Re: [RFA:] Make ix86-pc-linux-gnulibc1 buildable [fixinc fix included]

[Hans-Peter Nilsson <hp at bitrange dot com>]

On Thu, 24 Oct 2002, Joseph S. Myers wrote:
> On Thu, 24 Oct 2002, Hans-Peter Nilsson wrote:
>
> > 	* config/i386/linux.h: Move MD_FALLBACK_FRAME_STATE_FOR inside
> > 	ifndef IN_LIBGCC2.  Wrap it together with signal.h and
> > 	sys/ucontext.h inclusion in ifndef USE_GNULIBC_1.
>
> I think the same lack of <sys/ucontext.h> problem also applies with glibc
> 2.0 (which can be tested with (__GLIBC__ == 2 && __GLIBC_MINOR__ == 0)).
> Is there someone here with a glibc 2.0 system who can test whether this
> patch fixes the problem there?

I guess it should, if the line "#ifndef USE_GNULIBC_1" is changed to
"#if !defined (USE_GNULIBC_1) || (__GLIBC__ == 2 && __GLIBC_MINOR__ == 0)"

> PR other/3337 is an open bug report for the linux-gnulibc1 problem.  If
> i*86-pc-linux-gnulibc1 now bootstrap, PRs other/3338, other/3339 and
> other/3340 are probably fixed.  PR bootstrap/6763 is another report on
> i586-pc-linux-gnulibc1, as is bootstrap/8122.  PR bootstrap/4068 is a bug
> report for the problem with glibc 2.0.7 (Red Hat 5.2).  PR bootstrap/8145
> looks like the same problem, for whatever libc version (glibc 2.0?) "Suse
> linux 6.0" used.
>
> The ChangeLog should list the PRs fixed.

I'll add notes for and close 3337 and duplicates 3338, 3339,
though these PR:s are for 3.0.  Likewise 6763 for 3.1; after all
those release-branches are closed.  I'll add a note for 8122 for
3.2, though I won't close it; the patch should be applied to 3.2
for the PR to be ripe for closing.

3340 seems unrelated and I can't check if your hinted fix solves
4068 and 8145 (though it seems probable) so I'll leave those.

Thanks for reminding me about checking PR:s (doh!).

> (I'm not aware of any user interest in libc1 on non-x86 systems, which
> probably can still be deprecated, as can I expect all variants of a.out
> Linux.)

FWIW, I agree.

brgds, H-P