Port-sandpoint archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: sd* at atapibus and 2200bg driver issues.



Thank you Frank.
I think that the heavy disk i/o could be the cause, but please consider this:
1) I could reproduce the hang only via wifi network, using the intel wifi card (802.11g).
2) using the gigabit ethernet connection i haven't got the problem.
3) during the test of the sd-at-atapi kernel mod, I have managed to send and receive between my internal disk and the usb disk a burst of 5Mb files and my beloved :) 20Gb file, repeatedly:
no hangs at all. Maybe the long duration of the wireless file transfert could affect this issue?
May I ask you how I can activate the kernel/device driver debugger? I tried to recompile the kernel with debug option enabled, and defined also iwi_debug in source driver, but I'm only able to obtain messages during the ifconfig/wpa supplicant phase.
In driver source there's a NDBGL? macro: how can I enable and use it??? Google doesn't help in this case! :)
Now i'll leave my dsm box under heavy disk load via gigabit ethernet all night long, hope I could reproduce the same problem.
Bye!

Il giorno 29/apr/2012 18:09, "Frank Wille" <frank%phoenix.owl.de@localhost> ha scritto:
On Sun, 29 Apr 2012 15:56:36 +0200
Charles Brown <obigiankenobi%gmail.com@localhost> wrote:

> sd*   at atapibus? drive ? flags 0x0000
>
> [...]
> Please, could you consider to enable "sd at atapi" support on GENERIC?

Sure. This should be in. Done.
We will try to get it into the 6.0 release too.


> When I try to send or receive via FTP a 20GB test file, the system
> hangs. [...]
> the shell hangs and If I press <CTRL+T> this is what SIGINFO output
> reports:
>
> dsm-g600# ls
> load: 0.07  cmd: ls 911 [biowait] 0.00u 0.03s 0% 1168k

I guess it has nothing to do with the network, but might be the same
lockup under heavy disk i/o load, which Toru Nishimura reported on the
port-powerpc list yesterday:
http://mail-index.netbsd.org/port-powerpc/2012/04/28/msg003187.html

A serious problem indeed.

--
Frank Wille


Home | Main Index | Thread Index | Old Index