[Gnewsense-dev] Re: [lemote] gNewSense Lemote Yeeloong 2.6.30 linux - please test this build

[Danny Clark]

Danny Clark wrote:
> Hongbing Hu wrote:
>> Danny Clark 写道:
>>> Is there a patch to ec_modules to get it to compile against 2.6.30? I'm
>>> getting a build error that looks like it is related to a change between
>>> 2.6.27 and 2.6.30:
>>>
>>>   
>> Yes，the struct proc_dir_entry have been changed.
>> the new version deletes the member named "owner".
>>
>> try to delete the "bat_proc_entry->owner = THIS_MODULE",i think, it will
>> be ok!
> 
> Seems to work, although had to remove that line in a few files:
> http://config.fsf.org/trac/public/wiki/RmsLinux#Tocompileagainst2.6.30
> 
> The 2.6.30 linux for gnewsense is up now; I'm trying to get people to
> test, and am personally running a massive compile on it now (compile of
> cross-compile toolchain for SH2, to compile the GPLed firmware for a
> atheros ar9170 based USB 802.11n adapter - I'll post about that if it
> starts working - http://linuxwireless.org/en/users/Drivers/ar9170.fw ).
> 
> Install instructions are up on:
> 
> http://config.fsf.org/trac/public/wiki/RmsLinuxForYou
> 
> Note that this test build is not deblobbed (not linux-libre), so
> practice diligence in what USB devices you use :) - I will build a
> deblobbed version after some testing of this linux by myself and others.
> 
> Also not sure if I've posted about this before, but currently the best
> wifi performance I've seen on the yeeloong has come from this, which rms
> is carrying with him:
> 
> Alfa 500mW USB Wireless-G Adapter w/ threaded RP-SMA Jack & Antenna AWUS036H
> http://www.data-alliance.net/-strse-73/Alfa-500mW-AWUS036H-USB/Detail.bok
> 
> There is a printk statement you need to comment out in the source so it
> doesn't spam the consoles (which is where rms usually works):
> 
> http://config.fsf.org/trac/public/wiki/RmsLinux#Alfa500mWpatch
> 
> I'm hoping this build will not have the rare but noticeable (rms is
> seeing around once every 1-3 weeks) linux 2.6.27-STD hard lockup that
> seems to be somewhat correlated with large amounts of (perhaps USB
> flash) I/O.

Drat, just had a hard lockup during the compile. I don't have more time
to look at this now, but if you do, here is an email that describes how
we may be able to get some debug information for the hangs:

-------- Original Message --------
Subject: Re: [lemote] about hard disk freeze
Date: Wed, 01 Jul 2009 10:32:28 +0800
From: yanhua <address@hidden>
Reply-To: rms / fsf sysadmin / fsf volunteers / lemote <address@hidden>
To: rms / fsf sysadmin / fsf volunteers / lemote <address@hidden>
References: <address@hidden> <address@hidden>

Danny Clark 写道:
> Unfortunately not much information to share; it's happened with
> multiple disk types (the 160gb hitachis (I think) that shipped with
> the units, as well as a 64gb samsung slc ssd), and I've never got any
> useful error messages (even with linux compiled with detect soft
> lockups; I tried and failed to get netconsole to work - have you ever
> gotten that working / in general are there any debug .config options
> you'd recommend?)

1. sorry, I have no specific debug .config recommend.
I also just the kernel debug options.
2. the netconsole is working. I have tested it.

you'd better compile it into kernel.
when boots. use the following command line to kernel
g no_auto_cmd console=tty address@hidden/eth0,address@hidden/

in the machine which is used as console output:
nc -u -l -p 4444

then you should get the kernel output.

BTW: you say there is one drive which is OK, what is its type?

you can get drive model by /proc/ide/hda/model

signature.asc
Description: OpenPGP digital signature