[lm-sensors] fscpos and fscher drivers

[lm-sensors] fscpos and fscher drivers

[Jean Delvare]

Hi Hans,

Do you have a plan to phase out the fscpos and fscher drivers? Now that
your fschmd driver supports all known FSC hardware monitoring chips, it
would be great if we could get rid of the legacy fscpos and fscher
drivers to make the configuration, maintenance and support easier.

A first step could be to mark both drivers DEPRECATED in Kconfig, and
add an entry for their removal in
Documentation/feature-removal-schedule.txt. You could also add a
warning message when these drivers are loaded, pointing the user to the
replacement fschmd driver. I don't think these chips are very popular
so it will take some time before your new driver gets enough testing
and we can safely remove the legacy drivers. But the earlier we ask
the users to migrate, the earlier we will be able to remove the legacy
drivers.

Thanks,
-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] fscpos and fscher drivers

[Hans de Goede]

Jean Delvare wrote:
> Hi Hans,
> 
> Do you have a plan to phase out the fscpos and fscher drivers? Now that
> your fschmd driver supports all known FSC hardware monitoring chips, it
> would be great if we could get rid of the legacy fscpos and fscher
> drivers to make the configuration, maintenance and support easier.
> 

Agreed, the reason this hasn't been done until now is because the fscpos and 
fscher drivers support the watchdog function of the fscXXX family whereas my 
fschmd driver doesn't (note the fscpos and fscher drivers do not support the 
watchdog driver through the standard watchdog interface, instead they use 
custom sysfs attr under the hwmon class).

I've adding support for the watchdog part on my todo list for some time now. 
But I have too many OSS projects, so I have to prioritize and sofar haven't 
given adding watchdog support a priority.

> A first step could be to mark both drivers DEPRECATED in Kconfig, and
> add an entry for their removal in
> Documentation/feature-removal-schedule.txt. You could also add a
> warning message when these drivers are loaded, pointing the user to the
> replacement fschmd driver. I don't think these chips are very popular
> so it will take some time before your new driver gets enough testing
> and we can safely remove the legacy drivers. But the earlier we ask
> the users to migrate, the earlier we will be able to remove the legacy
> drivers.
> 

I don't think many people if any at all use the watchdog feature, esp. as its 
exported using a non standard interface, so if this is not a blocker, we could 
still schedule fscher and fscpos for removal.

Regards,

Hans


_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] fscpos and fscher drivers

[Jean Delvare]

On Thu, 26 Jun 2008 09:09:34 +0200, Hans de Goede wrote:
> Jean Delvare wrote:
> > Hi Hans,
> > 
> > Do you have a plan to phase out the fscpos and fscher drivers? Now that
> > your fschmd driver supports all known FSC hardware monitoring chips, it
> > would be great if we could get rid of the legacy fscpos and fscher
> > drivers to make the configuration, maintenance and support easier.
> > 
> 
> Agreed, the reason this hasn't been done until now is because the fscpos and 
> fscher drivers support the watchdog function of the fscXXX family whereas my 
> fschmd driver doesn't (note the fscpos and fscher drivers do not support the 
> watchdog driver through the standard watchdog interface, instead they use 
> custom sysfs attr under the hwmon class).

Ah, I had forgotten about this watchdog issue. This makes the situation
a bit more complex of course.

> I've adding support for the watchdog part on my todo list for some time now. 
> But I have too many OSS projects, so I have to prioritize and sofar haven't 
> given adding watchdog support a priority.
> 
> > A first step could be to mark both drivers DEPRECATED in Kconfig, and
> > add an entry for their removal in
> > Documentation/feature-removal-schedule.txt. You could also add a
> > warning message when these drivers are loaded, pointing the user to the
> > replacement fschmd driver. I don't think these chips are very popular
> > so it will take some time before your new driver gets enough testing
> > and we can safely remove the legacy drivers. But the earlier we ask
> > the users to migrate, the earlier we will be able to remove the legacy
> > drivers.
> 
> I don't think many people if any at all use the watchdog feature, esp. as its 
> exported using a non standard interface, so if this is not a blocker, we could 
> still schedule fscher and fscpos for removal.

I guess you could add the DEPRECATED marks in Kconfig already, and
maybe the warnings when loading the drivers as well, but without
scheduling the drivers for removal yet. If users need the watchdog
function, they'll see it is missing from the new driver and will
hopefully report to you. This will give you an idea if implementing the
watchdog feature in the new driver is worth your time at all, and if it
is, where in your todo list it should go.

-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] fscpos and fscher drivers

[Hans de Goede]

Jean Delvare wrote:
> On Thu, 26 Jun 2008 09:09:34 +0200, Hans de Goede wrote:
>> Jean Delvare wrote:
>>> Hi Hans,
>>>
>>> Do you have a plan to phase out the fscpos and fscher drivers? Now that
>>> your fschmd driver supports all known FSC hardware monitoring chips, it
>>> would be great if we could get rid of the legacy fscpos and fscher
>>> drivers to make the configuration, maintenance and support easier.
>>>
>> Agreed, the reason this hasn't been done until now is because the fscpos and 
>> fscher drivers support the watchdog function of the fscXXX family whereas my 
>> fschmd driver doesn't (note the fscpos and fscher drivers do not support the 
>> watchdog driver through the standard watchdog interface, instead they use 
>> custom sysfs attr under the hwmon class).
> 
> Ah, I had forgotten about this watchdog issue. This makes the situation
> a bit more complex of course.
> 

Actually this has sparked me into looking into the watchdog stuff. So I'm 
working on adding (proper) watchdog support right now.

Regards,

Hans


_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors