Fix for x64 Android. __unused is #defined elsewhere.

Fix for x64. __unused is #defined elsewhere.
Recent versions of NDK do not define __off64_t, changed to int64_t.

BUG=346626
R=mseaborn@chromium.org

Committed: 25

[Anton, 2014-04-01 12:42:18]

Hi,
This fix is needed for compilation of Android for x64.
It looks like try bots don't work here. I have done
a local build for Linux. Note I am not a committer so 
will need you to land the patch, assuming that there 
is not CQ here.
Anton

[Mark Seaborn, 2014-04-02 16:20:28]

https://codereview.chromium.org/219483003/diff/1/linux_syscall_support.h
File linux_syscall_support.h (right):

https://codereview.chromium.org/219483003/diff/1/linux_syscall_support.h#newc...
linux_syscall_support.h:2933: off64_t o) {
You didn't comment this part in the commit message.  Presumably Bionic doesn't
define __off64_t?  If I make this change, though, I get:

$ gcc -c linux_syscall_support.h
linux_syscall_support.h:2933:37: error: unknown type name ‘off64_t’

[Anton, 2014-04-02 16:29:51]

On 2014/04/02 16:20:28, Mark Seaborn wrote:
> https://codereview.chromium.org/219483003/diff/1/linux_syscall_support.h
> File linux_syscall_support.h (right):
> 
>
https://codereview.chromium.org/219483003/diff/1/linux_syscall_support.h#newc...
> linux_syscall_support.h:2933: off64_t o) {
> You didn't comment this part in the commit message.  Presumably Bionic doesn't
> define __off64_t?  If I make this change, though, I get:
> 
> $ gcc -c linux_syscall_support.h
> linux_syscall_support.h:2933:37: error: unknown type name ‘off64_t’

Is that a relevant test? compiler is never invoked this way by the build.

[Mark Seaborn, 2014-04-02 17:19:02]

On 2 April 2014 09:29, <anton@chromium.org> wrote:

> On 2014/04/02 16:20:28, Mark Seaborn wrote:
>
>> https://codereview.chromium.org/219483003/diff/1/linux_syscall_support.h
>> File linux_syscall_support.h (right):
>>
>
>
> https://codereview.chromium.org/219483003/diff/1/linux_
> syscall_support.h#newcode2933
>
>> linux_syscall_support.h:2933: off64_t o) {
>> You didn't comment this part in the commit message.  Presumably Bionic
>> doesn't
>> define __off64_t?  If I make this change, though, I get:
>>
>
>  $ gcc -c linux_syscall_support.h
>> linux_syscall_support.h:2933:37: error: unknown type name 'off64_t'
>>
>
> Is that a relevant test? compiler is never invoked this way by the build.


That was just a way of testing #includability.  Equivalently:

$ gcc -include linux_syscall_support.h -x c - </dev/null -c
In file included from <command-line>:0:0:
./linux_syscall_support.h:2934:37: error: unknown type name ‘off64_t’

And I get the same if I do "gcc -c test.c" on a C file that #includes
linux_syscall_support.h.

Mark

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Anton, 2014-04-02 17:25:00]

On 2014/04/02 17:19:02, Mark Seaborn wrote:
> On 2 April 2014 09:29, <mailto:anton@chromium.org> wrote:
> 
> > On 2014/04/02 16:20:28, Mark Seaborn wrote:
> >
> >> https://codereview.chromium.org/219483003/diff/1/linux_syscall_support.h
> >> File linux_syscall_support.h (right):
> >>
> >
> >
> > https://codereview.chromium.org/219483003/diff/1/linux_
> > syscall_support.h#newcode2933
> >
> >> linux_syscall_support.h:2933: off64_t o) {
> >> You didn't comment this part in the commit message.  Presumably Bionic
> >> doesn't
> >> define __off64_t?  If I make this change, though, I get:
> >>
> >
> >  $ gcc -c linux_syscall_support.h
> >> linux_syscall_support.h:2933:37: error: unknown type name 'off64_t'
> >>
> >
> > Is that a relevant test? compiler is never invoked this way by the build.
> 
> 
> That was just a way of testing #includability.  Equivalently:
> 
> $ gcc -include linux_syscall_support.h -x c - </dev/null -c
> In file included from <command-line>:0:0:
> ./linux_syscall_support.h:2934:37: error: unknown type name ‘off64_t’
> 
> And I get the same if I do "gcc -c test.c" on a C file that #includes
> linux_syscall_support.h.
> 
> Mark
> 
> To unsubscribe from this group and stop receiving emails from it, send an
email
> to mailto:chromium-reviews+unsubscribe@chromium.org.

PTAL I have upload patch #2 to address this. Conditional is unavoidable
as there is no overlap remaining.

[Mark Seaborn, 2014-04-02 17:27:37]

https://codereview.chromium.org/219483003/diff/20001/linux_syscall_support.h
File linux_syscall_support.h (right):

https://codereview.chromium.org/219483003/diff/20001/linux_syscall_support.h#...
linux_syscall_support.h:2936: off64_t o) {
How about just using int64_t here?

[Anton, 2014-04-02 17:43:35]

On 2014/04/02 17:27:37, Mark Seaborn wrote:
> https://codereview.chromium.org/219483003/diff/20001/linux_syscall_support.h
> File linux_syscall_support.h (right):
> 
>
https://codereview.chromium.org/219483003/diff/20001/linux_syscall_support.h#...
> linux_syscall_support.h:2936: off64_t o) {
> How about just using int64_t here?

In the NDK off64_t is a long long where as int64_t is a long. This may not
matter, but it will depend on the calling code. Since it is converted to
uint_64_t (@L2942) seems like int64_t is ok for local correctness. But I need to
recompile everything to make sure it doesn't break any caller.

[Anton, 2014-04-02 18:26:24]

On 2014/04/02 17:43:35, Anton wrote:
> On 2014/04/02 17:27:37, Mark Seaborn wrote:
> > https://codereview.chromium.org/219483003/diff/20001/linux_syscall_support.h
> > File linux_syscall_support.h (right):
> > 
> >
>
https://codereview.chromium.org/219483003/diff/20001/linux_syscall_support.h#...
> > linux_syscall_support.h:2936: off64_t o) {
> > How about just using int64_t here?
> 
> In the NDK off64_t is a long long where as int64_t is a long. This may not
> matter, but it will depend on the calling code. Since it is converted to
> uint_64_t (@L2942) seems like int64_t is ok for local correctness. But I need
to
> recompile everything to make sure it doesn't break any caller.

Done. PTAL.

[Mark Seaborn, 2014-04-03 14:44:41]

LGTM.  I've given you commit rights, so you can commit this.

Can you update lss_revision in Chromium's DEPS after committing this, please?

[Anton, 2014-04-04 11:20:54]

Committed patchset #3 manually as r25 (presubmit successful).