Fwd: Using select in boolean dependents of a tristate symbol

Fwd: Using select in boolean dependents of a tristate symbol

[Dmitry Torokhov]

[Argh.. resending to correct LKML address]

Roman,

Question for you as Kconfig maintainer - I have a module (HID) that
has a few sub-options. Now HID can be built as a module but
sub-options are suimple booleans. Some of these sub-options will
depend on a common code which is moved out of HID driver. Now I want
to use "select" like this:

 config THRUSTMASTER_FF
       bool "ThrustMaster FireStorm Dual Power 2 support (EXPERIMENTAL)"
       depends on HID_FF && EXPERIMENTAL
+       select INPUT_FF_MEMLESS
       help
         Say Y here if you have a THRUSTMASTER FireStore Dual Power 2,
         and want to enable force feedback support for it.

Unfortunately this forces INPUT_FF_MEMLESS to always be built-in,
although if HID is a module it could be a module as well. Do you have
any suggestions as to how allow INPUT_FF_MEMLESS to be compiled as a
module?

Thanks!

--
Dmitry


-- 
Dmitry

Re: Fwd: Using select in boolean dependents of a tristate symbol

[Roman Zippel]

Hi,

On Thu, 13 Jul 2006, Dmitry Torokhov wrote:

> config THRUSTMASTER_FF
>       bool "ThrustMaster FireStorm Dual Power 2 support (EXPERIMENTAL)"
>       depends on HID_FF && EXPERIMENTAL
> +       select INPUT_FF_MEMLESS
>       help
>         Say Y here if you have a THRUSTMASTER FireStore Dual Power 2,
>         and want to enable force feedback support for it.
> 
> Unfortunately this forces INPUT_FF_MEMLESS to always be built-in,
> although if HID is a module it could be a module as well. Do you have
> any suggestions as to how allow INPUT_FF_MEMLESS to be compiled as a
> module?

You need to directly include HID into the dependencies, only the direct 
dependencies for config entry are used for the select.

bye, Roman

Re: Fwd: Using select in boolean dependents of a tristate symbol

[Dmitry Torokhov]

On Thursday 13 July 2006 18:58, Roman Zippel wrote:
> Hi,
> 
> On Thu, 13 Jul 2006, Dmitry Torokhov wrote:
> 
> > config THRUSTMASTER_FF
> >       bool "ThrustMaster FireStorm Dual Power 2 support (EXPERIMENTAL)"
> >       depends on HID_FF && EXPERIMENTAL
> > +       select INPUT_FF_MEMLESS
> >       help
> >         Say Y here if you have a THRUSTMASTER FireStore Dual Power 2,
> >         and want to enable force feedback support for it.
> > 
> > Unfortunately this forces INPUT_FF_MEMLESS to always be built-in,
> > although if HID is a module it could be a module as well. Do you have
> > any suggestions as to how allow INPUT_FF_MEMLESS to be compiled as a
> > module?
> 
> You need to directly include HID into the dependencies, only the direct 
> dependencies for config entry are used for the select.
>

Oh, this indeed works, thanks a lot! And I was thinking I would need to
implement something like "select <expr> as <expr>" in kconfig ;)  

-- 
Dmitry

Re: Fwd: Using select in boolean dependents of a tristate symbol

[Roman Zippel]

Hi,

On Thu, 13 Jul 2006, Dmitry Torokhov wrote:

> On Thursday 13 July 2006 18:58, Roman Zippel wrote:
> > Hi,
> > 
> > On Thu, 13 Jul 2006, Dmitry Torokhov wrote:
> > 
> > > config THRUSTMASTER_FF
> > >       bool "ThrustMaster FireStorm Dual Power 2 support (EXPERIMENTAL)"
> > >       depends on HID_FF && EXPERIMENTAL
> > > +       select INPUT_FF_MEMLESS
> > >       help
> > >         Say Y here if you have a THRUSTMASTER FireStore Dual Power 2,
> > >         and want to enable force feedback support for it.
> > > 
> > > Unfortunately this forces INPUT_FF_MEMLESS to always be built-in,
> > > although if HID is a module it could be a module as well. Do you have
> > > any suggestions as to how allow INPUT_FF_MEMLESS to be compiled as a
> > > module?
> > 
> > You need to directly include HID into the dependencies, only the direct 
> > dependencies for config entry are used for the select.
> >
> 
> Oh, this indeed works, thanks a lot! And I was thinking I would need to
> implement something like "select <expr> as <expr>" in kconfig ;)  

What you could do is to use "select INPUT_FF_MEMLESS if HID" to make it 
visible that this dependency is actually for select.
This point is a little subtle and I'm not completely happy with it, maybe 
I'm going to split this into two select variations - one which includes 
all the dependencies and one which only uses the config symbol to select.

bye, Roman

Re: Fwd: Using select in boolean dependents of a tristate symbol

[Dmitry Torokhov]

On 7/14/06, Roman Zippel <zippel@linux-m68k.org> wrote:
>
> What you could do is to use "select INPUT_FF_MEMLESS if HID" to make it
> visible that this dependency is actually for select.
> This point is a little subtle and I'm not completely happy with it, maybe
> I'm going to split this into two select variations - one which includes
> all the dependencies and one which only uses the config symbol to select.
>

Roman,

Another question for you  - what is the best way to describe
dependancy of a sub-option on a subsystem so you won't end up with the
subsystem as a module and user built in. Something like

config IBM_ASM
        tristate "Device driver for IBM RSA service processor"
        depends on X86 && PCI && EXPERIMENTAL
...
config IBM_ASM_INPUT
        bool "Support for remote keyboard/mouse"
        depends on IBM_ASM && (INPUT=y || INPUT=IMB_ASM)

But the above feels yucky. Could we have something like:

         depends on matching(INPUT, IBM_ASM)

Thank you.

