From: Cheng Chao <cs.os.kernel@gmail.com>
To: Marc Zyngier <marc.zyngier@arm.com>
Cc: tglx@linutronix.de, jason@lakedaemon.net,
linux-kernel@vger.kernel.org, cs.os.kernel@gmail.com
Subject: Re: [PATCH] irqchip/gic: Enable gic_set_affinity set more than one cpu
Date: Fri, 14 Oct 2016 10:08:10 +0800 [thread overview]
Message-ID: <1c4adb8a-f7f7-3474-273a-edf34f575b8d@gmail.com> (raw)
In-Reply-To: <20161013163140.5f23abce@arm.com>
Marc,
Thanks for your comments.
Cheng
on 10/13/2016 11:31 PM, Marc Zyngier wrote:
> On Thu, 13 Oct 2016 18:57:14 +0800
> Cheng Chao <cs.os.kernel@gmail.com> wrote:
>
>> GIC can distribute an interrupt to more than one cpu,
>> but now, gic_set_affinity sets only one cpu to handle interrupt.
>
> What makes you think this is a good idea? What purpose does it serves?
> I can only see drawbacks to this: You're waking up more than one CPU,
> wasting power, adding jitter and clobbering the cache.
>
> I assume you see a benefit to that approach, so can you please spell it
> out?
>
Ok, You are right, but the performance is another point that we should consider.
We use E1 device to transmit/receive video stream. we find that E1's interrupt is
only on the one cpu that cause this cpu usage is almost 100%,
but other cpus is much lower load, so the performance is not good.
the cpu is 4-core.
so add CONFIG_ARM_GIC_AFFINITY_SINGLE_CPU is better?
thus we can make a trade-off between the performance with the power etc.
>>
>> Signed-off-by: Cheng Chao <cs.os.kernel@gmail.com>
>> ---
>> drivers/irqchip/irq-gic.c | 28 ++++++++++++++++++++++++----
>> 1 file changed, 24 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/irqchip/irq-gic.c b/drivers/irqchip/irq-gic.c
>> index 58e5b4e..198d33f 100644
>> --- a/drivers/irqchip/irq-gic.c
>> +++ b/drivers/irqchip/irq-gic.c
>> @@ -328,18 +328,38 @@ static int gic_set_affinity(struct irq_data *d, const struct cpumask *mask_val,
>> unsigned int cpu, shift = (gic_irq(d) % 4) * 8;
>> u32 val, mask, bit;
>> unsigned long flags;
>> + u32 valid_mask;
>>
>> - if (!force)
>> - cpu = cpumask_any_and(mask_val, cpu_online_mask);
>> - else
>> + if (!force) {
>> + valid_mask = cpumask_bits(mask_val)[0];
>> + valid_mask &= cpumask_bits(cpu_online_mask)[0];
>> +
>> + cpu = cpumask_any((struct cpumask *)&valid_mask);
>
> What is wrong with with cpumask_any_and?
>
#define cpumask_any_and(mask1, mask2) cpumask_first_and((mask1), (mask2))
#define cpumask_any(srcp) cpumask_first(srcp)
There is no wrong with the cpumask_any_and.
>> + } else {
>> cpu = cpumask_first(mask_val);
>> + }
>>
>> if (cpu >= NR_GIC_CPU_IF || cpu >= nr_cpu_ids)
>> return -EINVAL;
>>
>> gic_lock_irqsave(flags);
>> mask = 0xff << shift;
>> - bit = gic_cpu_map[cpu] << shift;
>> +
>> + if (!force) {
>> + bit = 0;
>> +
>> + for_each_cpu(cpu, (struct cpumask *)&valid_mask) {
>> + if (cpu >= NR_GIC_CPU_IF || cpu >= nr_cpu_ids)
>> + break;
>
> Shouldn't that be an error?
>
tested, no error.
at the beginning, I code such like,
cpumask_var_t valid_mask;
alloc_cpumask_var(&valid_mask, GFP_KERNEL);
cpumask_and(valid_mask, mask_val, cpu_online_mask);
for_each_cpu(cpu, valid_mask) {
}
but alloc_cpumask_var maybe fail, so
if (!alloc_cpumask_var(&valid_mask, GFP_KERNEL)) {
/* fail*/
} else {
}
a little more complex.
>> +
>> + bit |= gic_cpu_map[cpu];
>> + }
>> +
>> + bit = bit << shift;
>> + } else {
>> + bit = gic_cpu_map[cpu] << shift;
>> + }
>> +
>> val = readl_relaxed(reg) & ~mask;
>> writel_relaxed(val | bit, reg);
>> gic_unlock_irqrestore(flags);
>
> Thanks,
>
> M.
>
next prev parent reply other threads:[~2016-10-14 2:09 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-10-13 10:57 [PATCH] irqchip/gic: Enable gic_set_affinity set more than one cpu Cheng Chao
2016-10-13 11:11 ` Cheng Chao
2016-10-13 15:31 ` Marc Zyngier
2016-10-14 2:08 ` Cheng Chao [this message]
2016-10-14 17:33 ` Marc Zyngier
2016-10-15 7:23 ` Cheng Chao
2016-10-25 10:09 ` Marc Zyngier
2016-10-26 2:04 ` Cheng Chao
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1c4adb8a-f7f7-3474-273a-edf34f575b8d@gmail.com \
--to=cs.os.kernel@gmail.com \
--cc=jason@lakedaemon.net \
--cc=linux-kernel@vger.kernel.org \
--cc=marc.zyngier@arm.com \
--cc=tglx@linutronix.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.