[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <56be3f06-4d4a-4128-a8b4-dbe12d37ce78@infradead.org>
Date: Tue, 18 Aug 2020 08:52:38 -0700
From: Randy Dunlap <rdunlap@...radead.org>
To: Peter Zijlstra <peterz@...radead.org>, mingo@...nel.org,
torvalds@...ux-foundation.org
Cc: linux-kernel@...r.kernel.org, will@...nel.org, paulmck@...nel.org,
hch@....de, axboe@...nel.dk, chris@...is-wilson.co.uk,
davem@...emloft.net, kuba@...nel.org, fweisbec@...il.com,
oleg@...hat.com, vincent.guittot@...aro.org
Subject: Re: [RFC][PATCH v2 10/10] irq_work: Add a few comments
Hi Peter,
a few typos below...
On 8/18/20 3:51 AM, Peter Zijlstra wrote:
>
> Signed-off-by: Peter Zijlstra (Intel) <peterz@...radead.org>
> ---
> include/linux/irq_work.h | 80 +++++++++++++++++++++++++++++++++++++++++------
> kernel/irq_work.c | 10 +++++
> 2 files changed, 80 insertions(+), 10 deletions(-)
>
> --- a/include/linux/irq_work.h
> +++ b/include/linux/irq_work.h
> @@ -48,12 +47,45 @@ static inline bool irq_work_is_busy(stru
>
> #ifdef CONFIG_IRQ_WORK
>
> +/*
> + * irq_work_queue(): run @work in IRQ context on this CPU
> + * @work: work to run
> + *
> + * Self-IPI, NMI-safe
> + *
> + * When the function returns false; @work is already queued and
> + * any eventual execution of it's func() is guaranteed to see
its
> + * any state before the failing enqueue.
> + */
> bool irq_work_queue(struct irq_work *work);
> +
> +/*
> + * irq_work_queue_on(): run @work in IRQ context on @cpu
> + * @work: work to run
> + * @cpu: cpu to run @work on
> + *
> + * *NOT* NMI-safe
> + *
> + * When the function returns false; @work is already queued and
false,
> + * any eventual execution of it's func() is guaranteed to see
its
> + * any state before the failing enqueue.
> + */
> @@ -67,8 +99,36 @@ static inline bool irq_work_needs_cpu(vo
> +/*
> + * irq_work_queue_remote_state(): like irq_work_queue_remote() except dangerous
> + * @cpu:
> + * @work:
> + *
> + * DO NOT USE, this function is horrible/dangerous.
> + *
> + * The test-and-set-PENDING is not atomic, it also doesn't set
> + * the BUSY bit and with that breaks irq_work_sync().
> + *
> + * This means that the caller needs external serialization; life-time,
> + * where relevant, also needs to be externally orchestated.
orchestrated.
> + *
> + * There is no validation/debugging to help you if you get it wrong.
> + */
> extern int irq_work_queue_remote_static(int cpu, struct irq_work *work);
> +
> extern void irq_work_single(void *arg);
> #endif
>
--
~Randy
Powered by blists - more mailing lists