Commit | Line | Data |
---|---|---|
a246b010 CL |
1 | /* |
2 | * linux/net/sunrpc/xprtsock.c | |
3 | * | |
4 | * Client-side transport implementation for sockets. | |
5 | * | |
113aa838 AC |
6 | * TCP callback races fixes (C) 1998 Red Hat |
7 | * TCP send fixes (C) 1998 Red Hat | |
a246b010 CL |
8 | * TCP NFS related read + write fixes |
9 | * (C) 1999 Dave Airlie, University of Limerick, Ireland <airlied@linux.ie> | |
10 | * | |
11 | * Rewrite of larges part of the code in order to stabilize TCP stuff. | |
12 | * Fix behaviour when socket buffer is full. | |
13 | * (C) 1999 Trond Myklebust <trond.myklebust@fys.uio.no> | |
55aa4f58 CL |
14 | * |
15 | * IP socket transport implementation, (C) 2005 Chuck Lever <cel@netapp.com> | |
8f9d5b1a CL |
16 | * |
17 | * IPv6 support contributed by Gilles Quillard, Bull Open Source, 2005. | |
18 | * <gilles.quillard@bull.net> | |
a246b010 CL |
19 | */ |
20 | ||
21 | #include <linux/types.h> | |
22 | #include <linux/slab.h> | |
bc25571e | 23 | #include <linux/module.h> |
a246b010 | 24 | #include <linux/capability.h> |
a246b010 CL |
25 | #include <linux/pagemap.h> |
26 | #include <linux/errno.h> | |
27 | #include <linux/socket.h> | |
28 | #include <linux/in.h> | |
29 | #include <linux/net.h> | |
30 | #include <linux/mm.h> | |
31 | #include <linux/udp.h> | |
32 | #include <linux/tcp.h> | |
33 | #include <linux/sunrpc/clnt.h> | |
02107148 | 34 | #include <linux/sunrpc/sched.h> |
49c36fcc | 35 | #include <linux/sunrpc/xprtsock.h> |
a246b010 | 36 | #include <linux/file.h> |
44b98efd RL |
37 | #ifdef CONFIG_NFS_V4_1 |
38 | #include <linux/sunrpc/bc_xprt.h> | |
39 | #endif | |
a246b010 CL |
40 | |
41 | #include <net/sock.h> | |
42 | #include <net/checksum.h> | |
43 | #include <net/udp.h> | |
44 | #include <net/tcp.h> | |
45 | ||
c556b754 CL |
46 | /* |
47 | * xprtsock tunables | |
48 | */ | |
49 | unsigned int xprt_udp_slot_table_entries = RPC_DEF_SLOT_TABLE; | |
50 | unsigned int xprt_tcp_slot_table_entries = RPC_DEF_SLOT_TABLE; | |
51 | ||
52 | unsigned int xprt_min_resvport = RPC_DEF_MIN_RESVPORT; | |
53 | unsigned int xprt_max_resvport = RPC_DEF_MAX_RESVPORT; | |
54 | ||
7d1e8255 | 55 | #define XS_TCP_LINGER_TO (15U * HZ) |
25fe6142 | 56 | static unsigned int xs_tcp_fin_timeout __read_mostly = XS_TCP_LINGER_TO; |
7d1e8255 | 57 | |
fbf76683 CL |
58 | /* |
59 | * We can register our own files under /proc/sys/sunrpc by | |
60 | * calling register_sysctl_table() again. The files in that | |
61 | * directory become the union of all files registered there. | |
62 | * | |
63 | * We simply need to make sure that we don't collide with | |
64 | * someone else's file names! | |
65 | */ | |
66 | ||
67 | #ifdef RPC_DEBUG | |
68 | ||
69 | static unsigned int min_slot_table_size = RPC_MIN_SLOT_TABLE; | |
70 | static unsigned int max_slot_table_size = RPC_MAX_SLOT_TABLE; | |
71 | static unsigned int xprt_min_resvport_limit = RPC_MIN_RESVPORT; | |
72 | static unsigned int xprt_max_resvport_limit = RPC_MAX_RESVPORT; | |
73 | ||
74 | static struct ctl_table_header *sunrpc_table_header; | |
75 | ||
76 | /* | |
77 | * FIXME: changing the UDP slot table size should also resize the UDP | |
78 | * socket buffers for existing UDP transports | |
79 | */ | |
80 | static ctl_table xs_tunables_table[] = { | |
81 | { | |
82 | .ctl_name = CTL_SLOTTABLE_UDP, | |
83 | .procname = "udp_slot_table_entries", | |
84 | .data = &xprt_udp_slot_table_entries, | |
85 | .maxlen = sizeof(unsigned int), | |
86 | .mode = 0644, | |
87 | .proc_handler = &proc_dointvec_minmax, | |
88 | .strategy = &sysctl_intvec, | |
89 | .extra1 = &min_slot_table_size, | |
90 | .extra2 = &max_slot_table_size | |
91 | }, | |
92 | { | |
93 | .ctl_name = CTL_SLOTTABLE_TCP, | |
94 | .procname = "tcp_slot_table_entries", | |
95 | .data = &xprt_tcp_slot_table_entries, | |
96 | .maxlen = sizeof(unsigned int), | |
97 | .mode = 0644, | |
98 | .proc_handler = &proc_dointvec_minmax, | |
99 | .strategy = &sysctl_intvec, | |
100 | .extra1 = &min_slot_table_size, | |
101 | .extra2 = &max_slot_table_size | |
102 | }, | |
103 | { | |
104 | .ctl_name = CTL_MIN_RESVPORT, | |
105 | .procname = "min_resvport", | |
106 | .data = &xprt_min_resvport, | |
107 | .maxlen = sizeof(unsigned int), | |
108 | .mode = 0644, | |
109 | .proc_handler = &proc_dointvec_minmax, | |
110 | .strategy = &sysctl_intvec, | |
111 | .extra1 = &xprt_min_resvport_limit, | |
112 | .extra2 = &xprt_max_resvport_limit | |
113 | }, | |
114 | { | |
115 | .ctl_name = CTL_MAX_RESVPORT, | |
116 | .procname = "max_resvport", | |
117 | .data = &xprt_max_resvport, | |
118 | .maxlen = sizeof(unsigned int), | |
119 | .mode = 0644, | |
120 | .proc_handler = &proc_dointvec_minmax, | |
121 | .strategy = &sysctl_intvec, | |
122 | .extra1 = &xprt_min_resvport_limit, | |
123 | .extra2 = &xprt_max_resvport_limit | |
124 | }, | |
25fe6142 TM |
125 | { |
126 | .procname = "tcp_fin_timeout", | |
127 | .data = &xs_tcp_fin_timeout, | |
128 | .maxlen = sizeof(xs_tcp_fin_timeout), | |
129 | .mode = 0644, | |
130 | .proc_handler = &proc_dointvec_jiffies, | |
131 | .strategy = sysctl_jiffies | |
132 | }, | |
fbf76683 CL |
133 | { |
134 | .ctl_name = 0, | |
135 | }, | |
136 | }; | |
137 | ||
138 | static ctl_table sunrpc_table[] = { | |
139 | { | |
140 | .ctl_name = CTL_SUNRPC, | |
141 | .procname = "sunrpc", | |
142 | .mode = 0555, | |
143 | .child = xs_tunables_table | |
144 | }, | |
145 | { | |
146 | .ctl_name = 0, | |
147 | }, | |
148 | }; | |
149 | ||
150 | #endif | |
151 | ||
03bf4b70 CL |
152 | /* |
153 | * Time out for an RPC UDP socket connect. UDP socket connects are | |
154 | * synchronous, but we set a timeout anyway in case of resource | |
155 | * exhaustion on the local host. | |
156 | */ | |
157 | #define XS_UDP_CONN_TO (5U * HZ) | |
158 | ||
159 | /* | |
160 | * Wait duration for an RPC TCP connection to be established. Solaris | |
161 | * NFS over TCP uses 60 seconds, for example, which is in line with how | |
162 | * long a server takes to reboot. | |
163 | */ | |
164 | #define XS_TCP_CONN_TO (60U * HZ) | |
165 | ||
166 | /* | |
167 | * Wait duration for a reply from the RPC portmapper. | |
168 | */ | |
169 | #define XS_BIND_TO (60U * HZ) | |
170 | ||
171 | /* | |
172 | * Delay if a UDP socket connect error occurs. This is most likely some | |
173 | * kind of resource problem on the local host. | |
174 | */ | |
175 | #define XS_UDP_REEST_TO (2U * HZ) | |
176 | ||
177 | /* | |
178 | * The reestablish timeout allows clients to delay for a bit before attempting | |
179 | * to reconnect to a server that just dropped our connection. | |
180 | * | |
181 | * We implement an exponential backoff when trying to reestablish a TCP | |
182 | * transport connection with the server. Some servers like to drop a TCP | |
183 | * connection when they are overworked, so we start with a short timeout and | |
184 | * increase over time if the server is down or not responding. | |
185 | */ | |
186 | #define XS_TCP_INIT_REEST_TO (3U * HZ) | |
187 | #define XS_TCP_MAX_REEST_TO (5U * 60 * HZ) | |
188 | ||
189 | /* | |
190 | * TCP idle timeout; client drops the transport socket if it is idle | |
191 | * for this long. Note that we also timeout UDP sockets to prevent | |
192 | * holding port numbers when there is no RPC traffic. | |
193 | */ | |
194 | #define XS_IDLE_DISC_TO (5U * 60 * HZ) | |
195 | ||
a246b010 CL |
196 | #ifdef RPC_DEBUG |
197 | # undef RPC_DEBUG_DATA | |
9903cd1c | 198 | # define RPCDBG_FACILITY RPCDBG_TRANS |
a246b010 CL |
199 | #endif |
200 | ||
a246b010 | 201 | #ifdef RPC_DEBUG_DATA |
9903cd1c | 202 | static void xs_pktdump(char *msg, u32 *packet, unsigned int count) |
a246b010 | 203 | { |
9903cd1c CL |
204 | u8 *buf = (u8 *) packet; |
205 | int j; | |
a246b010 | 206 | |
46121cf7 | 207 | dprintk("RPC: %s\n", msg); |
a246b010 CL |
208 | for (j = 0; j < count && j < 128; j += 4) { |
209 | if (!(j & 31)) { | |
210 | if (j) | |
211 | dprintk("\n"); | |
212 | dprintk("0x%04x ", j); | |
213 | } | |
214 | dprintk("%02x%02x%02x%02x ", | |
215 | buf[j], buf[j+1], buf[j+2], buf[j+3]); | |
216 | } | |
217 | dprintk("\n"); | |
218 | } | |
219 | #else | |
9903cd1c | 220 | static inline void xs_pktdump(char *msg, u32 *packet, unsigned int count) |
a246b010 CL |
221 | { |
222 | /* NOP */ | |
223 | } | |
224 | #endif | |
225 | ||
ffc2e518 CL |
226 | struct sock_xprt { |
227 | struct rpc_xprt xprt; | |
ee0ac0c2 CL |
228 | |
229 | /* | |
230 | * Network layer | |
231 | */ | |
232 | struct socket * sock; | |
233 | struct sock * inet; | |
51971139 CL |
234 | |
235 | /* | |
236 | * State of TCP reply receive | |
237 | */ | |
238 | __be32 tcp_fraghdr, | |
239 | tcp_xid; | |
240 | ||
241 | u32 tcp_offset, | |
242 | tcp_reclen; | |
243 | ||
244 | unsigned long tcp_copied, | |
245 | tcp_flags; | |
c8475461 CL |
246 | |
247 | /* | |
248 | * Connection of transports | |
249 | */ | |
34161db6 | 250 | struct delayed_work connect_worker; |
d3bc9a1d | 251 | struct sockaddr_storage addr; |
c8475461 | 252 | unsigned short port; |
7c6e066e CL |
253 | |
254 | /* | |
255 | * UDP socket buffer size parameters | |
256 | */ | |
257 | size_t rcvsize, | |
258 | sndsize; | |
314dfd79 CL |
259 | |
260 | /* | |
261 | * Saved socket callback addresses | |
262 | */ | |
263 | void (*old_data_ready)(struct sock *, int); | |
264 | void (*old_state_change)(struct sock *); | |
265 | void (*old_write_space)(struct sock *); | |
2a9e1cfa | 266 | void (*old_error_report)(struct sock *); |
ffc2e518 CL |
267 | }; |
268 | ||
e136d092 CL |
269 | /* |
270 | * TCP receive state flags | |
271 | */ | |
272 | #define TCP_RCV_LAST_FRAG (1UL << 0) | |
273 | #define TCP_RCV_COPY_FRAGHDR (1UL << 1) | |
274 | #define TCP_RCV_COPY_XID (1UL << 2) | |
275 | #define TCP_RCV_COPY_DATA (1UL << 3) | |
f4a2e418 RL |
276 | #define TCP_RCV_READ_CALLDIR (1UL << 4) |
277 | #define TCP_RCV_COPY_CALLDIR (1UL << 5) | |
18dca02a RL |
278 | |
279 | /* | |
280 | * TCP RPC flags | |
281 | */ | |
f4a2e418 | 282 | #define TCP_RPC_REPLY (1UL << 6) |
e136d092 | 283 | |
95392c59 CL |
284 | static inline struct sockaddr *xs_addr(struct rpc_xprt *xprt) |
285 | { | |
286 | return (struct sockaddr *) &xprt->addr; | |
287 | } | |
288 | ||
289 | static inline struct sockaddr_in *xs_addr_in(struct rpc_xprt *xprt) | |
edb267a6 | 290 | { |
95392c59 CL |
291 | return (struct sockaddr_in *) &xprt->addr; |
292 | } | |
293 | ||
294 | static inline struct sockaddr_in6 *xs_addr_in6(struct rpc_xprt *xprt) | |
295 | { | |
296 | return (struct sockaddr_in6 *) &xprt->addr; | |
297 | } | |
298 | ||
c877b849 | 299 | static void xs_format_common_peer_addresses(struct rpc_xprt *xprt) |
edb267a6 | 300 | { |
c877b849 CL |
301 | struct sockaddr *sap = xs_addr(xprt); |
302 | char buf[128]; | |
edb267a6 | 303 | |
c877b849 CL |
304 | (void)rpc_ntop(sap, buf, sizeof(buf)); |
305 | xprt->address_strings[RPC_DISPLAY_ADDR] = kstrdup(buf, GFP_KERNEL); | |
edb267a6 | 306 | |
c877b849 CL |
307 | (void)snprintf(buf, sizeof(buf), "%u", rpc_get_port(sap)); |
308 | xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL); | |
edb267a6 | 309 | |
c877b849 CL |
310 | (void)snprintf(buf, sizeof(buf), "addr=%s port=%s proto=%s", |
311 | xprt->address_strings[RPC_DISPLAY_ADDR], | |
312 | xprt->address_strings[RPC_DISPLAY_PORT], | |
313 | xprt->address_strings[RPC_DISPLAY_PROTO]); | |
314 | xprt->address_strings[RPC_DISPLAY_ALL] = kstrdup(buf, GFP_KERNEL); | |
fbfe3cc6 | 315 | |
c877b849 CL |
316 | (void)snprintf(buf, sizeof(buf), "%4hx", rpc_get_port(sap)); |
317 | xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL); | |
318 | } | |
fbfe3cc6 | 319 | |
