From: Michael Kelley <mikelley@xxxxxxxxxxxxx>
Add comments describing intricacies of Hyper-V ring buffer
signaling code. This information is not in Hyper-V public
documents, so include here to capture the knowledge for
future coders.
There are no code changes in this commit.
Signed-off-by: Michael Kelley <mikelley@xxxxxxxxxxxxx>
---
drivers/hv/ring_buffer.c | 65 ++++++++++++++++++++++++++++++++++++++++--------
include/linux/hyperv.h | 31 +++++++++++++++++------
2 files changed, 77 insertions(+), 19 deletions(-)
diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
index 8699bb9..ce6ea6d 100644
--- a/drivers/hv/ring_buffer.c
+++ b/drivers/hv/ring_buffer.c
@@ -429,7 +429,24 @@ static u32 hv_pkt_iter_bytes_read(const struct hv_ring_buffer_info *rbi,
}
/*
- * Update host ring buffer after iterating over packets.
+ * Update host ring buffer after iterating over packets. If the host has
+ * stopped queuing new entries because it found the ring buffer full, and
+ * sufficient space is being freed up, signal the host. But be careful to
+ * only signal the host when necessary, both for performance reasons and
+ * because Hyper-V protects itself by throttling guests that signal
+ * inappropriately.
+ *
+ * Determining when to signal is tricky. There are three key data inputs
+ * that must be handled in this order to avoid race conditions:
+ *
+ * 1. Update the read_index
+ * 2. Read the pending_send_sz
+ * 3. Read the current write_index
+ *
+ * The interrupt_mask is not used to determine when to signal. The
+ * interrupt_mask is used only on the guest->host ring buffer when
+ * sending requests to the host. The host does not use it on the host->
+ * guest ring buffer to indicate whether it should be signaled.
*/
void hv_pkt_iter_close(struct vmbus_channel *channel)
{
@@ -445,22 +462,30 @@ void hv_pkt_iter_close(struct vmbus_channel *channel)
start_read_index = rbi->ring_buffer->read_index;
rbi->ring_buffer->read_index = rbi->priv_read_index;
+ /*
+ * Older versions of Hyper-V (before WS2102 and Win8) do not
+ * implement pending_send_sz and simply poll if the host->guest
+ * ring buffer is full. No signaling is needed or expected.
+ */
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
+ * If reading pending_send_sz were to be reordered and happen
+ * before we commit the new read_index, a race could occur. If the
+ * host were to set the pending_send_sz after we have sampled
+ * pending_send_sz, and the ring buffer blocks before we commit the
* read index, we could miss sending the interrupt. Issue a full
* memory barrier to address this.
*/
virt_mb();
+ /*
+ * If the pending_send_sz is zero, then the ring buffer is not
+ * blocked and there is no need to signal. This is far by the
+ * most common case, so exit quickly for best performance.
+ */
pending_sz = READ_ONCE(rbi->ring_buffer->pending_send_sz);
if (!pending_sz)
return;
@@ -474,14 +499,32 @@ void hv_pkt_iter_close(struct vmbus_channel *channel)
bytes_read = hv_pkt_iter_bytes_read(rbi, start_read_index);
/*
- * If there was space before we began iteration,
- * then host was not blocked.
+ * We want to signal the host only if we're transitioning
+ * from a "not enough free space" state to a "enough free
+ * space" state. For example, it's possible that this function
+ * could run and free up enough space to signal the host, and then
+ * run again and free up additional space before the host has a
+ * chance to clear the pending_send_sz. The 2nd invocation would
+ * be a null transition from "enough free space" to "enough free
+ * space", which doesn't warrant a signal.
+ *
+ * Exactly filling the ring buffer is treated as "not enough
+ * space". The ring buffer always must have at least one byte
+ * empty so the empty and full conditions are distinguishable.
+ * hv_get_bytes_to_write() doesn't fully tell the truth in
+ * this regard.
+ *
+ * So first check if we were in the "enough free space" state
+ * before we began the iteration. If so, the host was not
+ * blocked, and there's no need to signal.
*/
-
if (curr_write_sz - bytes_read > pending_sz)
return;
- /* If pending write will not fit, don't give false hope. */
+ /*
+ * Similarly, if the new state is "not enough space", then
+ * there's no need to signal.
+ */
if (curr_write_sz <= pending_sz)
return;
diff --git a/include/linux/hyperv.h b/include/linux/hyperv.h
index 2048f3c..a0dfe78 100644
--- a/include/linux/hyperv.h
+++ b/include/linux/hyperv.h
@@ -89,18 +89,33 @@ struct hv_ring_buffer {
u32 interrupt_mask;
/*
- * Win8 uses some of the reserved bits to implement
- * interrupt driven flow management. On the send side
- * we can request that the receiver interrupt the sender
- * when the ring transitions from being full to being able
- * to handle a message of size "pending_send_sz".
+ * WS2012/Win8 and later versions of Hyper-V implement interrupt
+ * driven flow management. The feature bit feat_pending_send_sz
+ * is set by the host on the host->guest ring buffer, and by the
+ * guest on the guest->host ring buffer.
*
- * Add necessary state for this enhancement.
+ * The meaning of the feature bit is a bit complex in that it has
+ * semantics that apply to both ring buffers. If the guest sets
+ * the feature bit in the guest->host ring buffer, the guest is
+ * telling the host that:
+ * 1) It will set the pending_send_sz field in the guest->host ring
+ * buffer when it is waiting for space to become available, and
+ * 2) It will read the pending_send_sz field in the host->guest
+ * ring buffer and interrupt the host when it frees enough space
+ *
+ * Similarly, if the host sets the feature bit in the host->guest
+ * ring buffer, the host is telling the guest that:
+ * 1) It will set the pending_send_sz field in the host->guest ring
+ * buffer when it is waiting for space to become available, and
+ * 2) It will read the pending_send_sz field in the guest->host
+ * ring buffer and interrupt the guest when it frees enough space
+ *
+ * If either the guest or host does not set the feature bit that it
+ * owns, that guest or host must do polling if it encounters a full
+ * ring buffer, and not signal the other end with an interrupt.
*/
u32 pending_send_sz;
-
u32 reserved1[12];
-
union {
struct {
u32 feat_pending_send_sz:1;
--
2.7.4
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
