[PATCH] Re: KDB in ameslab

linas at austin.ibm.com linas at austin.ibm.com
Thu Mar 4 07:30:25 EST 2004


Resending because the ppc64 mail list manager bounced the message
because the patch was too big.

This patch brings the ameslab ppc64 tree up to the current
level of KDB (v 4.3 hot from the sgi ftp site as of two days ago)
Note that since ameslab currently has kdb 4.1(?) in it, that
this patch has the effect of undoing the old kdb as it adds the
new kdb.

Rather than attaching, because of file size limits, I've put a new,
slightly improved patch at:

(it may take 1-24 hours for the above URL to bcome valid).

The 'slight improvement' includes:

-- addition of missing dis-asm.h
-- fixing CONFIG_KDB_MODULES so that kdb modules work

This patch should apply cleanly to the ameslab tree.


On Tue, Mar 02, 2004 at 07:59:46PM -0600, linas at austin.ibm.com wrote:
> On Tue, Feb 17, 2004 at 03:35:27PM +1100, Anton Blanchard wrote:
> > A side effect of this is that KDB is probably broken. I started looking
> > into fixing it however I noticed it looks out of date. Does someone have
> > the urge to update it?
> I notice that you still haven't applied my old "append __ to the
> debug handles in arch/ppc64/kdb/kdbmain.c patch" to fix the above. :-(
> No matter: here's a big honking patch.
> -- It updates yesterdays ameslab ppc64 bk tree to KDB version 4.3
> -- It compiles, it runs, it seems to work.
>    Caveats:
>     o I have not tested on lpars yet
>     o I stubbed out the TCE code, because that's all changed.
>     o I haven't tried CONFIG_KDB_MODULES which are probably broken.
>     o Its got a couple of other messy areas that need some tweaking,
>       which I may try to do tommorrow, or maybe IBM India might fix
>       later.
>     o It seems to take about 10 seconds between typing in 'startKDB'
>       and the time you get the kdb prompt. Don't know why.  It also
>       takes 10 seconds to switch cpus (with the kdb cpu command).
> *Please* apply this ASAP, before it bit-rots, and I have to do the
> work all over again.
> --linas
> p.s. Keith, I will try to create and send you the corresponding
> ppc64 architecture patch; however, this might be ugly, because
> I suspect the andrew morton kernels are still fairly out of sync
> with the BK ameslab trees.

** Sent via the linuxppc64-dev mail list. See http://lists.linuxppc.org/

More information about the Linuxppc64-dev mailing list