Re: [alsa-devel] [PATCH] sound/tlv320dac33: Add device tree support
On Tue, Jan 30, 2018 at 10:11:02AM +0100, Filip Matijević wrote:
Hi,
On 01/30/2018 09:53 AM, Pavel Machek wrote:
On Tue 2018-01-30 09:34:46, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 12:33:01AM +0100, Pavel Machek wrote:
On Tue 2018-01-30 00:20:31, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 12:05:39AM +0100, Pavel Machek wrote:
This adds device tree support to tlv320dac33.c.
Signed-off-by: Pavel Machek pavel@ucw.cz
diff --git a/Documentation/devicetree/bindings/sound/tlv320dac33.txt b/Documentation/devicetree/bindings/sound/tlv320dac33.txt new file mode 100644 index 0000000..6cbd311 --- /dev/null +++ b/Documentation/devicetree/bindings/sound/tlv320dac33.txt @@ -0,0 +1,32 @@ +Texas Instruments - tlv320dac33 Codec module
+The tlv320dac33 serial control bus communicates through I2C protocols.
+Required properties:
+- compatible - "ti,tlv320dac33" +- reg - I2C slave address
+Optional properties:
+- power-gpios - gpio pin to power the device, active high
While driver used gpio in platform data, isn't it more likely regulator which powers device?
power-gpios = <&gpio2 28 0>; /* gpio_60 */Looks like GPIO to me -- example is from Nokia N9. So this appears to be correct.
Device datasheet doesn't list any pin which looks like "power-gpio" http://www.ti.com/lit/ds/symlink/tlv320dac32.pdf Unfortunately I do not know much about N9, but was able to find Nokia 5610 scheme to get clue how could be tlv320dac33 hardwired (see page 2): http://mastermobile.spb.ru/service/nokia_5610_rm-242_service_schematics.pdf Here AVDD is powered by LP3985 voltage regulator which is enabled using VEN pin which might be connected to gpio. Or there could be completely different voltage regulator with different controls. And since Linux already has voltage regulator class, lets not limit ourselves to gpio pins.
Well, notice I'm converting existing driver to device tree. And that one already has GPIO dependency. It is possible that more work needs to be done there, but that should not be a reason to delay this. Feel free to help.
Adding DT properties that need to be maintained for compatibility reasons is a bad idea and very good reason to delay merging unfinished stuff. And meanwhile it turned out it is not power-gpio :)
According to N9 schematics http://www.s-manuals.com/manuals/phone/nokia/nokia_n9_rm-696_service_schemat... it's in fact GPIO pin that is connected to reset line (labeled CODEC_RST). So calling it "power" might be misleading, but the driver code is quite clear as it labels that GPIO as "tlv320dac33 reset"
Thanks a lot! So now it is clear that gpio is connected to codec's nRESET pin, therefore "reset-gpios" property seems reasonable here.
ladis
On Tue 2018-01-30 10:38:38, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 10:11:02AM +0100, Filip Matijević wrote:
Hi,
Well, notice I'm converting existing driver to device tree. And that one already has GPIO dependency. It is possible that more work needs to be done there, but that should not be a reason to delay this. Feel free to help.
Adding DT properties that need to be maintained for compatibility reasons is a bad idea and very good reason to delay merging unfinished stuff. And meanwhile it turned out it is not power-gpio :)
I believe reset-gpios and power-gpios are commonly used like this... and that's what the old code does.
You are not helping.
Pavel
On Tue, Jan 30, 2018 at 11:00:23AM +0100, Pavel Machek wrote:
On Tue 2018-01-30 10:38:38, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 10:11:02AM +0100, Filip Matijević wrote:
Hi,
Well, notice I'm converting existing driver to device tree. And that one already has GPIO dependency. It is possible that more work needs to be done there, but that should not be a reason to delay this. Feel free to help.
Adding DT properties that need to be maintained for compatibility reasons is a bad idea and very good reason to delay merging unfinished stuff. And meanwhile it turned out it is not power-gpio :)
I believe reset-gpios and power-gpios are commonly used like this... and that's what the old code does.
Why do you care about old code when introducing new DT property? Either it is reset, then lets call it reset-gpios or it is power supply and then voltage regulator should be used (VAUX4.OUT is such a regulator although it is unclear to me how it is controlled (*)).
You are not helping.
The only way I can help here is to resend your patch with "reset-gpios" used, which I'm pretty sure you can handle yourself.
(*) Remark "DAC33 enabled (for example, plug in a headset to the 3.5mm AV connector and start music playback from the device)" in schematics suggests it is controlled somehow.
On Tue 2018-01-30 11:10:46, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 11:00:23AM +0100, Pavel Machek wrote:
On Tue 2018-01-30 10:38:38, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 10:11:02AM +0100, Filip Matijević wrote:
Hi,
Well, notice I'm converting existing driver to device tree. And that one already has GPIO dependency. It is possible that more work needs to be done there, but that should not be a reason to delay this. Feel free to help.
Adding DT properties that need to be maintained for compatibility reasons is a bad idea and very good reason to delay merging unfinished stuff. And meanwhile it turned out it is not power-gpio :)
I believe reset-gpios and power-gpios are commonly used like this... and that's what the old code does.
Why do you care about old code when introducing new DT property? Either it is reset, then lets call it reset-gpios or it is power supply and then voltage regulator should be used (VAUX4.OUT is such a regulator although it is unclear to me how it is controlled (*)).
power gpio = !reset gpio. Difference is only in polarity.
You are not helping.
The only way I can help here is to resend your patch with "reset-gpios" used, which I'm pretty sure you can handle yourself.
Well, you can do that, and then you can argue with the next person who feels one of the properties has to get his preferred color. Hard part is not changing code :-(.
Pavel
(*) Remark "DAC33 enabled (for example, plug in a headset to the 3.5mm AV connector and start music playback from the device)" in schematics suggests it is controlled somehow.
On Tue, Jan 30, 2018 at 11:35:38AM +0100, Pavel Machek wrote:
On Tue 2018-01-30 11:10:46, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 11:00:23AM +0100, Pavel Machek wrote:
On Tue 2018-01-30 10:38:38, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 10:11:02AM +0100, Filip Matijević wrote:
Hi,
Well, notice I'm converting existing driver to device tree. And that one already has GPIO dependency. It is possible that more work needs to be done there, but that should not be a reason to delay this. Feel free to help.
Adding DT properties that need to be maintained for compatibility reasons is a bad idea and very good reason to delay merging unfinished stuff. And meanwhile it turned out it is not power-gpio :)
I believe reset-gpios and power-gpios are commonly used like this... and that's what the old code does.
Why do you care about old code when introducing new DT property? Either it is reset, then lets call it reset-gpios or it is power supply and then voltage regulator should be used (VAUX4.OUT is such a regulator although it is unclear to me how it is controlled (*)).
power gpio = !reset gpio. Difference is only in polarity.
Quick grep: Documentation/devicetree/bindings/net/smsc-lan91c111.txt - power-gpios: GPIO to control the PWRDWN pin - reset-gpios: GPIO to control the RESET pin Documentation/devicetree/bindings/net/wireless/ti,wl1251.txt - ti,power-gpio : GPIO connected to chip's PMEN pin Documentation/devicetree/bindings/input/touchscreen/silead_gsl1680.txt - power-gpios : Specification for the pin connected to the gsl1680's shutdown input. This needs to be driven high to take the gsl1680 out of its low power state Documentation/devicetree/bindings/input/touchscreen/ektf2127.txt - power-gpios : GPIO specification for the pin connected to the ektf2127's wake input. This needs to be driven high to take ektf2127 out of it's low power state
As you can see others are using it to drive pins different from nRESET. As stated in datasheet: "The TLV320DAC32 requires a hardware reset after power-up for proper operation. After all power supplies are at their specified values, the nRESET pin must be driven low for at least 10ns. If this reset sequence is not performed, the DAC32 may not respond properly to register reads/writes". That does not sound like anything to do with power.
(It seems the only difference between TLV320DAC33 and TLV320DAC32 is uses a ball grid array package vs QFN32)
You are not helping.
The only way I can help here is to resend your patch with "reset-gpios" used, which I'm pretty sure you can handle yourself.
Well, you can do that, and then you can argue with the next person who feels one of the properties has to get his preferred color. Hard part is not changing code :-(.
Indeed, hard part is not to break DT compatibility later. Consider someone will have to add regulator support later, which you omitted to do - having both power supply and power gpio is a bit confusing, don't you think?
Anyway, time to stop arguing, feel free to do what you think is right, I do not have anything important to add :)
ladis
On 2018-01-30 13:38, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 11:35:38AM +0100, Pavel Machek wrote:
On Tue 2018-01-30 11:10:46, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 11:00:23AM +0100, Pavel Machek wrote:
On Tue 2018-01-30 10:38:38, Ladislav Michl wrote:
On Tue, Jan 30, 2018 at 10:11:02AM +0100, Filip Matijević wrote:
Hi,
> Well, notice I'm converting existing driver to device tree. And that > one already has GPIO dependency. It is possible that more work needs > to be done there, but that should not be a reason to delay this. Feel > free to help.
Adding DT properties that need to be maintained for compatibility reasons is a bad idea and very good reason to delay merging unfinished stuff. And meanwhile it turned out it is not power-gpio :)
I believe reset-gpios and power-gpios are commonly used like this... and that's what the old code does.
Why do you care about old code when introducing new DT property? Either it is reset, then lets call it reset-gpios or it is power supply and then voltage regulator should be used (VAUX4.OUT is such a regulator although it is unclear to me how it is controlled (*)).
power gpio = !reset gpio. Difference is only in polarity.
Quick grep: Documentation/devicetree/bindings/net/smsc-lan91c111.txt
- power-gpios: GPIO to control the PWRDWN pin
- reset-gpios: GPIO to control the RESET pin
Documentation/devicetree/bindings/net/wireless/ti,wl1251.txt
- ti,power-gpio : GPIO connected to chip's PMEN pin
Documentation/devicetree/bindings/input/touchscreen/silead_gsl1680.txt
- power-gpios : Specification for the pin connected to the gsl1680's shutdown input. This needs to be driven high to take the gsl1680 out of its low power state
Documentation/devicetree/bindings/input/touchscreen/ektf2127.txt
- power-gpios : GPIO specification for the pin connected to the ektf2127's wake input. This needs to be driven high to take ektf2127 out of it's low power state
As you can see others are using it to drive pins different from nRESET. As stated in datasheet: "The TLV320DAC32 requires a hardware reset after power-up for proper operation. After all power supplies are at their specified values, the nRESET pin must be driven low for at least 10ns. If this reset sequence is not performed, the DAC32 may not respond properly to register reads/writes". That does not sound like anything to do with power.
(It seems the only difference between TLV320DAC33 and TLV320DAC32 is uses a ball grid array package vs QFN32)
Note that tlv320dac32 != tlv320dac33, they are not even register compatible devices... dac33 was created _for_ Nokia back in the days and as far as I know there is no publicly available documentation for dac33.
And yes, my pick to use power_gpio in the code was wrong, it should have been named as reset_gpio.
You are not helping.
The only way I can help here is to resend your patch with "reset-gpios" used, which I'm pretty sure you can handle yourself.
Well, you can do that, and then you can argue with the next person who feels one of the properties has to get his preferred color. Hard part is not changing code :-(.
Indeed, hard part is not to break DT compatibility later. Consider someone will have to add regulator support later, which you omitted to do - having both power supply and power gpio is a bit confusing, don't you think?
Anyway, time to stop arguing, feel free to do what you think is right, I do not have anything important to add :)
ladis _______________________________________________ Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
- Péter
Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki. Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki
participants (3)
-
Ladislav Michl -
Pavel Machek -
Peter Ujfalusi