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

Re: Minidisk support

On 2009-12-15, Stephen Powell wrote:
> My own attempts to backport the official fix to 2.6.26 have so far been
> unsuccessful due to my minuscule C skills.  I'm getting compile errors
> that I have so far not been able to resolve.  Fortunately, I have a
> simpler patch that works well enough for me.

Well, I finally succeeded in backporting the official DIAG patch from
2.6.33 to 2.6.26.  I was having trouble earlier with substitution parameters
in messages (like %s, %d, etc.).  I finally figured out how that stuff
works well enough to backport the fix to 2.6.26 (i.e. Lenny).  Of course,
it's unofficial.  It doesn't come directly from the kernel people.  But
I have tested it, and it applies cleanly, compiles cleanly, and appears
to execute correctly.  I tested it as well as I could on my system, and
it works great.  Here is a link to the backported patch, if anyone wants it:


For reference again, here is the upstream commit link, courtesy of
Peter Oberparleiter:


P.S. I wonder if we could make a case for including this fix in Lenny
on "security" grounds?  The rationale would be that without this patch,
sharing minidisks with the DIAG driver requires multiple simultaneous
read/write links (access mode MW), which carries with it the risk that
two different servers might accidentally mount the file system read/write
at the same time, which will corrupt the minidisk.  With the patch, the
minidisks can be linked read-only, which eliminates the risk.

Of course, I am not an upstream kernel person, nor a Debian developer,
nor a Debian security person; so someone would have to step up and
sponsor this.  Just a thought.  Suse apparently thought the fix important
enough to backport it to SLES10 (2.6.16) and SLES11 (2.6.27).  So if
Debian decided to do this they would be in good company.

Reply to: