Re: Fix sysdeps/unix/sysv/linux/arm/libc-do-syscall.S warning

[Mike Frysinger <vapier at gentoo dot org>]

On 05 Mar 2015 17:22, Carlos O'Donell wrote:
> On 03/05/2015 05:14 PM, Joseph Myers wrote:
> > On Thu, 5 Mar 2015, Carlos O'Donell wrote:
> >>>> should we update features.h to avoid that define when __ASSEMBLER__ is defined ?
> >>>> seems like this warning can impact external code too.
> >>>
> >>> I don't know.  To what extent do we support users including glibc header 
> >>> files from assembler code?
> >>
> >> To the extent that it builds glibc without warnings?
> > 
> > That answer bears no relation to my intended question.  Users = users of 
> > glibc including headers in non-glibc applications, built outside of the 
> > glibc build process.  Such uses are completely independent of whether 
> > glibc itself builds without warnings.
> 
> If compiling libc-do-syscall.S generates a warning, we should fix it so it
> doesn't generate a warning. Any fixes beyond those which fix the warning
> are superfluous.
> 
> We should not do any active work to support including glibc header files
> from assembly unless we have specific requests from users for specific
> use cases.
> 
> Does that answer your question?

not exactly ...

Joseph fixed the warning in this one .S file by undefining the symbol.
since the warning originated in features.h and can be triggered by any
.S file (in glibc or non-glibc code), i suggested maybe we change the
features.h header instead so we never have to worry about it again.  i
don't see that as superfluous ... more fixing it in the right place.

to the larger question Joseph and you raise, i don't have an answer for
that.  i don't think we've even thought about it beyond specific requests.
-mike

Attachment: signature.asc
Description: Digital signature