Andrea Arcangeli (arcangeli@mbox.queen.it)
Wed, 14 Jan 1998 17:56:51 +0100 (CET)
On Mon, 12 Jan 1998, Philip Blundell wrote:
>>On Mon, 12 Jan 1998, Philip Blundell wrote:
>>
>>> Be warned that save_and_cli() is frowned upon by Linus, which is why I
>>> took it out of the vger tree again.
>>
>>Oh? How come?
>
>Alan explained that it's not so good on SMP. Linus wants people to use local
>IRQ spinlocks instead. Take a look at the PLIP changes in 2.1.78 for an
>example.
So cli() don' t mask well critical code that can be managed also from
others CPU, right? cli() clear interrupts only on the CPU that executed
cli(), or no?
Since we can insmod two pardevice on two different CPU at the same time,
we should also mask the code that update the pardevice/parport list (in
parport_register_...()) with the cheap spin_lock()/spin_unlock() version
of spinlocks I suppose...
Andrea[s] Arcangeli
-- To unsubscribe, send mail to: linux-parport-request@torque.net --
-- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2.0b3 on Wed 30 Dec 1998 - 10:17:19 EST