* [PATCH] RDMA/iwcm: Allow AFONLY binding for IPv6 addresses.
@ 2021-02-19 14:34 Bernard Metzler
2021-03-10 19:49 ` Jason Gunthorpe
0 siblings, 1 reply; 4+ messages in thread
From: Bernard Metzler @ 2021-02-19 14:34 UTC (permalink / raw)
To: linux-rdma; +Cc: jgg, Bernard Metzler, Chuck Lever, Benjamin Coddington
Binding IPv6 address/port to AF_INET6 domain only is provided
via rdma_set_afonly(), but was not signalled to the provider.
Applications like NFS/RDMA bind the same port to both IPv4
and IPv6 addresses simultaneously and thus rely on it working
correctly.
Tested-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Bernard Metzler <bmt@zurich.ibm.com>
---
drivers/infiniband/core/cma.c | 1 +
drivers/infiniband/sw/siw/siw_cm.c | 19 +++++++++++++++++--
include/rdma/iw_cm.h | 1 +
3 files changed, 19 insertions(+), 2 deletions(-)
diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index c51b84b2d2f3..046d42c1f8f3 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -2466,6 +2466,7 @@ static int cma_iw_listen(struct rdma_id_private *id_priv, int backlog)
id->tos = id_priv->tos;
id->tos_set = id_priv->tos_set;
+ id->afonly = id_priv->afonly;
id_priv->cm_id.iw = id;
memcpy(&id_priv->cm_id.iw->local_addr, cma_src_addr(id_priv),
diff --git a/drivers/infiniband/sw/siw/siw_cm.c b/drivers/infiniband/sw/siw/siw_cm.c
index 1f9e15b71504..8bd0d0785433 100644
--- a/drivers/infiniband/sw/siw/siw_cm.c
+++ b/drivers/infiniband/sw/siw/siw_cm.c
@@ -1300,7 +1300,7 @@ static void siw_cm_llp_state_change(struct sock *sk)
}
static int kernel_bindconnect(struct socket *s, struct sockaddr *laddr,
- struct sockaddr *raddr)
+ struct sockaddr *raddr, bool afonly)
{
int rv, flags = 0;
size_t size = laddr->sa_family == AF_INET ?
@@ -1311,6 +1311,12 @@ static int kernel_bindconnect(struct socket *s, struct sockaddr *laddr,
*/
sock_set_reuseaddr(s->sk);
+ if (afonly) {
+ rv = ip6_sock_set_v6only(s->sk);
+ if (rv)
+ return rv;
+ }
+
rv = s->ops->bind(s, laddr, size);
if (rv < 0)
return rv;
@@ -1371,7 +1377,7 @@ int siw_connect(struct iw_cm_id *id, struct iw_cm_conn_param *params)
* mode. Might be reconsidered for async connection setup at
* TCP level.
*/
- rv = kernel_bindconnect(s, laddr, raddr);
+ rv = kernel_bindconnect(s, laddr, raddr, id->afonly);
if (rv != 0) {
siw_dbg_qp(qp, "kernel_bindconnect: error %d\n", rv);
goto error;
@@ -1786,6 +1792,15 @@ int siw_create_listen(struct iw_cm_id *id, int backlog)
} else {
struct sockaddr_in6 *laddr = &to_sockaddr_in6(id->local_addr);
+ if (id->afonly) {
+ rv = ip6_sock_set_v6only(s->sk);
+ if (rv) {
+ siw_dbg(id->device,
+ "ip6_sock_set_v6only erro: %d\n", rv);
+ goto error;
+ }
+ }
+
/* 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;
diff --git a/include/rdma/iw_cm.h b/include/rdma/iw_cm.h
index 91975400e1b3..03abd30e6c8c 100644
--- a/include/rdma/iw_cm.h
+++ b/include/rdma/iw_cm.h
@@ -70,6 +70,7 @@ struct iw_cm_id {
u8 tos;
bool tos_set:1;
bool mapped:1;
+ bool afonly:1;
};
struct iw_cm_conn_param {
--
2.17.2
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] RDMA/iwcm: Allow AFONLY binding for IPv6 addresses.
2021-02-19 14:34 [PATCH] RDMA/iwcm: Allow AFONLY binding for IPv6 addresses Bernard Metzler
@ 2021-03-10 19:49 ` Jason Gunthorpe
0 siblings, 0 replies; 4+ messages in thread
From: Jason Gunthorpe @ 2021-03-10 19:49 UTC (permalink / raw)
To: Bernard Metzler; +Cc: linux-rdma, Chuck Lever, Benjamin Coddington
On Fri, Feb 19, 2021 at 03:34:41PM +0100, Bernard Metzler wrote:
> Binding IPv6 address/port to AF_INET6 domain only is provided
> via rdma_set_afonly(), but was not signalled to the provider.
> Applications like NFS/RDMA bind the same port to both IPv4
> and IPv6 addresses simultaneously and thus rely on it working
> correctly.
