BLK_DEV_MD with CONFIG_NET

BLK_DEV_MD with CONFIG_NET

[Randy Dunlap]

drivers/md/md.c calls csum_partial().

IF CONFIG_NET=n and BLK_DEV_MD=y, if arch/*/lib/Makefile
puts csum-partial.o or checksum.o into lib-y, the function
is present.  (Of course, if the function is placed in
obj-y, there is no problem.)

If CONFIG_NET=n and BLK_DEV_MD=n, if arch/*/lib/Makefile
puts csum-partial.o or checksum.o into lib-y, the function
is removed from the kernel image due to having no built-in
callers.

Build a kernel with CONFIG_NET-n and CONFIG_BLK_DEV_MD=m.
Unless csum_partial() is built and kept by some arch Makefile,
the result is:
ERROR: "csum_partial" [drivers/md/md-mod.ko] undefined!
make[1]: *** [__modpost] Error 1
make: *** [modules] Error 2


Any suggested solutions?


Recorded for posterity here:
  http://bugzilla.kernel.org/show_bug.cgi?id=8242

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

Re: BLK_DEV_MD with CONFIG_NET

[David Miller]

From: Randy Dunlap <randy.dunlap@oracle.com>
Date: Tue, 20 Mar 2007 20:05:38 -0700

> Build a kernel with CONFIG_NET-n and CONFIG_BLK_DEV_MD=m.
> Unless csum_partial() is built and kept by some arch Makefile,
> the result is:
> ERROR: "csum_partial" [drivers/md/md-mod.ko] undefined!
> make[1]: *** [__modpost] Error 1
> make: *** [modules] Error 2
> 
> 
> Any suggested solutions?

Anything which is every exported to modules, which ought to
be the situation in this case, should be obj-y not lib-y
right?

Re: BLK_DEV_MD with CONFIG_NET

[Sam Ravnborg]

On Tue, Mar 20, 2007 at 08:26:21PM -0700, David Miller wrote:
> From: Randy Dunlap <randy.dunlap@oracle.com>
> Date: Tue, 20 Mar 2007 20:05:38 -0700
> 
> > Build a kernel with CONFIG_NET-n and CONFIG_BLK_DEV_MD=m.
> > Unless csum_partial() is built and kept by some arch Makefile,
> > the result is:
> > ERROR: "csum_partial" [drivers/md/md-mod.ko] undefined!
> > make[1]: *** [__modpost] Error 1
> > make: *** [modules] Error 2
> > 
> > 
> > Any suggested solutions?
> 
> Anything which is every exported to modules, which ought to
> be the situation in this case, should be obj-y not lib-y
> right?
That is also my understanding of lib-y - I should update makefiles.txt
to reflect this..

	Sam

Re: BLK_DEV_MD with CONFIG_NET

[Arnd Bergmann]

On Wednesday 21 March 2007 13:02:46 Sam Ravnborg wrote:
> > Anything which is every exported to modules, which ought to
> > be the situation in this case, should be obj-y not lib-y
> > right?
>
> That is also my understanding of lib-y - I should update makefiles.txt
> to reflect this..

Strictly speaking, it could well be obj-m instead of obj-y if it
is _only_ used by modules. OTOH, it makes the Makefile a lot simpler
to not optimize for this case.

	Arnd <><

Re: BLK_DEV_MD with CONFIG_NET

[Randy Dunlap]

On Wed, 21 Mar 2007 13:02:46 +0100 Sam Ravnborg wrote:

> On Tue, Mar 20, 2007 at 08:26:21PM -0700, David Miller wrote:
> > From: Randy Dunlap <randy.dunlap@oracle.com>
> > Date: Tue, 20 Mar 2007 20:05:38 -0700
> > 
> > > Build a kernel with CONFIG_NET-n and CONFIG_BLK_DEV_MD=m.
> > > Unless csum_partial() is built and kept by some arch Makefile,
> > > the result is:
> > > ERROR: "csum_partial" [drivers/md/md-mod.ko] undefined!
> > > make[1]: *** [__modpost] Error 1
> > > make: *** [modules] Error 2
> > > 
> > > 
> > > Any suggested solutions?
> > 
> > Anything which is every exported to modules, which ought to
> > be the situation in this case, should be obj-y not lib-y
> > right?
> That is also my understanding of lib-y - I should update makefiles.txt
> to reflect this..

I concur, but there seems to be quite a bit of change needed in
Makefile*s for this.  I'll begin with csum_partial() and trying
to use Arnd's suggestion as well.


---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

Re: BLK_DEV_MD with CONFIG_NET

[Adrian Bunk]

On Wed, Mar 21, 2007 at 11:30:24PM +0100, Arnd Bergmann wrote:
> On Wednesday 21 March 2007 13:02:46 Sam Ravnborg wrote:
> > > Anything which is every exported to modules, which ought to
> > > be the situation in this case, should be obj-y not lib-y
> > > right?
> >
> > That is also my understanding of lib-y - I should update makefiles.txt
> > to reflect this..
> 
> Strictly speaking, it could well be obj-m instead of obj-y if it
> is _only_ used by modules. OTOH, it makes the Makefile a lot simpler
> to not optimize for this case.

No, it is only used by modules only if CONFIG_BLK_DEV_MD=m and CONFIG_NET=n...

And except for some legacy drivers, there shouldn't be any module-only 
code in the kernel.

The solution is either obj-y or obj-$(CONFIG_CSUM_PARIAL).

> 	Arnd

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed