Skip to contentSkip to conversation

ath9k: fix build error with LEDS_CLASS=m

PATCH In review Arnd Bergmann 4 participants 10 messages Jan 25, 2021 – Jan 28, 2021

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ath9k: fix build error with LEDS_CLASS=m

From: Arnd Bergmann <[email protected]>

When CONFIG_ATH9K is built-in but LED support is in a loadable module, both ath9k drivers fails to link:

x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_deinit_leds':
gpio.c:(.text+0x36): undefined reference to `led_classdev_unregister'
x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_init_leds':
gpio.c:(.text+0x179): undefined reference to `led_classdev_register_ext'

The problem is that the 'imply' keyword does not enforce any dependency but is only a weak hint to Kconfig to enable another symbol from a defconfig file.

Change imply to a 'depends on LEDS_CLASS' that prevents the incorrect configuration but still allows building the driver without LED support.

The 'select MAC80211_LEDS' is now ensures that the LED support is actually used if it is present, and the added Kconfig dependency on MAC80211_LEDS ensures that it cannot be enabled manually when it has no effect.

Fixes 197f466e93f5 ("ath9k_htc: Do not select MAC80211_LEDS by default")
---
 drivers/net/wireless/ath/ath9k/Kconfig | 8 ++------
 net/mac80211/Kconfig                   | 2 +-
 2 files changed, 3 insertions(+), 7 deletions(-)
drivers/net/wireless/ath/ath9k/Kconfig +2 -6
index a84bb9b6573f..e150d82eddb6 100644
@@ -21,11 +21,9 @@ config ATH9K_BTCOEX_SUPPORT
 config ATH9K
 	tristate "Atheros 802.11n wireless cards support"
 	depends on MAC80211 && HAS_DMA
+	select MAC80211_LEDS if LEDS_CLASS=y || LEDS_CLASS=MAC80211
 	select ATH9K_HW
 	select ATH9K_COMMON
-	imply NEW_LEDS
-	imply LEDS_CLASS
-	imply MAC80211_LEDS
 	help
 	  This module adds support for wireless adapters based on
 	  Atheros IEEE 802.11n AR5008, AR9001 and AR9002 family
@@ -176,11 +174,9 @@ config ATH9K_PCI_NO_EEPROM
 config ATH9K_HTC
 	tristate "Atheros HTC based wireless cards support"
 	depends on USB && MAC80211
+	select MAC80211_LEDS if LEDS_CLASS=y || LEDS_CLASS=MAC80211
 	select ATH9K_HW
 	select ATH9K_COMMON
-	imply NEW_LEDS
-	imply LEDS_CLASS
-	imply MAC80211_LEDS
 	help
 	  Support for Atheros HTC based cards.
 	  Chipsets supported: AR9271
net/mac80211/Kconfig +1 -1
index cd9a9bd242ba..51ec8256b7fa 100644
@@ -69,7 +69,7 @@ config MAC80211_MESH
 config MAC80211_LEDS
 	bool "Enable LED triggers"
 	depends on MAC80211
-	depends on LEDS_CLASS
+	depends on LEDS_CLASS=y || LEDS_CLASS=MAC80211
 	select LEDS_TRIGGERS
 	help
 	  This option enables a few LED triggers for different
Signature · 1 line
2.29.2
#1 lore raw
Quoted 15 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

When CONFIG_ATH9K is built-in but LED support is in a loadable module, both ath9k drivers fails to link:

x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_deinit_leds':
gpio.c:(.text+0x36): undefined reference to `led_classdev_unregister'
x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_init_leds':
gpio.c:(.text+0x179): undefined reference to `led_classdev_register_ext'

The problem is that the 'imply' keyword does not enforce any dependency but is only a weak hint to Kconfig to enable another symbol from a defconfig file.

Quoted 7 lines

Change imply to a 'depends on LEDS_CLASS' that prevents the incorrect configuration but still allows building the driver without LED support.

The 'select MAC80211_LEDS' is now ensures that the LED support is actually used if it is present, and the added Kconfig dependency on MAC80211_LEDS ensures that it cannot be enabled manually when it has no effect.

But we do not want to have this dependency (selecting MAC80211_LEDS). I fixed this problem here:
https://lore.kernel.org/lkml/[email protected]/
Maybe let's take this approach?

Best regards,
Krzysztof

Quoted 54 lines
Fixes 197f466e93f5 ("ath9k_htc: Do not select MAC80211_LEDS by default")
---
 drivers/net/wireless/ath/ath9k/Kconfig | 8 ++------
 net/mac80211/Kconfig                   | 2 +-
 2 files changed, 3 insertions(+), 7 deletions(-)
drivers/net/wireless/ath/ath9k/Kconfig +2 -6
index a84bb9b6573f..e150d82eddb6 100644
@@ -21,11 +21,9 @@ config ATH9K_BTCOEX_SUPPORT
 config ATH9K
        tristate "Atheros 802.11n wireless cards support"
        depends on MAC80211 && HAS_DMA
+       select MAC80211_LEDS if LEDS_CLASS=y || LEDS_CLASS=MAC80211
        select ATH9K_HW
        select ATH9K_COMMON
-       imply NEW_LEDS
-       imply LEDS_CLASS
-       imply MAC80211_LEDS
        help
          This module adds support for wireless adapters based on
          Atheros IEEE 802.11n AR5008, AR9001 and AR9002 family
@@ -176,11 +174,9 @@ config ATH9K_PCI_NO_EEPROM
 config ATH9K_HTC
        tristate "Atheros HTC based wireless cards support"
        depends on USB && MAC80211
+       select MAC80211_LEDS if LEDS_CLASS=y || LEDS_CLASS=MAC80211
        select ATH9K_HW
        select ATH9K_COMMON
-       imply NEW_LEDS
-       imply LEDS_CLASS
-       imply MAC80211_LEDS
        help
          Support for Atheros HTC based cards.
          Chipsets supported: AR9271
net/mac80211/Kconfig +1 -2
index cd9a9bd242ba..51ec8256b7fa 100644
@@ -69,7 +69,7 @@ config MAC80211_MESH
 config MAC80211_LEDS
        bool "Enable LED triggers"
        depends on MAC80211
-       depends on LEDS_CLASS
+       depends on LEDS_CLASS=y || LEDS_CLASS=MAC80211
        select LEDS_TRIGGERS
        help
          This option enables a few LED triggers for different
--

2.29.2

#2 lore raw
Quoted 24 lines · Krzysztof Kozlowski
Quoted 15 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

When CONFIG_ATH9K is built-in but LED support is in a loadable module, both ath9k drivers fails to link:

x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_deinit_leds':
gpio.c:(.text+0x36): undefined reference to `led_classdev_unregister'
x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_init_leds':
gpio.c:(.text+0x179): undefined reference to `led_classdev_register_ext'

The problem is that the 'imply' keyword does not enforce any dependency but is only a weak hint to Kconfig to enable another symbol from a defconfig file.

Quoted 7 lines

Change imply to a 'depends on LEDS_CLASS' that prevents the incorrect configuration but still allows building the driver without LED support.

The 'select MAC80211_LEDS' is now ensures that the LED support is actually used if it is present, and the added Kconfig dependency on MAC80211_LEDS ensures that it cannot be enabled manually when it has no effect.

Quoted 4 lines

But we do not want to have this dependency (selecting MAC80211_LEDS). I fixed this problem here:
https://lore.kernel.org/lkml/[email protected]/
Maybe let's take this approach?

Generally speaking, I don't like to have a device driver specific Kconfig setting 'select' a subsystem', for two reasons:

- you suddenly get asked for tons of new LED specific options when enabling seemingly benign options

- Mixing 'depends on' and 'select' leads to bugs with circular dependencies that usually require turning some other 'select' into 'depends on'.

The problem with LEDS_CLASS in particular is that there is a mix of drivers using one vs the other roughly 50:50.

Arnd

#3 lore raw
Quoted 37 lines · Arnd Bergmann
Quoted 24 lines · Krzysztof Kozlowski
Quoted 15 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

When CONFIG_ATH9K is built-in but LED support is in a loadable module, both ath9k drivers fails to link:

x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_deinit_leds':
gpio.c:(.text+0x36): undefined reference to `led_classdev_unregister'
x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_init_leds':
gpio.c:(.text+0x179): undefined reference to `led_classdev_register_ext'

The problem is that the 'imply' keyword does not enforce any dependency but is only a weak hint to Kconfig to enable another symbol from a defconfig file.

Quoted 7 lines

Change imply to a 'depends on LEDS_CLASS' that prevents the incorrect configuration but still allows building the driver without LED support.

The 'select MAC80211_LEDS' is now ensures that the LED support is actually used if it is present, and the added Kconfig dependency on MAC80211_LEDS ensures that it cannot be enabled manually when it has no effect.

Quoted 4 lines

But we do not want to have this dependency (selecting MAC80211_LEDS). I fixed this problem here:
https://lore.kernel.org/lkml/[email protected]/
Maybe let's take this approach?

