#include <asm/div64.h>
static lnd_t the_o2iblnd = {
- .lnd_type = O2IBLND,
- .lnd_startup = kiblnd_startup,
- .lnd_shutdown = kiblnd_shutdown,
- .lnd_ctl = kiblnd_ctl,
- .lnd_query = kiblnd_query,
- .lnd_send = kiblnd_send,
- .lnd_recv = kiblnd_recv,
+ .lnd_type = O2IBLND,
+ .lnd_startup = kiblnd_startup,
+ .lnd_shutdown = kiblnd_shutdown,
+ .lnd_ctl = kiblnd_ctl,
+ .lnd_query = kiblnd_query,
+ .lnd_send = kiblnd_send,
+ .lnd_recv = kiblnd_recv,
};
-kib_data_t kiblnd_data;
+kib_data_t kiblnd_data;
static __u32 kiblnd_cksum(void *ptr, int nob)
{
- char *c = ptr;
- __u32 sum = 0;
+ char *c = ptr;
+ __u32 sum = 0;
while (nob-- > 0)
sum = ((sum << 1) | (sum >> 31)) + *c++;
static int kiblnd_unpack_rd(kib_msg_t *msg, int flip)
{
- kib_rdma_desc_t *rd;
- int nob;
- int n;
- int i;
+ kib_rdma_desc_t *rd;
+ int nob;
+ int n;
+ int i;
LASSERT(msg->ibm_type == IBLND_MSG_GET_REQ ||
msg->ibm_type == IBLND_MSG_PUT_ACK);
int kiblnd_unpack_msg(kib_msg_t *msg, int nob)
{
const int hdr_size = offsetof(kib_msg_t, ibm_u);
- __u32 msg_cksum;
- __u16 version;
- int msg_nob;
- int flip;
+ __u32 msg_cksum;
+ __u16 version;
+ int msg_nob;
+ int flip;
/* 6 bytes are enough to have received magic + version */
if (nob < 6) {
int kiblnd_create_peer(lnet_ni_t *ni, kib_peer_t **peerp, lnet_nid_t nid)
{
- kib_peer_t *peer;
- kib_net_t *net = ni->ni_data;
- int cpt = lnet_cpt_of_nid(nid);
- unsigned long flags;
+ kib_peer_t *peer;
+ kib_net_t *net = ni->ni_data;
+ int cpt = lnet_cpt_of_nid(nid);
+ unsigned long flags;
LASSERT(net != NULL);
LASSERT(nid != LNET_NID_ANY);
{
/* the caller is responsible for accounting the additional reference
* that this creates */
- struct list_head *peer_list = kiblnd_nid2peerlist(nid);
- struct list_head *tmp;
- kib_peer_t *peer;
+ struct list_head *peer_list = kiblnd_nid2peerlist(nid);
+ struct list_head *tmp;
+ kib_peer_t *peer;
list_for_each(tmp, peer_list) {
static int kiblnd_get_peer_info(lnet_ni_t *ni, int index,
lnet_nid_t *nidp, int *count)
{
- kib_peer_t *peer;
- struct list_head *ptmp;
- int i;
- unsigned long flags;
+ kib_peer_t *peer;
+ struct list_head *ptmp;
+ int i;
+ unsigned long flags;
read_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
static void kiblnd_del_peer_locked(kib_peer_t *peer)
{
- struct list_head *ctmp;
- struct list_head *cnxt;
- kib_conn_t *conn;
+ struct list_head *ctmp;
+ struct list_head *cnxt;
+ kib_conn_t *conn;
if (list_empty(&peer->ibp_conns)) {
kiblnd_unlink_peer_locked(peer);
static int kiblnd_del_peer(lnet_ni_t *ni, lnet_nid_t nid)
{
LIST_HEAD(zombies);
- struct list_head *ptmp;
- struct list_head *pnxt;
- kib_peer_t *peer;
- int lo;
- int hi;
- int i;
- unsigned long flags;
- int rc = -ENOENT;
+ struct list_head *ptmp;
+ struct list_head *pnxt;
+ kib_peer_t *peer;
+ int lo;
+ int hi;
+ int i;
+ unsigned long flags;
+ int rc = -ENOENT;
write_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
static kib_conn_t *kiblnd_get_conn_by_idx(lnet_ni_t *ni, int index)
{
- kib_peer_t *peer;
- struct list_head *ptmp;
- kib_conn_t *conn;
- struct list_head *ctmp;
- int i;
- unsigned long flags;
+ kib_peer_t *peer;
+ struct list_head *ptmp;
+ kib_conn_t *conn;
+ struct list_head *ctmp;
+ int i;
+ unsigned long flags;
read_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
static void kiblnd_setup_mtu_locked(struct rdma_cm_id *cmid)
{
- int mtu;
+ int mtu;
/* XXX There is no path record for iWARP, set by netdev->change_mtu? */
if (cmid->route.path_rec == NULL)
static int kiblnd_get_completion_vector(kib_conn_t *conn, int cpt)
{
- cpumask_t *mask;
- int vectors;
- int off;
- int i;
- lnet_nid_t nid = conn->ibc_peer->ibp_nid;
+ cpumask_t *mask;
+ int vectors;
+ int off;
+ int i;
+ lnet_nid_t nid = conn->ibc_peer->ibp_nid;
vectors = conn->ibc_cmid->device->num_comp_vectors;
if (vectors <= 1)
* she must dispose of 'cmid'. (Actually I'd block forever if I tried
* to destroy 'cmid' here since I'm called from the CM which still has
* its ref on 'cmid'). */
- rwlock_t *glock = &kiblnd_data.kib_global_lock;
- kib_net_t *net = peer->ibp_ni->ni_data;
- kib_dev_t *dev;
+ rwlock_t *glock = &kiblnd_data.kib_global_lock;
+ kib_net_t *net = peer->ibp_ni->ni_data;
+ kib_dev_t *dev;
struct ib_qp_init_attr *init_qp_attr;
- struct kib_sched_info *sched;
- kib_conn_t *conn;
- struct ib_cq *cq;
- unsigned long flags;
- int cpt;
- int rc;
- int i;
+ struct kib_sched_info *sched;
+ kib_conn_t *conn;
+ struct ib_cq *cq;
+ unsigned long flags;
+ int cpt;
+ int rc;
+ int i;
LASSERT(net != NULL);
LASSERT(!in_interrupt());
void kiblnd_destroy_conn(kib_conn_t *conn)
{
struct rdma_cm_id *cmid = conn->ibc_cmid;
- kib_peer_t *peer = conn->ibc_peer;
- int rc;
+ kib_peer_t *peer = conn->ibc_peer;
+ int rc;
LASSERT(!in_interrupt());
LASSERT(atomic_read(&conn->ibc_refcount) == 0);
int kiblnd_close_peer_conns_locked(kib_peer_t *peer, int why)
{
- kib_conn_t *conn;
- struct list_head *ctmp;
- struct list_head *cnxt;
- int count = 0;
+ kib_conn_t *conn;
+ struct list_head *ctmp;
+ struct list_head *cnxt;
+ int count = 0;
list_for_each_safe(ctmp, cnxt, &peer->ibp_conns) {
conn = list_entry(ctmp, kib_conn_t, ibc_list);
int kiblnd_close_stale_conns_locked(kib_peer_t *peer,
int version, __u64 incarnation)
{
- kib_conn_t *conn;
- struct list_head *ctmp;
- struct list_head *cnxt;
- int count = 0;
+ kib_conn_t *conn;
+ struct list_head *ctmp;
+ struct list_head *cnxt;
+ int count = 0;
list_for_each_safe(ctmp, cnxt, &peer->ibp_conns) {
conn = list_entry(ctmp, kib_conn_t, ibc_list);
static int kiblnd_close_matching_conns(lnet_ni_t *ni, lnet_nid_t nid)
{
- kib_peer_t *peer;
- struct list_head *ptmp;
- struct list_head *pnxt;
- int lo;
- int hi;
- int i;
- unsigned long flags;
- int count = 0;
+ kib_peer_t *peer;
+ struct list_head *ptmp;
+ struct list_head *pnxt;
+ int lo;
+ int hi;
+ int i;
+ unsigned long flags;
+ int count = 0;
write_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
int kiblnd_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg)
{
struct libcfs_ioctl_data *data = arg;
- int rc = -EINVAL;
+ int rc = -EINVAL;
switch (cmd) {
case IOC_LIBCFS_GET_PEER: {
- lnet_nid_t nid = 0;
- int count = 0;
+ lnet_nid_t nid = 0;
+ int count = 0;
rc = kiblnd_get_peer_info(ni, data->ioc_count,
&nid, &count);
- data->ioc_nid = nid;
- data->ioc_count = count;
+ data->ioc_nid = nid;
+ data->ioc_count = count;
break;
}
void kiblnd_query(lnet_ni_t *ni, lnet_nid_t nid, unsigned long *when)
{
- unsigned long last_alive = 0;
- unsigned long now = cfs_time_current();
- rwlock_t *glock = &kiblnd_data.kib_global_lock;
- kib_peer_t *peer;
- unsigned long flags;
+ unsigned long last_alive = 0;
+ unsigned long now = cfs_time_current();
+ rwlock_t *glock = &kiblnd_data.kib_global_lock;
+ kib_peer_t *peer;
+ unsigned long flags;
read_lock_irqsave(glock, flags);
void kiblnd_free_pages(kib_pages_t *p)
{
- int npages = p->ibp_npages;
- int i;
+ int npages = p->ibp_npages;
+ int i;
for (i = 0; i < npages; i++) {
if (p->ibp_pages[i] != NULL)
int kiblnd_alloc_pages(kib_pages_t **pp, int cpt, int npages)
{
- kib_pages_t *p;
- int i;
+ kib_pages_t *p;
+ int i;
LIBCFS_CPT_ALLOC(p, lnet_cpt_table(), cpt,
offsetof(kib_pages_t, ibp_pages[npages]));
void kiblnd_unmap_rx_descs(kib_conn_t *conn)
{
kib_rx_t *rx;
- int i;
+ int i;
LASSERT(conn->ibc_rxs != NULL);
LASSERT(conn->ibc_hdev != NULL);
void kiblnd_map_rx_descs(kib_conn_t *conn)
{
- kib_rx_t *rx;
- struct page *pg;
- int pg_off;
- int ipg;
- int i;
+ kib_rx_t *rx;
+ struct page *pg;
+ int pg_off;
+ int ipg;
+ int i;
- for (pg_off = ipg = i = 0;
- i < IBLND_RX_MSGS(conn->ibc_version); i++) {
+ for (pg_off = ipg = i = 0; i < IBLND_RX_MSGS(conn->ibc_version); i++) {
pg = conn->ibc_rx_pages->ibp_pages[ipg];
rx = &conn->ibc_rxs[i];
static void kiblnd_unmap_tx_pool(kib_tx_pool_t *tpo)
{
- kib_hca_dev_t *hdev = tpo->tpo_hdev;
- kib_tx_t *tx;
- int i;
+ kib_hca_dev_t *hdev = tpo->tpo_hdev;
+ kib_tx_t *tx;
+ int i;
LASSERT(tpo->tpo_pool.po_allocated == 0);
static kib_hca_dev_t *kiblnd_current_hdev(kib_dev_t *dev)
{
kib_hca_dev_t *hdev;
- unsigned long flags;
- int i = 0;
+ unsigned long flags;
+ int i = 0;
read_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
while (dev->ibd_failover) {
static void kiblnd_map_tx_pool(kib_tx_pool_t *tpo)
{
- kib_pages_t *txpgs = tpo->tpo_tx_pages;
- kib_pool_t *pool = &tpo->tpo_pool;
- kib_net_t *net = pool->po_owner->ps_net;
- kib_dev_t *dev;
- struct page *page;
- kib_tx_t *tx;
- int page_offset;
- int ipage;
- int i;
+ kib_pages_t *txpgs = tpo->tpo_tx_pages;
+ kib_pool_t *pool = &tpo->tpo_pool;
+ kib_net_t *net = pool->po_owner->ps_net;
+ kib_dev_t *dev;
+ struct page *page;
+ kib_tx_t *tx;
+ int page_offset;
+ int ipage;
+ int i;
LASSERT(net != NULL);
struct ib_mr *kiblnd_find_dma_mr(kib_hca_dev_t *hdev, __u64 addr, __u64 size)
{
- __u64 index;
+ __u64 index;
LASSERT(hdev->ibh_mrs[0] != NULL);
{
struct ib_mr *prev_mr;
struct ib_mr *mr;
- int i;
+ int i;
LASSERT(hdev->ibh_mrs[0] != NULL);
kib_fmr_pool_t **pp_fpo)
{
/* FMR pool for RDMA */
- kib_dev_t *dev = fps->fps_net->ibn_dev;
- kib_fmr_pool_t *fpo;
+ kib_dev_t *dev = fps->fps_net->ibn_dev;
+ kib_fmr_pool_t *fpo;
struct ib_fmr_pool_param param = {
.max_pages_per_fmr = LNET_MAX_PAYLOAD/PAGE_SIZE,
- .page_shift = PAGE_SHIFT,
- .access = (IB_ACCESS_LOCAL_WRITE |
- IB_ACCESS_REMOTE_WRITE),
- .pool_size = fps->fps_pool_size,
+ .page_shift = PAGE_SHIFT,
+ .access = (IB_ACCESS_LOCAL_WRITE |
+ IB_ACCESS_REMOTE_WRITE),
+ .pool_size = fps->fps_pool_size,
.dirty_watermark = fps->fps_flush_trigger,
.flush_function = NULL,
- .flush_arg = NULL,
- .cache = !!*kiblnd_tunables.kib_fmr_cache};
+ .flush_arg = NULL,
+ .cache = !!*kiblnd_tunables.kib_fmr_cache};
int rc;
LIBCFS_CPT_ALLOC(fpo, lnet_cpt_table(), fps->fps_cpt, sizeof(*fpo));
int flush_trigger)
{
kib_fmr_pool_t *fpo;
- int rc;
+ int rc;
memset(fps, 0, sizeof(kib_fmr_poolset_t));
void kiblnd_fmr_pool_unmap(kib_fmr_t *fmr, int status)
{
LIST_HEAD(zombies);
- kib_fmr_pool_t *fpo = fmr->fmr_pool;
+ kib_fmr_pool_t *fpo = fmr->fmr_pool;
kib_fmr_poolset_t *fps = fpo->fpo_owner;
- unsigned long now = cfs_time_current();
- kib_fmr_pool_t *tmp;
- int rc;
+ unsigned long now = cfs_time_current();
+ kib_fmr_pool_t *tmp;
+ int rc;
rc = ib_fmr_pool_unmap(fmr->fmr_pfmr);
LASSERT(rc == 0);
__u64 iov, kib_fmr_t *fmr)
{
struct ib_pool_fmr *pfmr;
- kib_fmr_pool_t *fpo;
- __u64 version;
- int rc;
+ kib_fmr_pool_t *fpo;
+ __u64 version;
+ int rc;
again:
spin_lock(&fps->fps_lock);
kib_ps_node_init_t nd_init,
kib_ps_node_fini_t nd_fini)
{
- kib_pool_t *pool;
- int rc;
+ kib_pool_t *pool;
+ int rc;
memset(ps, 0, sizeof(kib_poolset_t));
- ps->ps_cpt = cpt;
- ps->ps_net = net;
+ ps->ps_cpt = cpt;
+ ps->ps_net = net;
ps->ps_pool_create = po_create;
ps->ps_pool_destroy = po_destroy;
ps->ps_node_init = nd_init;
void kiblnd_pool_free_node(kib_pool_t *pool, struct list_head *node)
{
LIST_HEAD(zombies);
- kib_poolset_t *ps = pool->po_owner;
- kib_pool_t *tmp;
- unsigned long now = cfs_time_current();
+ kib_poolset_t *ps = pool->po_owner;
+ kib_pool_t *tmp;
+ unsigned long now = cfs_time_current();
spin_lock(&ps->ps_lock);
struct list_head *kiblnd_pool_alloc_node(kib_poolset_t *ps)
{
- struct list_head *node;
- kib_pool_t *pool;
- int rc;
+ struct list_head *node;
+ kib_pool_t *pool;
+ int rc;
again:
spin_lock(&ps->ps_lock);
void kiblnd_pmr_pool_unmap(kib_phys_mr_t *pmr)
{
- kib_pmr_pool_t *ppo = pmr->pmr_pool;
- struct ib_mr *mr = pmr->pmr_mr;
+ kib_pmr_pool_t *ppo = pmr->pmr_pool;
+ struct ib_mr *mr = pmr->pmr_mr;
pmr->pmr_mr = NULL;
kiblnd_pool_free_node(&ppo->ppo_pool, &pmr->pmr_list);
kib_rdma_desc_t *rd, __u64 *iova, kib_phys_mr_t **pp_pmr)
{
kib_phys_mr_t *pmr;
- struct list_head *node;
- int rc;
- int i;
+ struct list_head *node;
+ int rc;
+ int i;
node = kiblnd_pool_alloc_node(&pps->pps_poolset);
if (node == NULL) {
static void kiblnd_destroy_pmr_pool(kib_pool_t *pool)
{
kib_pmr_pool_t *ppo = container_of(pool, kib_pmr_pool_t, ppo_pool);
- kib_phys_mr_t *pmr;
+ kib_phys_mr_t *pmr;
kib_phys_mr_t *tmp;
LASSERT(pool->po_allocated == 0);
static int kiblnd_create_pmr_pool(kib_poolset_t *ps, int size,
kib_pool_t **pp_po)
{
- struct kib_pmr_pool *ppo;
- struct kib_pool *pool;
- kib_phys_mr_t *pmr;
- int i;
+ struct kib_pmr_pool *ppo;
+ struct kib_pool *pool;
+ kib_phys_mr_t *pmr;
+ int i;
LIBCFS_CPT_ALLOC(ppo, lnet_cpt_table(),
ps->ps_cpt, sizeof(kib_pmr_pool_t));
static void kiblnd_destroy_tx_pool(kib_pool_t *pool)
{
- kib_tx_pool_t *tpo = container_of(pool, kib_tx_pool_t, tpo_pool);
- int i;
+ kib_tx_pool_t *tpo = container_of(pool, kib_tx_pool_t, tpo_pool);
+ int i;
LASSERT(pool->po_allocated == 0);
static int kiblnd_create_tx_pool(kib_poolset_t *ps, int size,
kib_pool_t **pp_po)
{
- int i;
- int npg;
- kib_pool_t *pool;
+ int i;
+ int npg;
+ kib_pool_t *pool;
kib_tx_pool_t *tpo;
LIBCFS_CPT_ALLOC(tpo, lnet_cpt_table(), ps->ps_cpt, sizeof(*tpo));
{
kib_tx_poolset_t *tps = container_of(pool->po_owner, kib_tx_poolset_t,
tps_poolset);
- kib_tx_t *tx = list_entry(node, kib_tx_t, tx_list);
+ kib_tx_t *tx = list_entry(node, kib_tx_t, tx_list);
tx->tx_cookie = tps->tps_next_tx_cookie++;
}
static void kiblnd_net_fini_pools(kib_net_t *net)
{
- int i;
+ int i;
cfs_cpt_for_each(i, lnet_cpt_table()) {
- kib_tx_poolset_t *tps;
- kib_fmr_poolset_t *fps;
- kib_pmr_poolset_t *pps;
+ kib_tx_poolset_t *tps;
+ kib_fmr_poolset_t *fps;
+ kib_pmr_poolset_t *pps;
if (net->ibn_tx_ps != NULL) {
tps = net->ibn_tx_ps[i];
static int kiblnd_net_init_pools(kib_net_t *net, __u32 *cpts, int ncpts)
{
- unsigned long flags;
- int cpt;
- int rc;
- int i;
+ unsigned long flags;
+ int cpt;
+ int rc;
+ int i;
read_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
if (*kiblnd_tunables.kib_map_on_demand == 0 &&
net->ibn_dev->ibd_hdev->ibh_nmrs == 1) {
- read_unlock_irqrestore(&kiblnd_data.kib_global_lock,
- flags);
+ read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
goto create_tx_pool;
}
static int kiblnd_hdev_get_attr(kib_hca_dev_t *hdev)
{
struct ib_device_attr *attr;
- int rc;
+ int rc;
/* It's safe to assume a HCA can handle a page size
* matching that of the native system */
static void kiblnd_hdev_cleanup_mrs(kib_hca_dev_t *hdev)
{
- int i;
+ int i;
if (hdev->ibh_nmrs == 0 || hdev->ibh_mrs == NULL)
return;
static int kiblnd_hdev_setup_mrs(kib_hca_dev_t *hdev)
{
struct ib_mr *mr;
- int i;
- int rc;
- __u64 mm_size;
- __u64 mr_size;
- int acflags = IB_ACCESS_LOCAL_WRITE |
- IB_ACCESS_REMOTE_WRITE;
+ int i;
+ int rc;
+ __u64 mm_size;
+ __u64 mr_size;
+ int acflags = IB_ACCESS_LOCAL_WRITE | IB_ACCESS_REMOTE_WRITE;
rc = kiblnd_hdev_get_attr(hdev);
if (rc != 0)
for (i = 0; i < hdev->ibh_nmrs; i++) {
struct ib_phys_buf ipb;
- __u64 iova;
+ __u64 iova;
ipb.size = hdev->ibh_mr_size;
ipb.addr = i * mr_size;
- iova = ipb.addr;
+ iova = ipb.addr;
mr = ib_reg_phys_mr(hdev->ibh_pd, &ipb, 1, acflags, &iova);
if (IS_ERR(mr)) {
static int kiblnd_dev_need_failover(kib_dev_t *dev)
{
- struct rdma_cm_id *cmid;
- struct sockaddr_in srcaddr;
- struct sockaddr_in dstaddr;
- int rc;
+ struct rdma_cm_id *cmid;
+ struct sockaddr_in srcaddr;
+ struct sockaddr_in dstaddr;
+ int rc;
if (dev->ibd_hdev == NULL || /* initializing */
dev->ibd_hdev->ibh_cmid == NULL || /* listener is dead */
}
memset(&srcaddr, 0, sizeof(srcaddr));
- srcaddr.sin_family = AF_INET;
+ srcaddr.sin_family = AF_INET;
srcaddr.sin_addr.s_addr = (__force u32)htonl(dev->ibd_ifip);
memset(&dstaddr, 0, sizeof(dstaddr));
LIST_HEAD(zombie_tpo);
LIST_HEAD(zombie_ppo);
LIST_HEAD(zombie_fpo);
- struct rdma_cm_id *cmid = NULL;
- kib_hca_dev_t *hdev = NULL;
- kib_hca_dev_t *old;
- struct ib_pd *pd;
- kib_net_t *net;
- struct sockaddr_in addr;
- unsigned long flags;
- int rc = 0;
- int i;
+ struct rdma_cm_id *cmid = NULL;
+ kib_hca_dev_t *hdev = NULL;
+ kib_hca_dev_t *old;
+ struct ib_pd *pd;
+ kib_net_t *net;
+ struct sockaddr_in addr;
+ unsigned long flags;
+ int rc = 0;
+ int i;
LASSERT(*kiblnd_tunables.kib_dev_failover > 1 ||
dev->ibd_can_failover ||
static kib_dev_t *kiblnd_create_dev(char *ifname)
{
struct net_device *netdev;
- kib_dev_t *dev;
- __u32 netmask;
- __u32 ip;
- int up;
- int rc;
+ kib_dev_t *dev;
+ __u32 netmask;
+ __u32 ip;
+ int up;
+ int rc;
rc = libcfs_ipif_query(ifname, &up, &ip, &netmask);
if (rc != 0) {
static void kiblnd_base_shutdown(void)
{
- struct kib_sched_info *sched;
- int i;
+ struct kib_sched_info *sched;
+ int i;
LASSERT(list_empty(&kiblnd_data.kib_devs));
void kiblnd_shutdown(lnet_ni_t *ni)
{
- kib_net_t *net = ni->ni_data;
- rwlock_t *g_lock = &kiblnd_data.kib_global_lock;
- int i;
- unsigned long flags;
+ kib_net_t *net = ni->ni_data;
+ rwlock_t *g_lock = &kiblnd_data.kib_global_lock;
+ int i;
+ unsigned long flags;
LASSERT(kiblnd_data.kib_init == IBLND_INIT_ALL);
static int kiblnd_base_startup(void)
{
- struct kib_sched_info *sched;
- int rc;
- int i;
+ struct kib_sched_info *sched;
+ int rc;
+ int i;
LASSERT(kiblnd_data.kib_init == IBLND_INIT_NOTHING);
kiblnd_data.kib_peer_hash_size = IBLND_PEER_HASH_SIZE;
LIBCFS_ALLOC(kiblnd_data.kib_peers,
- sizeof(struct list_head) *
- kiblnd_data.kib_peer_hash_size);
+ sizeof(struct list_head) * kiblnd_data.kib_peer_hash_size);
if (kiblnd_data.kib_peers == NULL)
goto failed;
for (i = 0; i < kiblnd_data.kib_peer_hash_size; i++)
goto failed;
cfs_percpt_for_each(sched, i, kiblnd_data.kib_scheds) {
- int nthrs;
+ int nthrs;
spin_lock_init(&sched->ibs_lock);
INIT_LIST_HEAD(&sched->ibs_conns);
static int kiblnd_start_schedulers(struct kib_sched_info *sched)
{
- int rc = 0;
- int nthrs;
- int i;
+ int rc = 0;
+ int nthrs;
+ int i;
if (sched->ibs_nthreads == 0) {
if (*kiblnd_tunables.kib_nscheds > 0) {
}
for (i = 0; i < nthrs; i++) {
- long id;
- char name[20];
+ long id;
+ char name[20];
id = KIB_THREAD_ID(sched->ibs_cpt, sched->ibs_nthreads + i);
snprintf(name, sizeof(name), "kiblnd_sd_%02ld_%02ld",
static int kiblnd_dev_start_threads(kib_dev_t *dev, int newdev, __u32 *cpts,
int ncpts)
{
- int cpt;
- int rc;
- int i;
+ int cpt;
+ int rc;
+ int i;
for (i = 0; i < ncpts; i++) {
struct kib_sched_info *sched;
static kib_dev_t *kiblnd_dev_search(char *ifname)
{
- kib_dev_t *alias = NULL;
- kib_dev_t *dev;
- char *colon;
- char *colon2;
+ kib_dev_t *alias = NULL;
+ kib_dev_t *dev;
+ char *colon;
+ char *colon2;
colon = strchr(ifname, ':');
list_for_each_entry(dev, &kiblnd_data.kib_devs, ibd_list) {
int kiblnd_startup(lnet_ni_t *ni)
{
- char *ifname;
- kib_dev_t *ibdev = NULL;
- kib_net_t *net;
- struct timeval tv;
- unsigned long flags;
- int rc;
- int newdev;
+ char *ifname;
+ kib_dev_t *ibdev = NULL;
+ kib_net_t *net;
+ struct timeval tv;
+ unsigned long flags;
+ int rc;
+ int newdev;
LASSERT(ni->ni_lnd == &the_o2iblnd);
static int __init kiblnd_module_init(void)
{
- int rc;
+ int rc;
CLASSERT(sizeof(kib_msg_t) <= IBLND_MSG_SIZE);
CLASSERT(offsetof(kib_msg_t,
#define IBLND_N_SCHED_HIGH 4
typedef struct {
- int *kib_dev_failover; /* HCA failover */
- unsigned int *kib_service; /* IB service number */
- int *kib_min_reconnect_interval; /* first failed connection retry... */
- int *kib_max_reconnect_interval; /* ...exponentially increasing to this */
- int *kib_cksum; /* checksum kib_msg_t? */
- int *kib_timeout; /* comms timeout (seconds) */
- int *kib_keepalive; /* keepalive timeout (seconds) */
- int *kib_ntx; /* # tx descs */
- int *kib_credits; /* # concurrent sends */
- int *kib_peertxcredits; /* # concurrent sends to 1 peer */
- int *kib_peerrtrcredits; /* # per-peer router buffer credits */
- int *kib_peercredits_hiw; /* # when eagerly to return credits */
- int *kib_peertimeout; /* seconds to consider peer dead */
- char **kib_default_ipif; /* default IPoIB interface */
- int *kib_retry_count;
- int *kib_rnr_retry_count;
- int *kib_concurrent_sends; /* send work queue sizing */
- int *kib_ib_mtu; /* IB MTU */
- int *kib_map_on_demand; /* map-on-demand if RD has more fragments
- * than this value, 0 disable map-on-demand */
- int *kib_pmr_pool_size; /* # physical MR in pool */
- int *kib_fmr_pool_size; /* # FMRs in pool */
- int *kib_fmr_flush_trigger; /* When to trigger FMR flush */
- int *kib_fmr_cache; /* enable FMR pool cache? */
- int *kib_require_priv_port;/* accept only privileged ports */
- int *kib_use_priv_port; /* use privileged port for active connect */
- /* # threads on each CPT */
- int *kib_nscheds;
+ int *kib_dev_failover; /* HCA failover */
+ unsigned int *kib_service; /* IB service number */
+ int *kib_min_reconnect_interval; /* first failed connection
+ * retry... */
+ int *kib_max_reconnect_interval; /* ...exponentially increasing
+ * to this */
+ int *kib_cksum; /* checksum kib_msg_t? */
+ int *kib_timeout; /* comms timeout (seconds) */
+ int *kib_keepalive; /* keepalive timeout (seconds) */
+ int *kib_ntx; /* # tx descs */
+ int *kib_credits; /* # concurrent sends */
+ int *kib_peertxcredits; /* # concurrent sends to 1 peer */
+ int *kib_peerrtrcredits; /* # per-peer router buffer
+ * credits */
+ int *kib_peercredits_hiw; /* # when eagerly to return
+ * credits */
+ int *kib_peertimeout; /* seconds to consider peer dead */
+ char **kib_default_ipif; /* default IPoIB interface */
+ int *kib_retry_count;
+ int *kib_rnr_retry_count;
+ int *kib_concurrent_sends; /* send work queue sizing */
+ int *kib_ib_mtu; /* IB MTU */
+ int *kib_map_on_demand; /* map-on-demand if RD has more
+ * fragments than this value, 0
+ * disable map-on-demand */
+ int *kib_pmr_pool_size; /* # physical MR in pool */
+ int *kib_fmr_pool_size; /* # FMRs in pool */
+ int *kib_fmr_flush_trigger; /* When to trigger FMR flush */
+ int *kib_fmr_cache; /* enable FMR pool cache? */
+ int *kib_require_priv_port; /* accept only privileged ports */
+ int *kib_use_priv_port; /* use privileged port for active
+ * connect */
+ int *kib_nscheds; /* # threads on each CPT */
} kib_tunables_t;
extern kib_tunables_t kiblnd_tunables;
-#define IBLND_MSG_QUEUE_SIZE_V1 8 /* V1 only : # messages/RDMAs in-flight */
-#define IBLND_CREDIT_HIGHWATER_V1 7 /* V1 only : when eagerly to return credits */
+#define IBLND_MSG_QUEUE_SIZE_V1 8 /* V1 only : # messages/RDMAs in-flight */
+#define IBLND_CREDIT_HIGHWATER_V1 7 /* V1 only : when eagerly to return credits */
-#define IBLND_CREDITS_DEFAULT 8 /* default # of peer credits */
+#define IBLND_CREDITS_DEFAULT 8 /* default # of peer credits */
#define IBLND_CREDITS_MAX ((typeof(((kib_msg_t*) 0)->ibm_credits)) - 1) /* Max # of peer credits */
#define IBLND_MSG_QUEUE_SIZE(v) ((v) == IBLND_MSG_VERSION_1 ? \
#endif
typedef struct {
- struct list_head ibd_list; /* chain on kib_devs */
- struct list_head ibd_fail_list; /* chain on kib_failed_devs */
- __u32 ibd_ifip; /* IPoIB interface IP */
- /** IPoIB interface name */
- char ibd_ifname[KIB_IFNAME_SIZE];
- int ibd_nnets; /* # nets extant */
-
- unsigned long ibd_next_failover;
- int ibd_failed_failover; /* # failover failures */
- unsigned int ibd_failover; /* failover in progress */
- unsigned int ibd_can_failover; /* IPoIB interface is a bonding master */
- struct list_head ibd_nets;
- struct kib_hca_dev *ibd_hdev;
+ struct list_head ibd_list; /* chain on kib_devs */
+ struct list_head ibd_fail_list; /* chain on kib_failed_devs */
+ __u32 ibd_ifip; /* IPoIB interface IP */
+
+ /* IPoIB interface name */
+ char ibd_ifname[KIB_IFNAME_SIZE];
+ int ibd_nnets; /* # nets extant */
+
+ unsigned long ibd_next_failover;
+ int ibd_failed_failover; /* # failover failures */
+ unsigned int ibd_failover; /* failover in progress */
+ unsigned int ibd_can_failover; /* IPoIB interface is a bonding
+ * master */
+ struct list_head ibd_nets;
+ struct kib_hca_dev *ibd_hdev;
} kib_dev_t;
typedef struct kib_hca_dev {
- struct rdma_cm_id *ibh_cmid; /* listener cmid */
- struct ib_device *ibh_ibdev; /* IB device */
- int ibh_page_shift; /* page shift of current HCA */
- int ibh_page_size; /* page size of current HCA */
- __u64 ibh_page_mask; /* page mask of current HCA */
- int ibh_mr_shift; /* bits shift of max MR size */
- __u64 ibh_mr_size; /* size of MR */
- int ibh_nmrs; /* # of global MRs */
- struct ib_mr **ibh_mrs; /* global MR */
- struct ib_pd *ibh_pd; /* PD */
- kib_dev_t *ibh_dev; /* owner */
- atomic_t ibh_ref; /* refcount */
+ struct rdma_cm_id *ibh_cmid; /* listener cmid */
+ struct ib_device *ibh_ibdev; /* IB device */
+ int ibh_page_shift; /* page shift of current HCA */
+ int ibh_page_size; /* page size of current HCA */
+ __u64 ibh_page_mask; /* page mask of current HCA */
+ int ibh_mr_shift; /* bits shift of max MR size */
+ __u64 ibh_mr_size; /* size of MR */
+ int ibh_nmrs; /* # of global MRs */
+ struct ib_mr **ibh_mrs; /* global MR */
+ struct ib_pd *ibh_pd; /* PD */
+ kib_dev_t *ibh_dev; /* owner */
+ atomic_t ibh_ref; /* refcount */
} kib_hca_dev_t;
/** # of seconds to keep pool alive */
#define IBLND_POOL_RETRY 1
typedef struct {
- int ibp_npages; /* # pages */
- struct page *ibp_pages[0]; /* page array */
+ int ibp_npages; /* # pages */
+ struct page *ibp_pages[0]; /* page array */
} kib_pages_t;
struct kib_pmr_pool;
typedef struct {
- struct list_head pmr_list; /* chain node */
- struct ib_phys_buf *pmr_ipb; /* physical buffer */
- struct ib_mr *pmr_mr; /* IB MR */
- struct kib_pmr_pool *pmr_pool; /* owner of this MR */
- __u64 pmr_iova; /* Virtual I/O address */
- int pmr_refcount; /* reference count */
+ struct list_head pmr_list; /* chain node */
+ struct ib_phys_buf *pmr_ipb; /* physical buffer */
+ struct ib_mr *pmr_mr; /* IB MR */
+ struct kib_pmr_pool *pmr_pool; /* owner of this MR */
+ __u64 pmr_iova; /* Virtual I/O address */
+ int pmr_refcount; /* reference count */
} kib_phys_mr_t;
struct kib_pool;
#define IBLND_POOL_NAME_LEN 32
typedef struct kib_poolset {
- spinlock_t ps_lock; /* serialize */
- struct kib_net *ps_net; /* network it belongs to */
- char ps_name[IBLND_POOL_NAME_LEN]; /* pool set name */
- struct list_head ps_pool_list; /* list of pools */
- struct list_head ps_failed_pool_list; /* failed pool list */
- unsigned long ps_next_retry; /* time stamp for retry if failed to allocate */
- int ps_increasing; /* is allocating new pool */
- int ps_pool_size; /* new pool size */
- int ps_cpt; /* CPT id */
-
- kib_ps_pool_create_t ps_pool_create; /* create a new pool */
- kib_ps_pool_destroy_t ps_pool_destroy; /* destroy a pool */
- kib_ps_node_init_t ps_node_init; /* initialize new allocated node */
- kib_ps_node_fini_t ps_node_fini; /* finalize node */
+ spinlock_t ps_lock; /* serialize */
+ struct kib_net *ps_net; /* network it belongs to */
+ char ps_name[IBLND_POOL_NAME_LEN]; /* pool set name */
+ struct list_head ps_pool_list; /* list of pools */
+ struct list_head ps_failed_pool_list;/* failed pool list */
+ unsigned long ps_next_retry; /* time stamp for retry if
+ * failed to allocate */
+ int ps_increasing; /* is allocating new pool */
+ int ps_pool_size; /* new pool size */
+ int ps_cpt; /* CPT id */
+
+ kib_ps_pool_create_t ps_pool_create; /* create a new pool */
+ kib_ps_pool_destroy_t ps_pool_destroy; /* destroy a pool */
+ kib_ps_node_init_t ps_node_init; /* initialize new allocated
+ * node */
+ kib_ps_node_fini_t ps_node_fini; /* finalize node */
} kib_poolset_t;
typedef struct kib_pool {
- struct list_head po_list; /* chain on pool list */
- struct list_head po_free_list; /* pre-allocated node */
- kib_poolset_t *po_owner; /* pool_set of this pool */
- unsigned long po_deadline; /* deadline of this pool */
- int po_allocated; /* # of elements in use */
- int po_failed; /* pool is created on failed HCA */
- int po_size; /* # of pre-allocated elements */
+ struct list_head po_list; /* chain on pool list */
+ struct list_head po_free_list; /* pre-allocated node */
+ kib_poolset_t *po_owner; /* pool_set of this pool */
+ unsigned long po_deadline; /* deadline of this pool */
+ int po_allocated; /* # of elements in use */
+ int po_failed; /* pool is created on failed
+ * HCA */
+ int po_size; /* # of pre-allocated elements */
} kib_pool_t;
typedef struct {
- kib_poolset_t tps_poolset; /* pool-set */
- __u64 tps_next_tx_cookie; /* cookie of TX */
+ kib_poolset_t tps_poolset; /* pool-set */
+ __u64 tps_next_tx_cookie; /* cookie of TX */
