Hi Milan,
thanks for your fast reply, I will check it out as soon as posible.
When I read your comments on the patch proposal, I still get the feeling
that the kernel uevent interface is broken. Imho, if the kernel sends
out uevents to a netlink interface as well as calling the program stated
in /proc/sys/kernel/hotplug in an asynchronous manner, there should be a
mechanism for proper synchronization.
I might be wrong though.
Thanks & Regards
-Sven
Milan Broz schrieb:
Sven Eschenberg wrote:
Gentoo ships with cryptsetup 1.0.6. This version introduced a call to
udevsettle, which got 'corrected' to a call to udevadm. As far as I can
see, this is supposed to avoid a race condition.
What I am interested in is, which race condition that exactly is, and
what exactly cryptsetup still needs to do after that point.
I am asking since udev is not the only hotplugging mechanism. Personally
I feel that calling a specific external program is just not the right
way, to handle something like that.
For my usage case busybox is used together with busybox' mdev, thus a
call to udevsettle (or udevadm for that matter) is pretty much pointless.
Hi,
Clemens already merged my patch which removes this udev thing at all,
see my mail http://article.gmane.org/gmane.linux.kernel.device-mapper.dm-crypt/2804
Try version from upstream svn, it should work much more better with
your config
http://code.google.com/p/cryptsetup/source/checkout
We have cryptsetup with such patch in Fedora for a few weeks and
no bug reports.
Milan
--
mbroz@xxxxxxxxxx
---------------------------------------------------------------------
dm-crypt mailing list - http://www.saout.de/misc/dm-crypt/
To unsubscribe, e-mail: dm-crypt-unsubscribe@xxxxxxxx
For additional commands, e-mail: dm-crypt-help@xxxxxxxx
---------------------------------------------------------------------
dm-crypt mailing list - http://www.saout.de/misc/dm-crypt/
To unsubscribe, e-mail: dm-crypt-unsubscribe@xxxxxxxx
For additional commands, e-mail: dm-crypt-help@xxxxxxxx
