Drivers: hv: ring_buffer: eliminate hv_ringbuffer_peek()
Currently, there is only one user for hv_ringbuffer_read()/
hv_ringbuffer_peak() functions and the usage of these functions is:
- insecure as we drop ring_lock between them, someone else (in theory
only) can acquire it in between;
- non-optimal as we do a number of things (acquire/release the above
mentioned lock, calculate available space on the ring, ...) twice and
this path is performance-critical.
Remove hv_ringbuffer_peek() moving the logic from __vmbus_recvpacket() to
hv_ringbuffer_read().
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
diff --git a/drivers/hv/channel.c b/drivers/hv/channel.c
index dd6de7f..1161d68 100644
--- a/drivers/hv/channel.c
+++ b/drivers/hv/channel.c
@@ -927,37 +927,11 @@
u32 bufferlen, u32 *buffer_actual_len, u64 *requestid,
bool raw)
{
- struct vmpacket_descriptor desc;
- u32 packetlen;
- u32 userlen;
int ret;
bool signal = false;
- *buffer_actual_len = 0;
- *requestid = 0;
-
-
- ret = hv_ringbuffer_peek(&channel->inbound, &desc,
- sizeof(struct vmpacket_descriptor));
- if (ret != 0)
- return 0;
-
- packetlen = desc.len8 << 3;
- if (!raw)
- userlen = packetlen - (desc.offset8 << 3);
- else
- userlen = packetlen;
-
- *buffer_actual_len = userlen;
-
- if (userlen > bufferlen)
- return -ENOBUFS;
-
- *requestid = desc.trans_id;
-
- /* Copy over the packet to the user buffer */
- ret = hv_ringbuffer_read(&channel->inbound, buffer, userlen,
- raw ? 0 : desc.offset8 << 3, &signal);
+ ret = hv_ringbuffer_read(&channel->inbound, buffer, bufferlen,
+ buffer_actual_len, requestid, &signal, raw);
if (signal)
vmbus_setevent(channel);