Commit 46b6599c authored by Bernard Metzler's avatar Bernard Metzler Committed by Liu Jian
Browse files

RDMA/siw: Remove direct link to net_device 

mainline inclusion
from mainline-v6.13-rc6
commit 16b87037b48889d21854c8e97aec8a1baf2642b3
category: bugfix
bugzilla: https://gitee.com/src-openeuler/kernel/issues/IBI67U
CVE: CVE-2024-57857

Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=16b87037b48889d21854c8e97aec8a1baf2642b3



-------------------------------------------------

Do not manage a per device direct link to net_device. Rely
on associated ib_devices net_device management, not doubling
the effort locally. A badly managed local link to net_device
was causing a 'KASAN: slab-use-after-free' exception during
siw_query_port() call.

Fixes: bdcf26bf ("rdma/siw: network and RDMA core interface")
Reported-by: default avatar <syzbot+4b87489410b4efd181bf@syzkaller.appspotmail.com>
Closes: https://syzkaller.appspot.com/bug?extid=4b87489410b4efd181bf


Signed-off-by: default avatarBernard Metzler <bmt@zurich.ibm.com>
Link: https://patch.msgid.link/20241212151848.564872-1-bmt@zurich.ibm.com


Reviewed-by: default avatarZhu Yanjun <yanjun.zhu@linux.dev>
Signed-off-by: default avatarLeon Romanovsky <leon@kernel.org>

Conflicts:
	drivers/infiniband/sw/siw/siw_main.c
	drivers/infiniband/sw/siw/siw_verbs.c
	drivers/infiniband/sw/siw/siw_cm.c
[Did not backport af48f95492dc, 0b988c1bee28, b5c91543204c.]
Signed-off-by: default avatarLiu Jian <liujian56@huawei.com>
parent 4a28050d

drivers/infiniband/sw/siw/siw.h

+3 −4
Original line number Diff line number Diff line
@@ -46,6 +46,9 @@ 
 */

#define SIW_IRQ_MAXBURST_SQ_ACTIVE 4



/* There is always only a port 1 per siw device */

#define SIW_PORT 1



struct siw_dev_cap {

	int max_qp;

	int max_qp_wr;
@@ -69,16 +72,12 @@ struct siw_pd { 


struct siw_device {

	struct ib_device base_dev;

	struct net_device *netdev;

	struct siw_dev_cap attrs;



	u32 vendor_part_id;

	int numa_node;

	char raw_gid[ETH_ALEN];



	/* physical port state (only one port per device) */

	enum ib_port_state state;



	spinlock_t lock;



	struct xarray qp_xa;

drivers/infiniband/sw/siw/siw_cm.c

+21 −6
Original line number Diff line number Diff line
@@ -1767,6 +1767,7 @@ int siw_create_listen(struct iw_cm_id *id, int backlog) 
{

	struct socket *s;

	struct siw_cep *cep = NULL;

	struct net_device *ndev = NULL;

	struct siw_device *sdev = to_siw_dev(id->device);

	int addr_family = id->local_addr.ss_family;

	int rv = 0;
@@ -1787,18 +1788,30 @@ int siw_create_listen(struct iw_cm_id *id, int backlog) 
		struct sockaddr_in *laddr = &to_sockaddr_in(id->local_addr);



		/* For wildcard addr, limit binding to current device only */

		if (ipv4_is_zeronet(laddr->sin_addr.s_addr))

			s->sk->sk_bound_dev_if = sdev->netdev->ifindex;



		if (ipv4_is_zeronet(laddr->sin_addr.s_addr)) {

			ndev = ib_device_get_netdev(id->device, SIW_PORT);

			if (ndev) {

				s->sk->sk_bound_dev_if = ndev->ifindex;

			} else {

				rv = -ENODEV;

				goto error;

			}

		}

		rv = s->ops->bind(s, (struct sockaddr *)laddr,

				  sizeof(struct sockaddr_in));

	} else {

		struct sockaddr_in6 *laddr = &to_sockaddr_in6(id->local_addr);



		/* For wildcard addr, limit binding to current device only */

		if (ipv6_addr_any(&laddr->sin6_addr))

			s->sk->sk_bound_dev_if = sdev->netdev->ifindex;



		if (ipv6_addr_any(&laddr->sin6_addr)) {

			ndev = ib_device_get_netdev(id->device, SIW_PORT);

			if (ndev) {

				s->sk->sk_bound_dev_if = ndev->ifindex;

			} else {

				rv = -ENODEV;

				goto error;

			}

		}

		rv = s->ops->bind(s, (struct sockaddr *)laddr,

				  sizeof(struct sockaddr_in6));

	}
@@ -1859,6 +1872,7 @@ int siw_create_listen(struct iw_cm_id *id, int backlog) 
	}

	list_add_tail(&cep->listenq, (struct list_head *)id->provider_data);

	cep->state = SIW_EPSTATE_LISTENING;

	dev_put(ndev);



	siw_dbg(id->device, "Listen at laddr %pISp\n", &id->local_addr);
@@ -1882,6 +1896,7 @@ int siw_create_listen(struct iw_cm_id *id, int backlog) 
		siw_cep_put(cep);

	}

	sock_release(s);

	dev_put(ndev);



	return rv;

}

drivers/infiniband/sw/siw/siw_main.c

+1 −13
Original line number Diff line number Diff line
@@ -312,7 +312,6 @@ static struct siw_device *siw_device_create(struct net_device *netdev) 
		return NULL;



	base_dev = &sdev->base_dev;

	sdev->netdev = netdev;



	if (netdev->addr_len) {

		memcpy(sdev->raw_gid, netdev->dev_addr,
@@ -462,7 +461,6 @@ static int siw_netdev_event(struct notifier_block *nb, unsigned long event, 


	switch (event) {

	case NETDEV_UP:

		sdev->state = IB_PORT_ACTIVE;

		siw_port_event(sdev, 1, IB_EVENT_PORT_ACTIVE);

		break;
@@ -471,7 +469,6 @@ static int siw_netdev_event(struct notifier_block *nb, unsigned long event, 
		break;



	case NETDEV_DOWN:

		sdev->state = IB_PORT_DOWN;

		siw_port_event(sdev, 1, IB_EVENT_PORT_ERR);

		break;
@@ -492,12 +489,8 @@ static int siw_netdev_event(struct notifier_block *nb, unsigned long event, 
		siw_port_event(sdev, 1, IB_EVENT_LID_CHANGE);

		break;

	/*

	 * Todo: Below netdev events are currently not handled.

	 * All other events are not handled

	 */

	case NETDEV_CHANGEMTU:

	case NETDEV_CHANGE:

		break;



	default:

		break;

	}
@@ -528,11 +521,6 @@ static int siw_newlink(const char *basedev_name, struct net_device *netdev) 
	if (sdev) {

		dev_dbg(&netdev->dev, "siw: new device\n");



		if (netif_running(netdev) && netif_carrier_ok(netdev))

			sdev->state = IB_PORT_ACTIVE;

		else

			sdev->state = IB_PORT_DOWN;



		rv = siw_device_register(sdev, basedev_name);

		if (rv)

			ib_dealloc_device(&sdev->base_dev);

drivers/infiniband/sw/siw/siw_verbs.c

+24 −11
Original line number Diff line number Diff line
@@ -165,21 +165,29 @@ int siw_query_device(struct ib_device *base_dev, struct ib_device_attr *attr, 
int siw_query_port(struct ib_device *base_dev, u8 port,

		   struct ib_port_attr *attr)

{

	struct siw_device *sdev = to_siw_dev(base_dev);

	struct net_device *ndev;

	int rv;



	memset(attr, 0, sizeof(*attr));



	rv = ib_get_eth_speed(base_dev, port, &attr->active_speed,

			 &attr->active_width);

	if (rv)

		return rv;



	ndev = ib_device_get_netdev(base_dev, SIW_PORT);

	if (!ndev)

		return -ENODEV;



	attr->gid_tbl_len = 1;

	attr->max_msg_sz = -1;

	attr->max_mtu = ib_mtu_int_to_enum(sdev->netdev->mtu);

	attr->active_mtu = ib_mtu_int_to_enum(sdev->netdev->mtu);

	attr->phys_state = sdev->state == IB_PORT_ACTIVE ?

	attr->max_mtu = ib_mtu_int_to_enum(ndev->max_mtu);

	attr->active_mtu = ib_mtu_int_to_enum(READ_ONCE(ndev->mtu));

	attr->phys_state = (netif_running(ndev) && netif_carrier_ok(ndev)) ?

		IB_PORT_PHYS_STATE_LINK_UP : IB_PORT_PHYS_STATE_DISABLED;

	attr->state = attr->phys_state == IB_PORT_PHYS_STATE_LINK_UP ?

		IB_PORT_ACTIVE : IB_PORT_DOWN;

	attr->port_cap_flags = IB_PORT_CM_SUP | IB_PORT_DEVICE_MGMT_SUP;

	attr->state = sdev->state;

	/*

	 * All zero

	 *
@@ -193,6 +201,7 @@ int siw_query_port(struct ib_device *base_dev, u8 port, 
	 * attr->subnet_timeout = 0;

	 * attr->init_type_repy = 0;

	 */

	dev_put(ndev);

	return rv;

}
@@ -509,20 +518,23 @@ int siw_query_qp(struct ib_qp *base_qp, struct ib_qp_attr *qp_attr, 
		 int qp_attr_mask, struct ib_qp_init_attr *qp_init_attr)

{

	struct siw_qp *qp;

	struct siw_device *sdev;

	struct net_device *ndev;



	if (base_qp && qp_attr && qp_init_attr) {

	if (base_qp && qp_attr && qp_init_attr)

		qp = to_siw_qp(base_qp);

		sdev = to_siw_dev(base_qp->device);

	} else {

	else

		return -EINVAL;

	}



	ndev = ib_device_get_netdev(base_qp->device, SIW_PORT);

	if (!ndev)

		return -ENODEV;



	qp_attr->cap.max_inline_data = SIW_MAX_INLINE;

	qp_attr->cap.max_send_wr = qp->attrs.sq_size;

	qp_attr->cap.max_send_sge = qp->attrs.sq_max_sges;

	qp_attr->cap.max_recv_wr = qp->attrs.rq_size;

	qp_attr->cap.max_recv_sge = qp->attrs.rq_max_sges;

	qp_attr->path_mtu = ib_mtu_int_to_enum(sdev->netdev->mtu);

	qp_attr->path_mtu = ib_mtu_int_to_enum(READ_ONCE(ndev->mtu));

	qp_attr->max_rd_atomic = qp->attrs.irq_size;

	qp_attr->max_dest_rd_atomic = qp->attrs.orq_size;
@@ -537,6 +549,7 @@ int siw_query_qp(struct ib_qp *base_qp, struct ib_qp_attr *qp_attr, 


	qp_init_attr->cap = qp_attr->cap;



	dev_put(ndev);

	return 0;

}