} kib_tx_poolset_t;
typedef struct {
- kib_pool_t tpo_pool; /* pool */
- struct kib_hca_dev *tpo_hdev; /* device for this pool */
- struct kib_tx *tpo_tx_descs; /* all the tx descriptors */
- kib_pages_t *tpo_tx_pages; /* premapped tx msg pages */
+ kib_pool_t tpo_pool; /* pool */
+ struct kib_hca_dev *tpo_hdev; /* device for this pool */
+ struct kib_tx *tpo_tx_descs; /* all the tx descriptors */
+ kib_pages_t *tpo_tx_pages; /* premapped tx msg pages */
} kib_tx_pool_t;
typedef struct {
- kib_poolset_t pps_poolset; /* pool-set */
+ kib_poolset_t pps_poolset; /* pool-set */
} kib_pmr_poolset_t;
typedef struct kib_pmr_pool {
- struct kib_hca_dev *ppo_hdev; /* device for this pool */
- kib_pool_t ppo_pool; /* pool */
+ struct kib_hca_dev *ppo_hdev; /* device for this pool */
+ kib_pool_t ppo_pool; /* pool */
} kib_pmr_pool_t;
typedef struct {
- spinlock_t fps_lock; /* serialize */
- struct kib_net *fps_net; /* IB network */
- struct list_head fps_pool_list; /* FMR pool list */
- struct list_head fps_failed_pool_list; /* FMR pool list */
- __u64 fps_version; /* validity stamp */
- int fps_cpt; /* CPT id */
- int fps_pool_size;
- int fps_flush_trigger;
- /* is allocating new pool */
- int fps_increasing;
- /* time stamp for retry if failed to allocate */
- unsigned long fps_next_retry;
+ spinlock_t fps_lock; /* serialize */
+ struct kib_net *fps_net; /* IB network */
+ struct list_head fps_pool_list; /* FMR pool list */
+ struct list_head fps_failed_pool_list;/* FMR pool list */
+ __u64 fps_version; /* validity stamp */
+ int fps_cpt; /* CPT id */
+ int fps_pool_size;
+ int fps_flush_trigger;
+ int fps_increasing; /* is allocating new pool */
+ unsigned long fps_next_retry; /* time stamp for retry if
+ * failed to allocate */
} kib_fmr_poolset_t;
typedef struct {
- struct list_head fpo_list; /* chain on pool list */
- struct kib_hca_dev *fpo_hdev; /* device for this pool */
- kib_fmr_poolset_t *fpo_owner; /* owner of this pool */
- struct ib_fmr_pool *fpo_fmr_pool; /* IB FMR pool */
- unsigned long fpo_deadline; /* deadline of this pool */
- int fpo_failed; /* fmr pool is failed */
- int fpo_map_count; /* # of mapped FMR */
+ struct list_head fpo_list; /* chain on pool list */
+ struct kib_hca_dev *fpo_hdev; /* device for this pool */
+ kib_fmr_poolset_t *fpo_owner; /* owner of this pool */
+ struct ib_fmr_pool *fpo_fmr_pool; /* IB FMR pool */
+ unsigned long fpo_deadline; /* deadline of this pool */
+ int fpo_failed; /* fmr pool is failed */
+ int fpo_map_count; /* # of mapped FMR */
} kib_fmr_pool_t;
typedef struct {
- struct ib_pool_fmr *fmr_pfmr; /* IB pool fmr */
- kib_fmr_pool_t *fmr_pool; /* pool of FMR */
+ struct ib_pool_fmr *fmr_pfmr; /* IB pool fmr */
+ kib_fmr_pool_t *fmr_pool; /* pool of FMR */
} kib_fmr_t;
typedef struct kib_net {
- struct list_head ibn_list; /* chain on kib_dev_t::ibd_nets */
- __u64 ibn_incarnation; /* my epoch */
- int ibn_init; /* initialisation state */
- int ibn_shutdown; /* shutting down? */
+ struct list_head ibn_list; /* chain on kib_dev_t::ibd_nets */
+ __u64 ibn_incarnation;/* my epoch */
+ int ibn_init; /* initialisation state */
+ int ibn_shutdown; /* shutting down? */
- atomic_t ibn_npeers; /* # peers extant */
- atomic_t ibn_nconns; /* # connections extant */
+ atomic_t ibn_npeers; /* # peers extant */
+ atomic_t ibn_nconns; /* # connections extant */
- kib_tx_poolset_t **ibn_tx_ps; /* tx pool-set */
- kib_fmr_poolset_t **ibn_fmr_ps; /* fmr pool-set */
- kib_pmr_poolset_t **ibn_pmr_ps; /* pmr pool-set */
+ kib_tx_poolset_t **ibn_tx_ps; /* tx pool-set */
+ kib_fmr_poolset_t **ibn_fmr_ps; /* fmr pool-set */
+ kib_pmr_poolset_t **ibn_pmr_ps; /* pmr pool-set */
- kib_dev_t *ibn_dev; /* underlying IB device */
+ kib_dev_t *ibn_dev; /* underlying IB device */
} kib_net_t;
#define KIB_THREAD_SHIFT 16
#define KIB_THREAD_TID(id) ((id) & ((1UL << KIB_THREAD_SHIFT) - 1))
struct kib_sched_info {
- /* serialise */
- spinlock_t ibs_lock;
- /* schedulers sleep here */
- wait_queue_head_t ibs_waitq;
- /* conns to check for rx completions */
- struct list_head ibs_conns;
- /* number of scheduler threads */
- int ibs_nthreads;
- /* max allowed scheduler threads */
- int ibs_nthreads_max;
- int ibs_cpt; /* CPT id */
+ spinlock_t ibs_lock; /* serialise */
+ wait_queue_head_t ibs_waitq; /* schedulers sleep here */
+ struct list_head ibs_conns; /* conns to check for rx completions */
+ int ibs_nthreads; /* number of scheduler threads */
+ int ibs_nthreads_max; /* max allowed scheduler threads */
+ int ibs_cpt; /* CPT id */
};
typedef struct {
- int kib_init; /* initialisation state */
- int kib_shutdown; /* shut down? */
- struct list_head kib_devs; /* IB devices extant */
- /* list head of failed devices */
- struct list_head kib_failed_devs;
- /* schedulers sleep here */
- wait_queue_head_t kib_failover_waitq;
- atomic_t kib_nthreads; /* # live threads */
- /* stabilize net/dev/peer/conn ops */
- rwlock_t kib_global_lock;
- /* hash table of all my known peers */
- struct list_head *kib_peers;
- /* size of kib_peers */
- int kib_peer_hash_size;
- /* the connd task (serialisation assertions) */
- void *kib_connd;
- /* connections to setup/teardown */
- struct list_head kib_connd_conns;
- /* connections with zero refcount */
- struct list_head kib_connd_zombies;
- /* connection daemon sleeps here */
- wait_queue_head_t kib_connd_waitq;
- spinlock_t kib_connd_lock; /* serialise */
- struct ib_qp_attr kib_error_qpa; /* QP->ERROR */
- /* percpt data for schedulers */
- struct kib_sched_info **kib_scheds;
+ int kib_init; /* initialisation state */
+ int kib_shutdown; /* shut down? */
+ struct list_head kib_devs; /* IB devices extant */
+ struct list_head kib_failed_devs; /* list head of failed
+ * devices */
+ wait_queue_head_t kib_failover_waitq; /* schedulers sleep here */
+ atomic_t kib_nthreads; /* # live threads */
+ rwlock_t kib_global_lock; /* stabilize net/dev/peer/conn
+ * ops */
+ struct list_head *kib_peers; /* hash table of all my known
+ * peers */
+ int kib_peer_hash_size; /* size of kib_peers */
+ void *kib_connd; /* the connd task
+ * (serialisation assertions)
+ */
+ struct list_head kib_connd_conns; /* connections to
+ * setup/teardown */
+ struct list_head kib_connd_zombies; /* connections with zero
+ * refcount */
+ wait_queue_head_t kib_connd_waitq; /* connection daemon sleeps
+ * here */
+ spinlock_t kib_connd_lock; /* serialise */
+ struct ib_qp_attr kib_error_qpa; /* QP->ERROR */
+ struct kib_sched_info **kib_scheds; /* percpt data for schedulers
+ */
} kib_data_t;
-#define IBLND_INIT_NOTHING 0
-#define IBLND_INIT_DATA 1
-#define IBLND_INIT_ALL 2
+#define IBLND_INIT_NOTHING 0
+#define IBLND_INIT_DATA 1
+#define IBLND_INIT_ALL 2
/************************************************************************
* IB Wire message format.
*/
typedef struct kib_connparams {
- __u16 ibcp_queue_depth;
- __u16 ibcp_max_frags;
- __u32 ibcp_max_msg_size;
+ __u16 ibcp_queue_depth;
+ __u16 ibcp_max_frags;
+ __u32 ibcp_max_msg_size;
} WIRE_ATTR kib_connparams_t;
typedef struct {
- lnet_hdr_t ibim_hdr; /* portals header */
- char ibim_payload[0]; /* piggy-backed payload */
+ lnet_hdr_t ibim_hdr; /* portals header */
+ char ibim_payload[0]; /* piggy-backed payload */
} WIRE_ATTR kib_immediate_msg_t;
typedef struct {
- __u32 rf_nob; /* # bytes this frag */
- __u64 rf_addr; /* CAVEAT EMPTOR: misaligned!! */
+ __u32 rf_nob; /* # bytes this frag */
+ __u64 rf_addr; /* CAVEAT EMPTOR: misaligned!! */
} WIRE_ATTR kib_rdma_frag_t;
typedef struct {
- __u32 rd_key; /* local/remote key */
- __u32 rd_nfrags; /* # fragments */
- kib_rdma_frag_t rd_frags[0]; /* buffer frags */
+ __u32 rd_key; /* local/remote key */
+ __u32 rd_nfrags; /* # fragments */
+ kib_rdma_frag_t rd_frags[0]; /* buffer frags */
} WIRE_ATTR kib_rdma_desc_t;
typedef struct {
- lnet_hdr_t ibprm_hdr; /* portals header */
- __u64 ibprm_cookie; /* opaque completion cookie */
+ lnet_hdr_t ibprm_hdr; /* portals header */
+ __u64 ibprm_cookie; /* opaque completion cookie */
} WIRE_ATTR kib_putreq_msg_t;
typedef struct {
- __u64 ibpam_src_cookie; /* reflected completion cookie */
- __u64 ibpam_dst_cookie; /* opaque completion cookie */
- kib_rdma_desc_t ibpam_rd; /* sender's sink buffer */
+ __u64 ibpam_src_cookie; /* reflected completion cookie */
+ __u64 ibpam_dst_cookie; /* opaque completion cookie */
+ kib_rdma_desc_t ibpam_rd; /* sender's sink buffer */
} WIRE_ATTR kib_putack_msg_t;
typedef struct {
- lnet_hdr_t ibgm_hdr; /* portals header */
- __u64 ibgm_cookie; /* opaque completion cookie */
- kib_rdma_desc_t ibgm_rd; /* rdma descriptor */
+ lnet_hdr_t ibgm_hdr; /* portals header */
+ __u64 ibgm_cookie; /* opaque completion cookie */
+ kib_rdma_desc_t ibgm_rd; /* rdma descriptor */
} WIRE_ATTR kib_get_msg_t;
typedef struct {
- __u64 ibcm_cookie; /* opaque completion cookie */
- __s32 ibcm_status; /* < 0 failure: >= 0 length */
+ __u64 ibcm_cookie; /* opaque completion cookie */
+ __s32 ibcm_status; /* < 0 failure: >= 0 length */
} WIRE_ATTR kib_completion_msg_t;
typedef struct {
/* First 2 fields fixed FOR ALL TIME */
- __u32 ibm_magic; /* I'm an ibnal message */
- __u16 ibm_version; /* this is my version number */
-
- __u8 ibm_type; /* msg type */
- __u8 ibm_credits; /* returned credits */
- __u32 ibm_nob; /* # bytes in whole message */
- __u32 ibm_cksum; /* checksum (0 == no checksum) */
- __u64 ibm_srcnid; /* sender's NID */
- __u64 ibm_srcstamp; /* sender's incarnation */
- __u64 ibm_dstnid; /* destination's NID */
- __u64 ibm_dststamp; /* destination's incarnation */
+ __u32 ibm_magic; /* I'm an ibnal message */
+ __u16 ibm_version; /* this is my version number */
+
+ __u8 ibm_type; /* msg type */
+ __u8 ibm_credits; /* returned credits */
+ __u32 ibm_nob; /* # bytes in whole message */
+ __u32 ibm_cksum; /* checksum (0 == no checksum) */
+ __u64 ibm_srcnid; /* sender's NID */
+ __u64 ibm_srcstamp; /* sender's incarnation */
+ __u64 ibm_dstnid; /* destination's NID */
+ __u64 ibm_dststamp; /* destination's incarnation */
union {
- kib_connparams_t connparams;
- kib_immediate_msg_t immediate;
- kib_putreq_msg_t putreq;
- kib_putack_msg_t putack;
- kib_get_msg_t get;
- kib_completion_msg_t completion;
+ kib_connparams_t connparams;
+ kib_immediate_msg_t immediate;
+ kib_putreq_msg_t putreq;
+ kib_putack_msg_t putack;
+ kib_get_msg_t get;
+ kib_completion_msg_t completion;
} WIRE_ATTR ibm_u;
} WIRE_ATTR kib_msg_t;
-#define IBLND_MSG_MAGIC LNET_PROTO_IB_MAGIC /* unique magic */
+#define IBLND_MSG_MAGIC LNET_PROTO_IB_MAGIC /* unique magic */
-#define IBLND_MSG_VERSION_1 0x11
-#define IBLND_MSG_VERSION_2 0x12
-#define IBLND_MSG_VERSION IBLND_MSG_VERSION_2
+#define IBLND_MSG_VERSION_1 0x11
+#define IBLND_MSG_VERSION_2 0x12
+#define IBLND_MSG_VERSION IBLND_MSG_VERSION_2
-#define IBLND_MSG_CONNREQ 0xc0 /* connection request */
-#define IBLND_MSG_CONNACK 0xc1 /* connection acknowledge */
-#define IBLND_MSG_NOOP 0xd0 /* nothing (just credits) */
-#define IBLND_MSG_IMMEDIATE 0xd1 /* immediate */
-#define IBLND_MSG_PUT_REQ 0xd2 /* putreq (src->sink) */
-#define IBLND_MSG_PUT_NAK 0xd3 /* completion (sink->src) */
-#define IBLND_MSG_PUT_ACK 0xd4 /* putack (sink->src) */
-#define IBLND_MSG_PUT_DONE 0xd5 /* completion (src->sink) */
-#define IBLND_MSG_GET_REQ 0xd6 /* getreq (sink->src) */
-#define IBLND_MSG_GET_DONE 0xd7 /* completion (src->sink: all OK) */
+#define IBLND_MSG_CONNREQ 0xc0 /* connection request */
+#define IBLND_MSG_CONNACK 0xc1 /* connection acknowledge */
+#define IBLND_MSG_NOOP 0xd0 /* nothing (just credits) */
+#define IBLND_MSG_IMMEDIATE 0xd1 /* immediate */
+#define IBLND_MSG_PUT_REQ 0xd2 /* putreq (src->sink) */
+#define IBLND_MSG_PUT_NAK 0xd3 /* completion (sink->src) */
+#define IBLND_MSG_PUT_ACK 0xd4 /* putack (sink->src) */
+#define IBLND_MSG_PUT_DONE 0xd5 /* completion (src->sink) */
+#define IBLND_MSG_GET_REQ 0xd6 /* getreq (sink->src) */
+#define IBLND_MSG_GET_DONE 0xd7 /* completion (src->sink: all OK) */
typedef struct {
- __u32 ibr_magic; /* sender's magic */
- __u16 ibr_version; /* sender's version */
- __u8 ibr_why; /* reject reason */
- __u8 ibr_padding; /* padding */
- __u64 ibr_incarnation; /* incarnation of peer */
- kib_connparams_t ibr_cp; /* connection parameters */
+ __u32 ibr_magic; /* sender's magic */
+ __u16 ibr_version; /* sender's version */
+ __u8 ibr_why; /* reject reason */
+ __u8 ibr_padding; /* padding */
+ __u64 ibr_incarnation; /* incarnation of peer */
+ kib_connparams_t ibr_cp; /* connection parameters */
} WIRE_ATTR kib_rej_t;
/* connection rejection reasons */
-#define IBLND_REJECT_CONN_RACE 1 /* You lost connection race */
-#define IBLND_REJECT_NO_RESOURCES 2 /* Out of memory/conns etc */
-#define IBLND_REJECT_FATAL 3 /* Anything else */
-
-#define IBLND_REJECT_CONN_UNCOMPAT 4 /* incompatible version peer */
-#define IBLND_REJECT_CONN_STALE 5 /* stale peer */
-
-#define IBLND_REJECT_RDMA_FRAGS 6 /* Fatal: peer's rdma frags can't match mine */
-#define IBLND_REJECT_MSG_QUEUE_SIZE 7 /* Fatal: peer's msg queue size can't match mine */
+#define IBLND_REJECT_CONN_RACE 1 /* You lost connection race */
+#define IBLND_REJECT_NO_RESOURCES 2 /* Out of memory/conns etc */
+#define IBLND_REJECT_FATAL 3 /* Anything else */
+#define IBLND_REJECT_CONN_UNCOMPAT 4 /* incompatible version peer */
+#define IBLND_REJECT_CONN_STALE 5 /* stale peer */
+#define IBLND_REJECT_RDMA_FRAGS 6 /* Fatal: peer's rdma frags can't match
+ * mine */
+#define IBLND_REJECT_MSG_QUEUE_SIZE 7 /* Fatal: peer's msg queue size can't
+ * match mine */
/***********************************************************************/
-typedef struct kib_rx /* receive message */
+typedef struct kib_rx /* receive message */
{
- struct list_head rx_list; /* queue for attention */
- struct kib_conn *rx_conn; /* owning conn */
- int rx_nob; /* # bytes received (-1 while posted) */
- enum ib_wc_status rx_status; /* completion status */
- kib_msg_t *rx_msg; /* message buffer (host vaddr) */
- __u64 rx_msgaddr; /* message buffer (I/O addr) */
- DECLARE_PCI_UNMAP_ADDR (rx_msgunmap); /* for dma_unmap_single() */
- struct ib_recv_wr rx_wrq; /* receive work item... */
- struct ib_sge rx_sge; /* ...and its memory */
+ struct list_head rx_list; /* queue for attention */
+ struct kib_conn *rx_conn; /* owning conn */
+ int rx_nob; /* # bytes received (-1 while
+ * posted) */
+ enum ib_wc_status rx_status; /* completion status */
+ kib_msg_t *rx_msg; /* message buffer (host vaddr) */
+ __u64 rx_msgaddr; /* message buffer (I/O addr) */
+ DECLARE_PCI_UNMAP_ADDR (rx_msgunmap); /* for dma_unmap_single() */
+ struct ib_recv_wr rx_wrq; /* receive work item... */
+ struct ib_sge rx_sge; /* ...and its memory */
} kib_rx_t;
-#define IBLND_POSTRX_DONT_POST 0 /* don't post */
-#define IBLND_POSTRX_NO_CREDIT 1 /* post: no credits */
-#define IBLND_POSTRX_PEER_CREDIT 2 /* post: give peer back 1 credit */
-#define IBLND_POSTRX_RSRVD_CREDIT 3 /* post: give myself back 1 reserved credit */
+#define IBLND_POSTRX_DONT_POST 0 /* don't post */
+#define IBLND_POSTRX_NO_CREDIT 1 /* post: no credits */
+#define IBLND_POSTRX_PEER_CREDIT 2 /* post: give peer back 1 credit */
+#define IBLND_POSTRX_RSRVD_CREDIT 3 /* post: give myself back 1 reserved
+ * credit */
-typedef struct kib_tx /* transmit message */
+typedef struct kib_tx /* transmit message */
{
- struct list_head tx_list; /* queue on idle_txs ibc_tx_queue etc. */
- kib_tx_pool_t *tx_pool; /* pool I'm from */
- struct kib_conn *tx_conn; /* owning conn */
- short tx_sending; /* # tx callbacks outstanding */
- short tx_queued; /* queued for sending */
- short tx_waiting; /* waiting for peer */
- int tx_status; /* LNET completion status */
- unsigned long tx_deadline; /* completion deadline */
- __u64 tx_cookie; /* completion cookie */
- lnet_msg_t *tx_lntmsg[2]; /* lnet msgs to finalize on completion */
- kib_msg_t *tx_msg; /* message buffer (host vaddr) */
- __u64 tx_msgaddr; /* message buffer (I/O addr) */
- DECLARE_PCI_UNMAP_ADDR (tx_msgunmap); /* for dma_unmap_single() */
- int tx_nwrq; /* # send work items */
- struct ib_send_wr *tx_wrq; /* send work items... */
- struct ib_sge *tx_sge; /* ...and their memory */
- kib_rdma_desc_t *tx_rd; /* rdma descriptor */
- int tx_nfrags; /* # entries in... */
- struct scatterlist *tx_frags; /* dma_map_sg descriptor */
- __u64 *tx_pages; /* rdma phys page addrs */
+ struct list_head tx_list; /* queue on idle_txs ibc_tx_queue
+ * etc. */
+ kib_tx_pool_t *tx_pool; /* pool I'm from */
+ struct kib_conn *tx_conn; /* owning conn */
+ short tx_sending; /* # tx callbacks outstanding */
+ short tx_queued; /* queued for sending */
+ short tx_waiting; /* waiting for peer */
+ int tx_status; /* LNET completion status */
+ unsigned long tx_deadline; /* completion deadline */
+ __u64 tx_cookie; /* completion cookie */
+ lnet_msg_t *tx_lntmsg[2]; /* lnet msgs to finalize on
+ * completion */
+ kib_msg_t *tx_msg; /* message buffer (host vaddr) */
+ __u64 tx_msgaddr; /* message buffer (I/O addr) */
+ DECLARE_PCI_UNMAP_ADDR (tx_msgunmap); /* for dma_unmap_single() */
+ int tx_nwrq; /* # send work items */
+ struct ib_send_wr *tx_wrq; /* send work items... */
+ struct ib_sge *tx_sge; /* ...and their memory */
+ kib_rdma_desc_t *tx_rd; /* rdma descriptor */
+ int tx_nfrags; /* # entries in... */
+ struct scatterlist *tx_frags; /* dma_map_sg descriptor */
+ __u64 *tx_pages; /* rdma phys page addrs */
union {
- kib_phys_mr_t *pmr; /* MR for physical buffer */
- kib_fmr_t fmr; /* FMR */
- } tx_u;
- int tx_dmadir; /* dma direction */
+ kib_phys_mr_t *pmr; /* MR for physical buffer */
+ kib_fmr_t fmr; /* FMR */
+ } tx_u;
+ int tx_dmadir; /* dma direction */
} kib_tx_t;
typedef struct kib_connvars {
- /* connection-in-progress variables */
- kib_msg_t cv_msg;
+ kib_msg_t cv_msg; /* connection-in-progress variables */
} kib_connvars_t;
typedef struct kib_conn {
- struct kib_sched_info *ibc_sched; /* scheduler information */
- struct kib_peer *ibc_peer; /* owning peer */
- kib_hca_dev_t *ibc_hdev; /* HCA bound on */
- struct list_head ibc_list; /* stash on peer's conn list */
- struct list_head ibc_sched_list; /* schedule for attention */
- __u16 ibc_version; /* version of connection */
- __u64 ibc_incarnation; /* which instance of the peer */
- atomic_t ibc_refcount; /* # users */
- int ibc_state; /* what's happening */
- int ibc_nsends_posted; /* # uncompleted sends */
- int ibc_noops_posted; /* # uncompleted NOOPs */
- int ibc_credits; /* # credits I have */
- int ibc_outstanding_credits; /* # credits to return */
- int ibc_reserved_credits;/* # ACK/DONE msg credits */
- int ibc_comms_error; /* set on comms error */
- unsigned int ibc_nrx:16; /* receive buffers owned */
- unsigned int ibc_scheduled:1; /* scheduled for attention */
- unsigned int ibc_ready:1; /* CQ callback fired */
- /* time of last send */
- unsigned long ibc_last_send;
- /** link chain for kiblnd_check_conns only */
- struct list_head ibc_connd_list;
- /** rxs completed before ESTABLISHED */
- struct list_head ibc_early_rxs;
- /** IBLND_MSG_NOOPs for IBLND_MSG_VERSION_1 */
- struct list_head ibc_tx_noops;
- struct list_head ibc_tx_queue; /* sends that need a credit */
- struct list_head ibc_tx_queue_nocred;/* sends that don't need a credit */
- struct list_head ibc_tx_queue_rsrvd; /* sends that need to reserve an ACK/DONE msg */
- struct list_head ibc_active_txs; /* active tx awaiting completion */
- spinlock_t ibc_lock; /* serialise */
- kib_rx_t *ibc_rxs; /* the rx descs */
- kib_pages_t *ibc_rx_pages; /* premapped rx msg pages */
-
- struct rdma_cm_id *ibc_cmid; /* CM id */
- struct ib_cq *ibc_cq; /* completion queue */
-
- kib_connvars_t *ibc_connvars; /* in-progress connection state */
+ struct kib_sched_info *ibc_sched; /* scheduler information */
+ struct kib_peer *ibc_peer; /* owning peer */
+ kib_hca_dev_t *ibc_hdev; /* HCA bound on */
+ struct list_head ibc_list; /* stash on peer's conn
+ * list */
+ struct list_head ibc_sched_list; /* schedule for attention */
+ __u16 ibc_version; /* version of connection */
+ __u64 ibc_incarnation; /* which instance of the
+ * peer */
+ atomic_t ibc_refcount; /* # users */
+ int ibc_state; /* what's happening */
+ int ibc_nsends_posted; /* # uncompleted sends */
+ int ibc_noops_posted; /* # uncompleted NOOPs */
+ int ibc_credits; /* # credits I have */
+ int ibc_outstanding_credits; /* # credits to return */
+ int ibc_reserved_credits; /* # ACK/DONE msg credits */
+ int ibc_comms_error; /* set on comms error */
+ unsigned int ibc_nrx:16; /* receive buffers owned */
+ unsigned int ibc_scheduled:1; /* scheduled for attention
+ */
+ unsigned int ibc_ready:1; /* CQ callback fired */
+ unsigned long ibc_last_send; /* time of last send */
+ struct list_head ibc_connd_list; /* link chain for
+ * kiblnd_check_conns only
+ */
+ struct list_head ibc_early_rxs; /* rxs completed before
+ * ESTABLISHED */
+ struct list_head ibc_tx_noops; /* IBLND_MSG_NOOPs for
+ * IBLND_MSG_VERSION_1 */
+ struct list_head ibc_tx_queue; /* sends that need a credit
+ */
+ struct list_head ibc_tx_queue_nocred; /* sends that don't need a
+ * credit */
+ struct list_head ibc_tx_queue_rsrvd; /* sends that need to
+ * reserve an ACK/DONE msg
+ */
+ struct list_head ibc_active_txs; /* active tx awaiting
+ * completion */
+ spinlock_t ibc_lock; /* serialise */
+ kib_rx_t *ibc_rxs; /* the rx descs */
+ kib_pages_t *ibc_rx_pages; /* premapped rx msg pages */
+
+ struct rdma_cm_id *ibc_cmid; /* CM id */
+ struct ib_cq *ibc_cq; /* completion queue */
+
+ kib_connvars_t *ibc_connvars; /* in-progress connection
+ * state */
} kib_conn_t;
-#define IBLND_CONN_INIT 0 /* being initialised */
-#define IBLND_CONN_ACTIVE_CONNECT 1 /* active sending req */
-#define IBLND_CONN_PASSIVE_WAIT 2 /* passive waiting for rtu */
-#define IBLND_CONN_ESTABLISHED 3 /* connection established */
-#define IBLND_CONN_CLOSING 4 /* being closed */
-#define IBLND_CONN_DISCONNECTED 5 /* disconnected */
+#define IBLND_CONN_INIT 0 /* being initialised */
+#define IBLND_CONN_ACTIVE_CONNECT 1 /* active sending req */
+#define IBLND_CONN_PASSIVE_WAIT 2 /* passive waiting for rtu */
+#define IBLND_CONN_ESTABLISHED 3 /* connection established */
+#define IBLND_CONN_CLOSING 4 /* being closed */
+#define IBLND_CONN_DISCONNECTED 5 /* disconnected */
typedef struct kib_peer {
- struct list_head ibp_list; /* stash on global peer list */
- lnet_nid_t ibp_nid; /* who's on the other end(s) */
- lnet_ni_t *ibp_ni; /* LNet interface */
- atomic_t ibp_refcount; /* # users */
- struct list_head ibp_conns; /* all active connections */
- struct list_head ibp_tx_queue; /* msgs waiting for a conn */
- __u16 ibp_version; /* version of peer */
- __u64 ibp_incarnation; /* incarnation of peer */
- int ibp_connecting; /* current active connection attempts */
- int ibp_accepting; /* current passive connection attempts */
- int ibp_error; /* errno on closing this peer */
- unsigned long ibp_last_alive; /* when (in jiffies) I was last alive */
+ struct list_head ibp_list; /* stash on global peer list */
+ lnet_nid_t ibp_nid; /* who's on the other end(s) */
+ lnet_ni_t *ibp_ni; /* LNet interface */
+ atomic_t ibp_refcount; /* # users */
+ struct list_head ibp_conns; /* all active connections */
+ struct list_head ibp_tx_queue; /* msgs waiting for a conn */
+ __u16 ibp_version; /* version of peer */
+ __u64 ibp_incarnation; /* incarnation of peer */
+ int ibp_connecting; /* current active connection attempts
+ */
+ int ibp_accepting; /* current passive connection attempts
+ */
+ int ibp_error; /* errno on closing this peer */
+ unsigned long ibp_last_alive; /* when (in jiffies) I was last alive
+ */
} kib_peer_t;
-extern kib_data_t kiblnd_data;
+extern kib_data_t kiblnd_data;
extern void kiblnd_hdev_destroy(kib_hca_dev_t *hdev);
* right because OFED1.2 defines it as const, to use it we have to add
* (void *) cast to overcome "const" */
-#define KIBLND_CONN_PARAM(e) ((e)->param.conn.private_data)
-#define KIBLND_CONN_PARAM_LEN(e) ((e)->param.conn.private_data_len)
+#define KIBLND_CONN_PARAM(e) ((e)->param.conn.private_data)
+#define KIBLND_CONN_PARAM_LEN(e) ((e)->param.conn.private_data_len)
struct ib_mr *kiblnd_find_rd_dma_mr(kib_hca_dev_t *hdev,
kiblnd_tx_done(lnet_ni_t *ni, kib_tx_t *tx)
{
lnet_msg_t *lntmsg[2];
- kib_net_t *net = ni->ni_data;
- int rc;
- int i;
+ kib_net_t *net = ni->ni_data;
+ int rc;
+ int i;
LASSERT(net != NULL);
LASSERT(!in_interrupt());
static kib_tx_t *
kiblnd_get_idle_tx(lnet_ni_t *ni, lnet_nid_t target)
{
- kib_net_t *net = (kib_net_t *)ni->ni_data;
- struct list_head *node;
- kib_tx_t *tx;
- kib_tx_poolset_t *tps;
+ kib_net_t *net = (kib_net_t *)ni->ni_data;
+ struct list_head *node;
+ kib_tx_t *tx;
+ kib_tx_poolset_t *tps;
tps = net->ibn_tx_ps[lnet_cpt_of_nid(target)];
node = kiblnd_pool_alloc_node(&tps->tps_poolset);
static void
kiblnd_drop_rx(kib_rx_t *rx)
{
- kib_conn_t *conn = rx->rx_conn;
- struct kib_sched_info *sched = conn->ibc_sched;
- unsigned long flags;
+ kib_conn_t *conn = rx->rx_conn;
+ struct kib_sched_info *sched = conn->ibc_sched;
+ unsigned long flags;
spin_lock_irqsave(&sched->ibs_lock, flags);
LASSERT(conn->ibc_nrx > 0);
int
kiblnd_post_rx(kib_rx_t *rx, int credit)
{
- kib_conn_t *conn = rx->rx_conn;
- kib_net_t *net = conn->ibc_peer->ibp_ni->ni_data;
- struct ib_recv_wr *bad_wrq = NULL;
- struct ib_mr *mr;
- int rc;
+ kib_conn_t *conn = rx->rx_conn;
+ kib_net_t *net = conn->ibc_peer->ibp_ni->ni_data;
+ struct ib_recv_wr *bad_wrq = NULL;
+ struct ib_mr *mr;
+ int rc;
LASSERT(net != NULL);
LASSERT(!in_interrupt());
rx->rx_sge.addr = rx->rx_msgaddr;
rx->rx_sge.length = IBLND_MSG_SIZE;
- rx->rx_wrq.next = NULL;
+ rx->rx_wrq.next = NULL;
rx->rx_wrq.sg_list = &rx->rx_sge;
rx->rx_wrq.num_sge = 1;
- rx->rx_wrq.wr_id = kiblnd_ptr2wreqid(rx, IBLND_WID_RX);
+ rx->rx_wrq.wr_id = kiblnd_ptr2wreqid(rx, IBLND_WID_RX);
LASSERT(conn->ibc_state >= IBLND_CONN_INIT);
LASSERT(rx->rx_nob >= 0); /* not posted */
static kib_tx_t *
kiblnd_find_waiting_tx_locked(kib_conn_t *conn, int txtype, __u64 cookie)
{
- struct list_head *tmp;
+ struct list_head *tmp;
list_for_each(tmp, &conn->ibc_active_txs) {
kib_tx_t *tx = list_entry(tmp, kib_tx_t, tx_list);
static void
kiblnd_handle_completion(kib_conn_t *conn, int txtype, int status, __u64 cookie)
{
- kib_tx_t *tx;
- lnet_ni_t *ni = conn->ibc_peer->ibp_ni;
- int idle;
+ kib_tx_t *tx;
+ lnet_ni_t *ni = conn->ibc_peer->ibp_ni;
+ int idle;
spin_lock(&conn->ibc_lock);
static void
kiblnd_send_completion(kib_conn_t *conn, int type, int status, __u64 cookie)
{
- lnet_ni_t *ni = conn->ibc_peer->ibp_ni;
- kib_tx_t *tx = kiblnd_get_idle_tx(ni, conn->ibc_peer->ibp_nid);
+ lnet_ni_t *ni = conn->ibc_peer->ibp_ni;
+ kib_tx_t *tx = kiblnd_get_idle_tx(ni, conn->ibc_peer->ibp_nid);
if (tx == NULL) {
CERROR("Can't get tx for completion %x for %s\n",
static void
kiblnd_handle_rx(kib_rx_t *rx)
{
- kib_msg_t *msg = rx->rx_msg;
- kib_conn_t *conn = rx->rx_conn;
- lnet_ni_t *ni = conn->ibc_peer->ibp_ni;
- int credits = msg->ibm_credits;
- kib_tx_t *tx;
- int rc = 0;
- int rc2;
- int post_credit;
+ kib_msg_t *msg = rx->rx_msg;
+ kib_conn_t *conn = rx->rx_conn;
+ lnet_ni_t *ni = conn->ibc_peer->ibp_ni;
+ int credits = msg->ibm_credits;
+ kib_tx_t *tx;
+ int rc = 0;
+ int rc2;
+ int post_credit;
LASSERT(conn->ibc_state >= IBLND_CONN_ESTABLISHED);
static void
kiblnd_rx_complete(kib_rx_t *rx, int status, int nob)
{
- kib_msg_t *msg = rx->rx_msg;
- kib_conn_t *conn = rx->rx_conn;
- lnet_ni_t *ni = conn->ibc_peer->ibp_ni;
- kib_net_t *net = ni->ni_data;
- int rc;
- int err = -EIO;
+ kib_msg_t *msg = rx->rx_msg;
+ kib_conn_t *conn = rx->rx_conn;
+ lnet_ni_t *ni = conn->ibc_peer->ibp_ni;
+ kib_net_t *net = ni->ni_data;
+ int rc;
+ int err = -EIO;
LASSERT(net != NULL);
LASSERT(rx->rx_nob < 0); /* was posted */
/* racing with connection establishment/teardown! */
if (conn->ibc_state < IBLND_CONN_ESTABLISHED) {
- rwlock_t *g_lock = &kiblnd_data.kib_global_lock;
- unsigned long flags;
+ rwlock_t *g_lock = &kiblnd_data.kib_global_lock;
+ unsigned long flags;
write_lock_irqsave(g_lock, flags);
/* must check holding global lock to eliminate race */
static int
kiblnd_fmr_map_tx(kib_net_t *net, kib_tx_t *tx, kib_rdma_desc_t *rd, int nob)
{
- kib_hca_dev_t *hdev;
- __u64 *pages = tx->tx_pages;
- kib_fmr_poolset_t *fps;
- int npages;
- int size;
- int cpt;
- int rc;
- int i;
+ kib_hca_dev_t *hdev;
+ __u64 *pages = tx->tx_pages;
+ kib_fmr_poolset_t *fps;
+ int npages;
+ int size;
+ int cpt;
+ int rc;
+ int i;
LASSERT(tx->tx_pool != NULL);
LASSERT(tx->tx_pool->tpo_pool.po_owner != NULL);
- hdev = tx->tx_pool->tpo_hdev;
+ hdev = tx->tx_pool->tpo_hdev;
for (i = 0, npages = 0; i < rd->rd_nfrags; i++) {
for (size = 0; size < rd->rd_frags[i].rf_nob;
rd->rd_key = (rd != tx->tx_rd) ? tx->tx_u.fmr.fmr_pfmr->fmr->rkey :
tx->tx_u.fmr.fmr_pfmr->fmr->lkey;
rd->rd_frags[0].rf_addr &= ~hdev->ibh_page_mask;
- rd->rd_frags[0].rf_nob = nob;
+ rd->rd_frags[0].rf_nob = nob;
rd->rd_nfrags = 1;
return 0;
static int
kiblnd_pmr_map_tx(kib_net_t *net, kib_tx_t *tx, kib_rdma_desc_t *rd, int nob)
{
- kib_hca_dev_t *hdev;
- kib_pmr_poolset_t *pps;
- __u64 iova;
- int cpt;
- int rc;
+ kib_hca_dev_t *hdev;
+ kib_pmr_poolset_t *pps;
+ __u64 iova;
+ int cpt;
+ int rc;
LASSERT(tx->tx_pool != NULL);
LASSERT(tx->tx_pool->tpo_pool.po_owner != NULL);
tx->tx_u.pmr->pmr_mr->lkey;
rd->rd_nfrags = 1;
rd->rd_frags[0].rf_addr = iova;
- rd->rd_frags[0].rf_nob = nob;
+ rd->rd_frags[0].rf_nob = nob;
return 0;
}
void
kiblnd_unmap_tx(lnet_ni_t *ni, kib_tx_t *tx)
{
- kib_net_t *net = ni->ni_data;
+ kib_net_t *net = ni->ni_data;
LASSERT(net != NULL);
kiblnd_map_tx(lnet_ni_t *ni, kib_tx_t *tx,
kib_rdma_desc_t *rd, int nfrags)
{
- kib_hca_dev_t *hdev = tx->tx_pool->tpo_hdev;
- kib_net_t *net = ni->ni_data;
- struct ib_mr *mr = NULL;
- __u32 nob;
- int i;
+ kib_hca_dev_t *hdev = tx->tx_pool->tpo_hdev;
+ kib_net_t *net = ni->ni_data;
+ struct ib_mr *mr = NULL;
+ __u32 nob;
+ int i;
/* If rd is not tx_rd, it's going to get sent to a peer and I'm the
* RDMA sink */
tx->tx_dmadir = (rd != tx->tx_rd) ? DMA_FROM_DEVICE : DMA_TO_DEVICE;
tx->tx_nfrags = nfrags;
- rd->rd_nfrags =
- kiblnd_dma_map_sg(hdev->ibh_ibdev,
- tx->tx_frags, tx->tx_nfrags, tx->tx_dmadir);
+ rd->rd_nfrags = kiblnd_dma_map_sg(hdev->ibh_ibdev, tx->tx_frags,
+ tx->tx_nfrags, tx->tx_dmadir);
for (i = 0, nob = 0; i < rd->rd_nfrags; i++) {
rd->rd_frags[i].rf_nob = kiblnd_sg_dma_len(
kiblnd_setup_rd_iov(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd,
unsigned int niov, struct kvec *iov, int offset, int nob)
{
- kib_net_t *net = ni->ni_data;
- struct page *page;
+ kib_net_t *net = ni->ni_data;
+ struct page *page;
struct scatterlist *sg;
- unsigned long vaddr;
- int fragnob;
- int page_offset;
+ unsigned long vaddr;
+ int fragnob;
+ int page_offset;
LASSERT(nob > 0);
LASSERT(niov > 0);
kiblnd_setup_rd_kiov(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd,
int nkiov, lnet_kiov_t *kiov, int offset, int nob)
{
- kib_net_t *net = ni->ni_data;
+ kib_net_t *net = ni->ni_data;
struct scatterlist *sg;
- int fragnob;
+ int fragnob;
CDEBUG(D_NET, "niov %d offset %d nob %d\n", nkiov, offset, nob);
__releases(conn->ibc_lock)
__acquires(conn->ibc_lock)
{
- kib_msg_t *msg = tx->tx_msg;
- kib_peer_t *peer = conn->ibc_peer;
- int ver = conn->ibc_version;
- int rc;
- int done;
+ kib_msg_t *msg = tx->tx_msg;
+ kib_peer_t *peer = conn->ibc_peer;
+ int ver = conn->ibc_version;
+ int rc;
+ int done;
struct ib_send_wr *bad_wrq;
LASSERT(tx->tx_queued);
/* close_conn will launch failover */
rc = -ENETDOWN;
} else {
- rc = ib_post_send(conn->ibc_cmid->qp,
- tx->tx_wrq, &bad_wrq);
+ rc = ib_post_send(conn->ibc_cmid->qp, tx->tx_wrq, &bad_wrq);
}
conn->ibc_last_send = jiffies;
void
kiblnd_check_sends(kib_conn_t *conn)
{
- int ver = conn->ibc_version;
+ int ver = conn->ibc_version;
lnet_ni_t *ni = conn->ibc_peer->ibp_ni;
- kib_tx_t *tx;
+ kib_tx_t *tx;
/* Don't send anything until after the connection is established */
if (conn->ibc_state < IBLND_CONN_ESTABLISHED) {
static void
kiblnd_tx_complete(kib_tx_t *tx, int status)
{
- int failed = (status != IB_WC_SUCCESS);
- kib_conn_t *conn = tx->tx_conn;
- int idle;
+ int failed = (status != IB_WC_SUCCESS);
+ kib_conn_t *conn = tx->tx_conn;
+ int idle;
LASSERT(tx->tx_sending > 0);
void
kiblnd_init_tx_msg(lnet_ni_t *ni, kib_tx_t *tx, int type, int body_nob)
{
- kib_hca_dev_t *hdev = tx->tx_pool->tpo_hdev;
- struct ib_sge *sge = &tx->tx_sge[tx->tx_nwrq];
+ kib_hca_dev_t *hdev = tx->tx_pool->tpo_hdev;
+ struct ib_sge *sge = &tx->tx_sge[tx->tx_nwrq];
struct ib_send_wr *wrq = &tx->tx_wrq[tx->tx_nwrq];
- int nob = offsetof(kib_msg_t, ibm_u) + body_nob;
- struct ib_mr *mr;
+ int nob = offsetof(kib_msg_t, ibm_u) + body_nob;
+ struct ib_mr *mr;
LASSERT(tx->tx_nwrq >= 0);
LASSERT(tx->tx_nwrq < IBLND_MAX_RDMA_FRAGS + 1);
kiblnd_init_rdma(kib_conn_t *conn, kib_tx_t *tx, int type,
int resid, kib_rdma_desc_t *dstrd, __u64 dstcookie)
{
- kib_msg_t *ibmsg = tx->tx_msg;
- kib_rdma_desc_t *srcrd = tx->tx_rd;
- struct ib_sge *sge = &tx->tx_sge[0];
+ kib_msg_t *ibmsg = tx->tx_msg;
+ kib_rdma_desc_t *srcrd = tx->tx_rd;
+ struct ib_sge *sge = &tx->tx_sge[0];
struct ib_send_wr *wrq = &tx->tx_wrq[0];
- int rc = resid;
- int srcidx;
- int dstidx;
- int wrknob;
+ int rc = resid;
+ int srcidx;
+ int dstidx;
+ int wrknob;
LASSERT(!in_interrupt());
LASSERT(tx->tx_nwrq == 0);
wrq->send_flags = 0;
wrq->wr.rdma.remote_addr = kiblnd_rd_frag_addr(dstrd, dstidx);
- wrq->wr.rdma.rkey = kiblnd_rd_frag_key(dstrd, dstidx);
+ wrq->wr.rdma.rkey = kiblnd_rd_frag_key(dstrd, dstidx);
srcidx = kiblnd_rd_consume_frag(srcrd, srcidx, wrknob);
dstidx = kiblnd_rd_consume_frag(dstrd, dstidx, wrknob);
void
kiblnd_queue_tx_locked(kib_tx_t *tx, kib_conn_t *conn)
{
- struct list_head *q;
+ struct list_head *q;
LASSERT(tx->tx_nwrq > 0); /* work items set up */
LASSERT(!tx->tx_queued); /* not queued for sending already */
kiblnd_connect_peer(kib_peer_t *peer)
{
struct rdma_cm_id *cmid;
- kib_dev_t *dev;
- kib_net_t *net = peer->ibp_ni->ni_data;
+ kib_dev_t *dev;
+ kib_net_t *net = peer->ibp_ni->ni_data;
struct sockaddr_in srcaddr;
struct sockaddr_in dstaddr;
- int rc;
+ int rc;
LASSERT(net != NULL);
LASSERT(peer->ibp_connecting > 0);
void
kiblnd_launch_tx(lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid)
{
- kib_peer_t *peer;
- kib_peer_t *peer2;
- kib_conn_t *conn;
- rwlock_t *g_lock = &kiblnd_data.kib_global_lock;
- unsigned long flags;
- int rc;
+ kib_peer_t *peer;
+ kib_peer_t *peer2;
+ kib_conn_t *conn;
+ rwlock_t *g_lock = &kiblnd_data.kib_global_lock;
+ unsigned long flags;
+ int rc;
/* If I get here, I've committed to send, so I complete the tx with
* failure on any problems */
int
kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg)
{
- lnet_hdr_t *hdr = &lntmsg->msg_hdr;
- int type = lntmsg->msg_type;
+ lnet_hdr_t *hdr = &lntmsg->msg_hdr;
+ int type = lntmsg->msg_type;
lnet_process_id_t target = lntmsg->msg_target;
- int target_is_router = lntmsg->msg_target_is_router;
- int routing = lntmsg->msg_routing;
- unsigned int payload_niov = lntmsg->msg_niov;
- struct kvec *payload_iov = lntmsg->msg_iov;
- lnet_kiov_t *payload_kiov = lntmsg->msg_kiov;
- unsigned int payload_offset = lntmsg->msg_offset;
- unsigned int payload_nob = lntmsg->msg_len;
- kib_msg_t *ibmsg;
- kib_tx_t *tx;
- int nob;
- int rc;
+ int target_is_router = lntmsg->msg_target_is_router;
+ int routing = lntmsg->msg_routing;
+ unsigned int payload_niov = lntmsg->msg_niov;
+ struct kvec *payload_iov = lntmsg->msg_iov;
+ lnet_kiov_t *payload_kiov = lntmsg->msg_kiov;
+ unsigned int payload_offset = lntmsg->msg_offset;
+ unsigned int payload_nob = lntmsg->msg_len;
+ kib_msg_t *ibmsg;
+ kib_tx_t *tx;
+ int nob;
+ int rc;
/* NB 'private' is different depending on what we're sending.... */
kiblnd_reply(lnet_ni_t *ni, kib_rx_t *rx, lnet_msg_t *lntmsg)
{
lnet_process_id_t target = lntmsg->msg_target;
- unsigned int niov = lntmsg->msg_niov;
- struct kvec *iov = lntmsg->msg_iov;
- lnet_kiov_t *kiov = lntmsg->msg_kiov;
- unsigned int offset = lntmsg->msg_offset;
- unsigned int nob = lntmsg->msg_len;
- kib_tx_t *tx;
- int rc;
+ unsigned int niov = lntmsg->msg_niov;
+ struct kvec *iov = lntmsg->msg_iov;
+ lnet_kiov_t *kiov = lntmsg->msg_kiov;
+ unsigned int offset = lntmsg->msg_offset;
+ unsigned int nob = lntmsg->msg_len;
+ kib_tx_t *tx;
+ int rc;
tx = kiblnd_get_idle_tx(ni, rx->rx_conn->ibc_peer->ibp_nid);
if (tx == NULL) {
unsigned int niov, struct kvec *iov, lnet_kiov_t *kiov,
unsigned int offset, unsigned int mlen, unsigned int rlen)
{
- kib_rx_t *rx = private;
- kib_msg_t *rxmsg = rx->rx_msg;
- kib_conn_t *conn = rx->rx_conn;
- kib_tx_t *tx;
- kib_msg_t *txmsg;
- int nob;
- int post_credit = IBLND_POSTRX_PEER_CREDIT;
- int rc = 0;
+ kib_rx_t *rx = private;
+ kib_msg_t *rxmsg = rx->rx_msg;
+ kib_conn_t *conn = rx->rx_conn;
+ kib_tx_t *tx;
+ kib_msg_t *txmsg;
+ int nob;
+ int post_credit = IBLND_POSTRX_PEER_CREDIT;
+ int rc = 0;
LASSERT(mlen <= rlen);
LASSERT(!in_interrupt());
static void
kiblnd_peer_notify(kib_peer_t *peer)
{
- int error = 0;
- unsigned long last_alive = 0;
+ int error = 0;
+ unsigned long last_alive = 0;
unsigned long flags;
read_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
* connection to be finished off by the connd. Otherwise the connd is
* already dealing with it (either to set it up or tear it down).
* Caller holds kib_global_lock exclusively in irq context */
- kib_peer_t *peer = conn->ibc_peer;
- kib_dev_t *dev;
- unsigned long flags;
+ kib_peer_t *peer = conn->ibc_peer;
+ kib_dev_t *dev;
+ unsigned long flags;
LASSERT(error != 0 || conn->ibc_state >= IBLND_CONN_ESTABLISHED);
static void
kiblnd_handle_early_rxs(kib_conn_t *conn)
{
- unsigned long flags;
- kib_rx_t *rx;
+ unsigned long flags;
+ kib_rx_t *rx;
kib_rx_t *tmp;
LASSERT(!in_interrupt());
kiblnd_abort_txs(kib_conn_t *conn, struct list_head *txs)
{
LIST_HEAD(zombies);
- struct list_head *tmp;
- struct list_head *nxt;
- kib_tx_t *tx;
+ struct list_head *tmp;
+ struct list_head *nxt;
+ kib_tx_t *tx;
spin_lock(&conn->ibc_lock);
kiblnd_peer_connect_failed(kib_peer_t *peer, int active, int error)
{
LIST_HEAD(zombies);
- unsigned long flags;
+ unsigned long flags;
LASSERT(error != 0);
LASSERT(!in_interrupt());
void
kiblnd_connreq_done(kib_conn_t *conn, int status)
{
- kib_peer_t *peer = conn->ibc_peer;
- kib_tx_t *tx;
+ kib_peer_t *peer = conn->ibc_peer;
+ kib_tx_t *tx;
kib_tx_t *tmp;
- struct list_head txs;
- unsigned long flags;
- int active;
+ struct list_head txs;
+ unsigned long flags;
+ int active;
active = (conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT);
static void
kiblnd_reject(struct rdma_cm_id *cmid, kib_rej_t *rej)
{
- int rc;
+ int rc;
rc = rdma_reject(cmid, rej, sizeof(*rej));
static int
kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob)
{
- rwlock_t *g_lock = &kiblnd_data.kib_global_lock;
- kib_msg_t *reqmsg = priv;
- kib_msg_t *ackmsg;
- kib_dev_t *ibdev;
- kib_peer_t *peer;
- kib_peer_t *peer2;
- kib_conn_t *conn;
- lnet_ni_t *ni = NULL;
- kib_net_t *net = NULL;
- lnet_nid_t nid;
+ rwlock_t *g_lock = &kiblnd_data.kib_global_lock;
+ kib_msg_t *reqmsg = priv;
+ kib_msg_t *ackmsg;
+ kib_dev_t *ibdev;
+ kib_peer_t *peer;
+ kib_peer_t *peer2;
+ kib_conn_t *conn;
+ lnet_ni_t *ni = NULL;
+ kib_net_t *net = NULL;
+ lnet_nid_t nid;
struct rdma_conn_param cp;
- kib_rej_t rej;
- int version = IBLND_MSG_VERSION;
- unsigned long flags;
- int rc;
- struct sockaddr_in *peer_addr;
+ kib_rej_t rej;
+ int version = IBLND_MSG_VERSION;
+ unsigned long flags;
+ int rc;
+ struct sockaddr_in *peer_addr;
LASSERT(!in_interrupt());
/* cmid inherits 'context' from the corresponding listener id */
LASSERT(ibdev != NULL);
memset(&rej, 0, sizeof(rej));
- rej.ibr_magic = IBLND_MSG_MAGIC;
- rej.ibr_why = IBLND_REJECT_FATAL;
+ rej.ibr_magic = IBLND_MSG_MAGIC;
+ rej.ibr_why = IBLND_REJECT_FATAL;
rej.ibr_cp.ibcp_max_msg_size = IBLND_MSG_SIZE;
peer_addr = (struct sockaddr_in *)&(cmid->route.addr.dst_addr);
}
nid = reqmsg->ibm_srcnid;
- ni = lnet_net2ni(LNET_NIDNET(reqmsg->ibm_dstnid));
+ ni = lnet_net2ni(LNET_NIDNET(reqmsg->ibm_dstnid));
if (ni != NULL) {
net = (kib_net_t *)ni->ni_data;
* CM callback doesn't destroy cmid. */
conn->ibc_incarnation = reqmsg->ibm_srcstamp;
- conn->ibc_credits = IBLND_MSG_QUEUE_SIZE(version);
+ conn->ibc_credits = IBLND_MSG_QUEUE_SIZE(version);
conn->ibc_reserved_credits = IBLND_MSG_QUEUE_SIZE(version);
LASSERT(conn->ibc_credits + conn->ibc_reserved_credits + IBLND_OOB_MSGS(version)
<= IBLND_RX_MSGS(version));
memset(&cp, 0, sizeof(cp));
cp.private_data = ackmsg;
- cp.private_data_len = ackmsg->ibm_nob;
+ cp.private_data_len = ackmsg->ibm_nob;
cp.responder_resources = 0; /* No atomic ops or RDMA reads */
- cp.initiator_depth = 0;
+ cp.initiator_depth = 0;
cp.flow_control = 1;
- cp.retry_count = *kiblnd_tunables.kib_retry_count;
- cp.rnr_retry_count = *kiblnd_tunables.kib_rnr_retry_count;
+ cp.retry_count = *kiblnd_tunables.kib_retry_count;
+ cp.rnr_retry_count = *kiblnd_tunables.kib_rnr_retry_count;
CDEBUG(D_NET, "Accept %s\n", libcfs_nid2str(nid));
if (ni != NULL)
lnet_ni_decref(ni);
- rej.ibr_version = version;
+ rej.ibr_version = version;
rej.ibr_cp.ibcp_queue_depth = IBLND_MSG_QUEUE_SIZE(version);
rej.ibr_cp.ibcp_max_frags = IBLND_RDMA_FRAGS(version);
kiblnd_reject(cmid, &rej);
kiblnd_reconnect(kib_conn_t *conn, int version,
__u64 incarnation, int why, kib_connparams_t *cp)
{
- kib_peer_t *peer = conn->ibc_peer;
- char *reason;
- int retry = 0;
- unsigned long flags;
+ kib_peer_t *peer = conn->ibc_peer;
+ char *reason;
+ int retry = 0;
+ unsigned long flags;
LASSERT(conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT);
LASSERT(peer->ibp_connecting > 0); /* 'conn' at least */
static void
kiblnd_rejected(kib_conn_t *conn, int reason, void *priv, int priv_nob)
{
- kib_peer_t *peer = conn->ibc_peer;
+ kib_peer_t *peer = conn->ibc_peer;
LASSERT(!in_interrupt());
LASSERT(conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT);
case IB_CM_REJ_CONSUMER_DEFINED:
if (priv_nob >= offsetof(kib_rej_t, ibr_padding)) {
- kib_rej_t *rej = priv;
- kib_connparams_t *cp = NULL;
- int flip = 0;
- __u64 incarnation = -1;
+ kib_rej_t *rej = priv;
+ kib_connparams_t *cp = NULL;
+ int flip = 0;
+ __u64 incarnation = -1;
/* NB. default incarnation is -1 because:
* a) V1 will ignore dst incarnation in connreq.
static void
kiblnd_check_connreply(kib_conn_t *conn, void *priv, int priv_nob)
{
- kib_peer_t *peer = conn->ibc_peer;
- lnet_ni_t *ni = peer->ibp_ni;
- kib_net_t *net = ni->ni_data;
- kib_msg_t *msg = priv;
- int ver = conn->ibc_version;
- int rc = kiblnd_unpack_msg(msg, priv_nob);
- unsigned long flags;
+ kib_peer_t *peer = conn->ibc_peer;
+ lnet_ni_t *ni = peer->ibp_ni;
+ kib_net_t *net = ni->ni_data;
+ kib_msg_t *msg = priv;
+ int ver = conn->ibc_version;
+ int rc = kiblnd_unpack_msg(msg, priv_nob);
+ unsigned long flags;
LASSERT(net != NULL);
goto failed;
}
- conn->ibc_incarnation = msg->ibm_srcstamp;
- conn->ibc_credits =
+ conn->ibc_incarnation = msg->ibm_srcstamp;
+ conn->ibc_credits =
conn->ibc_reserved_credits = IBLND_MSG_QUEUE_SIZE(ver);
LASSERT(conn->ibc_credits + conn->ibc_reserved_credits + IBLND_OOB_MSGS(ver)
<= IBLND_RX_MSGS(ver));
static int
kiblnd_active_connect(struct rdma_cm_id *cmid)
{
- kib_peer_t *peer = (kib_peer_t *)cmid->context;
- kib_conn_t *conn;
- kib_msg_t *msg;
- struct rdma_conn_param cp;
- int version;
- __u64 incarnation;
- unsigned long flags;
- int rc;
+ kib_peer_t *peer = (kib_peer_t *)cmid->context;
+ kib_conn_t *conn;
+ kib_msg_t *msg;
+ struct rdma_conn_param cp;
+ int version;
+ __u64 incarnation;
+ unsigned long flags;
+ int rc;
read_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
incarnation = peer->ibp_incarnation;
- version = (peer->ibp_version == 0) ? IBLND_MSG_VERSION :
- peer->ibp_version;
+ version = (peer->ibp_version == 0) ? IBLND_MSG_VERSION :
+ peer->ibp_version;
read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
cp.private_data_len = msg->ibm_nob;
cp.responder_resources = 0; /* No atomic ops or RDMA reads */
cp.initiator_depth = 0;
- cp.flow_control = 1;
- cp.retry_count = *kiblnd_tunables.kib_retry_count;
+ cp.flow_control = 1;
+ cp.retry_count = *kiblnd_tunables.kib_retry_count;
cp.rnr_retry_count = *kiblnd_tunables.kib_rnr_retry_count;
LASSERT(cmid->context == (void *)conn);
int
kiblnd_cm_callback(struct rdma_cm_id *cmid, struct rdma_cm_event *event)
{
- kib_peer_t *peer;
- kib_conn_t *conn;
- int rc;
+ kib_peer_t *peer;
+ kib_conn_t *conn;
+ int rc;
switch (event->event) {
default:
static int
kiblnd_check_txs_locked(kib_conn_t *conn, struct list_head *txs)
{
- kib_tx_t *tx;
- struct list_head *ttmp;
+ kib_tx_t *tx;
+ struct list_head *ttmp;
list_for_each(ttmp, txs) {
tx = list_entry(ttmp, kib_tx_t, tx_list);
{
LIST_HEAD(closes);
LIST_HEAD(checksends);
- struct list_head *peers = &kiblnd_data.kib_peers[idx];
- struct list_head *ptmp;
- kib_peer_t *peer;
- kib_conn_t *conn;
+ struct list_head *peers = &kiblnd_data.kib_peers[idx];
+ struct list_head *ptmp;
+ kib_peer_t *peer;
+ kib_conn_t *conn;
kib_conn_t *tmp;
- struct list_head *ctmp;
- unsigned long flags;
+ struct list_head *ctmp;
+ unsigned long flags;
/* NB. We expect to have a look at all the peers and not find any
* RDMAs to time out, so we just use a shared lock while we
int
kiblnd_connd(void *arg)
{
- wait_queue_t wait;
- unsigned long flags;
- kib_conn_t *conn;
- int timeout;
- int i;
- int dropped_lock;
- int peer_index = 0;
- unsigned long deadline = jiffies;
+ wait_queue_t wait;
+ unsigned long flags;
+ kib_conn_t *conn;
+ int timeout;
+ int i;
+ int dropped_lock;
+ int peer_index = 0;
+ unsigned long deadline = jiffies;
cfs_block_allsigs();
if (timeout <= 0) {
const int n = 4;
const int p = 1;
- int chunk = kiblnd_data.kib_peer_hash_size;
+ int chunk = kiblnd_data.kib_peer_hash_size;
spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, flags);
dropped_lock = 1;
* consuming my CQ I could be called after all completions have
* occurred. But in this case, ibc_nrx == 0 && ibc_nsends_posted == 0
* and this CQ is about to be destroyed so I NOOP. */
- kib_conn_t *conn = (kib_conn_t *)arg;
- struct kib_sched_info *sched = conn->ibc_sched;
- unsigned long flags;
+ kib_conn_t *conn = (kib_conn_t *)arg;
+ struct kib_sched_info *sched = conn->ibc_sched;
+ unsigned long flags;
LASSERT(cq == conn->ibc_cq);
int
kiblnd_scheduler(void *arg)
{
- long id = (long)arg;
- struct kib_sched_info *sched;
- kib_conn_t *conn;
- wait_queue_t wait;
- unsigned long flags;
- struct ib_wc wc;
- int did_something;
- int busy_loops = 0;
- int rc;
+ long id = (long)arg;
+ struct kib_sched_info *sched;
+ kib_conn_t *conn;
+ wait_queue_t wait;
+ unsigned long flags;
+ struct ib_wc wc;
+ int did_something;
+ int busy_loops = 0;
+ int rc;
cfs_block_allsigs();
int
kiblnd_failover_thread(void *arg)
{
- rwlock_t *glock = &kiblnd_data.kib_global_lock;
- kib_dev_t *dev;
- wait_queue_t wait;
- unsigned long flags;
- int rc;
+ rwlock_t *glock = &kiblnd_data.kib_global_lock;
+ kib_dev_t *dev;
+ wait_queue_t wait;
+ unsigned long flags;
+ int rc;
LASSERT(*kiblnd_tunables.kib_dev_failover != 0);
write_lock_irqsave(glock, flags);
while (!kiblnd_data.kib_shutdown) {
- int do_failover = 0;
- int long_sleep;
+ int do_failover = 0;
+ int long_sleep;
list_for_each_entry(dev, &kiblnd_data.kib_failed_devs,
ibd_fail_list) {
MODULE_PARM_DESC(use_privileged_port, "use privileged port when initiating connection");
kib_tunables_t kiblnd_tunables = {
- .kib_dev_failover = &dev_failover,
- .kib_service = &service,
- .kib_cksum = &cksum,
- .kib_timeout = &timeout,
- .kib_keepalive = &keepalive,
- .kib_ntx = &ntx,
- .kib_credits = &credits,
- .kib_peertxcredits = &peer_credits,
- .kib_peercredits_hiw = &peer_credits_hiw,
- .kib_peerrtrcredits = &peer_buffer_credits,
- .kib_peertimeout = &peer_timeout,
- .kib_default_ipif = &ipif_name,
- .kib_retry_count = &retry_count,
- .kib_rnr_retry_count = &rnr_retry_count,
- .kib_concurrent_sends = &concurrent_sends,
- .kib_ib_mtu = &ib_mtu,
- .kib_map_on_demand = &map_on_demand,
- .kib_fmr_pool_size = &fmr_pool_size,
- .kib_fmr_flush_trigger = &fmr_flush_trigger,
- .kib_fmr_cache = &fmr_cache,
- .kib_pmr_pool_size = &pmr_pool_size,
- .kib_require_priv_port = &require_privileged_port,
- .kib_use_priv_port = &use_privileged_port,
- .kib_nscheds = &nscheds
+ .kib_dev_failover = &dev_failover,
+ .kib_service = &service,
+ .kib_cksum = &cksum,
+ .kib_timeout = &timeout,
+ .kib_keepalive = &keepalive,
+ .kib_ntx = &ntx,
+ .kib_credits = &credits,
+ .kib_peertxcredits = &peer_credits,
+ .kib_peercredits_hiw = &peer_credits_hiw,
+ .kib_peerrtrcredits = &peer_buffer_credits,
+ .kib_peertimeout = &peer_timeout,
+ .kib_default_ipif = &ipif_name,
+ .kib_retry_count = &retry_count,
+ .kib_rnr_retry_count = &rnr_retry_count,
+ .kib_concurrent_sends = &concurrent_sends,
+ .kib_ib_mtu = &ib_mtu,
+ .kib_map_on_demand = &map_on_demand,
+ .kib_fmr_pool_size = &fmr_pool_size,
+ .kib_fmr_flush_trigger = &fmr_flush_trigger,
+ .kib_fmr_cache = &fmr_cache,
+ .kib_pmr_pool_size = &pmr_pool_size,
+ .kib_require_priv_port = &require_privileged_port,
+ .kib_use_priv_port = &use_privileged_port,
+ .kib_nscheds = &nscheds
};
int
projects / openwrt / staging / blogic.git / commitdiff