[PATCH 26/26] rdma/ucm: allow user space to specify AF_IB when joining multicast

[PATCH 26/26] rdma/ucm: allow user space to specify AF_IB when joining multicast

[Sean Hefty]

Allow user space applications to join multicast groups using MGIDs
directly.  MGIDs may be passed using AF_IB addresses.  Since the
current multicast join command only supports addresses as large as
sockaddr_in6, define a new structure for joining addresses specified
using sockaddr_ib.

Signed-off-by: Sean Hefty <sean.hefty-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---

 drivers/infiniband/core/cma.c  |    8 ++++--
 drivers/infiniband/core/ucma.c |   55 ++++++++++++++++++++++++++++++++--------
 include/rdma/rdma_user_cm.h    |   12 ++++++++-
 3 files changed, 60 insertions(+), 15 deletions(-)

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index e0bdb80..4adad37 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -2928,14 +2928,16 @@ static void cma_set_mgid(struct rdma_id_private *id_priv,
 								 0xFF10A01B)) {
 		/* IPv6 address is an SA assigned MGID. */
 		memcpy(mgid, &sin6->sin6_addr, sizeof *mgid);
+	} else if (addr->sa_family == AF_IB) {
+		memcpy(mgid, &((struct sockaddr_ib *) addr)->sib_addr, sizeof *mgid);
 	} else if ((addr->sa_family == AF_INET6)) {
 		ipv6_ib_mc_map(&sin6->sin6_addr, dev_addr->broadcast, mc_map);
-		if (id_priv->id.ps == RDMA_PS_UDP)
+		if (id_priv->id.ps == RDMA_PS_UDP || id_priv->id.ps == RDMA_PS_IB)
 			mc_map[7] = 0x01;	/* Use RDMA CM signature */
 		*mgid = *(union ib_gid *) (mc_map + 4);
 	} else {
 		ip_ib_mc_map(sin->sin_addr.s_addr, dev_addr->broadcast, mc_map);
-		if (id_priv->id.ps == RDMA_PS_UDP)
+		if (id_priv->id.ps == RDMA_PS_UDP || id_priv->id.ps == RDMA_PS_IB)
 			mc_map[7] = 0x01;	/* Use RDMA CM signature */
 		*mgid = *(union ib_gid *) (mc_map + 4);
 	}
@@ -2956,7 +2958,7 @@ static int cma_join_ib_multicast(struct rdma_id_private *id_priv,
 		return ret;
 
 	cma_set_mgid(id_priv, (struct sockaddr *) &mc->addr, &rec.mgid);
-	if (id_priv->id.ps == RDMA_PS_UDP)
+	if (id_priv->id.ps == RDMA_PS_UDP || id_priv->id.ps == RDMA_PS_IB)
 		rec.qkey = cpu_to_be32(RDMA_UDP_QKEY);
 	rdma_addr_get_sgid(dev_addr, &rec.port_gid);
 	rec.pkey = cpu_to_be16(ib_addr_get_pkey(dev_addr));
diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c
index f735150..8cf0d70 100644
--- a/drivers/infiniband/core/ucma.c
+++ b/drivers/infiniband/core/ucma.c
@@ -1161,23 +1161,23 @@ static ssize_t ucma_notify(struct ucma_file *file, const char __user *inbuf,
 	return ret;
 }
 
-static ssize_t ucma_join_ip_multicast(struct ucma_file *file,
-				      const char __user *inbuf,
-				      int in_len, int out_len)
+static ssize_t ucma_process_join(struct ucma_file *file,
+				 struct rdma_ucm_join_mcast *cmd,  int out_len)
 {
-	struct rdma_ucm_join_ip_mcast cmd;
 	struct rdma_ucm_create_id_resp resp;
 	struct ucma_context *ctx;
 	struct ucma_multicast *mc;
+	struct sockaddr *addr;
 	int ret;
 
 	if (out_len < sizeof(resp))
 		return -ENOSPC;
 
-	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
-		return -EFAULT;
+	addr = (struct sockaddr *) &cmd->addr;
+	if (cmd->reserved || !cmd->addr_size || (cmd->addr_size != rdma_addr_size(addr)))
+		return -EINVAL;
 
-	ctx = ucma_get_ctx(file, cmd.id);
+	ctx = ucma_get_ctx(file, cmd->id);
 	if (IS_ERR(ctx))
 		return PTR_ERR(ctx);
 