c877b849 CL |
320 | static void xs_format_ipv4_peer_addresses(struct rpc_xprt *xprt, |
321 | const char *protocol, | |
322 | const char *netid) | |
323 | { | |
324 | struct sockaddr_in *sin = xs_addr_in(xprt); | |
325 | char buf[16]; | |
756805e7 | 326 | |
c877b849 | 327 | xprt->address_strings[RPC_DISPLAY_PROTO] = protocol; |
b454ae90 | 328 | xprt->address_strings[RPC_DISPLAY_NETID] = netid; |
c877b849 CL |
329 | |
330 | (void)snprintf(buf, sizeof(buf), "%02x%02x%02x%02x", | |
331 | NIPQUAD(sin->sin_addr.s_addr)); | |
332 | xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = kstrdup(buf, GFP_KERNEL); | |
333 | ||
334 | xs_format_common_peer_addresses(xprt); | |
edb267a6 CL |
335 | } |
336 | ||
b454ae90 CL |
337 | static void xs_format_ipv6_peer_addresses(struct rpc_xprt *xprt, |
338 | const char *protocol, | |
339 | const char *netid) | |
4b6473fb | 340 | { |
c877b849 CL |
341 | struct sockaddr_in6 *sin6 = xs_addr_in6(xprt); |
342 | char buf[48]; | |
4b6473fb | 343 | |
b454ae90 | 344 | xprt->address_strings[RPC_DISPLAY_PROTO] = protocol; |
c877b849 | 345 | xprt->address_strings[RPC_DISPLAY_NETID] = netid; |
4b6473fb | 346 | |
c877b849 CL |
347 | (void)snprintf(buf, sizeof(buf), "%pi6", &sin6->sin6_addr); |
348 | xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = kstrdup(buf, GFP_KERNEL); | |
756805e7 | 349 | |
c877b849 | 350 | xs_format_common_peer_addresses(xprt); |
edb267a6 CL |
351 | } |
352 | ||
353 | static void xs_free_peer_addresses(struct rpc_xprt *xprt) | |
354 | { | |
33e01dc7 CL |
355 | unsigned int i; |
356 | ||
357 | for (i = 0; i < RPC_DISPLAY_MAX; i++) | |
358 | switch (i) { | |
359 | case RPC_DISPLAY_PROTO: | |
360 | case RPC_DISPLAY_NETID: | |
361 | continue; | |
362 | default: | |
363 | kfree(xprt->address_strings[i]); | |
364 | } | |
edb267a6 CL |
365 | } |
366 | ||
b4b5cc85 CL |
367 | #define XS_SENDMSG_FLAGS (MSG_DONTWAIT | MSG_NOSIGNAL) |
368 | ||
24c5684b | 369 | static int xs_send_kvec(struct socket *sock, struct sockaddr *addr, int addrlen, struct kvec *vec, unsigned int base, int more) |
b4b5cc85 | 370 | { |
b4b5cc85 CL |
371 | struct msghdr msg = { |
372 | .msg_name = addr, | |
373 | .msg_namelen = addrlen, | |
24c5684b TM |
374 | .msg_flags = XS_SENDMSG_FLAGS | (more ? MSG_MORE : 0), |
375 | }; | |
376 | struct kvec iov = { | |
377 | .iov_base = vec->iov_base + base, | |
378 | .iov_len = vec->iov_len - base, | |
b4b5cc85 CL |
379 | }; |
380 | ||
24c5684b | 381 | if (iov.iov_len != 0) |
b4b5cc85 CL |
382 | return kernel_sendmsg(sock, &msg, &iov, 1, iov.iov_len); |
383 | return kernel_sendmsg(sock, &msg, NULL, 0, 0); | |
384 | } | |
385 | ||
24c5684b | 386 | static int xs_send_pagedata(struct socket *sock, struct xdr_buf *xdr, unsigned int base, int more) |
b4b5cc85 | 387 | { |
24c5684b TM |
388 | struct page **ppage; |
389 | unsigned int remainder; | |
390 | int err, sent = 0; | |
391 | ||
392 | remainder = xdr->page_len - base; | |
393 | base += xdr->page_base; | |
394 | ppage = xdr->pages + (base >> PAGE_SHIFT); | |
395 | base &= ~PAGE_MASK; | |
396 | for(;;) { | |
397 | unsigned int len = min_t(unsigned int, PAGE_SIZE - base, remainder); | |
398 | int flags = XS_SENDMSG_FLAGS; | |
b4b5cc85 | 399 | |
24c5684b TM |
400 | remainder -= len; |
401 | if (remainder != 0 || more) | |
402 | flags |= MSG_MORE; | |
403 | err = sock->ops->sendpage(sock, *ppage, base, len, flags); | |
404 | if (remainder == 0 || err != len) | |
405 | break; | |
406 | sent += err; | |
407 | ppage++; | |
408 | base = 0; | |
409 | } | |
410 | if (sent == 0) | |
411 | return err; | |
412 | if (err > 0) | |
413 | sent += err; | |
414 | return sent; | |
b4b5cc85 CL |
415 | } |
416 | ||
9903cd1c CL |
417 | /** |
418 | * xs_sendpages - write pages directly to a socket | |
419 | * @sock: socket to send on | |
420 | * @addr: UDP only -- address of destination | |
421 | * @addrlen: UDP only -- length of destination address | |
422 | * @xdr: buffer containing this request | |
423 | * @base: starting position in the buffer | |
424 | * | |
a246b010 | 425 | */ |
24c5684b | 426 | static int xs_sendpages(struct socket *sock, struct sockaddr *addr, int addrlen, struct xdr_buf *xdr, unsigned int base) |
a246b010 | 427 | { |
24c5684b TM |
428 | unsigned int remainder = xdr->len - base; |
429 | int err, sent = 0; | |
a246b010 | 430 | |
262965f5 | 431 | if (unlikely(!sock)) |
fba91afb | 432 | return -ENOTSOCK; |
262965f5 CL |
433 | |
434 | clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags); | |
24c5684b TM |
435 | if (base != 0) { |
436 | addr = NULL; | |
437 | addrlen = 0; | |
438 | } | |
262965f5 | 439 | |
24c5684b TM |
440 | if (base < xdr->head[0].iov_len || addr != NULL) { |
441 | unsigned int len = xdr->head[0].iov_len - base; | |
442 | remainder -= len; | |
443 | err = xs_send_kvec(sock, addr, addrlen, &xdr->head[0], base, remainder != 0); | |
444 | if (remainder == 0 || err != len) | |
a246b010 | 445 | goto out; |
24c5684b | 446 | sent += err; |
a246b010 CL |
447 | base = 0; |
448 | } else | |
24c5684b | 449 | base -= xdr->head[0].iov_len; |
a246b010 | 450 | |
24c5684b TM |
451 | if (base < xdr->page_len) { |
452 | unsigned int len = xdr->page_len - base; | |
453 | remainder -= len; | |
454 | err = xs_send_pagedata(sock, xdr, base, remainder != 0); | |
455 | if (remainder == 0 || err != len) | |
a246b010 | 456 | goto out; |
24c5684b | 457 | sent += err; |
a246b010 | 458 | base = 0; |
24c5684b TM |
459 | } else |
460 | base -= xdr->page_len; | |
461 | ||
462 | if (base >= xdr->tail[0].iov_len) | |
463 | return sent; | |
464 | err = xs_send_kvec(sock, NULL, 0, &xdr->tail[0], base, 0); | |
a246b010 | 465 | out: |
24c5684b TM |
466 | if (sent == 0) |
467 | return err; | |
468 | if (err > 0) | |
469 | sent += err; | |
470 | return sent; | |
a246b010 CL |
471 | } |
472 | ||
b6ddf64f TM |
473 | static void xs_nospace_callback(struct rpc_task *task) |
474 | { | |
475 | struct sock_xprt *transport = container_of(task->tk_rqstp->rq_xprt, struct sock_xprt, xprt); | |
476 | ||
477 | transport->inet->sk_write_pending--; | |
478 | clear_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags); | |
479 | } | |
480 | ||
9903cd1c | 481 | /** |
262965f5 CL |
482 | * xs_nospace - place task on wait queue if transmit was incomplete |
483 | * @task: task to put to sleep | |
9903cd1c | 484 | * |
a246b010 | 485 | */ |
5e3771ce | 486 | static int xs_nospace(struct rpc_task *task) |
a246b010 | 487 | { |
262965f5 CL |
488 | struct rpc_rqst *req = task->tk_rqstp; |
489 | struct rpc_xprt *xprt = req->rq_xprt; | |
ee0ac0c2 | 490 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
5e3771ce | 491 | int ret = 0; |
a246b010 | 492 | |
46121cf7 | 493 | dprintk("RPC: %5u xmit incomplete (%u left of %u)\n", |
262965f5 CL |
494 | task->tk_pid, req->rq_slen - req->rq_bytes_sent, |
495 | req->rq_slen); | |
496 | ||
b6ddf64f TM |
497 | /* Protect against races with write_space */ |
498 | spin_lock_bh(&xprt->transport_lock); | |
499 | ||
500 | /* Don't race with disconnect */ | |
501 | if (xprt_connected(xprt)) { | |
502 | if (test_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags)) { | |
5e3771ce | 503 | ret = -EAGAIN; |
b6ddf64f TM |
504 | /* |
505 | * Notify TCP that we're limited by the application | |
506 | * window size | |
507 | */ | |
508 | set_bit(SOCK_NOSPACE, &transport->sock->flags); | |
509 | transport->inet->sk_write_pending++; | |
510 | /* ...and wait for more buffer space */ | |
511 | xprt_wait_for_buffer_space(task, xs_nospace_callback); | |
512 | } | |
513 | } else { | |
514 | clear_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags); | |
5e3771ce | 515 | ret = -ENOTCONN; |
b6ddf64f | 516 | } |
262965f5 | 517 | |
b6ddf64f | 518 | spin_unlock_bh(&xprt->transport_lock); |
5e3771ce | 519 | return ret; |
262965f5 CL |
520 | } |
521 | ||
522 | /** | |
523 | * xs_udp_send_request - write an RPC request to a UDP socket | |
524 | * @task: address of RPC task that manages the state of an RPC request | |
525 | * | |
526 | * Return values: | |
527 | * 0: The request has been sent | |
528 | * EAGAIN: The socket was blocked, please call again later to | |
529 | * complete the request | |
530 | * ENOTCONN: Caller needs to invoke connect logic then call again | |
531 | * other: Some other error occured, the request was not sent | |
532 | */ | |
533 | static int xs_udp_send_request(struct rpc_task *task) | |
534 | { | |
535 | struct rpc_rqst *req = task->tk_rqstp; | |
536 | struct rpc_xprt *xprt = req->rq_xprt; | |
ee0ac0c2 | 537 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
262965f5 CL |
538 | struct xdr_buf *xdr = &req->rq_snd_buf; |
539 | int status; | |
a246b010 | 540 | |
9903cd1c | 541 | xs_pktdump("packet data:", |
a246b010 CL |
542 | req->rq_svec->iov_base, |
543 | req->rq_svec->iov_len); | |
544 | ||
01d37c42 TM |
545 | if (!xprt_bound(xprt)) |
546 | return -ENOTCONN; | |
ee0ac0c2 | 547 | status = xs_sendpages(transport->sock, |
95392c59 | 548 | xs_addr(xprt), |
ee0ac0c2 CL |
549 | xprt->addrlen, xdr, |
550 | req->rq_bytes_sent); | |
a246b010 | 551 | |
46121cf7 | 552 | dprintk("RPC: xs_udp_send_request(%u) = %d\n", |
262965f5 | 553 | xdr->len - req->rq_bytes_sent, status); |
a246b010 | 554 | |
2199700f TM |
555 | if (status >= 0) { |
556 | task->tk_bytes_sent += status; | |
557 | if (status >= req->rq_slen) | |
558 | return 0; | |
559 | /* Still some bytes left; set up for a retry later. */ | |
262965f5 | 560 | status = -EAGAIN; |
2199700f | 561 | } |
c8485e4d TM |
562 | if (!transport->sock) |
563 | goto out; | |
a246b010 | 564 | |
262965f5 | 565 | switch (status) { |
fba91afb TM |
566 | case -ENOTSOCK: |
567 | status = -ENOTCONN; | |
568 | /* Should we call xs_close() here? */ | |
569 | break; | |
b6ddf64f | 570 | case -EAGAIN: |
5e3771ce | 571 | status = xs_nospace(task); |
b6ddf64f | 572 | break; |
c8485e4d TM |
573 | default: |
574 | dprintk("RPC: sendmsg returned unrecognized error %d\n", | |
575 | -status); | |
262965f5 CL |
576 | case -ENETUNREACH: |
577 | case -EPIPE: | |
a246b010 CL |
578 | case -ECONNREFUSED: |
579 | /* When the server has died, an ICMP port unreachable message | |
9903cd1c | 580 | * prompts ECONNREFUSED. */ |
b6ddf64f | 581 | clear_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags); |
a246b010 | 582 | } |
c8485e4d | 583 | out: |
262965f5 | 584 | return status; |
a246b010 CL |
585 | } |
586 | ||
e06799f9 TM |
587 | /** |
588 | * xs_tcp_shutdown - gracefully shut down a TCP socket | |
589 | * @xprt: transport | |
590 | * | |
591 | * Initiates a graceful shutdown of the TCP socket by calling the | |
592 | * equivalent of shutdown(SHUT_WR); | |
593 | */ | |
594 | static void xs_tcp_shutdown(struct rpc_xprt *xprt) | |
595 | { | |
596 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); | |
597 | struct socket *sock = transport->sock; | |
598 | ||
599 | if (sock != NULL) | |
600 | kernel_sock_shutdown(sock, SHUT_WR); | |
601 | } | |
602 | ||
808012fb CL |
603 | static inline void xs_encode_tcp_record_marker(struct xdr_buf *buf) |
604 | { | |
605 | u32 reclen = buf->len - sizeof(rpc_fraghdr); | |
606 | rpc_fraghdr *base = buf->head[0].iov_base; | |
607 | *base = htonl(RPC_LAST_STREAM_FRAGMENT | reclen); | |
608 | } | |
609 | ||
9903cd1c | 610 | /** |
262965f5 | 611 | * xs_tcp_send_request - write an RPC request to a TCP socket |
9903cd1c CL |
612 | * @task: address of RPC task that manages the state of an RPC request |
613 | * | |
614 | * Return values: | |
262965f5 CL |
615 | * 0: The request has been sent |
616 | * EAGAIN: The socket was blocked, please call again later to | |
617 | * complete the request | |
618 | * ENOTCONN: Caller needs to invoke connect logic then call again | |
619 | * other: Some other error occured, the request was not sent | |
9903cd1c CL |
620 | * |
621 | * XXX: In the case of soft timeouts, should we eventually give up | |
262965f5 | 622 | * if sendmsg is not able to make progress? |
9903cd1c | 623 | */ |
262965f5 | 624 | static int xs_tcp_send_request(struct rpc_task *task) |
a246b010 CL |
625 | { |
626 | struct rpc_rqst *req = task->tk_rqstp; | |
627 | struct rpc_xprt *xprt = req->rq_xprt; | |
ee0ac0c2 | 628 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
262965f5 | 629 | struct xdr_buf *xdr = &req->rq_snd_buf; |
b595bb15 | 630 | int status; |
a246b010 | 631 | |
808012fb | 632 | xs_encode_tcp_record_marker(&req->rq_snd_buf); |
a246b010 | 633 | |
262965f5 CL |
634 | xs_pktdump("packet data:", |
635 | req->rq_svec->iov_base, | |
636 | req->rq_svec->iov_len); | |
a246b010 CL |
637 | |
638 | /* Continue transmitting the packet/record. We must be careful | |
639 | * to cope with writespace callbacks arriving _after_ we have | |
262965f5 | 640 | * called sendmsg(). */ |
a246b010 | 641 | while (1) { |
ee0ac0c2 CL |
642 | status = xs_sendpages(transport->sock, |
643 | NULL, 0, xdr, req->rq_bytes_sent); | |
a246b010 | 644 | |
46121cf7 | 645 | dprintk("RPC: xs_tcp_send_request(%u) = %d\n", |
262965f5 | 646 | xdr->len - req->rq_bytes_sent, status); |
a246b010 | 647 | |
262965f5 | 648 | if (unlikely(status < 0)) |
a246b010 | 649 | break; |
a246b010 | 650 | |
262965f5 CL |
651 | /* If we've sent the entire packet, immediately |
652 | * reset the count of bytes sent. */ | |
653 | req->rq_bytes_sent += status; | |
ef759a2e | 654 | task->tk_bytes_sent += status; |
262965f5 CL |
655 | if (likely(req->rq_bytes_sent >= req->rq_slen)) { |
656 | req->rq_bytes_sent = 0; | |
657 | return 0; | |
658 | } | |
a246b010 | 659 | |
06b4b681 TM |
660 | if (status != 0) |
661 | continue; | |
a246b010 | 662 | status = -EAGAIN; |
06b4b681 | 663 | break; |
a246b010 | 664 | } |
c8485e4d TM |
665 | if (!transport->sock) |
666 | goto out; | |
a246b010 | 667 | |
262965f5 | 668 | switch (status) { |
fba91afb TM |
669 | case -ENOTSOCK: |
670 | status = -ENOTCONN; | |
671 | /* Should we call xs_close() here? */ | |
672 | break; | |
262965f5 | 673 | case -EAGAIN: |
5e3771ce | 674 | status = xs_nospace(task); |
262965f5 | 675 | break; |
c8485e4d TM |
676 | default: |
677 | dprintk("RPC: sendmsg returned unrecognized error %d\n", | |
678 | -status); | |
262965f5 | 679 | case -ECONNRESET: |
55420c24 | 680 | case -EPIPE: |
2a9e1cfa TM |
681 | xs_tcp_shutdown(xprt); |
682 | case -ECONNREFUSED: | |
262965f5 | 683 | case -ENOTCONN: |
b6ddf64f | 684 | clear_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags); |
a246b010 | 685 | } |
c8485e4d | 686 | out: |
a246b010 CL |
687 | return status; |
688 | } | |
689 | ||
e0ab53de TM |
690 | /** |
691 | * xs_tcp_release_xprt - clean up after a tcp transmission | |
692 | * @xprt: transport | |
693 | * @task: rpc task | |
694 | * | |
695 | * This cleans up if an error causes us to abort the transmission of a request. | |
696 | * In this case, the socket may need to be reset in order to avoid confusing | |
697 | * the server. | |
698 | */ | |
699 | static void xs_tcp_release_xprt(struct rpc_xprt *xprt, struct rpc_task *task) | |
700 | { | |
701 | struct rpc_rqst *req; | |
702 | ||
703 | if (task != xprt->snd_task) | |
704 | return; | |
705 | if (task == NULL) | |
706 | goto out_release; | |
707 | req = task->tk_rqstp; | |
708 | if (req->rq_bytes_sent == 0) | |
709 | goto out_release; | |
710 | if (req->rq_bytes_sent == req->rq_snd_buf.len) | |
711 | goto out_release; | |
712 | set_bit(XPRT_CLOSE_WAIT, &task->tk_xprt->state); | |
713 | out_release: | |
714 | xprt_release_xprt(xprt, task); | |
715 | } | |
716 | ||
2a9e1cfa TM |
717 | static void xs_save_old_callbacks(struct sock_xprt *transport, struct sock *sk) |
718 | { | |
719 | transport->old_data_ready = sk->sk_data_ready; | |
720 | transport->old_state_change = sk->sk_state_change; | |
721 | transport->old_write_space = sk->sk_write_space; | |
722 | transport->old_error_report = sk->sk_error_report; | |
723 | } | |
724 | ||
725 | static void xs_restore_old_callbacks(struct sock_xprt *transport, struct sock *sk) | |
726 | { | |
727 | sk->sk_data_ready = transport->old_data_ready; | |
728 | sk->sk_state_change = transport->old_state_change; | |
729 | sk->sk_write_space = transport->old_write_space; | |
730 | sk->sk_error_report = transport->old_error_report; | |
731 | } | |
732 | ||
fe315e76 | 733 | static void xs_reset_transport(struct sock_xprt *transport) |
a246b010 | 734 | { |
ee0ac0c2 CL |
735 | struct socket *sock = transport->sock; |
736 | struct sock *sk = transport->inet; | |
a246b010 | 737 | |
fe315e76 CL |
738 | if (sk == NULL) |
739 | return; | |
9903cd1c | 740 | |
a246b010 | 741 | write_lock_bh(&sk->sk_callback_lock); |
ee0ac0c2 CL |
742 | transport->inet = NULL; |
743 | transport->sock = NULL; | |
a246b010 | 744 | |
9903cd1c | 745 | sk->sk_user_data = NULL; |
2a9e1cfa TM |
746 | |
747 | xs_restore_old_callbacks(transport, sk); | |
a246b010 CL |
748 | write_unlock_bh(&sk->sk_callback_lock); |
749 | ||
9903cd1c | 750 | sk->sk_no_check = 0; |
a246b010 CL |
751 | |
752 | sock_release(sock); | |
fe315e76 CL |
753 | } |
754 | ||
755 | /** | |
756 | * xs_close - close a socket | |
757 | * @xprt: transport | |
758 | * | |
759 | * This is used when all requests are complete; ie, no DRC state remains | |
760 | * on the server we want to save. | |
f75e6745 TM |
761 | * |
762 | * The caller _must_ be holding XPRT_LOCKED in order to avoid issues with | |
763 | * xs_reset_transport() zeroing the socket from underneath a writer. | |
fe315e76 CL |
764 | */ |
765 | static void xs_close(struct rpc_xprt *xprt) | |
766 | { | |
767 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); | |
768 | ||
769 | dprintk("RPC: xs_close xprt %p\n", xprt); | |
770 | ||
771 | xs_reset_transport(transport); | |
772 | ||
632e3bdc | 773 | smp_mb__before_clear_bit(); |
7d1e8255 | 774 | clear_bit(XPRT_CONNECTION_ABORT, &xprt->state); |
632e3bdc | 775 | clear_bit(XPRT_CLOSE_WAIT, &xprt->state); |
3b948ae5 | 776 | clear_bit(XPRT_CLOSING, &xprt->state); |
632e3bdc | 777 | smp_mb__after_clear_bit(); |
62da3b24 | 778 | xprt_disconnect_done(xprt); |
a246b010 CL |
779 | } |
780 | ||
f75e6745 TM |
781 | static void xs_tcp_close(struct rpc_xprt *xprt) |
782 | { | |
783 | if (test_and_clear_bit(XPRT_CONNECTION_CLOSE, &xprt->state)) | |
784 | xs_close(xprt); | |
785 | else | |
786 | xs_tcp_shutdown(xprt); | |
787 | } | |
788 | ||
9903cd1c CL |
789 | /** |
790 | * xs_destroy - prepare to shutdown a transport | |
791 | * @xprt: doomed transport | |
792 | * | |
793 | */ | |
794 | static void xs_destroy(struct rpc_xprt *xprt) | |
a246b010 | 795 | { |
c8475461 CL |
796 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
797 | ||
46121cf7 | 798 | dprintk("RPC: xs_destroy xprt %p\n", xprt); |
9903cd1c | 799 | |
c1384c9c | 800 | cancel_rearming_delayed_work(&transport->connect_worker); |
a246b010 | 801 | |
9903cd1c | 802 | xs_close(xprt); |
edb267a6 | 803 | xs_free_peer_addresses(xprt); |
a246b010 | 804 | kfree(xprt->slot); |
c8541ecd | 805 | kfree(xprt); |
bc25571e | 806 | module_put(THIS_MODULE); |
a246b010 CL |
807 | } |
808 | ||
9903cd1c CL |
809 | static inline struct rpc_xprt *xprt_from_sock(struct sock *sk) |
810 | { | |
811 | return (struct rpc_xprt *) sk->sk_user_data; | |
812 | } | |
813 | ||
814 | /** | |
815 | * xs_udp_data_ready - "data ready" callback for UDP sockets | |
816 | * @sk: socket with data to read | |
817 | * @len: how much data to read | |
818 | * | |
a246b010 | 819 | */ |
9903cd1c | 820 | static void xs_udp_data_ready(struct sock *sk, int len) |
a246b010 | 821 | { |
9903cd1c CL |
822 | struct rpc_task *task; |
823 | struct rpc_xprt *xprt; | |
a246b010 | 824 | struct rpc_rqst *rovr; |
9903cd1c | 825 | struct sk_buff *skb; |
a246b010 | 826 | int err, repsize, copied; |
d8ed029d AD |
827 | u32 _xid; |
828 | __be32 *xp; | |
a246b010 CL |
829 | |
830 | read_lock(&sk->sk_callback_lock); | |
46121cf7 | 831 | dprintk("RPC: xs_udp_data_ready...\n"); |
9903cd1c | 832 | if (!(xprt = xprt_from_sock(sk))) |
a246b010 | 833 | goto out; |
a246b010 CL |
834 | |
835 | if ((skb = skb_recv_datagram(sk, 0, 1, &err)) == NULL) | |
836 | goto out; | |
837 | ||
838 | if (xprt->shutdown) | |
839 | goto dropit; | |
840 | ||
841 | repsize = skb->len - sizeof(struct udphdr); | |
842 | if (repsize < 4) { | |
46121cf7 | 843 | dprintk("RPC: impossible RPC reply size %d!\n", repsize); |
a246b010 CL |
844 | goto dropit; |
845 | } | |
846 | ||
847 | /* Copy the XID from the skb... */ | |
848 | xp = skb_header_pointer(skb, sizeof(struct udphdr), | |
849 | sizeof(_xid), &_xid); | |
850 | if (xp == NULL) | |
851 | goto dropit; | |
852 | ||
853 | /* Look up and lock the request corresponding to the given XID */ | |
4a0f8c04 | 854 | spin_lock(&xprt->transport_lock); |
a246b010 CL |
855 | rovr = xprt_lookup_rqst(xprt, *xp); |
856 | if (!rovr) | |
857 | goto out_unlock; | |
858 | task = rovr->rq_task; | |
859 | ||
a246b010 CL |
860 | if ((copied = rovr->rq_private_buf.buflen) > repsize) |
861 | copied = repsize; | |
862 | ||
863 | /* Suck it into the iovec, verify checksum if not done by hw. */ | |
1781f7f5 HX |
864 | if (csum_partial_copy_to_xdr(&rovr->rq_private_buf, skb)) { |
865 | UDPX_INC_STATS_BH(sk, UDP_MIB_INERRORS); | |
a246b010 | 866 | goto out_unlock; |
1781f7f5 HX |
867 | } |
868 | ||
869 | UDPX_INC_STATS_BH(sk, UDP_MIB_INDATAGRAMS); | |
a246b010 CL |
870 | |
871 | /* Something worked... */ | |
adf30907 | 872 | dst_confirm(skb_dst(skb)); |
a246b010 | 873 | |
1570c1e4 CL |
874 | xprt_adjust_cwnd(task, copied); |
875 | xprt_update_rtt(task); | |
876 | xprt_complete_rqst(task, copied); | |
a246b010 CL |
877 | |
878 | out_unlock: | |
4a0f8c04 | 879 | spin_unlock(&xprt->transport_lock); |
a246b010 CL |
880 | dropit: |
881 | skb_free_datagram(sk, skb); | |
882 | out: | |
883 | read_unlock(&sk->sk_callback_lock); | |
884 | } | |
885 | ||
dd456471 | 886 | static inline void xs_tcp_read_fraghdr(struct rpc_xprt *xprt, struct xdr_skb_reader *desc) |
a246b010 | 887 | { |
51971139 | 888 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
a246b010 CL |
889 | size_t len, used; |
890 | char *p; | |
891 | ||
51971139 CL |
892 | p = ((char *) &transport->tcp_fraghdr) + transport->tcp_offset; |
893 | len = sizeof(transport->tcp_fraghdr) - transport->tcp_offset; | |
9d292316 | 894 | used = xdr_skb_read_bits(desc, p, len); |
51971139 | 895 | transport->tcp_offset += used; |
a246b010 CL |
896 | if (used != len) |
897 | return; | |
808012fb | 898 | |
51971139 CL |
899 | transport->tcp_reclen = ntohl(transport->tcp_fraghdr); |
900 | if (transport->tcp_reclen & RPC_LAST_STREAM_FRAGMENT) | |
e136d092 | 901 | transport->tcp_flags |= TCP_RCV_LAST_FRAG; |
a246b010 | 902 | else |
e136d092 | 903 | transport->tcp_flags &= ~TCP_RCV_LAST_FRAG; |
51971139 | 904 | transport->tcp_reclen &= RPC_FRAGMENT_SIZE_MASK; |
808012fb | 905 | |
e136d092 | 906 | transport->tcp_flags &= ~TCP_RCV_COPY_FRAGHDR; |
51971139 | 907 | transport->tcp_offset = 0; |
808012fb | 908 | |
a246b010 | 909 | /* Sanity check of the record length */ |
18dca02a | 910 | if (unlikely(transport->tcp_reclen < 8)) { |
46121cf7 | 911 | dprintk("RPC: invalid TCP record fragment length\n"); |
3ebb067d | 912 | xprt_force_disconnect(xprt); |
9903cd1c | 913 | return; |
a246b010 | 914 | } |
46121cf7 | 915 | dprintk("RPC: reading TCP record fragment of length %d\n", |
51971139 | 916 | transport->tcp_reclen); |
a246b010 CL |
917 | } |
918 | ||
51971139 | 919 | static void xs_tcp_check_fraghdr(struct sock_xprt *transport) |
a246b010 | 920 | { |
51971139 | 921 | if (transport->tcp_offset == transport->tcp_reclen) { |
e136d092 | 922 | transport->tcp_flags |= TCP_RCV_COPY_FRAGHDR; |
51971139 | 923 | transport->tcp_offset = 0; |
e136d092 CL |
924 | if (transport->tcp_flags & TCP_RCV_LAST_FRAG) { |
925 | transport->tcp_flags &= ~TCP_RCV_COPY_DATA; | |
926 | transport->tcp_flags |= TCP_RCV_COPY_XID; | |
51971139 | 927 | transport->tcp_copied = 0; |
a246b010 CL |
928 | } |
929 | } | |
930 | } | |
931 | ||
dd456471 | 932 | static inline void xs_tcp_read_xid(struct sock_xprt *transport, struct xdr_skb_reader *desc) |
a246b010 CL |
933 | { |
934 | size_t len, used; | |
935 | char *p; | |
936 | ||
51971139 | 937 | len = sizeof(transport->tcp_xid) - transport->tcp_offset; |
46121cf7 | 938 | dprintk("RPC: reading XID (%Zu bytes)\n", len); |
51971139 | 939 | p = ((char *) &transport->tcp_xid) + transport->tcp_offset; |
9d292316 | 940 | used = xdr_skb_read_bits(desc, p, len); |
51971139 | 941 | transport->tcp_offset += used; |
a246b010 CL |
942 | if (used != len) |
943 | return; | |
e136d092 | 944 | transport->tcp_flags &= ~TCP_RCV_COPY_XID; |
f4a2e418 | 945 | transport->tcp_flags |= TCP_RCV_READ_CALLDIR; |
51971139 | 946 | transport->tcp_copied = 4; |
18dca02a RL |
947 | dprintk("RPC: reading %s XID %08x\n", |
948 | (transport->tcp_flags & TCP_RPC_REPLY) ? "reply for" | |
949 | : "request with", | |
51971139 CL |
950 | ntohl(transport->tcp_xid)); |
951 | xs_tcp_check_fraghdr(transport); | |
a246b010 CL |
952 | } |
953 | ||
18dca02a RL |
954 | static inline void xs_tcp_read_calldir(struct sock_xprt *transport, |
955 | struct xdr_skb_reader *desc) | |
a246b010 | 956 | { |
18dca02a RL |
957 | size_t len, used; |
958 | u32 offset; | |
959 | __be32 calldir; | |
960 | ||
961 | /* | |
962 | * We want transport->tcp_offset to be 8 at the end of this routine | |
963 | * (4 bytes for the xid and 4 bytes for the call/reply flag). | |
964 | * When this function is called for the first time, | |
965 | * transport->tcp_offset is 4 (after having already read the xid). | |
966 | */ | |
967 | offset = transport->tcp_offset - sizeof(transport->tcp_xid); | |
968 | len = sizeof(calldir) - offset; | |
969 | dprintk("RPC: reading CALL/REPLY flag (%Zu bytes)\n", len); | |
970 | used = xdr_skb_read_bits(desc, &calldir, len); | |
971 | transport->tcp_offset += used; | |
972 | if (used != len) | |
973 | return; | |
f4a2e418 RL |
974 | transport->tcp_flags &= ~TCP_RCV_READ_CALLDIR; |
975 | transport->tcp_flags |= TCP_RCV_COPY_CALLDIR; | |
18dca02a | 976 | transport->tcp_flags |= TCP_RCV_COPY_DATA; |
f4a2e418 RL |
977 | /* |
978 | * We don't yet have the XDR buffer, so we will write the calldir | |
979 | * out after we get the buffer from the 'struct rpc_rqst' | |
980 | */ | |
18dca02a RL |
981 | if (ntohl(calldir) == RPC_REPLY) |
982 | transport->tcp_flags |= TCP_RPC_REPLY; | |
983 | else | |
984 | transport->tcp_flags &= ~TCP_RPC_REPLY; | |
985 | dprintk("RPC: reading %s CALL/REPLY flag %08x\n", | |
986 | (transport->tcp_flags & TCP_RPC_REPLY) ? | |
987 | "reply for" : "request with", calldir); | |
988 | xs_tcp_check_fraghdr(transport); | |
989 | } | |
990 | ||
44b98efd RL |
991 | static inline void xs_tcp_read_common(struct rpc_xprt *xprt, |
992 | struct xdr_skb_reader *desc, | |
993 | struct rpc_rqst *req) | |
a246b010 | 994 | { |
44b98efd RL |
995 | struct sock_xprt *transport = |
996 | container_of(xprt, struct sock_xprt, xprt); | |
a246b010 CL |
997 | struct xdr_buf *rcvbuf; |
998 | size_t len; | |
999 | ssize_t r; | |
1000 | ||
a246b010 | 1001 | rcvbuf = &req->rq_private_buf; |
f4a2e418 RL |
1002 | |
1003 | if (transport->tcp_flags & TCP_RCV_COPY_CALLDIR) { | |
1004 | /* | |
1005 | * Save the RPC direction in the XDR buffer | |
1006 | */ | |
1007 | __be32 calldir = transport->tcp_flags & TCP_RPC_REPLY ? | |
1008 | htonl(RPC_REPLY) : 0; | |
1009 | ||
1010 | memcpy(rcvbuf->head[0].iov_base + transport->tcp_copied, | |
1011 | &calldir, sizeof(calldir)); | |
1012 | transport->tcp_copied += sizeof(calldir); | |
1013 | transport->tcp_flags &= ~TCP_RCV_COPY_CALLDIR; | |
a246b010 CL |
1014 | } |
1015 | ||
a246b010 | 1016 | len = desc->count; |
51971139 | 1017 | if (len > transport->tcp_reclen - transport->tcp_offset) { |
dd456471 | 1018 | struct xdr_skb_reader my_desc; |
a246b010 | 1019 | |
51971139 | 1020 | len = transport->tcp_reclen - transport->tcp_offset; |
a246b010 CL |
1021 | memcpy(&my_desc, desc, sizeof(my_desc)); |
1022 | my_desc.count = len; | |
51971139 | 1023 | r = xdr_partial_copy_from_skb(rcvbuf, transport->tcp_copied, |
9d292316 | 1024 | &my_desc, xdr_skb_read_bits); |
a246b010 CL |
1025 | desc->count -= r; |
1026 | desc->offset += r; | |
1027 | } else | |
51971139 | 1028 | r = xdr_partial_copy_from_skb(rcvbuf, transport->tcp_copied, |
9d292316 | 1029 | desc, xdr_skb_read_bits); |
a246b010 CL |
1030 | |
1031 | if (r > 0) { | |
51971139 CL |
1032 | transport->tcp_copied += r; |
1033 | transport->tcp_offset += r; | |
a246b010 CL |
1034 | } |
1035 | if (r != len) { | |
1036 | /* Error when copying to the receive buffer, | |
1037 | * usually because we weren't able to allocate | |
1038 | * additional buffer pages. All we can do now | |
e136d092 | 1039 | * is turn off TCP_RCV_COPY_DATA, so the request |
a246b010 CL |
1040 | * will not receive any additional updates, |
1041 | * and time out. | |
1042 | * Any remaining data from this record will | |
1043 | * be discarded. | |
1044 | */ | |
e136d092 | 1045 | transport->tcp_flags &= ~TCP_RCV_COPY_DATA; |
46121cf7 | 1046 | dprintk("RPC: XID %08x truncated request\n", |
51971139 | 1047 | ntohl(transport->tcp_xid)); |
46121cf7 CL |
1048 | dprintk("RPC: xprt = %p, tcp_copied = %lu, " |
1049 | "tcp_offset = %u, tcp_reclen = %u\n", | |
1050 | xprt, transport->tcp_copied, | |
1051 | transport->tcp_offset, transport->tcp_reclen); | |
44b98efd | 1052 | return; |
a246b010 CL |
1053 | } |
1054 | ||
46121cf7 | 1055 | dprintk("RPC: XID %08x read %Zd bytes\n", |
51971139 | 1056 | ntohl(transport->tcp_xid), r); |
46121cf7 CL |
1057 | dprintk("RPC: xprt = %p, tcp_copied = %lu, tcp_offset = %u, " |
1058 | "tcp_reclen = %u\n", xprt, transport->tcp_copied, | |
1059 | transport->tcp_offset, transport->tcp_reclen); | |
51971139 CL |
1060 | |
1061 | if (transport->tcp_copied == req->rq_private_buf.buflen) | |
e136d092 | 1062 | transport->tcp_flags &= ~TCP_RCV_COPY_DATA; |
51971139 | 1063 | else if (transport->tcp_offset == transport->tcp_reclen) { |
e136d092 CL |
1064 | if (transport->tcp_flags & TCP_RCV_LAST_FRAG) |
1065 | transport->tcp_flags &= ~TCP_RCV_COPY_DATA; | |
a246b010 CL |
1066 | } |
1067 | ||
44b98efd RL |
1068 | return; |
1069 | } | |
1070 | ||
1071 | /* | |
1072 | * Finds the request corresponding to the RPC xid and invokes the common | |
1073 | * tcp read code to read the data. | |
1074 | */ | |
1075 | static inline int xs_tcp_read_reply(struct rpc_xprt *xprt, | |
1076 | struct xdr_skb_reader *desc) | |
1077 | { | |
1078 | struct sock_xprt *transport = | |
1079 | container_of(xprt, struct sock_xprt, xprt); | |
1080 | struct rpc_rqst *req; | |
1081 | ||
1082 | dprintk("RPC: read reply XID %08x\n", ntohl(transport->tcp_xid)); | |
1083 | ||
1084 | /* Find and lock the request corresponding to this xid */ | |
1085 | spin_lock(&xprt->transport_lock); | |
1086 | req = xprt_lookup_rqst(xprt, transport->tcp_xid); | |
1087 | if (!req) { | |
1088 | dprintk("RPC: XID %08x request not found!\n", | |
1089 | ntohl(transport->tcp_xid)); | |
1090 | spin_unlock(&xprt->transport_lock); | |
1091 | return -1; | |
1092 | } | |
1093 | ||
1094 | xs_tcp_read_common(xprt, desc, req); | |
1095 | ||
e136d092 | 1096 | if (!(transport->tcp_flags & TCP_RCV_COPY_DATA)) |
51971139 | 1097 | xprt_complete_rqst(req->rq_task, transport->tcp_copied); |
44b98efd | 1098 | |
4a0f8c04 | 1099 | spin_unlock(&xprt->transport_lock); |
44b98efd RL |
1100 | return 0; |
1101 | } | |
1102 | ||
1103 | #if defined(CONFIG_NFS_V4_1) | |
1104 | /* | |
1105 | * Obtains an rpc_rqst previously allocated and invokes the common | |
1106 | * tcp read code to read the data. The result is placed in the callback | |
1107 | * queue. | |
1108 | * If we're unable to obtain the rpc_rqst we schedule the closing of the | |
1109 | * connection and return -1. | |
1110 | */ | |
1111 | static inline int xs_tcp_read_callback(struct rpc_xprt *xprt, | |
1112 | struct xdr_skb_reader *desc) | |
1113 | { | |
1114 | struct sock_xprt *transport = | |
1115 | container_of(xprt, struct sock_xprt, xprt); | |
1116 | struct rpc_rqst *req; | |
1117 | ||
1118 | req = xprt_alloc_bc_request(xprt); | |
1119 | if (req == NULL) { | |
1120 | printk(KERN_WARNING "Callback slot table overflowed\n"); | |
1121 | xprt_force_disconnect(xprt); | |
1122 | return -1; | |
1123 | } | |
1124 | ||
1125 | req->rq_xid = transport->tcp_xid; | |
1126 | dprintk("RPC: read callback XID %08x\n", ntohl(req->rq_xid)); | |
1127 | xs_tcp_read_common(xprt, desc, req); | |
1128 | ||
1129 | if (!(transport->tcp_flags & TCP_RCV_COPY_DATA)) { | |
1130 | struct svc_serv *bc_serv = xprt->bc_serv; | |
1131 | ||
1132 | /* | |
1133 | * Add callback request to callback list. The callback | |
1134 | * service sleeps on the sv_cb_waitq waiting for new | |
1135 | * requests. Wake it up after adding enqueing the | |
1136 | * request. | |
1137 | */ | |
1138 | dprintk("RPC: add callback request to list\n"); | |
1139 | spin_lock(&bc_serv->sv_cb_lock); | |
1140 | list_add(&req->rq_bc_list, &bc_serv->sv_cb_list); | |
1141 | spin_unlock(&bc_serv->sv_cb_lock); | |
1142 | wake_up(&bc_serv->sv_cb_waitq); | |
1143 | } | |
1144 | ||
1145 | req->rq_private_buf.len = transport->tcp_copied; | |
1146 | ||
1147 | return 0; | |
1148 | } | |
1149 | ||
1150 | static inline int _xs_tcp_read_data(struct rpc_xprt *xprt, | |
1151 | struct xdr_skb_reader *desc) | |
1152 | { | |
1153 | struct sock_xprt *transport = | |
1154 | container_of(xprt, struct sock_xprt, xprt); | |
1155 | ||
1156 | return (transport->tcp_flags & TCP_RPC_REPLY) ? | |
1157 | xs_tcp_read_reply(xprt, desc) : | |
1158 | xs_tcp_read_callback(xprt, desc); | |
1159 | } | |
1160 | #else | |
1161 | static inline int _xs_tcp_read_data(struct rpc_xprt *xprt, | |
1162 | struct xdr_skb_reader *desc) | |
1163 | { | |
1164 | return xs_tcp_read_reply(xprt, desc); | |
1165 | } | |
1166 | #endif /* CONFIG_NFS_V4_1 */ | |
1167 | ||
1168 | /* | |
1169 | * Read data off the transport. This can be either an RPC_CALL or an | |
1170 | * RPC_REPLY. Relay the processing to helper functions. | |
1171 | */ | |
1172 | static void xs_tcp_read_data(struct rpc_xprt *xprt, | |
1173 | struct xdr_skb_reader *desc) | |
1174 | { | |
1175 | struct sock_xprt *transport = | |
1176 | container_of(xprt, struct sock_xprt, xprt); | |
1177 | ||
1178 | if (_xs_tcp_read_data(xprt, desc) == 0) | |
1179 | xs_tcp_check_fraghdr(transport); | |
1180 | else { | |
1181 | /* | |
1182 | * The transport_lock protects the request handling. | |
1183 | * There's no need to hold it to update the tcp_flags. | |
1184 | */ | |
1185 | transport->tcp_flags &= ~TCP_RCV_COPY_DATA; | |
1186 | } | |
a246b010 CL |
1187 | } |
1188 | ||
dd456471 | 1189 | static inline void xs_tcp_read_discard(struct sock_xprt *transport, struct xdr_skb_reader *desc) |
a246b010 CL |
1190 | { |
1191 | size_t len; | |
1192 | ||
51971139 | 1193 | len = transport->tcp_reclen - transport->tcp_offset; |
a246b010 CL |
1194 | if (len > desc->count) |
1195 | len = desc->count; | |
