Skip to contentSkip to conversation

Kconfig Q: expressing a modularity dependency

DISCUSSION Tilman Schmidt 2 participants 4 messages Apr 15, 2009 – Aug 23, 2009

lore
Share & tools
lore
#0 lore raw

My current attempts to port the Gigaset driver from isdn4linux to CAPI require the following Kconfig dependency:

- If isdn4linux or kernelcapi or both are built as a module (CONFIG_ISDN_I4L=m || CONFIG_ISDN_CAPI=m)
then gigaset can only be built as a module (CONFIG_ISDN_DRV_GIGASET depends on m).

- If neither isdn4linux nor kernelcapi ar built as a module (CONFIG_ISDN_I4L!=m && CONFIG_ISDN_CAPI!=m), including the case that neither is being built at all (CONFIG_ISDN_I4L=n && CONFIG_ISDN_CAPI=n),
then gigaset may be built statically or as a module (all three tristate settings for CONFIG_ISDN_DRV_GIGASET are legal).

How is that best expressed in Kconfig?

Thanks,
Tilman

Signature · 4 lines
Tilman Schmidt                    E-Mail: [email protected]
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Ungeöffnet mindestens haltbar bis: (siehe Rückseite)
#1 lore raw

/me wrote, four months ago:

Quoted 16 lines

My current attempts to port the Gigaset driver from isdn4linux to CAPI require the following Kconfig dependency:

- If isdn4linux or kernelcapi or both are built as a module (CONFIG_ISDN_I4L=m || CONFIG_ISDN_CAPI=m)
then gigaset can only be built as a module (CONFIG_ISDN_DRV_GIGASET depends on m).

- If neither isdn4linux nor kernelcapi ar built as a module (CONFIG_ISDN_I4L!=m && CONFIG_ISDN_CAPI!=m), including the case that neither is being built at all (CONFIG_ISDN_I4L=n && CONFIG_ISDN_CAPI=n),
then gigaset may be built statically or as a module (all three tristate settings for CONFIG_ISDN_DRV_GIGASET are legal).

Quoted 1 line

How is that best expressed in Kconfig?

Judging from the complete silence following that question, the answer is, apparently: not at all.

So I'll just put a notice in the help text then, warning people not to set CONFIG_ISDN_DRV_GIGASET=y if their ISDN subsystem is built as a module. If anybody has a better idea I'll be glad to hear it.

Thx
T.

Signature · 4 lines
Tilman Schmidt                    E-Mail: [email protected]
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Ungeöffnet mindestens haltbar bis: (siehe Rückseite)
#2 lore raw

Hi!

Quoted 19 lines

/me wrote, four months ago:

Quoted 16 lines

My current attempts to port the Gigaset driver from isdn4linux to CAPI require the following Kconfig dependency:

- If isdn4linux or kernelcapi or both are built as a module (CONFIG_ISDN_I4L=m || CONFIG_ISDN_CAPI=m)
then gigaset can only be built as a module (CONFIG_ISDN_DRV_GIGASET depends on m).

- If neither isdn4linux nor kernelcapi ar built as a module (CONFIG_ISDN_I4L!=m && CONFIG_ISDN_CAPI!=m), including the case that neither is being built at all (CONFIG_ISDN_I4L=n && CONFIG_ISDN_CAPI=n),
then gigaset may be built statically or as a module (all three tristate settings for CONFIG_ISDN_DRV_GIGASET are legal).

Quoted 1 line

How is that best expressed in Kconfig?

Quoted 7 lines

Judging from the complete silence following that question, the answer is, apparently: not at all.

So I'll just put a notice in the help text then, warning people not to set CONFIG_ISDN_DRV_GIGASET=y if their ISDN subsystem is built as a module. If anybody has a better idea I'll be glad to hear it.

That will break randconfig testing, AFAICT.

AFAIK Kconfig supports if statement just fine, so can you just

if something
bool GIGASET
else
tristate GIGASET

?

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#3 lore raw
Quoted 6 lines · Pavel Machek
Quoted 4 lines

So I'll just put a notice in the help text then, warning people not to set CONFIG_ISDN_DRV_GIGASET=y if their ISDN subsystem is built as a module. If anybody has a better idea I'll be glad to hear it.

That will break randconfig testing, AFAICT.

I don't think so. Ignoring the warning will not result in a build failure, just in reduced functionality, and AFAIK that isn't an issue in randconfig testing.

Quoted 6 lines

AFAIK Kconfig supports if statement just fine, so can you just

if something
bool GIGASET
else
tristate GIGASET

Interesting idea. I'll try that. It'll probably not make it into my submission for 2.6.32 anymore, but perhaps it can be retrofitted later.

Thanks,
Tilman

Signature · 4 lines
Tilman Schmidt                    E-Mail: [email protected]
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Ungeöffnet mindestens haltbar bis: (siehe Rückseite)