Leo Weppelman wrote:
> 
> Hi all,

Hi back ;)
> 
> I just uploaded the following new things to ftp.netbsd.org (directory:
> pub/NetBSD/arch/atari/...):...
> The last kernel I released about 2 months ago, had some serious problems on
> some of the Falcons. It also had problems in using removable-drives on both
> TT and Falcon. Markus Kilbinger and I have been working hard since to tackle
> the SCSI-problems and we at least discovered (and fixed ;-) ) a bug in the
> DMA handling on the Falcon. At least Markus's Syquest now works ok. But
> this is no guarantee that all Falcons behave like that (we found that out
> the hard way with the last kernel :-( ).OH YEAH
>					 So my request to all Falcon owners
> is: please try the new kernels. Start with 'netbsd.BOOTX.poll'. Some testing
> with Dan's Falcon showed that the problems he had with the last kernel
> disappeared when the kernel operated in polling DMA mode. If that works,
> also try the 'netbsd.BOOTX.intr' kernel. As this is definitely what we
> want to have in the long run.

OK,  here are my first results :

First boot of 'BOOTX.poll' kernel, delivers 

 - on boot sequence the same behavior on my ZIP, this is
   wainting about 20 seconds then showing the error-messages
sd1 at scsibus0 targ 6 lun 0: <IOMEGA, ZIP 100, R.41> SCSI2 0/direct removable
sd1: sd1(ncrscsi0:6:0): illegal request, 
	data = ff fe 01 05 c9 00 00 00 00 00 00 00 00 00 00 00 00
sd1: could not mode sense (4); using fictitious geometry
This messages were also shown when I 'disklabel' or 'newfs' the ZIP

 - fsck /home (= /dev/sd0f on SCSI disk) brought many 'st_pool_size' mesages
   on screen takes a long time and then kernel crashed with 'MMU fault'.
But I think the reason could be an incorrect /home-filesystem.

This crash was reproductable with the other kernel BOOTX.intr

But then reboot with both kernel works ok, also fsck in boot phase of all
partitions ( / /usr and also /home) works fine.

After this, reading a tar-file from a ZIP-disk in MSDOS format brought ONE
kernel hung, but after the next reboot I got NO hung or crash either :)

With kernel BOOTX.poll I then did :
	- binpatch'ing _st_pool_size with several values (see my mail before)
	- making a new filesystem on ZIP
	- writing tar-files of / and /usr on this new ZIP partition 
	  and reading that files.
NO ONE error message appeared and no more kernel crash or hung - up to now

Tomorrow I continue testing e.g. together with my second harddisk
and the other kernel (BOOTX.intr). When it will did also fine you recieve
my congratulations ;-)  but now I must go to bed.

good night
-- 
          ___   __   Helmar Goettsch (Admin HEG: Lehrstuhl III, Informatik)
  /   /  /     /  '  Orleansstr. 34, D-81667 Muenchen, Phone: 089/48095-190  
 /---/--/-----/--+   email: goettsch@informatik.tu-muenchen.de
/   /  /___  /__/    WWW:   http://www3.informatik.tu-muenchen.de/~goettsch