@@ -1188,14 +1188,14 @@ static ssize_t ucma_join_ip_multicast(struct ucma_file *file,
 		goto err1;
 	}
 
-	mc->uid = cmd.uid;
-	memcpy(&mc->addr, &cmd.addr, sizeof cmd.addr);
+	mc->uid = cmd->uid;
+	memcpy(&mc->addr, addr, cmd->addr_size);
 	ret = rdma_join_multicast(ctx->cm_id, (struct sockaddr *) &mc->addr, mc);
 	if (ret)
 		goto err2;
 
 	resp.id = mc->id;
-	if (copy_to_user((void __user *)(unsigned long)cmd.response,
+	if (copy_to_user((void __user *)(unsigned long) cmd->response,
 			 &resp, sizeof(resp))) {
 		ret = -EFAULT;
 		goto err3;
@@ -1220,6 +1220,38 @@ err1:
 	return ret;
 }
 
+static ssize_t ucma_join_ip_multicast(struct ucma_file *file,
+				      const char __user *inbuf,
+				      int in_len, int out_len)
+{
+	struct rdma_ucm_join_ip_mcast cmd;
+	struct rdma_ucm_join_mcast join_cmd;
+
+	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
+		return -EFAULT;
+
+	join_cmd.response = cmd.response;
+	join_cmd.uid = cmd.uid;
+	join_cmd.id = cmd.id;
+	join_cmd.addr_size = rdma_addr_size((struct sockaddr *) &cmd.addr);
+	join_cmd.reserved = 0;
+	memcpy(&join_cmd.addr, &cmd.addr, join_cmd.addr_size);
+
+	return ucma_process_join(file, &join_cmd, out_len);
+}
+
+static ssize_t ucma_join_multicast(struct ucma_file *file,
+				   const char __user *inbuf,
+				   int in_len, int out_len)
+{
+	struct rdma_ucm_join_mcast cmd;
+
+	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
+		return -EFAULT;
+
+	return ucma_process_join(file, &cmd, out_len);
+}
+
 static ssize_t ucma_leave_multicast(struct ucma_file *file,
 				    const char __user *inbuf,
 				    int in_len, int out_len)
@@ -1383,7 +1415,8 @@ static ssize_t (*ucma_cmd_table[])(struct ucma_file *file,
 	[RDMA_USER_CM_CMD_MIGRATE_ID]	 = ucma_migrate_id,
 	[RDMA_USER_CM_CMD_QUERY]	 = ucma_query,
 	[RDMA_USER_CM_CMD_BIND]		 = ucma_bind,
-	[RDMA_USER_CM_CMD_RESOLVE_ADDR]	 = ucma_resolve_addr
+	[RDMA_USER_CM_CMD_RESOLVE_ADDR]	 = ucma_resolve_addr,
+	[RDMA_USER_CM_CMD_JOIN_MCAST]	 = ucma_join_multicast
 };
 
 static ssize_t ucma_write(struct file *filp, const char __user *buf,
diff --git a/include/rdma/rdma_user_cm.h b/include/rdma/rdma_user_cm.h
index d488184..c699b4e 100644
--- a/include/rdma/rdma_user_cm.h
+++ b/include/rdma/rdma_user_cm.h
@@ -64,7 +64,8 @@ enum {
 	RDMA_USER_CM_CMD_MIGRATE_ID,
 	RDMA_USER_CM_CMD_QUERY,
 	RDMA_USER_CM_CMD_BIND,
-	RDMA_USER_CM_CMD_RESOLVE_ADDR
+	RDMA_USER_CM_CMD_RESOLVE_ADDR,
+	RDMA_USER_CM_CMD_JOIN_MCAST
 };
 
 /*
@@ -242,6 +243,15 @@ struct rdma_ucm_join_ip_mcast {
 	__u32 id;
 };
 
+struct rdma_ucm_join_mcast {
+	__u64 response;		/* rdma_ucma_create_id_resp */
+	__u64 uid;
+	__u32 id;
+	__u16 addr_size;
+	__u16 reserved;
+	struct sockaddr_storage addr;
+};
+
 struct rdma_ucm_get_event {
 	__u64 response;
 };



--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 26/26] rdma/ucm: allow user space to specify AF_IB when joining multicast

[Jason Gunthorpe]

On Thu, Apr 01, 2010 at 10:08:41AM -0700, Sean Hefty wrote:

>  	} else if ((addr->sa_family == AF_INET6)) {
>  		ipv6_ib_mc_map(&sin6->sin6_addr, dev_addr->broadcast, mc_map);
> -		if (id_priv->id.ps == RDMA_PS_UDP)
> +		if (id_priv->id.ps == RDMA_PS_UDP || id_priv->id.ps == RDMA_PS_IB)
>  			mc_map[7] = 0x01;	/* Use RDMA CM signature */
>  		*mgid = *(union ib_gid *) (mc_map + 4);

What does the combination of AF_INET/INET6 and RDMA_PS_IB mean?

