On Wed 24 Jun 2009 16:13, Mike Frysinger pondered:
> On Wed, Jun 24, 2009 at 16:11, Robin Getz wrote:
> > On Thu 18 Jun 2009 09:24, [email protected] pondered:
> >>       Revision
> >>       6784
> >>       Author
> >>       vapier <http://blackfin.uclinux.org/gf/user/vapier/>
> >>       Date
> >>       2009-06-18 08:24:18 -0500 (Thu, 18 Jun 2009)
> >>
> >>
> >> Log Message
> >>
> >> Blackfin: unify memory region checks between kgdb and traps (part 2)
> >>
> >> This fixes up some issues pointed out by Robin (obscure return values)
> >> and some issues pointed out by Enrik (ambiguous/missing checks for Core
> >> B on BF561).  It also cleans up the Blackfin kgdb immensely.
> >
> > Now that this is here - I think we can use in in
> > arch/blackfin/kernel/ptrace.c?
> >
> > That should make most of the nastiness go away there
> 
> indeed ... i had looked at that before, but forgot about it completely
> when doing this second part
> 
> > and the same time - clean up bug 3417
> >
> > http://blackfin.uclinux.org/gf/tracker/3417
> 
> if you say so ... i dont really get what's going on there and our
> module code looks like it only processes specific section names, not
> addresses.  so i'm not sure we can leverage these new functions.

Sorry - what I meant was (after looking at the top of the bug in a little more 
detail) was also change/add 
arch/blackfin/kernel/module.c:apply_relocate_add()

All the :
  *location16 = (value & 0x3ff);

should be figuring out where things are in memory, and then (if necessary) 
using isram_memcpy, or dma_memcpy()

We are only doing it in a two places (R_BFIN_LUIMM16 & R_BFIN_HUIMM16) and we 
need to do it for all...

-Robin

_______________________________________________
Linux-kernel-commits mailing list
[email protected]
https://blackfin.uclinux.org/mailman/listinfo/linux-kernel-commits

Reply via email to