Re: [alsa-devel] [PATCH 2/2] ARM: clk: imx35 sound admux_gate bugfix
On Wed, Mar 27, 2013 at 12:57:00PM +0100, Philipp Zabel wrote:
Am Mittwoch, den 27.03.2013, 12:22 +0100 schrieb Markus Pargmann:
imx-ssi needs admux_gate clock to work on imx35. This patch registers this clock and enables it in the imx-ssi driver.
I have no idea what the admux_gate really is, but the ssi driver doesn't seem like the right place. Isn't this more closely related to the imx-audmux.c driver?
This is a patch to the ASoC driver? I just ignored it since it said it was an ARM patch... Anyway, what I'd expect should be happening here is that the clock API walks up the tree enabling supply clocks - that would mean that either that code isn't there or there's a missing parent child relationship in the data.
Hi Mark,
On Wed, Mar 27, 2013 at 12:37:00PM +0000, Mark Brown wrote:
On Wed, Mar 27, 2013 at 12:57:00PM +0100, Philipp Zabel wrote:
Am Mittwoch, den 27.03.2013, 12:22 +0100 schrieb Markus Pargmann:
imx-ssi needs admux_gate clock to work on imx35. This patch registers this clock and enables it in the imx-ssi driver.
I have no idea what the admux_gate really is, but the ssi driver doesn't seem like the right place. Isn't this more closely related to the imx-audmux.c driver?
This is a patch to the ASoC driver? I just ignored it since it said it was an ARM patch... Anyway, what I'd expect should be happening here is that the clock API walks up the tree enabling supply clocks - that would mean that either that code isn't there or there's a missing parent child relationship in the data.
It changes some things in imx-ssi, sorry, perhaps I should have splitted it into two patches.
admux clock does not seem to be a physical parent of the ssi clock, so I don't think it should be a parent in the clock tree. But imx-ssi (or some other evice) needs this clock to work. As mentioned in the other response, I will put it directly into clk-imx35.c so this won't be a patch for ASoC in the next version.
Regards,
Markus
On Wed, Mar 27, 2013 at 04:04:20PM +0100, Markus Pargmann wrote:
On Wed, Mar 27, 2013 at 12:37:00PM +0000, Mark Brown wrote:
This is a patch to the ASoC driver? I just ignored it since it said it was an ARM patch... Anyway, what I'd expect should be happening here is that the clock API walks up the tree enabling supply clocks - that would mean that either that code isn't there or there's a missing parent child relationship in the data.
It changes some things in imx-ssi, sorry, perhaps I should have splitted it into two patches.
admux clock does not seem to be a physical parent of the ssi clock, so I don't think it should be a parent in the clock tree. But imx-ssi (or some other evice) needs this clock to work. As mentioned in the other response, I will put it directly into clk-imx35.c so this won't be a patch for ASoC in the next version.
If it's the IP that needs the clock then I'd expect the IP to be requesting it so changing the driver does make sense. Do we actually know what the clock does?
On Wed, Mar 27, 2013 at 03:50:11PM +0000, Mark Brown wrote:
On Wed, Mar 27, 2013 at 04:04:20PM +0100, Markus Pargmann wrote:
On Wed, Mar 27, 2013 at 12:37:00PM +0000, Mark Brown wrote:
This is a patch to the ASoC driver? I just ignored it since it said it was an ARM patch... Anyway, what I'd expect should be happening here is that the clock API walks up the tree enabling supply clocks - that would mean that either that code isn't there or there's a missing parent child relationship in the data.
It changes some things in imx-ssi, sorry, perhaps I should have splitted it into two patches.
admux clock does not seem to be a physical parent of the ssi clock, so I don't think it should be a parent in the clock tree. But imx-ssi (or some other evice) needs this clock to work. As mentioned in the other response, I will put it directly into clk-imx35.c so this won't be a patch for ASoC in the next version.
If it's the IP that needs the clock then I'd expect the IP to be requesting it so changing the driver does make sense. Do we actually know what the clock does?
No we don't know exactly. imx35 reference manual does not give any further explanation about admux. But I just found this old patch: http://lists.infradead.org/pipermail/linux-arm-kernel/2009-November/003843.h...
So I tested to register admux as audmux clock for imx-audmux, which seems to solve the problem. However, I am not sure why the old patch of Sascha was not applied.
Regards,
Markus
participants (2)
-
Mark Brown -
Markus Pargmann