summaryrefslogtreecommitdiffstats
path: root/net
diff options
context:
space:
mode:
authorDavid S. Miller <davem@davemloft.net>2018-11-13 08:43:05 -0800
committerDavid S. Miller <davem@davemloft.net>2018-11-13 08:43:05 -0800
commit11123ab9d90c5f426beadfe01507616c3654b06d (patch)
treed22e05eb886a670a887ee3464e17bdf307b5c178 /net
parent77e461d14ed141253573eeeb4d34eccc51e38328 (diff)
parentd788905f68fd4714c82936f6f7f1d3644d7ae7ef (diff)
downloadlinux-11123ab9d90c5f426beadfe01507616c3654b06d.tar.bz2
Merge tag 'linux-can-fixes-for-4.20-20181109' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can
Marc Kleine-Budde says: ==================== pull-request: can 2018-11-09 this is a pull request of 20 patches for net/master. First we have a patch by Oliver Hartkopp which changes the raw socket's raw_sendmsg() to return an error value if the user tries to send a CANFD frame to a CAN-2.0 device. The next two patches are by Jimmy Assarsson and fix potential problems in the kvaser_usb driver. YueHaibing's patches for the ucan driver fix a compile time warning and remove a duplicate include. Eugeniu Rosca patch adds more binding documentation to the rcar_can driver bindings. The next two patches are by Fabrizio Castro for the rcar_can driver and fixes a problem in the driver's probe function and document the r8a774a1 binding. Lukas Wunner's patch fixes a recpetion problem in hi311x driver by switching from edge to level triggered interruts. The next three patches all target the flexcan driver. Pankaj Bansal's patch unconditionally unlocks the last mailbox used for RX. Alexander Stein provides a better workaround for a hardware limitation when sending RTR frames, by using the last mailbox for TX, resulting in fewer lost frames. The patch by me simplyfies the driver, by making a runtime value a compile time constant. The following 4 patches are by me and provide the groundwork for the next patches by Oleksij Rempel. To avoid code duplication common code in the common CAN driver infrastructure is factured out and error handling is cleaned up. The next 4 patches are by Oleksij Rempel and fix the problem in the flexcan driver that other processes see TX frames arrive out of order with ragards to a RX'ed frame (which are send by a different system on the CAN bus as the result of our TX frame). ==================== Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net')
-rw-r--r--net/can/raw.c15
1 files changed, 8 insertions, 7 deletions
diff --git a/net/can/raw.c b/net/can/raw.c
index 1051eee82581..3aab7664933f 100644
--- a/net/can/raw.c
+++ b/net/can/raw.c
@@ -745,18 +745,19 @@ static int raw_sendmsg(struct socket *sock, struct msghdr *msg, size_t size)
} else
ifindex = ro->ifindex;
- if (ro->fd_frames) {
+ dev = dev_get_by_index(sock_net(sk), ifindex);
+ if (!dev)
+ return -ENXIO;
+
+ err = -EINVAL;
+ if (ro->fd_frames && dev->mtu == CANFD_MTU) {
if (unlikely(size != CANFD_MTU && size != CAN_MTU))
- return -EINVAL;
+ goto put_dev;
} else {
if (unlikely(size != CAN_MTU))
- return -EINVAL;
+ goto put_dev;
}
- dev = dev_get_by_index(sock_net(sk), ifindex);
- if (!dev)
- return -ENXIO;
-
skb = sock_alloc_send_skb(sk, size + sizeof(struct can_skb_priv),
msg->msg_flags & MSG_DONTWAIT, &err);
if (!skb)