Generally speaking, I don't like to have a device driver specific Kconfig setting 'select' a subsystem', for two reasons:

- you suddenly get asked for tons of new LED specific options when enabling seemingly benign options

Quoted 6 lines

- Mixing 'depends on' and 'select' leads to bugs with circular dependencies that usually require turning some other 'select' into 'depends on'.

The problem with LEDS_CLASS in particular is that there is a mix of drivers using one vs the other roughly 50:50.

Yes, you are right, I also don't like it. However it was like this before my commit so I am not introducing a new issue. The point is that in your choice the MAC80211_LEDS will be selected if LEDS_CLASS is present, which is exactly what I was trying to fix/remove. My WiFi dongle does not have a LED and it causes a periodic (every second) event. However I still have LEDS_CLASS for other LEDS in the system.

Best regards,
Krzysztof

Quoted 45 lines · Arnd Bergmann
Quoted 24 lines · Krzysztof Kozlowski
Quoted 15 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

When CONFIG_ATH9K is built-in but LED support is in a loadable module, both ath9k drivers fails to link:

x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_deinit_leds':
gpio.c:(.text+0x36): undefined reference to `led_classdev_unregister'
x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_init_leds':
gpio.c:(.text+0x179): undefined reference to `led_classdev_register_ext'

The problem is that the 'imply' keyword does not enforce any dependency but is only a weak hint to Kconfig to enable another symbol from a defconfig file.

Quoted 7 lines

Change imply to a 'depends on LEDS_CLASS' that prevents the incorrect configuration but still allows building the driver without LED support.

The 'select MAC80211_LEDS' is now ensures that the LED support is actually used if it is present, and the added Kconfig dependency on MAC80211_LEDS ensures that it cannot be enabled manually when it has no effect.

Quoted 4 lines

But we do not want to have this dependency (selecting MAC80211_LEDS). I fixed this problem here:
https://lore.kernel.org/lkml/[email protected]/
Maybe let's take this approach?

Generally speaking, I don't like to have a device driver specific Kconfig setting 'select' a subsystem', for two reasons:

- you suddenly get asked for tons of new LED specific options when enabling seemingly benign options

- Mixing 'depends on' and 'select' leads to bugs with circular dependencies that usually require turning some other 'select' into 'depends on'.

The problem with LEDS_CLASS in particular is that there is a mix of drivers using one vs the other roughly 50:50.

Arnd

#4 lore raw
Quoted 21 lines · Krzysztof Kozlowski
Quoted 13 lines · Arnd Bergmann
Quoted 5 lines · Krzysztof Kozlowski

On Mon, 25 Jan 2021 at 12:36, Arnd Bergmann <[email protected]> wrote:
But we do not want to have this dependency (selecting MAC80211_LEDS). I fixed this problem here:
https://lore.kernel.org/lkml/[email protected]/
Maybe let's take this approach?

Generally speaking, I don't like to have a device driver specific Kconfig setting 'select' a subsystem', for two reasons:

- you suddenly get asked for tons of new LED specific options when enabling seemingly benign options

Quoted 6 lines

- Mixing 'depends on' and 'select' leads to bugs with circular dependencies that usually require turning some other 'select' into 'depends on'.

The problem with LEDS_CLASS in particular is that there is a mix of drivers using one vs the other roughly 50:50.

Quoted 6 lines

Yes, you are right, I also don't like it. However it was like this before my commit so I am not introducing a new issue. The point is that in your choice the MAC80211_LEDS will be selected if LEDS_CLASS is present, which is exactly what I was trying to fix/remove. My WiFi dongle does not have a LED and it causes a periodic (every second) event. However I still have LEDS_CLASS for other LEDS in the system.

What is the effect of this lost event every second? If it causes some runtime warning or other problem, then neither of our fixes would solve it completely, because someone with a distro kernel would see the same issue when they have the symbol enabled but no physical LED in the device.

Arnd

#5 lore raw
Quoted 30 lines · Arnd Bergmann
Quoted 21 lines · Krzysztof Kozlowski
Quoted 13 lines · Arnd Bergmann
Quoted 5 lines · Krzysztof Kozlowski

On Mon, 25 Jan 2021 at 12:36, Arnd Bergmann <[email protected]> wrote:
But we do not want to have this dependency (selecting MAC80211_LEDS). I fixed this problem here:
https://lore.kernel.org/lkml/[email protected]/
Maybe let's take this approach?

Generally speaking, I don't like to have a device driver specific Kconfig setting 'select' a subsystem', for two reasons:

- you suddenly get asked for tons of new LED specific options when enabling seemingly benign options

Quoted 6 lines

- Mixing 'depends on' and 'select' leads to bugs with circular dependencies that usually require turning some other 'select' into 'depends on'.

The problem with LEDS_CLASS in particular is that there is a mix of drivers using one vs the other roughly 50:50.

Quoted 6 lines

Yes, you are right, I also don't like it. However it was like this before my commit so I am not introducing a new issue. The point is that in your choice the MAC80211_LEDS will be selected if LEDS_CLASS is present, which is exactly what I was trying to fix/remove. My WiFi dongle does not have a LED and it causes a periodic (every second) event. However I still have LEDS_CLASS for other LEDS in the system.

Quoted 5 lines

What is the effect of this lost event every second? If it causes some runtime warning or other problem, then neither of our fixes would solve it completely, because someone with a distro kernel would see the same issue when they have the symbol enabled but no physical LED in the device.

I meant that having MAC80211_LEDS selected causes the ath9k driver to toggle on/off the WiFi LED. Every second, regardless whether it's doing something or not. In my setup, I have problems with a WiFi dongle somehow crashing (WiFi disappears, nothing comes from the dongle... maybe it's Atheros FW, maybe some HW problem) and I found this LED on/off slightly increases the chances of this dongle-crash. That was the actual reason behind my commits.

Second reason is that I don't want to send USB commands every second when the device is idle. It unnecessarily consumes power on my low-power device.

Of course another solution is to just disable the trigger via sysfs LED API. It would also work but my patch allows entire code to be compiled-out (which was conditional in ath9k already).

Therefore the patch I sent allows the ath9k LED option to be fully choosable. Someone wants every-second-LED-blink, sure, enable ATH9K_LEDS and you have it. Someone wants to reduce the kernel size, don't enable ATH9K_LEDS.

Best regards,
Krzysztof

#6 lore raw
Quoted 11 lines · Krzysztof Kozlowski
Quoted 1 line · Arnd Bergmann

On Mon, Jan 25, 2021 at 2:27 PM Krzysztof Kozlowski <[email protected]> wrote:

I meant that having MAC80211_LEDS selected causes the ath9k driver to toggle on/off the WiFi LED. Every second, regardless whether it's doing something or not. In my setup, I have problems with a WiFi dongle somehow crashing (WiFi disappears, nothing comes from the dongle... maybe it's Atheros FW, maybe some HW problem) and I found this LED on/off slightly increases the chances of this dongle-crash. That was the actual reason behind my commits.

Quoted 3 lines

Second reason is that I don't want to send USB commands every second when the device is idle. It unnecessarily consumes power on my low-power device.

Ok, I see.

Quoted 8 lines

Of course another solution is to just disable the trigger via sysfs LED API. It would also work but my patch allows entire code to be compiled-out (which was conditional in ath9k already).

Therefore the patch I sent allows the ath9k LED option to be fully choosable. Someone wants every-second-LED-blink, sure, enable ATH9K_LEDS and you have it. Someone wants to reduce the kernel size, don't enable ATH9K_LEDS.

Originally, I think this is what CONFIG_MAC80211_LEDS was meant for, but it seems that this is not actually practical, since this also gets selected by half of the drivers using it, while the other half have a dependency on it. Out of the ones that select it, some in turn select LEDS_CLASS, while some depend on it.

I think this needs a larger-scale cleanup for consistency between (at least) all the wireless drivers using LEDs. Either your patch or mine should get applied in the meantime, and I don't care much which one in this case, as we still have the remaining inconsistency.

Arnd

#7 lore raw
Quoted 27 lines · Arnd Bergmann
Quoted 11 lines · Krzysztof Kozlowski
Quoted 1 line · Arnd Bergmann

On Mon, Jan 25, 2021 at 2:27 PM Krzysztof Kozlowski <[email protected]> wrote:

I meant that having MAC80211_LEDS selected causes the ath9k driver to toggle on/off the WiFi LED. Every second, regardless whether it's doing something or not. In my setup, I have problems with a WiFi dongle somehow crashing (WiFi disappears, nothing comes from the dongle... maybe it's Atheros FW, maybe some HW problem) and I found this LED on/off slightly increases the chances of this dongle-crash. That was the actual reason behind my commits.

Quoted 3 lines

Second reason is that I don't want to send USB commands every second when the device is idle. It unnecessarily consumes power on my low-power device.

Ok, I see.

Quoted 8 lines

Of course another solution is to just disable the trigger via sysfs LED API. It would also work but my patch allows entire code to be compiled-out (which was conditional in ath9k already).

Therefore the patch I sent allows the ath9k LED option to be fully choosable. Someone wants every-second-LED-blink, sure, enable ATH9K_LEDS and you have it. Someone wants to reduce the kernel size, don't enable ATH9K_LEDS.

Quoted 8 lines

Originally, I think this is what CONFIG_MAC80211_LEDS was meant for, but it seems that this is not actually practical, since this also gets selected by half of the drivers using it, while the other half have a dependency on it. Out of the ones that select it, some in turn select LEDS_CLASS, while some depend on it.

I think this needs a larger-scale cleanup for consistency between (at least) all the wireless drivers using LEDs.

I agree, this needs cleanup.

Quoted 3 lines

Either your patch or mine should get applied in the meantime, and I don't care much which one in this case, as we still have the remaining inconsistency.

My problem with Krzysztof's patch[1] is that it adds a new Kconfig option for ath9k, is that really necessary? Like Arnd said, we should fix drivers to use CONFIG_MAC80211_LEDS instead of having driver specific options.

So I would prefer take this Arnd's patch instead and queue it for v5.11. But as it modifies mac80211 I'll need an ack from Johannes, what do you think?

[1] https://patchwork.kernel.org/project/linux-wireless/patch/[email protected]/

Signature · 3 lines
https://patchwork.kernel.org/project/linux-wireless/list/

https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches
maintainer Acked-by
#8 lore raw
Quoted 44 lines · Kalle Valo
Quoted 27 lines · Arnd Bergmann
Quoted 11 lines · Krzysztof Kozlowski
Quoted 1 line · Arnd Bergmann

On Mon, Jan 25, 2021 at 2:27 PM Krzysztof Kozlowski <[email protected]> wrote:

I meant that having MAC80211_LEDS selected causes the ath9k driver to toggle on/off the WiFi LED. Every second, regardless whether it's doing something or not. In my setup, I have problems with a WiFi dongle somehow crashing (WiFi disappears, nothing comes from the dongle... maybe it's Atheros FW, maybe some HW problem) and I found this LED on/off slightly increases the chances of this dongle-crash. That was the actual reason behind my commits.

Quoted 3 lines

Second reason is that I don't want to send USB commands every second when the device is idle. It unnecessarily consumes power on my low-power device.

Ok, I see.

Quoted 8 lines

Of course another solution is to just disable the trigger via sysfs LED API. It would also work but my patch allows entire code to be compiled-out (which was conditional in ath9k already).

Therefore the patch I sent allows the ath9k LED option to be fully choosable. Someone wants every-second-LED-blink, sure, enable ATH9K_LEDS and you have it. Someone wants to reduce the kernel size, don't enable ATH9K_LEDS.

Quoted 8 lines

Originally, I think this is what CONFIG_MAC80211_LEDS was meant for, but it seems that this is not actually practical, since this also gets selected by half of the drivers using it, while the other half have a dependency on it. Out of the ones that select it, some in turn select LEDS_CLASS, while some depend on it.

I think this needs a larger-scale cleanup for consistency between (at least) all the wireless drivers using LEDs.

I agree, this needs cleanup.

Quoted 3 lines

Either your patch or mine should get applied in the meantime, and I don't care much which one in this case, as we still have the remaining inconsistency.

Quoted 8 lines

My problem with Krzysztof's patch[1] is that it adds a new Kconfig option for ath9k, is that really necessary? Like Arnd said, we should fix drivers to use CONFIG_MAC80211_LEDS instead of having driver specific options.

So I would prefer take this Arnd's patch instead and queue it for v5.11. But as it modifies mac80211 I'll need an ack from Johannes, what do you think?

Sure, that seems fine.

johannes

#9 lore raw
In reply to Arnd Bergmann
Quoted 17 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

When CONFIG_ATH9K is built-in but LED support is in a loadable module, both ath9k drivers fails to link:

x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_deinit_leds':
gpio.c:(.text+0x36): undefined reference to `led_classdev_unregister'
x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_init_leds':
gpio.c:(.text+0x179): undefined reference to `led_classdev_register_ext'

The problem is that the 'imply' keyword does not enforce any dependency but is only a weak hint to Kconfig to enable another symbol from a defconfig file.

Change imply to a 'depends on LEDS_CLASS' that prevents the incorrect configuration but still allows building the driver without LED support.

Quoted 8 lines

The 'select MAC80211_LEDS' is now ensures that the LED support is actually used if it is present, and the added Kconfig dependency on MAC80211_LEDS ensures that it cannot be enabled manually when it has no effect.

Fixes 197f466e93f5 ("ath9k_htc: Do not select MAC80211_LEDS by default")

Patch applied to wireless-drivers.git, thanks.

b64acb28da83 ath9k: fix build error with LEDS_CLASS=m

Signature · 3 lines
https://patchwork.kernel.org/project/linux-wireless/patch/[email protected]/

https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches