Mailing List Archive

Re: uclibc ebuild (also cvs)
On Wed, 1 Sep 2004, Mike Frysinger wrote:

> > -the socket patch is also needed, else redefines in ssp.c
> > __attribute_used__ part of the frandom patch is not needed (defined in
> > cdefs.h already)
>
> this needs to be handled differently ... like i said earlier, ifdef-ing it out
> causes build failure in userland apps because that header wasnt included ...
> could you post the build failure and how you got it ? (what app, what build
> options, etc...)

Are we speaking here of the socket patch?

+#ifndef _LIBC
#include <asm/socket.h>
+#endif

Than you misunderstand me, the patch is needed to build ssp.c itself
part of uClibc (UCLIBC_PROPOLICE enabled) w/o warnings of redefines,
nothing else does have problems, the userland apps are not influenced by
this patch, because none will/should (hopefully) define _LIBC

If though we are speaking of the frandom patch, then I do not understand
what's not clear.

> > -ppc is also a pie candidate.
>
> i tried a buildroot real quick on a ppc and iirc, i had failures in basic apps
> just running :)
> i'll try to check it out again

Which gcc? Only gcc-3.4.x would work correctly w/ -pie on ppc.

>
> > -PROPOLICE_BLOCK_SEGV is better for use debug
>
> i'm wondering if we could also move the PTHREAD debug support to `use debug`

I am using it like this already ;) (and had it in the originally proposed
uclibc ebuild, someone removed it, :-( )

Peter

--
Peter S. Mazinger <ps dot m at gmx dot net> ID: 0xA5F059F2
Key fingerprint = 92A4 31E1 56BC 3D5A 2D08 BB6E C389 975E A5F0 59F2


____________________________________________________________________
Miert fizetsz az internetert? Korlatlan, ingyenes internet hozzaferes a FreeStarttol.
Probald ki most! http://www.freestart.hu

--
gentoo-embedded@gentoo.org mailing list
Re: uclibc ebuild (also cvs) [ In reply to ]
On Monday 06 September 2004 07:20 am, Peter S. Mazinger wrote:
> Are we speaking here of the socket patch?
>
> +#ifndef _LIBC
> #include <asm/socket.h>
> +#endif

yes

> Than you misunderstand me, the patch is needed to build ssp.c itself
> part of uClibc (UCLIBC_PROPOLICE enabled) w/o warnings of redefines,
> nothing else does have problems, the userland apps are not influenced by
> this patch, because none will/should (hopefully) define _LIBC

i forget exactly what bombed but lemme see if i can dig up the package which
had the problem ...

all i do remember is that using that patch caused things to fail and taking it
out caused things to work :)
-mike

--
gentoo-embedded@gentoo.org mailing list