broken nouveau dependency on power supply

DISCUSSION Benjamin Herrenschmidt 3 participants 5 messages Apr 2, 2012 – May 5, 2012 Power supply

lore

Share & tools

lore

Benjamin Herrenschmidt

author

2012-04-02 01:06 UTC #0 lore raw

Hi folks !

With CONFIG_POWER_SUPPLY=m & nouveau built-in we get a build failure:

drivers/built-in.o: In function `.nouveau_pm_trigger':
(.text+0xa56e8): undefined reference to `.power_supply_is_system_supplied'

nouveau probably needs to depends on CONFIG_POWER_SUPPLY to force a module build with the latter is =m

Cheers,
Ben.

Benjamin Herrenschmidt

author

2012-04-02 03:53 UTC #1 lore raw

Quoted 9 lines · Benjamin Herrenschmidt

Hi folks !
With CONFIG_POWER_SUPPLY=m & nouveau built-in we get a build failure:
drivers/built-in.o: In function `.nouveau_pm_trigger':
(.text+0xa56e8): undefined reference to `.power_supply_is_system_supplied'
nouveau probably needs to depends on CONFIG_POWER_SUPPLY to force a module build with the latter is =m

Ok, not that trivial...

The problem is more like POWER_SUPPLY should be a bool, not a tristate.

If you think about it: you don't want things like nouveau to depend on a random subsystem like that, people will never get it. In fact, POWER_SUPPLY provides empty inline stubs when not enabled, so that's really designed to not have depends...

However that -cannot- work if POWER_SUPPLY is modular and the drivers who use it are not. The only fixes here that make sense I can think of that don't also involve Kconfig horrors are:

- Ugly: in power_supply.h, use the extern variant if

	defined(CONFIG_POWER_SUPPLY) ||
	 (defined(CONFIG_POWER_SUPPLY_MODULE) && defined(MODULE))

IE. use the stub if power supply is a module and what is being built is built-in. Of course that's not only ugly, it somewhat sucks from a user perspective as the subsystem now exists but can't be used by some drivers...

- Better: Just make the bloody thing a bool :-) The power supply
framework itself is small enough, just make it a boolean option and avoid the problem entirely. The actual power supply sub drivers can remain modular of course.

Cheers,
Ben.

David Airlie

2012-04-02 09:00 UTC #2 lore raw

Quoted 30 lines

Ok, not that trivial...
The problem is more like POWER_SUPPLY should be a bool, not a tristate.
If you think about it: you don't want things like nouveau to depend on a
random subsystem like that, people will never get it. In fact, POWER_SUPPLY provides empty inline stubs when not enabled, so that's really designed to not have depends...
However that -cannot- work if POWER_SUPPLY is modular and the drivers who use it are not. The only fixes here that make sense I can think of
that don't also involve Kconfig horrors are:
- Ugly: in power_supply.h, use the extern variant if
	defined(CONFIG_POWER_SUPPLY) ||
	 (defined(CONFIG_POWER_SUPPLY_MODULE) && defined(MODULE))
IE. use the stub if power supply is a module and what is being built is
built-in. Of course that's not only ugly, it somewhat sucks from a user
perspective as the subsystem now exists but can't be used by some drivers...

Quoted 4 lines

- Better: Just make the bloody thing a bool :-) The power supply
framework itself is small enough, just make it a boolean option and avoid the problem entirely. The actual power supply sub drivers can remain modular of course.

We can just do select POWER_SUPPLY.

Yes it reduces the option range for some stupid corner case but really I don't care, removing features from the kernel that a driver depends on is just leading to insane state combination and QA problems.

Dave.

Benjamin Herrenschmidt

author

2012-04-02 09:48 UTC #3 lore raw

Quoted 6 lines · David Airlie

Quoted 4 lines
- Better: Just make the bloody thing a bool :-) The power supply
framework itself is small enough, just make it a boolean option and avoid the problem entirely. The actual power supply sub drivers can remain modular of course.
We can just do select POWER_SUPPLY.

