lkml.org 
[lkml]   [2019]   [Oct]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH tip/core/rcu 1/9] rcu: Upgrade rcu_swap_protected() to rcu_replace()
----- On Oct 3, 2019, at 3:09 PM, paulmck paulmck@kernel.org wrote:

> On Thu, Oct 03, 2019 at 01:21:14PM -0400, Mathieu Desnoyers wrote:
>> ----- On Oct 3, 2019, at 12:52 PM, paulmck paulmck@kernel.org wrote:
>>
>> > On Thu, Oct 03, 2019 at 12:35:30PM -0400, Mathieu Desnoyers wrote:
>> >> ----- On Oct 2, 2019, at 9:43 PM, paulmck paulmck@kernel.org wrote:
>> >>
>> >> > From: "Paul E. McKenney" <paulmck@kernel.org>
>> >> >
>> >> > Although the rcu_swap_protected() macro follows the example of swap(),
>> >> > the interactions with RCU make its update of its argument somewhat
>> >> > counter-intuitive. This commit therefore introduces an rcu_replace()
>> >> > that returns the old value of the RCU pointer instead of doing the
>> >> > argument update. Once all the uses of rcu_swap_protected() are updated
>> >> > to instead use rcu_replace(), rcu_swap_protected() will be removed.
>> >>
>> >> We expose the rcu_xchg_pointer() API in liburcu (Userspace RCU) project.
>> >> Any reason for not going that way and keep the kernel and user-space RCU
>> >> APIs alike ?
>> >>
>> >> It's of course fine if they diverge, but we might want to at least consider
>> >> if using the same API name would be OK.
>> >
>> > Different semantics. An rcu_xchg_pointer() allows concurrent updates,
>> > and rcu_replace() does not.
>> >
>> > But yes, if someone needs the concurrent updates, rcu_xchg_pointer()
>> > would certainly be my choice for the name.
>>
>> Then considering that its assignment counterpart is "rcu_assign_pointer()"
>> (and not "rcu_assign()"), would "rcu_replace_pointer()" be less ambiguous
>> about its intended use ?
>
> The sequence was rcu_swap_protected() -> rcu_swap() -> rcu_replace().
> Because that rcu_replace(), unlike rcu_swap_protected(), returns a
> value, the shorter name is valuable.
>
> Maybe we should have used rcu_assign() instead of rcu_assign_pointer(),
> but there is no point in that sort of change at this late date!

I agree that having both the "rcu_" prefix and the "_pointer" suffix is
somewhat redundant. Indeed it's not a worthwhile change for a pre-existing
API, but it is welcome for a new API.

Thanks!

Mathieu

>
> Thanx, Paul
>
>> Thanks,
>>
>> Mathieu
>>
>>
>> >
>> > Thanx, Paul
>> >
>> >> Thanks,
>> >>
>> >> Mathieu
>> >>
>> >>
>> >> >
>> >> > Link:
>> >> > https://lore.kernel.org/lkml/CAHk-=wiAsJLw1egFEE=Z7-GGtM6wcvtyytXZA1+BHqta4gg6Hw@mail.gmail.com/
>> >> > Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
>> >> > Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
>> >> > Cc: Bart Van Assche <bart.vanassche@wdc.com>
>> >> > Cc: Christoph Hellwig <hch@lst.de>
>> >> > Cc: Hannes Reinecke <hare@suse.de>
>> >> > Cc: Johannes Thumshirn <jthumshirn@suse.de>
>> >> > Cc: Shane M Seymour <shane.seymour@hpe.com>
>> >> > Cc: Martin K. Petersen <martin.petersen@oracle.com>
>> >> > ---
>> >> > include/linux/rcupdate.h | 18 ++++++++++++++++++
>> >> > 1 file changed, 18 insertions(+)
>> >> >
>> >> > diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
>> >> > index 75a2ede..3b73287 100644
>> >> > --- a/include/linux/rcupdate.h
>> >> > +++ b/include/linux/rcupdate.h
>> >> > @@ -383,6 +383,24 @@ do { \
>> >> > } while (0)
>> >> >
>> >> > /**
>> >> > + * rcu_replace() - replace an RCU pointer, returning its old value
>> >> > + * @rcu_ptr: RCU pointer, whose old value is returned
>> >> > + * @ptr: regular pointer
>> >> > + * @c: the lockdep conditions under which the dereference will take place
>> >> > + *
>> >> > + * Perform a replacement, where @rcu_ptr is an RCU-annotated
>> >> > + * pointer and @c is the lockdep argument that is passed to the
>> >> > + * rcu_dereference_protected() call used to read that pointer. The old
>> >> > + * value of @rcu_ptr is returned, and @rcu_ptr is set to @ptr.
>> >> > + */
>> >> > +#define rcu_replace(rcu_ptr, ptr, c) \
>> >> > +({ \
>> >> > + typeof(ptr) __tmp = rcu_dereference_protected((rcu_ptr), (c)); \
>> >> > + rcu_assign_pointer((rcu_ptr), (ptr)); \
>> >> > + __tmp; \
>> >> > +})
>> >> > +
>> >> > +/**
>> >> > * rcu_swap_protected() - swap an RCU and a regular pointer
>> >> > * @rcu_ptr: RCU pointer
>> >> > * @ptr: regular pointer
>> >> > --
>> >> > 2.9.5
>> >>
>> >> --
>> >> Mathieu Desnoyers
>> >> EfficiOS Inc.
>> > > http://www.efficios.com
>>
>> --
>> Mathieu Desnoyers
>> EfficiOS Inc.
> > http://www.efficios.com

--
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com

\
 
 \ /
  Last update: 2019-10-03 21:15    [W:0.186 / U:0.016 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site