From: Kay Sievers <kay.sievers@vrfy.org>
To: linux-hotplug@vger.kernel.org
Subject: Re: udev-067 and 2.6.12?
Date: Wed, 17 Aug 2005 12:35:07 +0000 [thread overview]
Message-ID: <20050817123507.GA17712@vrfy.org> (raw)
In-Reply-To: <200508162349.01610.s0348365@sms.ed.ac.uk>
On Wed, Aug 17, 2005 at 09:26:08AM +0100, Alistair John Strachan wrote:
> On Wednesday 17 August 2005 00:55, Kay Sievers wrote:
> > On Wed, Aug 17, 2005 at 12:26:11AM +0100, Alistair John Strachan wrote:
> > > On Tuesday 16 August 2005 23:49, Alistair John Strachan wrote:
> > > # make
> > > cc -DUSE_LOG -O2 -march=k8 -mfpmath=sse -s -pipe -c -o run_directory.o
> > > run_directory.c
> > > ld -s -o udev_run_devd udev_run_devd.o
> > > run_directory.o ../../udev.a ../../libsysfs/sysfs.a
> > > ld: warning: cannot find entry symbol _start; defaulting to
> > > 00000000004000f0 udev_run_devd.o: In function `log_message':
> > > udev_run_devd.c:(.text+0x9d): undefined reference to `vsyslog'
> > > udev_run_devd.c:(.text+0xb6): undefined reference to `getenv'
> > >
> > > GCC 4.0.1/binutils 2.16.91.0.2. I worked around it by linking with gcc
> > > instead (like the rest of udev does). This worked fine. I didn't compile
> > > udev or the run_directory stuff against klibc.
> >
> > Hmm, no other distro seems to have this problem. Did you overwrite
> > CFLAGS for the make?
>
> No. I typed "make" and edited nothing. I'm sceptical that this is a problem
> with "only" my configuration, since I've compiled literally hundreds of
> packages with a problem, and run_directory/ is the only thing I've ever seen
> not link bizarrely. I think the problem is that it's a mistake to use ld
> explicitly to link in the case where I want to also link in libc, and not use
> klibc. Many of the missing symbols were from libc.
You may need to run make from the top-level directory of of the udev
tree with "make EXTRAS=extras/run_directory"?
> > > Finally, I'm sure this is my fault, but modules like "mousedev" and
> > > "sd_mod" don't get loaded by hotplug/udev when I plug in my mouse or
> > > pendrive after booting. This is because they are not actually
> > > dependencies of the modules usbhid and usb-storage, so they don't get
> > > loaded by modprobe.
> >
> > For scsi see here:
> >
> > http://www.kernel.org/git/?p=linux/hotplug/udev.git;a=blob;hh3a22a2bb7565
> >66b7e18a57231bae9913e61da2;hbV18b5611bccf64a1e22f8cbcebe8ecba9713c9c;f=etc
> >/udev/suse/udev.rules#l158
>
> This seems like a step backwards from what 053 was doing. I can't see how it's
> beneficial to not be intuitively loading useful modules without an explicit
> rule.
I don't see any problem here. If the kernel has no dependency to load
the module, _you_ need to load it.
> > And better compile mousedev in the kernel. That does not make a lot of
> > sense as a module. If you really want a module, load it with an
> > init-script.
> >
>
> Agreed, I used to run with an allbuiltin kernel because back in the early
> 2.5s, it was your only option. When I migrated to udev I switched to modules
> before the udevstart program was mature, and was probably a bit overzealous
> compiling out as much as possible.
>
> I still think this is a counter intuitive change.
You can still use the old hotplug multiplexer /sbin/hotplug and the
shell scripts and use udev only in as a device node manager, if you
like that more.
> > > udev 053 used to do this, 067 doesn't any more. I've got both the
> > > udev_run_devd and udev_run_hotplugd binaries compiled and in the right
> > > place; I stole the configuration from the gentoo rules. Any idea what's
> > > changed?
> >
> > No idea what's changed. Probably one of the hotplug agents is not
> > called, but we want to get rid of them anyway. :)
>
> Well, I suppose this makes sense. What would really be nice is if udev didn't
> abandon people running none of the six supported distributions, and instead
> provided broken apart rules files (much like how debian does it), so that at
> least the raw, basic functionality is always going to be installed.
>
> Things like automatically loading sd_mod when a scsi disk is added, and
> running the legacy devd/hotplugd helpers should really be done until there
> are 100% suitable replacements for them. udev promised not only automation,
> but an unparalleled minimum required configuration to work (I don't even have
> a modules.conf, or have to explicitly load any modules. It's great.)
Well, what people are calling "great" varies so much, that I decided to
remove the broken "default" rules, cause too many people complained about
these rules and it's simply wrong to claim to have a default, when nobody
cares what's in the default.
I'm happy to put your working rules for your distro in the tree if you
send it, but there will be no "default" rules anymore as the distro
rules are well maintained and are the best examples what may be needed
to get udev integrated.
Kay
-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel
next prev parent reply other threads:[~2005-08-17 12:35 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-08-16 22:49 udev-067 and 2.6.12? Alistair John Strachan
2005-08-16 23:00 ` David Brown
2005-08-16 23:26 ` Alistair John Strachan
2005-08-16 23:31 ` David Brown
2005-08-16 23:55 ` Kay Sievers
2005-08-17 8:09 ` Arioch
2005-08-17 8:26 ` Alistair John Strachan
2005-08-17 12:35 ` Kay Sievers [this message]
2005-08-17 14:52 ` Alistair John Strachan
2005-08-17 15:10 ` Kay Sievers
2005-08-17 23:48 ` Greg KH
2005-08-18 15:27 ` David Brown
2005-08-18 15:38 ` Greg KH
2005-08-18 15:53 ` David Brown
2005-08-18 16:32 ` Alistair John Strachan
2005-08-18 16:56 ` Greg KH
2005-08-18 17:19 ` Alistair John Strachan
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=20050817123507.GA17712@vrfy.org \
--to=kay.sievers@vrfy.org \
--cc=linux-hotplug@vger.kernel.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;
as well as URLs for NNTP newsgroup(s).