[FREEZE] 2.4.19-pre10 + Promise ATA100 tx2 ver 2.20

[FREEZE] 2.4.19-pre10 + Promise ATA100 tx2 ver 2.20

[Fabio Massimo Di Nitto]

Hi all,
            probably was already discussed (I hope not).

I get a really bad freeze when there's heavy writing over the 2 x 120GB 
WD disks.
reading seems to be ok.

Since Im really not an expert in setting up IDE stuff the only things I 
can say is that
there's no OOPS or error or nothing on the screen/log.

The freeze is reproducible both with 2.4.19-pre10 and 2.4.18.
In my config I have tryed with/without DMA/Promise special options and 
different combinations
of them with no results.

I will be glad to offer more info and help if someone can kindly tell me 
how to debug
this. If it's needed I can consider providing access to the machine.

Thanks
Fabio

Re: [FREEZE] 2.4.19-pre10 + Promise ATA100 tx2 ver 2.20

[Måns Rullgård]

Fabio Massimo Di Nitto <fabbione@fabbione.net> writes:

> Hi all,
>             probably was already discussed (I hope not).
> 
> I get a really bad freeze when there's heavy writing over the 2 x
> 120GB WD disks.
> 
> reading seems to be ok.
> 
> Since Im really not an expert in setting up IDE stuff the only things
> I can say is that
> 
> there's no OOPS or error or nothing on the screen/log.
> 
> The freeze is reproducible both with 2.4.19-pre10 and 2.4.18.
> In my config I have tryed with/without DMA/Promise special options and
> different combinations

Those have no effect in 2.4.19-pre10 and a few earlier pres; I don't
remember the exact number.

> of them with no results.
> 
> I will be glad to offer more info and help if someone can kindly tell
> me how to debug
> 
> this. If it's needed I can consider providing access to the machine.

Does the machine recover from the freeze or do you have to reboot?

-- 
Måns Rullgård
mru@users.sf.net

Re: [FREEZE] 2.4.19-pre10 + Promise ATA100 tx2 ver 2.20

[Fabio Massimo Di Nitto]

Måns Rullgård wrote:

>  
>
>Does the machine recover from the freeze or do you have to reboot?
>
>  
>

I have to hard reset the machine. Keyboard is unuseable at that stage.
not even Magic sysreq works.

Fabio

Re: [FREEZE] 2.4.19-pre10 + Promise ATA100 tx2 ver 2.20

[Måns Rullgård]

Fabio Massimo Di Nitto <fabbione@fabbione.net> writes:

> Måns Rullgård wrote:
> >Does the machine recover from the freeze or do you have to reboot?
>
> I have to hard reset the machine. Keyboard is unuseable at that stage.
> not even Magic sysreq works.

Are you running X? If yes, try reproducing the error while at a text
console. It's possible some panic message will make it that far. Messages
from a dying kernel will often not make it into the syslog.

-- 
Måns Rullgård
mru@users.sf.net

Re: [FREEZE] 2.4.19-pre10 + Promise ATA100 tx2 ver 2.20

[Fabio Massimo Di Nitto]

Måns Rullgård wrote:

>
>Are you running X? If yes, try reproducing the error while at a text
>console. It's possible some panic message will make it that far. Messages
>from a dying kernel will often not make it into the syslog.
>
>  
>
Unfortunatly there's no X and no space to install it. That box should be 
my home
(read: cheap) fileserver and runs just the minimum installation. It 
boots from a
mostly full 500Mb hd and the idea was to use the rest for serving the 
other boxes.

Fabio

Re: [FREEZE] 2.4.19-pre10 + Promise ATA100 tx2 ver 2.20

[Dieter Stueken]

Jussi Laako wrote:

> No matter which one of the
> installations is booted, the system will lock up if the VIA controller is in
> use at the same time as the PDC. However, system seems to be stable (short
> test priod though) if only the PDC controller is in use. (I left the old
> drives attached to the VIA controller unmounted.)
> 
> Does this problem sound similar to yours?

seems similar to my problem. If I connect large (75-120G) disks on
both PDC channels and run /sbin/badblocks in parallel (dd might do, too)
I get a kernel panic/freeze. I get it even with no disk on the onboard
controller. I can reproduce this on two different VIA boards with both,
a PDC20267 and a PDC20262, too. On a board with Ali chiset, however
no error occured during a short (5 minutes) test. On a board with an
Intel
chipset I got some "DMA timeout" after 3 hours, but may be this was
an unrelated problem.

Dieter.