On Tue, 05.02.13 06:42, Sérgio Basto (sergio(a)serjux.com) wrote:
On Ter, 2013-01-29 at 17:59 +0000, Sérgio Basto wrote:
> On Ter, 2013-01-29 at 01:52 +0100, Lennart Poettering wrote:
> > Yes, even then. udev will notice rules dropped there.
>
> OK I will test that
I just test it and doesn't assign USB devices.
Yes, and you shouldn't trigger those devices.
It's NOT OK to randomly trigger USB devices, that's a total NO-NO. You
can suggest the user to reboot or so, or replug the device, but just
retriggering the whole USB subsystem is explicitly not OK, and if Kay
finds you doing this in packager installer codepaths you might catch
yourself a bloody nose. ;-)
But seeing what udev.service , udev-trigger.service and
udev-settle.service do on :
systemctl restart udev.service
Nope. Stay away from that. That's even worse.
systemctl restart udev-trigger.service
Stay away from this too, see above.
systemctl restart udev-settle.service
And definitely stay away from this as well. udev-settle is a legacy
service for broken code such as LVM, and nothing else really needs that anymore.
I back to old scripts
# Assign USB devices
if /sbin/udevadm control --reload-rules >/dev/null 2>&1
This bit is entirely unnecessary. udev does this implicitly.
then
/sbin/udevadm trigger --subsystem-match=usb >/dev/null 2>&1 || :
Nope, forbidden. Don't do this.
/sbin/udevadm settle >/dev/null 2>&1 || :
Totally pointless... You are not LVM during boot, so stay away from this.
Lennart
--
Lennart Poettering - Red Hat, Inc.