Fix tcmalloc arm clang build

Fix tcmalloc arm clang build

Disable clang's internal assembler when building tcmalloc
for arm.  A bug in the internal assembler was causing the build
fail even though the assembly code is valid.

This change reverts the previous fix 
(https://codereview.chromium.org/2515503002) which was to
disable thumb when building tcmalloc.  Disabling thumb in
this way didn't work for chromeos where the compilers default
is thumb.

BUG=564059

[Sam Clegg, 2016-11-22 18:07:13]

Description was changed from

==========
Fix tcmalloc arm clang build

Disable clang's internal assembler when building tcmalloc
for arm.  A bug in the internal assembler was causing the build
fail even though that assmemly code is valid.

This change reverts the previous fix which was to disable
thumb when building tcmalloc.  Disabling thumb in this way
didn't work for chromeos where the compilers default is
thumb.

BUG=564059
==========

to

==========
Fix tcmalloc arm clang build

Disable clang's internal assembler when building tcmalloc
for arm.  A bug in the internal assembler was causing the build
fail even though the assembly code is valid.

This change reverts the previous fix which was to disable
thumb when building tcmalloc.  Disabling thumb in this way
didn't work for chromeos where the compilers default is
thumb.

BUG=564059
==========

[Sam Clegg, 2016-11-22 18:08:00]

Description was changed from

==========
Fix tcmalloc arm clang build

Disable clang's internal assembler when building tcmalloc
for arm.  A bug in the internal assembler was causing the build
fail even though the assembly code is valid.

This change reverts the previous fix which was to disable
thumb when building tcmalloc.  Disabling thumb in this way
didn't work for chromeos where the compilers default is
thumb.

BUG=564059
==========

to

==========
Fix tcmalloc arm clang build

Disable clang's internal assembler when building tcmalloc
for arm.  A bug in the internal assembler was causing the build
fail even though the assembly code is valid.

This change reverts the previous fix 
(https://codereview.chromium.org/2515503002) which was to
disable thumb when building tcmalloc.  Disabling thumb in
this way didn't work for chromeos where the compilers default
is thumb.

BUG=564059
==========

[Sam Clegg, 2016-11-22 19:08:20]

sbc@chromium.org changed reviewers:
+ brettw@chromium.org, dpranke@chromium.org, llozano@chromium.org

[Sam Clegg, 2016-11-22 19:10:27]

Hmm, this seems to have issues on goma:

GOMA:clang++ compiling
../../third_party/tcmalloc/chromium/src/raw_printer.cc:*ERROR*:
compiler_proxy:3536ms: No matching subprograms found in server

I guess arm-linux-gnueabihf-as is not part of the goma clang goma toolchain.

[llozano, 2016-11-22 19:42:53]

https://codereview.chromium.org/2518423002/diff/40001/base/allocator/BUILD.gn
File base/allocator/BUILD.gn (right):

https://codereview.chromium.org/2518423002/diff/40001/base/allocator/BUILD.gn...
base/allocator/BUILD.gn:79: cflags += [ "-fno-integrated-as" ]
I thought the option is -no-integrated-as but this also works for some reason.
I think it is better to just use -no-integrated-as. That is how I find it in the
documentation...

[llozano, 2016-11-22 19:47:35]

On 2016/11/22 19:10:27, Sam Clegg wrote:
> Hmm, this seems to have issues on goma:
> 
> GOMA:clang++ compiling
> ../../third_party/tcmalloc/chromium/src/raw_printer.cc:*ERROR*:
> compiler_proxy:3536ms: No matching subprograms found in server
> 
> I guess arm-linux-gnueabihf-as is not part of the goma clang goma toolchain.

I dont know anything about this..
Probably ask Nico or Hans

[llozano, 2016-11-22 19:48:08]

On 2016/11/22 19:47:35, llozano wrote:
> On 2016/11/22 19:10:27, Sam Clegg wrote:
> > Hmm, this seems to have issues on goma:
> > 
> > GOMA:clang++ compiling
> > ../../third_party/tcmalloc/chromium/src/raw_printer.cc:*ERROR*:
> > compiler_proxy:3536ms: No matching subprograms found in server
> > 
> > I guess arm-linux-gnueabihf-as is not part of the goma clang goma toolchain.
> 
> I dont know anything about this..
> Probably ask Nico or Hans

and if this is too complicated, then lets use the other route.

[Sam Clegg, 2016-11-22 19:59:53]

On 2016/11/22 19:48:08, llozano wrote:
> On 2016/11/22 19:47:35, llozano wrote:
> > On 2016/11/22 19:10:27, Sam Clegg wrote:
> > > Hmm, this seems to have issues on goma:
> > > 
> > > GOMA:clang++ compiling
> > > ../../third_party/tcmalloc/chromium/src/raw_printer.cc:*ERROR*:
> > > compiler_proxy:3536ms: No matching subprograms found in server
> > > 
> > > I guess arm-linux-gnueabihf-as is not part of the goma clang goma
toolchain.
> > 
> > I dont know anything about this..
> > Probably ask Nico or Hans
> 
> and if this is too complicated, then lets use the other route.

OK closing in favour of https://codereview.chromium.org/2522973002