On 04/05/2018 05:44 PM, Lars-Peter Clausen wrote: > On 04/03/2018 10:05 AM, Robert Rosengren wrote: >> From: Danny Smith <dannys@xxxxxxxx> >> >> DSP_RUN needs to be disabled during firmware write otherwise >> we can end up with undefined behavior if writing to a dsp which >> is already running firmware. >> > > Good point, thanks. > > I believe there is a possible (but rare under normal circumstances) race > condition here though, where DAPM could run at the same time and > enable/disable the DSP while the firmware is being loaded. > > To avoid this the firmware load sequence should be encapsulated in > snd_soc_dapm_mutex_lock()/snd_soc_dapm_mutex_unlock(). > > Pass struct snd_soc_component instead of struct adau to this function and > then use snd_soc_component_get_dapm() to get the DAPM context. > > For the next version of the patches please also add the ASoC maintainers to > the reciver list since they have to apply the patch. The maintainers are: > > Liam Girdwood <lgirdwood@xxxxxxxxx> > Mark Brown <broonie@xxxxxxxxxx> > Thanks for your comments! New patches are already sent for your review :) / Robert > >> Signed-off-by: Danny Smith <dannys@xxxxxxxx> >> --- >> sound/soc/codecs/adau17x1.c | 8 +++++++- >> 1 file changed, 7 insertions(+), 1 deletion(-) _______________________________________________ Alsa-devel mailing list Alsa-devel@xxxxxxxxxxxxxxxx http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
