On 04/20/2011 07:09 AM, Layne Edwards wrote:
On Tue, 05 Apr 2011 13:58:49 +0200, D.S. Ljungmark wrote:
On Tue, 2011-04-05 at 13:48 +0200, Jo-Philipp Wich wrote:
Hi.
Currently the ext-blockroot solution feels very racy, and the init
scripts sometimes end up in a fun state due to being started before
overlays are in place and not. ( what is it, S93 and forwards that
are "safe" to use? Not much )
You're doing something very wrong then as init is not even started
before the external overlay is finished, all the preparation happens
in preinit.
Hmm, really? From what I can see, after /etc/init.d/done is the first
place in where you can use services that assume the overlay is mounted
and ready?
I'm having a similar problem. My 3G modem does not switch modes on
coldboot (usb-modeswitch works great on a post-boot hotplug).
However, it looks like the usb-modeswitch hotplug event is being
executed on coldboot (per the syslog)... but it occurs before the USB
serial drivers are loaded, and it does not switch. I'm not sure what
the solution for this is... but a slight delay in the modeswitch
hotplug event may solve the problem (however that can be
accomplished). Although not an elegant solution, building the USB
serial drivers in the kernel should load them quicker (before the
hotplug event).
Suggestions?
Layne
Since I didn't get any real feedback on how it _should_ be solved
properly, I just added an ugly shellscript to parse through the usb-bus
and perform the same operations that I wanted coldplug to do. It's
non-portable and ugly, but got the job done enough. Painful though as I
hate having to maintain such a thing in the future :/
//D.S.
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel
