Drivers: hv: vmbus: finally fix hv_need_to_signal_on_read()

Commit a389fcfd2cb5 ("Drivers: hv: vmbus: Fix signaling logic in
hv_need_to_signal_on_read()")
added the proper mb(), but removed the test "prev_write_sz < pending_sz"
when making the signal decision.

As a result, the guest can signal the host unnecessarily,
and then the host can throttle the guest because the host
thinks the guest is buggy or malicious; finally the user
running stress test can perceive intermittent freeze of
the guest.

This patch brings back the test, and properly handles the
in-place consumption APIs used by NetVSC (see get_next_pkt_raw(),
put_pkt_raw() and commit_rd_index()).

Fixes: a389fcfd2cb5 ("Drivers: hv: vmbus: Fix signaling logic in
hv_need_to_signal_on_read()")

Signed-off-by: Dexuan Cui <decui@microsoft.com>
Reported-by: Rolf Neugebauer <rolf.neugebauer@docker.com>
Tested-by: Rolf Neugebauer <rolf.neugebauer@docker.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

3 files changed, 37 insertions, 2 deletions

diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
index cd49cb17eb7f..308dbda700eb 100644
--- a/drivers/hv/ring_buffer.c
+++ b/drivers/hv/ring_buffer.c
@@ -383,6 +383,7 @@ int hv_ringbuffer_read(struct vmbus_channel *channel,
                 return ret;
         }
 
+        init_cached_read_index(channel);
         next_read_location = hv_get_next_read_location(inring_info);
         next_read_location = hv_copyfrom_ringbuffer(inring_info, &desc,
                                                     sizeof(desc),
diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c
index 5a1cc089acb7..86e5749226ef 100644
--- a/drivers/net/hyperv/netvsc.c
+++ b/drivers/net/hyperv/netvsc.c
@@ -1295,6 +1295,9 @@ void netvsc_channel_cb(void *context)
         ndev = hv_get_drvdata(device);
         buffer = get_per_channel_state(channel);
 
+        /* commit_rd_index() -> hv_signal_on_read() needs this. */
+        init_cached_read_index(channel);
+
         do {
                 desc = get_next_pkt_raw(channel);
                 if (desc != NULL) {
@@ -1347,6 +1350,9 @@ void netvsc_channel_cb(void *context)
 
                         bufferlen = bytes_recvd;
                 }
+
+                init_cached_read_index(channel);
+
         } while (1);
 
         if (bufferlen > NETVSC_PACKET_SIZE)
diff --git a/include/linux/hyperv.h b/include/linux/hyperv.h
index 42fe43fb0c80..183efde54269 100644
--- a/include/linux/hyperv.h
+++ b/include/linux/hyperv.h
@@ -128,6 +128,7 @@ struct hv_ring_buffer_info {
         u32 ring_data_startoffset;
         u32 priv_write_index;
         u32 priv_read_index;
+        u32 cached_read_index;
 };
 
 /*
@@ -180,6 +181,19 @@ static inline u32 hv_get_bytes_to_write(struct hv_ring_buffer_info *rbi)
         return write;
 }
 
+static inline u32 hv_get_cached_bytes_to_write(
+        const struct hv_ring_buffer_info *rbi)
+{
+        u32 read_loc, write_loc, dsize, write;
+
+        dsize = rbi->ring_datasize;
+        read_loc = rbi->cached_read_index;
+        write_loc = rbi->ring_buffer->write_index;
+
+        write = write_loc >= read_loc ? dsize - (write_loc - read_loc) :
+                read_loc - write_loc;
+        return write;
+}
 /*
  * VMBUS version is 32 bit entity broken up into
  * two 16 bit quantities: major_number. minor_number.
@@ -1488,7 +1502,7 @@ hv_get_ring_buffer(struct hv_ring_buffer_info *ring_info)
 
 static inline  void hv_signal_on_read(struct vmbus_channel *channel)
 {
-        u32 cur_write_sz;
+        u32 cur_write_sz, cached_write_sz;
         u32 pending_sz;
         struct hv_ring_buffer_info *rbi = &channel->inbound;
 
@@ -1512,12 +1526,24 @@ static inline  void hv_signal_on_read(struct vmbus_channel *channel)
 
         cur_write_sz = hv_get_bytes_to_write(rbi);
 
-        if (cur_write_sz >= pending_sz)
+        if (cur_write_sz < pending_sz)
+                return;
+
+        cached_write_sz = hv_get_cached_bytes_to_write(rbi);
+        if (cached_write_sz < pending_sz)
                 vmbus_setevent(channel);
 
         return;
 }
 
+static inline void
+init_cached_read_index(struct vmbus_channel *channel)
+{
+        struct hv_ring_buffer_info *rbi = &channel->inbound;
+
+        rbi->cached_read_index = rbi->ring_buffer->read_index;
+}
+
 /*
  * An API to support in-place processing of incoming VMBUS packets.
  */
@@ -1569,6 +1595,8 @@ static inline void put_pkt_raw(struct vmbus_channel *channel,
  * This call commits the read index and potentially signals the host.
  * Here is the pattern for using the "in-place" consumption APIs:
  *
+ * init_cached_read_index();
+ *
  * while (get_next_pkt_raw() {
  *      process the packet "in-place";
  *      put_pkt_raw();