Re: Falcon panics, DEBUG & ST_POOL_SIZE (Was: Testing sysinst.fs)

["T. Makinen" <tjamaloo%gmail.com@localhost>]

On Fri, Nov 14, 2008 at 9:29 PM, David Brownlee <abs%netbsd.org@localhost> 
wrote:
> On Fri, 14 Nov 2008, T. Makinen wrote:
>        I think the best option would be to change REAL_DMA from
>        a compiletime to a runtime test and default it on for TT
>        and off for Falcon. Do you want to take a pass at that? :)

Sure, it seems that adding return value of 0 for Falcon from scsi_dmaok() fixes
this as you suggested  :)

>        If there are specific patterns which are more likely to
>        cause a failure then one option would be to have some code
>        at bootup which would switch to a high bandwidth video mode
>        and then run some DMA scsi tests to try to provoke a failure.
>        If it fails then it disables DMA and continues booting. Of
>        course for that to work it must be easy to induce the
>        failure (so we don't get any false "it works not but fails
>        later"), and we need to be able to recover from the failure
>        (hopefully just reset the SCSI controller). I don't really
>        think that would be worth the effort tho...

I agree. It sounds cumbersome.

>        Sounds like a plan. Do you want to cleanup the code and params
>        and maybe build a kernel for DavidR to test as well?

Here's link to diff that includes pmap related changes and Falcon DMA change:
http://koti.welho.com/tmakinen/atari/changes.diff

Here's link to patched ATARITT DEBUG kernel which has also ST_POOL_SIZE
set to 24:
http://koti.welho.com/tmakinen/atari/ataritt-24-DEBUG.gz

-Tuomo