Subject: SCSI failure going from 1.5.1 to 1.5Q
To: None <port-sparc@netbsd.org>
From: William O Ferry <woferry@iname.com>
List: port-sparc
Date: 01/18/2001 23:47:11
	With the sup changeover I grabbed the 1.5Q sources and built a kernel for my 
SPARCbook.  Switching from my 1.5.1_ALPHA2 kernel sup'ed and built 1/7/2001 to 
Tuesday's 1.5Q kernel the machine paused and then panicked during boot.  I'm 
running / on my SPARCbook off an Orb drive, and the Orb drive does not seem to 
work properly under 1.5Q.

	With 1.5.1_ALPHA2 I get the following messages during boot but everything 
works fine:

sd1 at scsibus0 target 4 lun 0: <CWS ORB2, -SE U ID 4, D01> SCSI2 0/direct 
removable
sd1: mode sense (4) returned nonsense; using fictitious geometry
sd1: 2103 MB, 2103 cyl, 64 head, 32 sec, 512 bytes/sect x 4307184 sectors
root on sd1a dumps on sd1b
sd1: mode sense (4) returned nonsense; using fictitious geometry
sd1: mode sense (4) returned nonsense; using fictitious geometry
sd1: mode sense (4) returned nonsense; using fictitious geometry
sd1: mode sense (4) returned nonsense; using fictitious geometry

	With 1.5Q I get:

sd1 at scsibus0 target 4 lun 0: <CWS ORB2, -SE U ID 4, D01> SCSI2 0/direct 
removable
sd1(esp0:4:0): max sync rate 7.57MB/s
esp0: unexpected disconnect: sending REQUEST SENSE
esp0: target didn't send tag: 0 bytes in fifo
esp0: SCSI bus reset

	At this point the machine hangs for a minute or so, then displays:

sd1(esp0:4:0): esp0: timed out [ecb 0xf02f7000 (flags 0x1, dleft 0, stat 0)], 
<state 2 nexus 0xf02f7000, phase(l 1, c 101, p 101), resid 0, msg(q 0, o 100) 
DMA active>
sd1(esp0:4:0): esp0: timed out [ecb 0xf02f7000 (flags 0x1, dleft 0, stat 0)], 
<state 2 nexus 0xf02f7000, phase(l 1, c 101, p 101), resid 0, msg(q 0, o 100) 
DMA active> AGAIN
esp0: SCSI bus reset
sd1: mode sense (4) returned nonsense; using fictitious geometry
sd1: 2103 MB, 2103 cyl, 64 head, 32 sec, 512 bytes/sect x 4307184 sectors
root on sd1a dumps on sd1b
sd1: mode sense (4) returned nonsense; using fictitious geometry
sd1: mode sense (4) returned nonsense; using fictitious geometry
sd1: mode sense (4) returned nonsense; using fictitious geometry
sd1: mode sense (4) returned nonsense; using fictitious geometry
panic: esp0: invalid state: 6

	It displays some panic information and attempts a dump, but obviously gets 
stuck again.  The values are slightly different this time:

sd1(esp0:4:0): esp0: timed out [ecb 0xf02f7070 (flags 0x3, dleft 200, stat 
0)], <state 6 nexus 0xf02f7070, phase(l 17, c 7, p 7), resid 6000, msg(q 0, o 
100) DMA active>
sd1(esp0:4:0): esp0: timed out [ecb 0xf02f7070 (flags 0x3, dleft 200, stat 
0)], <state 6 nexus 0xf02f7070, phase(l 17, c 7, p 7), resid 6000, msg(q 0, o 
100) DMA active> AGAIN

	And that's about as far as it gets.  Switching back to the 1.5.1_ALPHA2 
kernel it boots just fine.

	Does this drive need some sort of quirk entry, or is there something else I 
can do to get this to work under 1.5Q?

	Thanks in advance.
                                          Will Ferry