Hello Fabio,
Hi Marek,
(Sorry for the top-posting here)
There is a delay of 23 seconds, which is 1 second for each sgtl500- register write.
This is an I2C issue and we have already talked about this on the linux-arm-kernel list.
Alexandre Belloni also sees this 1 second on another I2C device connected to mx28.
There was no problem without these patches though. With your patches, I see NAK happening on the I2C lines upon first 2-byte write.
Regards,
Fabio Estevam ________________________________________ From: Marek Vasut [marex@denx.de] Sent: Thursday, May 30, 2013 5:36 PM To: alsa-devel@alsa-project.org Cc: Mark Brown; Estevam Fabio-R49496 Subject: Re: [alsa-devel] [PATCH] ASoC: sgtl5000: Do not read registers prior to turning on the supplies
Hi Mark, Fabio,
On Tue, May 28, 2013 at 11:04:18AM -0300, Fabio Estevam wrote:
Commit b871f1ad (ASoC: sgtl5000: Read SGTL5000_CHIP_ID in i2c_probe()) placed the code for reading the codec revision prior to turning on the power supplies.
Even though this works on some systems that always have the codec power supplies enabled, this is not correct, so revert this commit.
It seems like a better fix for this is to just enable the supplies while doing the device identification?
This patch does kinda fix it for me, but the system takes quite some time to init the soundcard now (a few seconds). After reverting these two patches, the soundcard works just fine (like before):
ASoC: sgtl5000: Fix driver probe after reset ASoC: sgtl5000: Read SGTL5000_CHIP_ID in i2c_probe()Best regards, Marek Vasut
Best regards, Marek Vasut