>
> Tested-by: Chuck Lever <chuck.lever@oracle.com>
> Tested-by: Benjamin Coddington <bcodding@redhat.com>
> Signed-off-by: Bernard Metzler <bmt@zurich.ibm.com>
> ---
> drivers/infiniband/core/cma.c | 1 +
> drivers/infiniband/sw/siw/siw_cm.c | 19 +++++++++++++++++--
> include/rdma/iw_cm.h | 1 +
> 3 files changed, 19 insertions(+), 2 deletions(-)
Applied to for-next, thanks
Jason
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] RDMA/iwcm: Allow AFONLY binding for IPv6 addresses.
2021-02-19 14:02 Bernard Metzler
@ 2021-02-19 14:28 ` Bernard Metzler
0 siblings, 0 replies; 4+ messages in thread
From: Bernard Metzler @ 2021-02-19 14:28 UTC (permalink / raw)
To: linux-rdma; +Cc: jgg, Chuck Lever, Benjamin Coddington
Sorry I have to resend this patch, since I mistyped Jason's address.
Please don't reply to the first send.
Sorry and thanks
Bernard.
-----"Bernard Metzler" <bmt@zurich.ibm.com> wrote: -----
To: linux-rdma@vger.kernel.org
From: "Bernard Metzler" <bmt@zurich.ibm.com>
Date: 02/19/2021 03:05PM
Cc: gg@ziepe.ca, "Bernard Metzler" <bmt@zurich.ibm.com>, "Chuck Lever" <chuck.lever@oracle.com>, "Benjamin Coddington" <bcodding@redhat.com>
Subject: [EXTERNAL] [PATCH] RDMA/iwcm: Allow AFONLY binding for IPv6 addresses.
Binding IPv6 address/port to AF_INET6 domain only is provided
via rdma_set_afonly(), but was not signalled to the provider.
Applications like NFS/RDMA bind the same port to both IPv4
and IPv6 addresses simultaneously and thus rely on it working
correctly.
Tested-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Bernard Metzler <bmt@zurich.ibm.com>
---
drivers/infiniband/core/cma.c | 1 +
drivers/infiniband/sw/siw/siw_cm.c | 19 +++++++++++++++++--
include/rdma/iw_cm.h | 1 +
3 files changed, 19 insertions(+), 2 deletions(-)
diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index c51b84b2d2f3..046d42c1f8f3 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -2466,6 +2466,7 @@ static int cma_iw_listen(struct rdma_id_private *id_priv, int backlog)
id->tos = id_priv->tos;
id->tos_set = id_priv->tos_set;
+ id->afonly = id_priv->afonly;
id_priv->cm_id.iw = id;
memcpy(&id_priv->cm_id.iw->local_addr, cma_src_addr(id_priv),
diff --git a/drivers/infiniband/sw/siw/siw_cm.c b/drivers/infiniband/sw/siw/siw_cm.c
index 1f9e15b71504..8bd0d0785433 100644
--- a/drivers/infiniband/sw/siw/siw_cm.c
+++ b/drivers/infiniband/sw/siw/siw_cm.c
@@ -1300,7 +1300,7 @@ static void siw_cm_llp_state_change(struct sock *sk)
}
static int kernel_bindconnect(struct socket *s, struct sockaddr *laddr,
- struct sockaddr *raddr)
+ struct sockaddr *raddr, bool afonly)
{
int rv, flags = 0;
size_t size = laddr->sa_family == AF_INET ?
@@ -1311,6 +1311,12 @@ static int kernel_bindconnect(struct socket *s, struct sockaddr *laddr,
*/
sock_set_reuseaddr(s->sk);
+ if (afonly) {
+ rv = ip6_sock_set_v6only(s->sk);
+ if (rv)
+ return rv;
+ }
+
rv = s->ops->bind(s, laddr, size);
if (rv < 0)
return rv;
@@ -1371,7 +1377,7 @@ int siw_connect(struct iw_cm_id *id, struct iw_cm_conn_param *params)
* mode. Might be reconsidered for async connection setup at
* TCP level.
*/
- rv = kernel_bindconnect(s, laddr, raddr);
+ rv = kernel_bindconnect(s, laddr, raddr, id->afonly);
if (rv != 0) {
siw_dbg_qp(qp, "kernel_bindconnect: error %d\n", rv);
goto error;
@@ -1786,6 +1792,15 @@ int siw_create_listen(struct iw_cm_id *id, int backlog)
} else {
struct sockaddr_in6 *laddr = &to_sockaddr_in6(id->local_addr);
+ if (id->afonly) {
+ rv = ip6_sock_set_v6only(s->sk);
+ if (rv) {
+ siw_dbg(id->device,
+ "ip6_sock_set_v6only erro: %d\n", rv);
+ goto error;
+ }
+ }
+
/* 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;
diff --git a/include/rdma/iw_cm.h b/include/rdma/iw_cm.h
index 91975400e1b3..03abd30e6c8c 100644
--- a/include/rdma/iw_cm.h
+++ b/include/rdma/iw_cm.h
@@ -70,6 +70,7 @@ struct iw_cm_id {
u8 tos;
bool tos_set:1;
bool mapped:1;
+ bool afonly:1;
};
struct iw_cm_conn_param {
--
2.17.2
^ permalink raw reply related [flat|nested] 4+ messages in thread
* [PATCH] RDMA/iwcm: Allow AFONLY binding for IPv6 addresses.
