On 3/31/2011 7:28 AM, Carsten Munk wrote:
2011/3/31 Arjan van de Ven<[email protected]>:
On 3/31/2011 7:07 AM, Markus Lehtonen wrote:
Hi,
I have made the following changes to fastinit in project Trunk:Testing.
Please review and accept ASAP.
Thank You,
Markus Lehtonen
[This message was auto-generated]
---
Request #15538:
submit: home:mlehtonen:branches:Trunk:Testing/fastinit(r2)(cleanup) ->
Trunk:Testing/fastinit
Message:
* Thu Mar 31 2011 Markus Lehtonen<[email protected]> - 1.28
- Sensorfw: support for a device-specific .modules file to load required
kernel modules before sensord starts. Needed to fix BMC#11699
this is the wrong solution though; udev not fastinit need to take care fo
this
Agreed, the problem here is that udev has probably not settled by the
time sensord starts (it is in rc.sysinit afterall and start_udev is in
a backgrounded nice -20) and finished it's initialization.
so call udev settle either as part of sensord or as part of the start
lines that start sensord ?
the current thing gets you two concurrent module loading paths --> bad.
The best thing would be if sensord had udev support instead of a
static configuration file, or in some way wait for the devices
mentioned in the configuration instead of aborting when it can't find
them?
I would say that is a hard requirement for sensord to be useful..
That'd probably require substantial code changes which may breach
feature freeze - should we have a proper solution in 1.3 instead?
I would call this a serious design bug, and one that needs fixing.
_______________________________________________
MeeGo-packaging mailing list
[email protected]
http://lists.meego.com/listinfo/meego-packaging