Well, select'ing otherwise user configurable options was still frowned on last we discussed that... and it makes the whole inline stubs in power_supply.h totally pointless :-)

Quoted 3 lines

Yes it reduces the option range for some stupid corner case but really I don't care, removing features from the kernel that a driver depends on is just leading to insane state combination and QA problems.

Well, the power supply stuff only works if you have a backend for it, which not all platforms are, and it's fairly safe to assume AC whenever it's not actually supported.

(Talking of which, we should be able to do a PowerBook backend reasonably easily).

I don't care which solution you guys end up choosing though, just fix it :-)

Cheers,
Ben.

Anton Vorontsov

2012-05-05 04:29 UTC #4 lore raw

In reply to Benjamin Herrenschmidt

Hello Benjamin,

Sorry, it took me some time to get to it.

On Mon, Apr 02, 2012 at 01:53:23PM +1000, Benjamin Herrenschmidt wrote:
[...]

Quoted 11 lines

Quoted 7 lines
With CONFIG_POWER_SUPPLY=m & nouveau built-in we get a build failure:
drivers/built-in.o: In function `.nouveau_pm_trigger':
(.text+0xa56e8): undefined reference to `.power_supply_is_system_supplied'
nouveau probably needs to depends on CONFIG_POWER_SUPPLY to force a module build with the latter is =m
Ok, not that trivial...
The problem is more like POWER_SUPPLY should be a bool, not a tristate.

I see that nouveau already selects POWER_SUPPLY, but your points are still valid. Let's make the power supply thing simple. I've applied the following patch:

- - - -
[PATCH] power_supply: Make the core a boolean instead of a tristate

Quoted 31 lines · Benjamin Herrenschmidt

Quoted 5 lines
drivers/built-in.o: In function `.nouveau_pm_trigger':
(.text+0xa56e8): undefined reference to `.power_supply_is_system_supplied'
nouveau probably needs to depends on CONFIG_POWER_SUPPLY to force a module build with the latter is =m
Ok, not that trivial...
The problem is more like POWER_SUPPLY should be a bool, not a tristate.
If you think about it: you don't want things like nouveau to depend on a random subsystem like that, people will never get it. In fact, POWER_SUPPLY provides empty inline stubs when not enabled, so that's really designed to not have depends...
However that -cannot- work if POWER_SUPPLY is modular and the drivers who use it are not.
The only fixes here that make sense I can think of that don't also involve Kconfig horrors are:
- Ugly: in power_supply.h, use the extern variant if
      defined(CONFIG_POWER_SUPPLY) ||
       (defined(CONFIG_POWER_SUPPLY_MODULE) && defined(MODULE))
IE. use the stub if power supply is a module and what is being built is built-in. Of course that's not only ugly, it somewhat sucks from a user perspective as the subsystem now exists but can't be used by some drivers...

Quoted 4 lines

- Better: Just make the bloody thing a bool :-) The power supply
framework itself is small enough, just make it a boolean option and avoid the problem entirely. The actual power supply sub drivers can remain modular of course.

Signed-off-by Anton Vorontsov <[email protected]>

---
 drivers/power/Kconfig        |    2 +-
 include/linux/power_supply.h |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

drivers/power/Kconfig +1 -1

index 99dc29f..0c52a40 100644
@@ -1,5 +1,5 @@
 menuconfig POWER_SUPPLY
-	tristate "Power supply class support"
+	bool "Power supply class support"
 	help
 	  Say Y here to enable power supply class support. This allows
 	  power supply (batteries, AC, USB) monitoring by userspace

include/linux/power_supply.h +1 -1

index fd17ae0..3b912be 100644
@@ -212,7 +212,7 @@ extern void power_supply_changed(struct power_supply *psy);
 extern int power_supply_am_i_supplied(struct power_supply *psy);
 extern int power_supply_set_battery_charged(struct power_supply *psy);
 
-#if defined(CONFIG_POWER_SUPPLY) || defined(CONFIG_POWER_SUPPLY_MODULE)
+#ifdef CONFIG_POWER_SUPPLY
 extern int power_supply_is_system_supplied(void);
 #else
 static inline int power_supply_is_system_supplied(void) { return -ENOSYS; }

Signature · 1 line

1.7.9.2