iio: st_pressure_spi: Add missing entries SPI to device ID table
PATCH 1/1 iio: st_pressure_spi: Add missing entries SPI to device ID table
Currently autoloading for SPI devices does not use the DT ID table, it uses SPI modalises. Supporting OF modalises is going to be difficult if not impractical, an attempt was made but has been reverted, so ensure that module autoloading works for this driver by adding SPI IDs for parts that only have a compatible listed.
--- drivers/iio/pressure/st_pressure_spi.c | 4 ++++ 1 file changed, 4 insertions(+)
drivers/iio/pressure/st_pressure_spi.c +4 -0
@@ -97,6 +97,10 @@ static const struct spi_device_id st_press_id_table[] = { { LPS33HW_PRESS_DEV_NAME }, { LPS35HW_PRESS_DEV_NAME }, { LPS22HH_PRESS_DEV_NAME }, + { "lps001wp-press" }, + { "lps25h-press", }, + { "lps331ap-press" }, + { "lps22hb-press" }, {}, };MODULE_DEVICE_TABLE (spi, st_press_id_table);
Signature · 1 line
2.20.1
Jonathan Cameron
maintainerQuoted 8 lines · On Mon, 27 Sep 2021 14:41:53 +0100 Mark Brown <[email protected]> wrote:
Currently autoloading for SPI devices does not use the DT ID table, it uses SPI modalises. Supporting OF modalises is going to be difficult if not impractical, an attempt was made but has been reverted, so ensure that module autoloading works for this driver by adding SPI IDs for parts that only have a compatible listed.
Fixes 96c8395e2166 ("spi: Revert modalias changes")Signed-off-by Mark Brown <[email protected]>
Whilst these IDs are deprecated, we should at least be consistent that they either work or not rather than current situation.
+CC Denis as driver author. I'll let it sit on list a little longer so Denis can take a look.
Thanks
Jonathan
Quoted 19 lines
--- drivers/iio/pressure/st_pressure_spi.c | 4 ++++ 1 file changed, 4 insertions(+)drivers/iio/pressure/st_pressure_spi.c +4
-0@@ -97,6 +97,10 @@ static const struct spi_device_id st_press_id_table[] = { { LPS33HW_PRESS_DEV_NAME }, { LPS35HW_PRESS_DEV_NAME }, { LPS22HH_PRESS_DEV_NAME }, + { "lps001wp-press" }, + { "lps25h-press", }, + { "lps331ap-press" }, + { "lps22hb-press" }, {}, };MODULE_DEVICE_TABLE (spi, st_press_id_table);
Hello Jonathan, Mark,
I am not very familiar with how much the kernel would like to keep 'probing id' consistent. I perfectly understand the value of doing this (maintain ID compatibility) but I also see increase confusion in maintaining half in a way and half in another. I personally think that we should drop the '-press' thing for all the devices since they all are single-chip (meaning that the name used identify univocally that is a pressure sensor).
If you think that compatibility is more important here, I think the patch is fine but this should be done in the i2c part as well so that it's at least congruent withing the driver.
Let me know what you think.
Thanks,
Denis
Quoted 53 lines
-----Original Message-----
From: Jonathan Cameron <[email protected]>
Sent: Thursday, September 30, 2021 9:39 AM
To: Mark Brown <[email protected]>
Cc: Lars-Peter Clausen <[email protected]>; [email protected]; Denis CIOCCA <[email protected]>
Subject: Re: [PATCH] iio:st_pressure_spi: Add missing entries SPI to device ID tableQuoted 8 lines · On Mon, 27 Sep 2021 14:41:53 +0100 Mark Brown <[email protected]> wrote:
Currently autoloading for SPI devices does not use the DT ID table, it uses SPI modalises. Supporting OF modalises is going to be difficult if not impractical, an attempt was made but has been reverted, so ensure that module autoloading works for this driver by adding SPI IDs for parts that only have a compatible listed.
Fixes 96c8395e2166 ("spi: Revert modalias changes")Signed-off-by Mark Brown <[email protected]>Whilst these IDs are deprecated, we should at least be consistent that they either work or not rather than current situation.
(quoted diff) +2
-0+CC Denis as driver author. I'll let it sit on list a little longer so +Denis cantake a look.
Thanks
Jonathan
drivers/iio/pressure/st_pressure_spi.c static const struct spi_device_id st_press_id_table[]
Quoted 10 lines
--- drivers/iio/pressure/st_pressure_spi.c | 4 ++++ 1 file changed, 4 insertions(+)drivers/iio/pressure/st_pressure_spi.c +0
-0@@ -97,6 +97,10 @@ static const struct spi_device_id st_press_id_table[]= {
Quoted 10 lines
drivers/iio/pressure/st_pressure_spi.c +4
-0{ LPS33HW_PRESS_DEV_NAME }, { LPS35HW_PRESS_DEV_NAME }, { LPS22HH_PRESS_DEV_NAME }, + { "lps001wp-press" }, + { "lps25h-press", }, + { "lps331ap-press" }, + { "lps22hb-press" }, {}, };MODULE_DEVICE_TABLE (spi, st_press_id_table);
Quoted 1 line · Denis CIOCCA
I am not very familiar with how much the kernel would like to keep 'probing id' consistent. I perfectly understand the value of doing this (maintain ID compatibility) but I also see increase confusion in maintaining half in a way and half in another.
The goal is not to maintain compatibility, the goal is to be able to load the driver as a module on DT systems. For historical reasons SPI uses the platform device IDs to load modules bound with DT, if there is no platform ID for a DT ID then userspace won't be able to find and load the module.
Quoted 1 line
I personally think that we should drop the '-press' thing for all the devices since they all are single-chip (meaning that the name used identify univocally that is a pressure sensor).
The DT bindings are an ABI, you can't really remove compatibles only deprecate them.
Quoted 1 line
If you think that compatibility is more important here, I think the patch is fine but this should be done in the i2c part as well so that it's at least congruent withing the driver.
I2C doesn't have this issue with modaliases so it's not an issue there.
Please fix your mail client to word wrap within paragraphs at something substantially less than 80 columns. Doing this makes your messages much easier to read and reply to.
Hello Mark,
Quoted 16 lines
-----Original Message-----
From: Mark Brown <[email protected]>
Sent: Friday, October 1, 2021 5:20 AM
To: Denis CIOCCA <[email protected]>
Cc: Jonathan Cameron <[email protected]>; Lars-Peter Clausen <[email protected]>; [email protected]
Subject: Re: [PATCH] iio:st_pressure_spi: Add missing entries SPI to device ID tableQuoted 1 line · Denis CIOCCA
I am not very familiar with how much the kernel would like to keep
'probing id' consistent. I perfectly understand the value of doing this (maintain ID compatibility) but I also see increase confusion in maintaining half in a way and half in another.
Quoted 5 lines
The goal is not to maintain compatibility, the goal is to be able to load the driver as a module on DT systems. For historical reasons SPI uses the platform device IDs to load modules bound with DT, if there is no platform ID for a DT ID then userspace won't be able to find and load the module.
Ok now it is clear. I wasn't aware of that.
In this case it is good to me (I didn't do any testing).
Quoted 7 lines
Quoted 2 lines
I personally think that we should drop the '-press' thing for all the devices
since they all are single-chip (meaning that the name used identify univocally that is a pressure sensor).
The DT bindings are an ABI, you can't really remove compatibles only deprecate them.
Yeah I guessed this was the case.
Quoted 6 lines
Quoted 2 lines
If you think that compatibility is more important here, I think the patch is
fine but this should be done in the i2c part as well so that it's at least congruent withing the driver.
I2C doesn't have this issue with modaliases so it's not an issue there.
Clear.
Quoted 4 lines
Please fix your mail client to word wrap within paragraphs at something substantially less than 80 columns. Doing this makes your messages much easier to read and reply to.
Quoted 18 lines · On Thu, 7 Oct 2021 00:47:56 +0000 Denis CIOCCA <[email protected]> wrote:
Hello Mark,
Quoted 16 lines
-----Original Message-----
From: Mark Brown <[email protected]>
Sent: Friday, October 1, 2021 5:20 AM
To: Denis CIOCCA <[email protected]>
Cc: Jonathan Cameron <[email protected]>; Lars-Peter Clausen <[email protected]>; [email protected]
Subject: Re: [PATCH] iio:st_pressure_spi: Add missing entries SPI to device ID tableQuoted 1 line · Denis CIOCCA
I am not very familiar with how much the kernel would like to keep
'probing id' consistent. I perfectly understand the value of doing this (maintain ID compatibility) but I also see increase confusion in maintaining half in a way and half in another.
Quoted 8 lines
Quoted 5 lines
The goal is not to maintain compatibility, the goal is to be able to load the driver as a module on DT systems. For historical reasons SPI uses the platform device IDs to load modules bound with DT, if there is no platform ID for a DT ID then userspace won't be able to find and load the module.
Ok now it is clear. I wasn't aware of that.
In this case it is good to me (I didn't do any testing).
Applied to the fixes-togreg branch of iio.git. I haven't explicitly marked it for stable but it may make sense to backport at somepoint.
Thanks,
Jonathan
Quoted 24 lines
Quoted 8 lines
Quoted 3 lines
I personally think that we should drop the '-press' thing for all the devices
since they all are single-chip (meaning that the name used identify univocally that is a pressure sensor).
The DT bindings are an ABI, you can't really remove compatibles only deprecate them.
Yeah I guessed this was the case.
Quoted 6 lines
Quoted 2 lines
If you think that compatibility is more important here, I think the patch is
fine but this should be done in the i2c part as well so that it's at least congruent withing the driver.
I2C doesn't have this issue with modaliases so it's not an issue there.
Clear.
Quoted 4 lines
Please fix your mail client to word wrap within paragraphs at something substantially less than 80 columns. Doing this makes your messages much easier to read and reply to.