Thorsten Glaser
2012-09-30 14:41:18 UTC
Hi,
it=E2=80=99s apparently armhf fixing weekend for me (first klibc, then
dietlibc, and funnily, both have bugs in setjmp), so I=E2=80=99d like
to ask, with my Debian Developer hat on of the many I wear,
that you look at the patches we carry in Debian again:
http://anonscm.debian.org/gitweb/?p=3Dcollab-maint/dietlibc.git;a=3Dtree;f=
=3Ddebian/patches;h=3D47f36734c30bc03c01f9f95ee50c884fb6468b08;hb=3DHEAD
These (in the order found in the =E2=80=9Cseries=E2=80=9D file) make a patc=
h
queue against CVS HEAD from 2012-03-25 (in the current tree),
but I think they should mostly apply against HEAD, looking at
https://code.launchpad.net/~mirabilos/dietlibc/MAIN (which I
set up as publicly accessible mirror so we have some sort of
overview on what was committed when considering GMane doesn=E2=80=99t
carry a gmane.linux.lib.dietlibc.cvs list with commitmails).
In addition to those patches, the current Debian build script
still deletes parisc/strstr.S as it=E2=80=99s buggy/broken. (That was
not doable using a diff.) Of course, fixing is preferable, but
I=E2=80=99d rather use the MI version than a broken but optimised MD
version.
This has been tested on most all Debian architectures, but only
statically linked, dynamic linking has been broken for a while
and not available everywhere, so we don=E2=80=99t offer it at the moment.
Most of the _new_ patches should have somewhat proper documentation
in the patch header (and possibly, the git commit adding it), and
IIRC some time ago I wrote what I could about the older patches
that we already carried back then. (Note recent changes, e.g. to
the ARMv7 support to unbreak ARMv6, as used on the Raspberry Pi.)
Thanks in advance,
//mirabilos
--=20
<Natureshadow> Dann mach ich git annex copy --to shore und fertig ist das
<Natureshadow> das ist ja viel cooler als ownCloud ...
<mirabilos> sag ich doch
<Natureshadow> ja wieso stimmt das denn immer was du sagst ...
it=E2=80=99s apparently armhf fixing weekend for me (first klibc, then
dietlibc, and funnily, both have bugs in setjmp), so I=E2=80=99d like
to ask, with my Debian Developer hat on of the many I wear,
that you look at the patches we carry in Debian again:
http://anonscm.debian.org/gitweb/?p=3Dcollab-maint/dietlibc.git;a=3Dtree;f=
=3Ddebian/patches;h=3D47f36734c30bc03c01f9f95ee50c884fb6468b08;hb=3DHEAD
These (in the order found in the =E2=80=9Cseries=E2=80=9D file) make a patc=
h
queue against CVS HEAD from 2012-03-25 (in the current tree),
but I think they should mostly apply against HEAD, looking at
https://code.launchpad.net/~mirabilos/dietlibc/MAIN (which I
set up as publicly accessible mirror so we have some sort of
overview on what was committed when considering GMane doesn=E2=80=99t
carry a gmane.linux.lib.dietlibc.cvs list with commitmails).
In addition to those patches, the current Debian build script
still deletes parisc/strstr.S as it=E2=80=99s buggy/broken. (That was
not doable using a diff.) Of course, fixing is preferable, but
I=E2=80=99d rather use the MI version than a broken but optimised MD
version.
This has been tested on most all Debian architectures, but only
statically linked, dynamic linking has been broken for a while
and not available everywhere, so we don=E2=80=99t offer it at the moment.
Most of the _new_ patches should have somewhat proper documentation
in the patch header (and possibly, the git commit adding it), and
IIRC some time ago I wrote what I could about the older patches
that we already carried back then. (Note recent changes, e.g. to
the ARMv7 support to unbreak ARMv6, as used on the Raspberry Pi.)
Thanks in advance,
//mirabilos
--=20
<Natureshadow> Dann mach ich git annex copy --to shore und fertig ist das
<Natureshadow> das ist ja viel cooler als ownCloud ...
<mirabilos> sag ich doch
<Natureshadow> ja wieso stimmt das denn immer was du sagst ...