Hi Prakash, On Thu, Nov 10, 2011 at 11:43:21, Manjunathappa, Prakash wrote: > Since there was bug in the audio driver, it was not picking the eventq > specified(EVENTQ_1) via platform data and was using EVENTQ_0. And in > system scenario other modules(say video) were using EVENTQ_1. > 48519f0ae03bc7e86b3dc93e56f1334d53803770(ASoC: davinci: let platform > data define edma queue numbers) fixes the bug in driver to pick > specified eventq via platform data. As a result starvation issue is > observed on audio side when audio/video uses same eventq for transfers. > Patch fixes the issue by changing eventq to EVENTQ_0 for audio. > > Signed-off-by: Manjunathappa, Prakash <[email protected]>
This description was tough for me to parse. After reading through it couple of times, here is what I think the description could have been: " On OMAP-L138 platform, it is desirable to use EDMA event queue 0 for audio transfers so that they are not starved by video data moving on event queue 1. Commit 48519f0ae03bc7e86b3dc93e56f1334d53803770 (ASoC: davinci: let platform data define edma queue numbers) had a side-effect of changing this behavior by making the driver actually honor the platform data passed. Fix this now by passing event queue 0 as the queue to be used for audio transfers. " The headline should have been: ARM: davinci: da850 evm: change audio edma event queue to EVENTQ_0 Notice the use of proper prefixes per current conventions being used in ARM and the dropping of tautological "queue eventq" Also, in future, please do not fail to copy Linux ARM kernel list on kernel patches. I am making an exception this time and not asking you to resubmit since this a fairly trivial change and I got to reviewing this pretty late. Since the bug fix is applicable to previous kernel versions, I will add the stable tag while committing. Thanks, Sekhar _______________________________________________ Davinci-linux-open-source mailing list [email protected] http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
