Merge branches 'mlx5', 'qib' and 'srp' into for-next
diff --git a/Documentation/ABI/stable/sysfs-driver-ib_srp b/Documentation/ABI/stable/sysfs-driver-ib_srp
index 481aae9..5c53d28 100644
--- a/Documentation/ABI/stable/sysfs-driver-ib_srp
+++ b/Documentation/ABI/stable/sysfs-driver-ib_srp
@@ -54,6 +54,13 @@
ib_srp. Specifying a value that exceeds cmd_sg_entries is
only safe with partial memory descriptor list support enabled
(allow_ext_sg=1).
+ * comp_vector, a number in the range 0..n-1 specifying the
+ MSI-X completion vector. Some HCA's allocate multiple (n)
+ MSI-X vectors per HCA port. If the IRQ affinity masks of
+ these interrupts have been configured such that each MSI-X
+ interrupt is handled by a different CPU then the comp_vector
+ parameter can be used to spread the SRP completion workload
+ over multiple CPU's.
What: /sys/class/infiniband_srp/srp-<hca>-<port_number>/ibdev
Date: January 2, 2006
diff --git a/drivers/infiniband/core/addr.c b/drivers/infiniband/core/addr.c
index eaec8d7..e90f2b2 100644
--- a/drivers/infiniband/core/addr.c
+++ b/drivers/infiniband/core/addr.c
@@ -45,6 +45,7 @@
#include <net/addrconf.h>
#include <net/ip6_route.h>
#include <rdma/ib_addr.h>
+#include <rdma/ib.h>
MODULE_AUTHOR("Sean Hefty");
MODULE_DESCRIPTION("IB Address Translation");
@@ -70,6 +71,21 @@
static DECLARE_DELAYED_WORK(work, process_req);
static struct workqueue_struct *addr_wq;
+int rdma_addr_size(struct sockaddr *addr)
+{
+ switch (addr->sa_family) {
+ case AF_INET:
+ return sizeof(struct sockaddr_in);
+ case AF_INET6:
+ return sizeof(struct sockaddr_in6);
+ case AF_IB:
+ return sizeof(struct sockaddr_ib);
+ default:
+ return 0;
+ }
+}
+EXPORT_SYMBOL(rdma_addr_size);
+
void rdma_addr_register_client(struct rdma_addr_client *client)
{
atomic_set(&client->refcount, 1);
@@ -369,12 +385,12 @@
goto err;
}
- memcpy(src_in, src_addr, ip_addr_size(src_addr));
+ memcpy(src_in, src_addr, rdma_addr_size(src_addr));
} else {
src_in->sa_family = dst_addr->sa_family;
}
- memcpy(dst_in, dst_addr, ip_addr_size(dst_addr));
+ memcpy(dst_in, dst_addr, rdma_addr_size(dst_addr));
req->addr = addr;
req->callback = callback;
req->context = context;
diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index 34fbc2f..f1c279f 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -50,6 +50,7 @@
#include <rdma/rdma_cm.h>
#include <rdma/rdma_cm_ib.h>
#include <rdma/rdma_netlink.h>
+#include <rdma/ib.h>
#include <rdma/ib_cache.h>
#include <rdma/ib_cm.h>
#include <rdma/ib_sa.h>
@@ -79,7 +80,6 @@
static LIST_HEAD(listen_any_list);
static DEFINE_MUTEX(lock);
static struct workqueue_struct *cma_wq;
-static DEFINE_IDR(sdp_ps);
static DEFINE_IDR(tcp_ps);
static DEFINE_IDR(udp_ps);
static DEFINE_IDR(ipoib_ps);
@@ -195,24 +195,7 @@
union cma_ip_addr dst_addr;
};
-struct sdp_hh {
- u8 bsdh[16];
- u8 sdp_version; /* Major version: 7:4 */
- u8 ip_version; /* IP version: 7:4 */
- u8 sdp_specific1[10];
- __be16 port;
- __be16 sdp_specific2;
- union cma_ip_addr src_addr;
- union cma_ip_addr dst_addr;
-};
-
-struct sdp_hah {
- u8 bsdh[16];
- u8 sdp_version;
-};
-
#define CMA_VERSION 0x00
-#define SDP_MAJ_VERSION 0x2
static int cma_comp(struct rdma_id_private *id_priv, enum rdma_cm_state comp)
{
@@ -261,21 +244,6 @@
hdr->ip_version = (ip_ver << 4) | (hdr->ip_version & 0xF);
}
-static inline u8 sdp_get_majv(u8 sdp_version)
-{
- return sdp_version >> 4;
-}
-
-static inline u8 sdp_get_ip_ver(struct sdp_hh *hh)
-{
- return hh->ip_version >> 4;
-}
-
-static inline void sdp_set_ip_ver(struct sdp_hh *hh, u8 ip_ver)
-{
- hh->ip_version = (ip_ver << 4) | (hh->ip_version & 0xF);
-}
-
static void cma_attach_to_dev(struct rdma_id_private *id_priv,
struct cma_device *cma_dev)
{
@@ -310,16 +278,40 @@
mutex_unlock(&lock);
}
-static int cma_set_qkey(struct rdma_id_private *id_priv)
+static inline struct sockaddr *cma_src_addr(struct rdma_id_private *id_priv)
+{
+ return (struct sockaddr *) &id_priv->id.route.addr.src_addr;
+}
+
+static inline struct sockaddr *cma_dst_addr(struct rdma_id_private *id_priv)
+{
+ return (struct sockaddr *) &id_priv->id.route.addr.dst_addr;
+}
+
+static inline unsigned short cma_family(struct rdma_id_private *id_priv)
+{
+ return id_priv->id.route.addr.src_addr.ss_family;
+}
+
+static int cma_set_qkey(struct rdma_id_private *id_priv, u32 qkey)
{
struct ib_sa_mcmember_rec rec;
int ret = 0;
- if (id_priv->qkey)
+ if (id_priv->qkey) {
+ if (qkey && id_priv->qkey != qkey)
+ return -EINVAL;
return 0;
+ }
+
+ if (qkey) {
+ id_priv->qkey = qkey;
+ return 0;
+ }
switch (id_priv->id.ps) {
case RDMA_PS_UDP:
+ case RDMA_PS_IB:
id_priv->qkey = RDMA_UDP_QKEY;
break;
case RDMA_PS_IPOIB:
@@ -358,6 +350,27 @@
return -EADDRNOTAVAIL;
}
+static void cma_translate_ib(struct sockaddr_ib *sib, struct rdma_dev_addr *dev_addr)
+{
+ dev_addr->dev_type = ARPHRD_INFINIBAND;
+ rdma_addr_set_sgid(dev_addr, (union ib_gid *) &sib->sib_addr);
+ ib_addr_set_pkey(dev_addr, ntohs(sib->sib_pkey));
+}
+
+static int cma_translate_addr(struct sockaddr *addr, struct rdma_dev_addr *dev_addr)
+{
+ int ret;
+
+ if (addr->sa_family != AF_IB) {
+ ret = rdma_translate_ip(addr, dev_addr);
+ } else {
+ cma_translate_ib((struct sockaddr_ib *) addr, dev_addr);
+ ret = 0;
+ }
+
+ return ret;
+}
+
static int cma_acquire_dev(struct rdma_id_private *id_priv)
{
struct rdma_dev_addr *dev_addr = &id_priv->id.route.addr.dev_addr;
@@ -401,6 +414,61 @@
return ret;
}
+/*
+ * Select the source IB device and address to reach the destination IB address.
+ */
+static int cma_resolve_ib_dev(struct rdma_id_private *id_priv)
+{
+ struct cma_device *cma_dev, *cur_dev;
+ struct sockaddr_ib *addr;
+ union ib_gid gid, sgid, *dgid;
+ u16 pkey, index;
+ u8 port, p;
+ int i;
+
+ cma_dev = NULL;
+ addr = (struct sockaddr_ib *) cma_dst_addr(id_priv);
+ dgid = (union ib_gid *) &addr->sib_addr;
+ pkey = ntohs(addr->sib_pkey);
+
+ list_for_each_entry(cur_dev, &dev_list, list) {
+ if (rdma_node_get_transport(cur_dev->device->node_type) != RDMA_TRANSPORT_IB)
+ continue;
+
+ for (p = 1; p <= cur_dev->device->phys_port_cnt; ++p) {
+ if (ib_find_cached_pkey(cur_dev->device, p, pkey, &index))
+ continue;
+
+ for (i = 0; !ib_get_cached_gid(cur_dev->device, p, i, &gid); i++) {
+ if (!memcmp(&gid, dgid, sizeof(gid))) {
+ cma_dev = cur_dev;
+ sgid = gid;
+ port = p;
+ goto found;
+ }
+
+ if (!cma_dev && (gid.global.subnet_prefix ==
+ dgid->global.subnet_prefix)) {
+ cma_dev = cur_dev;
+ sgid = gid;
+ port = p;
+ }
+ }
+ }
+ }
+
+ if (!cma_dev)
+ return -ENODEV;
+
+found:
+ cma_attach_to_dev(id_priv, cma_dev);
+ id_priv->id.port_num = port;
+ addr = (struct sockaddr_ib *) cma_src_addr(id_priv);
+ memcpy(&addr->sib_addr, &sgid, sizeof sgid);
+ cma_translate_ib(addr, &id_priv->id.route.addr.dev_addr);
+ return 0;
+}
+
static void cma_deref_id(struct rdma_id_private *id_priv)
{
if (atomic_dec_and_test(&id_priv->refcount))
@@ -630,7 +698,7 @@
*qp_attr_mask = IB_QP_STATE | IB_QP_PKEY_INDEX | IB_QP_PORT;
if (id_priv->id.qp_type == IB_QPT_UD) {
- ret = cma_set_qkey(id_priv);
+ ret = cma_set_qkey(id_priv, 0);
if (ret)
return ret;
@@ -679,26 +747,30 @@
static inline int cma_zero_addr(struct sockaddr *addr)
{
- struct in6_addr *ip6;
-
- if (addr->sa_family == AF_INET)
- return ipv4_is_zeronet(
- ((struct sockaddr_in *)addr)->sin_addr.s_addr);
- else {
- ip6 = &((struct sockaddr_in6 *) addr)->sin6_addr;
- return (ip6->s6_addr32[0] | ip6->s6_addr32[1] |
- ip6->s6_addr32[2] | ip6->s6_addr32[3]) == 0;
+ switch (addr->sa_family) {
+ case AF_INET:
+ return ipv4_is_zeronet(((struct sockaddr_in *)addr)->sin_addr.s_addr);
+ case AF_INET6:
+ return ipv6_addr_any(&((struct sockaddr_in6 *) addr)->sin6_addr);
+ case AF_IB:
+ return ib_addr_any(&((struct sockaddr_ib *) addr)->sib_addr);
+ default:
+ return 0;
}
}
static inline int cma_loopback_addr(struct sockaddr *addr)
{
- if (addr->sa_family == AF_INET)
- return ipv4_is_loopback(
- ((struct sockaddr_in *) addr)->sin_addr.s_addr);
- else
- return ipv6_addr_loopback(
- &((struct sockaddr_in6 *) addr)->sin6_addr);
+ switch (addr->sa_family) {
+ case AF_INET:
+ return ipv4_is_loopback(((struct sockaddr_in *) addr)->sin_addr.s_addr);
+ case AF_INET6:
+ return ipv6_addr_loopback(&((struct sockaddr_in6 *) addr)->sin6_addr);
+ case AF_IB:
+ return ib_addr_loopback(&((struct sockaddr_ib *) addr)->sib_addr);
+ default:
+ return 0;
+ }
}
static inline int cma_any_addr(struct sockaddr *addr)
@@ -715,18 +787,31 @@
case AF_INET:
return ((struct sockaddr_in *) src)->sin_addr.s_addr !=
((struct sockaddr_in *) dst)->sin_addr.s_addr;
- default:
+ case AF_INET6:
return ipv6_addr_cmp(&((struct sockaddr_in6 *) src)->sin6_addr,
&((struct sockaddr_in6 *) dst)->sin6_addr);
+ default:
+ return ib_addr_cmp(&((struct sockaddr_ib *) src)->sib_addr,
+ &((struct sockaddr_ib *) dst)->sib_addr);
}
}
-static inline __be16 cma_port(struct sockaddr *addr)
+static __be16 cma_port(struct sockaddr *addr)
{
- if (addr->sa_family == AF_INET)
+ struct sockaddr_ib *sib;
+
+ switch (addr->sa_family) {
+ case AF_INET:
return ((struct sockaddr_in *) addr)->sin_port;
- else
+ case AF_INET6:
return ((struct sockaddr_in6 *) addr)->sin6_port;
+ case AF_IB:
+ sib = (struct sockaddr_ib *) addr;
+ return htons((u16) (be64_to_cpu(sib->sib_sid) &
+ be64_to_cpu(sib->sib_sid_mask)));
+ default:
+ return 0;
+ }
}
static inline int cma_any_port(struct sockaddr *addr)
@@ -734,83 +819,92 @@
return !cma_port(addr);
}
-static int cma_get_net_info(void *hdr, enum rdma_port_space ps,
- u8 *ip_ver, __be16 *port,
- union cma_ip_addr **src, union cma_ip_addr **dst)
+static void cma_save_ib_info(struct rdma_cm_id *id, struct rdma_cm_id *listen_id,
+ struct ib_sa_path_rec *path)
{
- switch (ps) {
- case RDMA_PS_SDP:
- if (sdp_get_majv(((struct sdp_hh *) hdr)->sdp_version) !=
- SDP_MAJ_VERSION)
- return -EINVAL;
+ struct sockaddr_ib *listen_ib, *ib;
- *ip_ver = sdp_get_ip_ver(hdr);
- *port = ((struct sdp_hh *) hdr)->port;
- *src = &((struct sdp_hh *) hdr)->src_addr;
- *dst = &((struct sdp_hh *) hdr)->dst_addr;
- break;
- default:
- if (((struct cma_hdr *) hdr)->cma_version != CMA_VERSION)
- return -EINVAL;
+ listen_ib = (struct sockaddr_ib *) &listen_id->route.addr.src_addr;
+ ib = (struct sockaddr_ib *) &id->route.addr.src_addr;
+ ib->sib_family = listen_ib->sib_family;
+ ib->sib_pkey = path->pkey;
+ ib->sib_flowinfo = path->flow_label;
+ memcpy(&ib->sib_addr, &path->sgid, 16);
+ ib->sib_sid = listen_ib->sib_sid;
+ ib->sib_sid_mask = cpu_to_be64(0xffffffffffffffffULL);
+ ib->sib_scope_id = listen_ib->sib_scope_id;
- *ip_ver = cma_get_ip_ver(hdr);
- *port = ((struct cma_hdr *) hdr)->port;
- *src = &((struct cma_hdr *) hdr)->src_addr;
- *dst = &((struct cma_hdr *) hdr)->dst_addr;
- break;
+ ib = (struct sockaddr_ib *) &id->route.addr.dst_addr;
+ ib->sib_family = listen_ib->sib_family;
+ ib->sib_pkey = path->pkey;
+ ib->sib_flowinfo = path->flow_label;
+ memcpy(&ib->sib_addr, &path->dgid, 16);
+}
+
+static void cma_save_ip4_info(struct rdma_cm_id *id, struct rdma_cm_id *listen_id,
+ struct cma_hdr *hdr)
+{
+ struct sockaddr_in *listen4, *ip4;
+
+ listen4 = (struct sockaddr_in *) &listen_id->route.addr.src_addr;
+ ip4 = (struct sockaddr_in *) &id->route.addr.src_addr;
+ ip4->sin_family = listen4->sin_family;
+ ip4->sin_addr.s_addr = hdr->dst_addr.ip4.addr;
+ ip4->sin_port = listen4->sin_port;
+
+ ip4 = (struct sockaddr_in *) &id->route.addr.dst_addr;
+ ip4->sin_family = listen4->sin_family;
+ ip4->sin_addr.s_addr = hdr->src_addr.ip4.addr;
+ ip4->sin_port = hdr->port;
+}
+
+static void cma_save_ip6_info(struct rdma_cm_id *id, struct rdma_cm_id *listen_id,
+ struct cma_hdr *hdr)
+{
+ struct sockaddr_in6 *listen6, *ip6;
+
+ listen6 = (struct sockaddr_in6 *) &listen_id->route.addr.src_addr;
+ ip6 = (struct sockaddr_in6 *) &id->route.addr.src_addr;
+ ip6->sin6_family = listen6->sin6_family;
+ ip6->sin6_addr = hdr->dst_addr.ip6;
+ ip6->sin6_port = listen6->sin6_port;
+
+ ip6 = (struct sockaddr_in6 *) &id->route.addr.dst_addr;
+ ip6->sin6_family = listen6->sin6_family;
+ ip6->sin6_addr = hdr->src_addr.ip6;
+ ip6->sin6_port = hdr->port;
+}
+
+static int cma_save_net_info(struct rdma_cm_id *id, struct rdma_cm_id *listen_id,
+ struct ib_cm_event *ib_event)
+{
+ struct cma_hdr *hdr;
+
+ if (listen_id->route.addr.src_addr.ss_family == AF_IB) {
+ cma_save_ib_info(id, listen_id, ib_event->param.req_rcvd.primary_path);
+ return 0;
}
- if (*ip_ver != 4 && *ip_ver != 6)
+ hdr = ib_event->private_data;
+ if (hdr->cma_version != CMA_VERSION)
return -EINVAL;
+
+ switch (cma_get_ip_ver(hdr)) {
+ case 4:
+ cma_save_ip4_info(id, listen_id, hdr);
+ break;
+ case 6:
+ cma_save_ip6_info(id, listen_id, hdr);
+ break;
+ default:
+ return -EINVAL;
+ }
return 0;
}
-static void cma_save_net_info(struct rdma_addr *addr,
- struct rdma_addr *listen_addr,
- u8 ip_ver, __be16 port,
- union cma_ip_addr *src, union cma_ip_addr *dst)
+static inline int cma_user_data_offset(struct rdma_id_private *id_priv)
{
- struct sockaddr_in *listen4, *ip4;
- struct sockaddr_in6 *listen6, *ip6;
-
- switch (ip_ver) {
- case 4:
- listen4 = (struct sockaddr_in *) &listen_addr->src_addr;
- ip4 = (struct sockaddr_in *) &addr->src_addr;
- ip4->sin_family = listen4->sin_family;
- ip4->sin_addr.s_addr = dst->ip4.addr;
- ip4->sin_port = listen4->sin_port;
-
- ip4 = (struct sockaddr_in *) &addr->dst_addr;
- ip4->sin_family = listen4->sin_family;
- ip4->sin_addr.s_addr = src->ip4.addr;
- ip4->sin_port = port;
- break;
- case 6:
- listen6 = (struct sockaddr_in6 *) &listen_addr->src_addr;
- ip6 = (struct sockaddr_in6 *) &addr->src_addr;
- ip6->sin6_family = listen6->sin6_family;
- ip6->sin6_addr = dst->ip6;
- ip6->sin6_port = listen6->sin6_port;
-
- ip6 = (struct sockaddr_in6 *) &addr->dst_addr;
- ip6->sin6_family = listen6->sin6_family;
- ip6->sin6_addr = src->ip6;
- ip6->sin6_port = port;
- break;
- default:
- break;
- }
-}
-
-static inline int cma_user_data_offset(enum rdma_port_space ps)
-{
- switch (ps) {
- case RDMA_PS_SDP:
- return 0;
- default:
- return sizeof(struct cma_hdr);
- }
+ return cma_family(id_priv) == AF_IB ? 0 : sizeof(struct cma_hdr);
}
static void cma_cancel_route(struct rdma_id_private *id_priv)
@@ -861,8 +955,7 @@
cma_cancel_route(id_priv);
break;
case RDMA_CM_LISTEN:
- if (cma_any_addr((struct sockaddr *) &id_priv->id.route.addr.src_addr)
- && !id_priv->cma_dev)
+ if (cma_any_addr(cma_src_addr(id_priv)) && !id_priv->cma_dev)
cma_cancel_listens(id_priv);
break;
default:
@@ -977,16 +1070,6 @@
return ret;
}
-static int cma_verify_rep(struct rdma_id_private *id_priv, void *data)
-{
- if (id_priv->id.ps == RDMA_PS_SDP &&
- sdp_get_majv(((struct sdp_hah *) data)->sdp_version) !=
- SDP_MAJ_VERSION)
- return -EINVAL;
-
- return 0;
-}
-
static void cma_set_rep_event_data(struct rdma_cm_event *event,
struct ib_cm_rep_event_param *rep_data,
void *private_data)
@@ -1021,15 +1104,13 @@
event.status = -ETIMEDOUT;
break;
case IB_CM_REP_RECEIVED:
- event.status = cma_verify_rep(id_priv, ib_event->private_data);
- if (event.status)
- event.event = RDMA_CM_EVENT_CONNECT_ERROR;
- else if (id_priv->id.qp && id_priv->id.ps != RDMA_PS_SDP) {
+ if (id_priv->id.qp) {
event.status = cma_rep_recv(id_priv);
event.event = event.status ? RDMA_CM_EVENT_CONNECT_ERROR :
RDMA_CM_EVENT_ESTABLISHED;
- } else
+ } else {
event.event = RDMA_CM_EVENT_CONNECT_RESPONSE;
+ }
cma_set_rep_event_data(&event, &ib_event->param.rep_rcvd,
ib_event->private_data);
break;
@@ -1085,22 +1166,16 @@
struct rdma_id_private *id_priv;
struct rdma_cm_id *id;
struct rdma_route *rt;
- union cma_ip_addr *src, *dst;
- __be16 port;
- u8 ip_ver;
int ret;
- if (cma_get_net_info(ib_event->private_data, listen_id->ps,
- &ip_ver, &port, &src, &dst))
- return NULL;
-
id = rdma_create_id(listen_id->event_handler, listen_id->context,
listen_id->ps, ib_event->param.req_rcvd.qp_type);
if (IS_ERR(id))
return NULL;
- cma_save_net_info(&id->route.addr, &listen_id->route.addr,
- ip_ver, port, src, dst);
+ id_priv = container_of(id, struct rdma_id_private, id);
+ if (cma_save_net_info(id, listen_id, ib_event))
+ goto err;
rt = &id->route;
rt->num_paths = ib_event->param.req_rcvd.alternate_path ? 2 : 1;
@@ -1113,19 +1188,17 @@
if (rt->num_paths == 2)
rt->path_rec[1] = *ib_event->param.req_rcvd.alternate_path;
- if (cma_any_addr((struct sockaddr *) &rt->addr.src_addr)) {
+ if (cma_any_addr(cma_src_addr(id_priv))) {
rt->addr.dev_addr.dev_type = ARPHRD_INFINIBAND;
rdma_addr_set_sgid(&rt->addr.dev_addr, &rt->path_rec[0].sgid);
ib_addr_set_pkey(&rt->addr.dev_addr, be16_to_cpu(rt->path_rec[0].pkey));
} else {
- ret = rdma_translate_ip((struct sockaddr *) &rt->addr.src_addr,
- &rt->addr.dev_addr);
+ ret = cma_translate_addr(cma_src_addr(id_priv), &rt->addr.dev_addr);
if (ret)
goto err;
}
rdma_addr_set_dgid(&rt->addr.dev_addr, &rt->path_rec[0].dgid);
- id_priv = container_of(id, struct rdma_id_private, id);
id_priv->state = RDMA_CM_CONNECT;
return id_priv;
@@ -1139,9 +1212,6 @@
{
struct rdma_id_private *id_priv;
struct rdma_cm_id *id;
- union cma_ip_addr *src, *dst;
- __be16 port;
- u8 ip_ver;
int ret;
id = rdma_create_id(listen_id->event_handler, listen_id->context,
@@ -1149,22 +1219,16 @@
if (IS_ERR(id))
return NULL;
-
- if (cma_get_net_info(ib_event->private_data, listen_id->ps,
- &ip_ver, &port, &src, &dst))
+ id_priv = container_of(id, struct rdma_id_private, id);
+ if (cma_save_net_info(id, listen_id, ib_event))
goto err;
- cma_save_net_info(&id->route.addr, &listen_id->route.addr,
- ip_ver, port, src, dst);
-
if (!cma_any_addr((struct sockaddr *) &id->route.addr.src_addr)) {
- ret = rdma_translate_ip((struct sockaddr *) &id->route.addr.src_addr,
- &id->route.addr.dev_addr);
+ ret = cma_translate_addr(cma_src_addr(id_priv), &id->route.addr.dev_addr);
if (ret)
goto err;
}
- id_priv = container_of(id, struct rdma_id_private, id);
id_priv->state = RDMA_CM_CONNECT;
return id_priv;
err:
@@ -1210,7 +1274,7 @@
return -ECONNABORTED;
memset(&event, 0, sizeof event);
- offset = cma_user_data_offset(listen_id->id.ps);
+ offset = cma_user_data_offset(listen_id);
event.event = RDMA_CM_EVENT_CONNECT_REQUEST;
if (ib_event->event == IB_CM_SIDR_REQ_RECEIVED) {
conn_id = cma_new_udp_id(&listen_id->id, ib_event);
@@ -1272,58 +1336,44 @@
return ret;
}
-static __be64 cma_get_service_id(enum rdma_port_space ps, struct sockaddr *addr)
+__be64 rdma_get_service_id(struct rdma_cm_id *id, struct sockaddr *addr)
{
- return cpu_to_be64(((u64)ps << 16) + be16_to_cpu(cma_port(addr)));
+ if (addr->sa_family == AF_IB)
+ return ((struct sockaddr_ib *) addr)->sib_sid;
+
+ return cpu_to_be64(((u64)id->ps << 16) + be16_to_cpu(cma_port(addr)));
}
+EXPORT_SYMBOL(rdma_get_service_id);
static void cma_set_compare_data(enum rdma_port_space ps, struct sockaddr *addr,
struct ib_cm_compare_data *compare)
{
struct cma_hdr *cma_data, *cma_mask;
- struct sdp_hh *sdp_data, *sdp_mask;
__be32 ip4_addr;
struct in6_addr ip6_addr;
memset(compare, 0, sizeof *compare);
cma_data = (void *) compare->data;
cma_mask = (void *) compare->mask;
- sdp_data = (void *) compare->data;
- sdp_mask = (void *) compare->mask;
switch (addr->sa_family) {
case AF_INET:
ip4_addr = ((struct sockaddr_in *) addr)->sin_addr.s_addr;
- if (ps == RDMA_PS_SDP) {
- sdp_set_ip_ver(sdp_data, 4);
- sdp_set_ip_ver(sdp_mask, 0xF);
- sdp_data->dst_addr.ip4.addr = ip4_addr;
- sdp_mask->dst_addr.ip4.addr = htonl(~0);
- } else {
- cma_set_ip_ver(cma_data, 4);
- cma_set_ip_ver(cma_mask, 0xF);
- if (!cma_any_addr(addr)) {
- cma_data->dst_addr.ip4.addr = ip4_addr;
- cma_mask->dst_addr.ip4.addr = htonl(~0);
- }
+ cma_set_ip_ver(cma_data, 4);
