Re: [PATCH char-misc 1/1] Drivers: hv: vmbus: Fix ring buffer signaling

[Stephen Hemminger]

What about the following simplification. If on older hosts you can avoid more 
code
and the additional barrier. Also cleanup comment wording somewhat.



>From de2d566d9092cbbc8e5974dea581617ef787ff69 Mon Sep 17 00:00:00 2001
From: Michael Kelley 
Date: Sat, 10 Feb 2018 20:48:49 +
Subject: [PATCH 1/9] Drivers: hv: vmbus: Fix ring buffer signaling

Fix bugs in signaling the Hyper-V host when freeing space in the
host->guest ring buffer:

1. The interrupt_mask must not be used to determine whether to signal
   on the host->guest ring buffer
2. The ring buffer write_index must be read (via hv_get_bytes_to_write)
   *after* pending_send_sz is read in order to avoid a race condition
3. Comparisons with pending_send_sz must treat the "equals" case as
   not-enough-space
4. Don't signal if the pending_send_sz feature is not present. Older
   versions of Hyper-V that don't implement this feature will poll.

Fixes: 03bad714a161 ("vmbus: more host signalling avoidance")
Signed-off-by: Michael Kelley 
---
 drivers/hv/ring_buffer.c | 67 +---
 1 file changed, 40 insertions(+), 27 deletions(-)

diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
index 50e071444a5c..d27cbb66279c 100644
--- a/drivers/hv/ring_buffer.c
+++ b/drivers/hv/ring_buffer.c
@@ -417,13 +417,25 @@ __hv_pkt_iter_next(struct vmbus_channel *channel,
 }
 EXPORT_SYMBOL_GPL(__hv_pkt_iter_next);
 
+/* How many bytes were read in this iterator cycle */
+static u32 hv_pkt_iter_bytes_read(const struct hv_ring_buffer_info *rbi)
+{
+   if (rbi->ring_buffer->read_index < rbi->priv_read_index)
+   return rbi->priv_read_index - rbi->ring_buffer->read_index;
+   else
+   return rbi->ring_datasize -
+   rbi->ring_buffer->read_index + rbi->priv_read_index;
+}
+
 /*
  * Update host ring buffer after iterating over packets.
  */
 void hv_pkt_iter_close(struct vmbus_channel *channel)
 {
struct hv_ring_buffer_info *rbi = >inbound;
-   u32 orig_write_sz = hv_get_bytes_to_write(rbi);
+   u32 curr_write_sz;
+   u32 delta = hv_pkt_iter_bytes_read(rbi);
+   u32 pending_sz;
 
/*
 * Make sure all reads are done before we update the read index since
@@ -433,39 +445,40 @@ void hv_pkt_iter_close(struct vmbus_channel *channel)
virt_rmb();
rbi->ring_buffer->read_index = rbi->priv_read_index;
 
+   /* Older versions do not require signalling */
+   if (!rbi->ring_buffer->feature_bits.feat_pending_send_sz)
+   return;
+
/*
-* 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.
+* If the reading of the pend_sz were to be reordered and read
+* before we commit the new read index then we would 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.
 */
virt_mb();
 
-   /* If host has disabled notifications then skip */
-   if (rbi->ring_buffer->interrupt_mask)
-   return;
+   pending_sz = READ_ONCE(rbi->ring_buffer->pending_send_sz);
 
-   if (rbi->ring_buffer->feature_bits.feat_pending_send_sz) {
-   u32 pending_sz = READ_ONCE(rbi->ring_buffer->pending_send_sz);
+   /*
+* Ensure the read of write_index in hv_get_bytes_to_write()
+* happens after the read of pending_send_sz.
+*/
+   virt_rmb();
+   curr_write_sz = hv_get_bytes_to_write(rbi);
 
-   /*
-* If there was space before we began iteration,
-* then host was not blocked. Also handles case where
-* pending_sz is zero then host has nothing pending
-* and does not need to be signaled.
-*/
-   if (orig_write_sz > pending_sz)
-   return;
+   /*
+* If there was space before we began iteration,
+* then host was not blocked. Also handles case where
+* pending_sz is zero then host has nothing pending and does
+* not need to be  signaled.
+*/
+   if (curr_write_sz - delta > pending_sz)
+   return;
 
-   /* If pending write will not fit, don't give false hope. */
-   if (hv_get_bytes_to_write(rbi) < pending_sz)
-  

Re: [PATCH char-misc 1/1] Drivers: hv: vmbus: Fix ring buffer signaling

[Stephen Hemminger]

>   if (rbi->ring_buffer->feature_bits.feat_pending_send_sz) {
>   u32 pending_sz = READ_ONCE(rbi->ring_buffer->pending_send_sz);
>  
>   /*
> +  * Ensure the read of write_index in hv_get_bytes_to_write()
> +  * happens after the read of pending_send_sz.
> +  */
> + virt_rmb();
> + curr_write_sz = hv_get_bytes_to_write(rbi);
> +
> + /*
>* If there was space before we began iteration,
>* then host was not blocked. Also handles case where
>* pending_sz is zero then host has nothing pending
>* and does not need to be signaled.
>*/
> - if (orig_write_sz > pending_sz)
> + if (curr_write_sz - delta > pending_sz)
>   return;
>  
>   /* If pending write will not fit, don't give false hope. */
> - if (hv_get_bytes_to_write(rbi) < pending_sz)
> + if (curr_write_sz <= pending_sz)
>   return;
> +
> + vmbus_setevent(channel);
>   }
>  
> - vmbus_setevent(channel);
>  }

I think this won't work on older versions of Windows where feat_pending_sz
is never set.  On those versions vmbus_setevent needs to always be called.
___
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

RE: [PATCH char-misc 1/1] Drivers: hv: vmbus: Fix ring buffer signaling

[Michael Kelley (EOSG)]

> -Original Message-
> From: KY Srinivasan
> Sent: Sunday, February 11, 2018 5:14 PM

--- snip ---

> > if (rbi->ring_buffer->feature_bits.feat_pending_send_sz) {
> > u32 pending_sz = READ_ONCE(rbi->ring_buffer-
> > >pending_send_sz);
> >
> > /*
> > +* Ensure the read of write_index in
> > hv_get_bytes_to_write()
> > +* happens after the read of pending_send_sz.
> > +*/
> > +   virt_rmb();
> We can avoid the read barrier by making the initialization of curr_write_sz 
> conditional
> on pending_send_sz being non-zero. Indeed you can make all the signaling code 
> conditional
> on pending_send_sz being non-zero.
> 
> 

I agree that we can immediately test pending_send_sz for zero, and exit
if zero.  A  zero value will be by far the most common path, and would
save executing the read barrier and hv_get_bytes_to_write().  Can also
move the calculation of "delta" to after the test for zero.

But I believe the read barrier is still needed on the path where
pending_send_sz is non-zero.  Just the testing the value of pending_send_sz
doesn't guarantee that the write_index wouldn't be speculatively read,
and potentially out-of-order.   And we have to consider the out-of-order
behaviors on ARM64 as well.

> 
> > +   curr_write_sz = hv_get_bytes_to_write(rbi);
> > +
> > +   /*
> >  * If there was space before we began iteration,
> >  * then host was not blocked. Also handles case where
> >  * pending_sz is zero then host has nothing pending
> >  * and does not need to be signaled.
> >  */
> > -   if (orig_write_sz > pending_sz)
> > +   if (curr_write_sz - delta > pending_sz)
> > return;
> >
> > /* If pending write will not fit, don't give false hope. */
> > -   if (hv_get_bytes_to_write(rbi) < pending_sz)
> > +   if (curr_write_sz <= pending_sz)
> > return;
> > +
> > +   vmbus_setevent(channel);
> > }
> >
> > -   vmbus_setevent(channel);
> >  }
> >  EXPORT_SYMBOL_GPL(hv_pkt_iter_close);
> > --
> > 1.8.3.1

___
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

RE: [PATCH char-misc 1/1] Drivers: hv: vmbus: Fix ring buffer signaling

[KY Srinivasan]

> -Original Message-
> From: Michael Kelley [mailto:mhkel...@outlook.com]
> Sent: Saturday, February 10, 2018 12:49 PM
> To: gre...@linuxfoundation.org; linux-ker...@vger.kernel.org;
> de...@linuxdriverproject.org; o...@aepfle.de; a...@canonical.com;
> vkuzn...@redhat.com; jasow...@redhat.com;
> leann.ogasaw...@canonical.com; marcelo.ce...@canonical.com; Stephen
> Hemminger <sthem...@microsoft.com>; KY Srinivasan
> <k...@microsoft.com>
> Cc: Michael Kelley (EOSG) <michael.h.kel...@microsoft.com>
> Subject: [PATCH char-misc 1/1] Drivers: hv: vmbus: Fix ring buffer signaling
> 
> Fix bugs in signaling the Hyper-V host when freeing space in the
> host->guest ring buffer:
> 
> 1. The interrupt_mask must not be used to determine whether to signal
>on the host->guest ring buffer
> 2. The ring buffer write_index must be read (via hv_get_bytes_to_write)
>*after* pending_send_sz is read in order to avoid a race condition
> 3. Comparisons with pending_send_sz must treat the "equals" case as
>not-enough-space
> 4. Don't signal if the pending_send_sz feature is not present. Older
>versions of Hyper-V that don't implement this feature will poll.
> 
> Fixes: 03bad714a161 ("vmbus: more host signalling avoidance")
> Signed-off-by: Michael Kelley <mhkel...@outlook.com>
> ---
>  drivers/hv/ring_buffer.c | 24 
>  1 file changed, 16 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
> index 50e0714..b64be18 100644
> --- a/drivers/hv/ring_buffer.c
> +++ b/drivers/hv/ring_buffer.c
> @@ -423,7 +423,11 @@ struct vmpacket_descriptor *
>  void hv_pkt_iter_close(struct vmbus_channel *channel)
>  {
>   struct hv_ring_buffer_info *rbi = >inbound;
> - u32 orig_write_sz = hv_get_bytes_to_write(rbi);
> + u32 curr_write_sz;
> + u32 delta = rbi->ring_buffer->read_index < rbi->priv_read_index ?
> + (rbi->priv_read_index - rbi->ring_buffer-
> >read_index) :
> + (rbi->ring_datasize - rbi->ring_buffer->read_index +
> + rbi->priv_read_index);
> 
>   /*
>* Make sure all reads are done before we update the read index
> since
> @@ -446,27 +450,31 @@ void hv_pkt_iter_close(struct vmbus_channel
> *channel)
>*/
>   virt_mb();
> 
> - /* If host has disabled notifications then skip */
> - if (rbi->ring_buffer->interrupt_mask)
> - return;
> -
>   if (rbi->ring_buffer->feature_bits.feat_pending_send_sz) {
>   u32 pending_sz = READ_ONCE(rbi->ring_buffer-
> >pending_send_sz);
> 
>   /*
> +  * Ensure the read of write_index in
> hv_get_bytes_to_write()
> +  * happens after the read of pending_send_sz.
> +  */
> + virt_rmb();
We can avoid the read barrier by making the initialization of curr_write_sz 
conditional
on pending_send_sz being non-zero. Indeed you can make all the signaling code 
conditional
on pending_send_sz being non-zero.



> + curr_write_sz = hv_get_bytes_to_write(rbi);
> +
> + /*
>* If there was space before we began iteration,
>* then host was not blocked. Also handles case where
>* pending_sz is zero then host has nothing pending
>* and does not need to be signaled.
>*/
> - if (orig_write_sz > pending_sz)
> + if (curr_write_sz - delta > pending_sz)
>   return;
> 
>   /* If pending write will not fit, don't give false hope. */
> - if (hv_get_bytes_to_write(rbi) < pending_sz)
> + if (curr_write_sz <= pending_sz)
>   return;
> +
> + vmbus_setevent(channel);
>   }
> 
> - vmbus_setevent(channel);
>  }
>  EXPORT_SYMBOL_GPL(hv_pkt_iter_close);
> --
> 1.8.3.1

___
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Re: [PATCH char-misc 1/1] Drivers: hv: vmbus: Fix ring buffer signaling

[Stephen Hemminger]

On Sat, 10 Feb 2018 20:48:49 +
Michael Kelley  wrote:

> + u32 delta = rbi->ring_buffer->read_index < rbi->priv_read_index ?
> + (rbi->priv_read_index - rbi->ring_buffer->read_index) :
> + (rbi->ring_datasize - rbi->ring_buffer->read_index +
> + rbi->priv_read_index);
>  

Would it would make sense for this to be a wrapper function like other ring 
operations?
___
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

[PATCH char-misc 1/1] Drivers: hv: vmbus: Fix ring buffer signaling

[Michael Kelley]

Fix bugs in signaling the Hyper-V host when freeing space in the
host->guest ring buffer:

1. The interrupt_mask must not be used to determine whether to signal
   on the host->guest ring buffer
2. The ring buffer write_index must be read (via hv_get_bytes_to_write)
   *after* pending_send_sz is read in order to avoid a race condition
3. Comparisons with pending_send_sz must treat the "equals" case as
   not-enough-space
4. Don't signal if the pending_send_sz feature is not present. Older
   versions of Hyper-V that don't implement this feature will poll.

Fixes: 03bad714a161 ("vmbus: more host signalling avoidance")
Signed-off-by: Michael Kelley 
---
 drivers/hv/ring_buffer.c | 24 
 1 file changed, 16 insertions(+), 8 deletions(-)

diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
index 50e0714..b64be18 100644
--- a/drivers/hv/ring_buffer.c
+++ b/drivers/hv/ring_buffer.c
@@ -423,7 +423,11 @@ struct vmpacket_descriptor *
 void hv_pkt_iter_close(struct vmbus_channel *channel)
 {
struct hv_ring_buffer_info *rbi = >inbound;
-   u32 orig_write_sz = hv_get_bytes_to_write(rbi);
+   u32 curr_write_sz;
+   u32 delta = rbi->ring_buffer->read_index < rbi->priv_read_index ?
+   (rbi->priv_read_index - rbi->ring_buffer->read_index) :
+   (rbi->ring_datasize - rbi->ring_buffer->read_index +
+   rbi->priv_read_index);
 
/*
 * Make sure all reads are done before we update the read index since
@@ -446,27 +450,31 @@ void hv_pkt_iter_close(struct vmbus_channel *channel)
 */
virt_mb();
 
-   /* If host has disabled notifications then skip */
-   if (rbi->ring_buffer->interrupt_mask)
-   return;
-
if (rbi->ring_buffer->feature_bits.feat_pending_send_sz) {
u32 pending_sz = READ_ONCE(rbi->ring_buffer->pending_send_sz);
 
/*
+* Ensure the read of write_index in hv_get_bytes_to_write()
+* happens after the read of pending_send_sz.
+*/
+   virt_rmb();
+   curr_write_sz = hv_get_bytes_to_write(rbi);
+
+   /*
 * If there was space before we began iteration,
 * then host was not blocked. Also handles case where
 * pending_sz is zero then host has nothing pending
 * and does not need to be signaled.
 */
-   if (orig_write_sz > pending_sz)
+   if (curr_write_sz - delta > pending_sz)
return;
 
/* If pending write will not fit, don't give false hope. */
-   if (hv_get_bytes_to_write(rbi) < pending_sz)
+   if (curr_write_sz <= pending_sz)
return;
+
+   vmbus_setevent(channel);
}
 
-   vmbus_setevent(channel);
 }
 EXPORT_SYMBOL_GPL(hv_pkt_iter_close);
-- 
1.8.3.1

___
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel