[Pkg-alsa-devel] alsa-lib ftbfs

[Jordi Mallach]

--gKMricLos+KVdGMg
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Thomas, this is what vorlon said about it.

23:39 < vorlon> ok, the problem is in alsa-symbols.h
23:39 < vorlon> at the very bottom; use_default_symbol_version() is defined=
 to
                a chunk of inlined assembly which doesn't appear to be
                portable. :P
23:52  * jordi checks how recently that changed.
23:53 < jordi> date: 2003/08/13 13:18:36;  author: tiwai;  state: Exp;  lin=
es:
               +19 -2
23:53 < jordi> added workaround for ppc64.
23:54 < jordi> hmm.
23:54 < vorlon> right, there's a separate set of #defines for ppc64, but th=
ey
                don't work for alpha. :)
23:54 < jordi> I mean, hasn't changed since 2003
23:55 < vorlon> how about the src/control/control.c file?
23:55 < jordi> let's see
23:56 < jordi> date: 2004/04/20 13:42:48;  author: perex;  state: Exp;  lin=
es:
               +1 -1
23:56 < jordi> Fixed bugs detected by Intel compiler
23:56  * jordi looks at the whole dir.
23:56 < vorlon> it may also be that USE_VERSIONED_SYMBOLS was always being =
set
                before, even when not building PIC code...
23:57 < jordi> no changes since March.
23:57 < jordi> err, april
23:57  * jordi has a look at the whole alsa-lib
23:57 < vorlon> looks like that may be the ticket -- if I add
                -DUSE_VERSIONED_SYMBOLS, the damn thing compiles w/o errors.
23:58 < jordi> hmm.
23:58 < vorlon> I don't know if the static lib generated that way is *usabl=
e*;
                I have no idea what the effect of using .symver is in a sta=
tic
                lib.
Day changed to 25 Sep 2004
00:01 < jordi> cvs changelogs don't give any hint...
00:17 < vorlon> jordi: the installed version of libasound.a here does indeed
                have snd_ctl_elem_info_get_dimension@@ALSA_0.9.3 and
                snd_ctl_elem_info_get_dimensions@@ALSA_0.9.3
[...]
00:39 < vorlon> and I've confirmed that the old static lib was usable.
00:40 < jordi> vorlon: do you know who I might ask to get the assembler thi=
ng
               fixed?
00:40 < vorlon> I've already asked p2-mate to look at it.
00:40 < jordi> ah, thanks
00:41 < vorlon> if he isn't able to come up with anything, we can hack
                something together to always use .symver on alpha.
00:41 < jordi> Why do you think it might be breaking now, if this hasn't
               changed in a while?
00:41 < vorlon> because *something* changed which bit of assembly is being
                included; the answer probably lies in a header file or in t=
he
                build scripts.
00:41 < jordi> I don't see changes to the USE_VERSIONED_SYMBOLS stuff eithe=
r.
00:42 < vorlon> Right, I think it's that somehow PIC was being defined befo=
re.

Let's see what p2 says.

Jordi
--=20
Jordi Mallach P=E9rez  --  Debian developer     http://www.debian.org/
jordi@sindominio.net     jordi@debian.org     http://www.sindominio.net/
GnuPG public key information available at http://oskuro.net/~jordi/

--gKMricLos+KVdGMg
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: Digital signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)

iD8DBQFBVJOfJYSUupF6Il4RAtukAJ9cQZYhnUkK17xLg91v5h2EXEIauQCdHzBw
9Bcg4tIQmxfxMYXj9wOAxdo=
=JOa3
-----END PGP SIGNATURE-----

--gKMricLos+KVdGMg--