Hi Matthieu
On Wednesday 31 of August 2005 11:25, you wrote:
> yes but as we call wait_cmv_ack before request_cmvs, we should wait that 
> uea_load_page upload enough dsp for the modem to be ready, and we 
> shouldn't have a race ?
Right, problem was because we don't check wait_cmv_ack return value, 
when wait_cmv_ack fail on timeout there was possibility to simultaneous 
request_firmware() calls. 

> > Other solution is fix firmware_class such the directory have name 
> > bus_id/firmware_name
> > but this is quite difficult and of course there are problems to put patch 
> > upstream.
> We could report this on ML, we'll saw what think firmware_class mainteners.
I think firmware_class is orphaned, for us it doesn't matter now.

Regards
Staszek Gruszka

Reply via email to