On Wed, Mar 01, 2000 at 10:06:36PM -0500, Jake Luck wrote:
> 
> Just compiled today's 1.4T GENERIC Kernel, however
> up on boot, it no longer recognized my harddrive any more
> 
> ---- cut from boot ------------
> scsibus0: waiting 2 seconds for device to settle...
> sd0 at scsibus0 targ 2 lun 0: <, , > SCSI0 0/direct fixed
> sd0: mode sense (4) returned nonsense; using fictitious geometry
> sd0: 8201 MMB, 8201 cyl, 64 head, 32 sec, 512 bytes/sec x 16796929 sectors
> probe(mesh0:4:0): Sense Error Code 0x0
> mesh: timeout state=3
> mesh: resetting dma
> 
> <hangs indefinitely>
> ---- end cut from boot -------
> 
> i tried rebooting with the 20000205 snapshot kernel and everything seems
> to work. would somone please shed some light on this?

Confirmation may not be worth much, but I had identical problems (and
had to resort to the 1.4R GENERIC kernel - is it possible to get
kernel source other than -current?). Well, I get each mesh error twice
before the hang, but whatever. 

I've a 7500 (OF 1.0.5) with a G3 processor card and no unusual PCI
cards (that is, no other SCSI chains than the standard to be probing).

Is this a global mesh SCSI thing or an oddity of OF 1.0.5-era
powermacs? (That is, x500.)

       ~ g r @ eclipsed.net

PS, I'm back alive... untarring even base.tgz by hand rather than
letting sysinst do it (it didn't want to fsck my /usr partition) is a
bad idea for two reasons. First, it overwrites /etc/postfix with an
empty directory (even though /etc/postfix existed before, yick, and
it's in /etc, yicker). Second, if you're like me, you'll forget tar is
actually pax on the install floppy (I think?) and that you really want
the -pe options if you'd like to get at kmem (well, actually, if you'd
like ownership and mode of installed files to be correct). So I
learned how to use pax and did it the right way... the second time.
And rewrote my postfix config.