>
> probably cygwin maintains their own set of binutils patches...? they
> need rather special stuff anyway, i guess. i don't think that it will be
> easy to get this building out of the box. ask somewhere in the cygwin
> realm, whether it's possible to obtain patches and/or build without them...
>

Send a posting to cygwin list.

Al

Somebody did already start to strip this stuff, but did not go to end
for some reasons:

work/binutils-2.20.51.0.8/bfd/ChangeLog-9495: * som.c: Don't include
<sys/dir.h> or <sys/user.h>.
work/binutils-2.20.51.0.8/bfd/hppabsd-core.c:#include <sys/user.h> /*
After a.out.h */
work/binutils-2.20.51.0.8/bfd/osf-core.c:#include <sys/user.h>
work/binutils-2.20.51.0.8/bfd/rs6000-core.c:#include <sys/user.h>
work/binutils-2.20.51.0.8/bfd/sco5-core.c:#include <sys/user.h> /*
After a.out.h */
work/binutils-2.20.51.0.8/bfd/trad-core.c:#include <sys/user.h> /*
After a.out.h */

Al

2010/8/26 Al <oss.elmar@googlemail.com>:
> Somebody did already start to strip this stuff, but did not go to end
> for some reasons:
>


Oh, that was in 1994. :-)

Wed May 18 08:29:04 1994 Ian Lance Taylor (ian@cygnus.com)

* som.c: Don't include <sys/dir.h> or <sys/user.h>.
(som_reloc_queue_find): Call memcmp instead of bcmp.
(som_bfd_reloc_type_lookup): Change first argument to bfd *.
(compare_syms): Change types of arguments to const void *.
(bfd_section_from_som_symbol): Removed unused local found.
(som_write_armap): Add elength, map, orl_count and int arguments.
(som_write_armap): Use %ld and cast to long for getuid result.

> Does it work when you just drop including sys/user.h in binutils?
>
> That did work for hpux11.31 some time ago, where sys/user.h was dropped
> (did exist in hpux11.23):
> http://sourceware.org/ml/binutils/2008-11/msg00310.html

The solution was to set the matching CHOST. Obviously it is used as
"host" setting for "configure", but I didn't find that documentet. In
my case it is "i686-pc-cygwin" as Fabian already pointed out in a
differend thread.

By using "i686-pc-cygwin" the trad-core module is not created at all.
It is optional and heavily depends on <user.h>.

I also find settings for hppa*hpux* in binutils/configure. Maybe it
would be the clean solution to bring in your patches here in a similar
way.


*-*-elf* | *-*-sysv4* | *-*-unixware* | *-*-eabi* | hppa*64*-*-hpux*
hppa*-hp-hpux11*)
hppa*-hp-hpux10*)
hppa*-hp-hpux* | hppa*-*-hiux*)
hppa*64*-*-hpux11*)


Al

On 08/27/2010 10:30 AM, Al wrote:
>> Does it work when you just drop including sys/user.h in binutils?
>>
>> That did work for hpux11.31 some time ago, where sys/user.h was dropped
>> (did exist in hpux11.23):
>> http://sourceware.org/ml/binutils/2008-11/msg00310.html
>
> The solution was to set the matching CHOST.

In this case I fail to see any relation to the hpux fix.

> Obviously it is used as
> "host" setting for "configure", but I didn't find that documentet.

I'm wondering what exactly you miss in the docs when it is obvious...

> I also find settings for hppa*hpux* in binutils/configure. Maybe it
> would be the clean solution to bring in your patches here in a similar
> way.

As the fix was to use correct CHOST: which patches do you mean here?

/haubi/
--
Michael Haubenwallner
Gentoo on a different level

2010/8/27 Michael Haubenwallner <haubi@gentoo.org>:
>
>
> On 08/27/2010 10:30 AM, Al wrote:
>>> Does it work when you just drop including sys/user.h in binutils?
>>>
>>> That did work for hpux11.31 some time ago, where sys/user.h was dropped
>>> (did exist in hpux11.23):
>>> http://sourceware.org/ml/binutils/2008-11/msg00310.html
>>
>> The solution was to set the matching CHOST.
>
> In this case I fail to see any relation to the hpux fix.

Me neither. I wanted to reach two goals with one posting.

1.) Send my solution into the archives.
2.) Give a hint that there may be an alternative lever in your case.

>
>> Obviously it is used as
>> "host" setting for "configure", but I didn't find that documentet.
>
> I'm wondering what exactly you miss in the docs when it is obvious...

Maybe it seems obvious for you. For me it becomes obvious by bad
experience. Not the way I prefer. ;-)

Al