1196 | desc->count -= len; | |
1197 | desc->offset += len; | |
51971139 | 1198 | transport->tcp_offset += len; |
46121cf7 | 1199 | dprintk("RPC: discarded %Zu bytes\n", len); |
51971139 | 1200 | xs_tcp_check_fraghdr(transport); |
a246b010 CL |
1201 | } |
1202 | ||
9903cd1c | 1203 | static int xs_tcp_data_recv(read_descriptor_t *rd_desc, struct sk_buff *skb, unsigned int offset, size_t len) |
a246b010 CL |
1204 | { |
1205 | struct rpc_xprt *xprt = rd_desc->arg.data; | |
51971139 | 1206 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
dd456471 | 1207 | struct xdr_skb_reader desc = { |
a246b010 CL |
1208 | .skb = skb, |
1209 | .offset = offset, | |
1210 | .count = len, | |
9903cd1c | 1211 | }; |
a246b010 | 1212 | |
46121cf7 | 1213 | dprintk("RPC: xs_tcp_data_recv started\n"); |
a246b010 CL |
1214 | do { |
1215 | /* Read in a new fragment marker if necessary */ | |
1216 | /* Can we ever really expect to get completely empty fragments? */ | |
e136d092 | 1217 | if (transport->tcp_flags & TCP_RCV_COPY_FRAGHDR) { |
9903cd1c | 1218 | xs_tcp_read_fraghdr(xprt, &desc); |
a246b010 CL |
1219 | continue; |
1220 | } | |
1221 | /* Read in the xid if necessary */ | |
e136d092 | 1222 | if (transport->tcp_flags & TCP_RCV_COPY_XID) { |
51971139 | 1223 | xs_tcp_read_xid(transport, &desc); |
a246b010 CL |
1224 | continue; |
1225 | } | |
18dca02a | 1226 | /* Read in the call/reply flag */ |
f4a2e418 | 1227 | if (transport->tcp_flags & TCP_RCV_READ_CALLDIR) { |
18dca02a RL |
1228 | xs_tcp_read_calldir(transport, &desc); |
1229 | continue; | |
1230 | } | |
a246b010 | 1231 | /* Read in the request data */ |
e136d092 | 1232 | if (transport->tcp_flags & TCP_RCV_COPY_DATA) { |
44b98efd | 1233 | xs_tcp_read_data(xprt, &desc); |
a246b010 CL |
1234 | continue; |
1235 | } | |
1236 | /* Skip over any trailing bytes on short reads */ | |
51971139 | 1237 | xs_tcp_read_discard(transport, &desc); |
a246b010 | 1238 | } while (desc.count); |
46121cf7 | 1239 | dprintk("RPC: xs_tcp_data_recv done\n"); |
a246b010 CL |
1240 | return len - desc.count; |
1241 | } | |
1242 | ||
9903cd1c CL |
1243 | /** |
1244 | * xs_tcp_data_ready - "data ready" callback for TCP sockets | |
1245 | * @sk: socket with data to read | |
1246 | * @bytes: how much data to read | |
1247 | * | |
1248 | */ | |
1249 | static void xs_tcp_data_ready(struct sock *sk, int bytes) | |
a246b010 CL |
1250 | { |
1251 | struct rpc_xprt *xprt; | |
1252 | read_descriptor_t rd_desc; | |
ff2d7db8 | 1253 | int read; |
a246b010 | 1254 | |
46121cf7 CL |
1255 | dprintk("RPC: xs_tcp_data_ready...\n"); |
1256 | ||
a246b010 | 1257 | read_lock(&sk->sk_callback_lock); |
9903cd1c | 1258 | if (!(xprt = xprt_from_sock(sk))) |
a246b010 | 1259 | goto out; |
a246b010 CL |
1260 | if (xprt->shutdown) |
1261 | goto out; | |
1262 | ||
9903cd1c | 1263 | /* We use rd_desc to pass struct xprt to xs_tcp_data_recv */ |
a246b010 | 1264 | rd_desc.arg.data = xprt; |
ff2d7db8 TM |
1265 | do { |
1266 | rd_desc.count = 65536; | |
1267 | read = tcp_read_sock(sk, &rd_desc, xs_tcp_data_recv); | |
1268 | } while (read > 0); | |
a246b010 CL |
1269 | out: |
1270 | read_unlock(&sk->sk_callback_lock); | |
1271 | } | |
1272 | ||
7d1e8255 TM |
1273 | /* |
1274 | * Do the equivalent of linger/linger2 handling for dealing with | |
1275 | * broken servers that don't close the socket in a timely | |
1276 | * fashion | |
1277 | */ | |
1278 | static void xs_tcp_schedule_linger_timeout(struct rpc_xprt *xprt, | |
1279 | unsigned long timeout) | |
1280 | { | |
1281 | struct sock_xprt *transport; | |
1282 | ||
1283 | if (xprt_test_and_set_connecting(xprt)) | |
1284 | return; | |
1285 | set_bit(XPRT_CONNECTION_ABORT, &xprt->state); | |
1286 | transport = container_of(xprt, struct sock_xprt, xprt); | |
1287 | queue_delayed_work(rpciod_workqueue, &transport->connect_worker, | |
1288 | timeout); | |
1289 | } | |
1290 | ||
1291 | static void xs_tcp_cancel_linger_timeout(struct rpc_xprt *xprt) | |
1292 | { | |
1293 | struct sock_xprt *transport; | |
1294 | ||
1295 | transport = container_of(xprt, struct sock_xprt, xprt); | |
1296 | ||
1297 | if (!test_bit(XPRT_CONNECTION_ABORT, &xprt->state) || | |
1298 | !cancel_delayed_work(&transport->connect_worker)) | |
1299 | return; | |
1300 | clear_bit(XPRT_CONNECTION_ABORT, &xprt->state); | |
1301 | xprt_clear_connecting(xprt); | |
1302 | } | |
1303 | ||
1304 | static void xs_sock_mark_closed(struct rpc_xprt *xprt) | |
1305 | { | |
1306 | smp_mb__before_clear_bit(); | |
1307 | clear_bit(XPRT_CLOSE_WAIT, &xprt->state); | |
1308 | clear_bit(XPRT_CLOSING, &xprt->state); | |
1309 | smp_mb__after_clear_bit(); | |
1310 | /* Mark transport as closed and wake up all pending tasks */ | |
1311 | xprt_disconnect_done(xprt); | |
1312 | } | |
1313 | ||
9903cd1c CL |
1314 | /** |
1315 | * xs_tcp_state_change - callback to handle TCP socket state changes | |
1316 | * @sk: socket whose state has changed | |
1317 | * | |
1318 | */ | |
1319 | static void xs_tcp_state_change(struct sock *sk) | |
a246b010 | 1320 | { |
9903cd1c | 1321 | struct rpc_xprt *xprt; |
a246b010 CL |
1322 | |
1323 | read_lock(&sk->sk_callback_lock); | |
1324 | if (!(xprt = xprt_from_sock(sk))) | |
1325 | goto out; | |
46121cf7 CL |
1326 | dprintk("RPC: xs_tcp_state_change client %p...\n", xprt); |
1327 | dprintk("RPC: state %x conn %d dead %d zapped %d\n", | |
1328 | sk->sk_state, xprt_connected(xprt), | |
1329 | sock_flag(sk, SOCK_DEAD), | |
1330 | sock_flag(sk, SOCK_ZAPPED)); | |
a246b010 CL |
1331 | |
1332 | switch (sk->sk_state) { | |
1333 | case TCP_ESTABLISHED: | |
4a0f8c04 | 1334 | spin_lock_bh(&xprt->transport_lock); |
a246b010 | 1335 | if (!xprt_test_and_set_connected(xprt)) { |
51971139 CL |
1336 | struct sock_xprt *transport = container_of(xprt, |
1337 | struct sock_xprt, xprt); | |
1338 | ||
a246b010 | 1339 | /* Reset TCP record info */ |
51971139 CL |
1340 | transport->tcp_offset = 0; |
1341 | transport->tcp_reclen = 0; | |
1342 | transport->tcp_copied = 0; | |
e136d092 CL |
1343 | transport->tcp_flags = |
1344 | TCP_RCV_COPY_FRAGHDR | TCP_RCV_COPY_XID; | |
51971139 | 1345 | |
2a491991 | 1346 | xprt_wake_pending_tasks(xprt, -EAGAIN); |
a246b010 | 1347 | } |
4a0f8c04 | 1348 | spin_unlock_bh(&xprt->transport_lock); |
a246b010 | 1349 | break; |
3b948ae5 TM |
1350 | case TCP_FIN_WAIT1: |
1351 | /* The client initiated a shutdown of the socket */ | |
7c1d71cf | 1352 | xprt->connect_cookie++; |
663b8858 | 1353 | xprt->reestablish_timeout = 0; |
3b948ae5 TM |
1354 | set_bit(XPRT_CLOSING, &xprt->state); |
1355 | smp_mb__before_clear_bit(); | |
1356 | clear_bit(XPRT_CONNECTED, &xprt->state); | |
ef803670 | 1357 | clear_bit(XPRT_CLOSE_WAIT, &xprt->state); |
3b948ae5 | 1358 | smp_mb__after_clear_bit(); |
25fe6142 | 1359 | xs_tcp_schedule_linger_timeout(xprt, xs_tcp_fin_timeout); |
a246b010 | 1360 | break; |
632e3bdc | 1361 | case TCP_CLOSE_WAIT: |
3b948ae5 | 1362 | /* The server initiated a shutdown of the socket */ |
66af1e55 | 1363 | xprt_force_disconnect(xprt); |
663b8858 | 1364 | case TCP_SYN_SENT: |
7c1d71cf | 1365 | xprt->connect_cookie++; |
663b8858 TM |
1366 | case TCP_CLOSING: |
1367 | /* | |
1368 | * If the server closed down the connection, make sure that | |
1369 | * we back off before reconnecting | |
1370 | */ | |
1371 | if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO) | |
1372 | xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO; | |
3b948ae5 TM |
1373 | break; |
1374 | case TCP_LAST_ACK: | |
670f9457 | 1375 | set_bit(XPRT_CLOSING, &xprt->state); |
25fe6142 | 1376 | xs_tcp_schedule_linger_timeout(xprt, xs_tcp_fin_timeout); |
3b948ae5 TM |
1377 | smp_mb__before_clear_bit(); |
1378 | clear_bit(XPRT_CONNECTED, &xprt->state); | |
1379 | smp_mb__after_clear_bit(); | |
1380 | break; | |
1381 | case TCP_CLOSE: | |
7d1e8255 TM |
1382 | xs_tcp_cancel_linger_timeout(xprt); |
1383 | xs_sock_mark_closed(xprt); | |
a246b010 CL |
1384 | } |
1385 | out: | |
1386 | read_unlock(&sk->sk_callback_lock); | |
1387 | } | |
1388 | ||
2a9e1cfa | 1389 | /** |
482f32e6 | 1390 | * xs_error_report - callback mainly for catching socket errors |
2a9e1cfa TM |
1391 | * @sk: socket |
1392 | */ | |
482f32e6 | 1393 | static void xs_error_report(struct sock *sk) |
2a9e1cfa TM |
1394 | { |
1395 | struct rpc_xprt *xprt; | |
1396 | ||
1397 | read_lock(&sk->sk_callback_lock); | |
2a9e1cfa TM |
1398 | if (!(xprt = xprt_from_sock(sk))) |
1399 | goto out; | |
1400 | dprintk("RPC: %s client %p...\n" | |
1401 | "RPC: error %d\n", | |
1402 | __func__, xprt, sk->sk_err); | |
482f32e6 | 1403 | xprt_wake_pending_tasks(xprt, -EAGAIN); |
2a9e1cfa TM |
1404 | out: |
1405 | read_unlock(&sk->sk_callback_lock); | |
1406 | } | |
1407 | ||
1f0fa154 IJ |
1408 | static void xs_write_space(struct sock *sk) |
1409 | { | |
1410 | struct socket *sock; | |
1411 | struct rpc_xprt *xprt; | |
1412 | ||
1413 | if (unlikely(!(sock = sk->sk_socket))) | |
1414 | return; | |
1415 | clear_bit(SOCK_NOSPACE, &sock->flags); | |
1416 | ||
1417 | if (unlikely(!(xprt = xprt_from_sock(sk)))) | |
1418 | return; | |
1419 | if (test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags) == 0) | |
1420 | return; | |
1421 | ||
1422 | xprt_write_space(xprt); | |
1423 | } | |
1424 | ||
9903cd1c | 1425 | /** |
c7b2cae8 CL |
1426 | * xs_udp_write_space - callback invoked when socket buffer space |
1427 | * becomes available | |
9903cd1c CL |
1428 | * @sk: socket whose state has changed |
1429 | * | |
a246b010 CL |
1430 | * Called when more output buffer space is available for this socket. |
1431 | * We try not to wake our writers until they can make "significant" | |
c7b2cae8 | 1432 | * progress, otherwise we'll waste resources thrashing kernel_sendmsg |
a246b010 CL |
1433 | * with a bunch of small requests. |
1434 | */ | |
c7b2cae8 | 1435 | static void xs_udp_write_space(struct sock *sk) |
a246b010 | 1436 | { |
a246b010 | 1437 | read_lock(&sk->sk_callback_lock); |
a246b010 | 1438 | |
c7b2cae8 | 1439 | /* from net/core/sock.c:sock_def_write_space */ |
1f0fa154 IJ |
1440 | if (sock_writeable(sk)) |
1441 | xs_write_space(sk); | |
a246b010 | 1442 | |
c7b2cae8 CL |
1443 | read_unlock(&sk->sk_callback_lock); |
1444 | } | |
a246b010 | 1445 | |
c7b2cae8 CL |
1446 | /** |
1447 | * xs_tcp_write_space - callback invoked when socket buffer space | |
1448 | * becomes available | |
1449 | * @sk: socket whose state has changed | |
1450 | * | |
1451 | * Called when more output buffer space is available for this socket. | |
1452 | * We try not to wake our writers until they can make "significant" | |
1453 | * progress, otherwise we'll waste resources thrashing kernel_sendmsg | |
1454 | * with a bunch of small requests. | |
1455 | */ | |
1456 | static void xs_tcp_write_space(struct sock *sk) | |
1457 | { | |
1458 | read_lock(&sk->sk_callback_lock); | |
1459 | ||
1460 | /* from net/core/stream.c:sk_stream_write_space */ | |
1f0fa154 IJ |
1461 | if (sk_stream_wspace(sk) >= sk_stream_min_wspace(sk)) |
1462 | xs_write_space(sk); | |
b6ddf64f | 1463 | |
a246b010 CL |
1464 | read_unlock(&sk->sk_callback_lock); |
1465 | } | |
1466 | ||
470056c2 | 1467 | static void xs_udp_do_set_buffer_size(struct rpc_xprt *xprt) |
a246b010 | 1468 | { |
ee0ac0c2 CL |
1469 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
1470 | struct sock *sk = transport->inet; | |
a246b010 | 1471 | |
7c6e066e | 1472 | if (transport->rcvsize) { |
a246b010 | 1473 | sk->sk_userlocks |= SOCK_RCVBUF_LOCK; |
7c6e066e | 1474 | sk->sk_rcvbuf = transport->rcvsize * xprt->max_reqs * 2; |
a246b010 | 1475 | } |
7c6e066e | 1476 | if (transport->sndsize) { |
a246b010 | 1477 | sk->sk_userlocks |= SOCK_SNDBUF_LOCK; |
7c6e066e | 1478 | sk->sk_sndbuf = transport->sndsize * xprt->max_reqs * 2; |
a246b010 CL |
1479 | sk->sk_write_space(sk); |
1480 | } | |
1481 | } | |
1482 | ||
43118c29 | 1483 | /** |
470056c2 | 1484 | * xs_udp_set_buffer_size - set send and receive limits |
43118c29 | 1485 | * @xprt: generic transport |
470056c2 CL |
1486 | * @sndsize: requested size of send buffer, in bytes |
1487 | * @rcvsize: requested size of receive buffer, in bytes | |
43118c29 | 1488 | * |
470056c2 | 1489 | * Set socket send and receive buffer size limits. |
43118c29 | 1490 | */ |
470056c2 | 1491 | static void xs_udp_set_buffer_size(struct rpc_xprt *xprt, size_t sndsize, size_t rcvsize) |
43118c29 | 1492 | { |
7c6e066e CL |
1493 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
1494 | ||
1495 | transport->sndsize = 0; | |
470056c2 | 1496 | if (sndsize) |
7c6e066e CL |
1497 | transport->sndsize = sndsize + 1024; |
1498 | transport->rcvsize = 0; | |
470056c2 | 1499 | if (rcvsize) |
7c6e066e | 1500 | transport->rcvsize = rcvsize + 1024; |
470056c2 CL |
1501 | |
1502 | xs_udp_do_set_buffer_size(xprt); | |
43118c29 CL |
1503 | } |
1504 | ||
46c0ee8b CL |
1505 | /** |
1506 | * xs_udp_timer - called when a retransmit timeout occurs on a UDP transport | |
1507 | * @task: task that timed out | |
1508 | * | |
1509 | * Adjust the congestion window after a retransmit timeout has occurred. | |
1510 | */ | |
1511 | static void xs_udp_timer(struct rpc_task *task) | |
1512 | { | |
1513 | xprt_adjust_cwnd(task, -ETIMEDOUT); | |
1514 | } | |
1515 | ||
b85d8806 CL |
1516 | static unsigned short xs_get_random_port(void) |
1517 | { | |
1518 | unsigned short range = xprt_max_resvport - xprt_min_resvport; | |
1519 | unsigned short rand = (unsigned short) net_random() % range; | |
1520 | return rand + xprt_min_resvport; | |
1521 | } | |
1522 | ||
92200412 CL |
1523 | /** |
1524 | * xs_set_port - reset the port number in the remote endpoint address | |
1525 | * @xprt: generic transport | |
1526 | * @port: new port number | |
1527 | * | |
1528 | */ | |
1529 | static void xs_set_port(struct rpc_xprt *xprt, unsigned short port) | |
1530 | { | |
95392c59 | 1531 | struct sockaddr *addr = xs_addr(xprt); |
c4efcb1d | 1532 | |
46121cf7 | 1533 | dprintk("RPC: setting port for xprt %p to %u\n", xprt, port); |
c4efcb1d | 1534 | |
20612005 CL |
1535 | switch (addr->sa_family) { |
1536 | case AF_INET: | |
1537 | ((struct sockaddr_in *)addr)->sin_port = htons(port); | |
1538 | break; | |
1539 | case AF_INET6: | |
1540 | ((struct sockaddr_in6 *)addr)->sin6_port = htons(port); | |
1541 | break; | |
1542 | default: | |
1543 | BUG(); | |
1544 | } | |
92200412 CL |
1545 | } |
1546 | ||
67a391d7 TM |
1547 | static unsigned short xs_get_srcport(struct sock_xprt *transport, struct socket *sock) |
1548 | { | |
1549 | unsigned short port = transport->port; | |
1550 | ||
1551 | if (port == 0 && transport->xprt.resvport) | |
1552 | port = xs_get_random_port(); | |
1553 | return port; | |
1554 | } | |
1555 | ||
1556 | static unsigned short xs_next_srcport(struct sock_xprt *transport, struct socket *sock, unsigned short port) | |
1557 | { | |
1558 | if (transport->port != 0) | |
1559 | transport->port = 0; | |
1560 | if (!transport->xprt.resvport) | |
1561 | return 0; | |
1562 | if (port <= xprt_min_resvport || port > xprt_max_resvport) | |
1563 | return xprt_max_resvport; | |
1564 | return --port; | |
1565 | } | |
1566 | ||
7dc753f0 | 1567 | static int xs_bind4(struct sock_xprt *transport, struct socket *sock) |
a246b010 CL |
1568 | { |
1569 | struct sockaddr_in myaddr = { | |
1570 | .sin_family = AF_INET, | |
1571 | }; | |
d3bc9a1d | 1572 | struct sockaddr_in *sa; |
67a391d7 TM |
1573 | int err, nloop = 0; |
1574 | unsigned short port = xs_get_srcport(transport, sock); | |
1575 | unsigned short last; | |
a246b010 | 1576 | |
d3bc9a1d FM |
1577 | sa = (struct sockaddr_in *)&transport->addr; |
1578 | myaddr.sin_addr = sa->sin_addr; | |
a246b010 CL |
1579 | do { |
1580 | myaddr.sin_port = htons(port); | |
e6242e92 | 1581 | err = kernel_bind(sock, (struct sockaddr *) &myaddr, |
a246b010 | 1582 | sizeof(myaddr)); |
67a391d7 | 1583 | if (port == 0) |
d3bc9a1d | 1584 | break; |
a246b010 | 1585 | if (err == 0) { |
c8475461 | 1586 | transport->port = port; |
d3bc9a1d | 1587 | break; |
a246b010 | 1588 | } |
67a391d7 TM |
1589 | last = port; |
1590 | port = xs_next_srcport(transport, sock, port); | |
1591 | if (port > last) | |
1592 | nloop++; | |
1593 | } while (err == -EADDRINUSE && nloop != 2); | |
21454aaa HH |
1594 | dprintk("RPC: %s %pI4:%u: %s (%d)\n", |
1595 | __func__, &myaddr.sin_addr, | |
7dc753f0 | 1596 | port, err ? "failed" : "ok", err); |
a246b010 CL |
1597 | return err; |
1598 | } | |
1599 | ||
90058d37 CL |
1600 | static int xs_bind6(struct sock_xprt *transport, struct socket *sock) |
1601 | { | |
1602 | struct sockaddr_in6 myaddr = { | |
1603 | .sin6_family = AF_INET6, | |
1604 | }; | |
1605 | struct sockaddr_in6 *sa; | |
67a391d7 TM |
1606 | int err, nloop = 0; |
1607 | unsigned short port = xs_get_srcport(transport, sock); | |
1608 | unsigned short last; | |
90058d37 | 1609 | |
90058d37 CL |
1610 | sa = (struct sockaddr_in6 *)&transport->addr; |
1611 | myaddr.sin6_addr = sa->sin6_addr; | |
1612 | do { | |
1613 | myaddr.sin6_port = htons(port); | |
1614 | err = kernel_bind(sock, (struct sockaddr *) &myaddr, | |
1615 | sizeof(myaddr)); | |
67a391d7 | 1616 | if (port == 0) |
90058d37 CL |
1617 | break; |
1618 | if (err == 0) { | |
1619 | transport->port = port; | |
1620 | break; | |
1621 | } | |
67a391d7 TM |
1622 | last = port; |
1623 | port = xs_next_srcport(transport, sock, port); | |
1624 | if (port > last) | |
1625 | nloop++; | |
1626 | } while (err == -EADDRINUSE && nloop != 2); | |
5b095d98 | 1627 | dprintk("RPC: xs_bind6 %pI6:%u: %s (%d)\n", |
fdb46ee7 | 1628 | &myaddr.sin6_addr, port, err ? "failed" : "ok", err); |
a246b010 CL |
1629 | return err; |
1630 | } | |
1631 | ||
ed07536e PZ |
1632 | #ifdef CONFIG_DEBUG_LOCK_ALLOC |
1633 | static struct lock_class_key xs_key[2]; | |
1634 | static struct lock_class_key xs_slock_key[2]; | |
1635 | ||
8945ee5e | 1636 | static inline void xs_reclassify_socket4(struct socket *sock) |
ed07536e PZ |
1637 | { |
1638 | struct sock *sk = sock->sk; | |
8945ee5e | 1639 | |
02b3d346 | 1640 | BUG_ON(sock_owned_by_user(sk)); |
8945ee5e CL |
1641 | sock_lock_init_class_and_name(sk, "slock-AF_INET-RPC", |
1642 | &xs_slock_key[0], "sk_lock-AF_INET-RPC", &xs_key[0]); | |
1643 | } | |
ed07536e | 1644 | |
8945ee5e CL |
1645 | static inline void xs_reclassify_socket6(struct socket *sock) |
1646 | { | |
1647 | struct sock *sk = sock->sk; | |
ed07536e | 1648 | |
f4921aff | 1649 | BUG_ON(sock_owned_by_user(sk)); |
8945ee5e CL |
1650 | sock_lock_init_class_and_name(sk, "slock-AF_INET6-RPC", |
1651 | &xs_slock_key[1], "sk_lock-AF_INET6-RPC", &xs_key[1]); | |
ed07536e PZ |
1652 | } |
1653 | #else | |
8945ee5e CL |
1654 | static inline void xs_reclassify_socket4(struct socket *sock) |
1655 | { | |
1656 | } | |
1657 | ||
1658 | static inline void xs_reclassify_socket6(struct socket *sock) | |
ed07536e PZ |
1659 | { |
1660 | } | |
1661 | #endif | |
1662 | ||
16be2d20 CL |
1663 | static void xs_udp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock) |
1664 | { | |
1665 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); | |
1666 | ||
1667 | if (!transport->inet) { | |
1668 | struct sock *sk = sock->sk; | |
1669 | ||
1670 | write_lock_bh(&sk->sk_callback_lock); | |
1671 | ||
2a9e1cfa TM |
1672 | xs_save_old_callbacks(transport, sk); |
1673 | ||
16be2d20 | 1674 | sk->sk_user_data = xprt; |
16be2d20 CL |
1675 | sk->sk_data_ready = xs_udp_data_ready; |
1676 | sk->sk_write_space = xs_udp_write_space; | |
482f32e6 | 1677 | sk->sk_error_report = xs_error_report; |
16be2d20 CL |
1678 | sk->sk_no_check = UDP_CSUM_NORCV; |
1679 | sk->sk_allocation = GFP_ATOMIC; | |
1680 | ||
1681 | xprt_set_connected(xprt); | |
1682 | ||
1683 | /* Reset to new socket */ | |
1684 | transport->sock = sock; | |
1685 | transport->inet = sk; | |
1686 | ||
1687 | write_unlock_bh(&sk->sk_callback_lock); | |
1688 | } | |
1689 | xs_udp_do_set_buffer_size(xprt); | |
1690 | } | |
1691 | ||
b0d93ad5 | 1692 | /** |
9c3d72de | 1693 | * xs_udp_connect_worker4 - set up a UDP socket |
65f27f38 | 1694 | * @work: RPC transport to connect |
b0d93ad5 CL |
1695 | * |
1696 | * Invoked by a work queue tasklet. | |
1697 | */ | |
9c3d72de | 1698 | static void xs_udp_connect_worker4(struct work_struct *work) |
a246b010 | 1699 | { |
34161db6 TM |
1700 | struct sock_xprt *transport = |
1701 | container_of(work, struct sock_xprt, connect_worker.work); | |
c8475461 | 1702 | struct rpc_xprt *xprt = &transport->xprt; |
ee0ac0c2 | 1703 | struct socket *sock = transport->sock; |
b0d93ad5 | 1704 | int err, status = -EIO; |
9903cd1c | 1705 | |
01d37c42 | 1706 | if (xprt->shutdown) |
b0d93ad5 | 1707 | goto out; |
9903cd1c | 1708 | |
b0d93ad5 | 1709 | /* Start by resetting any existing state */ |
fe315e76 | 1710 | xs_reset_transport(transport); |
9903cd1c | 1711 | |
fe315e76 CL |
1712 | err = sock_create_kern(PF_INET, SOCK_DGRAM, IPPROTO_UDP, &sock); |
1713 | if (err < 0) { | |
46121cf7 | 1714 | dprintk("RPC: can't create UDP transport socket (%d).\n", -err); |
b0d93ad5 CL |
1715 | goto out; |
1716 | } | |
8945ee5e | 1717 | xs_reclassify_socket4(sock); |
9903cd1c | 1718 | |
7dc753f0 | 1719 | if (xs_bind4(transport, sock)) { |
b0d93ad5 CL |
1720 | sock_release(sock); |
1721 | goto out; | |
1722 | } | |
9903cd1c | 1723 | |
46121cf7 | 1724 | dprintk("RPC: worker connecting xprt %p to address: %s\n", |
7559c7a2 | 1725 | xprt, xprt->address_strings[RPC_DISPLAY_ALL]); |
edb267a6 | 1726 | |
16be2d20 | 1727 | xs_udp_finish_connecting(xprt, sock); |
b0d93ad5 CL |
1728 | status = 0; |
1729 | out: | |
b0d93ad5 | 1730 | xprt_clear_connecting(xprt); |
7d1e8255 | 1731 | xprt_wake_pending_tasks(xprt, status); |
a246b010 | 1732 | } |
a246b010 | 1733 | |
68e220bd CL |
1734 | /** |
1735 | * xs_udp_connect_worker6 - set up a UDP socket | |
1736 | * @work: RPC transport to connect | |
1737 | * | |
1738 | * Invoked by a work queue tasklet. | |
1739 | */ | |
1740 | static void xs_udp_connect_worker6(struct work_struct *work) | |
1741 | { | |
1742 | struct sock_xprt *transport = | |
1743 | container_of(work, struct sock_xprt, connect_worker.work); | |
1744 | struct rpc_xprt *xprt = &transport->xprt; | |
1745 | struct socket *sock = transport->sock; | |
1746 | int err, status = -EIO; | |
a246b010 | 1747 | |
01d37c42 | 1748 | if (xprt->shutdown) |
68e220bd | 1749 | goto out; |
b0d93ad5 | 1750 | |
68e220bd | 1751 | /* Start by resetting any existing state */ |
fe315e76 | 1752 | xs_reset_transport(transport); |
a246b010 | 1753 | |
fe315e76 CL |
1754 | err = sock_create_kern(PF_INET6, SOCK_DGRAM, IPPROTO_UDP, &sock); |
1755 | if (err < 0) { | |
68e220bd CL |
1756 | dprintk("RPC: can't create UDP transport socket (%d).\n", -err); |
1757 | goto out; | |
1758 | } | |
8945ee5e | 1759 | xs_reclassify_socket6(sock); |
a246b010 | 1760 | |
68e220bd CL |
1761 | if (xs_bind6(transport, sock) < 0) { |
1762 | sock_release(sock); | |
1763 | goto out; | |
b0d93ad5 | 1764 | } |
68e220bd CL |
1765 | |
1766 | dprintk("RPC: worker connecting xprt %p to address: %s\n", | |
1767 | xprt, xprt->address_strings[RPC_DISPLAY_ALL]); | |
1768 | ||
1769 | xs_udp_finish_connecting(xprt, sock); | |
b0d93ad5 CL |
1770 | status = 0; |
1771 | out: | |
b0d93ad5 | 1772 | xprt_clear_connecting(xprt); |
7d1e8255 | 1773 | xprt_wake_pending_tasks(xprt, status); |
a246b010 CL |
1774 | } |
1775 | ||
3167e12c CL |
1776 | /* |
1777 | * We need to preserve the port number so the reply cache on the server can | |
1778 | * find our cached RPC replies when we get around to reconnecting. | |
1779 | */ | |
40d2549d | 1780 | static void xs_abort_connection(struct rpc_xprt *xprt, struct sock_xprt *transport) |
3167e12c CL |
1781 | { |
1782 | int result; | |
3167e12c CL |
1783 | struct sockaddr any; |
1784 | ||
46121cf7 | 1785 | dprintk("RPC: disconnecting xprt %p to reuse port\n", xprt); |
3167e12c CL |
1786 | |
1787 | /* | |
1788 | * Disconnect the transport socket by doing a connect operation | |
1789 | * with AF_UNSPEC. This should return immediately... | |
1790 | */ | |
1791 | memset(&any, 0, sizeof(any)); | |
1792 | any.sa_family = AF_UNSPEC; | |
ee0ac0c2 | 1793 | result = kernel_connect(transport->sock, &any, sizeof(any), 0); |
7d1e8255 TM |
1794 | if (!result) |
1795 | xs_sock_mark_closed(xprt); | |
1796 | else | |
46121cf7 | 1797 | dprintk("RPC: AF_UNSPEC connect return code %d\n", |
3167e12c CL |
1798 | result); |
1799 | } | |
1800 | ||
40d2549d TM |
1801 | static void xs_tcp_reuse_connection(struct rpc_xprt *xprt, struct sock_xprt *transport) |
1802 | { | |
1803 | unsigned int state = transport->inet->sk_state; | |
1804 | ||
1805 | if (state == TCP_CLOSE && transport->sock->state == SS_UNCONNECTED) | |
1806 | return; | |
1807 | if ((1 << state) & (TCPF_ESTABLISHED|TCPF_SYN_SENT)) | |
1808 | return; | |
1809 | xs_abort_connection(xprt, transport); | |
1810 | } | |
1811 | ||
16be2d20 | 1812 | static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock) |
a246b010 | 1813 | { |
16be2d20 | 1814 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
edb267a6 | 1815 | |
ee0ac0c2 | 1816 | if (!transport->inet) { |
b0d93ad5 CL |
1817 | struct sock *sk = sock->sk; |
1818 | ||
1819 | write_lock_bh(&sk->sk_callback_lock); | |
1820 | ||
2a9e1cfa TM |
1821 | xs_save_old_callbacks(transport, sk); |
1822 | ||
b0d93ad5 | 1823 | sk->sk_user_data = xprt; |
b0d93ad5 CL |
1824 | sk->sk_data_ready = xs_tcp_data_ready; |
1825 | sk->sk_state_change = xs_tcp_state_change; | |
1826 | sk->sk_write_space = xs_tcp_write_space; | |
482f32e6 | 1827 | sk->sk_error_report = xs_error_report; |
b079fa7b | 1828 | sk->sk_allocation = GFP_ATOMIC; |
3167e12c CL |
1829 | |
1830 | /* socket options */ | |
1831 | sk->sk_userlocks |= SOCK_BINDPORT_LOCK; | |
1832 | sock_reset_flag(sk, SOCK_LINGER); | |
1833 | tcp_sk(sk)->linger2 = 0; | |
1834 | tcp_sk(sk)->nonagle |= TCP_NAGLE_OFF; | |
b0d93ad5 CL |
1835 | |
1836 | xprt_clear_connected(xprt); | |
1837 | ||
1838 | /* Reset to new socket */ | |
ee0ac0c2 CL |
1839 | transport->sock = sock; |
1840 | transport->inet = sk; | |
b0d93ad5 CL |
1841 | |
1842 | write_unlock_bh(&sk->sk_callback_lock); | |
1843 | } | |
1844 | ||
01d37c42 TM |
1845 | if (!xprt_bound(xprt)) |
1846 | return -ENOTCONN; | |
1847 | ||
b0d93ad5 | 1848 | /* Tell the socket layer to start connecting... */ |
262ca07d CL |
1849 | xprt->stat.connect_count++; |
1850 | xprt->stat.connect_start = jiffies; | |
95392c59 | 1851 | return kernel_connect(sock, xs_addr(xprt), xprt->addrlen, O_NONBLOCK); |
16be2d20 CL |
1852 | } |
1853 | ||
9903cd1c | 1854 | /** |
b61d59ff TM |
1855 | * xs_tcp_setup_socket - create a TCP socket and connect to a remote endpoint |
1856 | * @xprt: RPC transport to connect | |
1857 | * @transport: socket transport to connect | |
1858 | * @create_sock: function to create a socket of the correct type | |
9903cd1c CL |
1859 | * |
1860 | * Invoked by a work queue tasklet. | |
a246b010 | 1861 | */ |
b61d59ff TM |
1862 | static void xs_tcp_setup_socket(struct rpc_xprt *xprt, |
1863 | struct sock_xprt *transport, | |
1864 | struct socket *(*create_sock)(struct rpc_xprt *, | |
1865 | struct sock_xprt *)) | |
a246b010 | 1866 | { |
ee0ac0c2 | 1867 | struct socket *sock = transport->sock; |
b61d59ff | 1868 | int status = -EIO; |
a246b010 | 1869 | |
01d37c42 | 1870 | if (xprt->shutdown) |
a246b010 CL |
1871 | goto out; |
1872 | ||
ee0ac0c2 | 1873 | if (!sock) { |
7d1e8255 | 1874 | clear_bit(XPRT_CONNECTION_ABORT, &xprt->state); |
b61d59ff TM |
1875 | sock = create_sock(xprt, transport); |
1876 | if (IS_ERR(sock)) { | |
1877 | status = PTR_ERR(sock); | |
3167e12c CL |
1878 | goto out; |
1879 | } | |
7d1e8255 TM |
1880 | } else { |
1881 | int abort_and_exit; | |
a246b010 | 1882 | |
7d1e8255 TM |
1883 | abort_and_exit = test_and_clear_bit(XPRT_CONNECTION_ABORT, |
1884 | &xprt->state); | |
3167e12c | 1885 | /* "close" the socket, preserving the local port */ |
40d2549d | 1886 | xs_tcp_reuse_connection(xprt, transport); |
a246b010 | 1887 | |
7d1e8255 TM |
1888 | if (abort_and_exit) |
1889 | goto out_eagain; | |
1890 | } | |
a246b010 | 1891 | |
46121cf7 | 1892 | dprintk("RPC: worker connecting xprt %p to address: %s\n", |
7559c7a2 | 1893 | xprt, xprt->address_strings[RPC_DISPLAY_ALL]); |
edb267a6 | 1894 | |
16be2d20 | 1895 | status = xs_tcp_finish_connecting(xprt, sock); |
46121cf7 CL |
1896 | dprintk("RPC: %p connect status %d connected %d sock state %d\n", |
1897 | xprt, -status, xprt_connected(xprt), | |
1898 | sock->sk->sk_state); | |
2a491991 | 1899 | switch (status) { |
f75e6745 TM |
1900 | default: |
1901 | printk("%s: connect returned unhandled error %d\n", | |
1902 | __func__, status); | |
1903 | case -EADDRNOTAVAIL: | |
1904 | /* We're probably in TIME_WAIT. Get rid of existing socket, | |
1905 | * and retry | |
1906 | */ | |
1907 | set_bit(XPRT_CONNECTION_CLOSE, &xprt->state); | |
1908 | xprt_force_disconnect(xprt); | |
88b5ed73 | 1909 | break; |
8a2cec29 TM |
1910 | case -ECONNREFUSED: |
1911 | case -ECONNRESET: | |
1912 | case -ENETUNREACH: | |
1913 | /* retry with existing socket, after a delay */ | |
2a491991 TM |
1914 | case 0: |
1915 | case -EINPROGRESS: | |
1916 | case -EALREADY: | |
7d1e8255 TM |
1917 | xprt_clear_connecting(xprt); |
1918 | return; | |
a246b010 | 1919 | } |
7d1e8255 | 1920 | out_eagain: |
2a491991 | 1921 | status = -EAGAIN; |
a246b010 | 1922 | out: |
68e220bd | 1923 | xprt_clear_connecting(xprt); |
7d1e8255 | 1924 | xprt_wake_pending_tasks(xprt, status); |
68e220bd | 1925 | } |
b0d93ad5 | 1926 | |
b61d59ff TM |
1927 | static struct socket *xs_create_tcp_sock4(struct rpc_xprt *xprt, |
1928 | struct sock_xprt *transport) | |
1929 | { | |
1930 | struct socket *sock; | |
1931 | int err; | |
1932 | ||
1933 | /* start from scratch */ | |
1934 | err = sock_create_kern(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock); | |
1935 | if (err < 0) { | |
1936 | dprintk("RPC: can't create TCP transport socket (%d).\n", | |
1937 | -err); | |
1938 | goto out_err; | |
1939 | } | |
1940 | xs_reclassify_socket4(sock); | |
1941 | ||
1942 | if (xs_bind4(transport, sock) < 0) { | |
1943 | sock_release(sock); | |
1944 | goto out_err; | |
1945 | } | |
1946 | return sock; | |
1947 | out_err: | |
1948 | return ERR_PTR(-EIO); | |
68e220bd | 1949 | } |
b0d93ad5 | 1950 | |
68e220bd | 1951 | /** |
b61d59ff | 1952 | * xs_tcp_connect_worker4 - connect a TCP socket to a remote endpoint |
68e220bd CL |
1953 | * @work: RPC transport to connect |
1954 | * | |
1955 | * Invoked by a work queue tasklet. | |
1956 | */ | |
b61d59ff | 1957 | static void xs_tcp_connect_worker4(struct work_struct *work) |
68e220bd CL |
1958 | { |
1959 | struct sock_xprt *transport = | |
1960 | container_of(work, struct sock_xprt, connect_worker.work); | |
1961 | struct rpc_xprt *xprt = &transport->xprt; | |
3167e12c | 1962 | |
b61d59ff TM |
1963 | xs_tcp_setup_socket(xprt, transport, xs_create_tcp_sock4); |
1964 | } | |
b0d93ad5 | 1965 | |
b61d59ff TM |
1966 | static struct socket *xs_create_tcp_sock6(struct rpc_xprt *xprt, |
1967 | struct sock_xprt *transport) | |
1968 | { | |
1969 | struct socket *sock; | |
1970 | int err; | |
7d1e8255 | 1971 | |
b61d59ff TM |
1972 | /* start from scratch */ |
1973 | err = sock_create_kern(PF_INET6, SOCK_STREAM, IPPROTO_TCP, &sock); | |
1974 | if (err < 0) { | |
1975 | dprintk("RPC: can't create TCP transport socket (%d).\n", | |
1976 | -err); | |
1977 | goto out_err; | |
1978 | } | |
1979 | xs_reclassify_socket6(sock); | |
b0d93ad5 | 1980 | |
b61d59ff TM |
1981 | if (xs_bind6(transport, sock) < 0) { |
1982 | sock_release(sock); | |
1983 | goto out_err; | |
7d1e8255 | 1984 | } |
b61d59ff TM |
1985 | return sock; |
1986 | out_err: | |
1987 | return ERR_PTR(-EIO); | |
1988 | } | |
b0d93ad5 | 1989 | |
b61d59ff TM |
1990 | /** |
1991 | * xs_tcp_connect_worker6 - connect a TCP socket to a remote endpoint | |
1992 | * @work: RPC transport to connect | |
1993 | * | |
1994 | * Invoked by a work queue tasklet. | |
1995 | */ | |
1996 | static void xs_tcp_connect_worker6(struct work_struct *work) | |
1997 | { | |
1998 | struct sock_xprt *transport = | |
1999 | container_of(work, struct sock_xprt, connect_worker.work); | |
2000 | struct rpc_xprt *xprt = &transport->xprt; | |
b0d93ad5 | 2001 | |
b61d59ff | 2002 | xs_tcp_setup_socket(xprt, transport, xs_create_tcp_sock6); |
a246b010 CL |
2003 | } |
2004 | ||
9903cd1c CL |
2005 | /** |
2006 | * xs_connect - connect a socket to a remote endpoint | |
2007 | * @task: address of RPC task that manages state of connect request | |
2008 | * | |
2009 | * TCP: If the remote end dropped the connection, delay reconnecting. | |
03bf4b70 CL |
2010 | * |
2011 | * UDP socket connects are synchronous, but we use a work queue anyway | |
2012 | * to guarantee that even unprivileged user processes can set up a | |
2013 | * socket on a privileged port. | |
2014 | * | |
2015 | * If a UDP socket connect fails, the delay behavior here prevents | |
2016 | * retry floods (hard mounts). | |
9903cd1c CL |
2017 | */ |
2018 | static void xs_connect(struct rpc_task *task) | |
a246b010 CL |
2019 | { |
2020 | struct rpc_xprt *xprt = task->tk_xprt; | |
ee0ac0c2 | 2021 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
a246b010 | 2022 | |
b0d93ad5 CL |
2023 | if (xprt_test_and_set_connecting(xprt)) |
2024 | return; | |
2025 | ||
ee0ac0c2 | 2026 | if (transport->sock != NULL) { |
46121cf7 CL |
2027 | dprintk("RPC: xs_connect delayed xprt %p for %lu " |
2028 | "seconds\n", | |
03bf4b70 | 2029 | xprt, xprt->reestablish_timeout / HZ); |
c1384c9c TM |
2030 | queue_delayed_work(rpciod_workqueue, |
2031 | &transport->connect_worker, | |
2032 | xprt->reestablish_timeout); | |
03bf4b70 CL |
2033 | xprt->reestablish_timeout <<= 1; |
2034 | if (xprt->reestablish_timeout > XS_TCP_MAX_REEST_TO) | |
2035 | xprt->reestablish_timeout = XS_TCP_MAX_REEST_TO; | |
b0d93ad5 | 2036 | } else { |
46121cf7 | 2037 | dprintk("RPC: xs_connect scheduled xprt %p\n", xprt); |
c1384c9c TM |
2038 | queue_delayed_work(rpciod_workqueue, |
2039 | &transport->connect_worker, 0); | |
a246b010 CL |
2040 | } |
2041 | } | |
2042 | ||
e06799f9 TM |
2043 | static void xs_tcp_connect(struct rpc_task *task) |
2044 | { | |
2045 | struct rpc_xprt *xprt = task->tk_xprt; | |
2046 | ||
e06799f9 TM |
2047 | /* Exit if we need to wait for socket shutdown to complete */ |
2048 | if (test_bit(XPRT_CLOSING, &xprt->state)) | |
2049 | return; | |
2050 | xs_connect(task); | |
2051 | } | |
2052 | ||
262ca07d CL |
2053 | /** |
2054 | * xs_udp_print_stats - display UDP socket-specifc stats | |
2055 | * @xprt: rpc_xprt struct containing statistics | |
2056 | * @seq: output file | |
2057 | * | |
2058 | */ | |
2059 | static void xs_udp_print_stats(struct rpc_xprt *xprt, struct seq_file *seq) | |
2060 | { | |
c8475461 CL |
2061 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
2062 | ||
262ca07d | 2063 | seq_printf(seq, "\txprt:\tudp %u %lu %lu %lu %lu %Lu %Lu\n", |
