Re: [BUG] 2.6.25-rc2-git3 broke cdrecord

[James Bottomley <James.Bottomley@HansenPartnership.com>]

On Mon, 2008-03-03 at 11:20 +0100, Mikael Pettersson wrote:
> Mikael Pettersson writes:
>  > Stefan Richter writes:
>  >  > FUJITA Tomonori wrote:
>  >  > > On Mon, 3 Mar 2008 02:48:23 +0100 (CET)
>  >  > > Stefan Richter <stefanr@s5r6.in-berlin.de> wrote:
>  >  > >> CD burning still fails for me via ide-cd, although in a
>  >  > >> different way than without the patch.  (Without the patch, ide-cd also
>  >  > >> fails like described by you for SATA.)
>  >  > >>
>  >  > >> With the patch, the failure is as follows.  I don't know if this is
>  >  > >> related to the data length arithmetic or something entirely different;
>  >  > >> all I learned so far is that 2.6.24 works.  Application log from
>  >  > >> 2.6.25-rc3 + Tejun's patch + Gentoo's cdrkit-1.1.6 + k3b:
>  >  > > 
>  >  > > Can you try the following patch?
>  >  > > 
>  >  > > http://lkml.org/lkml/2008/3/2/218
>  >  > 
>  >  > Fails just like with Tejun's patch, alas.
>  > 
>  > At this point I must confess I feel very uneasy about this situation.
>  > 
>  > Dare I suggest the perhaps hereditary solution of reverting these
>  > changes and doing them right after 2.6.25 final?
> 
> Doh! s/hereditary/heretical/

Unfortunately not.  This patch set was first introduced to fix a panic
in the SAS drivers that support SATAPI because they couldn't cope with
the way libata mangled the sg list.

We know what the problem is, we just have to find an acceptable way of
fixing it.

James