[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <49ca7920-d429-434a-aede-1a200e8d5ce8@linux.dev>
Date: Thu, 7 Mar 2024 12:26:00 -0500
From: Sean Anderson <sean.anderson@...ux.dev>
To: Christophe Leroy <christophe.leroy@...roup.eu>,
"David S . Miller" <davem@...emloft.net>, Eric Dumazet
<edumazet@...gle.com>, Jakub Kicinski <kuba@...nel.org>,
Paolo Abeni <pabeni@...hat.com>,
"netdev@...r.kernel.org" <netdev@...r.kernel.org>
Cc: Vladimir Oltean <vladimir.oltean@....com>, Roy Pledge
<roy.pledge@....com>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"stable@...r.kernel.org" <stable@...r.kernel.org>,
Li Yang <leoyang.li@....com>, Scott Wood <oss@...error.net>,
Claudiu Manoil <claudiu.manoil@....com>,
Camelia Groza <camelia.groza@....com>,
Steffen Trumtrar <s.trumtrar@...gutronix.de>,
"linuxppc-dev@...ts.ozlabs.org" <linuxppc-dev@...ts.ozlabs.org>,
"linux-arm-kernel@...ts.infradead.org" <linux-arm-kernel@...ts.infradead.org>
Subject: Re: [RESEND2 PATCH net v4 2/2] soc: fsl: qbman: Use raw spinlock for
cgr_lock
On 3/5/24 17:18, Christophe Leroy wrote:
>
>
> Le 05/03/2024 à 19:14, Sean Anderson a écrit :
>> [Vous ne recevez pas souvent de courriers de sean.anderson@...ux.dev. Découvrez pourquoi ceci est important à https://aka.ms/LearnAboutSenderIdentification ]
>>
>> Hi,
>>
>> On 2/23/24 11:02, Sean Anderson wrote:
>>> On 2/23/24 00:38, Christophe Leroy wrote:
>>>> Le 22/02/2024 à 18:07, Sean Anderson a écrit :
>>>>> [Vous ne recevez pas souvent de courriers de sean.anderson@...ux.dev. Découvrez pourquoi ceci est important à https://aka.ms/LearnAboutSenderIdentification ]
>>>>>
>>>>> cgr_lock may be locked with interrupts already disabled by
>>>>> smp_call_function_single. As such, we must use a raw spinlock to avoid
>>>>> problems on PREEMPT_RT kernels. Although this bug has existed for a
>>>>> while, it was not apparent until commit ef2a8d5478b9 ("net: dpaa: Adjust
>>>>> queue depth on rate change") which invokes smp_call_function_single via
>>>>> qman_update_cgr_safe every time a link goes up or down.
>>>>
>>>> Why a raw spinlock to avoid problems on PREEMPT_RT, can you elaborate ?
>>>
>>> smp_call_function always runs its callback in hard IRQ context, even on
>>> PREEMPT_RT, where spinlocks can sleep. So we need to use raw spinlocks
>>> to ensure we aren't waiting on a sleeping task. See the first bug report
>>> for more discussion.
>>>
>>> In the longer term it would be better to switch to some other
>>> abstraction.
>>
>> Does this make sense to you?
>
> Yes that fine, thanks for the clarification. Maybe you can explain that
> in the patch description in case you send a v5.
Hm, I thought I put this description in the commit message already.
Maybe something like
| smp_call_function always runs its callback in hard IRQ context, even on
| PREEMPT_RT, where spinlocks can sleep. So we need to use a raw spinlock
| for cgr_lock to ensure we aren't waiting on a sleeping task.
|
| Although this bug has existed for a while, it was not apparent until
| commit ef2a8d5478b9 ("net: dpaa: Adjust queue depth on rate change")
| which invokes smp_call_function_single via qman_update_cgr_safe every
| time a link goes up or down.
would be clearer.
--Sean
Powered by blists - more mailing lists