Hi,
On 27-06-18 06:40, Pierre-Louis Bossart wrote:
On 6/26/18 2:47 AM, Hans de Goede wrote:
Hi,
On 25-06-18 22:01, Pierre-Louis Bossart wrote:
On 6/24/18 9:06 AM, Hans de Goede wrote:
Hi Mark,
This series is mainly a cleanup series. In the beginning of the rt5651 machine driver some wrong assumptions were made, such as the headset mic being attached to IN2 (it is on IN3 on the 7 machines I have access to and on all otherwise known machines).
We use a rt5651 reference board for this codec connector to the MinnowBoard. I am pretty sure for that board IN2 is used for the headset. Please give me a couple of days to double-check.
According to this commit:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/so...
It is connected to IN3 on the Minnowboard Max with Realtek rt5651 eval board.
And you yourself added a quirk for it being on IN3 for the MinnowBoard Turbot: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/so...
Note I later renamed the IN3_MAP to IN1_HS_IN3 to make clear that it is for internal mic on IN1, hs-mis on IN3, as can be seen in: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/so... (which is the commit adding both the quirk and updating the Minnowboard Max DMI entry to use this quirk).
Yes, you are right that headset is connected on IN3 but I wasn't completely crazy with IN2 - it's used for the line in, I just confused connectors...
The cleanups look ok and well documented, I just find that the simplification of patch 7/9 goes one bridge too far. If we ever get a quirk where the headset is not on IN3 we'll have limited ways of finding the right UCM files. Keeping hs-in3 in the filename is not big deal - your call really.
Not having the hs-in3 in the longname matches what we are doing for the rt5640, where the headset mic also always is on the same input.
For consistency I think it is good to drop it, so lets keep this patch set as is.
FWIW, all Reviewed-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com
Thanks.
Regards,
Hans
And also adding a quirk for a machine with the intmic on IN2, which later got fixed with a new quirk for machines with 2 internal mics on both IN1 and IN2 and moving the one machine with the IN2 quirk over to the new IN1_IN2 quirk, leaving the IN2 quirk as an orphan quirk for non existing hardware.
Then I made a similar mistake adding the IN2_HS_IN3 quirk, while I should have used the IN1_HS_IN3 (which itself only exists because the original IN1 quirk has the headset input mapping wrong),
TL;DR: it is a bit of a mess due to a number of wrong assumptions about how the inputs where actually routed in the past.
This series cleans this all up in small commits / one bit a time and refers the original commit messages in its commit messages.
Note patch 10/11 is not a cleanup patch, but is more or less the reason I took a second look at all the quirks.
Regards,
Hans
Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel