On Sat, Jun 19, 2010 at 19:42, Lars-Peter Clausen <lars@xxxxxxxxxx> wrote:
> Marek Vasut wrote:
>> Dne So 19. června 2010 15:05:18 Lars-Peter Clausen napsal(a):
>>> Marek Vasut wrote:
>>>> Dne So 19. června 2010 07:08:20 Lars-Peter Clausen napsal(a):
>>>>> This patch adds support for the RTC unit on JZ4740 SoCs.
>>>>> +static void jz4740_rtc_ctrl_set_bits(struct jz4740_rtc *rtc, uint32_t
>>>>> mask, + uint32_t val)
>>>>> +{
>>>>> + unsigned long flags;
>>>>> + uint32_t ctrl;
>>>>> +
>>>>> + spin_lock_irqsave(&rtc->lock, flags);
>>>>>
>>>> Can't we use local_irq_save()/local_irq_restore() ?
>>>>
>>> Why would that be preferable? In the non-debug, non-rt case this will
>>> expand to local_irq_{save,restore} anyway, but you'll lose the semantics
>>> of an lock.
>>>
>>
>> I believe on SMP systems, local_irq_save will give you finer locking
>> granularity.
>>
> Hm, not sure about that. But this is on a non SMP system anyway.
If the driver will ever be used on a SMP system, local_irq_save() will
not protect
against concurrent accesses on different CPUs. So it's better to use
spin_lock_irqsave().
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
