1 /* 2 * Copyright (c) 1996 John S. Dyson 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice immediately at the beginning of the file, without modification, 10 * this list of conditions, and the following disclaimer. 11 * 2. Redistributions in binary form must reproduce the above copyright 12 * notice, this list of conditions and the following disclaimer in the 13 * documentation and/or other materials provided with the distribution. 14 * 3. Absolutely no warranty of function or purpose is made by the author 15 * John S. Dyson. 16 * 4. Modifications may be freely made to this file if the above conditions 17 * are met. 18 * 19 * $FreeBSD: src/sys/kern/sys_pipe.c,v 1.60.2.13 2002/08/05 15:05:15 des Exp $ 20 * $DragonFly: src/sys/kern/sys_pipe.c,v 1.50 2008/09/09 04:06:13 dillon Exp $ 21 */ 22 23 /* 24 * This file contains a high-performance replacement for the socket-based 25 * pipes scheme originally used in FreeBSD/4.4Lite. It does not support 26 * all features of sockets, but does do everything that pipes normally 27 * do. 28 */ 29 #include <sys/param.h> 30 #include <sys/systm.h> 31 #include <sys/kernel.h> 32 #include <sys/proc.h> 33 #include <sys/fcntl.h> 34 #include <sys/file.h> 35 #include <sys/filedesc.h> 36 #include <sys/filio.h> 37 #include <sys/ttycom.h> 38 #include <sys/stat.h> 39 #include <sys/signalvar.h> 40 #include <sys/sysproto.h> 41 #include <sys/pipe.h> 42 #include <sys/vnode.h> 43 #include <sys/uio.h> 44 #include <sys/event.h> 45 #include <sys/globaldata.h> 46 #include <sys/module.h> 47 #include <sys/malloc.h> 48 #include <sys/sysctl.h> 49 #include <sys/socket.h> 50 51 #include <vm/vm.h> 52 #include <vm/vm_param.h> 53 #include <sys/lock.h> 54 #include <vm/vm_object.h> 55 #include <vm/vm_kern.h> 56 #include <vm/vm_extern.h> 57 #include <vm/pmap.h> 58 #include <vm/vm_map.h> 59 #include <vm/vm_page.h> 60 #include <vm/vm_zone.h> 61 62 #include <sys/file2.h> 63 #include <sys/signal2.h> 64 65 #include <machine/cpufunc.h> 66 67 /* 68 * interfaces to the outside world 69 */ 70 static int pipe_read (struct file *fp, struct uio *uio, 71 struct ucred *cred, int flags); 72 static int pipe_write (struct file *fp, struct uio *uio, 73 struct ucred *cred, int flags); 74 static int pipe_close (struct file *fp); 75 static int pipe_shutdown (struct file *fp, int how); 76 static int pipe_kqfilter (struct file *fp, struct knote *kn); 77 static int pipe_stat (struct file *fp, struct stat *sb, struct ucred *cred); 78 static int pipe_ioctl (struct file *fp, u_long cmd, caddr_t data, 79 struct ucred *cred, struct sysmsg *msg); 80 81 static struct fileops pipeops = { 82 .fo_read = pipe_read, 83 .fo_write = pipe_write, 84 .fo_ioctl = pipe_ioctl, 85 .fo_kqfilter = pipe_kqfilter, 86 .fo_stat = pipe_stat, 87 .fo_close = pipe_close, 88 .fo_shutdown = pipe_shutdown 89 }; 90 91 static void filt_pipedetach(struct knote *kn); 92 static int filt_piperead(struct knote *kn, long hint); 93 static int filt_pipewrite(struct knote *kn, long hint); 94 95 static struct filterops pipe_rfiltops = 96 { FILTEROP_ISFD|FILTEROP_MPSAFE, NULL, filt_pipedetach, filt_piperead }; 97 static struct filterops pipe_wfiltops = 98 { FILTEROP_ISFD|FILTEROP_MPSAFE, NULL, filt_pipedetach, filt_pipewrite }; 99 100 MALLOC_DEFINE(M_PIPE, "pipe", "pipe structures"); 101 102 /* 103 * Default pipe buffer size(s), this can be kind-of large now because pipe 104 * space is pageable. The pipe code will try to maintain locality of 105 * reference for performance reasons, so small amounts of outstanding I/O 106 * will not wipe the cache. 107 */ 108 #define MINPIPESIZE (PIPE_SIZE/3) 109 #define MAXPIPESIZE (2*PIPE_SIZE/3) 110 111 /* 112 * Limit the number of "big" pipes 113 */ 114 #define LIMITBIGPIPES 64 115 #define PIPEQ_MAX_CACHE 16 /* per-cpu pipe structure cache */ 116 117 static int pipe_maxbig = LIMITBIGPIPES; 118 static int pipe_maxcache = PIPEQ_MAX_CACHE; 119 static int pipe_bigcount; 120 static int pipe_nbig; 121 static int pipe_bcache_alloc; 122 static int pipe_bkmem_alloc; 123 static int pipe_rblocked_count; 124 static int pipe_wblocked_count; 125 126 SYSCTL_NODE(_kern, OID_AUTO, pipe, CTLFLAG_RW, 0, "Pipe operation"); 127 SYSCTL_INT(_kern_pipe, OID_AUTO, nbig, 128 CTLFLAG_RD, &pipe_nbig, 0, "numer of big pipes allocated"); 129 SYSCTL_INT(_kern_pipe, OID_AUTO, bigcount, 130 CTLFLAG_RW, &pipe_bigcount, 0, "number of times pipe expanded"); 131 SYSCTL_INT(_kern_pipe, OID_AUTO, rblocked, 132 CTLFLAG_RW, &pipe_rblocked_count, 0, "number of times pipe expanded"); 133 SYSCTL_INT(_kern_pipe, OID_AUTO, wblocked, 134 CTLFLAG_RW, &pipe_wblocked_count, 0, "number of times pipe expanded"); 135 SYSCTL_INT(_kern_pipe, OID_AUTO, maxcache, 136 CTLFLAG_RW, &pipe_maxcache, 0, "max pipes cached per-cpu"); 137 SYSCTL_INT(_kern_pipe, OID_AUTO, maxbig, 138 CTLFLAG_RW, &pipe_maxbig, 0, "max number of big pipes"); 139 #ifdef SMP 140 static int pipe_delay = 5000; /* 5uS default */ 141 SYSCTL_INT(_kern_pipe, OID_AUTO, delay, 142 CTLFLAG_RW, &pipe_delay, 0, "SMP delay optimization in ns"); 143 #endif 144 #if !defined(NO_PIPE_SYSCTL_STATS) 145 SYSCTL_INT(_kern_pipe, OID_AUTO, bcache_alloc, 146 CTLFLAG_RW, &pipe_bcache_alloc, 0, "pipe buffer from pcpu cache"); 147 SYSCTL_INT(_kern_pipe, OID_AUTO, bkmem_alloc, 148 CTLFLAG_RW, &pipe_bkmem_alloc, 0, "pipe buffer from kmem"); 149 #endif 150 151 /* 152 * Auto-size pipe cache to reduce kmem allocations and frees. 153 */ 154 static 155 void 156 pipeinit(void *dummy) 157 { 158 size_t mbytes = kmem_lim_size(); 159 160 if (pipe_maxbig == LIMITBIGPIPES) { 161 if (mbytes >= 7 * 1024) 162 pipe_maxbig *= 2; 163 if (mbytes >= 15 * 1024) 164 pipe_maxbig *= 2; 165 } 166 if (pipe_maxcache == PIPEQ_MAX_CACHE) { 167 if (mbytes >= 7 * 1024) 168 pipe_maxcache *= 2; 169 if (mbytes >= 15 * 1024) 170 pipe_maxcache *= 2; 171 } 172 } 173 SYSINIT(kmem, SI_BOOT2_MACHDEP, SI_ORDER_ANY, pipeinit, NULL) 174 175 static void pipeclose (struct pipe *cpipe); 176 static void pipe_free_kmem (struct pipe *cpipe); 177 static int pipe_create (struct pipe **cpipep); 178 static int pipespace (struct pipe *cpipe, int size); 179 180 static __inline void 181 pipewakeup(struct pipe *cpipe, int dosigio) 182 { 183 if (dosigio && (cpipe->pipe_state & PIPE_ASYNC) && cpipe->pipe_sigio) { 184 lwkt_gettoken(&proc_token); 185 pgsigio(cpipe->pipe_sigio, SIGIO, 0); 186 lwkt_reltoken(&proc_token); 187 } 188 KNOTE(&cpipe->pipe_kq.ki_note, 0); 189 } 190 191 /* 192 * These routines are called before and after a UIO. The UIO 193 * may block, causing our held tokens to be lost temporarily. 194 * 195 * We use these routines to serialize reads against other reads 196 * and writes against other writes. 197 * 198 * The read token is held on entry so *ipp does not race. 199 */ 200 static __inline int 201 pipe_start_uio(struct pipe *cpipe, int *ipp) 202 { 203 int error; 204 205 while (*ipp) { 206 *ipp = -1; 207 error = tsleep(ipp, PCATCH, "pipexx", 0); 208 if (error) 209 return (error); 210 } 211 *ipp = 1; 212 return (0); 213 } 214 215 static __inline void 216 pipe_end_uio(struct pipe *cpipe, int *ipp) 217 { 218 if (*ipp < 0) { 219 *ipp = 0; 220 wakeup(ipp); 221 } else { 222 KKASSERT(*ipp > 0); 223 *ipp = 0; 224 } 225 } 226 227 /* 228 * The pipe system call for the DTYPE_PIPE type of pipes 229 * 230 * pipe_args(int dummy) 231 * 232 * MPSAFE 233 */ 234 int 235 sys_pipe(struct pipe_args *uap) 236 { 237 struct thread *td = curthread; 238 struct filedesc *fdp = td->td_proc->p_fd; 239 struct file *rf, *wf; 240 struct pipe *rpipe, *wpipe; 241 int fd1, fd2, error; 242 243 rpipe = wpipe = NULL; 244 if (pipe_create(&rpipe) || pipe_create(&wpipe)) { 245 pipeclose(rpipe); 246 pipeclose(wpipe); 247 return (ENFILE); 248 } 249 250 error = falloc(td->td_lwp, &rf, &fd1); 251 if (error) { 252 pipeclose(rpipe); 253 pipeclose(wpipe); 254 return (error); 255 } 256 uap->sysmsg_fds[0] = fd1; 257 258 /* 259 * Warning: once we've gotten past allocation of the fd for the 260 * read-side, we can only drop the read side via fdrop() in order 261 * to avoid races against processes which manage to dup() the read 262 * side while we are blocked trying to allocate the write side. 263 */ 264 rf->f_type = DTYPE_PIPE; 265 rf->f_flag = FREAD | FWRITE; 266 rf->f_ops = &pipeops; 267 rf->f_data = rpipe; 268 error = falloc(td->td_lwp, &wf, &fd2); 269 if (error) { 270 fsetfd(fdp, NULL, fd1); 271 fdrop(rf); 272 /* rpipe has been closed by fdrop(). */ 273 pipeclose(wpipe); 274 return (error); 275 } 276 wf->f_type = DTYPE_PIPE; 277 wf->f_flag = FREAD | FWRITE; 278 wf->f_ops = &pipeops; 279 wf->f_data = wpipe; 280 uap->sysmsg_fds[1] = fd2; 281 282 rpipe->pipe_slock = kmalloc(sizeof(struct lock), 283 M_PIPE, M_WAITOK|M_ZERO); 284 wpipe->pipe_slock = rpipe->pipe_slock; 285 rpipe->pipe_peer = wpipe; 286 wpipe->pipe_peer = rpipe; 287 lockinit(rpipe->pipe_slock, "pipecl", 0, 0); 288 289 /* 290 * Once activated the peer relationship remains valid until 291 * both sides are closed. 292 */ 293 fsetfd(fdp, rf, fd1); 294 fsetfd(fdp, wf, fd2); 295 fdrop(rf); 296 fdrop(wf); 297 298 return (0); 299 } 300 301 /* 302 * Allocate kva for pipe circular buffer, the space is pageable 303 * This routine will 'realloc' the size of a pipe safely, if it fails 304 * it will retain the old buffer. 305 * If it fails it will return ENOMEM. 306 */ 307 static int 308 pipespace(struct pipe *cpipe, int size) 309 { 310 struct vm_object *object; 311 caddr_t buffer; 312 int npages, error; 313 314 npages = round_page(size) / PAGE_SIZE; 315 object = cpipe->pipe_buffer.object; 316 317 /* 318 * [re]create the object if necessary and reserve space for it 319 * in the kernel_map. The object and memory are pageable. On 320 * success, free the old resources before assigning the new 321 * ones. 322 */ 323 if (object == NULL || object->size != npages) { 324 object = vm_object_allocate(OBJT_DEFAULT, npages); 325 buffer = (caddr_t)vm_map_min(&kernel_map); 326 327 error = vm_map_find(&kernel_map, object, 0, 328 (vm_offset_t *)&buffer, 329 size, PAGE_SIZE, 330 1, VM_MAPTYPE_NORMAL, 331 VM_PROT_ALL, VM_PROT_ALL, 332 0); 333 334 if (error != KERN_SUCCESS) { 335 vm_object_deallocate(object); 336 return (ENOMEM); 337 } 338 pipe_free_kmem(cpipe); 339 cpipe->pipe_buffer.object = object; 340 cpipe->pipe_buffer.buffer = buffer; 341 cpipe->pipe_buffer.size = size; 342 ++pipe_bkmem_alloc; 343 } else { 344 ++pipe_bcache_alloc; 345 } 346 cpipe->pipe_buffer.rindex = 0; 347 cpipe->pipe_buffer.windex = 0; 348 return (0); 349 } 350 351 /* 352 * Initialize and allocate VM and memory for pipe, pulling the pipe from 353 * our per-cpu cache if possible. For now make sure it is sized for the 354 * smaller PIPE_SIZE default. 355 */ 356 static int 357 pipe_create(struct pipe **cpipep) 358 { 359 globaldata_t gd = mycpu; 360 struct pipe *cpipe; 361 int error; 362 363 if ((cpipe = gd->gd_pipeq) != NULL) { 364 gd->gd_pipeq = cpipe->pipe_peer; 365 --gd->gd_pipeqcount; 366 cpipe->pipe_peer = NULL; 367 cpipe->pipe_wantwcnt = 0; 368 } else { 369 cpipe = kmalloc(sizeof(struct pipe), M_PIPE, M_WAITOK|M_ZERO); 370 } 371 *cpipep = cpipe; 372 if ((error = pipespace(cpipe, PIPE_SIZE)) != 0) 373 return (error); 374 vfs_timestamp(&cpipe->pipe_ctime); 375 cpipe->pipe_atime = cpipe->pipe_ctime; 376 cpipe->pipe_mtime = cpipe->pipe_ctime; 377 lwkt_token_init(&cpipe->pipe_rlock, "piper"); 378 lwkt_token_init(&cpipe->pipe_wlock, "pipew"); 379 return (0); 380 } 381 382 static int 383 pipe_read(struct file *fp, struct uio *uio, struct ucred *cred, int fflags) 384 { 385 struct pipe *rpipe; 386 struct pipe *wpipe; 387 int error; 388 size_t nread = 0; 389 int nbio; 390 u_int size; /* total bytes available */ 391 u_int nsize; /* total bytes to read */ 392 u_int rindex; /* contiguous bytes available */ 393 int notify_writer; 394 int bigread; 395 int bigcount; 396 397 if (uio->uio_resid == 0) 398 return(0); 399 400 /* 401 * Setup locks, calculate nbio 402 */ 403 rpipe = (struct pipe *)fp->f_data; 404 wpipe = rpipe->pipe_peer; 405 lwkt_gettoken(&rpipe->pipe_rlock); 406 407 if (fflags & O_FBLOCKING) 408 nbio = 0; 409 else if (fflags & O_FNONBLOCKING) 410 nbio = 1; 411 else if (fp->f_flag & O_NONBLOCK) 412 nbio = 1; 413 else 414 nbio = 0; 415 416 /* 417 * Reads are serialized. Note however that pipe_buffer.buffer and 418 * pipe_buffer.size can change out from under us when the number 419 * of bytes in the buffer are zero due to the write-side doing a 420 * pipespace(). 