Re: [PATCH 1/1] crypto: ux500: hash: Add namespacing to hash_init()

[Lee Jones]

On Tue, 30 Jun 2020, Herbert Xu wrote:

> On Mon, Jun 29, 2020 at 01:30:03PM +0100, Lee Jones wrote:
> > A recent change to the Regulator consumer API (which this driver
> > utilises) add prototypes for the some suspend functions.  These
> > functions require including header file include/linux/suspend.h.
> > 
> > The following tree of includes affecting this driver will be
> > present:
> > 
> >    In file included from include/linux/elevator.h:6,
> >                     from include/linux/blkdev.h:288,
> >                     from include/linux/blk-cgroup.h:23,
> >                     from include/linux/writeback.h:14,
> >                     from include/linux/memcontrol.h:22,
> >                     from include/linux/swap.h:9,
> >                     from include/linux/suspend.h:5,
> >                     from include/linux/regulator/consumer.h:35,
> >                     from drivers/crypto/ux500/hash/hash_core.c:28:
> > 
> > include/linux/elevator.h pulls in include/linux/hashtable.h which
> > contains its own version of hash_init().  This confuses the build
> > system and results in the following error (amongst others):
> > 
> >  drivers/crypto/ux500/hash/hash_core.c:1362:19: error: passing argument 1 of '__hash_init' from incompatible pointer type [-Werror=incompatible-pointer-types]
> >  1362 |  return hash_init(req);
> > 
> > Fix this by namespacing the local hash_init() such that the
> > source of confusion is removed.
> > 
> > Cc: Linus Walleij <linus.walleij@xxxxxxxxxx>
> > Cc: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>
> > Cc: David S. Miller <davem@xxxxxxxxxxxxx>
> > Cc: linux-crypto@xxxxxxxxxxxxxxx
> > Signed-off-by: Lee Jones <lee.jones@xxxxxxxxxx>
> > ---
> > 
> > Ideally this should go into v5.8's -rcs else it runs the risk of
> > breaking when Linus pulls everything in for v5.9-rc1.

[...]

> I also dislike pulling in the kitchen sink when all you need in
> consumer.h is the definition of suspend_state_t.  A better solution
> would be to move the definition of suspend_state_t into linux/types.h
> and including that instead of suspend.h in consumer.h.

IMHO, including (whole) headers into source/header files is the norm.
Even if only a small portion is actually referenced.  Very seldom do
consumers of an API use more than a fraction of what is available.
Whether it's a couple of function calls, a struct or a type.

Pulling headers apart and placing items in more convenient places
i.e. into headers which are more commonly included, messes with the
compartmentalisation of subsystems and sounds like more of a hack than
simply saying "to enable suspend functions we need to reference the
suspend API" like we are here.

> I have no objections to this patch.  However, I'd rather put
> it on a topic branch which you could pull rather than pushing
> it into 5.8 straight away.

An immutable branch sounds like a sensible solution.  Thanks.

-- 
Lee Jones [æçæ]
Senior Technical Lead - Developer Services
Linaro.org â Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog