* Re: waiting for udevstart
2004-10-15 21:43 waiting for udevstart Bill Nottingham
@ 2004-10-15 21:54 ` Bill Nottingham
2004-10-15 22:01 ` Greg KH
` (7 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: Bill Nottingham @ 2004-10-15 21:54 UTC (permalink / raw)
To: linux-hotplug
Bill Nottingham (notting@redhat.com) said:
> udevstart is called on startup to populate /dev.
>
> However, all it does is generate udev events - which means
> that when udevstart exits, /dev isn't necessarily populated
> at all.
Never mind. Different error going on here.
Bill
-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
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
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: waiting for udevstart
2004-10-15 21:43 waiting for udevstart Bill Nottingham
2004-10-15 21:54 ` Bill Nottingham
@ 2004-10-15 22:01 ` Greg KH
2004-10-15 22:46 ` Marco d'Itri
` (6 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: Greg KH @ 2004-10-15 22:01 UTC (permalink / raw)
To: linux-hotplug
On Fri, Oct 15, 2004 at 05:43:33PM -0400, Bill Nottingham wrote:
> udevstart is called on startup to populate /dev.
>
> However, all it does is generate udev events - which means
> that when udevstart exits, /dev isn't necessarily populated
> at all.
Are you sure? In looking at the most recent code (remember, udevstart
is a symlink to udev now) I don't see any exec or fork() path when
udevstart is running. But I might be missing something...
> You could do a hack like:
>
> - generate a list of all the events you're going to create
> - write a dev.d script that waits for the last one
> - while (1) sleep(500);
> - have that dev.d script kill udevstart
>
> But that's gross.
That is.
> Any better suggestions?
Make udevstart not return until it is done. I think that's what it does
today. But if not, I'll gladly accept patches to make it do so.
thanks,
greg k-h
-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
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
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: waiting for udevstart
2004-10-15 21:43 waiting for udevstart Bill Nottingham
2004-10-15 21:54 ` Bill Nottingham
2004-10-15 22:01 ` Greg KH
@ 2004-10-15 22:46 ` Marco d'Itri
2004-10-15 22:54 ` Kay Sievers
` (5 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: Marco d'Itri @ 2004-10-15 22:46 UTC (permalink / raw)
To: linux-hotplug
On Oct 16, Greg KH <greg@kroah.com> wrote:
> Make udevstart not return until it is done. I think that's what it does
> today. But if not, I'll gladly accept patches to make it do so.
Debian bug #250598 says it does not, but it may be old:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug%0598
--
ciao, |
Marco | [8582 giYtO/DRUAPco]
-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
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
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: waiting for udevstart
2004-10-15 21:43 waiting for udevstart Bill Nottingham
` (2 preceding siblings ...)
2004-10-15 22:46 ` Marco d'Itri
@ 2004-10-15 22:54 ` Kay Sievers
2004-10-15 23:03 ` Kay Sievers
` (4 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: Kay Sievers @ 2004-10-15 22:54 UTC (permalink / raw)
To: linux-hotplug
On Fri, Oct 15, 2004 at 03:01:38PM -0700, Greg KH wrote:
> On Fri, Oct 15, 2004 at 05:43:33PM -0400, Bill Nottingham wrote:
> > udevstart is called on startup to populate /dev.
> >
> > However, all it does is generate udev events - which means
> > that when udevstart exits, /dev isn't necessarily populated
> > at all.
>
> Are you sure? In looking at the most recent code (remember, udevstart
> is a symlink to udev now) I don't see any exec or fork() path when
> udevstart is running. But I might be missing something...
Right, we just prepare the list of devices to work on, and walk over it
without any external call. Just one binary. During that walk the dev.d/
scripts will be called.
> > You could do a hack like:
> >
> > - generate a list of all the events you're going to create
> > - write a dev.d script that waits for the last one
> > - while (1) sleep(500);
> > - have that dev.d script kill udevstart
> >
> > But that's gross.
>
> That is.
>
> > Any better suggestions?
>
> Make udevstart not return until it is done. I think that's what it does
> today. But if not, I'll gladly accept patches to make it do so.
No, it will not return until the last node is created!
Kay
-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
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
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: waiting for udevstart
2004-10-15 21:43 waiting for udevstart Bill Nottingham
` (3 preceding siblings ...)
2004-10-15 22:54 ` Kay Sievers
@ 2004-10-15 23:03 ` Kay Sievers
2004-10-15 23:24 ` Marco d'Itri
` (3 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: Kay Sievers @ 2004-10-15 23:03 UTC (permalink / raw)
To: linux-hotplug
On Sat, Oct 16, 2004 at 12:46:19AM +0200, Marco d'Itri wrote:
> On Oct 16, Greg KH <greg@kroah.com> wrote:
>
> > Make udevstart not return until it is done. I think that's what it does
> > today. But if not, I'll gladly accept patches to make it do so.
> Debian bug #250598 says it does not, but it may be old:
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug%0598
If Debian really runs udevstart and waits until it's finished this
should not happen? Are you sure, that udevstart (the binary and not
the old shellscript) is run as early as possible and not in the
background?
Kay
-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
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
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: waiting for udevstart
2004-10-15 21:43 waiting for udevstart Bill Nottingham
` (4 preceding siblings ...)
2004-10-15 23:03 ` Kay Sievers
@ 2004-10-15 23:24 ` Marco d'Itri
2004-10-15 23:40 ` Kay Sievers
` (2 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: Marco d'Itri @ 2004-10-15 23:24 UTC (permalink / raw)
To: linux-hotplug
On Oct 16, Kay Sievers <kay.sievers@vrfy.org> wrote:
> > Debian bug #250598 says it does not, but it may be old:
> > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug%0598
> If Debian really runs udevstart and waits until it's finished this
> should not happen? Are you sure, that udevstart (the binary and not
> the old shellscript) is run as early as possible and not in the
> background?
Yes, I am quite sure because I wrote the script. udevstart runs in
foreground and the only init script running before udev is the one
which mounts /sys and /proc.
/etc/rcS.d/:
README S20modutils@ S37hotplug@ S48console-screen.sh@
S02mountvirtfs@ S25raid2@ S38pppd-dns@ S50hwclock.sh@
S04udev@ S27evms@ S39atm@ S55bootmisc.sh@
S05bootlogd@ S30checkfs.sh@ S39ifupdown@ S55urandom@
S05keymap.sh@ S30procps.sh@ S40hostname.sh@ S70screen-cleanup@
S07hdparm@ S35mountall.sh@ S40network@ S70xfree86-common@
S10checkroot.sh@ S36discover@ S40networking@ S75sudo@
S18hwclockfirst.sh@ S36lm-sensors@ S43portmap@
S20module-init-tools@ S36mountvirtfs@ S45mountnfs.sh@
--
ciao, |
Marco | [8584 lu4i9vBNCcik.]
-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
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
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: waiting for udevstart
2004-10-15 21:43 waiting for udevstart Bill Nottingham
` (5 preceding siblings ...)
2004-10-15 23:24 ` Marco d'Itri
@ 2004-10-15 23:40 ` Kay Sievers
2004-10-15 23:45 ` Marco d'Itri
2004-10-16 2:54 ` Bill Nottingham
8 siblings, 0 replies; 10+ messages in thread
From: Kay Sievers @ 2004-10-15 23:40 UTC (permalink / raw)
To: linux-hotplug
On Sat, Oct 16, 2004 at 01:24:41AM +0200, Marco d'Itri wrote:
> On Oct 16, Kay Sievers <kay.sievers@vrfy.org> wrote:
>
> > > Debian bug #250598 says it does not, but it may be old:
> > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug%0598
> > If Debian really runs udevstart and waits until it's finished this
> > should not happen? Are you sure, that udevstart (the binary and not
> > the old shellscript) is run as early as possible and not in the
> > background?
> Yes, I am quite sure because I wrote the script. udevstart runs in
> foreground and the only init script running before udev is the one
> which mounts /sys and /proc.
>
> /etc/rcS.d/:
>
> README S20modutils@ S37hotplug@ S48console-screen.sh@
> S02mountvirtfs@ S25raid2@ S38pppd-dns@ S50hwclock.sh@
> S04udev@ S27evms@ S39atm@ S55bootmisc.sh@
> S05bootlogd@ S30checkfs.sh@ S39ifupdown@ S55urandom@
> S05keymap.sh@ S30procps.sh@ S40hostname.sh@ S70screen-cleanup@
> S07hdparm@ S35mountall.sh@ S40network@ S70xfree86-common@
> S10checkroot.sh@ S36discover@ S40networking@ S75sudo@
> S18hwclockfirst.sh@ S36lm-sensors@ S43portmap@
> S20module-init-tools@ S36mountvirtfs@ S45mountnfs.sh@
Looks good. I just looked into the last udev-deb and this error should not
happen.
The bug report is from May and the recent udevstart is from August, so I
think this is long solved, or do you get any recent reports like this?
Kay
-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
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
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: waiting for udevstart
2004-10-15 21:43 waiting for udevstart Bill Nottingham
` (6 preceding siblings ...)
2004-10-15 23:40 ` Kay Sievers
@ 2004-10-15 23:45 ` Marco d'Itri
2004-10-16 2:54 ` Bill Nottingham
8 siblings, 0 replies; 10+ messages in thread
From: Marco d'Itri @ 2004-10-15 23:45 UTC (permalink / raw)
To: linux-hotplug
On Oct 16, Kay Sievers <kay.sievers@vrfy.org> wrote:
> The bug report is from May and the recent udevstart is from August, so I
> think this is long solved, or do you get any recent reports like this?
I haven't got anything recent, but this may be because there is this bug
open. I will close it and see what happens.
--
ciao, |
Marco | [8585 denKuaohgKTzs]
-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
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
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: waiting for udevstart
2004-10-15 21:43 waiting for udevstart Bill Nottingham
` (7 preceding siblings ...)
2004-10-15 23:45 ` Marco d'Itri
@ 2004-10-16 2:54 ` Bill Nottingham
8 siblings, 0 replies; 10+ messages in thread
From: Bill Nottingham @ 2004-10-16 2:54 UTC (permalink / raw)
To: linux-hotplug
Kay Sievers (kay.sievers@vrfy.org) said:
> > Make udevstart not return until it is done. I think that's what it does
> > today. But if not, I'll gladly accept patches to make it do so.
>
> No, it will not return until the last node is created!
Correct; the problem being seen was:
a) run udevstart
b) load some modules
c) do stuff
The problem was the devices created by b), not a).
I have a strange feeling that this is going to uncover lots
of problems in various places. :/
Bill
-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
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
^ permalink raw reply [flat|nested] 10+ messages in thread