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

Re: About the current binary firmware drivers situation



On Thu, Mar 10, 2005 at 11:10:56PM +0100, Frederik Schueler wrote:
> Hello,
> 
> The general resolutions 2004-003 and 2004-004 in mind, I would like to
> find a consensus among the kernel-team members on how to deal with
> binary-only firmware blobs in the kernel, now for sarge, and for later 
> releases.
> 
> Bugs like #298743 show we have users needing the currently pruned 
> drivers, and I need some of them too myself - both for my private boxes,
> as well as at my workplace. Having the drivers removed without a 
> replacement in non-free, like for example custom buildable module 
> packages and udebs to be loaded at installation time, is a not tolerable 
> situation.
> 
> If you have a closer look at the kernel-sources we distribute, only a 
> few binary firmware drivers are removed at all. The dri modules for 
> radeon and mga, the advansys scsi driver, soundblaster 16 and many more 
> drivers with binary firmware are still in place. Even questionable 
> examples like the intel e100 driver, which applies a small (binary) 
> patch to the firmware at initialization time. 
> 
> In my opinion, this situation is discriminatory against users who want 
> to run Debian on a system containing affected hardware.
> 
> 
> In short, I would like to propose the following:
> 
> 1. we remove the pruning from current sarge kernels, and release kernel
>    sources complete with all drivers as distributed on kernel.org
> 
> 2. we agree on how to handle binary firmware in future releases,
>    in the light of GR 2004-004:

Notice that invoking 2004-004 is problematic, since it only overrides the
cleanup-solution of 2004-03, which applies to non-software in the old
interpretation.

I don't think you can argue that these firmware blobs are not software though,
altough the fact that the 2004-03 GR that 2004-04 overrides is about
'editorial changes' may let us some options to slit in stuff under cover of
service to the users.

In any case i think that keeping mutilated drivers in main is a problem, and
is going to cause us support nightmares in the long run, so i would rather we
don't build those in main at all.

Friendly,

Sven Luther



Reply to: