Skip to Content.
Sympa Menu

sm-discuss - Re: [SM-Discuss] eudev

sm-discuss AT lists.ibiblio.org

Subject: Public SourceMage Discussion List

List archive

Chronological Thread  
  • From: Javier Vasquez <j.e.vasquez.v AT gmail.com>
  • To: "sm, discuss" <sm-discuss AT lists.ibiblio.org>
  • Subject: Re: [SM-Discuss] eudev
  • Date: Sun, 18 Aug 2013 20:23:49 -0600

On Sun, Aug 18, 2013 at 5:26 PM, Javier Vasquez <j.e.vasquez.v AT gmail.com>
wrote:
> On Sun, Aug 18, 2013 at 4:58 PM, Javier Vasquez <j.e.vasquez.v AT gmail.com>
> wrote:
>> On Sun, Aug 18, 2013 at 4:34 PM, Javier Vasquez <j.e.vasquez.v AT gmail.com>
>> wrote:
>>> On Sun, Aug 18, 2013 at 4:28 PM, Javier Vasquez <j.e.vasquez.v AT gmail.com>
>>> wrote:
>>>> On Sun, Aug 18, 2013 at 4:15 PM, Javier Vasquez
>>>> <j.e.vasquez.v AT gmail.com> wrote:
>>>>>>>> ...
>>>>>>>
>>>>>>> docbook-xsl is a required dependency:
>>>>>>>
>>>>>>> make[2]: Entering directory `/home/sorcery/src/eudev-1.2/man'
>>>>>>> /usr/bin/xsltproc -o udev.7 --nonet --stringparam man.output.quietly 1
>>>>>>> --stringparam funcsynopsis.style ansi --stringparam
>>>>>>> man.th.extra1.suppress 1 --stringparam man.authors.section.enabled 0
>>>>>>> --stringparam man.copyright.section.enabled 0
>>>>>>> http://docbook.sourceforge.net/release/xsl/current/manpages/docbook.xsl
>>>>>>> udev.xml
>>>>>>> I/O error : Attempt to load network entity
>>>>>>> http://docbook.sourceforge.net/release/xsl/current/manpages/docbook.xsl
>>>>>>> warning: failed to load external entity
>>>>>>> "http://docbook.sourceforge.net/release/xsl/current/manpages/docbook.xsl";
>>>>>>> cannot parse
>>>>>>> http://docbook.sourceforge.net/release/xsl/current/manpages/docbook.xsl
>>>>>>>
>>>>>>> See grimoire bug 588...
>>>>>>
>>>>>> Two additinal problems I was not expecting given what you mentioned in
>>>>>> prior e-mails:
>>>>>>
>>>>>> 1.- eudev doesn't seem to be respecting /etc/udev/rules.d files.
>>>>>> 2.- eudev is not loading modules specified under /etc/modules.
>>>>>>
>>>>>> Those 2 things were working fine with kmod+udev.
>>>>>>
>>>>>> Notice I no longer has kmod neither udev installed. Instead I have
>>>>>> module-init-tools+eudev installed.
>>>>>
>>>>> Forgot to mention I filed grimoire bug 589 for this...
>>>>
>>>> BTW, /dev/shm is also missing as initially indicated, but somehow it
>>>> sounds like it was corrected, given that /dev/pts had a special
>>>> handling but /dev/shm didn't.
>>>>
>>>> I just copied/pasted the pts init script into a shm one and changed
>>>> the dev and target, we'll see if that works...
>>>
>>> I forgot again, see grimoire bug 590.
>>>
>>> BTW, I'm thinking that the /etc/modules might get loaded if I
>>> re-enable the init scrip for module-init-tools, I disabled it in favor
>>> of kmod+udev, but perhaps eudev no longer plays its part on it.
>>>
>>> Something else is that mountall init script fails, though I don't seem
>>> to get any log indicating why.
>>>
>>> I'll reboot again after and see if at least the modules are loaded,
>>> and /shm mounted.
>>
>> As indicated in 589, my mistake on /etc/modules. That was provided by
>> kmod, not udev, and by removing it, module-init-toos must take over,
>> so I needed to recast it enabling the init script.
>>
>> Also, copying and modifying the pts init script into a shm one, did
>> the magic to get /dev/shm mounted, as also indicated in 590. The
>> reason this is required might be that the device is not ready.
>>
>> Interestingly enough as well, as stated on 589, it seems eudev is
>> paying attention to the rules in /etc/udev/rules.d, since both the kvm
>> and the vhost-net modules, when loaded, generated the devices with the
>> right group ownership, which only happens if obeying the rules in
>> etc... That might mean the problem is in the way eudev renames the
>> network interfaces, :-(
>>
>> So, after the workarounds, the still pending issues are:
>>
>> 1.- Network interfaces renamed, no matter the rules in etc.
>> 2.- mountall still fails, I haven't found out why.
>>
>> /dev/shm of course requires the init script attached in previous e-mail...
>
> Just create 591, to reflect the fact mountall fails... Notice that
> might be related to the fact neither /dev/pts, nor /dev/shm get
> mounted, unless the 2 runlevel 3 scripts are added. It might be
> mountall should creat those sub-dirs if not present, prior to trying
> "mount -a ...". That might be an easy fix, but in a different spell
> then.

Found the issue why /dev/shm and /dev/pts are not mounted, and why the
init script mountall fails. I explained it in bug 591. Bug 590
becomes then a duplicate of 590. On 591 I describe the work around,
and the reason for the work around. Though I didn't understand why
INSTALL fails...

Any ways, the work around consists on copying the udev.conf provided
by the spell by hand on /etc/udev. That's required to define
$udev_root, which used by the init script "devices".

After doing that the "devpts" and in my case the "devshm" scripts are
removed, and still on boot mountall succeeds and /dev/shm and /dev/pts
get mounted. So the work around works OK.

I suspect that as the spell is "eudev", and the config file
"udev.conf", the method "install_config_file" used by INSTALL somehow
overrides udev.conf...

Besides fixing the spell so that INSTALL works as it should, the only
problem left for me would be the rename of interfaces, not paying
attention to the names in rules.d...

Thanks,

--
Javier.




Archive powered by MHonArc 2.6.24.

Top of Page