| Message ID | 637e945f-f28a-86d9-a242-1f4be85d9840@sberdevices.ru (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | virtio/vsock: use SO_RCVLOWAT to set POLLIN/POLLRDNORM | expand |
On Mon, Jul 18, 2022 at 08:15:42AM +0000, Arseniy Krasnov wrote: >Both bits indicate, that next data read call won't be blocked, >but when sk_rcvlowat is not 1,these bits will be set by poll >anyway,thus when user tries to dequeue data, it will wait until >sk_rcvlowat bytes of data will be available. > >Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru> >--- > net/vmw_vsock/af_vsock.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > >diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c >index f04abf662ec6..0225f3558e30 100644 >--- a/net/vmw_vsock/af_vsock.c >+++ b/net/vmw_vsock/af_vsock.c >@@ -1067,7 +1067,7 @@ static __poll_t vsock_poll(struct file *file, struct socket *sock, > !(sk->sk_shutdown & RCV_SHUTDOWN)) { > bool data_ready_now = false; > int ret = transport->notify_poll_in( >- vsk, 1, &data_ready_now); >+ vsk, sk->sk_rcvlowat, &data_ready_now); In tcp_poll() we have the following: int target = sock_rcvlowat(sk, 0, INT_MAX); Maybe we can do the same here. Thanks, Stefano > if (ret < 0) { > mask |= EPOLLERR; > } else { >-- >2.25.1
On 19.07.2022 15:44, Stefano Garzarella wrote: > On Mon, Jul 18, 2022 at 08:15:42AM +0000, Arseniy Krasnov wrote: >> Both bits indicate, that next data read call won't be blocked, >> but when sk_rcvlowat is not 1,these bits will be set by poll >> anyway,thus when user tries to dequeue data, it will wait until >> sk_rcvlowat bytes of data will be available. >> >> Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru> >> --- >> net/vmw_vsock/af_vsock.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c >> index f04abf662ec6..0225f3558e30 100644 >> --- a/net/vmw_vsock/af_vsock.c >> +++ b/net/vmw_vsock/af_vsock.c >> @@ -1067,7 +1067,7 @@ static __poll_t vsock_poll(struct file *file, struct socket *sock, >> !(sk->sk_shutdown & RCV_SHUTDOWN)) { >> bool data_ready_now = false; >> int ret = transport->notify_poll_in( >> - vsk, 1, &data_ready_now); >> + vsk, sk->sk_rcvlowat, &data_ready_now); > > In tcp_poll() we have the following: > int target = sock_rcvlowat(sk, 0, INT_MAX); > > Maybe we can do the same here. You are right, ack > > Thanks, > Stefano > >> if (ret < 0) { >> mask |= EPOLLERR; >> } else { >> -- >> 2.25.1 >
diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c index f04abf662ec6..0225f3558e30 100644 --- a/net/vmw_vsock/af_vsock.c +++ b/net/vmw_vsock/af_vsock.c @@ -1067,7 +1067,7 @@ static __poll_t vsock_poll(struct file *file, struct socket *sock, !(sk->sk_shutdown & RCV_SHUTDOWN)) { bool data_ready_now = false; int ret = transport->notify_poll_in( - vsk, 1, &data_ready_now); + vsk, sk->sk_rcvlowat, &data_ready_now); if (ret < 0) { mask |= EPOLLERR; } else {
Both bits indicate, that next data read call won't be blocked, but when sk_rcvlowat is not 1,these bits will be set by poll anyway,thus when user tries to dequeue data, it will wait until sk_rcvlowat bytes of data will be available. Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru> --- net/vmw_vsock/af_vsock.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)