Openembedded Devel Discussions
 help / color / mirror / Atom feed
From: Paul Sokolovsky <pmiscml@gmail.com>
To: openembedded-devel@lists.openembedded.org
Subject: Re: [RFC] Revamp module handling in OE
Date: Tue, 24 Jul 2007 00:54:33 +0300	[thread overview]
Message-ID: <86855446.20070724005433@gmail.com> (raw)
In-Reply-To: <77326513.20070720105101@gmail.com>

Hello OpenEmbedded,

Friday, July 20, 2007, 10:51:01 AM, you wrote:

> Hello openembedded-devel,

>   We have few issues with kernel module handling in OE which I'd like
> to bring to attention and discuss ways to resolve.

> 1. Module config option handling was not upgraded since 2.4.

> This is known issue, it surfaced several times already, and we
> discussed ways to solve it with Graeme Gregory on IRC recently. The
> issue is that OE currently writes module config options where 2.4
> modutils expect them (/etc/modutils/*.conf), not where 2.6
> module-init-tools do (/etc/modprobe.d/*.conf). The patch is at
> http://bugs.openembedded.org/show_bug.cgi?id=2669

  I got confirmation from Graeme that it works ok, so I'd like to
commit this tomorrow night.


> 2. Location of module autoload config.

> Module autoload is rather different feature. It actually never was
> part of modutils/module-init-tools, but is done via adhoc scripts per
> some distro's convention. Relatively randomly OE uses the same
> /etc/modutils dir to store that info (just in files w/o extension).
> Per-module data is then collated into /etc/modules file, and *that*
> file is Debian convention. But AFAICT, that file is supposed to
> contain local user's selection of autoload modules, whereas OE
> automatically overwrites it. Of course, this information rather be
> checked by people with more Debian background. Either way, at least
> following issues can be identified:

> 1. Location of per-module autoload data: /etc/modutils/ is rather
> confusing place for this, especially if we switch to 2.6 way.
> 2. /etc/modules is supposedly a user file.

> One of the possible solutions: if a file is named /etc/modules, then
> it's natural to name dir /etc/modules.d/ . If data from it would need
> to be collated, that would go to some /etc/modules.foo, leaving
> /etc/modules intact. FInally, both contents of files and dri would be
> taken into account.


  I hope that people are fully back from GUADEC now and can give this
some thought ;-).


> Thoughts?




-- 
Best regards,
 Paul                            mailto:pmiscml@gmail.com




      reply	other threads:[~2007-07-23 21:55 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-07-20  7:51 [RFC] Revamp module handling in OE Paul Sokolovsky
2007-07-23 21:54 ` Paul Sokolovsky [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=86855446.20070724005433@gmail.com \
    --to=pmiscml@gmail.com \
    --cc=openembedded-devel@lists.openembedded.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox