LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
From: Zenghui Yu <yuzenghui@huawei.com>
To: Andre Przywara <andre.przywara@arm.com>
Cc: <marc.zyngier@arm.com>, <eric.auger@redhat.com>,
	<drjones@redhat.com>, <tglx@linutronix.de>,
	<jason@lakedaemon.net>, <wanghaibin.wang@huawei.com>,
	<linux-kernel@vger.kernel.org>,
	<linux-arm-kernel@lists.infradead.org>
Subject: Re: [RFC PATCH] irqchip/gic-v3: Correct the usage of GICD_CTLR's RWP field
Date: Mon, 13 May 2019 19:55:34 +0800	[thread overview]
Message-ID: <0d1febde-30de-6474-4cca-a0a17963a329@huawei.com> (raw)
In-Reply-To: <20190513093704.0b293de0@donnerap.cambridge.arm.com>

Hi Andre,

On 2019/5/13 16:37, Andre Przywara wrote:
> On Mon, 13 May 2019 04:15:54 +0000
> Zenghui Yu <yuzenghui@huawei.com> wrote:
> 
> Hi,
> 
>> As per ARM IHI 0069D, GICD_CTLR's RWP field tracks updates to:
>>
>> 	GICD_CTLR's Group Enable bits, for transitions from 1 to 0 only
>> 	GICD_CTLR's ARE bits, E1NWF bit and DS bit (if we have)
>> 	GICD_ICENABLER<n>
>>
>> We seemed use this field in an inappropriate way, somewhere in the
>> GIC-v3 driver. For some examples:
>>
>> In gic_set_affinity(), if the interrupt was not enabled, we only need
>> to write GICD_IROUTER<n> with the appropriate mpidr value. Updates to
>> GICD_IROUTER will not be tracked by RWP field, and we can remove the
>> unnecessary RWP waiting.
> 
> I am not sure this is the proper fix, see below inline.
> 
>> In gic_dist_init(), we "Enable distributor with ARE, Group1" by writing
>> to GICD_CTLR, and we should use gic_dist_wait_for_rwp() here.
> 
> That looks reasonable, yes.
> 
>> These two are obvious cases, and there are some other cases where we don't
>> need to do RWP waiting, such as in gic_configure_irq() and gic_poke_irq().
>> But too many modifications makes me not confident. It's hard for me to say
>> whether this patch is doing the right thing and how does the RWP waiting
>> affect the system, thus RFC.
> 
> So did you actually see a problem, and this patch fixes it? Or was this
> just discovered by code inspection and comparing to the spec?

The latter ;-)

>> Signed-off-by: Zenghui Yu <yuzenghui@huawei.com>
>> ---
>>   drivers/irqchip/irq-gic-v3.c | 8 ++------
>>   1 file changed, 2 insertions(+), 6 deletions(-)
>>
>> diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
>> index 15e55d3..8d63950 100644
>> --- a/drivers/irqchip/irq-gic-v3.c
>> +++ b/drivers/irqchip/irq-gic-v3.c
>> @@ -600,6 +600,7 @@ static void __init gic_dist_init(void)
>>   	/* Enable distributor with ARE, Group1 */
>>   	writel_relaxed(GICD_CTLR_ARE_NS | GICD_CTLR_ENABLE_G1A |
>> GICD_CTLR_ENABLE_G1, base + GICD_CTLR);
>> +	gic_dist_wait_for_rwp();
>>   
>>   	/*
>>   	 * Set all global interrupts to the boot CPU only. ARE must be
>> @@ -986,14 +987,9 @@ static int gic_set_affinity(struct irq_data *d,
>> const struct cpumask *mask_val,
>>   	gic_write_irouter(val, reg);
>>   
>> -	/*
>> -	 * If the interrupt was enabled, enabled it again. Otherwise,
>> -	 * just wait for the distributor to have digested our changes.
>> -	 */
>> +	/* If the interrupt was enabled, enabled it again. */
>>   	if (enabled)
>>   		gic_unmask_irq(d);
>> -	else
>> -		gic_dist_wait_for_rwp();
> 
> I think you are right in this is not needed here.
> But I guess this call belongs further up in this function, after the
> gic_mask_irq() call, as this one writes to GICD_ICENABLER. So in case this
> IRQ was enabled, we should wait for the distributor to have properly
> disabled it, before changing its affinity.

I still think we don't need this call in gic_set_affinity().
Actually, the writes to GICD_ICENABLER happens in gic_poke_irq(). And we
already have a gic_dist_wait_for_rwp() there, named rwp_wait().


thanks,
zenghui

> 
> Cheers,
> Andre.
> 
>>   
>>   	irq_data_update_effective_affinity(d, cpumask_of(cpu));
>>   
> 
> 
> .
> 


      reply	other threads:[~2019-05-13 11:57 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-05-13  4:15 Zenghui Yu
2019-05-13  8:37 ` Andre Przywara
2019-05-13 11:55   ` Zenghui Yu [this message]

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=0d1febde-30de-6474-4cca-a0a17963a329@huawei.com \
    --to=yuzenghui@huawei.com \
    --cc=andre.przywara@arm.com \
    --cc=drjones@redhat.com \
    --cc=eric.auger@redhat.com \
    --cc=jason@lakedaemon.net \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=marc.zyngier@arm.com \
    --cc=tglx@linutronix.de \
    --cc=wanghaibin.wang@huawei.com \
    --subject='Re: [RFC PATCH] irqchip/gic-v3: Correct the usage of GICD_CTLR'\''s RWP field' \
    /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

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).