RE: [PATCH 1/1] Drivers: hv: vmbus: Fix signaling logic in hv_need_to_signal_on_read()

[KY Srinivasan]

> -----Original Message-----
> From: K. Y. Srinivasan [mailto:kys@xxxxxxxxxxxxx]
> Sent: Saturday, April 2, 2016 3:44 PM
> To: gregkh@xxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx;
> devel@xxxxxxxxxxxxxxxxxxxxxx; olaf@xxxxxxxxx; apw@xxxxxxxxxxxxx;
> vkuznets@xxxxxxxxxx; jasowang@xxxxxxxxxx
> Cc: KY Srinivasan <kys@xxxxxxxxxxxxx>; stable@xxxxxxxxxxxxxxx
> Subject: [PATCH 1/1] Drivers: hv: vmbus: Fix signaling logic in
> hv_need_to_signal_on_read()
> 
> On the consumer side, we have interrupt driven flow management of the
> producer. It is sufficient to base the signaling decision on the
> amount of space that is available to write after the read is complete.
> The current code samples the previous available space and uses this
> in making the signaling decision. This state can be stale and is
> unnecessary. Since the state can be stale, we end up not signaling
> the host (when we should) and this can result in a hang. Fix this
> problem by removing the unnecessary check. I would like to thank
> Arseney Romanenko <arseneyr@xxxxxxxxxxxxx> for pointing out this issue.
> 
> Also, issue a full memory barrier before making the signaling descision
> to correctly deal with potential reordering of the write (read index)
> followed by the read of pending_sz.

Greg,

Please drop this; I sent the wrong version of the patch. Sorry for
The confusion.

K. Y
> 
> Signed-off-by: K. Y. Srinivasan <kys@xxxxxxxxxxxxx>
> Tested-by: Dexuan Cui <decui@xxxxxxxxxxxxx>
> Cc: <stable@xxxxxxxxxxxxxxx>
> ---
>  drivers/hv/ring_buffer.c |   20 ++++++++++++++++----
>  1 files changed, 16 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
> index 5613e2b..e00b632 100644
> --- a/drivers/hv/ring_buffer.c
> +++ b/drivers/hv/ring_buffer.c
> @@ -103,8 +103,7 @@ static bool hv_need_to_signal(u32 old_write, struct
> hv_ring_buffer_info *rbi)
>   *    there is room for the producer to send the pending packet.
>   */
> 
> -static bool hv_need_to_signal_on_read(u32 prev_write_sz,
> -				      struct hv_ring_buffer_info *rbi)
> +static bool hv_need_to_signal_on_read(struct hv_ring_buffer_info *rbi)
>  {
>  	u32 cur_write_sz;
>  	u32 r_size;
> @@ -112,6 +111,19 @@ static bool hv_need_to_signal_on_read(u32
> prev_write_sz,
>  	u32 read_loc = rbi->ring_buffer->read_index;
>  	u32 pending_sz = rbi->ring_buffer->pending_send_sz;
> 
> +	/*
> +	 * Issue a full memory barrier before making the signaling decision.
> +	 * Here is the reason for having this barrier:
> +	 * If the reading of the pend_sz (in this function)
> +	 * were to be reordered and read before we commit the new read
> +	 * index (in the calling function)  we could
> +	 * have a problem. If the host were to set the pending_sz after we
> +	 * have sampled pending_sz and go to sleep before we commit the
> +	 * read index, we could miss sending the interrupt. Issue a full
> +	 * memory barrier to address this.
> +	 */
> +	mb();
> +
>  	/* If the other end is not blocked on write don't bother. */
>  	if (pending_sz == 0)
>  		return false;
> @@ -120,7 +132,7 @@ static bool hv_need_to_signal_on_read(u32
> prev_write_sz,
>  	cur_write_sz = write_loc >= read_loc ? r_size - (write_loc - read_loc)
> :
>  			read_loc - write_loc;
> 
> -	if ((prev_write_sz < pending_sz) && (cur_write_sz >= pending_sz))
> +	if (cur_write_sz >= pending_sz)
>  		return true;
> 
>  	return false;
> @@ -455,7 +467,7 @@ int hv_ringbuffer_read(struct hv_ring_buffer_info
> *inring_info,
>  	/* Update the read index */
>  	hv_set_next_read_location(inring_info, next_read_location);
> 
> -	*signal = hv_need_to_signal_on_read(bytes_avail_towrite,
> inring_info);
> +	*signal = hv_need_to_signal_on_read(inring_info);
> 
>  	return ret;
>  }
> --
> 1.7.4.1