I guess we should still strive to implement it with as good accuracy as hardware offers, in case there will be a change in-firmware to begin actually using this. Of course it won't be perfect on all hardware so we'll just do whatever is possible.
It might be useful in the "pause" case after all to be able to report an accurate position in the media file. (but I mean we already have an up-to-ms or up-to-2ms accuracy even without this, based on the size of the elem, of how much goes through the pipeline in a single pipeline_copy() call)
So I guess: Good accuracy within what the hardware tells us, but low priority esp. since for the moment it is unused.
________________________________________ From: Daniel Baluta daniel.baluta@gmail.com Sent: Tuesday, April 21, 2020 4:12 PM To: Paul Olaru (OSS) Cc: sound-open-firmware@alsa-project.org; Lauda, Tomasz; Liam Girdwood; Paul Olaru Subject: Re: [Sound-open-firmware] What is the role of dma_status function?
On Fri, Apr 17, 2020 at 1:53 PM Paul Olaru (OSS) paul.olaru@oss.nxp.com wrote:
I think there is an intended use which isn't implemented. If I look at the dai_position function, there is this comment:
/* TODO: improve accuracy by adding current DMA position */
This is probably a place where dma_status would be called in the future/is intended to be called in the future.
The same comment is visible in the host_position function.
Yes, this makes sense. Although it is not used.