+ cma_set_ip_ver(cma_mask, 0xF);
+ if (!cma_any_addr(addr)) {
+ cma_data->dst_addr.ip4.addr = ip4_addr;
+ cma_mask->dst_addr.ip4.addr = htonl(~0);
}
break;
case AF_INET6:
ip6_addr = ((struct sockaddr_in6 *) addr)->sin6_addr;
- if (ps == RDMA_PS_SDP) {
- sdp_set_ip_ver(sdp_data, 6);
- sdp_set_ip_ver(sdp_mask, 0xF);
- sdp_data->dst_addr.ip6 = ip6_addr;
- memset(&sdp_mask->dst_addr.ip6, 0xFF,
- sizeof sdp_mask->dst_addr.ip6);
- } else {
- cma_set_ip_ver(cma_data, 6);
- cma_set_ip_ver(cma_mask, 0xF);
- if (!cma_any_addr(addr)) {
- cma_data->dst_addr.ip6 = ip6_addr;
- memset(&cma_mask->dst_addr.ip6, 0xFF,
- sizeof cma_mask->dst_addr.ip6);
- }
+ cma_set_ip_ver(cma_data, 6);
+ cma_set_ip_ver(cma_mask, 0xF);
+ if (!cma_any_addr(addr)) {
+ cma_data->dst_addr.ip6 = ip6_addr;
+ memset(&cma_mask->dst_addr.ip6, 0xFF,
+ sizeof cma_mask->dst_addr.ip6);
}
break;
default:
@@ -1347,9 +1397,9 @@
event.event = RDMA_CM_EVENT_DISCONNECTED;
break;
case IW_CM_EVENT_CONNECT_REPLY:
- sin = (struct sockaddr_in *) &id_priv->id.route.addr.src_addr;
+ sin = (struct sockaddr_in *) cma_src_addr(id_priv);
*sin = iw_event->local_addr;
- sin = (struct sockaddr_in *) &id_priv->id.route.addr.dst_addr;
+ sin = (struct sockaddr_in *) cma_dst_addr(id_priv);
*sin = iw_event->remote_addr;
switch (iw_event->status) {
case 0:
@@ -1447,9 +1497,9 @@
cm_id->context = conn_id;
cm_id->cm_handler = cma_iw_handler;
- sin = (struct sockaddr_in *) &new_cm_id->route.addr.src_addr;
+ sin = (struct sockaddr_in *) cma_src_addr(conn_id);
*sin = iw_event->local_addr;
- sin = (struct sockaddr_in *) &new_cm_id->route.addr.dst_addr;
+ sin = (struct sockaddr_in *) cma_dst_addr(conn_id);
*sin = iw_event->remote_addr;
ret = ib_query_device(conn_id->id.device, &attr);
@@ -1506,8 +1556,8 @@
id_priv->cm_id.ib = id;
- addr = (struct sockaddr *) &id_priv->id.route.addr.src_addr;
- svc_id = cma_get_service_id(id_priv->id.ps, addr);
+ addr = cma_src_addr(id_priv);
+ svc_id = rdma_get_service_id(&id_priv->id, addr);
if (cma_any_addr(addr) && !id_priv->afonly)
ret = ib_cm_listen(id_priv->cm_id.ib, svc_id, 0, NULL);
else {
@@ -1537,7 +1587,7 @@
id_priv->cm_id.iw = id;
- sin = (struct sockaddr_in *) &id_priv->id.route.addr.src_addr;
+ sin = (struct sockaddr_in *) cma_src_addr(id_priv);
id_priv->cm_id.iw->local_addr = *sin;
ret = iw_cm_listen(id_priv->cm_id.iw, backlog);
@@ -1567,6 +1617,10 @@
struct rdma_cm_id *id;
int ret;
+ if (cma_family(id_priv) == AF_IB &&
+ rdma_node_get_transport(cma_dev->device->node_type) != RDMA_TRANSPORT_IB)
+ return;
+
id = rdma_create_id(cma_listen_handler, id_priv, id_priv->id.ps,
id_priv->id.qp_type);
if (IS_ERR(id))
@@ -1575,8 +1629,8 @@
dev_id_priv = container_of(id, struct rdma_id_private, id);
dev_id_priv->state = RDMA_CM_ADDR_BOUND;
- memcpy(&id->route.addr.src_addr, &id_priv->id.route.addr.src_addr,
- ip_addr_size((struct sockaddr *) &id_priv->id.route.addr.src_addr));
+ memcpy(cma_src_addr(dev_id_priv), cma_src_addr(id_priv),
+ rdma_addr_size(cma_src_addr(id_priv)));
cma_attach_to_dev(dev_id_priv, cma_dev);
list_add_tail(&dev_id_priv->listen_list, &id_priv->listen_list);
@@ -1634,31 +1688,39 @@
static int cma_query_ib_route(struct rdma_id_private *id_priv, int timeout_ms,
struct cma_work *work)
{
- struct rdma_addr *addr = &id_priv->id.route.addr;
+ struct rdma_dev_addr *dev_addr = &id_priv->id.route.addr.dev_addr;
struct ib_sa_path_rec path_rec;
ib_sa_comp_mask comp_mask;
struct sockaddr_in6 *sin6;
+ struct sockaddr_ib *sib;
memset(&path_rec, 0, sizeof path_rec);
- rdma_addr_get_sgid(&addr->dev_addr, &path_rec.sgid);
- rdma_addr_get_dgid(&addr->dev_addr, &path_rec.dgid);
- path_rec.pkey = cpu_to_be16(ib_addr_get_pkey(&addr->dev_addr));
+ rdma_addr_get_sgid(dev_addr, &path_rec.sgid);
+ rdma_addr_get_dgid(dev_addr, &path_rec.dgid);
+ path_rec.pkey = cpu_to_be16(ib_addr_get_pkey(dev_addr));
path_rec.numb_path = 1;
path_rec.reversible = 1;
- path_rec.service_id = cma_get_service_id(id_priv->id.ps,
- (struct sockaddr *) &addr->dst_addr);
+ path_rec.service_id = rdma_get_service_id(&id_priv->id, cma_dst_addr(id_priv));
comp_mask = IB_SA_PATH_REC_DGID | IB_SA_PATH_REC_SGID |
IB_SA_PATH_REC_PKEY | IB_SA_PATH_REC_NUMB_PATH |
IB_SA_PATH_REC_REVERSIBLE | IB_SA_PATH_REC_SERVICE_ID;
- if (addr->src_addr.ss_family == AF_INET) {
+ switch (cma_family(id_priv)) {
+ case AF_INET:
path_rec.qos_class = cpu_to_be16((u16) id_priv->tos);
comp_mask |= IB_SA_PATH_REC_QOS_CLASS;
- } else {
- sin6 = (struct sockaddr_in6 *) &addr->src_addr;
+ break;
+ case AF_INET6:
+ sin6 = (struct sockaddr_in6 *) cma_src_addr(id_priv);
path_rec.traffic_class = (u8) (be32_to_cpu(sin6->sin6_flowinfo) >> 20);
comp_mask |= IB_SA_PATH_REC_TRAFFIC_CLASS;
+ break;
+ case AF_IB:
+ sib = (struct sockaddr_ib *) cma_src_addr(id_priv);
+ path_rec.traffic_class = (u8) (be32_to_cpu(sib->sib_flowinfo) >> 20);
+ comp_mask |= IB_SA_PATH_REC_TRAFFIC_CLASS;
+ break;
}
id_priv->query_id = ib_sa_path_rec_get(&sa_client, id_priv->id.device,
@@ -1800,14 +1862,9 @@
struct rdma_addr *addr = &route->addr;
struct cma_work *work;
int ret;
- struct sockaddr_in *src_addr = (struct sockaddr_in *)&route->addr.src_addr;
- struct sockaddr_in *dst_addr = (struct sockaddr_in *)&route->addr.dst_addr;
struct net_device *ndev = NULL;
u16 vid;
- if (src_addr->sin_family != dst_addr->sin_family)
- return -EINVAL;
-
work = kzalloc(sizeof *work, GFP_KERNEL);
if (!work)
return -ENOMEM;
@@ -1913,28 +1970,57 @@
}
EXPORT_SYMBOL(rdma_resolve_route);
+static void cma_set_loopback(struct sockaddr *addr)
+{
+ switch (addr->sa_family) {
+ case AF_INET:
+ ((struct sockaddr_in *) addr)->sin_addr.s_addr = htonl(INADDR_LOOPBACK);
+ break;
+ case AF_INET6:
+ ipv6_addr_set(&((struct sockaddr_in6 *) addr)->sin6_addr,
+ 0, 0, 0, htonl(1));
+ break;
+ default:
+ ib_addr_set(&((struct sockaddr_ib *) addr)->sib_addr,
+ 0, 0, 0, htonl(1));
+ break;
+ }
+}
+
static int cma_bind_loopback(struct rdma_id_private *id_priv)
{
- struct cma_device *cma_dev;
+ struct cma_device *cma_dev, *cur_dev;
struct ib_port_attr port_attr;
union ib_gid gid;
u16 pkey;
int ret;
u8 p;
+ cma_dev = NULL;
mutex_lock(&lock);
- if (list_empty(&dev_list)) {
+ list_for_each_entry(cur_dev, &dev_list, list) {
+ if (cma_family(id_priv) == AF_IB &&
+ rdma_node_get_transport(cur_dev->device->node_type) != RDMA_TRANSPORT_IB)
+ continue;
+
+ if (!cma_dev)
+ cma_dev = cur_dev;
+
+ for (p = 1; p <= cur_dev->device->phys_port_cnt; ++p) {
+ if (!ib_query_port(cur_dev->device, p, &port_attr) &&
+ port_attr.state == IB_PORT_ACTIVE) {
+ cma_dev = cur_dev;
+ goto port_found;
+ }
+ }
+ }
+
+ if (!cma_dev) {
ret = -ENODEV;
goto out;
}
- list_for_each_entry(cma_dev, &dev_list, list)
- for (p = 1; p <= cma_dev->device->phys_port_cnt; ++p)
- if (!ib_query_port(cma_dev->device, p, &port_attr) &&
- port_attr.state == IB_PORT_ACTIVE)
- goto port_found;
p = 1;
- cma_dev = list_entry(dev_list.next, struct cma_device, list);
port_found:
ret = ib_get_cached_gid(cma_dev->device, p, 0, &gid);
@@ -1953,6 +2039,7 @@
ib_addr_set_pkey(&id_priv->id.route.addr.dev_addr, pkey);
id_priv->id.port_num = p;
cma_attach_to_dev(id_priv, cma_dev);
+ cma_set_loopback(cma_src_addr(id_priv));
out:
mutex_unlock(&lock);
return ret;
@@ -1980,8 +2067,7 @@
event.event = RDMA_CM_EVENT_ADDR_ERROR;
event.status = status;
} else {
- memcpy(&id_priv->id.route.addr.src_addr, src_addr,
- ip_addr_size(src_addr));
+ memcpy(cma_src_addr(id_priv), src_addr, rdma_addr_size(src_addr));
event.event = RDMA_CM_EVENT_ADDR_RESOLVED;
}
@@ -2000,7 +2086,6 @@
static int cma_resolve_loopback(struct rdma_id_private *id_priv)
{
struct cma_work *work;
- struct sockaddr *src, *dst;
union ib_gid gid;
int ret;
@@ -2017,18 +2102,36 @@
rdma_addr_get_sgid(&id_priv->id.route.addr.dev_addr, &gid);
rdma_addr_set_dgid(&id_priv->id.route.addr.dev_addr, &gid);
- src = (struct sockaddr *) &id_priv->id.route.addr.src_addr;
- if (cma_zero_addr(src)) {
- dst = (struct sockaddr *) &id_priv->id.route.addr.dst_addr;
- if ((src->sa_family = dst->sa_family) == AF_INET) {
- ((struct sockaddr_in *)src)->sin_addr =
- ((struct sockaddr_in *)dst)->sin_addr;
- } else {
- ((struct sockaddr_in6 *)src)->sin6_addr =
- ((struct sockaddr_in6 *)dst)->sin6_addr;
- }
+ work->id = id_priv;
+ INIT_WORK(&work->work, cma_work_handler);
+ work->old_state = RDMA_CM_ADDR_QUERY;
+ work->new_state = RDMA_CM_ADDR_RESOLVED;
+ work->event.event = RDMA_CM_EVENT_ADDR_RESOLVED;
+ queue_work(cma_wq, &work->work);
+ return 0;
+err:
+ kfree(work);
+ return ret;
+}
+
+static int cma_resolve_ib_addr(struct rdma_id_private *id_priv)
+{
+ struct cma_work *work;
+ int ret;
+
+ work = kzalloc(sizeof *work, GFP_KERNEL);
+ if (!work)
+ return -ENOMEM;
+
+ if (!id_priv->cma_dev) {
+ ret = cma_resolve_ib_dev(id_priv);
+ if (ret)
+ goto err;
}
+ rdma_addr_set_dgid(&id_priv->id.route.addr.dev_addr, (union ib_gid *)
+ &(((struct sockaddr_ib *) &id_priv->id.route.addr.dst_addr)->sib_addr));
+
work->id = id_priv;
INIT_WORK(&work->work, cma_work_handler);
work->old_state = RDMA_CM_ADDR_QUERY;
@@ -2046,9 +2149,13 @@
{
if (!src_addr || !src_addr->sa_family) {
src_addr = (struct sockaddr *) &id->route.addr.src_addr;
- if ((src_addr->sa_family = dst_addr->sa_family) == AF_INET6) {
+ src_addr->sa_family = dst_addr->sa_family;
+ if (dst_addr->sa_family == AF_INET6) {
((struct sockaddr_in6 *) src_addr)->sin6_scope_id =
((struct sockaddr_in6 *) dst_addr)->sin6_scope_id;
+ } else if (dst_addr->sa_family == AF_IB) {
+ ((struct sockaddr_ib *) src_addr)->sib_pkey =
+ ((struct sockaddr_ib *) dst_addr)->sib_pkey;
}
}
return rdma_bind_addr(id, src_addr);
@@ -2067,17 +2174,25 @@
return ret;
}
+ if (cma_family(id_priv) != dst_addr->sa_family)
+ return -EINVAL;
+
if (!cma_comp_exch(id_priv, RDMA_CM_ADDR_BOUND, RDMA_CM_ADDR_QUERY))
return -EINVAL;
atomic_inc(&id_priv->refcount);
- memcpy(&id->route.addr.dst_addr, dst_addr, ip_addr_size(dst_addr));
- if (cma_any_addr(dst_addr))
+ memcpy(cma_dst_addr(id_priv), dst_addr, rdma_addr_size(dst_addr));
+ if (cma_any_addr(dst_addr)) {
ret = cma_resolve_loopback(id_priv);
- else
- ret = rdma_resolve_ip(&addr_client, (struct sockaddr *) &id->route.addr.src_addr,
- dst_addr, &id->route.addr.dev_addr,
- timeout_ms, addr_handler, id_priv);
+ } else {
+ if (dst_addr->sa_family == AF_IB) {
+ ret = cma_resolve_ib_addr(id_priv);
+ } else {
+ ret = rdma_resolve_ip(&addr_client, cma_src_addr(id_priv),
+ dst_addr, &id->route.addr.dev_addr,
+ timeout_ms, addr_handler, id_priv);
+ }
+ }
if (ret)
goto err;
@@ -2097,7 +2212,7 @@
id_priv = container_of(id, struct rdma_id_private, id);
spin_lock_irqsave(&id_priv->lock, flags);
- if (id_priv->state == RDMA_CM_IDLE) {
+ if (reuse || id_priv->state == RDMA_CM_IDLE) {
id_priv->reuseaddr = reuse;
ret = 0;
} else {
@@ -2131,10 +2246,29 @@
static void cma_bind_port(struct rdma_bind_list *bind_list,
struct rdma_id_private *id_priv)
{
- struct sockaddr_in *sin;
+ struct sockaddr *addr;
+ struct sockaddr_ib *sib;
+ u64 sid, mask;
+ __be16 port;
- sin = (struct sockaddr_in *) &id_priv->id.route.addr.src_addr;
- sin->sin_port = htons(bind_list->port);
+ addr = cma_src_addr(id_priv);
+ port = htons(bind_list->port);
+
+ switch (addr->sa_family) {
+ case AF_INET:
+ ((struct sockaddr_in *) addr)->sin_port = port;
+ break;
+ case AF_INET6:
+ ((struct sockaddr_in6 *) addr)->sin6_port = port;
+ break;
+ case AF_IB:
+ sib = (struct sockaddr_ib *) addr;
+ sid = be64_to_cpu(sib->sib_sid);
+ mask = be64_to_cpu(sib->sib_sid_mask);
+ sib->sib_sid = cpu_to_be64((sid & mask) | (u64) ntohs(port));
+ sib->sib_sid_mask = cpu_to_be64(~0ULL);
+ break;
+ }
id_priv->bind_list = bind_list;
hlist_add_head(&id_priv->node, &bind_list->owners);
}
@@ -2205,7 +2339,7 @@
struct rdma_id_private *cur_id;
struct sockaddr *addr, *cur_addr;
- addr = (struct sockaddr *) &id_priv->id.route.addr.src_addr;
+ addr = cma_src_addr(id_priv);
hlist_for_each_entry(cur_id, &bind_list->owners, node) {
if (id_priv == cur_id)
continue;
@@ -2214,7 +2348,7 @@
cur_id->reuseaddr)
continue;
- cur_addr = (struct sockaddr *) &cur_id->id.route.addr.src_addr;
+ cur_addr = cma_src_addr(cur_id);
if (id_priv->afonly && cur_id->afonly &&
(addr->sa_family != cur_addr->sa_family))
continue;
@@ -2234,7 +2368,7 @@
unsigned short snum;
int ret;
- snum = ntohs(cma_port((struct sockaddr *) &id_priv->id.route.addr.src_addr));
+ snum = ntohs(cma_port(cma_src_addr(id_priv)));
if (snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE))
return -EACCES;
@@ -2261,33 +2395,67 @@
return ret;
}
+static struct idr *cma_select_inet_ps(struct rdma_id_private *id_priv)
+{
+ switch (id_priv->id.ps) {
+ case RDMA_PS_TCP:
+ return &tcp_ps;
+ case RDMA_PS_UDP:
+ return &udp_ps;
+ case RDMA_PS_IPOIB:
+ return &ipoib_ps;
+ case RDMA_PS_IB:
+ return &ib_ps;
+ default:
+ return NULL;
+ }
+}
+
+static struct idr *cma_select_ib_ps(struct rdma_id_private *id_priv)
+{
+ struct idr *ps = NULL;
+ struct sockaddr_ib *sib;
+ u64 sid_ps, mask, sid;
+
+ sib = (struct sockaddr_ib *) cma_src_addr(id_priv);
+ mask = be64_to_cpu(sib->sib_sid_mask) & RDMA_IB_IP_PS_MASK;
+ sid = be64_to_cpu(sib->sib_sid) & mask;
+
+ if ((id_priv->id.ps == RDMA_PS_IB) && (sid == (RDMA_IB_IP_PS_IB & mask))) {
+ sid_ps = RDMA_IB_IP_PS_IB;
+ ps = &ib_ps;
+ } else if (((id_priv->id.ps == RDMA_PS_IB) || (id_priv->id.ps == RDMA_PS_TCP)) &&
+ (sid == (RDMA_IB_IP_PS_TCP & mask))) {
+ sid_ps = RDMA_IB_IP_PS_TCP;
+ ps = &tcp_ps;
+ } else if (((id_priv->id.ps == RDMA_PS_IB) || (id_priv->id.ps == RDMA_PS_UDP)) &&
+ (sid == (RDMA_IB_IP_PS_UDP & mask))) {
+ sid_ps = RDMA_IB_IP_PS_UDP;
+ ps = &udp_ps;
+ }
+
+ if (ps) {
+ sib->sib_sid = cpu_to_be64(sid_ps | ntohs(cma_port((struct sockaddr *) sib)));
+ sib->sib_sid_mask = cpu_to_be64(RDMA_IB_IP_PS_MASK |
+ be64_to_cpu(sib->sib_sid_mask));
+ }
+ return ps;
+}
+
static int cma_get_port(struct rdma_id_private *id_priv)
{
struct idr *ps;
int ret;
- switch (id_priv->id.ps) {
- case RDMA_PS_SDP:
- ps = &sdp_ps;
- break;
- case RDMA_PS_TCP:
- ps = &tcp_ps;
- break;
- case RDMA_PS_UDP:
- ps = &udp_ps;
- break;
- case RDMA_PS_IPOIB:
- ps = &ipoib_ps;
- break;
- case RDMA_PS_IB:
- ps = &ib_ps;
- break;
- default:
+ if (cma_family(id_priv) != AF_IB)
+ ps = cma_select_inet_ps(id_priv);
+ else
+ ps = cma_select_ib_ps(id_priv);
+ if (!ps)
return -EPROTONOSUPPORT;
- }
mutex_lock(&lock);
- if (cma_any_port((struct sockaddr *) &id_priv->id.route.addr.src_addr))
+ if (cma_any_port(cma_src_addr(id_priv)))
ret = cma_alloc_any_port(ps, id_priv);
else
ret = cma_use_port(ps, id_priv);
@@ -2322,8 +2490,8 @@
id_priv = container_of(id, struct rdma_id_private, id);
if (id_priv->state == RDMA_CM_IDLE) {
- ((struct sockaddr *) &id->route.addr.src_addr)->sa_family = AF_INET;
- ret = rdma_bind_addr(id, (struct sockaddr *) &id->route.addr.src_addr);
+ id->route.addr.src_addr.ss_family = AF_INET;
+ ret = rdma_bind_addr(id, cma_src_addr(id_priv));
if (ret)
return ret;
}
@@ -2370,7 +2538,8 @@
struct rdma_id_private *id_priv;
int ret;
- if (addr->sa_family != AF_INET && addr->sa_family != AF_INET6)
+ if (addr->sa_family != AF_INET && addr->sa_family != AF_INET6 &&
+ addr->sa_family != AF_IB)
return -EAFNOSUPPORT;
id_priv = container_of(id, struct rdma_id_private, id);
@@ -2382,7 +2551,7 @@
goto err1;
if (!cma_any_addr(addr)) {
- ret = rdma_translate_ip(addr, &id->route.addr.dev_addr);
+ ret = cma_translate_addr(addr, &id->route.addr.dev_addr);
if (ret)
goto err1;
@@ -2391,7 +2560,7 @@
goto err1;
}
- memcpy(&id->route.addr.src_addr, addr, ip_addr_size(addr));
+ memcpy(cma_src_addr(id_priv), addr, rdma_addr_size(addr));
if (!(id_priv->options & (1 << CMA_OPTION_AFONLY))) {
if (addr->sa_family == AF_INET)
id_priv->afonly = 1;
@@ -2414,62 +2583,32 @@
}
EXPORT_SYMBOL(rdma_bind_addr);
-static int cma_format_hdr(void *hdr, enum rdma_port_space ps,
- struct rdma_route *route)
+static int cma_format_hdr(void *hdr, struct rdma_id_private *id_priv)
{
struct cma_hdr *cma_hdr;
- struct sdp_hh *sdp_hdr;
- if (route->addr.src_addr.ss_family == AF_INET) {
+ cma_hdr = hdr;
+ cma_hdr->cma_version = CMA_VERSION;
+ if (cma_family(id_priv) == AF_INET) {
struct sockaddr_in *src4, *dst4;
- src4 = (struct sockaddr_in *) &route->addr.src_addr;
- dst4 = (struct sockaddr_in *) &route->addr.dst_addr;
+ src4 = (struct sockaddr_in *) cma_src_addr(id_priv);
+ dst4 = (struct sockaddr_in *) cma_dst_addr(id_priv);
- switch (ps) {
- case RDMA_PS_SDP:
- sdp_hdr = hdr;
- if (sdp_get_majv(sdp_hdr->sdp_version) != SDP_MAJ_VERSION)
- return -EINVAL;
- sdp_set_ip_ver(sdp_hdr, 4);
- sdp_hdr->src_addr.ip4.addr = src4->sin_addr.s_addr;
- sdp_hdr->dst_addr.ip4.addr = dst4->sin_addr.s_addr;
- sdp_hdr->port = src4->sin_port;
- break;
- default:
- cma_hdr = hdr;
- cma_hdr->cma_version = CMA_VERSION;
- cma_set_ip_ver(cma_hdr, 4);
- cma_hdr->src_addr.ip4.addr = src4->sin_addr.s_addr;
- cma_hdr->dst_addr.ip4.addr = dst4->sin_addr.s_addr;
- cma_hdr->port = src4->sin_port;
- break;
- }
- } else {
+ cma_set_ip_ver(cma_hdr, 4);
+ cma_hdr->src_addr.ip4.addr = src4->sin_addr.s_addr;
+ cma_hdr->dst_addr.ip4.addr = dst4->sin_addr.s_addr;
+ cma_hdr->port = src4->sin_port;
+ } else if (cma_family(id_priv) == AF_INET6) {
struct sockaddr_in6 *src6, *dst6;
- src6 = (struct sockaddr_in6 *) &route->addr.src_addr;
- dst6 = (struct sockaddr_in6 *) &route->addr.dst_addr;
+ src6 = (struct sockaddr_in6 *) cma_src_addr(id_priv);
+ dst6 = (struct sockaddr_in6 *) cma_dst_addr(id_priv);
- switch (ps) {
- case RDMA_PS_SDP:
- sdp_hdr = hdr;
- if (sdp_get_majv(sdp_hdr->sdp_version) != SDP_MAJ_VERSION)
- return -EINVAL;
- sdp_set_ip_ver(sdp_hdr, 6);
- sdp_hdr->src_addr.ip6 = src6->sin6_addr;
- sdp_hdr->dst_addr.ip6 = dst6->sin6_addr;
- sdp_hdr->port = src6->sin6_port;
- break;
- default:
- cma_hdr = hdr;
- cma_hdr->cma_version = CMA_VERSION;
- cma_set_ip_ver(cma_hdr, 6);
- cma_hdr->src_addr.ip6 = src6->sin6_addr;
- cma_hdr->dst_addr.ip6 = dst6->sin6_addr;
- cma_hdr->port = src6->sin6_port;
- break;
- }
+ cma_set_ip_ver(cma_hdr, 6);
+ cma_hdr->src_addr.ip6 = src6->sin6_addr;
+ cma_hdr->dst_addr.ip6 = dst6->sin6_addr;
+ cma_hdr->port = src6->sin6_port;
}
return 0;
}
@@ -2499,15 +2638,10 @@
event.status = ib_event->param.sidr_rep_rcvd.status;
break;
}
- ret = cma_set_qkey(id_priv);
+ ret = cma_set_qkey(id_priv, rep->qkey);
if (ret) {
event.event = RDMA_CM_EVENT_ADDR_ERROR;
- event.status = -EINVAL;
- break;
- }
- if (id_priv->qkey != rep->qkey) {
- event.event = RDMA_CM_EVENT_UNREACHABLE;
- event.status = -EINVAL;
+ event.status = ret;
break;
}
ib_init_ah_from_path(id_priv->id.device, id_priv->id.port_num,
@@ -2542,27 +2676,31 @@
struct rdma_conn_param *conn_param)
{
struct ib_cm_sidr_req_param req;
- struct rdma_route *route;
struct ib_cm_id *id;
- int ret;
+ int offset, ret;
- req.private_data_len = sizeof(struct cma_hdr) +
- conn_param->private_data_len;
+ offset = cma_user_data_offset(id_priv);
+ req.private_data_len = offset + conn_param->private_data_len;
if (req.private_data_len < conn_param->private_data_len)
return -EINVAL;
- req.private_data = kzalloc(req.private_data_len, GFP_ATOMIC);
- if (!req.private_data)
- return -ENOMEM;
+ if (req.private_data_len) {
+ req.private_data = kzalloc(req.private_data_len, GFP_ATOMIC);
+ if (!req.private_data)
+ return -ENOMEM;
+ } else {
+ req.private_data = NULL;
+ }
if (conn_param->private_data && conn_param->private_data_len)
- memcpy((void *) req.private_data + sizeof(struct cma_hdr),
+ memcpy((void *) req.private_data + offset,
conn_param->private_data, conn_param->private_data_len);
- route = &id_priv->id.route;
- ret = cma_format_hdr((void *) req.private_data, id_priv->id.ps, route);
- if (ret)
- goto out;
+ if (req.private_data) {
+ ret = cma_format_hdr((void *) req.private_data, id_priv);
+ if (ret)
+ goto out;
+ }
id = ib_create_cm_id(id_priv->id.device, cma_sidr_rep_handler,
id_priv);
@@ -2572,9 +2710,8 @@
}
id_priv->cm_id.ib = id;
- req.path = route->path_rec;
- req.service_id = cma_get_service_id(id_priv->id.ps,
- (struct sockaddr *) &route->addr.dst_addr);
+ req.path = id_priv->id.route.path_rec;
+ req.service_id = rdma_get_service_id(&id_priv->id, cma_dst_addr(id_priv));
req.timeout_ms = 1 << (CMA_CM_RESPONSE_TIMEOUT - 8);
req.max_cm_retries = CMA_MAX_CM_RETRIES;
@@ -2598,14 +2735,18 @@
int offset, ret;
memset(&req, 0, sizeof req);
- offset = cma_user_data_offset(id_priv->id.ps);
+ offset = cma_user_data_offset(id_priv);
req.private_data_len = offset + conn_param->private_data_len;
if (req.private_data_len < conn_param->private_data_len)
return -EINVAL;
- private_data = kzalloc(req.private_data_len, GFP_ATOMIC);
- if (!private_data)
- return -ENOMEM;
+ if (req.private_data_len) {
+ private_data = kzalloc(req.private_data_len, GFP_ATOMIC);
+ if (!private_data)
+ return -ENOMEM;
+ } else {
+ private_data = NULL;
+ }
if (conn_param->private_data && conn_param->private_data_len)
memcpy(private_data + offset, conn_param->private_data,
@@ -2619,17 +2760,18 @@
id_priv->cm_id.ib = id;
route = &id_priv->id.route;
- ret = cma_format_hdr(private_data, id_priv->id.ps, route);
- if (ret)
- goto out;
- req.private_data = private_data;
+ if (private_data) {
+ ret = cma_format_hdr(private_data, id_priv);
+ if (ret)
+ goto out;
+ req.private_data = private_data;
+ }
req.primary_path = &route->path_rec[0];
if (route->num_paths == 2)
req.alternate_path = &route->path_rec[1];
- req.service_id = cma_get_service_id(id_priv->id.ps,
- (struct sockaddr *) &route->addr.dst_addr);
+ req.service_id = rdma_get_service_id(&id_priv->id, cma_dst_addr(id_priv));
req.qp_num = id_priv->qp_num;
req.qp_type = id_priv->id.qp_type;
req.starting_psn = id_priv->seq_num;
@@ -2668,10 +2810,10 @@
id_priv->cm_id.iw = cm_id;
- sin = (struct sockaddr_in*) &id_priv->id.route.addr.src_addr;
+ sin = (struct sockaddr_in *) cma_src_addr(id_priv);
cm_id->local_addr = *sin;
- sin = (struct sockaddr_in*) &id_priv->id.route.addr.dst_addr;
+ sin = (struct sockaddr_in *) cma_dst_addr(id_priv);
cm_id->remote_addr = *sin;
ret = cma_modify_qp_rtr(id_priv, conn_param);
@@ -2789,7 +2931,7 @@
}
static int cma_send_sidr_rep(struct rdma_id_private *id_priv,
- enum ib_cm_sidr_status status,
+ enum ib_cm_sidr_status status, u32 qkey,
const void *private_data, int private_data_len)
{
struct ib_cm_sidr_rep_param rep;
@@ -2798,7 +2940,7 @@
memset(&rep, 0, sizeof rep);
rep.status = status;
if (status == IB_SIDR_SUCCESS) {
- ret = cma_set_qkey(id_priv);
+ ret = cma_set_qkey(id_priv, qkey);
if (ret)
return ret;
rep.qp_num = id_priv->qp_num;
@@ -2832,11 +2974,12 @@
if (id->qp_type == IB_QPT_UD) {
if (conn_param)
ret = cma_send_sidr_rep(id_priv, IB_SIDR_SUCCESS,
+ conn_param->qkey,
conn_param->private_data,
conn_param->private_data_len);
else
ret = cma_send_sidr_rep(id_priv, IB_SIDR_SUCCESS,
- NULL, 0);
+ 0, NULL, 0);
} else {
if (conn_param)
ret = cma_accept_ib(id_priv, conn_param);
@@ -2897,7 +3040,7 @@
switch (rdma_node_get_transport(id->device->node_type)) {
case RDMA_TRANSPORT_IB:
if (id->qp_type == IB_QPT_UD)
- ret = cma_send_sidr_rep(id_priv, IB_SIDR_REJECT,
+ ret = cma_send_sidr_rep(id_priv, IB_SIDR_REJECT, 0,
private_data, private_data_len);
else
ret = ib_send_cm_rej(id_priv->cm_id.ib,
@@ -2958,6 +3101,8 @@
cma_disable_callback(id_priv, RDMA_CM_ADDR_RESOLVED))
return 0;
+ if (!status)
+ status = cma_set_qkey(id_priv, be32_to_cpu(multicast->rec.qkey));
mutex_lock(&id_priv->qp_mutex);
if (!status && id_priv->id.qp)
status = ib_attach_mcast(id_priv->id.qp, &multicast->rec.mgid,
@@ -3004,6 +3149,8 @@
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)
@@ -3031,9 +3178,12 @@
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;
@@ -3170,7 +3320,7 @@
if (!mc)
return -ENOMEM;
- memcpy(&mc->addr, addr, ip_addr_size(addr));
+ memcpy(&mc->addr, addr, rdma_addr_size(addr));
mc->context = context;
mc->id_priv = id_priv;
@@ -3215,7 +3365,7 @@
id_priv = container_of(id, struct rdma_id_private, id);
spin_lock_irq(&id_priv->lock);
list_for_each_entry(mc, &id_priv->mc_list, list) {
- if (!memcmp(&mc->addr, addr, ip_addr_size(addr))) {
+ if (!memcmp(&mc->addr, addr, rdma_addr_size(addr))) {
list_del(&mc->list);
spin_unlock_irq(&id_priv->lock);
@@ -3436,33 +3586,16 @@
id_stats->bound_dev_if =
id->route.addr.dev_addr.bound_dev_if;
- if (id->route.addr.src_addr.ss_family == AF_INET) {
- if (ibnl_put_attr(skb, nlh,
- sizeof(struct sockaddr_in),
- &id->route.addr.src_addr,
- RDMA_NL_RDMA_CM_ATTR_SRC_ADDR)) {
- goto out;
- }
- if (ibnl_put_attr(skb, nlh,
- sizeof(struct sockaddr_in),
- &id->route.addr.dst_addr,
- RDMA_NL_RDMA_CM_ATTR_DST_ADDR)) {
- goto out;
- }
- } else if (id->route.addr.src_addr.ss_family == AF_INET6) {
- if (ibnl_put_attr(skb, nlh,
- sizeof(struct sockaddr_in6),
- &id->route.addr.src_addr,
- RDMA_NL_RDMA_CM_ATTR_SRC_ADDR)) {
- goto out;
- }
- if (ibnl_put_attr(skb, nlh,
- sizeof(struct sockaddr_in6),
- &id->route.addr.dst_addr,
- RDMA_NL_RDMA_CM_ATTR_DST_ADDR)) {
- goto out;
- }
- }
+ if (ibnl_put_attr(skb, nlh,
+ rdma_addr_size(cma_src_addr(id_priv)),
+ cma_src_addr(id_priv),
+ RDMA_NL_RDMA_CM_ATTR_SRC_ADDR))
+ goto out;
+ if (ibnl_put_attr(skb, nlh,
+ rdma_addr_size(cma_src_addr(id_priv)),
+ cma_dst_addr(id_priv),
+ RDMA_NL_RDMA_CM_ATTR_DST_ADDR))
+ goto out;
id_stats->pid = id_priv->owner;
id_stats->port_space = id->ps;
@@ -3527,7 +3660,6 @@
rdma_addr_unregister_client(&addr_client);
ib_sa_unregister_client(&sa_client);
destroy_workqueue(cma_wq);
- idr_destroy(&sdp_ps);
idr_destroy(&tcp_ps);
idr_destroy(&udp_ps);
idr_destroy(&ipoib_ps);
diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c
index 934f45e..9838ca4 100644
--- a/drivers/infiniband/core/sa_query.c
+++ b/drivers/infiniband/core/sa_query.c
@@ -652,6 +652,12 @@
}
EXPORT_SYMBOL(ib_sa_unpack_path);
+void ib_sa_pack_path(struct ib_sa_path_rec *rec, void *attribute)
+{
+ ib_pack(path_rec_table, ARRAY_SIZE(path_rec_table), rec, attribute);
+}
+EXPORT_SYMBOL(ib_sa_pack_path);
+
static void ib_sa_path_rec_callback(struct ib_sa_query *sa_query,
int status,
struct ib_sa_mad *mad)
diff --git a/drivers/infiniband/core/sysfs.c b/drivers/infiniband/core/sysfs.c
index 246fdc1..d9b78c4 100644
--- a/drivers/infiniband/core/sysfs.c
+++ b/drivers/infiniband/core/sysfs.c
@@ -545,8 +545,10 @@
p->gid_group.name = "gids";
p->gid_group.attrs = alloc_group_attrs(show_port_gid, attr.gid_tbl_len);
- if (!p->gid_group.attrs)
+ if (!p->gid_group.attrs) {
+ ret = -ENOMEM;
goto err_remove_pma;
+ }
ret = sysfs_create_group(&p->kobj, &p->gid_group);
if (ret)
@@ -555,8 +557,10 @@
p->pkey_group.name = "pkeys";
p->pkey_group.attrs = alloc_group_attrs(show_port_pkey,
attr.pkey_tbl_len);
- if (!p->pkey_group.attrs)
+ if (!p->pkey_group.attrs) {
+ ret = -ENOMEM;
goto err_remove_gid;
+ }
ret = sysfs_create_group(&p->kobj, &p->pkey_group);
if (ret)
diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c
index 5ca44cd..b0f189b 100644
--- a/drivers/infiniband/core/ucma.c
+++ b/drivers/infiniband/core/ucma.c
@@ -47,6 +47,8 @@
#include <rdma/ib_marshall.h>
#include <rdma/rdma_cm.h>
#include <rdma/rdma_cm_ib.h>
+#include <rdma/ib_addr.h>
+#include <rdma/ib.h>
MODULE_AUTHOR("Sean Hefty");
MODULE_DESCRIPTION("RDMA Userspace Connection Manager Access");
@@ -510,10 +512,10 @@
return ret;
}
-static ssize_t ucma_bind_addr(struct ucma_file *file, const char __user *inbuf,
+static ssize_t ucma_bind_ip(struct ucma_file *file, const char __user *inbuf,
int in_len, int out_len)
{
- struct rdma_ucm_bind_addr cmd;
+ struct rdma_ucm_bind_ip cmd;
struct ucma_context *ctx;
int ret;
@@ -529,11 +531,35 @@
return ret;
}
-static ssize_t ucma_resolve_addr(struct ucma_file *file,
- const char __user *inbuf,
- int in_len, int out_len)
+static ssize_t ucma_bind(struct ucma_file *file, const char __user *inbuf,
+ int in_len, int out_len)
{
- struct rdma_ucm_resolve_addr cmd;
+ struct rdma_ucm_bind cmd;
+ struct sockaddr *addr;
+ struct ucma_context *ctx;
+ int ret;
+
+ 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);
+ if (IS_ERR(ctx))
+ return PTR_ERR(ctx);
+
+ ret = rdma_bind_addr(ctx->cm_id, addr);
+ ucma_put_ctx(ctx);
+ return ret;
+}
+
+static ssize_t ucma_resolve_ip(struct ucma_file *file,
+ const char __user *inbuf,
+ int in_len, int out_len)
+{
+ struct rdma_ucm_resolve_ip cmd;
struct ucma_context *ctx;
int ret;
@@ -551,6 +577,33 @@
return ret;
}
+static ssize_t ucma_resolve_addr(struct ucma_file *file,
+ const char __user *inbuf,
+ int in_len, int out_len)
+{
+ struct rdma_ucm_resolve_addr cmd;
+ struct sockaddr *src, *dst;
+ struct ucma_context *ctx;
+ int ret;
+
+ if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
+ return -EFAULT;
+
+ src = (struct sockaddr *) &cmd.src_addr;
+ dst = (struct sockaddr *) &cmd.dst_addr;
+ if (cmd.reserved || (cmd.src_size && (cmd.src_size != rdma_addr_size(src))) ||
+ !cmd.dst_size || (cmd.dst_size != rdma_addr_size(dst)))
+ return -EINVAL;
+
+ ctx = ucma_get_ctx(file, cmd.id);
+ if (IS_ERR(ctx))
+ return PTR_ERR(ctx);
+
+ ret = rdma_resolve_addr(ctx->cm_id, src, dst, cmd.timeout_ms);
+ ucma_put_ctx(ctx);
+ return ret;
+}
+
static ssize_t ucma_resolve_route(struct ucma_file *file,
const char __user *inbuf,
int in_len, int out_len)
@@ -649,7 +702,7 @@
const char __user *inbuf,
int in_len, int out_len)
{
- struct rdma_ucm_query_route cmd;
+ struct rdma_ucm_query cmd;
struct rdma_ucm_query_route_resp resp;
struct ucma_context *ctx;
struct sockaddr *addr;
@@ -709,7 +762,162 @@
return ret;
}
-static void ucma_copy_conn_param(struct rdma_conn_param *dst,
+static void ucma_query_device_addr(struct rdma_cm_id *cm_id,
+ struct rdma_ucm_query_addr_resp *resp)
+{
+ if (!cm_id->device)
+ return;
+
+ resp->node_guid = (__force __u64) cm_id->device->node_guid;
+ resp->port_num = cm_id->port_num;
+ resp->pkey = (__force __u16) cpu_to_be16(
+ ib_addr_get_pkey(&cm_id->route.addr.dev_addr));
+}
+
+static ssize_t ucma_query_addr(struct ucma_context *ctx,
+ void __user *response, int out_len)
+{
+ struct rdma_ucm_query_addr_resp resp;
+ struct sockaddr *addr;
+ int ret = 0;
+
+ if (out_len < sizeof(resp))
+ return -ENOSPC;
+
+ memset(&resp, 0, sizeof resp);
+
+ addr = (struct sockaddr *) &ctx->cm_id->route.addr.src_addr;
+ resp.src_size = rdma_addr_size(addr);
+ memcpy(&resp.src_addr, addr, resp.src_size);
+
+ addr = (struct sockaddr *) &ctx->cm_id->route.addr.dst_addr;
+ resp.dst_size = rdma_addr_size(addr);
+ memcpy(&resp.dst_addr, addr, resp.dst_size);
+
+ ucma_query_device_addr(ctx->cm_id, &resp);
+
+ if (copy_to_user(response, &resp, sizeof(resp)))
+ ret = -EFAULT;
+
+ return ret;
+}
+
+static ssize_t ucma_query_path(struct ucma_context *ctx,
+ void __user *response, int out_len)
+{
+ struct rdma_ucm_query_path_resp *resp;
+ int i, ret = 0;
+
+ if (out_len < sizeof(*resp))
+ return -ENOSPC;
+
+ resp = kzalloc(out_len, GFP_KERNEL);
+ if (!resp)
+ return -ENOMEM;
+
+ resp->num_paths = ctx->cm_id->route.num_paths;
+ for (i = 0, out_len -= sizeof(*resp);
+ i < resp->num_paths && out_len > sizeof(struct ib_path_rec_data);
+ i++, out_len -= sizeof(struct ib_path_rec_data)) {
+
+ resp->path_data[i].flags = IB_PATH_GMP | IB_PATH_PRIMARY |
+ IB_PATH_BIDIRECTIONAL;
+ ib_sa_pack_path(&ctx->cm_id->route.path_rec[i],
+ &resp->path_data[i].path_rec);
+ }
+
+ if (copy_to_user(response, resp,
+ sizeof(*resp) + (i * sizeof(struct ib_path_rec_data))))
+ ret = -EFAULT;
+
+ kfree(resp);
+ return ret;
+}
+
+static ssize_t ucma_query_gid(struct ucma_context *ctx,
+ void __user *response, int out_len)
+{
+ struct rdma_ucm_query_addr_resp resp;
+ struct sockaddr_ib *addr;
+ int ret = 0;
+
+ if (out_len < sizeof(resp))
+ return -ENOSPC;
+
+ memset(&resp, 0, sizeof resp);
+
+ ucma_query_device_addr(ctx->cm_id, &resp);
+
+ addr = (struct sockaddr_ib *) &resp.src_addr;
+ resp.src_size = sizeof(*addr);
+ if (ctx->cm_id->route.addr.src_addr.ss_family == AF_IB) {
+ memcpy(addr, &ctx->cm_id->route.addr.src_addr, resp.src_size);
+ } else {
+ addr->sib_family = AF_IB;
+ addr->sib_pkey = (__force __be16) resp.pkey;
+ rdma_addr_get_sgid(&ctx->cm_id->route.addr.dev_addr,
+ (union ib_gid *) &addr->sib_addr);
+ addr->sib_sid = rdma_get_service_id(ctx->cm_id, (struct sockaddr *)
+ &ctx->cm_id->route.addr.src_addr);
+ }
+
+ addr = (struct sockaddr_ib *) &resp.dst_addr;
+ resp.dst_size = sizeof(*addr);
+ if (ctx->cm_id->route.addr.dst_addr.ss_family == AF_IB) {
+ memcpy(addr, &ctx->cm_id->route.addr.dst_addr, resp.dst_size);
+ } else {
+ addr->sib_family = AF_IB;
+ addr->sib_pkey = (__force __be16) resp.pkey;
+ rdma_addr_get_dgid(&ctx->cm_id->route.addr.dev_addr,
+ (union ib_gid *) &addr->sib_addr);
+ addr->sib_sid = rdma_get_service_id(ctx->cm_id, (struct sockaddr *)
+ &ctx->cm_id->route.addr.dst_addr);
+ }
+
+ if (copy_to_user(response, &resp, sizeof(resp)))
+ ret = -EFAULT;
+
+ return ret;
+}
+
+static ssize_t ucma_query(struct ucma_file *file,
+ const char __user *inbuf,
+ int in_len, int out_len)
+{
+ struct rdma_ucm_query cmd;
+ struct ucma_context *ctx;
+ void __user *response;
+ int ret;
+
+ if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
+ return -EFAULT;
+
+ response = (void __user *)(unsigned long) cmd.response;
+ ctx = ucma_get_ctx(file, cmd.id);
+ if (IS_ERR(ctx))
+ return PTR_ERR(ctx);
+
+ switch (cmd.option) {
+ case RDMA_USER_CM_QUERY_ADDR:
+ ret = ucma_query_addr(ctx, response, out_len);
+ break;
+ case RDMA_USER_CM_QUERY_PATH:
+ ret = ucma_query_path(ctx, response, out_len);
+ break;
+ case RDMA_USER_CM_QUERY_GID:
+ ret = ucma_query_gid(ctx, response, out_len);
+ break;
+ default:
+ ret = -ENOSYS;
+ break;
+ }
+
+ ucma_put_ctx(ctx);
+ return ret;
+}
+
+static void ucma_copy_conn_param(struct rdma_cm_id *id,
+ struct rdma_conn_param *dst,
struct rdma_ucm_conn_param *src)
{
dst->private_data = src->private_data;
@@ -721,6 +929,7 @@
dst->rnr_retry_count = src->rnr_retry_count;
dst->srq = src->srq;
dst->qp_num = src->qp_num;
+ dst->qkey = (id->route.addr.src_addr.ss_family == AF_IB) ? src->qkey : 0;
}
static ssize_t ucma_connect(struct ucma_file *file, const char __user *inbuf,
@@ -741,7 +950,7 @@
if (IS_ERR(ctx))
return PTR_ERR(ctx);
- ucma_copy_conn_param(&conn_param, &cmd.conn_param);
+ ucma_copy_conn_param(ctx->cm_id, &conn_param, &cmd.conn_param);
ret = rdma_connect(ctx->cm_id, &conn_param);
ucma_put_ctx(ctx);
return ret;
@@ -784,7 +993,7 @@
return PTR_ERR(ctx);
if (cmd.conn_param.valid) {
- ucma_copy_conn_param(&conn_param, &cmd.conn_param);
+ ucma_copy_conn_param(ctx->cm_id, &conn_param, &cmd.conn_param);
mutex_lock(&file->mut);
ret = rdma_accept(ctx->cm_id, &conn_param);
if (!ret)
@@ -1020,23 +1229,23 @@
return ret;
}
-static ssize_t ucma_join_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_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);
@@ -1047,14 +1256,14 @@
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;
@@ -1079,6 +1288,38 @@
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)
@@ -1221,25 +1462,29 @@
static ssize_t (*ucma_cmd_table[])(struct ucma_file *file,
const char __user *inbuf,
int in_len, int out_len) = {
- [RDMA_USER_CM_CMD_CREATE_ID] = ucma_create_id,
- [RDMA_USER_CM_CMD_DESTROY_ID] = ucma_destroy_id,
- [RDMA_USER_CM_CMD_BIND_ADDR] = ucma_bind_addr,
- [RDMA_USER_CM_CMD_RESOLVE_ADDR] = ucma_resolve_addr,
- [RDMA_USER_CM_CMD_RESOLVE_ROUTE]= ucma_resolve_route,
- [RDMA_USER_CM_CMD_QUERY_ROUTE] = ucma_query_route,
- [RDMA_USER_CM_CMD_CONNECT] = ucma_connect,
- [RDMA_USER_CM_CMD_LISTEN] = ucma_listen,
- [RDMA_USER_CM_CMD_ACCEPT] = ucma_accept,
- [RDMA_USER_CM_CMD_REJECT] = ucma_reject,
- [RDMA_USER_CM_CMD_DISCONNECT] = ucma_disconnect,
- [RDMA_USER_CM_CMD_INIT_QP_ATTR] = ucma_init_qp_attr,
- [RDMA_USER_CM_CMD_GET_EVENT] = ucma_get_event,
- [RDMA_USER_CM_CMD_GET_OPTION] = NULL,
- [RDMA_USER_CM_CMD_SET_OPTION] = ucma_set_option,
- [RDMA_USER_CM_CMD_NOTIFY] = ucma_notify,
- [RDMA_USER_CM_CMD_JOIN_MCAST] = ucma_join_multicast,
- [RDMA_USER_CM_CMD_LEAVE_MCAST] = ucma_leave_multicast,
- [RDMA_USER_CM_CMD_MIGRATE_ID] = ucma_migrate_id
+ [RDMA_USER_CM_CMD_CREATE_ID] = ucma_create_id,
+ [RDMA_USER_CM_CMD_DESTROY_ID] = ucma_destroy_id,
+ [RDMA_USER_CM_CMD_BIND_IP] = ucma_bind_ip,
+ [RDMA_USER_CM_CMD_RESOLVE_IP] = ucma_resolve_ip,
+ [RDMA_USER_CM_CMD_RESOLVE_ROUTE] = ucma_resolve_route,
+ [RDMA_USER_CM_CMD_QUERY_ROUTE] = ucma_query_route,
+ [RDMA_USER_CM_CMD_CONNECT] = ucma_connect,
+ [RDMA_USER_CM_CMD_LISTEN] = ucma_listen,
+ [RDMA_USER_CM_CMD_ACCEPT] = ucma_accept,
+ [RDMA_USER_CM_CMD_REJECT] = ucma_reject,
+ [RDMA_USER_CM_CMD_DISCONNECT] = ucma_disconnect,
+ [RDMA_USER_CM_CMD_INIT_QP_ATTR] = ucma_init_qp_attr,
+ [RDMA_USER_CM_CMD_GET_EVENT] = ucma_get_event,
+ [RDMA_USER_CM_CMD_GET_OPTION] = NULL,
+ [RDMA_USER_CM_CMD_SET_OPTION] = ucma_set_option,
+ [RDMA_USER_CM_CMD_NOTIFY] = ucma_notify,
+ [RDMA_USER_CM_CMD_JOIN_IP_MCAST] = ucma_join_ip_multicast,
+ [RDMA_USER_CM_CMD_LEAVE_MCAST] = ucma_leave_multicast,
+ [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_JOIN_MCAST] = ucma_join_multicast
};
static ssize_t ucma_write(struct file *filp, const char __user *buf,
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index a7d00f6..b3c07b0 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -334,7 +334,7 @@
resp.num_comp_vectors = file->device->num_comp_vectors;
- ret = get_unused_fd();
+ ret = get_unused_fd_flags(O_CLOEXEC);
if (ret < 0)
goto err_free;
resp.async_fd = ret;
@@ -1184,7 +1184,7 @@
if (copy_from_user(&cmd, buf, sizeof cmd))
return -EFAULT;
- ret = get_unused_fd();
+ ret = get_unused_fd_flags(O_CLOEXEC);
if (ret < 0)
return ret;
resp.fd = ret;
diff --git a/drivers/infiniband/hw/cxgb3/iwch_qp.c b/drivers/infiniband/hw/cxgb3/iwch_qp.c
index e5649e8..b57c0be 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_qp.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_qp.c
@@ -883,7 +883,8 @@
{
union t3_wr *wqe = qhp->wq.queue;
u16 count = 0;
- while ((count+1) != 0 && fw_riwrh_opcode((struct fw_riwrh *)wqe) == T3_WR_RCV) {
+
+ while (count < USHRT_MAX && fw_riwrh_opcode((struct fw_riwrh *)wqe) == T3_WR_RCV) {
count++;
wqe++;
}
diff --git a/drivers/infiniband/hw/ehca/ehca_main.c b/drivers/infiniband/hw/ehca/ehca_main.c
index f8a6291..8f43093 100644
--- a/drivers/infiniband/hw/ehca/ehca_main.c
+++ b/drivers/infiniband/hw/ehca/ehca_main.c
@@ -211,6 +211,7 @@
if (!ctblk_cache) {
ehca_gen_err("Cannot create ctblk SLAB cache.");
ehca_cleanup_small_qp_cache();
+ ret = -ENOMEM;
goto create_slab_caches6;
}
#endif
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma.h b/drivers/infiniband/hw/ocrdma/ocrdma.h
index 48970af..d540180 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma.h
@@ -42,8 +42,6 @@
#define OCRDMA_ROCE_DEV_VERSION "1.0.0"
#define OCRDMA_NODE_DESC "Emulex OneConnect RoCE HCA"
-#define ocrdma_err(format, arg...) printk(KERN_ERR format, ##arg)
-
#define OCRDMA_MAX_AH 512
#define OCRDMA_UVERBS(CMD_NAME) (1ull << IB_USER_VERBS_CMD_##CMD_NAME)
@@ -97,7 +95,6 @@
u16 id; /* qid, where to ring the doorbell. */
u16 head, tail;
bool created;
- atomic_t used; /* Number of valid elements in the queue */
};
struct ocrdma_eq {
@@ -198,7 +195,6 @@
struct ocrdma_ucontext *ucontext;
dma_addr_t pa;
u32 len;
- atomic_t use_cnt;
/* head of all qp's sq and rq for which cqes need to be flushed
* by the software.
@@ -210,7 +206,6 @@
struct ib_pd ibpd;
struct ocrdma_dev *dev;
struct ocrdma_ucontext *uctx;
- atomic_t use_cnt;
u32 id;
int num_dpp_qp;
u32 dpp_page;
@@ -241,16 +236,16 @@
struct ib_srq ibsrq;
struct ocrdma_dev *dev;
u8 __iomem *db;
- /* provide synchronization to multiple context(s) posting rqe */
- spinlock_t q_lock ____cacheline_aligned;
-
struct ocrdma_qp_hwq_info rq;
- struct ocrdma_pd *pd;
- atomic_t use_cnt;
- u32 id;
u64 *rqe_wr_id_tbl;
u32 *idx_bit_fields;
u32 bit_fields_len;
+
+ /* provide synchronization to multiple context(s) posting rqe */
+ spinlock_t q_lock ____cacheline_aligned;
+
+ struct ocrdma_pd *pd;
+ u32 id;
};
struct ocrdma_qp {
@@ -258,8 +253,6 @@
struct ocrdma_dev *dev;
u8 __iomem *sq_db;
- /* provide synchronization to multiple context(s) posting wqe, rqe */
- spinlock_t q_lock ____cacheline_aligned;
struct ocrdma_qp_hwq_info sq;
struct {
uint64_t wrid;
@@ -269,6 +262,9 @@
uint8_t rsvd[3];
} *wqe_wr_id_tbl;
u32 max_inline_data;
+
+ /* provide synchronization to multiple context(s) posting wqe, rqe */
+ spinlock_t q_lock ____cacheline_aligned;
struct ocrdma_cq *sq_cq;
/* list maintained per CQ to flush SQ errors */
struct list_head sq_entry;
@@ -296,10 +292,6 @@
u8 *ird_q_va;
};
-#define OCRDMA_GET_NUM_POSTED_SHIFT_VAL(qp) \
- (((qp->dev->nic_info.dev_family == OCRDMA_GEN2_FAMILY) && \
- (qp->id < 64)) ? 24 : 16)
-
struct ocrdma_hw_mr {
struct ocrdma_dev *dev;
u32 lkey;
@@ -390,4 +382,43 @@
return container_of(ibsrq, struct ocrdma_srq, ibsrq);
}
+
+static inline int ocrdma_get_num_posted_shift(struct ocrdma_qp *qp)
+{
+ return ((qp->dev->nic_info.dev_family == OCRDMA_GEN2_FAMILY &&
+ qp->id < 64) ? 24 : 16);
+}
+
+static inline int is_cqe_valid(struct ocrdma_cq *cq, struct ocrdma_cqe *cqe)
+{
+ int cqe_valid;
+ cqe_valid = le32_to_cpu(cqe->flags_status_srcqpn) & OCRDMA_CQE_VALID;
+ return ((cqe_valid == cq->phase) ? 1 : 0);
+}
+
+static inline int is_cqe_for_sq(struct ocrdma_cqe *cqe)
+{
+ return (le32_to_cpu(cqe->flags_status_srcqpn) &
+ OCRDMA_CQE_QTYPE) ? 0 : 1;
+}
+
+static inline int is_cqe_invalidated(struct ocrdma_cqe *cqe)
+{
+ return (le32_to_cpu(cqe->flags_status_srcqpn) &
+ OCRDMA_CQE_INVALIDATE) ? 1 : 0;
+}
+
+static inline int is_cqe_imm(struct ocrdma_cqe *cqe)
+{
+ return (le32_to_cpu(cqe->flags_status_srcqpn) &
+ OCRDMA_CQE_IMM) ? 1 : 0;
+}
+
+static inline int is_cqe_wr_imm(struct ocrdma_cqe *cqe)
+{
+ return (le32_to_cpu(cqe->flags_status_srcqpn) &
+ OCRDMA_CQE_WRITE_IMM) ? 1 : 0;
+}
+
+
#endif
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_hw.c b/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
index 71942af..0965278 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
@@ -128,7 +128,6 @@
static inline void ocrdma_mq_inc_head(struct ocrdma_dev *dev)
{
dev->mq.sq.head = (dev->mq.sq.head + 1) & (OCRDMA_MQ_LEN - 1);
- atomic_inc(&dev->mq.sq.used);
}
static inline void *ocrdma_get_mqe_rsp(struct ocrdma_dev *dev)
@@ -564,32 +563,19 @@
memset(cmd, 0, sizeof(*cmd));
num_pages = PAGES_4K_SPANNED(mq->va, mq->size);
- if (dev->nic_info.dev_family == OCRDMA_GEN2_FAMILY) {
- ocrdma_init_mch(&cmd->req, OCRDMA_CMD_CREATE_MQ,
- OCRDMA_SUBSYS_COMMON, sizeof(*cmd));
- cmd->v0.pages = num_pages;
- cmd->v0.async_cqid_valid = OCRDMA_CREATE_MQ_ASYNC_CQ_VALID;
- cmd->v0.async_cqid_valid = (cq->id << 1);
- cmd->v0.cqid_ringsize |= (ocrdma_encoded_q_len(mq->len) <<
- OCRDMA_CREATE_MQ_RING_SIZE_SHIFT);
- cmd->v0.cqid_ringsize |=
- (cq->id << OCRDMA_CREATE_MQ_V0_CQ_ID_SHIFT);
- cmd->v0.valid = OCRDMA_CREATE_MQ_VALID;
- pa = &cmd->v0.pa[0];
- } else {
- ocrdma_init_mch(&cmd->req, OCRDMA_CMD_CREATE_MQ_EXT,
- OCRDMA_SUBSYS_COMMON, sizeof(*cmd));
- cmd->req.rsvd_version = 1;
- cmd->v1.cqid_pages = num_pages;
- cmd->v1.cqid_pages |= (cq->id << OCRDMA_CREATE_MQ_CQ_ID_SHIFT);
- cmd->v1.async_cqid_valid = OCRDMA_CREATE_MQ_ASYNC_CQ_VALID;
- cmd->v1.async_event_bitmap = Bit(20);
- cmd->v1.async_cqid_ringsize = cq->id;
- cmd->v1.async_cqid_ringsize |= (ocrdma_encoded_q_len(mq->len) <<
- OCRDMA_CREATE_MQ_RING_SIZE_SHIFT);
- cmd->v1.valid = OCRDMA_CREATE_MQ_VALID;
- pa = &cmd->v1.pa[0];
- }
+ ocrdma_init_mch(&cmd->req, OCRDMA_CMD_CREATE_MQ_EXT,
+ OCRDMA_SUBSYS_COMMON, sizeof(*cmd));
+ cmd->req.rsvd_version = 1;
+ cmd->cqid_pages = num_pages;
+ cmd->cqid_pages |= (cq->id << OCRDMA_CREATE_MQ_CQ_ID_SHIFT);
+ cmd->async_cqid_valid = OCRDMA_CREATE_MQ_ASYNC_CQ_VALID;
+ cmd->async_event_bitmap = Bit(20);
+ cmd->async_cqid_ringsize = cq->id;
+ cmd->async_cqid_ringsize |= (ocrdma_encoded_q_len(mq->len) <<
+ OCRDMA_CREATE_MQ_RING_SIZE_SHIFT);
+ cmd->valid = OCRDMA_CREATE_MQ_VALID;
+ pa = &cmd->pa[0];
+
ocrdma_build_q_pages(pa, num_pages, mq->dma, PAGE_SIZE_4K);
status = be_roce_mcc_cmd(dev->nic_info.netdev,
cmd, sizeof(*cmd), NULL, NULL);
@@ -745,7 +731,7 @@
qp_event = 0;
srq_event = 0;
dev_event = 0;
- ocrdma_err("%s() unknown type=0x%x\n", __func__, type);
+ pr_err("%s() unknown type=0x%x\n", __func__, type);
break;
}
@@ -775,8 +761,8 @@
if (evt_code == OCRDMA_ASYNC_EVE_CODE)
ocrdma_dispatch_ibevent(dev, cqe);
else
- ocrdma_err("%s(%d) invalid evt code=0x%x\n",
- __func__, dev->id, evt_code);
+ pr_err("%s(%d) invalid evt code=0x%x\n", __func__,
+ dev->id, evt_code);
}
static void ocrdma_process_mcqe(struct ocrdma_dev *dev, struct ocrdma_mcqe *cqe)
@@ -790,8 +776,8 @@
dev->mqe_ctx.cmd_done = true;
wake_up(&dev->mqe_ctx.cmd_wait);
} else
- ocrdma_err("%s() cqe for invalid tag0x%x.expected=0x%x\n",
- __func__, cqe->tag_lo, dev->mqe_ctx.tag);
+ pr_err("%s() cqe for invalid tag0x%x.expected=0x%x\n",
+ __func__, cqe->tag_lo, dev->mqe_ctx.tag);
}
static int ocrdma_mq_cq_handler(struct ocrdma_dev *dev, u16 cq_id)
@@ -810,7 +796,7 @@
else if (cqe->valid_ae_cmpl_cons & OCRDMA_MCQE_CMPL_MASK)
ocrdma_process_mcqe(dev, cqe);
else
- ocrdma_err("%s() cqe->compl is not set.\n", __func__);
+ pr_err("%s() cqe->compl is not set.\n", __func__);
memset(cqe, 0, sizeof(struct ocrdma_mcqe));
ocrdma_mcq_inc_tail(dev);
}
@@ -869,7 +855,7 @@
cq = dev->cq_tbl[cq_idx];
if (cq == NULL) {
- ocrdma_err("%s%d invalid id=0x%x\n", __func__, dev->id, cq_idx);
+ pr_err("%s%d invalid id=0x%x\n", __func__, dev->id, cq_idx);
return;
}
spin_lock_irqsave(&cq->cq_lock, flags);
@@ -971,7 +957,7 @@
rsp = ocrdma_get_mqe_rsp(dev);
ocrdma_copy_le32_to_cpu(mqe, rsp, (sizeof(*mqe)));
if (cqe_status || ext_status) {
- ocrdma_err
+ pr_err
("%s() opcode=0x%x, cqe_status=0x%x, ext_status=0x%x\n",
__func__,
(rsp->u.rsp.subsys_op & OCRDMA_MBX_RSP_OPCODE_MASK) >>
@@ -1353,8 +1339,8 @@
if (dpp_cq)
return -EINVAL;
if (entries > dev->attr.max_cqe) {
- ocrdma_err("%s(%d) max_cqe=0x%x, requester_cqe=0x%x\n",
- __func__, dev->id, dev->attr.max_cqe, entries);
+ pr_err("%s(%d) max_cqe=0x%x, requester_cqe=0x%x\n",
+ __func__, dev->id, dev->attr.max_cqe, entries);
return -EINVAL;
}
if (dpp_cq && (dev->nic_info.dev_family != OCRDMA_GEN2_FAMILY))
@@ -1621,7 +1607,7 @@
status = ocrdma_mbx_reg_mr(dev, hwmr, pdid,
cur_pbl_cnt, hwmr->pbe_size, last);
if (status) {
- ocrdma_err("%s() status=%d\n", __func__, status);
+ pr_err("%s() status=%d\n", __func__, status);
return status;
}
/* if there is no more pbls to register then exit. */
@@ -1644,7 +1630,7 @@
break;
}
if (status)
- ocrdma_err("%s() err. status=%d\n", __func__, status);
+ pr_err("%s() err. status=%d\n", __func__, status);
return status;
}
@@ -1841,8 +1827,8 @@
status = ocrdma_build_q_conf(&max_wqe_allocated,
dev->attr.wqe_size, &hw_pages, &hw_page_size);
if (status) {
- ocrdma_err("%s() req. max_send_wr=0x%x\n", __func__,
- max_wqe_allocated);
+ pr_err("%s() req. max_send_wr=0x%x\n", __func__,
+ max_wqe_allocated);
return -EINVAL;
}
qp->sq.max_cnt = max_wqe_allocated;
@@ -1891,8 +1877,8 @@
status = ocrdma_build_q_conf(&max_rqe_allocated, dev->attr.rqe_size,
&hw_pages, &hw_page_size);
if (status) {
- ocrdma_err("%s() req. max_recv_wr=0x%x\n", __func__,
- attrs->cap.max_recv_wr + 1);
+ pr_err("%s() req. max_recv_wr=0x%x\n", __func__,
+ attrs->cap.max_recv_wr + 1);
return status;
}
qp->rq.max_cnt = max_rqe_allocated;
@@ -1900,7 +1886,7 @@
qp->rq.va = dma_alloc_coherent(&pdev->dev, len, &pa, GFP_KERNEL);
if (!qp->rq.va)
- return status;
+ return -ENOMEM;
memset(qp->rq.va, 0, len);
qp->rq.pa = pa;
qp->rq.len = len;
@@ -2087,10 +2073,10 @@
if (qp->rq.va)
dma_free_coherent(&pdev->dev, qp->rq.len, qp->rq.va, qp->rq.pa);
rq_err:
- ocrdma_err("%s(%d) rq_err\n", __func__, dev->id);
+ pr_err("%s(%d) rq_err\n", __func__, dev->id);
dma_free_coherent(&pdev->dev, qp->sq.len, qp->sq.va, qp->sq.pa);
sq_err:
- ocrdma_err("%s(%d) sq_err\n", __func__, dev->id);
+ pr_err("%s(%d) sq_err\n", __func__, dev->id);
kfree(cmd);
return status;
}
@@ -2127,7 +2113,7 @@
else if (rdma_link_local_addr(&in6))
rdma_get_ll_mac(&in6, mac_addr);
else {
- ocrdma_err("%s() fail to resolve mac_addr.\n", __func__);
+ pr_err("%s() fail to resolve mac_addr.\n", __func__);
return -EINVAL;
}
return 0;
@@ -2362,8 +2348,8 @@
dev->attr.rqe_size,
&hw_pages, &hw_page_size);
if (status) {
- ocrdma_err("%s() req. max_wr=0x%x\n", __func__,
- srq_attr->attr.max_wr);
+ pr_err("%s() req. max_wr=0x%x\n", __func__,
+ srq_attr->attr.max_wr);
status = -EINVAL;
goto ret;
}
@@ -2614,7 +2600,7 @@
ocrdma_destroy_qp_eqs(dev);
qpeq_err:
ocrdma_destroy_eq(dev, &dev->meq);
- ocrdma_err("%s() status=%d\n", __func__, status);
+ pr_err("%s() status=%d\n", __func__, status);
return status;
}
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_main.c b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
index 48928c8..ded416f1 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_main.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
@@ -378,7 +378,7 @@
spin_lock_init(&dev->flush_q_lock);
return 0;
alloc_err:
- ocrdma_err("%s(%d) error.\n", __func__, dev->id);
+ pr_err("%s(%d) error.\n", __func__, dev->id);
return -ENOMEM;
}
@@ -396,7 +396,7 @@
dev = (struct ocrdma_dev *)ib_alloc_device(sizeof(struct ocrdma_dev));
if (!dev) {
- ocrdma_err("Unable to allocate ib device\n");
+ pr_err("Unable to allocate ib device\n");
return NULL;
}
dev->mbx_cmd = kzalloc(sizeof(struct ocrdma_mqe_emb_cmd), GFP_KERNEL);
@@ -437,7 +437,7 @@
idr_err:
kfree(dev->mbx_cmd);
ib_dealloc_device(&dev->ibdev);
- ocrdma_err("%s() leaving. ret=%d\n", __func__, status);
+ pr_err("%s() leaving. ret=%d\n", __func__, status);
return NULL;
}
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_sli.h b/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
index c75cbdf..36b062d 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
@@ -608,16 +608,8 @@
OCRDMA_CREATE_MQ_ASYNC_CQ_VALID = Bit(0)
};
-struct ocrdma_create_mq_v0 {
- u32 pages;
- u32 cqid_ringsize;
- u32 valid;
- u32 async_cqid_valid;
- u32 rsvd;
- struct ocrdma_pa pa[8];
-} __packed;
-
-struct ocrdma_create_mq_v1 {
+struct ocrdma_create_mq_req {
+ struct ocrdma_mbx_hdr req;
u32 cqid_pages;
u32 async_event_bitmap;
u32 async_cqid_ringsize;
@@ -627,14 +619,6 @@
struct ocrdma_pa pa[8];
} __packed;
-struct ocrdma_create_mq_req {
- struct ocrdma_mbx_hdr req;
- union {
- struct ocrdma_create_mq_v0 v0;
- struct ocrdma_create_mq_v1 v1;
- };
-} __packed;
-
struct ocrdma_create_mq_rsp {
struct ocrdma_mbx_rsp rsp;
u32 id;
@@ -1550,21 +1534,6 @@
u32 flags_status_srcqpn; /* w3 */
} __packed;
-#define is_cqe_valid(cq, cqe) \
- (((le32_to_cpu(cqe->flags_status_srcqpn) & OCRDMA_CQE_VALID)\
- == cq->phase) ? 1 : 0)
-#define is_cqe_for_sq(cqe) \
- ((le32_to_cpu(cqe->flags_status_srcqpn) & OCRDMA_CQE_QTYPE) ? 0 : 1)
-#define is_cqe_for_rq(cqe) \
- ((le32_to_cpu(cqe->flags_status_srcqpn) & OCRDMA_CQE_QTYPE) ? 1 : 0)
-#define is_cqe_invalidated(cqe) \
- ((le32_to_cpu(cqe->flags_status_srcqpn) & OCRDMA_CQE_INVALIDATE) ? \
- 1 : 0)
-#define is_cqe_imm(cqe) \
- ((le32_to_cpu(cqe->flags_status_srcqpn) & OCRDMA_CQE_IMM) ? 1 : 0)
-#define is_cqe_wr_imm(cqe) \
- ((le32_to_cpu(cqe->flags_status_srcqpn) & OCRDMA_CQE_WRITE_IMM) ? 1 : 0)
-
struct ocrdma_sge {
u32 addr_hi;
u32 addr_lo;
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
index b29a424..dcfbab1 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
@@ -114,8 +114,8 @@
dev = get_ocrdma_dev(ibdev);
if (port > 1) {
- ocrdma_err("%s(%d) invalid_port=0x%x\n", __func__,
- dev->id, port);
+ pr_err("%s(%d) invalid_port=0x%x\n", __func__,
+ dev->id, port);
return -EINVAL;
}
netdev = dev->nic_info.netdev;
@@ -155,8 +155,7 @@
dev = get_ocrdma_dev(ibdev);
if (port > 1) {
- ocrdma_err("%s(%d) invalid_port=0x%x\n", __func__,
- dev->id, port);
+ pr_err("%s(%d) invalid_port=0x%x\n", __func__, dev->id, port);
return -EINVAL;
}
return 0;
@@ -398,7 +397,6 @@
kfree(pd);
return ERR_PTR(status);
}
- atomic_set(&pd->use_cnt, 0);
if (udata && context) {
status = ocrdma_copy_pd_uresp(pd, context, udata);
@@ -419,12 +417,6 @@
int status;
u64 usr_db;
- if (atomic_read(&pd->use_cnt)) {
- ocrdma_err("%s(%d) pd=0x%x is in use.\n",
- __func__, dev->id, pd->id);
- status = -EFAULT;
- goto dealloc_err;
- }
status = ocrdma_mbx_dealloc_pd(dev, pd);
if (pd->uctx) {
u64 dpp_db = dev->nic_info.dpp_unmapped_addr +
@@ -436,7 +428,6 @@
ocrdma_del_mmap(pd->uctx, usr_db, dev->nic_info.db_page_size);
}
kfree(pd);
-dealloc_err:
return status;
}
@@ -450,8 +441,8 @@
struct ocrdma_dev *dev = pd->dev;
if (acc & IB_ACCESS_REMOTE_WRITE && !(acc & IB_ACCESS_LOCAL_WRITE)) {
- ocrdma_err("%s(%d) leaving err, invalid access rights\n",
- __func__, dev->id);
+ pr_err("%s(%d) leaving err, invalid access rights\n",
+ __func__, dev->id);
return ERR_PTR(-EINVAL);
}
@@ -474,7 +465,6 @@
return ERR_PTR(-ENOMEM);
}
mr->pd = pd;
- atomic_inc(&pd->use_cnt);
mr->ibmr.lkey = mr->hwmr.lkey;
if (mr->hwmr.remote_wr || mr->hwmr.remote_rd)
mr->ibmr.rkey = mr->hwmr.lkey;
@@ -664,7 +654,6 @@
if (status)
goto mbx_err;
mr->pd = pd;
- atomic_inc(&pd->use_cnt);
mr->ibmr.lkey = mr->hwmr.lkey;
if (mr->hwmr.remote_wr || mr->hwmr.remote_rd)
mr->ibmr.rkey = mr->hwmr.lkey;
@@ -689,7 +678,6 @@
if (mr->hwmr.fr_mr == 0)
ocrdma_free_mr_pbl_tbl(dev, &mr->hwmr);
- atomic_dec(&mr->pd->use_cnt);
/* it could be user registered memory. */
if (mr->umem)
ib_umem_release(mr->umem);
@@ -714,8 +702,8 @@
uresp.phase_change = cq->phase_change ? 1 : 0;
status = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
if (status) {
- ocrdma_err("%s(%d) copy error cqid=0x%x.\n",
- __func__, cq->dev->id, cq->id);
+ pr_err("%s(%d) copy error cqid=0x%x.\n",
+ __func__, cq->dev->id, cq->id);
goto err;
}
uctx = get_ocrdma_ucontext(ib_ctx);
@@ -752,7 +740,6 @@
spin_lock_init(&cq->cq_lock);
spin_lock_init(&cq->comp_handler_lock);
- atomic_set(&cq->use_cnt, 0);
INIT_LIST_HEAD(&cq->sq_head);
INIT_LIST_HEAD(&cq->rq_head);
cq->dev = dev;
@@ -799,9 +786,6 @@
struct ocrdma_cq *cq = get_ocrdma_cq(ibcq);
struct ocrdma_dev *dev = cq->dev;
- if (atomic_read(&cq->use_cnt))
- return -EINVAL;
-
status = ocrdma_mbx_destroy_cq(dev, cq);
if (cq->ucontext) {
@@ -837,57 +821,56 @@
if (attrs->qp_type != IB_QPT_GSI &&
attrs->qp_type != IB_QPT_RC &&
attrs->qp_type != IB_QPT_UD) {
- ocrdma_err("%s(%d) unsupported qp type=0x%x requested\n",
- __func__, dev->id, attrs->qp_type);
+ pr_err("%s(%d) unsupported qp type=0x%x requested\n",
+ __func__, dev->id, attrs->qp_type);
return -EINVAL;
}
if (attrs->cap.max_send_wr > dev->attr.max_wqe) {
- ocrdma_err("%s(%d) unsupported send_wr=0x%x requested\n",
- __func__, dev->id, attrs->cap.max_send_wr);
- ocrdma_err("%s(%d) supported send_wr=0x%x\n",
- __func__, dev->id, dev->attr.max_wqe);
+ pr_err("%s(%d) unsupported send_wr=0x%x requested\n",
+ __func__, dev->id, attrs->cap.max_send_wr);
+ pr_err("%s(%d) supported send_wr=0x%x\n",
+ __func__, dev->id, dev->attr.max_wqe);
return -EINVAL;
}
if (!attrs->srq && (attrs->cap.max_recv_wr > dev->attr.max_rqe)) {
- ocrdma_err("%s(%d) unsupported recv_wr=0x%x requested\n",
- __func__, dev->id, attrs->cap.max_recv_wr);
- ocrdma_err("%s(%d) supported recv_wr=0x%x\n",
- __func__, dev->id, dev->attr.max_rqe);
+ pr_err("%s(%d) unsupported recv_wr=0x%x requested\n",
+ __func__, dev->id, attrs->cap.max_recv_wr);
+ pr_err("%s(%d) supported recv_wr=0x%x\n",
+ __func__, dev->id, dev->attr.max_rqe);
return -EINVAL;
}
if (attrs->cap.max_inline_data > dev->attr.max_inline_data) {
- ocrdma_err("%s(%d) unsupported inline data size=0x%x"
- " requested\n", __func__, dev->id,
- attrs->cap.max_inline_data);
- ocrdma_err("%s(%d) supported inline data size=0x%x\n",
- __func__, dev->id, dev->attr.max_inline_data);
+ pr_err("%s(%d) unsupported inline data size=0x%x requested\n",
+ __func__, dev->id, attrs->cap.max_inline_data);
+ pr_err("%s(%d) supported inline data size=0x%x\n",
+ __func__, dev->id, dev->attr.max_inline_data);
return -EINVAL;
}
if (attrs->cap.max_send_sge > dev->attr.max_send_sge) {
- ocrdma_err("%s(%d) unsupported send_sge=0x%x requested\n",
- __func__, dev->id, attrs->cap.max_send_sge);
- ocrdma_err("%s(%d) supported send_sge=0x%x\n",
- __func__, dev->id, dev->attr.max_send_sge);
+ pr_err("%s(%d) unsupported send_sge=0x%x requested\n",
+ __func__, dev->id, attrs->cap.max_send_sge);
+ pr_err("%s(%d) supported send_sge=0x%x\n",
+ __func__, dev->id, dev->attr.max_send_sge);
return -EINVAL;
}
if (attrs->cap.max_recv_sge > dev->attr.max_recv_sge) {
- ocrdma_err("%s(%d) unsupported recv_sge=0x%x requested\n",
- __func__, dev->id, attrs->cap.max_recv_sge);
- ocrdma_err("%s(%d) supported recv_sge=0x%x\n",
- __func__, dev->id, dev->attr.max_recv_sge);
+ pr_err("%s(%d) unsupported recv_sge=0x%x requested\n",
+ __func__, dev->id, attrs->cap.max_recv_sge);
+ pr_err("%s(%d) supported recv_sge=0x%x\n",
+ __func__, dev->id, dev->attr.max_recv_sge);
return -EINVAL;
}
/* unprivileged user space cannot create special QP */
if (ibpd->uobject && attrs->qp_type == IB_QPT_GSI) {
- ocrdma_err
+ pr_err
("%s(%d) Userspace can't create special QPs of type=0x%x\n",
__func__, dev->id, attrs->qp_type);
return -EINVAL;
}
/* allow creating only one GSI type of QP */
if (attrs->qp_type == IB_QPT_GSI && dev->gsi_qp_created) {
- ocrdma_err("%s(%d) GSI special QPs already created.\n",
- __func__, dev->id);
+ pr_err("%s(%d) GSI special QPs already created.\n",
+ __func__, dev->id);
return -EINVAL;
}
/* verify consumer QPs are not trying to use GSI QP's CQ */
@@ -896,8 +879,8 @@
(dev->gsi_sqcq == get_ocrdma_cq(attrs->recv_cq)) ||
(dev->gsi_rqcq == get_ocrdma_cq(attrs->send_cq)) ||
(dev->gsi_rqcq == get_ocrdma_cq(attrs->recv_cq))) {
- ocrdma_err("%s(%d) Consumer QP cannot use GSI CQs.\n",
- __func__, dev->id);
+ pr_err("%s(%d) Consumer QP cannot use GSI CQs.\n",
+ __func__, dev->id);
return -EINVAL;
}
}
@@ -949,7 +932,7 @@
}
status = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
if (status) {
- ocrdma_err("%s(%d) user copy error.\n", __func__, dev->id);
+ pr_err("%s(%d) user copy error.\n", __func__, dev->id);
goto err;
}
status = ocrdma_add_mmap(pd->uctx, uresp.sq_page_addr[0],
@@ -1023,15 +1006,6 @@
qp->state = OCRDMA_QPS_RST;
}
-static void ocrdma_set_qp_use_cnt(struct ocrdma_qp *qp, struct ocrdma_pd *pd)
-{
- atomic_inc(&pd->use_cnt);
- atomic_inc(&qp->sq_cq->use_cnt);
- atomic_inc(&qp->rq_cq->use_cnt);
- if (qp->srq)
- atomic_inc(&qp->srq->use_cnt);
- qp->ibqp.qp_num = qp->id;
-}
static void ocrdma_store_gsi_qp_cq(struct ocrdma_dev *dev,
struct ib_qp_init_attr *attrs)
@@ -1099,7 +1073,7 @@
goto cpy_err;
}
ocrdma_store_gsi_qp_cq(dev, attrs);
- ocrdma_set_qp_use_cnt(qp, pd);
+ qp->ibqp.qp_num = qp->id;
mutex_unlock(&dev->dev_lock);
return &qp->ibqp;
@@ -1112,7 +1086,7 @@
kfree(qp->wqe_wr_id_tbl);
kfree(qp->rqe_wr_id_tbl);
kfree(qp);
- ocrdma_err("%s(%d) error=%d\n", __func__, dev->id, status);
+ pr_err("%s(%d) error=%d\n", __func__, dev->id, status);
gen_err:
return ERR_PTR(status);
}
@@ -1162,10 +1136,10 @@
spin_unlock_irqrestore(&qp->q_lock, flags);
if (!ib_modify_qp_is_ok(old_qps, new_qps, ibqp->qp_type, attr_mask)) {
- ocrdma_err("%s(%d) invalid attribute mask=0x%x specified for "
- "qpn=0x%x of type=0x%x old_qps=0x%x, new_qps=0x%x\n",
- __func__, dev->id, attr_mask, qp->id, ibqp->qp_type,
- old_qps, new_qps);
+ pr_err("%s(%d) invalid attribute mask=0x%x specified for\n"
+ "qpn=0x%x of type=0x%x old_qps=0x%x, new_qps=0x%x\n",
+ __func__, dev->id, attr_mask, qp->id, ibqp->qp_type,
+ old_qps, new_qps);
goto param_err;
}
@@ -1475,11 +1449,6 @@
ocrdma_del_flush_qp(qp);
- atomic_dec(&qp->pd->use_cnt);
- atomic_dec(&qp->sq_cq->use_cnt);
- atomic_dec(&qp->rq_cq->use_cnt);
- if (qp->srq)
- atomic_dec(&qp->srq->use_cnt);
kfree(qp->wqe_wr_id_tbl);
kfree(qp->rqe_wr_id_tbl);
kfree(qp);
@@ -1565,14 +1534,12 @@
goto arm_err;
}
- atomic_set(&srq->use_cnt, 0);
if (udata) {
status = ocrdma_copy_srq_uresp(srq, udata);
if (status)
goto arm_err;
}
- atomic_inc(&pd->use_cnt);
return &srq->ibsrq;
arm_err:
@@ -1618,18 +1585,12 @@
srq = get_ocrdma_srq(ibsrq);
dev = srq->dev;
- if (atomic_read(&srq->use_cnt)) {
- ocrdma_err("%s(%d) err, srq=0x%x in use\n",
- __func__, dev->id, srq->id);
- return -EAGAIN;
- }
status = ocrdma_mbx_destroy_srq(dev, srq);
if (srq->pd->uctx)
ocrdma_del_mmap(srq->pd->uctx, (u64) srq->rq.pa, srq->rq.len);
- atomic_dec(&srq->pd->use_cnt);
kfree(srq->idx_bit_fields);
kfree(srq->rqe_wr_id_tbl);
kfree(srq);
@@ -1677,9 +1638,9 @@
{
if (wr->send_flags & IB_SEND_INLINE) {
if (wr->sg_list[0].length > qp->max_inline_data) {
- ocrdma_err("%s() supported_len=0x%x,"
- " unspported len req=0x%x\n", __func__,
- qp->max_inline_data, wr->sg_list[0].length);
+ pr_err("%s() supported_len=0x%x,\n"
+ " unspported len req=0x%x\n", __func__,
+ qp->max_inline_data, wr->sg_list[0].length);
return -EINVAL;
}
memcpy(sge,
@@ -1773,12 +1734,14 @@
spin_lock_irqsave(&qp->q_lock, flags);
if (qp->state != OCRDMA_QPS_RTS && qp->state != OCRDMA_QPS_SQD) {
spin_unlock_irqrestore(&qp->q_lock, flags);
+ *bad_wr = wr;
return -EINVAL;
}
while (wr) {
if (ocrdma_hwq_free_cnt(&qp->sq) == 0 ||
wr->num_sge > qp->sq.max_sges) {
+ *bad_wr = wr;
status = -ENOMEM;
break;
}
@@ -1856,7 +1819,7 @@
static void ocrdma_ring_rq_db(struct ocrdma_qp *qp)
{
- u32 val = qp->rq.dbid | (1 << OCRDMA_GET_NUM_POSTED_SHIFT_VAL(qp));
+ u32 val = qp->rq.dbid | (1 << ocrdma_get_num_posted_shift(qp));
iowrite32(val, qp->rq_db);
}
@@ -2094,8 +2057,8 @@
break;
default:
ibwc->status = IB_WC_GENERAL_ERR;
- ocrdma_err("%s() invalid opcode received = 0x%x\n",
- __func__, hdr->cw & OCRDMA_WQE_OPCODE_MASK);
+ pr_err("%s() invalid opcode received = 0x%x\n",
+ __func__, hdr->cw & OCRDMA_WQE_OPCODE_MASK);
break;
};
}
diff --git a/drivers/infiniband/hw/qib/Kconfig b/drivers/infiniband/hw/qib/Kconfig
index 1e603a3..d03ca4c 100644
--- a/drivers/infiniband/hw/qib/Kconfig
+++ b/drivers/infiniband/hw/qib/Kconfig
@@ -5,3 +5,11 @@
This is a low-level driver for Intel PCIe QLE InfiniBand host
channel adapters. This driver does not support the Intel
HyperTransport card (model QHT7140).
+
+config INFINIBAND_QIB_DCA
+ bool "QIB DCA support"
+ depends on INFINIBAND_QIB && DCA && SMP && GENERIC_HARDIRQS && !(INFINIBAND_QIB=y && DCA=m)
+ default y
+ ---help---
+ Setting this enables DCA support on some Intel chip sets
+ with the iba7322 HCA.
diff --git a/drivers/infiniband/hw/qib/Makefile b/drivers/infiniband/hw/qib/Makefile
index f12d7bb..57f8103 100644
--- a/drivers/infiniband/hw/qib/Makefile
+++ b/drivers/infiniband/hw/qib/Makefile
@@ -13,3 +13,4 @@
ib_qib-$(CONFIG_X86_64) += qib_wc_x86_64.o
ib_qib-$(CONFIG_PPC64) += qib_wc_ppc64.o
+ib_qib-$(CONFIG_DEBUG_FS) += qib_debugfs.o
diff --git a/drivers/infiniband/hw/qib/qib.h b/drivers/infiniband/hw/qib/qib.h
index 4d11575..4a9af79 100644
--- a/drivers/infiniband/hw/qib/qib.h
+++ b/drivers/infiniband/hw/qib/qib.h
@@ -1,7 +1,7 @@
#ifndef _QIB_KERNEL_H
#define _QIB_KERNEL_H
/*
- * Copyright (c) 2012 Intel Corporation. All rights reserved.
+ * Copyright (c) 2012, 2013 Intel Corporation. All rights reserved.
* Copyright (c) 2006 - 2012 QLogic Corporation. All rights reserved.
* Copyright (c) 2003, 2004, 2005, 2006 PathScale, Inc. All rights reserved.
*
@@ -51,6 +51,7 @@
#include <linux/completion.h>
#include <linux/kref.h>
#include <linux/sched.h>
+#include <linux/kthread.h>
#include "qib_common.h"
#include "qib_verbs.h"
@@ -114,6 +115,11 @@
/*
* Below contains all data related to a single context (formerly called port).
*/
+
+#ifdef CONFIG_DEBUG_FS
+struct qib_opcode_stats_perctx;
+#endif
+
struct qib_ctxtdata {
void **rcvegrbuf;
dma_addr_t *rcvegrbuf_phys;
@@ -154,6 +160,8 @@
*/
/* instead of calculating it */
unsigned ctxt;
+ /* local node of context */
+ int node_id;
/* non-zero if ctxt is being shared. */
u16 subctxt_cnt;
/* non-zero if ctxt is being shared. */
@@ -222,12 +230,15 @@
u8 redirect_seq_cnt;
/* ctxt rcvhdrq head offset */
u32 head;
- u32 pkt_count;
/* lookaside fields */
struct qib_qp *lookaside_qp;
u32 lookaside_qpn;
/* QPs waiting for context processing */
struct list_head qp_wait_list;
+#ifdef CONFIG_DEBUG_FS
+ /* verbs stats per CTX */
+ struct qib_opcode_stats_perctx *opstats;
+#endif
};
struct qib_sge_state;
@@ -428,9 +439,19 @@
#define ACTIVITY_TIMER 5
#define MAX_NAME_SIZE 64
+
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+struct qib_irq_notify;
+#endif
+
struct qib_msix_entry {
struct msix_entry msix;
void *arg;
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ int dca;
+ int rcv;
+ struct qib_irq_notify *notifier;
+#endif
char name[MAX_NAME_SIZE];
cpumask_var_t mask;
};
@@ -828,6 +849,9 @@
struct qib_ctxtdata *);
void (*f_writescratch)(struct qib_devdata *, u32);
int (*f_tempsense_rd)(struct qib_devdata *, int regnum);
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ int (*f_notify_dca)(struct qib_devdata *, unsigned long event);
+#endif
char *boardname; /* human readable board info */
@@ -1075,6 +1099,10 @@
u16 psxmitwait_check_rate;
/* high volume overflow errors defered to tasklet */
struct tasklet_struct error_tasklet;
+ /* per device cq worker */
+ struct kthread_worker *worker;
+
+ int assigned_node_id; /* NUMA node closest to HCA */
};
/* hol_state values */
@@ -1154,7 +1182,7 @@
int qib_setup_eagerbufs(struct qib_ctxtdata *);
void qib_set_ctxtcnt(struct qib_devdata *);
int qib_create_ctxts(struct qib_devdata *dd);
-struct qib_ctxtdata *qib_create_ctxtdata(struct qib_pportdata *, u32);
+struct qib_ctxtdata *qib_create_ctxtdata(struct qib_pportdata *, u32, int);
void qib_init_pportdata(struct qib_pportdata *, struct qib_devdata *, u8, u8);
void qib_free_ctxtdata(struct qib_devdata *, struct qib_ctxtdata *);
@@ -1320,7 +1348,7 @@
return ppd->sdma_state.current_state == qib_sdma_state_s99_running;
}
int qib_sdma_running(struct qib_pportdata *);
-
+void dump_sdma_state(struct qib_pportdata *ppd);
void __qib_sdma_process_event(struct qib_pportdata *, enum qib_sdma_events);
void qib_sdma_process_event(struct qib_pportdata *, enum qib_sdma_events);
@@ -1445,6 +1473,7 @@
extern unsigned qib_sdma_fetch_arb;
extern unsigned qib_compat_ddr_negotiate;
extern int qib_special_trigger;
+extern unsigned qib_numa_aware;
extern struct mutex qib_mutex;
@@ -1474,27 +1503,23 @@
* first to avoid possible serial port delays from printk.
*/
#define qib_early_err(dev, fmt, ...) \
- do { \
- dev_err(dev, fmt, ##__VA_ARGS__); \
- } while (0)
+ dev_err(dev, fmt, ##__VA_ARGS__)
#define qib_dev_err(dd, fmt, ...) \
- do { \
- dev_err(&(dd)->pcidev->dev, "%s: " fmt, \
- qib_get_unit_name((dd)->unit), ##__VA_ARGS__); \
- } while (0)
+ dev_err(&(dd)->pcidev->dev, "%s: " fmt, \
+ qib_get_unit_name((dd)->unit), ##__VA_ARGS__)
+
+#define qib_dev_warn(dd, fmt, ...) \
+ dev_warn(&(dd)->pcidev->dev, "%s: " fmt, \
+ qib_get_unit_name((dd)->unit), ##__VA_ARGS__)
#define qib_dev_porterr(dd, port, fmt, ...) \
- do { \
- dev_err(&(dd)->pcidev->dev, "%s: IB%u:%u " fmt, \
- qib_get_unit_name((dd)->unit), (dd)->unit, (port), \
- ##__VA_ARGS__); \
- } while (0)
+ dev_err(&(dd)->pcidev->dev, "%s: IB%u:%u " fmt, \
+ qib_get_unit_name((dd)->unit), (dd)->unit, (port), \
+ ##__VA_ARGS__)
#define qib_devinfo(pcidev, fmt, ...) \
- do { \
- dev_info(&(pcidev)->dev, fmt, ##__VA_ARGS__); \
- } while (0)
+ dev_info(&(pcidev)->dev, fmt, ##__VA_ARGS__)
/*
* this is used for formatting hw error messages...
diff --git a/drivers/infiniband/hw/qib/qib_common.h b/drivers/infiniband/hw/qib/qib_common.h
index d39e018..4f255b7 100644
--- a/drivers/infiniband/hw/qib/qib_common.h
+++ b/drivers/infiniband/hw/qib/qib_common.h
@@ -279,7 +279,7 @@
* may not be implemented; the user code must deal with this if it
* cares, or it must abort after initialization reports the difference.
*/
-#define QIB_USER_SWMINOR 11
+#define QIB_USER_SWMINOR 12
#define QIB_USER_SWVERSION ((QIB_USER_SWMAJOR << 16) | QIB_USER_SWMINOR)
diff --git a/drivers/infiniband/hw/qib/qib_cq.c b/drivers/infiniband/hw/qib/qib_cq.c
index 5246aa4..ab4e11c 100644
--- a/drivers/infiniband/hw/qib/qib_cq.c
+++ b/drivers/infiniband/hw/qib/qib_cq.c
@@ -1,4 +1,5 @@
/*
+ * Copyright (c) 2013 Intel Corporation. All rights reserved.
* Copyright (c) 2006, 2007, 2008, 2010 QLogic Corporation. All rights reserved.
* Copyright (c) 2005, 2006 PathScale, Inc. All rights reserved.
*
@@ -34,8 +35,10 @@
#include <linux/err.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>
+#include <linux/kthread.h>
#include "qib_verbs.h"
+#include "qib.h"
/**
* qib_cq_enter - add a new entry to the completion queue
@@ -102,13 +105,18 @@
if (cq->notify == IB_CQ_NEXT_COMP ||
(cq->notify == IB_CQ_SOLICITED &&
(solicited || entry->status != IB_WC_SUCCESS))) {
- cq->notify = IB_CQ_NONE;
- cq->triggered++;
+ struct kthread_worker *worker;
/*
* This will cause send_complete() to be called in
* another thread.
*/
- queue_work(qib_cq_wq, &cq->comptask);
+ smp_rmb();
+ worker = cq->dd->worker;
+ if (likely(worker)) {
+ cq->notify = IB_CQ_NONE;
+ cq->triggered++;
+ queue_kthread_work(worker, &cq->comptask);
+ }
}
spin_unlock_irqrestore(&cq->lock, flags);
@@ -163,7 +171,7 @@
return npolled;
}
-static void send_complete(struct work_struct *work)
+static void send_complete(struct kthread_work *work)
{
struct qib_cq *cq = container_of(work, struct qib_cq, comptask);
@@ -287,11 +295,12 @@
* The number of entries should be >= the number requested or return
* an error.
*/
+ cq->dd = dd_from_dev(dev);
cq->ibcq.cqe = entries;
cq->notify = IB_CQ_NONE;
cq->triggered = 0;
spin_lock_init(&cq->lock);
- INIT_WORK(&cq->comptask, send_complete);
+ init_kthread_work(&cq->comptask, send_complete);
wc->head = 0;
wc->tail = 0;
cq->queue = wc;
@@ -323,7 +332,7 @@
struct qib_ibdev *dev = to_idev(ibcq->device);
struct qib_cq *cq = to_icq(ibcq);
- flush_work(&cq->comptask);
+ flush_kthread_work(&cq->comptask);
spin_lock(&dev->n_cqs_lock);
dev->n_cqs_allocated--;
spin_unlock(&dev->n_cqs_lock);
@@ -483,3 +492,49 @@
bail:
return ret;
}
+
+int qib_cq_init(struct qib_devdata *dd)
+{
+ int ret = 0;
+ int cpu;
+ struct task_struct *task;
+
+ if (dd->worker)
+ return 0;
+ dd->worker = kzalloc(sizeof(*dd->worker), GFP_KERNEL);
+ if (!dd->worker)
+ return -ENOMEM;
+ init_kthread_worker(dd->worker);
+ task = kthread_create_on_node(
+ kthread_worker_fn,
+ dd->worker,
+ dd->assigned_node_id,
+ "qib_cq%d", dd->unit);
+ if (IS_ERR(task))
+ goto task_fail;
+ cpu = cpumask_first(cpumask_of_node(dd->assigned_node_id));
+ kthread_bind(task, cpu);
+ wake_up_process(task);
+out:
+ return ret;
+task_fail:
+ ret = PTR_ERR(task);
+ kfree(dd->worker);
+ dd->worker = NULL;
+ goto out;
+}
+
+void qib_cq_exit(struct qib_devdata *dd)
+{
+ struct kthread_worker *worker;
+
+ worker = dd->worker;
+ if (!worker)
+ return;
+ /* blocks future queuing from send_complete() */
+ dd->worker = NULL;
+ smp_wmb();
+ flush_kthread_worker(worker);
+ kthread_stop(worker->task);
+ kfree(worker);
+}
diff --git a/drivers/infiniband/hw/qib/qib_debugfs.c b/drivers/infiniband/hw/qib/qib_debugfs.c
new file mode 100644
index 0000000..799a0c3
--- /dev/null
+++ b/drivers/infiniband/hw/qib/qib_debugfs.c
@@ -0,0 +1,283 @@
+#ifdef CONFIG_DEBUG_FS
+/*
+ * Copyright (c) 2013 Intel Corporation. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include <linux/debugfs.h>
+#include <linux/seq_file.h>
+#include <linux/kernel.h>
+#include <linux/export.h>
+
+#include "qib.h"
+#include "qib_verbs.h"
+#include "qib_debugfs.h"
+
+static struct dentry *qib_dbg_root;
+
+#define DEBUGFS_FILE(name) \
+static const struct seq_operations _##name##_seq_ops = { \
+ .start = _##name##_seq_start, \
+ .next = _##name##_seq_next, \
+ .stop = _##name##_seq_stop, \
+ .show = _##name##_seq_show \
+}; \
+static int _##name##_open(struct inode *inode, struct file *s) \
+{ \
+ struct seq_file *seq; \
+ int ret; \
+ ret = seq_open(s, &_##name##_seq_ops); \
+ if (ret) \
+ return ret; \
+ seq = s->private_data; \
+ seq->private = inode->i_private; \
+ return 0; \
+} \
+static const struct file_operations _##name##_file_ops = { \
+ .owner = THIS_MODULE, \
+ .open = _##name##_open, \
+ .read = seq_read, \
+ .llseek = seq_lseek, \
+ .release = seq_release \
+};
+
+#define DEBUGFS_FILE_CREATE(name) \
+do { \
+ struct dentry *ent; \
+ ent = debugfs_create_file(#name , 0400, ibd->qib_ibdev_dbg, \
+ ibd, &_##name##_file_ops); \
+ if (!ent) \
+ pr_warn("create of " #name " failed\n"); \
+} while (0)
+
+static void *_opcode_stats_seq_start(struct seq_file *s, loff_t *pos)
+{
+ struct qib_opcode_stats_perctx *opstats;
+
+ if (*pos >= ARRAY_SIZE(opstats->stats))
+ return NULL;
+ return pos;
+}
+
+static void *_opcode_stats_seq_next(struct seq_file *s, void *v, loff_t *pos)
+{
+ struct qib_opcode_stats_perctx *opstats;
+
+ ++*pos;
+ if (*pos >= ARRAY_SIZE(opstats->stats))
+ return NULL;
+ return pos;
+}
+
+
+static void _opcode_stats_seq_stop(struct seq_file *s, void *v)
+{
+ /* nothing allocated */
+}
+
+static int _opcode_stats_seq_show(struct seq_file *s, void *v)
+{
+ loff_t *spos = v;
+ loff_t i = *spos, j;
+ u64 n_packets = 0, n_bytes = 0;
+ struct qib_ibdev *ibd = (struct qib_ibdev *)s->private;
+ struct qib_devdata *dd = dd_from_dev(ibd);
+
+ for (j = 0; j < dd->first_user_ctxt; j++) {
+ if (!dd->rcd[j])
+ continue;
+ n_packets += dd->rcd[j]->opstats->stats[i].n_packets;
+ n_bytes += dd->rcd[j]->opstats->stats[i].n_bytes;
+ }
+ if (!n_packets && !n_bytes)
+ return SEQ_SKIP;
+ seq_printf(s, "%02llx %llu/%llu\n", i,
+ (unsigned long long) n_packets,
+ (unsigned long long) n_bytes);
+
+ return 0;
+}
+
+DEBUGFS_FILE(opcode_stats)
+
+static void *_ctx_stats_seq_start(struct seq_file *s, loff_t *pos)
+{
+ struct qib_ibdev *ibd = (struct qib_ibdev *)s->private;
+ struct qib_devdata *dd = dd_from_dev(ibd);
+
+ if (!*pos)
+ return SEQ_START_TOKEN;
+ if (*pos >= dd->first_user_ctxt)
+ return NULL;
+ return pos;
+}
+
+static void *_ctx_stats_seq_next(struct seq_file *s, void *v, loff_t *pos)
+{
+ struct qib_ibdev *ibd = (struct qib_ibdev *)s->private;
+ struct qib_devdata *dd = dd_from_dev(ibd);
+
+ if (v == SEQ_START_TOKEN)
+ return pos;
+
+ ++*pos;
+ if (*pos >= dd->first_user_ctxt)
+ return NULL;
+ return pos;
+}
+
+static void _ctx_stats_seq_stop(struct seq_file *s, void *v)
+{
+ /* nothing allocated */
+}
+
+static int _ctx_stats_seq_show(struct seq_file *s, void *v)
+{
+ loff_t *spos;
+ loff_t i, j;
+ u64 n_packets = 0;
+ struct qib_ibdev *ibd = (struct qib_ibdev *)s->private;
+ struct qib_devdata *dd = dd_from_dev(ibd);
+
+ if (v == SEQ_START_TOKEN) {
+ seq_puts(s, "Ctx:npkts\n");
+ return 0;
+ }
+
+ spos = v;
+ i = *spos;
+
+ if (!dd->rcd[i])
+ return SEQ_SKIP;
+
+ for (j = 0; j < ARRAY_SIZE(dd->rcd[i]->opstats->stats); j++)
+ n_packets += dd->rcd[i]->opstats->stats[j].n_packets;
+
+ if (!n_packets)
+ return SEQ_SKIP;
+
+ seq_printf(s, " %llu:%llu\n", i, n_packets);
+ return 0;
+}
+
+DEBUGFS_FILE(ctx_stats)
+
+static void *_qp_stats_seq_start(struct seq_file *s, loff_t *pos)
+{
+ struct qib_qp_iter *iter;
+ loff_t n = *pos;
+
+ iter = qib_qp_iter_init(s->private);
+ if (!iter)
+ return NULL;
+
+ while (n--) {
+ if (qib_qp_iter_next(iter)) {
+ kfree(iter);
+ return NULL;
+ }
+ }
+
+ return iter;
+}
+
+static void *_qp_stats_seq_next(struct seq_file *s, void *iter_ptr,
+ loff_t *pos)
+{
+ struct qib_qp_iter *iter = iter_ptr;
+
+ (*pos)++;
+
+ if (qib_qp_iter_next(iter)) {
+ kfree(iter);
+ return NULL;
+ }
+
+ return iter;
+}
+
+static void _qp_stats_seq_stop(struct seq_file *s, void *iter_ptr)
+{
+ /* nothing for now */
+}
+
+static int _qp_stats_seq_show(struct seq_file *s, void *iter_ptr)
+{
+ struct qib_qp_iter *iter = iter_ptr;
+
+ if (!iter)
+ return 0;
+
+ qib_qp_iter_print(s, iter);
+
+ return 0;
+}
+
+DEBUGFS_FILE(qp_stats)
+
+void qib_dbg_ibdev_init(struct qib_ibdev *ibd)
+{
+ char name[10];
+
+ snprintf(name, sizeof(name), "qib%d", dd_from_dev(ibd)->unit);
+ ibd->qib_ibdev_dbg = debugfs_create_dir(name, qib_dbg_root);
+ if (!ibd->qib_ibdev_dbg) {
+ pr_warn("create of %s failed\n", name);
+ return;
+ }
+ DEBUGFS_FILE_CREATE(opcode_stats);
+ DEBUGFS_FILE_CREATE(ctx_stats);
+ DEBUGFS_FILE_CREATE(qp_stats);
+ return;
+}
+
+void qib_dbg_ibdev_exit(struct qib_ibdev *ibd)
+{
+ if (!qib_dbg_root)
+ goto out;
+ debugfs_remove_recursive(ibd->qib_ibdev_dbg);
+out:
+ ibd->qib_ibdev_dbg = NULL;
+}
+
+void qib_dbg_init(void)
+{
+ qib_dbg_root = debugfs_create_dir(QIB_DRV_NAME, NULL);
+ if (!qib_dbg_root)
+ pr_warn("init of debugfs failed\n");
+}
+
+void qib_dbg_exit(void)
+{
+ debugfs_remove_recursive(qib_dbg_root);
+ qib_dbg_root = NULL;
+}
+
+#endif
+
diff --git a/drivers/infiniband/hw/qib/qib_debugfs.h b/drivers/infiniband/hw/qib/qib_debugfs.h
new file mode 100644
index 0000000..7ae983a
--- /dev/null
+++ b/drivers/infiniband/hw/qib/qib_debugfs.h
@@ -0,0 +1,45 @@
+#ifndef _QIB_DEBUGFS_H
+#define _QIB_DEBUGFS_H
+
+#ifdef CONFIG_DEBUG_FS
+/*
+ * Copyright (c) 2013 Intel Corporation. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+struct qib_ibdev;
+void qib_dbg_ibdev_init(struct qib_ibdev *ibd);
+void qib_dbg_ibdev_exit(struct qib_ibdev *ibd);
+void qib_dbg_init(void);
+void qib_dbg_exit(void);
+
+#endif
+
+#endif /* _QIB_DEBUGFS_H */
diff --git a/drivers/infiniband/hw/qib/qib_driver.c b/drivers/infiniband/hw/qib/qib_driver.c
index 2160924..5bee08f 100644
--- a/drivers/infiniband/hw/qib/qib_driver.c
+++ b/drivers/infiniband/hw/qib/qib_driver.c
@@ -558,7 +558,6 @@
}
rcd->head = l;
- rcd->pkt_count += i;
/*
* Iterate over all QPs waiting to respond.
diff --git a/drivers/infiniband/hw/qib/qib_file_ops.c b/drivers/infiniband/hw/qib/qib_file_ops.c
index b56c942..df3808a 100644
--- a/drivers/infiniband/hw/qib/qib_file_ops.c
+++ b/drivers/infiniband/hw/qib/qib_file_ops.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2012 Intel Corporation. All rights reserved.
+ * Copyright (c) 2012, 2013 Intel Corporation. All rights reserved.
* Copyright (c) 2006 - 2012 QLogic Corporation. All rights reserved.
* Copyright (c) 2003, 2004, 2005, 2006 PathScale, Inc. All rights reserved.
*
@@ -1155,6 +1155,49 @@
return pollflag;
}
+static void assign_ctxt_affinity(struct file *fp, struct qib_devdata *dd)
+{
+ struct qib_filedata *fd = fp->private_data;
+ const unsigned int weight = cpumask_weight(¤t->cpus_allowed);
+ const struct cpumask *local_mask = cpumask_of_pcibus(dd->pcidev->bus);
+ int local_cpu;
+
+ /*
+ * If process has NOT already set it's affinity, select and
+ * reserve a processor for it on the local NUMA node.
+ */
+ if ((weight >= qib_cpulist_count) &&
+ (cpumask_weight(local_mask) <= qib_cpulist_count)) {
+ for_each_cpu(local_cpu, local_mask)
+ if (!test_and_set_bit(local_cpu, qib_cpulist)) {
+ fd->rec_cpu_num = local_cpu;
+ return;
+ }
+ }
+
+ /*
+ * If process has NOT already set it's affinity, select and
+ * reserve a processor for it, as a rendevous for all
+ * users of the driver. If they don't actually later
+ * set affinity to this cpu, or set it to some other cpu,
+ * it just means that sooner or later we don't recommend
+ * a cpu, and let the scheduler do it's best.
+ */
+ if (weight >= qib_cpulist_count) {
+ int cpu;
+ cpu = find_first_zero_bit(qib_cpulist,
+ qib_cpulist_count);
+ if (cpu == qib_cpulist_count)
+ qib_dev_err(dd,
+ "no cpus avail for affinity PID %u\n",
+ current->pid);
+ else {
+ __set_bit(cpu, qib_cpulist);
+ fd->rec_cpu_num = cpu;
+ }
+ }
+}
+
/*
* Check that userland and driver are compatible for subcontexts.
*/
@@ -1259,12 +1302,20 @@
static int setup_ctxt(struct qib_pportdata *ppd, int ctxt,
struct file *fp, const struct qib_user_info *uinfo)
{
+ struct qib_filedata *fd = fp->private_data;
struct qib_devdata *dd = ppd->dd;
struct qib_ctxtdata *rcd;
void *ptmp = NULL;
int ret;
+ int numa_id;
- rcd = qib_create_ctxtdata(ppd, ctxt);
+ assign_ctxt_affinity(fp, dd);
+
+ numa_id = qib_numa_aware ? ((fd->rec_cpu_num != -1) ?
+ cpu_to_node(fd->rec_cpu_num) :
+ numa_node_id()) : dd->assigned_node_id;
+
+ rcd = qib_create_ctxtdata(ppd, ctxt, numa_id);
/*
* Allocate memory for use in qib_tid_update() at open to
@@ -1296,6 +1347,9 @@
goto bail;
bailerr:
+ if (fd->rec_cpu_num != -1)
+ __clear_bit(fd->rec_cpu_num, qib_cpulist);
+
dd->rcd[ctxt] = NULL;
kfree(rcd);
kfree(ptmp);
@@ -1485,6 +1539,57 @@
return fp->private_data ? 0 : -ENOMEM;
}
+static int find_hca(unsigned int cpu, int *unit)
+{
+ int ret = 0, devmax, npresent, nup, ndev;
+
+ *unit = -1;
+
+ devmax = qib_count_units(&npresent, &nup);
+ if (!npresent) {
+ ret = -ENXIO;
+ goto done;
+ }
+ if (!nup) {
+ ret = -ENETDOWN;
+ goto done;
+ }
+ for (ndev = 0; ndev < devmax; ndev++) {
+ struct qib_devdata *dd = qib_lookup(ndev);
+ if (dd) {
+ if (pcibus_to_node(dd->pcidev->bus) < 0) {
+ ret = -EINVAL;
+ goto done;
+ }
+ if (cpu_to_node(cpu) ==
+ pcibus_to_node(dd->pcidev->bus)) {
+ *unit = ndev;
+ goto done;
+ }
+ }
+ }
+done:
+ return ret;
+}
+
+static int do_qib_user_sdma_queue_create(struct file *fp)
+{
+ struct qib_filedata *fd = fp->private_data;
+ struct qib_ctxtdata *rcd = fd->rcd;
+ struct qib_devdata *dd = rcd->dd;
+
+ if (dd->flags & QIB_HAS_SEND_DMA)
+
+ fd->pq = qib_user_sdma_queue_create(&dd->pcidev->dev,
+ dd->unit,
+ rcd->ctxt,
+ fd->subctxt);
+ if (!fd->pq)
+ return -ENOMEM;
+
+ return 0;
+}
+
/*
* Get ctxt early, so can set affinity prior to memory allocation.
*/
@@ -1517,61 +1622,36 @@
if (qib_compatible_subctxts(swmajor, swminor) &&
uinfo->spu_subctxt_cnt) {
ret = find_shared_ctxt(fp, uinfo);
- if (ret) {
- if (ret > 0)
- ret = 0;
- goto done_chk_sdma;
+ if (ret > 0) {
+ ret = do_qib_user_sdma_queue_create(fp);
+ if (!ret)
+ assign_ctxt_affinity(fp, (ctxt_fp(fp))->dd);
+ goto done_ok;
}
}
i_minor = iminor(file_inode(fp)) - QIB_USER_MINOR_BASE;
if (i_minor)
ret = find_free_ctxt(i_minor - 1, fp, uinfo);
- else
+ else {
+ int unit;
+ const unsigned int cpu = cpumask_first(¤t->cpus_allowed);
+ const unsigned int weight =
+ cpumask_weight(¤t->cpus_allowed);
+
+ if (weight == 1 && !test_bit(cpu, qib_cpulist))
+ if (!find_hca(cpu, &unit) && unit >= 0)
+ if (!find_free_ctxt(unit, fp, uinfo)) {
+ ret = 0;
+ goto done_chk_sdma;
+ }
ret = get_a_ctxt(fp, uinfo, alg);
-
-done_chk_sdma:
- if (!ret) {
- struct qib_filedata *fd = fp->private_data;
- const struct qib_ctxtdata *rcd = fd->rcd;
- const struct qib_devdata *dd = rcd->dd;
- unsigned int weight;
-
- if (dd->flags & QIB_HAS_SEND_DMA) {
- fd->pq = qib_user_sdma_queue_create(&dd->pcidev->dev,
- dd->unit,
- rcd->ctxt,
- fd->subctxt);
- if (!fd->pq)
- ret = -ENOMEM;
- }
-
- /*
- * If process has NOT already set it's affinity, select and
- * reserve a processor for it, as a rendezvous for all
- * users of the driver. If they don't actually later
- * set affinity to this cpu, or set it to some other cpu,
- * it just means that sooner or later we don't recommend
- * a cpu, and let the scheduler do it's best.
- */
- weight = cpumask_weight(tsk_cpus_allowed(current));
- if (!ret && weight >= qib_cpulist_count) {
- int cpu;
- cpu = find_first_zero_bit(qib_cpulist,
- qib_cpulist_count);
- if (cpu != qib_cpulist_count) {
- __set_bit(cpu, qib_cpulist);
- fd->rec_cpu_num = cpu;
- }
- } else if (weight == 1 &&
- test_bit(cpumask_first(tsk_cpus_allowed(current)),
- qib_cpulist))
- qib_devinfo(dd->pcidev,
- "%s PID %u affinity set to cpu %d; already allocated\n",
- current->comm, current->pid,
- cpumask_first(tsk_cpus_allowed(current)));
}
+done_chk_sdma:
+ if (!ret)
+ ret = do_qib_user_sdma_queue_create(fp);
+done_ok:
mutex_unlock(&qib_mutex);
done:
diff --git a/drivers/infiniband/hw/qib/qib_iba6120.c b/drivers/infiniband/hw/qib/qib_iba6120.c
index 0232ae5..84e593d 100644
--- a/drivers/infiniband/hw/qib/qib_iba6120.c
+++ b/drivers/infiniband/hw/qib/qib_iba6120.c
@@ -3464,6 +3464,13 @@
return -ENXIO;
}
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+static int qib_6120_notify_dca(struct qib_devdata *dd, unsigned long event)
+{
+ return 0;
+}
+#endif
+
/* Dummy function, as 6120 boards never disable EEPROM Write */
static int qib_6120_eeprom_wen(struct qib_devdata *dd, int wen)
{
@@ -3539,6 +3546,9 @@
dd->f_xgxs_reset = qib_6120_xgxs_reset;
dd->f_writescratch = writescratch;
dd->f_tempsense_rd = qib_6120_tempsense_rd;
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ dd->f_notify_dca = qib_6120_notify_dca;
+#endif
/*
* Do remaining pcie setup and save pcie values in dd.
* Any error printing is already done by the init code.
diff --git a/drivers/infiniband/hw/qib/qib_iba7220.c b/drivers/infiniband/hw/qib/qib_iba7220.c
index 64d0ecb..454c2e7 100644
--- a/drivers/infiniband/hw/qib/qib_iba7220.c
+++ b/drivers/infiniband/hw/qib/qib_iba7220.c
@@ -4513,6 +4513,13 @@
return ret;
}
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+static int qib_7220_notify_dca(struct qib_devdata *dd, unsigned long event)
+{
+ return 0;
+}
+#endif
+
/* Dummy function, as 7220 boards never disable EEPROM Write */
static int qib_7220_eeprom_wen(struct qib_devdata *dd, int wen)
{
@@ -4587,6 +4594,9 @@
dd->f_xgxs_reset = qib_7220_xgxs_reset;
dd->f_writescratch = writescratch;
dd->f_tempsense_rd = qib_7220_tempsense_rd;
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ dd->f_notify_dca = qib_7220_notify_dca;
+#endif
/*
* Do remaining pcie setup and save pcie values in dd.
* Any error printing is already done by the init code.
diff --git a/drivers/infiniband/hw/qib/qib_iba7322.c b/drivers/infiniband/hw/qib/qib_iba7322.c
index 3f6b21e..21e8b09 100644
--- a/drivers/infiniband/hw/qib/qib_iba7322.c
+++ b/drivers/infiniband/hw/qib/qib_iba7322.c
@@ -44,6 +44,9 @@
#include <linux/module.h>
#include <rdma/ib_verbs.h>
#include <rdma/ib_smi.h>
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+#include <linux/dca.h>
+#endif
#include "qib.h"
#include "qib_7322_regs.h"
@@ -80,6 +83,7 @@
static void serdes_7322_los_enable(struct qib_pportdata *, int);
static int serdes_7322_init_old(struct qib_pportdata *);
static int serdes_7322_init_new(struct qib_pportdata *);
+static void dump_sdma_7322_state(struct qib_pportdata *);
#define BMASK(msb, lsb) (((1 << ((msb) + 1 - (lsb))) - 1) << (lsb))
@@ -519,6 +523,14 @@
[0x17] = IB_PHYSPORTSTATE_CFG_TRAIN
};
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+struct qib_irq_notify {
+ int rcv;
+ void *arg;
+ struct irq_affinity_notify notify;
+};
+#endif
+
struct qib_chip_specific {
u64 __iomem *cregbase;
u64 *cntrs;
@@ -546,6 +558,12 @@
u32 lastbuf_for_pio;
u32 stay_in_freeze;
u32 recovery_ports_initted;
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ u32 dca_ctrl;
+ int rhdr_cpu[18];
+ int sdma_cpu[2];
+ u64 dca_rcvhdr_ctrl[5]; /* B, C, D, E, F */
+#endif
struct qib_msix_entry *msix_entries;
unsigned long *sendchkenable;
unsigned long *sendgrhchk;
@@ -573,7 +591,7 @@
static void write_tx_serdes_param(struct qib_pportdata *, struct txdds_ent *);
#define TXDDS_TABLE_SZ 16 /* number of entries per speed in onchip table */
-#define TXDDS_EXTRA_SZ 13 /* number of extra tx settings entries */
+#define TXDDS_EXTRA_SZ 18 /* number of extra tx settings entries */
#define TXDDS_MFG_SZ 2 /* number of mfg tx settings entries */
#define SERDES_CHANS 4 /* yes, it's obvious, but one less magic number */
@@ -635,6 +653,7 @@
u8 ibmalfusesnap;
struct qib_qsfp_data qsfp_data;
char epmsgbuf[192]; /* for port error interrupt msg buffer */
+ char sdmamsgbuf[192]; /* for per-port sdma error messages */
};
static struct {
@@ -642,28 +661,76 @@
irq_handler_t handler;
int lsb;
int port; /* 0 if not port-specific, else port # */
+ int dca;
} irq_table[] = {
- { "", qib_7322intr, -1, 0 },
+ { "", qib_7322intr, -1, 0, 0 },
{ " (buf avail)", qib_7322bufavail,
- SYM_LSB(IntStatus, SendBufAvail), 0 },
+ SYM_LSB(IntStatus, SendBufAvail), 0, 0},
{ " (sdma 0)", sdma_intr,
- SYM_LSB(IntStatus, SDmaInt_0), 1 },
+ SYM_LSB(IntStatus, SDmaInt_0), 1, 1 },
{ " (sdma 1)", sdma_intr,
- SYM_LSB(IntStatus, SDmaInt_1), 2 },
+ SYM_LSB(IntStatus, SDmaInt_1), 2, 1 },
{ " (sdmaI 0)", sdma_idle_intr,
- SYM_LSB(IntStatus, SDmaIdleInt_0), 1 },
+ SYM_LSB(IntStatus, SDmaIdleInt_0), 1, 1},
{ " (sdmaI 1)", sdma_idle_intr,
- SYM_LSB(IntStatus, SDmaIdleInt_1), 2 },
+ SYM_LSB(IntStatus, SDmaIdleInt_1), 2, 1},
{ " (sdmaP 0)", sdma_progress_intr,
- SYM_LSB(IntStatus, SDmaProgressInt_0), 1 },
+ SYM_LSB(IntStatus, SDmaProgressInt_0), 1, 1 },
{ " (sdmaP 1)", sdma_progress_intr,
- SYM_LSB(IntStatus, SDmaProgressInt_1), 2 },
+ SYM_LSB(IntStatus, SDmaProgressInt_1), 2, 1 },
{ " (sdmaC 0)", sdma_cleanup_intr,
- SYM_LSB(IntStatus, SDmaCleanupDone_0), 1 },
+ SYM_LSB(IntStatus, SDmaCleanupDone_0), 1, 0 },
{ " (sdmaC 1)", sdma_cleanup_intr,
- SYM_LSB(IntStatus, SDmaCleanupDone_1), 2 },
+ SYM_LSB(IntStatus, SDmaCleanupDone_1), 2 , 0},
};
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+
+static const struct dca_reg_map {
+ int shadow_inx;
+ int lsb;
+ u64 mask;
+ u16 regno;
+} dca_rcvhdr_reg_map[] = {
+ { 0, SYM_LSB(DCACtrlB, RcvHdrq0DCAOPH),
+ ~SYM_MASK(DCACtrlB, RcvHdrq0DCAOPH) , KREG_IDX(DCACtrlB) },
+ { 0, SYM_LSB(DCACtrlB, RcvHdrq1DCAOPH),
+ ~SYM_MASK(DCACtrlB, RcvHdrq1DCAOPH) , KREG_IDX(DCACtrlB) },
+ { 0, SYM_LSB(DCACtrlB, RcvHdrq2DCAOPH),
+ ~SYM_MASK(DCACtrlB, RcvHdrq2DCAOPH) , KREG_IDX(DCACtrlB) },
+ { 0, SYM_LSB(DCACtrlB, RcvHdrq3DCAOPH),
+ ~SYM_MASK(DCACtrlB, RcvHdrq3DCAOPH) , KREG_IDX(DCACtrlB) },
+ { 1, SYM_LSB(DCACtrlC, RcvHdrq4DCAOPH),
+ ~SYM_MASK(DCACtrlC, RcvHdrq4DCAOPH) , KREG_IDX(DCACtrlC) },
+ { 1, SYM_LSB(DCACtrlC, RcvHdrq5DCAOPH),
+ ~SYM_MASK(DCACtrlC, RcvHdrq5DCAOPH) , KREG_IDX(DCACtrlC) },
+ { 1, SYM_LSB(DCACtrlC, RcvHdrq6DCAOPH),
+ ~SYM_MASK(DCACtrlC, RcvHdrq6DCAOPH) , KREG_IDX(DCACtrlC) },
+ { 1, SYM_LSB(DCACtrlC, RcvHdrq7DCAOPH),
+ ~SYM_MASK(DCACtrlC, RcvHdrq7DCAOPH) , KREG_IDX(DCACtrlC) },
+ { 2, SYM_LSB(DCACtrlD, RcvHdrq8DCAOPH),
+ ~SYM_MASK(DCACtrlD, RcvHdrq8DCAOPH) , KREG_IDX(DCACtrlD) },
+ { 2, SYM_LSB(DCACtrlD, RcvHdrq9DCAOPH),
+ ~SYM_MASK(DCACtrlD, RcvHdrq9DCAOPH) , KREG_IDX(DCACtrlD) },
+ { 2, SYM_LSB(DCACtrlD, RcvHdrq10DCAOPH),
+ ~SYM_MASK(DCACtrlD, RcvHdrq10DCAOPH) , KREG_IDX(DCACtrlD) },
+ { 2, SYM_LSB(DCACtrlD, RcvHdrq11DCAOPH),
+ ~SYM_MASK(DCACtrlD, RcvHdrq11DCAOPH) , KREG_IDX(DCACtrlD) },
+ { 3, SYM_LSB(DCACtrlE, RcvHdrq12DCAOPH),
+ ~SYM_MASK(DCACtrlE, RcvHdrq12DCAOPH) , KREG_IDX(DCACtrlE) },
+ { 3, SYM_LSB(DCACtrlE, RcvHdrq13DCAOPH),
+ ~SYM_MASK(DCACtrlE, RcvHdrq13DCAOPH) , KREG_IDX(DCACtrlE) },
+ { 3, SYM_LSB(DCACtrlE, RcvHdrq14DCAOPH),
+ ~SYM_MASK(DCACtrlE, RcvHdrq14DCAOPH) , KREG_IDX(DCACtrlE) },
+ { 3, SYM_LSB(DCACtrlE, RcvHdrq15DCAOPH),
+ ~SYM_MASK(DCACtrlE, RcvHdrq15DCAOPH) , KREG_IDX(DCACtrlE) },
+ { 4, SYM_LSB(DCACtrlF, RcvHdrq16DCAOPH),
+ ~SYM_MASK(DCACtrlF, RcvHdrq16DCAOPH) , KREG_IDX(DCACtrlF) },
+ { 4, SYM_LSB(DCACtrlF, RcvHdrq17DCAOPH),
+ ~SYM_MASK(DCACtrlF, RcvHdrq17DCAOPH) , KREG_IDX(DCACtrlF) },
+};
+#endif
+
/* ibcctrl bits */
#define QLOGIC_IB_IBCC_LINKINITCMD_DISABLE 1
/* cycle through TS1/TS2 till OK */
@@ -686,6 +753,13 @@
static void setup_7322_link_recovery(struct qib_pportdata *, u32);
static void check_7322_rxe_status(struct qib_pportdata *);
static u32 __iomem *qib_7322_getsendbuf(struct qib_pportdata *, u64, u32 *);
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+static void qib_setup_dca(struct qib_devdata *dd);
+static void setup_dca_notifier(struct qib_devdata *dd,
+ struct qib_msix_entry *m);
+static void reset_dca_notifier(struct qib_devdata *dd,
+ struct qib_msix_entry *m);
+#endif
/**
* qib_read_ureg32 - read 32-bit virtualized per-context register
@@ -1529,6 +1603,15 @@
spin_lock_irqsave(&ppd->sdma_lock, flags);
+ if (errs != QIB_E_P_SDMAHALT) {
+ /* SDMA errors have QIB_E_P_SDMAHALT and another bit set */
+ qib_dev_porterr(dd, ppd->port,
+ "SDMA %s 0x%016llx %s\n",
+ qib_sdma_state_names[ppd->sdma_state.current_state],
+ errs, ppd->cpspec->sdmamsgbuf);
+ dump_sdma_7322_state(ppd);
+ }
+
switch (ppd->sdma_state.current_state) {
case qib_sdma_state_s00_hw_down:
break;
@@ -2084,6 +2167,29 @@
qib_dev_err(dd, "%s hardware error\n", msg);
+ if (hwerrs &
+ (SYM_MASK(HwErrMask, SDmaMemReadErrMask_0) |
+ SYM_MASK(HwErrMask, SDmaMemReadErrMask_1))) {
+ int pidx = 0;
+ int err;
+ unsigned long flags;
+ struct qib_pportdata *ppd = dd->pport;
+ for (; pidx < dd->num_pports; ++pidx, ppd++) {
+ err = 0;
+ if (pidx == 0 && (hwerrs &
+ SYM_MASK(HwErrMask, SDmaMemReadErrMask_0)))
+ err++;
+ if (pidx == 1 && (hwerrs &
+ SYM_MASK(HwErrMask, SDmaMemReadErrMask_1)))
+ err++;
+ if (err) {
+ spin_lock_irqsave(&ppd->sdma_lock, flags);
+ dump_sdma_7322_state(ppd);
+ spin_unlock_irqrestore(&ppd->sdma_lock, flags);
+ }
+ }
+ }
+
if (isfatal && !dd->diag_client) {
qib_dev_err(dd,
"Fatal Hardware Error, no longer usable, SN %.16s\n",
@@ -2558,6 +2664,162 @@
qib_write_kreg_port(ppd, krp_rcvpktledcnt, ledblink);
}
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+
+static int qib_7322_notify_dca(struct qib_devdata *dd, unsigned long event)
+{
+ switch (event) {
+ case DCA_PROVIDER_ADD:
+ if (dd->flags & QIB_DCA_ENABLED)
+ break;
+ if (!dca_add_requester(&dd->pcidev->dev)) {
+ qib_devinfo(dd->pcidev, "DCA enabled\n");
+ dd->flags |= QIB_DCA_ENABLED;
+ qib_setup_dca(dd);
+ }
+ break;
+ case DCA_PROVIDER_REMOVE:
+ if (dd->flags & QIB_DCA_ENABLED) {
+ dca_remove_requester(&dd->pcidev->dev);
+ dd->flags &= ~QIB_DCA_ENABLED;
+ dd->cspec->dca_ctrl = 0;
+ qib_write_kreg(dd, KREG_IDX(DCACtrlA),
+ dd->cspec->dca_ctrl);
+ }
+ break;
+ }
+ return 0;
+}
+
+static void qib_update_rhdrq_dca(struct qib_ctxtdata *rcd, int cpu)
+{
+ struct qib_devdata *dd = rcd->dd;
+ struct qib_chip_specific *cspec = dd->cspec;
+
+ if (!(dd->flags & QIB_DCA_ENABLED))
+ return;
+ if (cspec->rhdr_cpu[rcd->ctxt] != cpu) {
+ const struct dca_reg_map *rmp;
+
+ cspec->rhdr_cpu[rcd->ctxt] = cpu;
+ rmp = &dca_rcvhdr_reg_map[rcd->ctxt];
+ cspec->dca_rcvhdr_ctrl[rmp->shadow_inx] &= rmp->mask;
+ cspec->dca_rcvhdr_ctrl[rmp->shadow_inx] |=
+ (u64) dca3_get_tag(&dd->pcidev->dev, cpu) << rmp->lsb;
+ qib_devinfo(dd->pcidev,
+ "Ctxt %d cpu %d dca %llx\n", rcd->ctxt, cpu,
+ (long long) cspec->dca_rcvhdr_ctrl[rmp->shadow_inx]);
+ qib_write_kreg(dd, rmp->regno,
+ cspec->dca_rcvhdr_ctrl[rmp->shadow_inx]);
+ cspec->dca_ctrl |= SYM_MASK(DCACtrlA, RcvHdrqDCAEnable);
+ qib_write_kreg(dd, KREG_IDX(DCACtrlA), cspec->dca_ctrl);
+ }
+}
+
+static void qib_update_sdma_dca(struct qib_pportdata *ppd, int cpu)
+{
+ struct qib_devdata *dd = ppd->dd;
+ struct qib_chip_specific *cspec = dd->cspec;
+ unsigned pidx = ppd->port - 1;
+
+ if (!(dd->flags & QIB_DCA_ENABLED))
+ return;
+ if (cspec->sdma_cpu[pidx] != cpu) {
+ cspec->sdma_cpu[pidx] = cpu;
+ cspec->dca_rcvhdr_ctrl[4] &= ~(ppd->hw_pidx ?
+ SYM_MASK(DCACtrlF, SendDma1DCAOPH) :
+ SYM_MASK(DCACtrlF, SendDma0DCAOPH));
+ cspec->dca_rcvhdr_ctrl[4] |=
+ (u64) dca3_get_tag(&dd->pcidev->dev, cpu) <<
+ (ppd->hw_pidx ?
+ SYM_LSB(DCACtrlF, SendDma1DCAOPH) :
+ SYM_LSB(DCACtrlF, SendDma0DCAOPH));
+ qib_devinfo(dd->pcidev,
+ "sdma %d cpu %d dca %llx\n", ppd->hw_pidx, cpu,
+ (long long) cspec->dca_rcvhdr_ctrl[4]);
+ qib_write_kreg(dd, KREG_IDX(DCACtrlF),
+ cspec->dca_rcvhdr_ctrl[4]);
+ cspec->dca_ctrl |= ppd->hw_pidx ?
+ SYM_MASK(DCACtrlA, SendDMAHead1DCAEnable) :
+ SYM_MASK(DCACtrlA, SendDMAHead0DCAEnable);
+ qib_write_kreg(dd, KREG_IDX(DCACtrlA), cspec->dca_ctrl);
+ }
+}
+
+static void qib_setup_dca(struct qib_devdata *dd)
+{
+ struct qib_chip_specific *cspec = dd->cspec;
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(cspec->rhdr_cpu); i++)
+ cspec->rhdr_cpu[i] = -1;
+ for (i = 0; i < ARRAY_SIZE(cspec->sdma_cpu); i++)
+ cspec->sdma_cpu[i] = -1;
+ cspec->dca_rcvhdr_ctrl[0] =
+ (1ULL << SYM_LSB(DCACtrlB, RcvHdrq0DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlB, RcvHdrq1DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlB, RcvHdrq2DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlB, RcvHdrq3DCAXfrCnt));
+ cspec->dca_rcvhdr_ctrl[1] =
+ (1ULL << SYM_LSB(DCACtrlC, RcvHdrq4DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlC, RcvHdrq5DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlC, RcvHdrq6DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlC, RcvHdrq7DCAXfrCnt));
+ cspec->dca_rcvhdr_ctrl[2] =
+ (1ULL << SYM_LSB(DCACtrlD, RcvHdrq8DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlD, RcvHdrq9DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlD, RcvHdrq10DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlD, RcvHdrq11DCAXfrCnt));
+ cspec->dca_rcvhdr_ctrl[3] =
+ (1ULL << SYM_LSB(DCACtrlE, RcvHdrq12DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlE, RcvHdrq13DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlE, RcvHdrq14DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlE, RcvHdrq15DCAXfrCnt));
+ cspec->dca_rcvhdr_ctrl[4] =
+ (1ULL << SYM_LSB(DCACtrlF, RcvHdrq16DCAXfrCnt)) |
+ (1ULL << SYM_LSB(DCACtrlF, RcvHdrq17DCAXfrCnt));
+ for (i = 0; i < ARRAY_SIZE(cspec->sdma_cpu); i++)
+ qib_write_kreg(dd, KREG_IDX(DCACtrlB) + i,
+ cspec->dca_rcvhdr_ctrl[i]);
+ for (i = 0; i < cspec->num_msix_entries; i++)
+ setup_dca_notifier(dd, &cspec->msix_entries[i]);
+}
+
+static void qib_irq_notifier_notify(struct irq_affinity_notify *notify,
+ const cpumask_t *mask)
+{
+ struct qib_irq_notify *n =
+ container_of(notify, struct qib_irq_notify, notify);
+ int cpu = cpumask_first(mask);
+
+ if (n->rcv) {
+ struct qib_ctxtdata *rcd = (struct qib_ctxtdata *)n->arg;
+ qib_update_rhdrq_dca(rcd, cpu);
+ } else {
+ struct qib_pportdata *ppd = (struct qib_pportdata *)n->arg;
+ qib_update_sdma_dca(ppd, cpu);
+ }
+}
+
+static void qib_irq_notifier_release(struct kref *ref)
+{
+ struct qib_irq_notify *n =
+ container_of(ref, struct qib_irq_notify, notify.kref);
+ struct qib_devdata *dd;
+
+ if (n->rcv) {
+ struct qib_ctxtdata *rcd = (struct qib_ctxtdata *)n->arg;
+ dd = rcd->dd;
+ } else {
+ struct qib_pportdata *ppd = (struct qib_pportdata *)n->arg;
+ dd = ppd->dd;
+ }
+ qib_devinfo(dd->pcidev,
+ "release on HCA notify 0x%p n 0x%p\n", ref, n);
+ kfree(n);
+}
+#endif
+
/*
* Disable MSIx interrupt if enabled, call generic MSIx code
* to cleanup, and clear pending MSIx interrupts.
@@ -2575,6 +2837,9 @@
dd->cspec->num_msix_entries = 0;
for (i = 0; i < n; i++) {
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ reset_dca_notifier(dd, &dd->cspec->msix_entries[i]);
+#endif
irq_set_affinity_hint(
dd->cspec->msix_entries[i].msix.vector, NULL);
free_cpumask_var(dd->cspec->msix_entries[i].mask);
@@ -2602,6 +2867,15 @@
{
int i;
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ if (dd->flags & QIB_DCA_ENABLED) {
+ dca_remove_requester(&dd->pcidev->dev);
+ dd->flags &= ~QIB_DCA_ENABLED;
+ dd->cspec->dca_ctrl = 0;
+ qib_write_kreg(dd, KREG_IDX(DCACtrlA), dd->cspec->dca_ctrl);
+ }
+#endif
+
qib_7322_free_irq(dd);
kfree(dd->cspec->cntrs);
kfree(dd->cspec->sendchkenable);
@@ -3068,6 +3342,53 @@
return IRQ_HANDLED;
}
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+
+static void reset_dca_notifier(struct qib_devdata *dd, struct qib_msix_entry *m)
+{
+ if (!m->dca)
+ return;
+ qib_devinfo(dd->pcidev,
+ "Disabling notifier on HCA %d irq %d\n",
+ dd->unit,
+ m->msix.vector);
+ irq_set_affinity_notifier(
+ m->msix.vector,
+ NULL);
+ m->notifier = NULL;
+}
+
+static void setup_dca_notifier(struct qib_devdata *dd, struct qib_msix_entry *m)
+{
+ struct qib_irq_notify *n;
+
+ if (!m->dca)
+ return;
+ n = kzalloc(sizeof(*n), GFP_KERNEL);
+ if (n) {
+ int ret;
+
+ m->notifier = n;
+ n->notify.irq = m->msix.vector;
+ n->notify.notify = qib_irq_notifier_notify;
+ n->notify.release = qib_irq_notifier_release;
+ n->arg = m->arg;
+ n->rcv = m->rcv;
+ qib_devinfo(dd->pcidev,
+ "set notifier irq %d rcv %d notify %p\n",
+ n->notify.irq, n->rcv, &n->notify);
+ ret = irq_set_affinity_notifier(
+ n->notify.irq,
+ &n->notify);
+ if (ret) {
+ m->notifier = NULL;
+ kfree(n);
+ }
+ }
+}
+
+#endif
+
/*
* Set up our chip-specific interrupt handler.
* The interrupt type has already been setup, so
@@ -3149,6 +3470,9 @@
void *arg;
u64 val;
int lsb, reg, sh;
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ int dca = 0;
+#endif
dd->cspec->msix_entries[msixnum].
name[sizeof(dd->cspec->msix_entries[msixnum].name) - 1]
@@ -3161,6 +3485,9 @@
arg = dd->pport + irq_table[i].port - 1;
} else
arg = dd;
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ dca = irq_table[i].dca;
+#endif
lsb = irq_table[i].lsb;
handler = irq_table[i].handler;
snprintf(dd->cspec->msix_entries[msixnum].name,
@@ -3178,6 +3505,9 @@
continue;
if (qib_krcvq01_no_msi && ctxt < 2)
continue;
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ dca = 1;
+#endif
lsb = QIB_I_RCVAVAIL_LSB + ctxt;
handler = qib_7322pintr;
snprintf(dd->cspec->msix_entries[msixnum].name,
@@ -3203,6 +3533,11 @@
goto try_intx;
}
dd->cspec->msix_entries[msixnum].arg = arg;
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ dd->cspec->msix_entries[msixnum].dca = dca;
+ dd->cspec->msix_entries[msixnum].rcv =
+ handler == qib_7322pintr;
+#endif
if (lsb >= 0) {
reg = lsb / IBA7322_REDIRECT_VEC_PER_REG;
sh = (lsb % IBA7322_REDIRECT_VEC_PER_REG) *
@@ -6452,6 +6787,86 @@
qib_write_kreg_port(ppd, krp_senddmadesccnt, cnt);
}
+/*
+ * sdma_lock should be acquired before calling this routine
+ */
+static void dump_sdma_7322_state(struct qib_pportdata *ppd)
+{
+ u64 reg, reg1, reg2;
+
+ reg = qib_read_kreg_port(ppd, krp_senddmastatus);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmastatus: 0x%016llx\n", reg);
+
+ reg = qib_read_kreg_port(ppd, krp_sendctrl);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA sendctrl: 0x%016llx\n", reg);
+
+ reg = qib_read_kreg_port(ppd, krp_senddmabase);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmabase: 0x%016llx\n", reg);
+
+ reg = qib_read_kreg_port(ppd, krp_senddmabufmask0);
+ reg1 = qib_read_kreg_port(ppd, krp_senddmabufmask1);
+ reg2 = qib_read_kreg_port(ppd, krp_senddmabufmask2);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmabufmask 0:%llx 1:%llx 2:%llx\n",
+ reg, reg1, reg2);
+
+ /* get bufuse bits, clear them, and print them again if non-zero */
+ reg = qib_read_kreg_port(ppd, krp_senddmabuf_use0);
+ qib_write_kreg_port(ppd, krp_senddmabuf_use0, reg);
+ reg1 = qib_read_kreg_port(ppd, krp_senddmabuf_use1);
+ qib_write_kreg_port(ppd, krp_senddmabuf_use0, reg1);
+ reg2 = qib_read_kreg_port(ppd, krp_senddmabuf_use2);
+ qib_write_kreg_port(ppd, krp_senddmabuf_use0, reg2);
+ /* 0 and 1 should always be zero, so print as short form */
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA current senddmabuf_use 0:%llx 1:%llx 2:%llx\n",
+ reg, reg1, reg2);
+ reg = qib_read_kreg_port(ppd, krp_senddmabuf_use0);
+ reg1 = qib_read_kreg_port(ppd, krp_senddmabuf_use1);
+ reg2 = qib_read_kreg_port(ppd, krp_senddmabuf_use2);
+ /* 0 and 1 should always be zero, so print as short form */
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA cleared senddmabuf_use 0:%llx 1:%llx 2:%llx\n",
+ reg, reg1, reg2);
+
+ reg = qib_read_kreg_port(ppd, krp_senddmatail);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmatail: 0x%016llx\n", reg);
+
+ reg = qib_read_kreg_port(ppd, krp_senddmahead);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmahead: 0x%016llx\n", reg);
+
+ reg = qib_read_kreg_port(ppd, krp_senddmaheadaddr);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmaheadaddr: 0x%016llx\n", reg);
+
+ reg = qib_read_kreg_port(ppd, krp_senddmalengen);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmalengen: 0x%016llx\n", reg);
+
+ reg = qib_read_kreg_port(ppd, krp_senddmadesccnt);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmadesccnt: 0x%016llx\n", reg);
+
+ reg = qib_read_kreg_port(ppd, krp_senddmaidlecnt);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmaidlecnt: 0x%016llx\n", reg);
+
+ reg = qib_read_kreg_port(ppd, krp_senddmaprioritythld);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmapriorityhld: 0x%016llx\n", reg);
+
+ reg = qib_read_kreg_port(ppd, krp_senddmareloadcnt);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA senddmareloadcnt: 0x%016llx\n", reg);
+
+ dump_sdma_state(ppd);
+}
+
static struct sdma_set_state_action sdma_7322_action_table[] = {
[qib_sdma_state_s00_hw_down] = {
.go_s99_running_tofalse = 1,
@@ -6885,6 +7300,9 @@
dd->f_sdma_init_early = qib_7322_sdma_init_early;
dd->f_writescratch = writescratch;
dd->f_tempsense_rd = qib_7322_tempsense_rd;
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ dd->f_notify_dca = qib_7322_notify_dca;
+#endif
/*
* Do remaining PCIe setup and save PCIe values in dd.
* Any error printing is already done by the init code.
@@ -6921,7 +7339,7 @@
actual_cnt -= dd->num_pports;
tabsize = actual_cnt;
- dd->cspec->msix_entries = kmalloc(tabsize *
+ dd->cspec->msix_entries = kzalloc(tabsize *
sizeof(struct qib_msix_entry), GFP_KERNEL);
if (!dd->cspec->msix_entries) {
qib_dev_err(dd, "No memory for MSIx table\n");
@@ -6941,7 +7359,13 @@
/* clear diagctrl register, in case diags were running and crashed */
qib_write_kreg(dd, kr_hwdiagctrl, 0);
-
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ if (!dca_add_requester(&pdev->dev)) {
+ qib_devinfo(dd->pcidev, "DCA enabled\n");
+ dd->flags |= QIB_DCA_ENABLED;
+ qib_setup_dca(dd);
+ }
+#endif
goto bail;
bail_cleanup:
@@ -7156,15 +7580,20 @@
{ 0, 0, 0, 1 }, /* QMH7342 backplane settings */
{ 0, 0, 0, 2 }, /* QMH7342 backplane settings */
{ 0, 0, 0, 2 }, /* QMH7342 backplane settings */
- { 0, 0, 0, 11 }, /* QME7342 backplane settings */
- { 0, 0, 0, 11 }, /* QME7342 backplane settings */
- { 0, 0, 0, 11 }, /* QME7342 backplane settings */
- { 0, 0, 0, 11 }, /* QME7342 backplane settings */
- { 0, 0, 0, 11 }, /* QME7342 backplane settings */
- { 0, 0, 0, 11 }, /* QME7342 backplane settings */
- { 0, 0, 0, 11 }, /* QME7342 backplane settings */
{ 0, 0, 0, 3 }, /* QMH7342 backplane settings */
{ 0, 0, 0, 4 }, /* QMH7342 backplane settings */
+ { 0, 1, 4, 15 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 3, 15 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 12 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 11 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 9 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 14 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 2, 15 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 11 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 7 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 9 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 6 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 8 }, /* QME7342 backplane settings 1.1 */
};
static const struct txdds_ent txdds_extra_ddr[TXDDS_EXTRA_SZ] = {
@@ -7173,15 +7602,20 @@
{ 0, 0, 0, 7 }, /* QMH7342 backplane settings */
{ 0, 0, 0, 8 }, /* QMH7342 backplane settings */
{ 0, 0, 0, 8 }, /* QMH7342 backplane settings */
- { 0, 0, 0, 13 }, /* QME7342 backplane settings */
- { 0, 0, 0, 13 }, /* QME7342 backplane settings */
- { 0, 0, 0, 13 }, /* QME7342 backplane settings */
- { 0, 0, 0, 13 }, /* QME7342 backplane settings */
- { 0, 0, 0, 13 }, /* QME7342 backplane settings */
- { 0, 0, 0, 13 }, /* QME7342 backplane settings */
- { 0, 0, 0, 13 }, /* QME7342 backplane settings */
{ 0, 0, 0, 9 }, /* QMH7342 backplane settings */
{ 0, 0, 0, 10 }, /* QMH7342 backplane settings */
+ { 0, 1, 4, 15 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 3, 15 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 12 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 11 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 9 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 14 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 2, 15 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 11 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 7 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 9 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 6 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 8 }, /* QME7342 backplane settings 1.1 */
};
static const struct txdds_ent txdds_extra_qdr[TXDDS_EXTRA_SZ] = {
@@ -7190,15 +7624,20 @@
{ 0, 1, 0, 5 }, /* QMH7342 backplane settings */
{ 0, 1, 0, 6 }, /* QMH7342 backplane settings */
{ 0, 1, 0, 8 }, /* QMH7342 backplane settings */
- { 0, 1, 12, 10 }, /* QME7342 backplane setting */
- { 0, 1, 12, 11 }, /* QME7342 backplane setting */
- { 0, 1, 12, 12 }, /* QME7342 backplane setting */
- { 0, 1, 12, 14 }, /* QME7342 backplane setting */
- { 0, 1, 12, 6 }, /* QME7342 backplane setting */
- { 0, 1, 12, 7 }, /* QME7342 backplane setting */
- { 0, 1, 12, 8 }, /* QME7342 backplane setting */
{ 0, 1, 0, 10 }, /* QMH7342 backplane settings */
{ 0, 1, 0, 12 }, /* QMH7342 backplane settings */
+ { 0, 1, 4, 15 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 3, 15 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 12 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 11 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 9 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 14 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 2, 15 }, /* QME7342 backplane settings 1.0 */
+ { 0, 1, 0, 11 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 7 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 9 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 6 }, /* QME7342 backplane settings 1.1 */
+ { 0, 1, 0, 8 }, /* QME7342 backplane settings 1.1 */
};
static const struct txdds_ent txdds_extra_mfg[TXDDS_MFG_SZ] = {
diff --git a/drivers/infiniband/hw/qib/qib_init.c b/drivers/infiniband/hw/qib/qib_init.c
index 173f805..36e048e 100644
--- a/drivers/infiniband/hw/qib/qib_init.c
+++ b/drivers/infiniband/hw/qib/qib_init.c
@@ -39,10 +39,17 @@
#include <linux/idr.h>
#include <linux/module.h>
#include <linux/printk.h>
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+#include <linux/dca.h>
+#endif
#include "qib.h"
#include "qib_common.h"
#include "qib_mad.h"
+#ifdef CONFIG_DEBUG_FS
+#include "qib_debugfs.h"
+#include "qib_verbs.h"
+#endif
#undef pr_fmt
#define pr_fmt(fmt) QIB_DRV_NAME ": " fmt
@@ -64,6 +71,11 @@
module_param_named(cfgctxts, qib_cfgctxts, ushort, S_IRUGO);
MODULE_PARM_DESC(cfgctxts, "Set max number of contexts to use");
+unsigned qib_numa_aware;
+module_param_named(numa_aware, qib_numa_aware, uint, S_IRUGO);
+MODULE_PARM_DESC(numa_aware,
+ "0 -> PSM allocation close to HCA, 1 -> PSM allocation local to process");
+
/*
* If set, do not write to any regs if avoidable, hack to allow
* check for deranged default register values.
@@ -89,8 +101,6 @@
module_param_named(wc_pat, qib_wc_pat, uint, S_IRUGO);
MODULE_PARM_DESC(wc_pat, "enable write-combining via PAT mechanism");
-struct workqueue_struct *qib_cq_wq;
-
static void verify_interrupt(unsigned long);
static struct idr qib_unit_table;
@@ -121,6 +131,11 @@
{
unsigned i;
int ret;
+ int local_node_id = pcibus_to_node(dd->pcidev->bus);
+
+ if (local_node_id < 0)
+ local_node_id = numa_node_id();
+ dd->assigned_node_id = local_node_id;
/*
* Allocate full ctxtcnt array, rather than just cfgctxts, because
@@ -143,7 +158,8 @@
continue;
ppd = dd->pport + (i % dd->num_pports);
- rcd = qib_create_ctxtdata(ppd, i);
+
+ rcd = qib_create_ctxtdata(ppd, i, dd->assigned_node_id);
if (!rcd) {
qib_dev_err(dd,
"Unable to allocate ctxtdata for Kernel ctxt, failing\n");
@@ -161,20 +177,33 @@
/*
* Common code for user and kernel context setup.
*/
-struct qib_ctxtdata *qib_create_ctxtdata(struct qib_pportdata *ppd, u32 ctxt)
+struct qib_ctxtdata *qib_create_ctxtdata(struct qib_pportdata *ppd, u32 ctxt,
+ int node_id)
{
struct qib_devdata *dd = ppd->dd;
struct qib_ctxtdata *rcd;
- rcd = kzalloc(sizeof(*rcd), GFP_KERNEL);
+ rcd = kzalloc_node(sizeof(*rcd), GFP_KERNEL, node_id);
if (rcd) {
INIT_LIST_HEAD(&rcd->qp_wait_list);
+ rcd->node_id = node_id;
rcd->ppd = ppd;
rcd->dd = dd;
rcd->cnt = 1;
rcd->ctxt = ctxt;
dd->rcd[ctxt] = rcd;
-
+#ifdef CONFIG_DEBUG_FS
+ if (ctxt < dd->first_user_ctxt) { /* N/A for PSM contexts */
+ rcd->opstats = kzalloc_node(sizeof(*rcd->opstats),
+ GFP_KERNEL, node_id);
+ if (!rcd->opstats) {
+ kfree(rcd);
+ qib_dev_err(dd,
+ "Unable to allocate per ctxt stats buffer\n");
+ return NULL;
+ }
+ }
+#endif
dd->f_init_ctxt(rcd);
/*
@@ -429,6 +458,7 @@
dd->intrchk_timer.function = verify_interrupt;
dd->intrchk_timer.data = (unsigned long) dd;
+ ret = qib_cq_init(dd);
done:
return ret;
}
@@ -944,6 +974,10 @@
vfree(rcd->subctxt_uregbase);
vfree(rcd->subctxt_rcvegrbuf);
vfree(rcd->subctxt_rcvhdr_base);
+#ifdef CONFIG_DEBUG_FS
+ kfree(rcd->opstats);
+ rcd->opstats = NULL;
+#endif
kfree(rcd);
}
@@ -1033,7 +1067,6 @@
dd->f_set_armlaunch(dd, 1);
}
-
void qib_free_devdata(struct qib_devdata *dd)
{
unsigned long flags;
@@ -1043,6 +1076,9 @@
list_del(&dd->list);
spin_unlock_irqrestore(&qib_devs_lock, flags);
+#ifdef CONFIG_DEBUG_FS
+ qib_dbg_ibdev_exit(&dd->verbs_dev);
+#endif
ib_dealloc_device(&dd->verbs_dev.ibdev);
}
@@ -1066,6 +1102,10 @@
goto bail;
}
+#ifdef CONFIG_DEBUG_FS
+ qib_dbg_ibdev_init(&dd->verbs_dev);
+#endif
+
idr_preload(GFP_KERNEL);
spin_lock_irqsave(&qib_devs_lock, flags);
@@ -1081,6 +1121,9 @@
if (ret < 0) {
qib_early_err(&pdev->dev,
"Could not allocate unit ID: error %d\n", -ret);
+#ifdef CONFIG_DEBUG_FS
+ qib_dbg_ibdev_exit(&dd->verbs_dev);
+#endif
ib_dealloc_device(&dd->verbs_dev.ibdev);
dd = ERR_PTR(ret);
goto bail;
@@ -1158,6 +1201,35 @@
.err_handler = &qib_pci_err_handler,
};
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+
+static int qib_notify_dca(struct notifier_block *, unsigned long, void *);
+static struct notifier_block dca_notifier = {
+ .notifier_call = qib_notify_dca,
+ .next = NULL,
+ .priority = 0
+};
+
+static int qib_notify_dca_device(struct device *device, void *data)
+{
+ struct qib_devdata *dd = dev_get_drvdata(device);
+ unsigned long event = *(unsigned long *)data;
+
+ return dd->f_notify_dca(dd, event);
+}
+
+static int qib_notify_dca(struct notifier_block *nb, unsigned long event,
+ void *p)
+{
+ int rval;
+
+ rval = driver_for_each_device(&qib_driver.driver, NULL,
+ &event, qib_notify_dca_device);
+ return rval ? NOTIFY_BAD : NOTIFY_DONE;
+}
+
+#endif
+
/*
* Do all the generic driver unit- and chip-independent memory
* allocation and initialization.
@@ -1170,22 +1242,22 @@
if (ret)
goto bail;
- qib_cq_wq = create_singlethread_workqueue("qib_cq");
- if (!qib_cq_wq) {
- ret = -ENOMEM;
- goto bail_dev;
- }
-
/*
* These must be called before the driver is registered with
* the PCI subsystem.
*/
idr_init(&qib_unit_table);
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ dca_register_notify(&dca_notifier);
+#endif
+#ifdef CONFIG_DEBUG_FS
+ qib_dbg_init();
+#endif
ret = pci_register_driver(&qib_driver);
if (ret < 0) {
pr_err("Unable to register driver: error %d\n", -ret);
- goto bail_unit;
+ goto bail_dev;
}
/* not fatal if it doesn't work */
@@ -1193,10 +1265,14 @@
pr_err("Unable to register ipathfs\n");
goto bail; /* all OK */
-bail_unit:
- idr_destroy(&qib_unit_table);
- destroy_workqueue(qib_cq_wq);
bail_dev:
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ dca_unregister_notify(&dca_notifier);
+#endif
+#ifdef CONFIG_DEBUG_FS
+ qib_dbg_exit();
+#endif
+ idr_destroy(&qib_unit_table);
qib_dev_cleanup();
bail:
return ret;
@@ -1217,9 +1293,13 @@
"Unable to cleanup counter filesystem: error %d\n",
-ret);
+#ifdef CONFIG_INFINIBAND_QIB_DCA
+ dca_unregister_notify(&dca_notifier);
+#endif
pci_unregister_driver(&qib_driver);
-
- destroy_workqueue(qib_cq_wq);
+#ifdef CONFIG_DEBUG_FS
+ qib_dbg_exit();
+#endif
qib_cpulist_count = 0;
kfree(qib_cpulist);
@@ -1270,7 +1350,7 @@
if (dd->pageshadow) {
struct page **tmpp = dd->pageshadow;
dma_addr_t *tmpd = dd->physshadow;
- int i, cnt = 0;
+ int i;
for (ctxt = 0; ctxt < dd->cfgctxts; ctxt++) {
int ctxt_tidbase = ctxt * dd->rcvtidcnt;
@@ -1283,13 +1363,13 @@
PAGE_SIZE, PCI_DMA_FROMDEVICE);
qib_release_user_pages(&tmpp[i], 1);
tmpp[i] = NULL;
- cnt++;
}
}
- tmpp = dd->pageshadow;
dd->pageshadow = NULL;
vfree(tmpp);
+ dd->physshadow = NULL;
+ vfree(tmpd);
}
/*
@@ -1311,6 +1391,7 @@
}
kfree(tmp);
kfree(dd->boardname);
+ qib_cq_exit(dd);
}
/*
@@ -1483,6 +1564,7 @@
int qib_create_rcvhdrq(struct qib_devdata *dd, struct qib_ctxtdata *rcd)
{
unsigned amt;
+ int old_node_id;
if (!rcd->rcvhdrq) {
dma_addr_t phys_hdrqtail;
@@ -1492,9 +1574,13 @@
sizeof(u32), PAGE_SIZE);
gfp_flags = (rcd->ctxt >= dd->first_user_ctxt) ?
GFP_USER : GFP_KERNEL;
+
+ old_node_id = dev_to_node(&dd->pcidev->dev);
+ set_dev_node(&dd->pcidev->dev, rcd->node_id);
rcd->rcvhdrq = dma_alloc_coherent(
&dd->pcidev->dev, amt, &rcd->rcvhdrq_phys,
gfp_flags | __GFP_COMP);
+ set_dev_node(&dd->pcidev->dev, old_node_id);
if (!rcd->rcvhdrq) {
qib_dev_err(dd,
@@ -1510,9 +1596,11 @@
}
if (!(dd->flags & QIB_NODMA_RTAIL)) {
+ set_dev_node(&dd->pcidev->dev, rcd->node_id);
rcd->rcvhdrtail_kvaddr = dma_alloc_coherent(
&dd->pcidev->dev, PAGE_SIZE, &phys_hdrqtail,
gfp_flags);
+ set_dev_node(&dd->pcidev->dev, old_node_id);
if (!rcd->rcvhdrtail_kvaddr)
goto bail_free;
rcd->rcvhdrqtailaddr_phys = phys_hdrqtail;
@@ -1556,6 +1644,7 @@
unsigned e, egrcnt, egrperchunk, chunk, egrsize, egroff;
size_t size;
gfp_t gfp_flags;
+ int old_node_id;
/*
* GFP_USER, but without GFP_FS, so buffer cache can be
@@ -1574,25 +1663,29 @@
size = rcd->rcvegrbuf_size;
if (!rcd->rcvegrbuf) {
rcd->rcvegrbuf =
- kzalloc(chunk * sizeof(rcd->rcvegrbuf[0]),
- GFP_KERNEL);
+ kzalloc_node(chunk * sizeof(rcd->rcvegrbuf[0]),
+ GFP_KERNEL, rcd->node_id);
if (!rcd->rcvegrbuf)
goto bail;
}
if (!rcd->rcvegrbuf_phys) {
rcd->rcvegrbuf_phys =
- kmalloc(chunk * sizeof(rcd->rcvegrbuf_phys[0]),
- GFP_KERNEL);
+ kmalloc_node(chunk * sizeof(rcd->rcvegrbuf_phys[0]),
+ GFP_KERNEL, rcd->node_id);
if (!rcd->rcvegrbuf_phys)
goto bail_rcvegrbuf;
}
for (e = 0; e < rcd->rcvegrbuf_chunks; e++) {
if (rcd->rcvegrbuf[e])
continue;
+
+ old_node_id = dev_to_node(&dd->pcidev->dev);
+ set_dev_node(&dd->pcidev->dev, rcd->node_id);
rcd->rcvegrbuf[e] =
dma_alloc_coherent(&dd->pcidev->dev, size,
&rcd->rcvegrbuf_phys[e],
gfp_flags);
+ set_dev_node(&dd->pcidev->dev, old_node_id);
if (!rcd->rcvegrbuf[e])
goto bail_rcvegrbuf_phys;
}
diff --git a/drivers/infiniband/hw/qib/qib_qp.c b/drivers/infiniband/hw/qib/qib_qp.c
index a6a2cc2..3cca55b 100644
--- a/drivers/infiniband/hw/qib/qib_qp.c
+++ b/drivers/infiniband/hw/qib/qib_qp.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2012 Intel Corporation. All rights reserved.
+ * Copyright (c) 2012, 2013 Intel Corporation. All rights reserved.
* Copyright (c) 2006 - 2012 QLogic Corporation. * All rights reserved.
* Copyright (c) 2005, 2006 PathScale, Inc. All rights reserved.