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: [PATCH 26/26] rdma/ucm: allow user space to specify AF_IB when joining multicast

[Sean Hefty]

>>  	} else if ((addr->sa_family == AF_INET6)) {
>>  		ipv6_ib_mc_map(&sin6->sin6_addr, dev_addr->broadcast, mc_map);
>> -		if (id_priv->id.ps == RDMA_PS_UDP)
>> +		if (id_priv->id.ps == RDMA_PS_UDP || id_priv->id.ps ==
RDMA_PS_IB)
>>  			mc_map[7] = 0x01;	/* Use RDMA CM signature */
>>  		*mgid = *(union ib_gid *) (mc_map + 4);
>
>What does the combination of AF_INET/INET6 and RDMA_PS_IB mean?

This indicates that the rdma_cm_id is bound to AF_IB/RDMA_PS_IB, but that the
multicast join address was provided using AF_INET6.  The 'else' code that
follows the above code:

	} else {
 		ip_ib_mc_map(sin->sin_addr.s_addr, dev_addr->broadcast, mc_map);
-		if (id_priv->id.ps == RDMA_PS_UDP)
+		if (id_priv->id.ps == RDMA_PS_UDP || id_priv->id.ps ==
RDMA_PS_IB)
 			mc_map[7] = 0x01;	/* Use RDMA CM signature */
 		*mgid = *(union ib_gid *) (mc_map + 4);
 	}

is similar, except that the multicast join address was specified using AF_INET.

- Sean

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 26/26] rdma/ucm: allow user space to specify AF_IB when joining multicast

[Jason Gunthorpe]

On Thu, Apr 01, 2010 at 10:46:06AM -0700, Sean Hefty wrote:
> >>  	} else if ((addr->sa_family == AF_INET6)) {
> >>  		ipv6_ib_mc_map(&sin6->sin6_addr, dev_addr->broadcast, mc_map);
> >> -		if (id_priv->id.ps == RDMA_PS_UDP)
> >> +		if (id_priv->id.ps == RDMA_PS_UDP || id_priv->id.ps ==
> RDMA_PS_IB)
> >>  			mc_map[7] = 0x01;	/* Use RDMA CM signature */
> >>  		*mgid = *(union ib_gid *) (mc_map + 4);
> >
> >What does the combination of AF_INET/INET6 and RDMA_PS_IB mean?
> 
> This indicates that the rdma_cm_id is bound to AF_IB/RDMA_PS_IB, but that the
> multicast join address was provided using AF_INET6.  The 'else' code that
> follows the above code:

Humm, interesting.. So you propose to accept addresses for families
other than the one used to create the cm_id?

Just from a simplicity of the future perspective I'd say not to bother
...

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: [PATCH 26/26] rdma/ucm: allow user space to specify AF_IB when joining multicast

[Sean Hefty]

>Humm, interesting.. So you propose to accept addresses for families
>other than the one used to create the cm_id?
>
>Just from a simplicity of the future perspective I'd say not to bother

Part of the reasoning behind the existing code is that users requested the
ability to join a multicast group using an address of 0, so the transport could
select what the address.  Once the address was selected, it was communicated to
remote peers, who would join using the assigned address.  Today, the assigned
address is passed as AF_INET6.  This patch tries to 'fix' that and allow the
address to be passed to the kernel as AF_IB instead.

For compatibility reasons, I think we want to continue to support AF_INET
rdma_cm_id's using AF_INET6 or AF_IB.  Off the top of my head, I can't think of
any reason why support for AF_IB id's would need AF_INET/AF_INET6.  I don't know
if IBoE would want this or not, but I can update the patch to remove this
support.