421 */ 422 error = pipe_start_uio(rpipe, &rpipe->pipe_rip); 423 if (error) { 424 lwkt_reltoken(&rpipe->pipe_rlock); 425 return (error); 426 } 427 notify_writer = 0; 428 429 bigread = (uio->uio_resid > 10 * 1024 * 1024); 430 bigcount = 10; 431 432 while (uio->uio_resid) { 433 /* 434 * Don't hog the cpu. 435 */ 436 if (bigread && --bigcount == 0) { 437 lwkt_user_yield(); 438 bigcount = 10; 439 if (CURSIG(curthread->td_lwp)) { 440 error = EINTR; 441 break; 442 } 443 } 444 445 size = rpipe->pipe_buffer.windex - rpipe->pipe_buffer.rindex; 446 cpu_lfence(); 447 if (size) { 448 rindex = rpipe->pipe_buffer.rindex & 449 (rpipe->pipe_buffer.size - 1); 450 nsize = size; 451 if (nsize > rpipe->pipe_buffer.size - rindex) 452 nsize = rpipe->pipe_buffer.size - rindex; 453 nsize = szmin(nsize, uio->uio_resid); 454 455 error = uiomove(&rpipe->pipe_buffer.buffer[rindex], 456 nsize, uio); 457 if (error) 458 break; 459 cpu_mfence(); 460 rpipe->pipe_buffer.rindex += nsize; 461 nread += nsize; 462 463 /* 464 * If the FIFO is still over half full just continue 465 * and do not try to notify the writer yet. 466 */ 467 if (size - nsize >= (rpipe->pipe_buffer.size >> 1)) { 468 notify_writer = 0; 469 continue; 470 } 471 472 /* 473 * When the FIFO is less then half full notify any 474 * waiting writer. WANTW can be checked while 475 * holding just the rlock. 476 */ 477 notify_writer = 1; 478 if ((rpipe->pipe_state & PIPE_WANTW) == 0) 479 continue; 480 } 481 482 /* 483 * If the "write-side" was blocked we wake it up. This code 484 * is reached either when the buffer is completely emptied 485 * or if it becomes more then half-empty. 486 * 487 * Pipe_state can only be modified if both the rlock and 488 * wlock are held. 489 */ 490 if (rpipe->pipe_state & PIPE_WANTW) { 491 lwkt_gettoken(&rpipe->pipe_wlock); 492 if (rpipe->pipe_state & PIPE_WANTW) { 493 rpipe->pipe_state &= ~PIPE_WANTW; 494 lwkt_reltoken(&rpipe->pipe_wlock); 495 wakeup(rpipe); 496 } else { 497 lwkt_reltoken(&rpipe->pipe_wlock); 498 } 499 } 500 501 /* 502 * Pick up our copy loop again if the writer sent data to 503 * us while we were messing around. 504 * 505 * On a SMP box poll up to pipe_delay nanoseconds for new 506 * data. Typically a value of 2000 to 4000 is sufficient 507 * to eradicate most IPIs/tsleeps/wakeups when a pipe 508 * is used for synchronous communications with small packets, 509 * and 8000 or so (8uS) will pipeline large buffer xfers 510 * between cpus over a pipe. 511 * 512 * For synchronous communications a hit means doing a 513 * full Awrite-Bread-Bwrite-Aread cycle in less then 2uS, 514 * where as miss requiring a tsleep/wakeup sequence 515 * will take 7uS or more. 516 */ 517 if (rpipe->pipe_buffer.windex != rpipe->pipe_buffer.rindex) 518 continue; 519 520 #if defined(SMP) && defined(_RDTSC_SUPPORTED_) 521 if (pipe_delay) { 522 int64_t tsc_target; 523 int good = 0; 524 525 tsc_target = tsc_get_target(pipe_delay); 526 while (tsc_test_target(tsc_target) == 0) { 527 if (rpipe->pipe_buffer.windex != 528 rpipe->pipe_buffer.rindex) { 529 good = 1; 530 break; 531 } 532 } 533 if (good) 534 continue; 535 } 536 #endif 537 538 /* 539 * Detect EOF condition, do not set error. 540 */ 541 if (rpipe->pipe_state & PIPE_REOF) 542 break; 543 544 /* 545 * Break if some data was read, or if this was a non-blocking 546 * read. 547 */ 548 if (nread > 0) 549 break; 550 551 if (nbio) { 552 error = EAGAIN; 553 break; 554 } 555 556 /* 557 * Last chance, interlock with WANTR. 558 */ 559 lwkt_gettoken(&rpipe->pipe_wlock); 560 size = rpipe->pipe_buffer.windex - rpipe->pipe_buffer.rindex; 561 if (size) { 562 lwkt_reltoken(&rpipe->pipe_wlock); 563 continue; 564 } 565 566 /* 567 * Retest EOF - acquiring a new token can temporarily release 568 * tokens already held. 569 */ 570 if (rpipe->pipe_state & PIPE_REOF) { 571 lwkt_reltoken(&rpipe->pipe_wlock); 572 break; 573 } 574 575 /* 576 * If there is no more to read in the pipe, reset its 577 * pointers to the beginning. This improves cache hit 578 * stats. 579 * 580 * We need both locks to modify both pointers, and there 581 * must also not be a write in progress or the uiomove() 582 * in the write might block and temporarily release 583 * its wlock, then reacquire and update windex. We are 584 * only serialized against reads, not writes. 585 * 586 * XXX should we even bother resetting the indices? It 587 * might actually be more cache efficient not to. 588 */ 589 if (rpipe->pipe_buffer.rindex == rpipe->pipe_buffer.windex && 590 rpipe->pipe_wip == 0) { 591 rpipe->pipe_buffer.rindex = 0; 592 rpipe->pipe_buffer.windex = 0; 593 } 594 595 /* 596 * Wait for more data. 597 * 598 * Pipe_state can only be set if both the rlock and wlock 599 * are held. 600 */ 601 rpipe->pipe_state |= PIPE_WANTR; 602 tsleep_interlock(rpipe, PCATCH); 603 lwkt_reltoken(&rpipe->pipe_wlock); 604 error = tsleep(rpipe, PCATCH | PINTERLOCKED, "piperd", 0); 605 ++pipe_rblocked_count; 606 if (error) 607 break; 608 } 609 pipe_end_uio(rpipe, &rpipe->pipe_rip); 610 611 /* 612 * Uptime last access time 613 */ 614 if (error == 0 && nread) 615 vfs_timestamp(&rpipe->pipe_atime); 616 617 /* 618 * If we drained the FIFO more then half way then handle 619 * write blocking hysteresis. 620 * 621 * Note that PIPE_WANTW cannot be set by the writer without 622 * it holding both rlock and wlock, so we can test it 623 * while holding just rlock. 