> >/ I'm particularly confused about when to get, set, keep,
> />/ and release client ID's. My dongle doesn't seem to require a client ID
> />/ except for starting and stopping the connection. Is this different for
> />/ other dongles?
> /
> You'll generally need a subsystem client ID for all commands.  The only
> exception is for the QMI_CTL system, which includes the get/release
> client ID commands and other generic management stuff.
>
> All devices I've encountered will release all resources allocated by a
> client when the client ID is released.  For the QMI_WDS system that
> means that any activated session will be taken down.  So you usually
> need to
>  - allocate a QMI_WDS client ID
>  - connect
>  - keep the client for as long as your session lasts
>  - (optionally disconnect)
>  - release  QMI_WDS client

Thanks. I see now that uqmi does the allocation and release for me
automatically. For keeping the connection up I obviously need to prevent
the release, and this seems to happen when I get or set the client ID
explicitly. For some reason I couldn't disconnect by explicitly
releasing the client ID, but maybe that is just a user error... That was
also why I implemented stop-network. I'll need to look at uqmi sources
more carefully to understand this.
_______________________________________________
openwrt-devel mailing list
[email protected]
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

Reply via email to