c8475461 | 2064 | transport->port, |
262ca07d CL |
2065 | xprt->stat.bind_count, |
2066 | xprt->stat.sends, | |
2067 | xprt->stat.recvs, | |
2068 | xprt->stat.bad_xids, | |
2069 | xprt->stat.req_u, | |
2070 | xprt->stat.bklog_u); | |
2071 | } | |
2072 | ||
2073 | /** | |
2074 | * xs_tcp_print_stats - display TCP socket-specifc stats | |
2075 | * @xprt: rpc_xprt struct containing statistics | |
2076 | * @seq: output file | |
2077 | * | |
2078 | */ | |
2079 | static void xs_tcp_print_stats(struct rpc_xprt *xprt, struct seq_file *seq) | |
2080 | { | |
c8475461 | 2081 | struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); |
262ca07d CL |
2082 | long idle_time = 0; |
2083 | ||
2084 | if (xprt_connected(xprt)) | |
2085 | idle_time = (long)(jiffies - xprt->last_used) / HZ; | |
2086 | ||
2087 | seq_printf(seq, "\txprt:\ttcp %u %lu %lu %lu %ld %lu %lu %lu %Lu %Lu\n", | |
c8475461 | 2088 | transport->port, |
262ca07d CL |
2089 | xprt->stat.bind_count, |
2090 | xprt->stat.connect_count, | |
2091 | xprt->stat.connect_time, | |
2092 | idle_time, | |
2093 | xprt->stat.sends, | |
2094 | xprt->stat.recvs, | |
2095 | xprt->stat.bad_xids, | |
2096 | xprt->stat.req_u, | |
2097 | xprt->stat.bklog_u); | |
2098 | } | |
2099 | ||
262965f5 | 2100 | static struct rpc_xprt_ops xs_udp_ops = { |
43118c29 | 2101 | .set_buffer_size = xs_udp_set_buffer_size, |
12a80469 | 2102 | .reserve_xprt = xprt_reserve_xprt_cong, |
49e9a890 | 2103 | .release_xprt = xprt_release_xprt_cong, |
45160d62 | 2104 | .rpcbind = rpcb_getport_async, |
92200412 | 2105 | .set_port = xs_set_port, |
262965f5 | 2106 | .connect = xs_connect, |
02107148 CL |
2107 | .buf_alloc = rpc_malloc, |
2108 | .buf_free = rpc_free, | |
262965f5 | 2109 | .send_request = xs_udp_send_request, |
fe3aca29 | 2110 | .set_retrans_timeout = xprt_set_retrans_timeout_rtt, |
46c0ee8b | 2111 | .timer = xs_udp_timer, |
a58dd398 | 2112 | .release_request = xprt_release_rqst_cong, |
262965f5 CL |
2113 | .close = xs_close, |
2114 | .destroy = xs_destroy, | |
262ca07d | 2115 | .print_stats = xs_udp_print_stats, |
262965f5 CL |
2116 | }; |
2117 | ||
2118 | static struct rpc_xprt_ops xs_tcp_ops = { | |
12a80469 | 2119 | .reserve_xprt = xprt_reserve_xprt, |
e0ab53de | 2120 | .release_xprt = xs_tcp_release_xprt, |
45160d62 | 2121 | .rpcbind = rpcb_getport_async, |
92200412 | 2122 | .set_port = xs_set_port, |
e06799f9 | 2123 | .connect = xs_tcp_connect, |
02107148 CL |
2124 | .buf_alloc = rpc_malloc, |
2125 | .buf_free = rpc_free, | |
262965f5 | 2126 | .send_request = xs_tcp_send_request, |
fe3aca29 | 2127 | .set_retrans_timeout = xprt_set_retrans_timeout_def, |
0d90ba1c RL |
2128 | #if defined(CONFIG_NFS_V4_1) |
2129 | .release_request = bc_release_request, | |
2130 | #endif /* CONFIG_NFS_V4_1 */ | |
f75e6745 | 2131 | .close = xs_tcp_close, |
9903cd1c | 2132 | .destroy = xs_destroy, |
262ca07d | 2133 | .print_stats = xs_tcp_print_stats, |
a246b010 CL |
2134 | }; |
2135 | ||
3c341b0b | 2136 | static struct rpc_xprt *xs_setup_xprt(struct xprt_create *args, |
bc25571e | 2137 | unsigned int slot_table_size) |
c8541ecd CL |
2138 | { |
2139 | struct rpc_xprt *xprt; | |
ffc2e518 | 2140 | struct sock_xprt *new; |
c8541ecd | 2141 | |
96802a09 | 2142 | if (args->addrlen > sizeof(xprt->addr)) { |
46121cf7 | 2143 | dprintk("RPC: xs_setup_xprt: address too large\n"); |
c8541ecd CL |
2144 | return ERR_PTR(-EBADF); |
2145 | } | |
2146 | ||
ffc2e518 CL |
2147 | new = kzalloc(sizeof(*new), GFP_KERNEL); |
2148 | if (new == NULL) { | |
46121cf7 CL |
2149 | dprintk("RPC: xs_setup_xprt: couldn't allocate " |
2150 | "rpc_xprt\n"); | |
c8541ecd CL |
2151 | return ERR_PTR(-ENOMEM); |
2152 | } | |
ffc2e518 | 2153 | xprt = &new->xprt; |
c8541ecd CL |
2154 | |
2155 | xprt->max_reqs = slot_table_size; | |
2156 | xprt->slot = kcalloc(xprt->max_reqs, sizeof(struct rpc_rqst), GFP_KERNEL); | |
2157 | if (xprt->slot == NULL) { | |
2158 | kfree(xprt); | |
46121cf7 CL |
2159 | dprintk("RPC: xs_setup_xprt: couldn't allocate slot " |
2160 | "table\n"); | |
c8541ecd CL |
2161 | return ERR_PTR(-ENOMEM); |
2162 | } | |
2163 | ||
96802a09 FM |
2164 | memcpy(&xprt->addr, args->dstaddr, args->addrlen); |
2165 | xprt->addrlen = args->addrlen; | |
d3bc9a1d FM |
2166 | if (args->srcaddr) |
2167 | memcpy(&new->addr, args->srcaddr, args->addrlen); | |
c8541ecd CL |
2168 | |
2169 | return xprt; | |
2170 | } | |
2171 | ||
2881ae74 TM |
2172 | static const struct rpc_timeout xs_udp_default_timeout = { |
2173 | .to_initval = 5 * HZ, | |
2174 | .to_maxval = 30 * HZ, | |
2175 | .to_increment = 5 * HZ, | |
2176 | .to_retries = 5, | |
2177 | }; | |
2178 | ||
9903cd1c CL |
2179 | /** |
2180 | * xs_setup_udp - Set up transport to use a UDP socket | |
96802a09 | 2181 | * @args: rpc transport creation arguments |
9903cd1c CL |
2182 | * |
2183 | */ | |
483066d6 | 2184 | static struct rpc_xprt *xs_setup_udp(struct xprt_create *args) |
a246b010 | 2185 | { |
8f9d5b1a | 2186 | struct sockaddr *addr = args->dstaddr; |
c8541ecd | 2187 | struct rpc_xprt *xprt; |
c8475461 | 2188 | struct sock_xprt *transport; |
a246b010 | 2189 | |
96802a09 | 2190 | xprt = xs_setup_xprt(args, xprt_udp_slot_table_entries); |
c8541ecd CL |
2191 | if (IS_ERR(xprt)) |
2192 | return xprt; | |
c8475461 | 2193 | transport = container_of(xprt, struct sock_xprt, xprt); |
a246b010 | 2194 | |
ec739ef0 | 2195 | xprt->prot = IPPROTO_UDP; |
808012fb | 2196 | xprt->tsh_size = 0; |
a246b010 CL |
2197 | /* XXX: header size can vary due to auth type, IPv6, etc. */ |
2198 | xprt->max_payload = (1U << 16) - (MAX_HEADER << 3); | |
2199 | ||
03bf4b70 CL |
2200 | xprt->bind_timeout = XS_BIND_TO; |
2201 | xprt->connect_timeout = XS_UDP_CONN_TO; | |
2202 | xprt->reestablish_timeout = XS_UDP_REEST_TO; | |
2203 | xprt->idle_timeout = XS_IDLE_DISC_TO; | |
a246b010 | 2204 | |
262965f5 | 2205 | xprt->ops = &xs_udp_ops; |
a246b010 | 2206 | |
ba7392bb | 2207 | xprt->timeout = &xs_udp_default_timeout; |
a246b010 | 2208 | |
8f9d5b1a CL |
2209 | switch (addr->sa_family) { |
2210 | case AF_INET: | |
2211 | if (((struct sockaddr_in *)addr)->sin_port != htons(0)) | |
2212 | xprt_set_bound(xprt); | |
2213 | ||
2214 | INIT_DELAYED_WORK(&transport->connect_worker, | |
2215 | xs_udp_connect_worker4); | |
b454ae90 | 2216 | xs_format_ipv4_peer_addresses(xprt, "udp", RPCBIND_NETID_UDP); |
8f9d5b1a CL |
2217 | break; |
2218 | case AF_INET6: | |
2219 | if (((struct sockaddr_in6 *)addr)->sin6_port != htons(0)) | |
2220 | xprt_set_bound(xprt); | |
2221 | ||
2222 | INIT_DELAYED_WORK(&transport->connect_worker, | |
2223 | xs_udp_connect_worker6); | |
b454ae90 | 2224 | xs_format_ipv6_peer_addresses(xprt, "udp", RPCBIND_NETID_UDP6); |
8f9d5b1a CL |
2225 | break; |
2226 | default: | |
2227 | kfree(xprt); | |
2228 | return ERR_PTR(-EAFNOSUPPORT); | |
2229 | } | |
2230 | ||
46121cf7 | 2231 | dprintk("RPC: set up transport to address %s\n", |
7559c7a2 | 2232 | xprt->address_strings[RPC_DISPLAY_ALL]); |
edb267a6 | 2233 | |
bc25571e TT |
2234 | if (try_module_get(THIS_MODULE)) |
2235 | return xprt; | |
2236 | ||
2237 | kfree(xprt->slot); | |
2238 | kfree(xprt); | |
2239 | return ERR_PTR(-EINVAL); | |
a246b010 CL |
2240 | } |
2241 | ||
2881ae74 TM |
2242 | static const struct rpc_timeout xs_tcp_default_timeout = { |
2243 | .to_initval = 60 * HZ, | |
2244 | .to_maxval = 60 * HZ, | |
2245 | .to_retries = 2, | |
2246 | }; | |
2247 | ||
9903cd1c CL |
2248 | /** |
2249 | * xs_setup_tcp - Set up transport to use a TCP socket | |
96802a09 | 2250 | * @args: rpc transport creation arguments |
9903cd1c CL |
2251 | * |
2252 | */ | |
483066d6 | 2253 | static struct rpc_xprt *xs_setup_tcp(struct xprt_create *args) |
a246b010 | 2254 | { |
8f9d5b1a | 2255 | struct sockaddr *addr = args->dstaddr; |
c8541ecd | 2256 | struct rpc_xprt *xprt; |
c8475461 | 2257 | struct sock_xprt *transport; |
a246b010 | 2258 | |
96802a09 | 2259 | xprt = xs_setup_xprt(args, xprt_tcp_slot_table_entries); |
c8541ecd CL |
2260 | if (IS_ERR(xprt)) |
2261 | return xprt; | |
c8475461 | 2262 | transport = container_of(xprt, struct sock_xprt, xprt); |
a246b010 | 2263 | |
ec739ef0 | 2264 | xprt->prot = IPPROTO_TCP; |
808012fb | 2265 | xprt->tsh_size = sizeof(rpc_fraghdr) / sizeof(u32); |
808012fb | 2266 | xprt->max_payload = RPC_MAX_FRAGMENT_SIZE; |
a246b010 | 2267 | |
03bf4b70 CL |
2268 | xprt->bind_timeout = XS_BIND_TO; |
2269 | xprt->connect_timeout = XS_TCP_CONN_TO; | |
2270 | xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO; | |
2271 | xprt->idle_timeout = XS_IDLE_DISC_TO; | |
a246b010 | 2272 | |
262965f5 | 2273 | xprt->ops = &xs_tcp_ops; |
ba7392bb | 2274 | xprt->timeout = &xs_tcp_default_timeout; |
a246b010 | 2275 | |
8f9d5b1a CL |
2276 | switch (addr->sa_family) { |
2277 | case AF_INET: | |
2278 | if (((struct sockaddr_in *)addr)->sin_port != htons(0)) | |
2279 | xprt_set_bound(xprt); | |
2280 | ||
2281 | INIT_DELAYED_WORK(&transport->connect_worker, xs_tcp_connect_worker4); | |
b454ae90 | 2282 | xs_format_ipv4_peer_addresses(xprt, "tcp", RPCBIND_NETID_TCP); |
8f9d5b1a CL |
2283 | break; |
2284 | case AF_INET6: | |
2285 | if (((struct sockaddr_in6 *)addr)->sin6_port != htons(0)) | |
2286 | xprt_set_bound(xprt); | |
2287 | ||
2288 | INIT_DELAYED_WORK(&transport->connect_worker, xs_tcp_connect_worker6); | |
b454ae90 | 2289 | xs_format_ipv6_peer_addresses(xprt, "tcp", RPCBIND_NETID_TCP6); |
8f9d5b1a CL |
2290 | break; |
2291 | default: | |
2292 | kfree(xprt); | |
2293 | return ERR_PTR(-EAFNOSUPPORT); | |
2294 | } | |
2295 | ||
46121cf7 | 2296 | dprintk("RPC: set up transport to address %s\n", |
7559c7a2 | 2297 | xprt->address_strings[RPC_DISPLAY_ALL]); |
edb267a6 | 2298 | |
bc25571e TT |
2299 | if (try_module_get(THIS_MODULE)) |
2300 | return xprt; | |
2301 | ||
2302 | kfree(xprt->slot); | |
2303 | kfree(xprt); | |
2304 | return ERR_PTR(-EINVAL); | |
a246b010 | 2305 | } |
282b32e1 | 2306 | |
bc25571e TT |
2307 | static struct xprt_class xs_udp_transport = { |
2308 | .list = LIST_HEAD_INIT(xs_udp_transport.list), | |
2309 | .name = "udp", | |
2310 | .owner = THIS_MODULE, | |
4fa016eb | 2311 | .ident = IPPROTO_UDP, |
bc25571e TT |
2312 | .setup = xs_setup_udp, |
2313 | }; | |
2314 | ||
2315 | static struct xprt_class xs_tcp_transport = { | |
2316 | .list = LIST_HEAD_INIT(xs_tcp_transport.list), | |
2317 | .name = "tcp", | |
2318 | .owner = THIS_MODULE, | |
4fa016eb | 2319 | .ident = IPPROTO_TCP, |
bc25571e TT |
2320 | .setup = xs_setup_tcp, |
2321 | }; | |
2322 | ||
282b32e1 | 2323 | /** |
bc25571e | 2324 | * init_socket_xprt - set up xprtsock's sysctls, register with RPC client |
282b32e1 CL |
2325 | * |
2326 | */ | |
2327 | int init_socket_xprt(void) | |
2328 | { | |
fbf76683 | 2329 | #ifdef RPC_DEBUG |
2b1bec5f | 2330 | if (!sunrpc_table_header) |
0b4d4147 | 2331 | sunrpc_table_header = register_sysctl_table(sunrpc_table); |
fbf76683 CL |
2332 | #endif |
2333 | ||
bc25571e TT |
2334 | xprt_register_transport(&xs_udp_transport); |
2335 | xprt_register_transport(&xs_tcp_transport); | |
2336 | ||
282b32e1 CL |
2337 | return 0; |
2338 | } | |
2339 | ||
2340 | /** | |
bc25571e | 2341 | * cleanup_socket_xprt - remove xprtsock's sysctls, unregister |
282b32e1 CL |
2342 | * |
2343 | */ | |
2344 | void cleanup_socket_xprt(void) | |
2345 | { | |
fbf76683 CL |
2346 | #ifdef RPC_DEBUG |
2347 | if (sunrpc_table_header) { | |
2348 | unregister_sysctl_table(sunrpc_table_header); | |
2349 | sunrpc_table_header = NULL; | |
2350 | } | |
2351 | #endif | |
bc25571e TT |
2352 | |
2353 | xprt_unregister_transport(&xs_udp_transport); | |
2354 | xprt_unregister_transport(&xs_tcp_transport); | |
282b32e1 | 2355 | } |