[Fwd: aic7xxx driver 5.0.19 -> 5.1.2 panic kernel with AHA2742T]

[Randy Gobbel]

On 20-Oct-98 at 22:10, Robert Baller (rdballer at home.com) wrote:
> It seems that others are seeing this problem on the PPC systems.  My
> system is a dual processor pentium 133 Tyan motherboard with a
> AHA2742A-T (Twin Channel) card.
> 
> I've even seen the boot on a AHA3940 system die (used the 5.0.19 driver
> on this system).  The AHA3940 boot problem only seem to occur on random
> boots. 
> 
> The 5.0.19 through 5.1.2 aic7xxx code will not work with my AHA2742A-T
> card.  
> 
> I have a temporary workaround in that I built a 2.0.34 kernel with the
> 4.1 aic7xxx code and that works.

I can't confirm it since my PowerPC Linux system is back in my office, but I
think I may have found what's causing the panic with the newest aic7xxx driver
on PowerPC's.  Doug recently added parity checking on the sequencer code load,
and also redefined the structs for sequencer instructions using explicit bit
fields.  I suspect that the way this was done makes an assumption about data
type endianness, and that the parity bits are being written over the data.

If I'm right about this, it would not be a problem on x86 systems.  We'll see
what Doug has to say.

-Randy


To Unsubscribe: send mail to majordomo at FreeBSD.org
with "unsubscribe freebsd-aic7xxx" in the body of the message