Re: landisk / Plextor Panicing problem

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

> > Could you try attached patch for netbsd-5?
> 
> Tried. fsck went fine, but when I tried to do stuff in the filesystem 
> (delete pkgsrc), I got strange errors like directory not empty. After 
> that, rebooting gave me:

I had the similar ffs corruption issue and I had to rerun newfs(8)
even fater bus_dma(9) fixes were applied.

> # reboot
> Jun 16 03:02:29 reboot: rebooted by root
> panic: tlb_exception: no copyin/out fault handler (page not found)
> expevt=c0 va=c0624000 ssr=400001c0 spc=8c1c2b60 lwp=0x8ff99d20 
> onfault=0x0
> Stopped in pid 0.2 (system) at  netbsd:cpu_Debugger+0x6:        mov 
> r14, r15

On my landisk sh_dcache_inv_range() in BUS_DMASYNC_POSTREAD
and BUS_DMASYNC_POSTREAD|BUS_DMASYNC_POSTWRITE cases are commented out
and I have no obvious problem on building several packages on 5.1.

The above panic is actually caused by the inv function in PREREAD
case, but anyway sh_dcache_inv_range() is harmful if buffer is
not cacheline aligned.
(cacheline at the boundary of the buffer might contain dirty data
 that should be written back)
---
Izumi Tsutsui