-- 
Dmitry

Re: Fwd: Using select in boolean dependents of a tristate symbol

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 796 bytes --]

On Wed, 19 Jul 2006 16:17:38 EDT, Dmitry Torokhov said:

> Another question for you  - what is the best way to describe
> dependancy of a sub-option on a subsystem so you won't end up with the
> subsystem as a module and user built in. Something like
> 
> config IBM_ASM
>         tristate "Device driver for IBM RSA service processor"
>         depends on X86 && PCI && EXPERIMENTAL
> ...
> config IBM_ASM_INPUT
>         bool "Support for remote keyboard/mouse"
>         depends on IBM_ASM && (INPUT=y || INPUT=IMB_ASM)
> 
> But the above feels yucky. Could we have something like:
> 
>          depends on matching(INPUT, IBM_ASM)

What feels yucky is the dependency of a 'bool' on a tristate.  Does the
ASM_INPUT get used in places where the source file can only be a builtin,
not a module?

[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]

Re: Fwd: Using select in boolean dependents of a tristate symbol

[Dmitry Torokhov]

On 7/19/06, Valdis.Kletnieks@vt.edu <Valdis.Kletnieks@vt.edu> wrote:
> On Wed, 19 Jul 2006 16:17:38 EDT, Dmitry Torokhov said:
>
> > Another question for you  - what is the best way to describe
> > dependancy of a sub-option on a subsystem so you won't end up with the
> > subsystem as a module and user built in. Something like
> >
> > config IBM_ASM
> >         tristate "Device driver for IBM RSA service processor"
> >         depends on X86 && PCI && EXPERIMENTAL
> > ...
> > config IBM_ASM_INPUT
> >         bool "Support for remote keyboard/mouse"
> >         depends on IBM_ASM && (INPUT=y || INPUT=IMB_ASM)
> >
> > But the above feels yucky. Could we have something like:
> >
> >          depends on matching(INPUT, IBM_ASM)
>
> What feels yucky is the dependency of a 'bool' on a tristate.  Does the
> ASM_INPUT get used in places where the source file can only be a builtin,
> not a module?
>

In this case ASM_INPUT enables an optional part of a bigger module
(IBM_ASM). Sometimes it is done because optional part is too small to
be split into a separately loadable module or because it is difficult
to implement "attaching" of the optional part at the later time if it
is compiled as a module.

-- 
Dmitry

Re: Fwd: Using select in boolean dependents of a tristate symbol

[Roman Zippel]

Hi,

On Wed, 19 Jul 2006, Dmitry Torokhov wrote:

> Another question for you  - what is the best way to describe
> dependancy of a sub-option on a subsystem so you won't end up with the
> subsystem as a module and user built in. Something like
> 
> config IBM_ASM
>        tristate "Device driver for IBM RSA service processor"
>        depends on X86 && PCI && EXPERIMENTAL
> ...
> config IBM_ASM_INPUT
>        bool "Support for remote keyboard/mouse"
>        depends on IBM_ASM && (INPUT=y || INPUT=IMB_ASM)
> 
> But the above feels yucky. Could we have something like:
> 
>         depends on matching(INPUT, IBM_ASM)

This is not really descriptive of what it does, is it?
Linus suggested a syntax like (IBM_ASM && IMB_ASM<=INPUT)
Another alternative which works now is to just disable the one invalid 
case explicitely:

	depends on IBM_ASM && INPUT
	depends on !(IBM_ASM=y && INPUT=m)

bye, Roman

Re: Fwd: Using select in boolean dependents of a tristate symbol

[Dmitry Torokhov]

On Wednesday 26 July 2006 20:42, Roman Zippel wrote:
> Hi,
> 
> On Wed, 19 Jul 2006, Dmitry Torokhov wrote:
> 
> > Another question for you  - what is the best way to describe
> > dependancy of a sub-option on a subsystem so you won't end up with the
> > subsystem as a module and user built in. Something like
> > 
> > config IBM_ASM
> >        tristate "Device driver for IBM RSA service processor"
> >        depends on X86 && PCI && EXPERIMENTAL
> > ...
> > config IBM_ASM_INPUT
> >        bool "Support for remote keyboard/mouse"
> >        depends on IBM_ASM && (INPUT=y || INPUT=IMB_ASM)
> > 
> > But the above feels yucky. Could we have something like:
> > 
> >         depends on matching(INPUT, IBM_ASM)
> 
> This is not really descriptive of what it does, is it?
> Linus suggested a syntax like (IBM_ASM && IMB_ASM<=INPUT)
> Another alternative which works now is to just disable the one invalid 
> case explicitely:
> 
> 	depends on IBM_ASM && INPUT
> 	depends on !(IBM_ASM=y && INPUT=m)
>

OK, then I'll be disabling invalid cases explicitly for now.

Thank you Roman.

-- 
Dmitry