Re: [PATCH V12 1/4] ptp: Added a brand new class driver for ptpclocks.

From: Richard Cochran
Date: Sat Mar 26 2011 - 09:13:10 EST


On Wed, Mar 23, 2011 at 02:19:20PM -0700, John Stultz wrote:
> On Mon, 2011-02-28 at 08:57 +0100, Richard Cochran wrote:
> > +++ b/drivers/ptp/ptp_clock.c
> > @@ -0,0 +1,320 @@
> [snip]
> > +static void enqueue_external_timestamp(struct timestamp_event_queue *queue,
> > + struct ptp_clock_event *src)
> > +{
> > + struct ptp_extts_event *dst;
> > + unsigned long flags;
> > + u32 remainder;
> > +
> > + dst = &queue->buf[queue->tail];
>
> Doesn't the lock need to happen before you access the
> queue->buf[queue->tail] ?
>
> For example: What happens if two cpus enter the function at the same
> time, both get the same tail index, one overwrite the other's data, then
> both take turns bumping up the tail pointer?

Yes, thanks for that catch.

> > +struct timestamp_event_queue {
> > + struct ptp_extts_event buf[PTP_MAX_TIMESTAMPS];
> > + int head;
> > + int tail;
> > + spinlock_t lock;
> > +};
> > +
> > +struct ptp_clock {
> > + struct posix_clock clock;
> > + struct device *dev;
> > + struct ptp_clock_info *info;
> > + dev_t devid;
> > + int index; /* index into clocks.map */
> > + struct pps_device *pps_source;
> > + struct timestamp_event_queue tsevq; /* simple fifo for time stamps */
> > + struct mutex tsevq_mux; /* one process at a time reading the fifo */
> > + wait_queue_head_t tsev_wq;
> > +};
> > +
> > +static inline int queue_cnt(struct timestamp_event_queue *q)
> > +{
> > + int cnt = q->tail - q->head;
> > + return cnt < 0 ? PTP_MAX_TIMESTAMPS + cnt : cnt;
> > +}
>
> q->tail and head access probably need to happen only when locked.
>
> So probably need a comment that queue_cnt must be called only when
> holding the proper lock.

In this case, calling without a lock is allowed. However, I'll add
comment like the following.

* The function queue_cnt() is safe for readers to call without
* holding q->lock. Readers use this function to verify that the queue
* is nonempty before proceeding with a dequeue operation. The fact
* that a writer might concurrently increment the tail does not
* matter, since the queue remains nonempty nonetheless.

Thanks for your feedback,

Richard
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/