Hi,

we are having some problems getting the Adaptec-2940U driver in a
recently installed NetBSD machine here to work reliably.  First
we tried -current code as of approx. 10 May, second we tried the
1.1 driver patched to recognize the 2940U as earlier instructed.
Both of these seem to lose in our case.

I saw that a new version of the Adaptec 2xxx / 7xxx driver code
went in recently, and decided to try it.

This is on a Digital Celebris 590 machine (90 MHz Pentium),
equpped with a Micropolis 532MB SCSI disk identifying itself as
<MICROP, 2105-08MQ1068605, 4849> and a small IDE disk as system
disk (but with /var moved to SCSI).

I have toyed a bit with the device configuration in the Adaptec
SCSISelect BIOS utility, where I can turn on or off sync
negotiation and use of SCSI disconnect.

With both sync and disconnect turned on, I rather promptly
(during startup) got the following console messages (coped from
paper...):

sd0(ahc0:3:0): timed out in dataout phase, SCSISIGI = 0x0
sd0(ahc0:3:0): BUS DEVICE RESET message queued
Bus Device Reset Message Sent
sd0(ahc0:3:0): Bus Device Reset delivered. 1 SCBs aborted
sd0(ahc0:3:0): timed out in message in phase, SCSISIGI = 0xe6
sd0(ahc0:3:0): asserted ATN - device reset in message buffer
sd0(ahc0:3:0): timed out in message in phase, SCSISIGI = 0xf6
ahc0: Issued Channel A Bus Reset #1. 1 SCBs aborted
ahc0: target 3 synchronous at 10.0MHz, offset = 0xf
First bad
/var: bad dir ino 3457 at offset 0: mangled entry
panic: bad dir
db> 

Not too good.

My next try was to turn off synchronous SCSI, and it relatively
quickly (during startup if I remember correctly) complained:

sd0(ahc0:3:0): timed out in dataout phase, SCSISIGI = 0x0
ahc0: Issued Channel A Bus Reset #2, 2 SCBs aborted

This one didn't crash, but the above looked serious enough that I
decided to try to turn off disconnect as well.  When I did that
the machine came up without a hitch, and when ran the drive for a
while (less than 10 minutes) I got:

sd0(ahc0:3:0): Target Busy

During some of the earlier testing we also got this message, and
we hoped it was non-fatal, which however appeared not to be the
case (it later resulted in a "mangled inode" panic).

So...  It seems that this combination of machine, controller and
disk doesn't go along too well with the current version of the
AHC driver.

The versions of the relevant (?) files used in this latest kernel
are:

/*      $NetBSD: ahc_pci.c,v 1.2 1996/05/20 00:56:39 thorpej Exp $      */
/*      $NetBSD: aic7xxx.c,v 1.8 1996/05/20 00:58:07 thorpej Exp $      */
/*      $NetBSD: aic7xxxreg.h,v 1.2 1996/05/20 00:58:10 thorpej Exp $   */
/*      $NetBSD: aic7xxxvar.h,v 1.7 1996/05/20 00:58:11 thorpej Exp $   */
/*      $NetBSD: aic7xxx.seq,v 1.3 1996/05/20 00:48:45 thorpej Exp $    */
/*      $NetBSD: aic7xxx_asm.c,v 1.4 1996/05/20 00:48:48 thorpej Exp $  */

If someone would like us to try out some tweaks to the driver, I
think we can offer to do that testing.

Regards,

- Havard