[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: patches not upstream



Stephen R. Marenka dixit:

>So I'm looking at the patches that are needed in the debian kernel and I
>noticed quite a few haven't been sent upstream.
>
>git format-patch -o ../tmp v2.6.36..m68k-v2.6.36 lists 85 patches.
>
>Some of those are already marked for 2.6.37, but the atari/aranym patches
>seem to have been sitting around a while. Any chance they can go upstream?
>It would simplify debian kernel maintenance immensely.

Definitively. FWIW, I’m building all flavours of the following .dsc at
the moment (the m68k patches have no diff between the kernel I’m running
(linux-image-2.6.32-5+m68k.1-atari_2.6.32-24+m68k.1_m68k.deb) and that,
only the rest is rebased on a proper Debian sid released kernel):
https://www.freewrt.org/~tg/debs68k/notyet/linux-2.6_2.6.32-26+m68k.2.dsc
https://www.freewrt.org/~tg/debs68k/notyet/linux-2.6_2.6.32-26+m68k.2.diff.gz

Stephen: Please do an SVN commit to the Debian repo, removing
bugfix/m68k/fp/0061-Keys-KEYCTL_SESSION_TO_PARENT-needs-TIF_NOTIFY_RESU.patch
and taking over my 0026 and 0052 to make them apply again. As they’re
currently disabled, this should be okay with the other maintainers as
it’s a nop. Please look at the debdiff too:
https://www.freewrt.org/~tg/debs68k/notyet/linux-2.6_2.6.32-26+m68k.2.debdiff
Ignore 26-extra, it was just renamed; config/defines, which makes things
build for me and get rid of the gcc-4.3 dependency in the bootstrap repo
I’m working on at the moment; config/i386/defines, which just made the
non-m68k part build faster (dpkg-buildpackage -A *still* does build the
binary-arch, it only doesn’t package them). The only other relevant one
besides the three bugfix/m68k/fp/ ones above is patches/series/26+m68k.2
which enables them. Once I finished building and can boot into that ker-
nel image, and Ingo can boot into the amiga flavour, we can probably ask
Ben/waldi to allow them in (although waldi said 2.6.32 is of no interest
any longer).

I suppose this also means that the ones listed in patches/series/26+m68k.2
will have been tested by me (or rather, actually are, just in a Debian
kernel, not a vanilla one).

FWIW: I’ve built like three dozen packages in cowbuilder now, satisfying
dependencies from debian-ports.org unstable+unreleased where needed, but
ordering them somewhat to get newer dependencies first, mostly. Another
half dozen needed for python2.6 and perl, but they all run the testsuite
(openssl takes a LOT), and things look good, minus the gcc problems I
wrote about on the mailing list. debs68k/sources.txt if someone wants
to peek, but this is like unstable, and I regularly move packages around
(to denote where a .deb came from: old sid, old unreleased, hand-built,
cowbuilder-built, or _all.deb copied from unstable), so apt-get update
is required relatively often.

bye,
//mirabilos
-- 
I believe no one can invent an algorithm. One just happens to hit upon it
when God enlightens him. Or only God invents algorithms, we merely copy them.
If you don't believe in God, just consider God as Nature if you won't deny
existence.		-- Coywolf Qi Hunt


Reply to: