Re: [PATCH v1 3/3] kernel.h: Split out container_of() and typeof_memeber() macros

[Andy Shevchenko <andriy.shevchenko@linux.intel.com>]

On Tue, Jul 13, 2021 at 12:37:46PM +0200, Greg Kroah-Hartman wrote:
> On Tue, Jul 13, 2021 at 11:45:41AM +0300, Andy Shevchenko wrote:
> > kernel.h is being used as a dump for all kinds of stuff for a long time.
> > Here is the attempt cleaning it up by splitting out container_of() and
> > typeof_memeber() macros.
> 
> That feels messy, why?

Because the headers in the kernel are messy.

> Reading one .h file for these common
> macros/defines is fine, why are container_of and typeof somehow
> deserving of their own .h files?

It's explained here. There are tons of drivers that includes kernel.h for only
a few or even solely for container_of() macro.

> What speedups are you seeing by
> splitting this up?

C preprocessing.

> > At the same time convert users in the header and other folders to use it.
> > Though for time being include new header back to kernel.h to avoid twisted
> > indirected includes for existing users.
> > 
> > Note, there are _a lot_ of headers and modules that include kernel.h solely
> > for one of these macros and this allows to unburden compiler for the twisted
> > inclusion paths and to make new code cleaner in the future.
> > 
> > Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> > ---
> >  include/kunit/test.h         | 14 ++++++++++++--
> >  include/linux/container_of.h | 37 ++++++++++++++++++++++++++++++++++++
> >  include/linux/kernel.h       | 31 +-----------------------------
> >  include/linux/kobject.h      | 14 +++++++-------
> 
> Why are all of these changes needed to kobject.h for this one change?
> This diff:
> 
> > --- a/include/linux/kobject.h
> > +++ b/include/linux/kobject.h
> > @@ -15,18 +15,18 @@
> >  #ifndef _KOBJECT_H_
> >  #define _KOBJECT_H_
> >  
> > -#include <linux/types.h>
> > -#include <linux/list.h>
> > -#include <linux/sysfs.h>
> > +#include <linux/atomic.h>
> >  #include <linux/compiler.h>
> > -#include <linux/spinlock.h>
> > +#include <linux/container_of.h>
> > +#include <linux/list.h>
> >  #include <linux/kref.h>
> >  #include <linux/kobject_ns.h>
> > -#include <linux/kernel.h>
> >  #include <linux/wait.h>
> > -#include <linux/atomic.h>
> > -#include <linux/workqueue.h>
> > +#include <linux/spinlock.h>
> > +#include <linux/sysfs.h>
> > +#include <linux/types.h>
> >  #include <linux/uidgid.h>
> > +#include <linux/workqueue.h>
> 
> Is a lot more changes than the "split the macros out" deserves.
> 
> Please make this a separate change, remember to only do one thing at a
> time (this patch is at least 2 changes...)
> 
> so NAK, this change isn't ok as-is.

Fair enough. I will remove these conversions from the patch in v2.

Thanks for review!

-- 
With Best Regards,
Andy Shevchenko