*
@@ -35,6 +35,9 @@
#include <linux/err.h>
#include <linux/vmalloc.h>
#include <linux/jhash.h>
+#ifdef CONFIG_DEBUG_FS
+#include <linux/seq_file.h>
+#endif
#include "qib.h"
@@ -222,8 +225,8 @@
unsigned long flags;
unsigned n = qpn_hash(dev, qp->ibqp.qp_num);
- spin_lock_irqsave(&dev->qpt_lock, flags);
atomic_inc(&qp->refcount);
+ spin_lock_irqsave(&dev->qpt_lock, flags);
if (qp->ibqp.qp_num == 0)
rcu_assign_pointer(ibp->qp0, qp);
@@ -235,7 +238,6 @@
}
spin_unlock_irqrestore(&dev->qpt_lock, flags);
- synchronize_rcu();
}
/*
@@ -247,36 +249,39 @@
struct qib_ibport *ibp = to_iport(qp->ibqp.device, qp->port_num);
unsigned n = qpn_hash(dev, qp->ibqp.qp_num);
unsigned long flags;
+ int removed = 1;
spin_lock_irqsave(&dev->qpt_lock, flags);
if (rcu_dereference_protected(ibp->qp0,
lockdep_is_held(&dev->qpt_lock)) == qp) {
- atomic_dec(&qp->refcount);
rcu_assign_pointer(ibp->qp0, NULL);
} else if (rcu_dereference_protected(ibp->qp1,
lockdep_is_held(&dev->qpt_lock)) == qp) {
- atomic_dec(&qp->refcount);
rcu_assign_pointer(ibp->qp1, NULL);
} else {
struct qib_qp *q;
struct qib_qp __rcu **qpp;
+ removed = 0;
qpp = &dev->qp_table[n];
for (; (q = rcu_dereference_protected(*qpp,
lockdep_is_held(&dev->qpt_lock))) != NULL;
qpp = &q->next)
if (q == qp) {
- atomic_dec(&qp->refcount);
rcu_assign_pointer(*qpp,
rcu_dereference_protected(qp->next,
lockdep_is_held(&dev->qpt_lock)));
+ removed = 1;
break;
}
}
spin_unlock_irqrestore(&dev->qpt_lock, flags);
- synchronize_rcu();
+ if (removed) {
+ synchronize_rcu();
+ atomic_dec(&qp->refcount);
+ }
}
/**
@@ -334,26 +339,25 @@
{
struct qib_qp *qp = NULL;
+ rcu_read_lock();
if (unlikely(qpn <= 1)) {
- rcu_read_lock();
if (qpn == 0)
qp = rcu_dereference(ibp->qp0);
else
qp = rcu_dereference(ibp->qp1);
+ if (qp)
+ atomic_inc(&qp->refcount);
} else {
struct qib_ibdev *dev = &ppd_from_ibp(ibp)->dd->verbs_dev;
unsigned n = qpn_hash(dev, qpn);
- rcu_read_lock();
for (qp = rcu_dereference(dev->qp_table[n]); qp;
qp = rcu_dereference(qp->next))
- if (qp->ibqp.qp_num == qpn)
+ if (qp->ibqp.qp_num == qpn) {
+ atomic_inc(&qp->refcount);
break;
+ }
}
- if (qp)
- if (unlikely(!atomic_inc_not_zero(&qp->refcount)))
- qp = NULL;
-
rcu_read_unlock();
return qp;
}
@@ -1286,3 +1290,94 @@
}
}
}
+
+#ifdef CONFIG_DEBUG_FS
+
+struct qib_qp_iter {
+ struct qib_ibdev *dev;
+ struct qib_qp *qp;
+ int n;
+};
+
+struct qib_qp_iter *qib_qp_iter_init(struct qib_ibdev *dev)
+{
+ struct qib_qp_iter *iter;
+
+ iter = kzalloc(sizeof(*iter), GFP_KERNEL);
+ if (!iter)
+ return NULL;
+
+ iter->dev = dev;
+ if (qib_qp_iter_next(iter)) {
+ kfree(iter);
+ return NULL;
+ }
+
+ return iter;
+}
+
+int qib_qp_iter_next(struct qib_qp_iter *iter)
+{
+ struct qib_ibdev *dev = iter->dev;
+ int n = iter->n;
+ int ret = 1;
+ struct qib_qp *pqp = iter->qp;
+ struct qib_qp *qp;
+
+ rcu_read_lock();
+ for (; n < dev->qp_table_size; n++) {
+ if (pqp)
+ qp = rcu_dereference(pqp->next);
+ else
+ qp = rcu_dereference(dev->qp_table[n]);
+ pqp = qp;
+ if (qp) {
+ if (iter->qp)
+ atomic_dec(&iter->qp->refcount);
+ atomic_inc(&qp->refcount);
+ rcu_read_unlock();
+ iter->qp = qp;
+ iter->n = n;
+ return 0;
+ }
+ }
+ rcu_read_unlock();
+ if (iter->qp)
+ atomic_dec(&iter->qp->refcount);
+ return ret;
+}
+
+static const char * const qp_type_str[] = {
+ "SMI", "GSI", "RC", "UC", "UD",
+};
+
+void qib_qp_iter_print(struct seq_file *s, struct qib_qp_iter *iter)
+{
+ struct qib_swqe *wqe;
+ struct qib_qp *qp = iter->qp;
+
+ wqe = get_swqe_ptr(qp, qp->s_last);
+ seq_printf(s,
+ "N %d QP%u %s %u %u %u f=%x %u %u %u %u %u PSN %x %x %x %x %x (%u %u %u %u %u %u) QP%u LID %x\n",
+ iter->n,
+ qp->ibqp.qp_num,
+ qp_type_str[qp->ibqp.qp_type],
+ qp->state,
+ wqe->wr.opcode,
+ qp->s_hdrwords,
+ qp->s_flags,
+ atomic_read(&qp->s_dma_busy),
+ !list_empty(&qp->iowait),
+ qp->timeout,
+ wqe->ssn,
+ qp->s_lsn,
+ qp->s_last_psn,
+ qp->s_psn, qp->s_next_psn,
+ qp->s_sending_psn, qp->s_sending_hpsn,
+ qp->s_last, qp->s_acked, qp->s_cur,
+ qp->s_tail, qp->s_head, qp->s_size,
+ qp->remote_qpn,
+ qp->remote_ah_attr.dlid);
+}
+
+#endif
diff --git a/drivers/infiniband/hw/qib/qib_sdma.c b/drivers/infiniband/hw/qib/qib_sdma.c
index 3fc5144..32162d3 100644
--- a/drivers/infiniband/hw/qib/qib_sdma.c
+++ b/drivers/infiniband/hw/qib/qib_sdma.c
@@ -708,6 +708,62 @@
return ret;
}
+/*
+ * sdma_lock should be acquired before calling this routine
+ */
+void dump_sdma_state(struct qib_pportdata *ppd)
+{
+ struct qib_sdma_desc *descq;
+ struct qib_sdma_txreq *txp, *txpnext;
+ __le64 *descqp;
+ u64 desc[2];
+ dma_addr_t addr;
+ u16 gen, dwlen, dwoffset;
+ u16 head, tail, cnt;
+
+ head = ppd->sdma_descq_head;
+ tail = ppd->sdma_descq_tail;
+ cnt = qib_sdma_descq_freecnt(ppd);
+ descq = ppd->sdma_descq;
+
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA ppd->sdma_descq_head: %u\n", head);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA ppd->sdma_descq_tail: %u\n", tail);
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA sdma_descq_freecnt: %u\n", cnt);
+
+ /* print info for each entry in the descriptor queue */
+ while (head != tail) {
+ char flags[6] = { 'x', 'x', 'x', 'x', 'x', 0 };
+
+ descqp = &descq[head].qw[0];
+ desc[0] = le64_to_cpu(descqp[0]);
+ desc[1] = le64_to_cpu(descqp[1]);
+ flags[0] = (desc[0] & 1<<15) ? 'I' : '-';
+ flags[1] = (desc[0] & 1<<14) ? 'L' : 'S';
+ flags[2] = (desc[0] & 1<<13) ? 'H' : '-';
+ flags[3] = (desc[0] & 1<<12) ? 'F' : '-';
+ flags[4] = (desc[0] & 1<<11) ? 'L' : '-';
+ addr = (desc[1] << 32) | ((desc[0] >> 32) & 0xfffffffcULL);
+ gen = (desc[0] >> 30) & 3ULL;
+ dwlen = (desc[0] >> 14) & (0x7ffULL << 2);
+ dwoffset = (desc[0] & 0x7ffULL) << 2;
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA sdmadesc[%u]: flags:%s addr:0x%016llx gen:%u len:%u bytes offset:%u bytes\n",
+ head, flags, addr, gen, dwlen, dwoffset);
+ if (++head == ppd->sdma_descq_cnt)
+ head = 0;
+ }
+
+ /* print dma descriptor indices from the TX requests */
+ list_for_each_entry_safe(txp, txpnext, &ppd->sdma_activelist,
+ list)
+ qib_dev_porterr(ppd->dd, ppd->port,
+ "SDMA txp->start_idx: %u txp->next_descq_idx: %u\n",
+ txp->start_idx, txp->next_descq_idx);
+}
+
void qib_sdma_process_event(struct qib_pportdata *ppd,
enum qib_sdma_events event)
{
diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c
index 904c384..092b0bb 100644
--- a/drivers/infiniband/hw/qib/qib_verbs.c
+++ b/drivers/infiniband/hw/qib/qib_verbs.c
@@ -645,9 +645,11 @@
} else
goto drop;
- opcode = be32_to_cpu(ohdr->bth[0]) >> 24;
- ibp->opstats[opcode & 0x7f].n_bytes += tlen;
- ibp->opstats[opcode & 0x7f].n_packets++;
+ opcode = (be32_to_cpu(ohdr->bth[0]) >> 24) & 0x7f;
+#ifdef CONFIG_DEBUG_FS
+ rcd->opstats->stats[opcode].n_bytes += tlen;
+ rcd->opstats->stats[opcode].n_packets++;
+#endif
/* Get the destination QP number. */
qp_num = be32_to_cpu(ohdr->bth[1]) & QIB_QPN_MASK;
diff --git a/drivers/infiniband/hw/qib/qib_verbs.h b/drivers/infiniband/hw/qib/qib_verbs.h
index aff8b2c..012e2c7 100644
--- a/drivers/infiniband/hw/qib/qib_verbs.h
+++ b/drivers/infiniband/hw/qib/qib_verbs.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2012 Intel Corporation. All rights reserved.
+ * Copyright (c) 2012, 2013 Intel Corporation. All rights reserved.
* Copyright (c) 2006 - 2012 QLogic Corporation. All rights reserved.
* Copyright (c) 2005, 2006 PathScale, Inc. All rights reserved.
*
@@ -41,6 +41,7 @@
#include <linux/interrupt.h>
#include <linux/kref.h>
#include <linux/workqueue.h>
+#include <linux/kthread.h>
#include <linux/completion.h>
#include <rdma/ib_pack.h>
#include <rdma/ib_user_verbs.h>
@@ -267,7 +268,8 @@
*/
struct qib_cq {
struct ib_cq ibcq;
- struct work_struct comptask;
+ struct kthread_work comptask;
+ struct qib_devdata *dd;
spinlock_t lock; /* protect changes in this struct */
u8 notify;
u8 triggered;
@@ -658,6 +660,10 @@
u64 n_bytes; /* total number of bytes */
};
+struct qib_opcode_stats_perctx {
+ struct qib_opcode_stats stats[128];
+};
+
struct qib_ibport {
struct qib_qp __rcu *qp0;
struct qib_qp __rcu *qp1;
@@ -724,7 +730,6 @@
u8 vl_high_limit;
u8 sl_to_vl[16];
- struct qib_opcode_stats opstats[128];
};
@@ -768,6 +773,10 @@
spinlock_t n_srqs_lock;
u32 n_mcast_grps_allocated; /* number of mcast groups allocated */
spinlock_t n_mcast_grps_lock;
+#ifdef CONFIG_DEBUG_FS
+ /* per HCA debugfs */
+ struct dentry *qib_ibdev_dbg;
+#endif
};
struct qib_verbs_counters {
@@ -832,8 +841,6 @@
!(qp->s_flags & QIB_S_ANY_WAIT_SEND));
}
-extern struct workqueue_struct *qib_cq_wq;
-
/*
* This must be called with s_lock held.
*/
@@ -910,6 +917,18 @@
void qib_free_qpn_table(struct qib_qpn_table *qpt);
+#ifdef CONFIG_DEBUG_FS
+
+struct qib_qp_iter;
+
+struct qib_qp_iter *qib_qp_iter_init(struct qib_ibdev *dev);
+
+int qib_qp_iter_next(struct qib_qp_iter *iter);
+
+void qib_qp_iter_print(struct seq_file *s, struct qib_qp_iter *iter);
+
+#endif
+
void qib_get_credit(struct qib_qp *qp, u32 aeth);
unsigned qib_pkt_delay(u32 plen, u8 snd_mult, u8 rcv_mult);
@@ -972,6 +991,10 @@
int qib_destroy_srq(struct ib_srq *ibsrq);
+int qib_cq_init(struct qib_devdata *dd);
+
+void qib_cq_exit(struct qib_devdata *dd);
+
void qib_cq_enter(struct qib_cq *cq, struct ib_wc *entry, int sig);
int qib_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry);
diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c
index 7ccf328..f93baf8 100644
--- a/drivers/infiniband/ulp/srp/ib_srp.c
+++ b/drivers/infiniband/ulp/srp/ib_srp.c
@@ -53,8 +53,8 @@
#define DRV_NAME "ib_srp"
#define PFX DRV_NAME ": "
-#define DRV_VERSION "0.2"
-#define DRV_RELDATE "November 1, 2005"
+#define DRV_VERSION "1.0"
+#define DRV_RELDATE "July 1, 2013"
MODULE_AUTHOR("Roland Dreier");
MODULE_DESCRIPTION("InfiniBand SCSI RDMA Protocol initiator "
@@ -231,14 +231,16 @@
return -ENOMEM;
recv_cq = ib_create_cq(target->srp_host->srp_dev->dev,
- srp_recv_completion, NULL, target, SRP_RQ_SIZE, 0);
+ srp_recv_completion, NULL, target, SRP_RQ_SIZE,
+ target->comp_vector);
if (IS_ERR(recv_cq)) {
ret = PTR_ERR(recv_cq);
goto err;
}
send_cq = ib_create_cq(target->srp_host->srp_dev->dev,
- srp_send_completion, NULL, target, SRP_SQ_SIZE, 0);
+ srp_send_completion, NULL, target, SRP_SQ_SIZE,
+ target->comp_vector);
if (IS_ERR(send_cq)) {
ret = PTR_ERR(send_cq);
goto err_recv_cq;
@@ -542,11 +544,11 @@
WARN_ON_ONCE(target->state != SRP_TARGET_REMOVED);
+ srp_remove_target(target);
+
spin_lock(&target->srp_host->target_lock);
list_del(&target->list);
spin_unlock(&target->srp_host->target_lock);
-
- srp_remove_target(target);
}
static void srp_rport_delete(struct srp_rport *rport)
@@ -1744,18 +1746,24 @@
{
struct srp_target_port *target = host_to_target(scmnd->device->host);
struct srp_request *req = (struct srp_request *) scmnd->host_scribble;
+ int ret;
shost_printk(KERN_ERR, target->scsi_host, "SRP abort called\n");
if (!req || !srp_claim_req(target, req, scmnd))
return FAILED;
- srp_send_tsk_mgmt(target, req->index, scmnd->device->lun,
- SRP_TSK_ABORT_TASK);
+ if (srp_send_tsk_mgmt(target, req->index, scmnd->device->lun,
+ SRP_TSK_ABORT_TASK) == 0)
+ ret = SUCCESS;
+ else if (target->transport_offline)
+ ret = FAST_IO_FAIL;
+ else
+ ret = FAILED;
srp_free_req(target, req, scmnd, 0);
scmnd->result = DID_ABORT << 16;
scmnd->scsi_done(scmnd);
- return SUCCESS;
+ return ret;
}
static int srp_reset_device(struct scsi_cmnd *scmnd)
@@ -1891,6 +1899,14 @@
return sprintf(buf, "%s\n", target->srp_host->srp_dev->dev->name);
}
+static ssize_t show_comp_vector(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct srp_target_port *target = host_to_target(class_to_shost(dev));
+
+ return sprintf(buf, "%d\n", target->comp_vector);
+}
+
static ssize_t show_cmd_sg_entries(struct device *dev,
struct device_attribute *attr, char *buf)
{
@@ -1917,6 +1933,7 @@
static DEVICE_ATTR(zero_req_lim, S_IRUGO, show_zero_req_lim, NULL);
static DEVICE_ATTR(local_ib_port, S_IRUGO, show_local_ib_port, NULL);
static DEVICE_ATTR(local_ib_device, S_IRUGO, show_local_ib_device, NULL);
+static DEVICE_ATTR(comp_vector, S_IRUGO, show_comp_vector, NULL);
static DEVICE_ATTR(cmd_sg_entries, S_IRUGO, show_cmd_sg_entries, NULL);
static DEVICE_ATTR(allow_ext_sg, S_IRUGO, show_allow_ext_sg, NULL);
@@ -1931,6 +1948,7 @@
&dev_attr_zero_req_lim,
&dev_attr_local_ib_port,
&dev_attr_local_ib_device,
+ &dev_attr_comp_vector,
&dev_attr_cmd_sg_entries,
&dev_attr_allow_ext_sg,
NULL
@@ -1946,6 +1964,7 @@
.eh_abort_handler = srp_abort,
.eh_device_reset_handler = srp_reset_device,
.eh_host_reset_handler = srp_reset_host,
+ .skip_settle_delay = true,
.sg_tablesize = SRP_DEF_SG_TABLESIZE,
.can_queue = SRP_CMD_SQ_SIZE,
.this_id = -1,
@@ -2001,6 +2020,36 @@
.dev_release = srp_release_dev
};
+/**
+ * srp_conn_unique() - check whether the connection to a target is unique
+ */
+static bool srp_conn_unique(struct srp_host *host,
+ struct srp_target_port *target)
+{
+ struct srp_target_port *t;
+ bool ret = false;
+
+ if (target->state == SRP_TARGET_REMOVED)
+ goto out;
+
+ ret = true;
+
+ spin_lock(&host->target_lock);
+ list_for_each_entry(t, &host->target_list, list) {
+ if (t != target &&
+ target->id_ext == t->id_ext &&
+ target->ioc_guid == t->ioc_guid &&
+ target->initiator_ext == t->initiator_ext) {
+ ret = false;
+ break;
+ }
+ }
+ spin_unlock(&host->target_lock);
+
+out:
+ return ret;
+}
+
/*
* Target ports are added by writing
*
@@ -2023,6 +2072,7 @@
SRP_OPT_CMD_SG_ENTRIES = 1 << 9,
SRP_OPT_ALLOW_EXT_SG = 1 << 10,
SRP_OPT_SG_TABLESIZE = 1 << 11,
+ SRP_OPT_COMP_VECTOR = 1 << 12,
SRP_OPT_ALL = (SRP_OPT_ID_EXT |
SRP_OPT_IOC_GUID |
SRP_OPT_DGID |
@@ -2043,6 +2093,7 @@
{ SRP_OPT_CMD_SG_ENTRIES, "cmd_sg_entries=%u" },
{ SRP_OPT_ALLOW_EXT_SG, "allow_ext_sg=%u" },
{ SRP_OPT_SG_TABLESIZE, "sg_tablesize=%u" },
+ { SRP_OPT_COMP_VECTOR, "comp_vector=%u" },
{ SRP_OPT_ERR, NULL }
};
@@ -2198,6 +2249,14 @@
target->sg_tablesize = token;
break;
+ case SRP_OPT_COMP_VECTOR:
+ if (match_int(args, &token) || token < 0) {
+ pr_warn("bad comp_vector parameter '%s'\n", p);
+ goto out;
+ }
+ target->comp_vector = token;
+ break;
+
default:
pr_warn("unknown parameter or missing value '%s' in target creation request\n",
p);
@@ -2257,6 +2316,16 @@
if (ret)
goto err;
+ if (!srp_conn_unique(target->srp_host, target)) {
+ shost_printk(KERN_INFO, target->scsi_host,
+ PFX "Already connected to target port with id_ext=%016llx;ioc_guid=%016llx;initiator_ext=%016llx\n",
+ be64_to_cpu(target->id_ext),
+ be64_to_cpu(target->ioc_guid),
+ be64_to_cpu(target->initiator_ext));
+ ret = -EEXIST;
+ goto err;
+ }
+
if (!host->srp_dev->fmr_pool && !target->allow_ext_sg &&
target->cmd_sg_cnt < target->sg_tablesize) {
pr_warn("No FMR pool and no external indirect descriptors, limiting sg_tablesize to cmd_sg_cnt\n");
@@ -2507,6 +2576,8 @@
struct srp_target_port *target;
srp_dev = ib_get_client_data(device, &srp_client);
+ if (!srp_dev)
+ return;
list_for_each_entry_safe(host, tmp_host, &srp_dev->dev_list, list) {
device_unregister(&host->dev);
diff --git a/drivers/infiniband/ulp/srp/ib_srp.h b/drivers/infiniband/ulp/srp/ib_srp.h
index 66fbedd..e641088 100644
--- a/drivers/infiniband/ulp/srp/ib_srp.h
+++ b/drivers/infiniband/ulp/srp/ib_srp.h
@@ -156,6 +156,7 @@
char target_name[32];
unsigned int scsi_id;
unsigned int sg_tablesize;
+ int comp_vector;
struct ib_sa_path_rec path;
__be16 orig_dgid[8];
diff --git a/include/linux/socket.h b/include/linux/socket.h
index b10ce4b..230c04b 100644
--- a/include/linux/socket.h
+++ b/include/linux/socket.h
@@ -167,6 +167,7 @@
#define AF_PPPOX 24 /* PPPoX sockets */
#define AF_WANPIPE 25 /* Wanpipe API Sockets */
#define AF_LLC 26 /* Linux LLC */
+#define AF_IB 27 /* Native InfiniBand address */
#define AF_CAN 29 /* Controller Area Network */
#define AF_TIPC 30 /* TIPC sockets */
#define AF_BLUETOOTH 31 /* Bluetooth sockets */
@@ -211,6 +212,7 @@
#define PF_PPPOX AF_PPPOX
#define PF_WANPIPE AF_WANPIPE
#define PF_LLC AF_LLC
+#define PF_IB AF_IB
#define PF_CAN AF_CAN
#define PF_TIPC AF_TIPC
#define PF_BLUETOOTH AF_BLUETOOTH
diff --git a/include/rdma/ib.h b/include/rdma/ib.h
new file mode 100644
index 0000000..cf8f9e7
--- /dev/null
+++ b/include/rdma/ib.h
@@ -0,0 +1,89 @@
+/*
+ * Copyright (c) 2010 Intel Corporation. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#if !defined(_RDMA_IB_H)
+#define _RDMA_IB_H
+
+#include <linux/types.h>
+
+struct ib_addr {
+ union {
+ __u8 uib_addr8[16];
+ __be16 uib_addr16[8];
+ __be32 uib_addr32[4];
+ __be64 uib_addr64[2];
+ } ib_u;
+#define sib_addr8 ib_u.uib_addr8
+#define sib_addr16 ib_u.uib_addr16
+#define sib_addr32 ib_u.uib_addr32
+#define sib_addr64 ib_u.uib_addr64
+#define sib_raw ib_u.uib_addr8
+#define sib_subnet_prefix ib_u.uib_addr64[0]
+#define sib_interface_id ib_u.uib_addr64[1]
+};
+
+static inline int ib_addr_any(const struct ib_addr *a)
+{
+ return ((a->sib_addr64[0] | a->sib_addr64[1]) == 0);
+}
+
+static inline int ib_addr_loopback(const struct ib_addr *a)
+{
+ return ((a->sib_addr32[0] | a->sib_addr32[1] |
+ a->sib_addr32[2] | (a->sib_addr32[3] ^ htonl(1))) == 0);
+}
+
+static inline void ib_addr_set(struct ib_addr *addr,
+ __be32 w1, __be32 w2, __be32 w3, __be32 w4)
+{
+ addr->sib_addr32[0] = w1;
+ addr->sib_addr32[1] = w2;
+ addr->sib_addr32[2] = w3;
+ addr->sib_addr32[3] = w4;
+}
+
+static inline int ib_addr_cmp(const struct ib_addr *a1, const struct ib_addr *a2)
+{
+ return memcmp(a1, a2, sizeof(struct ib_addr));
+}
+
+struct sockaddr_ib {
+ unsigned short int sib_family; /* AF_IB */
+ __be16 sib_pkey;
+ __be32 sib_flowinfo;
+ struct ib_addr sib_addr;
+ __be64 sib_sid;
+ __be64 sib_sid_mask;
+ __u64 sib_scope_id;
+};
+
+#endif /* _RDMA_IB_H */
diff --git a/include/rdma/ib_addr.h b/include/rdma/ib_addr.h
index 9996539..f3ac0f2 100644
--- a/include/rdma/ib_addr.h
+++ b/include/rdma/ib_addr.h
@@ -102,11 +102,7 @@
int rdma_copy_addr(struct rdma_dev_addr *dev_addr, struct net_device *dev,
const unsigned char *dst_dev_addr);
-static inline int ip_addr_size(struct sockaddr *addr)
-{
- return addr->sa_family == AF_INET6 ?
- sizeof(struct sockaddr_in6) : sizeof(struct sockaddr_in);
-}
+int rdma_addr_size(struct sockaddr *addr);
static inline u16 ib_addr_get_pkey(struct rdma_dev_addr *dev_addr)
{
diff --git a/include/rdma/ib_sa.h b/include/rdma/ib_sa.h
index 8275e53..125f871 100644
--- a/include/rdma/ib_sa.h
+++ b/include/rdma/ib_sa.h
@@ -402,6 +402,12 @@
struct ib_ah_attr *ah_attr);
/**
+ * ib_sa_pack_path - Conert a path record from struct ib_sa_path_rec
+ * to IB MAD wire format.
+ */
+void ib_sa_pack_path(struct ib_sa_path_rec *rec, void *attribute);
+
+/**
* ib_sa_unpack_path - Convert a path record from MAD format to struct
* ib_sa_path_rec.
*/
@@ -418,4 +424,5 @@
void *context),
void *context,
struct ib_sa_query **sa_query);
+
#endif /* IB_SA_H */
diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h
index ad3a314..1ed2088 100644
--- a/include/rdma/rdma_cm.h
+++ b/include/rdma/rdma_cm.h
@@ -70,6 +70,11 @@
RDMA_PS_UDP = 0x0111,
};
+#define RDMA_IB_IP_PS_MASK 0xFFFFFFFFFFFF0000ULL
+#define RDMA_IB_IP_PS_TCP 0x0000000001060000ULL
+#define RDMA_IB_IP_PS_UDP 0x0000000001110000ULL
+#define RDMA_IB_IP_PS_IB 0x00000000013F0000ULL
+
struct rdma_addr {
struct sockaddr_storage src_addr;
struct sockaddr_storage dst_addr;
@@ -93,6 +98,7 @@
/* Fields below ignored if a QP is created on the rdma_cm_id. */
u8 srq;
u32 qp_num;
+ u32 qkey;
};
struct rdma_ud_param {
@@ -367,4 +373,11 @@
*/
int rdma_set_afonly(struct rdma_cm_id *id, int afonly);
+ /**
+ * rdma_get_service_id - Return the IB service ID for a specified address.
+ * @id: Communication identifier associated with the address.
+ * @addr: Address for the service ID.
+ */
+__be64 rdma_get_service_id(struct rdma_cm_id *id, struct sockaddr *addr);
+
#endif /* RDMA_CM_H */
diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h
index 1ee9239..99b80ab 100644
--- a/include/uapi/rdma/rdma_user_cm.h
+++ b/include/uapi/rdma/rdma_user_cm.h
@@ -45,8 +45,8 @@
enum {
RDMA_USER_CM_CMD_CREATE_ID,
RDMA_USER_CM_CMD_DESTROY_ID,
- RDMA_USER_CM_CMD_BIND_ADDR,
- RDMA_USER_CM_CMD_RESOLVE_ADDR,
+ RDMA_USER_CM_CMD_BIND_IP,
+ RDMA_USER_CM_CMD_RESOLVE_IP,
RDMA_USER_CM_CMD_RESOLVE_ROUTE,
RDMA_USER_CM_CMD_QUERY_ROUTE,
RDMA_USER_CM_CMD_CONNECT,
@@ -59,9 +59,13 @@
RDMA_USER_CM_CMD_GET_OPTION,
RDMA_USER_CM_CMD_SET_OPTION,
RDMA_USER_CM_CMD_NOTIFY,
- RDMA_USER_CM_CMD_JOIN_MCAST,
+ RDMA_USER_CM_CMD_JOIN_IP_MCAST,
RDMA_USER_CM_CMD_LEAVE_MCAST,
- RDMA_USER_CM_CMD_MIGRATE_ID
+ 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_JOIN_MCAST
};
/*
@@ -95,28 +99,51 @@
__u32 events_reported;
};
-struct rdma_ucm_bind_addr {
+struct rdma_ucm_bind_ip {
__u64 response;
struct sockaddr_in6 addr;
__u32 id;
};
-struct rdma_ucm_resolve_addr {
+struct rdma_ucm_bind {
+ __u32 id;
+ __u16 addr_size;
+ __u16 reserved;
+ struct sockaddr_storage addr;
+};
+
+struct rdma_ucm_resolve_ip {
struct sockaddr_in6 src_addr;
struct sockaddr_in6 dst_addr;
__u32 id;
__u32 timeout_ms;
};
+struct rdma_ucm_resolve_addr {
+ __u32 id;
+ __u32 timeout_ms;
+ __u16 src_size;
+ __u16 dst_size;
+ __u32 reserved;
+ struct sockaddr_storage src_addr;
+ struct sockaddr_storage dst_addr;
+};
+
struct rdma_ucm_resolve_route {
__u32 id;
__u32 timeout_ms;
};
-struct rdma_ucm_query_route {
+enum {
+ RDMA_USER_CM_QUERY_ADDR,
+ RDMA_USER_CM_QUERY_PATH,
+ RDMA_USER_CM_QUERY_GID
+};
+
+struct rdma_ucm_query {
__u64 response;
__u32 id;
- __u32 reserved;
+ __u32 option;
};
struct rdma_ucm_query_route_resp {
@@ -129,9 +156,26 @@
__u8 reserved[3];
};
+struct rdma_ucm_query_addr_resp {
+ __u64 node_guid;
+ __u8 port_num;
+ __u8 reserved;
+ __u16 pkey;
+ __u16 src_size;
+ __u16 dst_size;
+ struct sockaddr_storage src_addr;
+ struct sockaddr_storage dst_addr;
+};
+
+struct rdma_ucm_query_path_resp {
+ __u32 num_paths;
+ __u32 reserved;
+ struct ib_path_rec_data path_data[0];
+};
+
struct rdma_ucm_conn_param {
__u32 qp_num;
- __u32 reserved;
+ __u32 qkey;
__u8 private_data[RDMA_MAX_PRIVATE_DATA];
__u8 private_data_len;
__u8 srq;
@@ -192,13 +236,22 @@
__u32 event;
};
-struct rdma_ucm_join_mcast {
+struct rdma_ucm_join_ip_mcast {
__u64 response; /* rdma_ucm_create_id_resp */
__u64 uid;
struct sockaddr_in6 addr;
__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;
};