Re: [PATCH v9 2/5] qrwlock x86: Enable x86 to use queue read/writelock

[Waiman Long]

On 01/20/2014 11:16 AM, Steven Rostedt wrote:
> On Mon, 20 Jan 2014 17:08:06 +0100
> Peter Zijlstra<peterz@xxxxxxxxxxxxx>  wrote:
>
> > On Tue, Jan 14, 2014 at 11:44:04PM -0500, Waiman Long wrote:
> > > This patch makes the necessary changes at the x86 architecture specific
> > > layer to enable the presence of the CONFIG_QUEUE_RWLOCK kernel option
> > > to replace the read/write lock by the queue read/write lock.
> > >
> > > It also enables the CONFIG_QUEUE_RWLOCK option by default for x86 which
> > > will force the use of queue read/write lock. That will greatly improve
> > > the fairness of read/write lock and eliminate live-lock situation
> > > where one task may not get the lock for an indefinite period of time.
> > >
> > > Signed-off-by: Waiman Long<Waiman.Long@xxxxxx>
> > > Reviewed-by: Paul E. McKenney<paulmck@xxxxxxxxxxxxxxxxxx>
> > > ---
> > >   arch/x86/Kconfig                      |    1 +
> > >   arch/x86/include/asm/spinlock.h       |    2 ++
> > >   arch/x86/include/asm/spinlock_types.h |    4 ++++
> > >   3 files changed, 7 insertions(+), 0 deletions(-)
> > This patch doesn't have nearly enough -, you should rip out all traces
> > of the old rwlock. There's no point in having it configurable, either
> > qrwlock is good or not.
> Although having something that switches back to the old method gives
> more users the way to compare. Especially to catch any regressions.
>
> Make one release with it configurable, then drop it.
>
> -- Steve

Yes, I agree that it will be more prudent to leave the original rwlock 
implementation there for one or two release cycles before getting rid of 
them entirely.

-Longman
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/