Re: [patch 1/2] show being-loaded/being-unloaded indic

Previous thread: [GIT PATCHES] V4L/DVB fixes by Mauro Carvalho Chehab on Monday, January 7, 2008 - 2:38 am. (3 messages)

Next thread: regression: 100% io-wait with 2.6.24-rcX by Joerg Platte on Monday, January 7, 2008 - 3:51 am. (11 messages)
From: devzero
Date: Monday, January 7, 2008 - 3:09 am

no wonder, as most modules seem to be very quiet on on load/unload.

some days ago i was curious how "consistent" modules would leave messages in syslog load/unload.
this is why i made a little statistic on that.

i`d like to share the essential part here (can provide more details if you like):

on load, 1204 on unload, 1655 (out of 2088) left absoultely _no_ message in syslog           


example:

3c359 - silence
3c501 - silence
3c503 - 2 lines on load, 0 on unload
3c505 - 3 lines on load, 0 on unload
3c507 - silence
3c509 - silence
3c515 - 1 line on load, 0 on unload
3c574_cs - silence
3c589_cs - silence
3c59x  - silence

ok, this is not too reliable, because modules might act differently if they find appropriate hardware.

but i wonder if some mandatory "print a message on init/exit" wouldn`t give a benefit to admins or end-users, too.
sure this would additionally clutter up syslog, but it would add transparency.


_____________________________________________________________________
Der WEB.DE SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
http://smartsurfer.web.de/?mc=100071&distributionid=000000000066

--

From: Arjan van de Ven
Date: Monday, January 7, 2008 - 7:21 am

On Mon, 07 Jan 2008 11:09:23 +0100

Regardless of the usefulness, really, if this would be mandatory for all modules, it should
be done in the module loader itself.... 
But I'm not entirely convinced how useful that is to do always, although I can see the 
point of having it as debug option, similar to the initcall debug stuff...


-- 
If you want to reach me at my work email, use arjan@linux.intel.com
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
--

From: Rusty Russell
Date: Monday, January 7, 2008 - 2:05 pm

We could overload initcall_debug for this if we wanted to be consistant.  But 
I'm just not sure it'd be clear when to turn it on...

Rusty.
--

From: Valdis.Kletnieks
Subject: Re:
Date: Monday, January 7, 2008 - 2:43 pm

With a 'printk(LOG_DEBUG,...' please.  Boot with initcall_debug sometime to
see why. ;)
Previous thread: [GIT PATCHES] V4L/DVB fixes by Mauro Carvalho Chehab on Monday, January 7, 2008 - 2:38 am. (3 messages)

Next thread: regression: 100% io-wait with 2.6.24-rcX by Joerg Platte on Monday, January 7, 2008 - 3:51 am. (11 messages)