but since we're
not compiling libclang_rt.so, that's not yet applicable. Many other
problems will have to be fixed first in compiler-rt (such as removing
the 'arch' name from it and making it trully multi-arch, moving it to
the default lib directory, make both .a and .so variants, etc).
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@201307 91177308-0d34-0410-b5e6-96231b3b80d8
These features are new in VS 2013 and are necessary in order to layout
std::ostream correctly. Currently we have an ABI incompatibility when
self-hosting with the 2013 stdlib in our convertible_fwd_ostream wrapper
in gtest.
This change adds another implicit attribute, MSVtorDispAttr, because
implicit attributes are currently the best way to make sure the
information stays on class templates through instantiation.
Reviewers: majnemer
Differential Revision: http://llvm-reviews.chandlerc.com/D2746
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@201274 91177308-0d34-0410-b5e6-96231b3b80d8
the build
When Clang loads the module, it verifies the user source files that the module
was built from. If any file was changed, the module is rebuilt. There are two
problems with this:
1. correctness: we don't verify system files (there are too many of them, and
stat'ing all of them would take a lot of time);
2. performance: the same module file is verified again and again during a
single build.
This change allows the build system to optimize source file verification. The
idea is based on the fact that while the project is being built, the source
files don't change. This allows us to verify the module only once during a
single build session. The build system passes a flag,
-fbuild-session-timestamp=, to inform Clang of the time when the build started.
The build system also requests to enable this feature by passing
-fmodules-validate-once-per-build-session. If these flags are not passed, the
behavior is not changed. When Clang verifies the module the first time, it
writes out a timestamp file. Then, when Clang loads the module the second
time, it finds a timestamp file, so it can compare the verification timestamp
of the module with the time when the build started. If the verification
timestamp is too old, the module is verified again, and the timestamp file is
updated.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@201224 91177308-0d34-0410-b5e6-96231b3b80d8
This patch improves the support for picking Multilibs from gcc installations.
It also provides a better approximation for the flags '-print-multi-directory'
and '-print-multi-lib'.
This reverts r201203 (i.e. re-applying r201202 with small fixes in
unittests/CMakeLists.txtto make the build bots happy).
review: http://llvm-reviews.chandlerc.com/D2538
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@201205 91177308-0d34-0410-b5e6-96231b3b80d8
hack of passing -fconst-strings to -cc1"
Passing or not a language option based on diagnostic settings is a bad idea, it breaks
using a PCH that was compiled with different diagnostic settings.
Also add a test case to make sure we don't regress.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@200964 91177308-0d34-0410-b5e6-96231b3b80d8
Use the verify hook rather than the compile hook to represent the
-verify-pch action, and move the exising --verify-debug-info action
into its own subclass of VerifyJobAction. Incidentally change the name
printed by -ccc-print-phases for --verify-debug-info.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@200938 91177308-0d34-0410-b5e6-96231b3b80d8
This option will:
- load the given pch file
- verify it is not out of date by stat'ing dependencies, and
- return 0 on success and non-zero on error
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@200884 91177308-0d34-0410-b5e6-96231b3b80d8
These changes make it impossible to use ARM EHABI for stack unwinding without
enabling exceptions and break AddressSanitizer on ARM.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@200627 91177308-0d34-0410-b5e6-96231b3b80d8
This is a simpler rule, broadly in line with previous Darwin (which chose
between "soft" and "softfp") but probably safer. In practice the only real
reason for "softfp" is ABI compatibility, not usually an issue on limited chips
like these, so anyone who wanted hard-float should already be saying so.
That's my story and I'm sticking to it.
rdar://problem/15887493
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@199896 91177308-0d34-0410-b5e6-96231b3b80d8
Using backend-option like a few other debug codegen flags. I believe
Eric Christopher's working at porting those over to something nicer
such as an API level CodeGenOptions or the like, so this can be
improved along with that work.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@199535 91177308-0d34-0410-b5e6-96231b3b80d8
Previously we had bodged together some hacks mapping MachO embedded
targets (i.e. mainly ARM v6M and v7M) to the "*-*-darwin-eabi" triple.
