Re: [PATCH 4/7] sempahore: add a helper for a concurrency limiter

[Luis Chamberlain]

On Wed, Mar 29, 2023 at 09:21:12AM +0200, Peter Zijlstra wrote:
> On Tue, Mar 28, 2023 at 10:31:46PM -0700, Luis Chamberlain wrote:
> > While I looked at re-using the old kernel/kmod.c (now kernel/module/kmod.c)
> > concurrency delimiter methodology for another place in the kernel Linus
> > noted that this could be simply replaced with a sempahore [0].
> > 
> > So add that so we we don't re-invent the wheel and make it obvious to use.
> > 
> > [0] https://lore.kernel.org/all/CAHk-=whkj6=wyi201JXkw9iT_eTUTsSx+Yb9d4OgmZFjDJA18g@xxxxxxxxxxxxxx/
> > 
> > Suggested-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> > Signed-off-by: Luis Chamberlain <mcgrof@xxxxxxxxxx>
> > ---
> >  include/linux/semaphore.h | 3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > diff --git a/include/linux/semaphore.h b/include/linux/semaphore.h
> > index 6694d0019a68..2ecdffdb9814 100644
> > --- a/include/linux/semaphore.h
> > +++ b/include/linux/semaphore.h
> > @@ -28,6 +28,9 @@ struct semaphore {
> >  #define DEFINE_SEMAPHORE(name)	\
> >  	struct semaphore name = __SEMAPHORE_INITIALIZER(name, 1)
> >  
> > +#define CONCURRENCY_LIMITER(name, n) \
> > +	struct semaphore name = __SEMAPHORE_INITIALIZER(name, n)
> > +
> 
> Why should this live in semaphore.h?

I have no preference, but sharing seems to have been better. Do you
have any recommendations?

  Luis