- Sean

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 26/26 v2] rdma/ucm: allow user space to specify AF_IB when joining multicast

[Sean Hefty]

Allow user space applications to join multicast groups using MGIDs
directly.  MGIDs may be passed using AF_IB addresses.  Since the
current multicast join command only supports addresses as large as
sockaddr_in6, define a new structure for joining addresses specified
using sockaddr_ib.

Since port space IB allows the user to specify the qkey when
resolving a remote UD QP address, when joining the multicast
group use the qkey value, if one has been assigned.

Signed-off-by: Sean Hefty <sean.hefty-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
Changes from v1:
Restrict PS IB to using AF_IB multicast addresses.
Use the qkey stored with the rdma_cm_id if it has been set.

Note: specifying transport specific multicast parameters, such
as the qkey, are outside of this series.  This patch simply ensures
that the qkey for a multicast join is in sync with whatever value
the user specified for unicast operation.  Future support for transport
specific parameters can either be accomplished by extending
struct rdma_ucm_join_mcast or through a set_option command. 

 drivers/infiniband/core/cma.c  |   12 +++++++--
 drivers/infiniband/core/ucma.c |   55 ++++++++++++++++++++++++++++++++--------
 include/rdma/rdma_user_cm.h    |   12 ++++++++-
 3 files changed, 65 insertions(+), 14 deletions(-)

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index a642325..a42dd94 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -2934,6 +2934,8 @@ static void cma_set_mgid(struct rdma_id_private *id_priv,
 								 0xFF10A01B)) {
 		/* IPv6 address is an SA assigned MGID. */
 		memcpy(mgid, &sin6->sin6_addr, sizeof *mgid);
+	} else if (addr->sa_family == AF_IB) {
+		memcpy(mgid, &((struct sockaddr_ib *) addr)->sib_addr, sizeof *mgid);
 	} else if ((addr->sa_family == AF_INET6)) {
 		ipv6_ib_mc_map(&sin6->sin6_addr, dev_addr->broadcast, mc_map);
 		if (id_priv->id.ps == RDMA_PS_UDP)
@@ -2961,9 +2963,12 @@ static int cma_join_ib_multicast(struct rdma_id_private *id_priv,
 	if (ret)
 		return ret;
 
+	ret = cma_set_qkey(id_priv, 0);
+	if (ret)
+		return ret;
+
 	cma_set_mgid(id_priv, (struct sockaddr *) &mc->addr, &rec.mgid);
-	if (id_priv->id.ps == RDMA_PS_UDP)
-		rec.qkey = cpu_to_be32(RDMA_UDP_QKEY);
+	rec.qkey = cpu_to_be32(id_priv->qkey);
 	rdma_addr_get_sgid(dev_addr, &rec.port_gid);
 	rec.pkey = cpu_to_be16(ib_addr_get_pkey(dev_addr));
 	rec.join_state = 1;
@@ -3000,6 +3005,9 @@ int rdma_join_multicast(struct rdma_cm_id *id, struct sockaddr *addr,
 	    !cma_comp(id_priv, CMA_ADDR_RESOLVED))
 		return -EINVAL;
 
+	if (id_priv->id.ps == RDMA_PS_IB && addr->sa_family != AF_IB)
+		return -EINVAL;
+
 	mc = kmalloc(sizeof *mc, GFP_KERNEL);
 	if (!mc)
 		return -ENOMEM;
diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c
index da85dee..ab0e2f9 100644
--- a/drivers/infiniband/core/ucma.c
+++ b/drivers/infiniband/core/ucma.c
@@ -1166,23 +1166,23 @@ static ssize_t ucma_notify(struct ucma_file *file, const char __user *inbuf,
 	return ret;
 }
 
-static ssize_t ucma_join_ip_multicast(struct ucma_file *file,
-				      const char __user *inbuf,
-				      int in_len, int out_len)
+static ssize_t ucma_process_join(struct ucma_file *file,
+				 struct rdma_ucm_join_mcast *cmd,  int out_len)
 {
-	struct rdma_ucm_join_ip_mcast cmd;
 	struct rdma_ucm_create_id_resp resp;
 	struct ucma_context *ctx;
 	struct ucma_multicast *mc;
+	struct sockaddr *addr;
 	int ret;
 
 	if (out_len < sizeof(resp))
 		return -ENOSPC;
 
-	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
-		return -EFAULT;
+	addr = (struct sockaddr *) &cmd->addr;
+	if (cmd->reserved || !cmd->addr_size || (cmd->addr_size != rdma_addr_size(addr)))
+		return -EINVAL;
 
-	ctx = ucma_get_ctx(file, cmd.id);
+	ctx = ucma_get_ctx(file, cmd->id);
 	if (IS_ERR(ctx))
 		return PTR_ERR(ctx);
 
@@ -1193,14 +1193,14 @@ static ssize_t ucma_join_ip_multicast(struct ucma_file *file,
 		goto err1;
 	}
 
-	mc->uid = cmd.uid;
-	memcpy(&mc->addr, &cmd.addr, sizeof cmd.addr);
+	mc->uid = cmd->uid;
+	memcpy(&mc->addr, addr, cmd->addr_size);
 	ret = rdma_join_multicast(ctx->cm_id, (struct sockaddr *) &mc->addr, mc);
 	if (ret)
 		goto err2;
 
 	resp.id = mc->id;
-	if (copy_to_user((void __user *)(unsigned long)cmd.response,
+	if (copy_to_user((void __user *)(unsigned long) cmd->response,
 			 &resp, sizeof(resp))) {
 		ret = -EFAULT;
 		goto err3;
@@ -1225,6 +1225,38 @@ err1:
 	return ret;
 }
 
+static ssize_t ucma_join_ip_multicast(struct ucma_file *file,
+				      const char __user *inbuf,
+				      int in_len, int out_len)
+{
+	struct rdma_ucm_join_ip_mcast cmd;
+	struct rdma_ucm_join_mcast join_cmd;
+
+	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
+		return -EFAULT;
+
+	join_cmd.response = cmd.response;
+	join_cmd.uid = cmd.uid;
+	join_cmd.id = cmd.id;
+	join_cmd.addr_size = rdma_addr_size((struct sockaddr *) &cmd.addr);
+	join_cmd.reserved = 0;
+	memcpy(&join_cmd.addr, &cmd.addr, join_cmd.addr_size);
+
+	return ucma_process_join(file, &join_cmd, out_len);
+}
+
+static ssize_t ucma_join_multicast(struct ucma_file *file,
+				   const char __user *inbuf,
+				   int in_len, int out_len)
+{
+	struct rdma_ucm_join_mcast cmd;
+
+	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
+		return -EFAULT;
+
+	return ucma_process_join(file, &cmd, out_len);
+}
+
 static ssize_t ucma_leave_multicast(struct ucma_file *file,
 				    const char __user *inbuf,
 				    int in_len, int out_len)
@@ -1388,7 +1420,8 @@ static ssize_t (*ucma_cmd_table[])(struct ucma_file *file,
 	[RDMA_USER_CM_CMD_MIGRATE_ID]	 = ucma_migrate_id,
 	[RDMA_USER_CM_CMD_QUERY]	 = ucma_query,
 	[RDMA_USER_CM_CMD_BIND]		 = ucma_bind,
-	[RDMA_USER_CM_CMD_RESOLVE_ADDR]	 = ucma_resolve_addr
+	[RDMA_USER_CM_CMD_RESOLVE_ADDR]	 = ucma_resolve_addr,
+	[RDMA_USER_CM_CMD_JOIN_MCAST]	 = ucma_join_multicast
 };
 
 static ssize_t ucma_write(struct file *filp, const char __user *buf,
diff --git a/include/rdma/rdma_user_cm.h b/include/rdma/rdma_user_cm.h
index 8b2e9fb..a35c2da 100644
--- a/include/rdma/rdma_user_cm.h
+++ b/include/rdma/rdma_user_cm.h
@@ -64,7 +64,8 @@ enum {
 	RDMA_USER_CM_CMD_MIGRATE_ID,
 	RDMA_USER_CM_CMD_QUERY,
 	RDMA_USER_CM_CMD_BIND,
-	RDMA_USER_CM_CMD_RESOLVE_ADDR
+	RDMA_USER_CM_CMD_RESOLVE_ADDR,
+	RDMA_USER_CM_CMD_JOIN_MCAST
 };
 
 /*
@@ -242,6 +243,15 @@ struct rdma_ucm_join_ip_mcast {
 	__u32 id;
 };
 
+struct rdma_ucm_join_mcast {
+	__u64 response;		/* rdma_ucma_create_id_resp */
+	__u64 uid;
+	__u32 id;
+	__u16 addr_size;
+	__u16 reserved;
+	struct sockaddr_storage addr;
+};
+
 struct rdma_ucm_get_event {
 	__u64 response;
 };



--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html