/* It shouldn't matter if the number of backchannel session slots
* doesn't match the number of RPC/RDMA credits. That just means
* one or the other will have extra slots that aren't used.
*/
static struct rpc_xprt *
xprt_setup_rdma_bc(struct xprt_create *args)
{
struct rpc_xprt *xprt;
struct rpcrdma_xprt *new_xprt;
if (args->addrlen > sizeof(xprt->addr)) {
dprintk("RPC: %s: address too large\n", __func__);
return ERR_PTR(-EBADF);
}
xprt = xprt_alloc(args->net, sizeof(*new_xprt),
RPCRDMA_MAX_BC_REQUESTS,
RPCRDMA_MAX_BC_REQUESTS);
if (!xprt) {
dprintk("RPC: %s: couldn't allocate rpc_xprt\n",
__func__);
return ERR_PTR(-ENOMEM);
}
xprt->timeout = &xprt_rdma_bc_timeout;
xprt_set_bound(xprt);
xprt_set_connected(xprt);
xprt->bind_timeout = RPCRDMA_BIND_TO;
xprt->reestablish_timeout = RPCRDMA_INIT_REEST_TO;
xprt->idle_timeout = RPCRDMA_IDLE_DISC_TO;
xprt->prot = XPRT_TRANSPORT_BC_RDMA;
xprt->tsh_size = RPCRDMA_HDRLEN_MIN / sizeof(__be32);
xprt->ops = &xprt_rdma_bc_procs;
memcpy(&xprt->addr, args->dstaddr, args->addrlen);
xprt->addrlen = args->addrlen;
xprt_rdma_format_addresses(xprt, (struct sockaddr *)&xprt->addr);
xprt->resvport = 0;
xprt->max_payload = xprt_rdma_max_inline_read;
new_xprt = rpcx_to_rdmax(xprt);
new_xprt->rx_buf.rb_bc_max_requests = xprt->max_reqs;
xprt_get(xprt);
args->bc_xprt->xpt_bc_xprt = xprt;
xprt->bc_xprt = args->bc_xprt;
if (!try_module_get(THIS_MODULE))
goto out_fail;
/* Final put for backchannel xprt is in __svc_rdma_free */
xprt_get(xprt);
return xprt;
out_fail:
xprt_rdma_free_addresses(xprt);
args->bc_xprt->xpt_bc_xprt = NULL;
args->bc_xprt->xpt_bc_xps = NULL;
xprt_put(xprt);
xprt_free(xprt);
return ERR_PTR(-EINVAL);
}
/**
* xprt_setup_rdma - Set up transport to use RDMA
*
* @args: rpc transport arguments
*/
static struct rpc_xprt *
xprt_setup_rdma(struct xprt_create *args)
{
struct rpcrdma_create_data_internal cdata;
struct rpc_xprt *xprt;
struct rpcrdma_xprt *new_xprt;
struct rpcrdma_ep *new_ep;
struct sockaddr *sap;
int rc;
if (args->addrlen > sizeof(xprt->addr)) {
dprintk("RPC: %s: address too large\n", __func__);
return ERR_PTR(-EBADF);
}
xprt = xprt_alloc(args->net, sizeof(struct rpcrdma_xprt), 0, 0);
if (xprt == NULL) {
dprintk("RPC: %s: couldn't allocate rpcrdma_xprt\n",
__func__);
return ERR_PTR(-ENOMEM);
}
/* 60 second timeout, no retries */
xprt->timeout = &xprt_rdma_default_timeout;
xprt->bind_timeout = RPCRDMA_BIND_TO;
xprt->reestablish_timeout = RPCRDMA_INIT_REEST_TO;
xprt->idle_timeout = RPCRDMA_IDLE_DISC_TO;
xprt->resvport = 0; /* privileged port not needed */
xprt->tsh_size = 0; /* RPC-RDMA handles framing */
xprt->ops = &xprt_rdma_procs;
/*
* Set up RDMA-specific connect data.
*/
sap = args->dstaddr;
/* Ensure xprt->addr holds valid server TCP (not RDMA)
* address, for any side protocols which peek at it */
xprt->prot = IPPROTO_TCP;
xprt->addrlen = args->addrlen;
memcpy(&xprt->addr, sap, xprt->addrlen);
if (rpc_get_port(sap))
xprt_set_bound(xprt);
xprt_rdma_format_addresses(xprt, sap);
cdata.max_requests = xprt_rdma_slot_table_entries;
cdata.rsize = RPCRDMA_MAX_SEGS * PAGE_SIZE; /* RDMA write max */
cdata.wsize = RPCRDMA_MAX_SEGS * PAGE_SIZE; /* RDMA read max */
cdata.inline_wsize = xprt_rdma_max_inline_write;
if (cdata.inline_wsize > cdata.wsize)
cdata.inline_wsize = cdata.wsize;
cdata.inline_rsize = xprt_rdma_max_inline_read;
if (cdata.inline_rsize > cdata.rsize)
cdata.inline_rsize = cdata.rsize;
/*
* Create new transport instance, which includes initialized
* o ia
* o endpoint
* o buffers
*/
new_xprt = rpcx_to_rdmax(xprt);
rc = rpcrdma_ia_open(new_xprt);
if (rc)
goto out1;
/*
* initialize and create ep
*/
new_xprt->rx_data = cdata;
new_ep = &new_xprt->rx_ep;
rc = rpcrdma_ep_create(&new_xprt->rx_ep,
&new_xprt->rx_ia, &new_xprt->rx_data);
if (rc)
goto out2;
rc = rpcrdma_buffer_create(new_xprt);
if (rc)
goto out3;
INIT_DELAYED_WORK(&new_xprt->rx_connect_worker,
xprt_rdma_connect_worker);
xprt->max_payload = new_xprt->rx_ia.ri_ops->ro_maxpages(new_xprt);
if (xprt->max_payload == 0)
goto out4;
xprt->max_payload <<= PAGE_SHIFT;
dprintk("RPC: %s: transport data payload maximum: %zu bytes\n",
__func__, xprt->max_payload);
if (!try_module_get(THIS_MODULE))
goto out4;
dprintk("RPC: %s: %s:%s\n", __func__,
xprt->address_strings[RPC_DISPLAY_ADDR],
xprt->address_strings[RPC_DISPLAY_PORT]);
trace_xprtrdma_create(new_xprt);
return xprt;
out4:
rpcrdma_buffer_destroy(&new_xprt->rx_buf);
rc = -ENODEV;
out3:
rpcrdma_ep_destroy(new_ep, &new_xprt->rx_ia);
out2:
rpcrdma_ia_close(&new_xprt->rx_ia);
out1:
trace_xprtrdma_destroy(new_xprt);
xprt_rdma_free_addresses(xprt);
xprt_free(xprt);
return ERR_PTR(rc);
}
/**
* rpcb_register - set or unset a port registration with the local rpcbind svc
* @prog: RPC program number to bind
* @vers: RPC version number to bind
* @prot: transport protocol to use to make this request
* @port: port value to register
* @okay: result code
*
* port == 0 means unregister, port != 0 means register.
*
* This routine supports only rpcbind version 2.
*/
int rpcb_register(u32 prog, u32 vers, int prot, unsigned short port, int *okay)
{
struct sockaddr_in sin = {
.sin_family = AF_INET,
.sin_addr.s_addr = htonl(INADDR_LOOPBACK),
};
struct rpcbind_args map = {
.r_prog = prog,
.r_vers = vers,
.r_prot = prot,
.r_port = port,
};
struct rpc_message msg = {
.rpc_proc = &rpcb_procedures2[port ?
RPCBPROC_SET : RPCBPROC_UNSET],
.rpc_argp = &map,
.rpc_resp = okay,
};
struct rpc_clnt *rpcb_clnt;
int error = 0;
dprintk("RPC: %sregistering (%u, %u, %d, %u) with local "
"rpcbind\n", (port ? "" : "un"),
prog, vers, prot, port);
rpcb_clnt = rpcb_create("localhost", (struct sockaddr *) &sin,
sizeof(sin), XPRT_TRANSPORT_UDP, 2, 1);
if (IS_ERR(rpcb_clnt))
return PTR_ERR(rpcb_clnt);
error = rpc_call_sync(rpcb_clnt, &msg, 0);
rpc_shutdown_client(rpcb_clnt);
if (error < 0)
printk(KERN_WARNING "RPC: failed to contact local rpcbind "
"server (errno %d).\n", -error);
dprintk("RPC: registration status %d/%d\n", error, *okay);
return error;
}
/**
* rpcb_getport_sync - obtain the port for an RPC service on a given host
* @sin: address of remote peer
* @prog: RPC program number to bind
* @vers: RPC version number to bind
* @prot: transport protocol to use to make this request
*
* Return value is the requested advertised port number,
* or a negative errno value.
*
* Called from outside the RPC client in a synchronous task context.
* Uses default timeout parameters specified by underlying transport.
*
* XXX: Needs to support IPv6
*/
int rpcb_getport_sync(struct sockaddr_in *sin, u32 prog, u32 vers, int prot)
{
struct rpcbind_args map = {
.r_prog = prog,
.r_vers = vers,
.r_prot = prot,
.r_port = 0,
};
struct rpc_message msg = {
.rpc_proc = &rpcb_procedures2[RPCBPROC_GETPORT],
.rpc_argp = &map,
.rpc_resp = &map.r_port,
};
struct rpc_clnt *rpcb_clnt;
int status;
dprintk("RPC: %s(" NIPQUAD_FMT ", %u, %u, %d)\n",
__FUNCTION__, NIPQUAD(sin->sin_addr.s_addr), prog, vers, prot);
rpcb_clnt = rpcb_create(NULL, (struct sockaddr *)sin,
sizeof(*sin), prot, 2, 0);
if (IS_ERR(rpcb_clnt))
return PTR_ERR(rpcb_clnt);
status = rpc_call_sync(rpcb_clnt, &msg, 0);
rpc_shutdown_client(rpcb_clnt);
if (status >= 0) {
if (map.r_port != 0)
return map.r_port;
status = -EACCES;
}
return status;
}
EXPORT_SYMBOL_GPL(rpcb_getport_sync);
static struct rpc_task *rpcb_call_async(struct rpc_clnt *rpcb_clnt, struct rpcbind_args *map, int version)
{
struct rpc_message msg = {
.rpc_proc = rpcb_next_version[version].rpc_proc,
.rpc_argp = map,
.rpc_resp = &map->r_port,
};
struct rpc_task_setup task_setup_data = {
.rpc_client = rpcb_clnt,
.rpc_message = &msg,
.callback_ops = &rpcb_getport_ops,
.callback_data = map,
.flags = RPC_TASK_ASYNC,
};
return rpc_run_task(&task_setup_data);
}
/**
* rpcb_getport_async - obtain the port for a given RPC service on a given host
* @task: task that is waiting for portmapper request
*
* This one can be called for an ongoing RPC request, and can be used in
* an async (rpciod) context.
*/
void rpcb_getport_async(struct rpc_task *task)
{
struct rpc_clnt *clnt = task->tk_client;
u32 bind_version;
struct rpc_xprt *xprt = task->tk_xprt;
struct rpc_clnt *rpcb_clnt;
static struct rpcbind_args *map;
struct rpc_task *child;
struct sockaddr_storage addr;
struct sockaddr *sap = (struct sockaddr *)&addr;
size_t salen;
int status;
struct rpcb_info *info;
dprintk("RPC: %5u %s(%s, %u, %u, %d)\n",
task->tk_pid, __FUNCTION__,
clnt->cl_server, clnt->cl_prog, clnt->cl_vers, xprt->prot);
/* Autobind on cloned rpc clients is discouraged */
BUG_ON(clnt->cl_parent != clnt);
if (xprt_test_and_set_binding(xprt)) {
status = -EAGAIN; /* tell caller to check again */
dprintk("RPC: %5u %s: waiting for another binder\n",
task->tk_pid, __FUNCTION__);
goto bailout_nowake;
}
/* Put self on queue before sending rpcbind request, in case
* rpcb_getport_done completes before we return from rpc_run_task */
rpc_sleep_on(&xprt->binding, task, NULL, NULL);
/* Someone else may have bound if we slept */
if (xprt_bound(xprt)) {
status = 0;
dprintk("RPC: %5u %s: already bound\n",
task->tk_pid, __FUNCTION__);
goto bailout_nofree;
}
salen = rpc_peeraddr(clnt, sap, sizeof(addr));
/* Don't ever use rpcbind v2 for AF_INET6 requests */
switch (sap->sa_family) {
case AF_INET:
info = rpcb_next_version;
break;
case AF_INET6:
info = rpcb_next_version6;
break;
default:
status = -EAFNOSUPPORT;
dprintk("RPC: %5u %s: bad address family\n",
task->tk_pid, __FUNCTION__);
goto bailout_nofree;
}
if (info[xprt->bind_index].rpc_proc == NULL) {
xprt->bind_index = 0;
status = -EPFNOSUPPORT;
dprintk("RPC: %5u %s: no more getport versions available\n",
task->tk_pid, __FUNCTION__);
goto bailout_nofree;
}
bind_version = info[xprt->bind_index].rpc_vers;
dprintk("RPC: %5u %s: trying rpcbind version %u\n",
task->tk_pid, __FUNCTION__, bind_version);
rpcb_clnt = rpcb_create(clnt->cl_server, sap, salen, xprt->prot,
bind_version, 0);
if (IS_ERR(rpcb_clnt)) {
status = PTR_ERR(rpcb_clnt);
dprintk("RPC: %5u %s: rpcb_create failed, error %ld\n",
task->tk_pid, __FUNCTION__, PTR_ERR(rpcb_clnt));
goto bailout_nofree;
}
map = kzalloc(sizeof(struct rpcbind_args), GFP_ATOMIC);
if (!map) {
status = -ENOMEM;
dprintk("RPC: %5u %s: no memory available\n",
task->tk_pid, __FUNCTION__);
goto bailout_nofree;
}
map->r_prog = clnt->cl_prog;
map->r_vers = clnt->cl_vers;
map->r_prot = xprt->prot;
map->r_port = 0;
map->r_xprt = xprt_get(xprt);
map->r_netid = rpc_peeraddr2str(clnt, RPC_DISPLAY_NETID);
map->r_addr = rpc_peeraddr2str(rpcb_clnt, RPC_DISPLAY_UNIVERSAL_ADDR);
map->r_owner = RPCB_OWNER_STRING; /* ignored for GETADDR */
child = rpcb_call_async(rpcb_clnt, map, xprt->bind_index);
rpc_release_client(rpcb_clnt);
if (IS_ERR(child)) {
status = -EIO;
dprintk("RPC: %5u %s: rpc_run_task failed\n",
task->tk_pid, __FUNCTION__);
goto bailout;
}
rpc_put_task(child);
task->tk_xprt->stat.bind_count++;
return;
bailout:
kfree(map);
xprt_put(xprt);
bailout_nofree:
rpcb_wake_rpcbind_waiters(xprt, status);
bailout_nowake:
task->tk_status = status;
}
EXPORT_SYMBOL_GPL(rpcb_getport_async);
/*
* Rpcbind child task calls this callback via tk_exit.
*/
static void rpcb_getport_done(struct rpc_task *child, void *data)
{
struct rpcbind_args *map = data;
struct rpc_xprt *xprt = map->r_xprt;
int status = child->tk_status;
/* Garbage reply: retry with a lesser rpcbind version */
if (status == -EIO)
status = -EPROTONOSUPPORT;
/* rpcbind server doesn't support this rpcbind protocol version */
if (status == -EPROTONOSUPPORT)
xprt->bind_index++;
if (status < 0) {
/* rpcbind server not available on remote host? */
xprt->ops->set_port(xprt, 0);
} else if (map->r_port == 0) {
/* Requested RPC service wasn't registered on remote host */
xprt->ops->set_port(xprt, 0);
status = -EACCES;
} else {
/* Succeeded */
xprt->ops->set_port(xprt, map->r_port);
xprt_set_bound(xprt);
status = 0;
}
dprintk("RPC: %5u rpcb_getport_done(status %d, port %u)\n",
child->tk_pid, status, map->r_port);
rpcb_wake_rpcbind_waiters(xprt, status);
}
static int rpcb_encode_mapping(struct rpc_rqst *req, __be32 *p,
struct rpcbind_args *rpcb)
{
dprintk("RPC: rpcb_encode_mapping(%u, %u, %d, %u)\n",
rpcb->r_prog, rpcb->r_vers, rpcb->r_prot, rpcb->r_port);
*p++ = htonl(rpcb->r_prog);
*p++ = htonl(rpcb->r_vers);
*p++ = htonl(rpcb->r_prot);
*p++ = htonl(rpcb->r_port);
req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
return 0;
}
static int rpcb_decode_getport(struct rpc_rqst *req, __be32 *p,
unsigned short *portp)
{
*portp = (unsigned short) ntohl(*p++);
dprintk("RPC: rpcb_decode_getport result %u\n",
*portp);
return 0;
}
static int rpcb_decode_set(struct rpc_rqst *req, __be32 *p,
unsigned int *boolp)
{
*boolp = (unsigned int) ntohl(*p++);
dprintk("RPC: rpcb_decode_set result %u\n",
*boolp);
return 0;
}
static int rpcb_encode_getaddr(struct rpc_rqst *req, __be32 *p,
struct rpcbind_args *rpcb)
{
dprintk("RPC: rpcb_encode_getaddr(%u, %u, %s)\n",
rpcb->r_prog, rpcb->r_vers, rpcb->r_addr);
*p++ = htonl(rpcb->r_prog);
*p++ = htonl(rpcb->r_vers);
p = xdr_encode_string(p, rpcb->r_netid);
p = xdr_encode_string(p, rpcb->r_addr);
p = xdr_encode_string(p, rpcb->r_owner);
req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
return 0;
}
/**
* xprt_setup_rdma - Set up transport to use RDMA
*
* @args: rpc transport arguments
*/
static struct rpc_xprt *
xprt_setup_rdma(struct xprt_create *args)
{
struct rpcrdma_create_data_internal cdata;
struct rpc_xprt *xprt;
struct rpcrdma_xprt *new_xprt;
struct rpcrdma_ep *new_ep;
struct sockaddr_in *sin;
int rc;
if (args->addrlen > sizeof(xprt->addr)) {
dprintk("RPC: %s: address too large\n", __func__);
return ERR_PTR(-EBADF);
}
xprt = xprt_alloc(args->net, sizeof(struct rpcrdma_xprt),
xprt_rdma_slot_table_entries,
xprt_rdma_slot_table_entries);
if (xprt == NULL) {
dprintk("RPC: %s: couldn't allocate rpcrdma_xprt\n",
__func__);
return ERR_PTR(-ENOMEM);
}
/* 60 second timeout, no retries */
xprt->timeout = &xprt_rdma_default_timeout;
xprt->bind_timeout = RPCRDMA_BIND_TO;
xprt->reestablish_timeout = RPCRDMA_INIT_REEST_TO;
xprt->idle_timeout = RPCRDMA_IDLE_DISC_TO;
xprt->resvport = 0; /* privileged port not needed */
xprt->tsh_size = 0; /* RPC-RDMA handles framing */
xprt->ops = &xprt_rdma_procs;
/*
* Set up RDMA-specific connect data.
*/
/* Put server RDMA address in local cdata */
memcpy(&cdata.addr, args->dstaddr, args->addrlen);
/* Ensure xprt->addr holds valid server TCP (not RDMA)
* address, for any side protocols which peek at it */
xprt->prot = IPPROTO_TCP;
xprt->addrlen = args->addrlen;
memcpy(&xprt->addr, &cdata.addr, xprt->addrlen);
sin = (struct sockaddr_in *)&cdata.addr;
if (ntohs(sin->sin_port) != 0)
xprt_set_bound(xprt);
dprintk("RPC: %s: %pI4:%u\n",
__func__, &sin->sin_addr.s_addr, ntohs(sin->sin_port));
/* Set max requests */
cdata.max_requests = xprt->max_reqs;
/* Set some length limits */
cdata.rsize = RPCRDMA_MAX_SEGS * PAGE_SIZE; /* RDMA write max */
cdata.wsize = RPCRDMA_MAX_SEGS * PAGE_SIZE; /* RDMA read max */
cdata.inline_wsize = xprt_rdma_max_inline_write;
if (cdata.inline_wsize > cdata.wsize)
cdata.inline_wsize = cdata.wsize;
cdata.inline_rsize = xprt_rdma_max_inline_read;
if (cdata.inline_rsize > cdata.rsize)
cdata.inline_rsize = cdata.rsize;
cdata.padding = xprt_rdma_inline_write_padding;
/*
* Create new transport instance, which includes initialized
* o ia
* o endpoint
* o buffers
*/
new_xprt = rpcx_to_rdmax(xprt);
rc = rpcrdma_ia_open(new_xprt, (struct sockaddr *) &cdata.addr,
xprt_rdma_memreg_strategy);
if (rc)
goto out1;
/*
* initialize and create ep
*/
new_xprt->rx_data = cdata;
new_ep = &new_xprt->rx_ep;
new_ep->rep_remote_addr = cdata.addr;
rc = rpcrdma_ep_create(&new_xprt->rx_ep,
&new_xprt->rx_ia, &new_xprt->rx_data);
if (rc)
goto out2;
/*
* Allocate pre-registered send and receive buffers for headers and
* any inline data. Also specify any padding which will be provided
* from a preregistered zero buffer.
*/
rc = rpcrdma_buffer_create(new_xprt);
if (rc)
goto out3;
/*
* Register a callback for connection events. This is necessary because
* connection loss notification is async. We also catch connection loss
* when reaping receives.
*/
INIT_DELAYED_WORK(&new_xprt->rx_connect_worker,
xprt_rdma_connect_worker);
xprt_rdma_format_addresses(xprt);
xprt->max_payload = rpcrdma_max_payload(new_xprt);
dprintk("RPC: %s: transport data payload maximum: %zu bytes\n",
__func__, xprt->max_payload);
if (!try_module_get(THIS_MODULE))
goto out4;
return xprt;
out4:
xprt_rdma_free_addresses(xprt);
rc = -EINVAL;
out3:
rpcrdma_ep_destroy(new_ep, &new_xprt->rx_ia);
out2:
rpcrdma_ia_close(&new_xprt->rx_ia);
out1:
xprt_free(xprt);
return ERR_PTR(rc);
}