@ 2021-02-19 14:02 Bernard Metzler
2021-02-19 14:28 ` Bernard Metzler
0 siblings, 1 reply; 4+ messages in thread
From: Bernard Metzler @ 2021-02-19 14:02 UTC (permalink / raw)
To: linux-rdma; +Cc: gg, Bernard Metzler, Chuck Lever, Benjamin Coddington
Binding IPv6 address/port to AF_INET6 domain only is provided
via rdma_set_afonly(), but was not signalled to the provider.
Applications like NFS/RDMA bind the same port to both IPv4
and IPv6 addresses simultaneously and thus rely on it working
correctly.
Tested-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Bernard Metzler <bmt@zurich.ibm.com>
---
drivers/infiniband/core/cma.c | 1 +
drivers/infiniband/sw/siw/siw_cm.c | 19 +++++++++++++++++--
include/rdma/iw_cm.h | 1 +
3 files changed, 19 insertions(+), 2 deletions(-)
diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index c51b84b2d2f3..046d42c1f8f3 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -2466,6 +2466,7 @@ static int cma_iw_listen(struct rdma_id_private *id_priv, int backlog)
id->tos = id_priv->tos;
id->tos_set = id_priv->tos_set;
+ id->afonly = id_priv->afonly;
id_priv->cm_id.iw = id;
memcpy(&id_priv->cm_id.iw->local_addr, cma_src_addr(id_priv),
diff --git a/drivers/infiniband/sw/siw/siw_cm.c b/drivers/infiniband/sw/siw/siw_cm.c
index 1f9e15b71504..8bd0d0785433 100644
--- a/drivers/infiniband/sw/siw/siw_cm.c
+++ b/drivers/infiniband/sw/siw/siw_cm.c
@@ -1300,7 +1300,7 @@ static void siw_cm_llp_state_change(struct sock *sk)
}
static int kernel_bindconnect(struct socket *s, struct sockaddr *laddr,
- struct sockaddr *raddr)
+ struct sockaddr *raddr, bool afonly)
{
int rv, flags = 0;
size_t size = laddr->sa_family == AF_INET ?
@@ -1311,6 +1311,12 @@ static int kernel_bindconnect(struct socket *s, struct sockaddr *laddr,
*/
sock_set_reuseaddr(s->sk);
+ if (afonly) {
+ rv = ip6_sock_set_v6only(s->sk);
+ if (rv)
+ return rv;
+ }
+
rv = s->ops->bind(s, laddr, size);
if (rv < 0)
return rv;
@@ -1371,7 +1377,7 @@ int siw_connect(struct iw_cm_id *id, struct iw_cm_conn_param *params)
* mode. Might be reconsidered for async connection setup at
* TCP level.
*/
- rv = kernel_bindconnect(s, laddr, raddr);
+ rv = kernel_bindconnect(s, laddr, raddr, id->afonly);
if (rv != 0) {
siw_dbg_qp(qp, "kernel_bindconnect: error %d\n", rv);
goto error;
@@ -1786,6 +1792,15 @@ int siw_create_listen(struct iw_cm_id *id, int backlog)
} else {
struct sockaddr_in6 *laddr = &to_sockaddr_in6(id->local_addr);
+ if (id->afonly) {
+ rv = ip6_sock_set_v6only(s->sk);
+ if (rv) {
+ siw_dbg(id->device,
+ "ip6_sock_set_v6only erro: %d\n", rv);
+ goto error;
+ }
+ }
+
/* 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;
diff --git a/include/rdma/iw_cm.h b/include/rdma/iw_cm.h
index 91975400e1b3..03abd30e6c8c 100644
--- a/include/rdma/iw_cm.h
+++ b/include/rdma/iw_cm.h
@@ -70,6 +70,7 @@ struct iw_cm_id {
u8 tos;
bool tos_set:1;
bool mapped:1;
+ bool afonly:1;
};
struct iw_cm_conn_param {
--
2.17.2
^ permalink raw reply related [flat|nested] 4+ messages in thread
end of thread, other threads:[~2021-03-10 19:50 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-02-19 14:34 [PATCH] RDMA/iwcm: Allow AFONLY binding for IPv6 addresses Bernard Metzler
2021-03-10 19:49 ` Jason Gunthorpe
-- strict thread matches above, loose matches on Subject: below --
2021-02-19 14:02 Bernard Metzler
2021-02-19 14:28 ` Bernard Metzler
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).