This is incorrect in both details (they don't run Darwin and they're
not EABI in any real sense).
This commit appropriates the existing "MachO" environment for the
purpose instead.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@199367 91177308-0d34-0410-b5e6-96231b3b80d8
This makes the C++ ABI depend entirely on the target: MS ABI for -win32 triples,
Itanium otherwise. It's no longer possible to do weird combinations.
To be able to run a test with a specific ABI without constraining it to a
specific triple, new substitutions are added to lit: %itanium_abi_triple and
%ms_abi_triple can be used to get the current target triple adjusted to the
desired ABI. For example, if the test suite is running with the i686-pc-win32
target, %itanium_abi_triple will expand to i686-pc-mingw32.
Differential Revision: http://llvm-reviews.chandlerc.com/D2545
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@199250 91177308-0d34-0410-b5e6-96231b3b80d8
encodes the canonical rules for LLVM's style. I noticed this had drifted
quite a bit when cleaning up LLVM, so wanted to clean up Clang as well.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@198686 91177308-0d34-0410-b5e6-96231b3b80d8
It controls everything that -flimit-debug-info used to, plus the
vtable type optimization. The old -fno-limit-debug-info option is now an
alias to -fstandalone-debug and vice versa.
Standalone is the default on Darwin until dtrace is updated to work with
non-standalone debug info (rdar://problem/15758808).
Note: I kept the LimitedDebugInfo name in CodeGenOptions::DebugInfoKind
because NoStandaloneDebugInfo sounded even more confusing.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@198655 91177308-0d34-0410-b5e6-96231b3b80d8
getARMCPU and getLLVMArchSuffixForARM existed as very similar functions
in both ToolChain.cpp and Tools.cpp. Create a single implementation of
each in Tools.cpp, eliminate the duplicate and share via Tools.h.
Creates an 'arm' namespace in Tools.h to be used by any ARM-targetting tools.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@197153 91177308-0d34-0410-b5e6-96231b3b80d8
Passing -mthumb with no explicit CPU on the command line
resulted in target CPU changing from the architecture
default to arm7tdmi. Now it does not.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@197151 91177308-0d34-0410-b5e6-96231b3b80d8
This refactors some of the Darwin toolchain classification to give a more solid
distinction between the three primary Darwin platforms (OS X, IOS and IOS
simulator) so that a 4th choice can be added temporarily: embedded MachO
targets.
Longer term, this support will be factored out into a separate class and no
longer classified as "darwin-eabi", but the refactoring should still be useful.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@197148 91177308-0d34-0410-b5e6-96231b3b80d8
- krait processor currently modeled with the same features as A9.
- Krait processor additionally has VFP4 (fused multiply add/sub)
and hardware division features enabled.
- krait has currently the same Schedule model as A9
- krait cpu flag is not recognized by the GNU assembler yet,
it is replaced with march=armv7-a to avoid a lower march
from being used.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@196618 91177308-0d34-0410-b5e6-96231b3b80d8
I happened to notice this while trying to write a test for an iOS simulator
target. I suspect we just missed this when we added separate "macosx" and "ios"
triples instead of the generic "darwin" OS.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@196527 91177308-0d34-0410-b5e6-96231b3b80d8
AddGoldPlugin() is the same on gnutools and freebsd so avoid the
copy-and-paste duplicate code.
No change in functionality.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@196096 91177308-0d34-0410-b5e6-96231b3b80d8
Clang still has support for running gcc for performing various stages
of a build. Right now it looks like this is used for
* Supporting Fortran in the clang driver
* Running an assembler or linker in systems we don't yet know how to
run them directly.
It looks like the gcc::Precompile is a vestige from the days when we
supported using clang for C and running gcc for c++. This patch removes it
(yes, we have no tests for it).
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@195586 91177308-0d34-0410-b5e6-96231b3b80d8
Clang knows how to use the gnu assembler directly from doing so on linux and
hurd. The existing support worked out of the box on cygwin and mingw and I was
able to bootstrap clang with it in both systems (with pending patches for the
new mingw abi, but that is independent of the assembler).
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@195554 91177308-0d34-0410-b5e6-96231b3b80d8
We are still using Dwarf Version 2 for Darwin systems, make it consistent
with -gline-tables-only.
This should fix an internal buildbot.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@195267 91177308-0d34-0410-b5e6-96231b3b80d8
should be isolated in the backend (r195123). From the frontend point
of view in case of "-mhard-float -mips16" combination of flags the float
ABI mode should remain unchanged.
The patch reviewed by Reed Kotler.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@195124 91177308-0d34-0410-b5e6-96231b3b80d8
the -Q flag to the as(1) assembler driver.
We will soon be switching the darwin as(1) assembler driver to call clang(1)
and use the intergated assembler by default. To do this and still support
clang(1)'s -no-integrated-as flag, when clang(1) runs the as(1) assembler
driver and -no-integrated-as is used it needs to pass the -Q flag to as(1)
so it uses its GNU based assembler, and not turn around and call clag(1)'s
integrated assembler.
rdar://15495921
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@195054 91177308-0d34-0410-b5e6-96231b3b80d8
This adds -freroll-loops (and -fno-reroll-loops in the usual way) to enable
loop rerolling as part of the optimization pass manager. This transformation
can enable vectorization, reduce code size (or both).
Briefly, loop rerolling can transform a loop like this:
for (int i = 0; i < 3200; i += 5) {
a[i] += alpha * b[i];
a[i + 1] += alpha * b[i + 1];
a[i + 2] += alpha * b[i + 2];
a[i + 3] += alpha * b[i + 3];
a[i + 4] += alpha * b[i + 4];
}
into this:
for (int i = 0; i < 3200; ++i) {
a[i] += alpha * b[i];
}
Loop rerolling is currently disabled by default at all optimization levels.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@194967 91177308-0d34-0410-b5e6-96231b3b80d8
Teach the '-arch' command line option to enable the compiler-friendly
features of core-avx2 CPUs on Darwin. Pass the information along in the
target triple like Darwin+ARM does.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@194907 91177308-0d34-0410-b5e6-96231b3b80d8
Trying to fix test failures since earlier today.
One of the tests added in this commit is outputting test/Driver/clang_f_opts.s
which the builders that build in-tree (eg. clang-native-arm-cortex-a9) are
trying to run as a test case, causing failures.
clang_f_opts.c:
If -### doesn't emit the warning then this test probably shouldn't be in
here in the first place. Frontend maybe?
invalid-o-level.c:
Running %clang_cc1 in the Driver tests doesn't make sense because -cc1
bypasses the driver. (I'm not reverting the commit that introduced this but
please fix instead of keeping it this way.)
Reverting to fix the build failures and also so that the tests can be thought
out more thoroughly.
This reverts commit r194817.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@194845 91177308-0d34-0410-b5e6-96231b3b80d8
This patch adds -I to the arguments that are passed to the
integrated assembler from -Wa, and -Xassembler args.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@194743 91177308-0d34-0410-b5e6-96231b3b80d8
This options accepts a path to a directory, collects the filenames of the files
it contains, and the migrator will only modify files with the same filename.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@194710 91177308-0d34-0410-b5e6-96231b3b80d8
This adds a new option -fprofile-sample-use=filename to Clang. It
tells the driver to schedule the SampleProfileLoader pass and passes
on the name of the profile file to use.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@194567 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Currently with clang:
$ clang -O20 foo.c
error: invalid value '20' in '-O20'
With the patch:
$ clang -O20 foo.c
warning: invalid value '20' in '-O20'. Fall back on value '3'
Reviewers: rengolin, hfinkel
Reviewed By: rengolin
CC: cfe-commits, hfinkel, rengolin
Differential Revision: http://llvm-reviews.chandlerc.com/D2125
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@194403 91177308-0d34-0410-b5e6-96231b3b80d8
hack of passing -fconst-strings to -cc1, but at least the driver uses
the regular warning checking code path.
Since we don't support a warning that is DefaultIgnore in one language
but not in another, this patch creates a dummy C only warning in the same
group as the existing one to get the desired effect.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@194097 91177308-0d34-0410-b5e6-96231b3b80d8
The thread, memory, dataflow and function sanitizers are now diagnosed if
enabled explicitly on an unsupported platform. Unsupported sanitizers which
are enabled implicitly (as part of a larger group) are silently disabled. As a
side effect, this makes SanitizerArgs parsing toolchain-dependent (and thus
essentially reverts r188058), and moves SanitizerArgs ownership to ToolChain.
Differential Revision: http://llvm-reviews.chandlerc.com/D1990
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@193875 91177308-0d34-0410-b5e6-96231b3b80d8
Enables the clang driver to begin targeting specific CPUs. Introduced a
"generic" CPU which will ensure that the optional FP feature is enabled
by default when it gets to LLVM, without needing any extra arguments.
Cortex-A53 and A-57 are also introduced with tests, although backend
handling of them does not yet exist.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@193740 91177308-0d34-0410-b5e6-96231b3b80d8
Although we wire up a bit for v8fp for macro setting
purposes, we don't set a macro yet. Need to ask list
about that.
Change-Id: Ic9819593ce00882fbec72757ffccc6f0b18160a0
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@193367 91177308-0d34-0410-b5e6-96231b3b80d8
Adds some Cortex-A53 strings where they were missing before.
Cortex-A57 is entirely new to clang.
Doesn't touch code only used by Darwin, in consequence of which
one of the A53 lines has been removed.
Change-Id: I5edb58f6eae93947334787e26a8772c736de6483
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@193364 91177308-0d34-0410-b5e6-96231b3b80d8
pick up the common bits ubsan actually needs. Also remove whole-archive when we
aren't trying to re-export the symbols.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@193022 91177308-0d34-0410-b5e6-96231b3b80d8
These options specify 64-bit FP registers and 32-bit FP registers respectively.
When using -mfp32, the FPU has 16x double-precision registers overlapping with
the 32x single-precision registers (each double-precision register overlaps
two single-precision registers).
When using -mfp64, the FPU has 32x double-precision registers overlapping with
the 32x single-precision registers (each double-precision register overlaps
with one single-precision register and has an additional 32-bits).
MSA requires -mfp64.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@192899 91177308-0d34-0410-b5e6-96231b3b80d8
Use -no-struct-path-tbaa to turn it off.
This is the same as r191695, which was reverted because it depends on a
commit that has issues.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@192497 91177308-0d34-0410-b5e6-96231b3b80d8
We want the modern behavior most of the time, so inverting the option simplifies
the driver and the tests.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@191551 91177308-0d34-0410-b5e6-96231b3b80d8
We were previously mostly passing it through, but -O0 and -O3 are not valid
options to cl.exe.
We should translate -O0 to /Od and -O3 to /Ox. -O{1,2,s} get passed through.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@191323 91177308-0d34-0410-b5e6-96231b3b80d8
Review: http://llvm-reviews.chandlerc.com/D1546.
I have picked up this patch form Lawrence
(http://llvm-reviews.chandlerc.com/D1063) and did a few changes.
From the original change description (updated as appropriate):
This patch adds a check that ensures that modules only use modules they
have so declared. To this end, it adds a statement on intended module
use to the module.map grammar:
use module-id
A module can then only use headers from other modules if it 'uses' them.
This enforcement is off by default, but may be turned on with the new
option -fmodules-decluse.
When enforcing the module semantics, we also need to consider a source
file part of a module. This is achieved with a compiler option
-fmodule-name=<module-id>.
The compiler at present only applies restrictions to the module directly
being built.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@191283 91177308-0d34-0410-b5e6-96231b3b80d8
gcc doesn't support "gcc -m sse" and this was not tested in clang and only
used for link argument on darwin, so this was very likely just a bug.
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@191251 91177308-0d34-0410-b5e6-96231b3b80d8
This solves two problems:
1) MSBuild will not flag the build as unsuccessful just because we print
an error in the output, since "error(clang):" doesn't seem to match
the regex it's using.
2) It becomes more clear that the diagnostic is coming from clang as
supposed to cl.exe.
Differential Revision: http://llvm-reviews.chandlerc.com/D1735
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@191250 91177308-0d34-0410-b5e6-96231b3b80d8
Duncan P. N. Exon Smith