Re: drivers/w1/: why is dscore.c not ds9490r.c ?

[Adrian Bunk <bunk@stusta.de>]

On Mon, Nov 22, 2004 at 08:37:00PM +0300, Evgeniy Polyakov wrote:
> On Mon, 2004-11-22 at 18:19 +0100, Adrian Bunk wrote:
> > On Mon, Nov 22, 2004 at 08:05:09PM +0300, Evgeniy Polyakov wrote:
> > > On Mon, 2004-11-22 at 17:51 +0100, Adrian Bunk wrote:
> > > > On Mon, Nov 22, 2004 at 07:25:45PM +0300, Evgeniy Polyakov wrote:
> > > > > 
> > > > > > How would a different w1 bus master chip look like in 
> > > > > > drivers/w1/Makefile?
> > > > > 
> > > > > obj-m: proprietary_module.o
> > > > > proprietary_module-objs: dscore.o proprietary_module_init.o
> > > > > 
> > > > > Actually it will live outside the kernel tree, but will require ds2490
> > > > > driver.
> > > > > It could be called ds2490.c but I think dscore is better name.
> > > > 
> > > > Why are you talking about proprietary modules living outside the kernel 
> > > > tree?
> > > > 
> > > > The only interesting case is the one of modules shipped with the kernel.
> > > > And for them, this will break at link time if two such modules are 
> > > > included statically into the kernel.
> > > 
> > > If we _currently_ do not have any open hw/module that depends on ds2490
> > > core then it does not
> > > mean that tomorrow noone will add it.
> > 
> > Once again:
> >   _this will break at link time if two such modules are included 
> >    statically into the kernel_
> > 
> > obj-$(CONFIG_W1_DS9490)         += ds9490r.o 
> > ds9490r-objs    := dscore.o
> > 
> > obj-$(CONFIG_W1_FOO)         += foo.o 
> > foo-objs    := dscore.o
> > 
> 
> that should be follwing:
> 
> Kconfig:
> foo depends on ds9490r
> 
> obj-$() += foo.o
> foo-objs := foo_1.o foo_2.o
> 
> It just happened that ds9490r does not need any other parts but
> dscore.o.
> That is why ds9490r.o have only dscore.o in it's dependency.

If foo_1 or foo_2 is dscore, you get exactly the compile breakage I 
described.

> > This will break with CONFIG_W1_DS9490=y and CONFIG_W1_FOO=y.
> > 
> > 
> > That drivers/w1/ contains many EXPORT_SYMBOL's with no in-kernel users 
> > is a different issue I might send a separate patch for (that besides 
> > proprietary modules there might come some day open source drivers using 
> > them is not a reason).
> 
> Why remove existing non disturbing set of exported functions?
> Are they violate some unknown issues?

If an export is currently unused, there's no need to export it.

If an export is only used for proprietary modules, that's a reason for 
an immediate removal of this export.

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