624 */ 625 if (notify_writer) { 626 /* 627 * Synchronous blocking is done on the pipe involved 628 */ 629 if (rpipe->pipe_state & PIPE_WANTW) { 630 lwkt_gettoken(&rpipe->pipe_wlock); 631 if (rpipe->pipe_state & PIPE_WANTW) { 632 rpipe->pipe_state &= ~PIPE_WANTW; 633 lwkt_reltoken(&rpipe->pipe_wlock); 634 wakeup(rpipe); 635 } else { 636 lwkt_reltoken(&rpipe->pipe_wlock); 637 } 638 } 639 640 /* 641 * But we may also have to deal with a kqueue which is 642 * stored on the same pipe as its descriptor, so a 643 * EVFILT_WRITE event waiting for our side to drain will 644 * be on the other side. 645 */ 646 lwkt_gettoken(&wpipe->pipe_wlock); 647 pipewakeup(wpipe, 0); 648 lwkt_reltoken(&wpipe->pipe_wlock); 649 } 650 /*size = rpipe->pipe_buffer.windex - rpipe->pipe_buffer.rindex;*/ 651 lwkt_reltoken(&rpipe->pipe_rlock); 652 653 return (error); 654 } 655 656 static int 657 pipe_write(struct file *fp, struct uio *uio, struct ucred *cred, int fflags) 658 { 659 int error; 660 int orig_resid; 661 int nbio; 662 struct pipe *wpipe; 663 struct pipe *rpipe; 664 u_int windex; 665 u_int space; 666 u_int wcount; 667 int bigwrite; 668 int bigcount; 669 670 /* 671 * Writes go to the peer. The peer will always exist. 672 */ 673 rpipe = (struct pipe *) fp->f_data; 674 wpipe = rpipe->pipe_peer; 675 lwkt_gettoken(&wpipe->pipe_wlock); 676 if (wpipe->pipe_state & PIPE_WEOF) { 677 lwkt_reltoken(&wpipe->pipe_wlock); 678 return (EPIPE); 679 } 680 681 /* 682 * Degenerate case (EPIPE takes prec) 683 */ 684 if (uio->uio_resid == 0) { 685 lwkt_reltoken(&wpipe->pipe_wlock); 686 return(0); 687 } 688 689 /* 690 * Writes are serialized (start_uio must be called with wlock) 691 */ 692 error = pipe_start_uio(wpipe, &wpipe->pipe_wip); 693 if (error) { 694 lwkt_reltoken(&wpipe->pipe_wlock); 695 return (error); 696 } 697 698 if (fflags & O_FBLOCKING) 699 nbio = 0; 700 else if (fflags & O_FNONBLOCKING) 701 nbio = 1; 702 else if (fp->f_flag & O_NONBLOCK) 703 nbio = 1; 704 else 705 nbio = 0; 706 707 /* 708 * If it is advantageous to resize the pipe buffer, do 709 * so. We are write-serialized so we can block safely. 710 */ 711 if ((wpipe->pipe_buffer.size <= PIPE_SIZE) && 712 (pipe_nbig < pipe_maxbig) && 713 wpipe->pipe_wantwcnt > 4 && 714 (wpipe->pipe_buffer.rindex == wpipe->pipe_buffer.windex)) { 715 /* 716 * Recheck after lock. 717 */ 718 lwkt_gettoken(&wpipe->pipe_rlock); 719 if ((wpipe->pipe_buffer.size <= PIPE_SIZE) && 720 (pipe_nbig < pipe_maxbig) && 721 (wpipe->pipe_buffer.rindex == wpipe->pipe_buffer.windex)) { 722 atomic_add_int(&pipe_nbig, 1); 723 if (pipespace(wpipe, BIG_PIPE_SIZE) == 0) 724 ++pipe_bigcount; 725 else 726 atomic_subtract_int(&pipe_nbig, 1); 727 } 728 lwkt_reltoken(&wpipe->pipe_rlock); 729 } 730 731 orig_resid = uio->uio_resid; 732 wcount = 0; 733 734 bigwrite = (uio->uio_resid > 10 * 1024 * 1024); 735 bigcount = 10; 736 737 while (uio->uio_resid) { 738 if (wpipe->pipe_state & PIPE_WEOF) { 739 error = EPIPE; 740 break; 741 } 742 743 /* 744 * Don't hog the cpu. 745 */ 746 if (bigwrite && --bigcount == 0) { 747 lwkt_user_yield(); 748 bigcount = 10; 749 if (CURSIG(curthread->td_lwp)) { 750 error = EINTR; 751 break; 752 } 753 } 754 755 windex = wpipe->pipe_buffer.windex & 756 (wpipe->pipe_buffer.size - 1); 757 space = wpipe->pipe_buffer.size - 758 (wpipe->pipe_buffer.windex - wpipe->pipe_buffer.rindex); 759 cpu_lfence(); 760 761 /* Writes of size <= PIPE_BUF must be atomic. */ 762 if ((space < uio->uio_resid) && (orig_resid <= PIPE_BUF)) 763 space = 0; 764 765 /* 766 * Write to fill, read size handles write hysteresis. Also 767 * additional restrictions can cause select-based non-blocking 768 * writes to spin. 769 */ 770 if (space > 0) { 771 u_int segsize; 772 773 /* 774 * Transfer size is minimum of uio transfer 775 * and free space in pipe buffer. 776 * 777 * Limit each uiocopy to no more then PIPE_SIZE 778 * so we can keep the gravy train going on a 779 * SMP box. This doubles the performance for 780 * write sizes > 16K. Otherwise large writes 781 * wind up doing an inefficient synchronous 782 * ping-pong. 783 */ 784 space = szmin(space, uio->uio_resid); 785 if (space > PIPE_SIZE) 786 space = PIPE_SIZE; 787 788 /* 789 * First segment to transfer is minimum of 790 * transfer size and contiguous space in 791 * pipe buffer. If first segment to transfer 792 * is less than the transfer size, we've got 793 * a wraparound in the buffer. 794 */ 795 segsize = wpipe->pipe_buffer.size - windex; 796 if (segsize > space) 797 segsize = space; 798 799 #ifdef SMP 800 /* 801 * If this is the first loop and the reader is 802 * blocked, do a preemptive wakeup of the reader. 803 * 804 * On SMP the IPI latency plus the wlock interlock 805 * on the reader side is the fastest way to get the 806 * reader going. (The scheduler will hard loop on 807 * lock tokens). 808 * 809 * NOTE: We can't clear WANTR here without acquiring 810 * the rlock, which we don't want to do here! 811 */ 812 if ((wpipe->pipe_state & PIPE_WANTR)) 813 wakeup(wpipe); 814 #endif 815 816 /* 817 * Transfer segment, which may include a wrap-around. 818 * Update windex to account for both all in one go 819 * so the reader can read() the data atomically. 820 */ 821 error = uiomove(&wpipe->pipe_buffer.buffer[windex], 822 segsize, uio); 823 if (error == 0 && segsize < space) { 824 segsize = space - segsize; 825 error = uiomove(&wpipe->pipe_buffer.buffer[0], 826 segsize, uio); 827 } 828 if (error) 829 break; 830 cpu_mfence(); 831 wpipe->pipe_buffer.windex += space; 832 wcount += space; 833 continue; 834 } 835 836 /* 837 * We need both the rlock and the wlock to interlock against 838 * the EOF, WANTW, and size checks, and to modify pipe_state. 839 * 840 * These are token locks so we do not have to worry about 841 * deadlocks. 842 */ 843 lwkt_gettoken(&wpipe->pipe_rlock); 844 845 /* 846 * If the "read-side" has been blocked, wake it up now 847 * and yield to let it drain synchronously rather 848 * then block. 849 */ 850 if (wpipe->pipe_state & PIPE_WANTR) { 851 wpipe->pipe_state &= ~PIPE_WANTR; 852 wakeup(wpipe); 853 } 854 855 /* 856 * don't block on non-blocking I/O 857 */ 858 if (nbio) { 859 lwkt_reltoken(&wpipe->pipe_rlock); 860 error = EAGAIN; 861 break; 862 } 863 864 /* 865 * re-test whether we have to block in the writer after 866 * acquiring both locks, in case the reader opened up 867 * some space. 868 */ 869 space = wpipe->pipe_buffer.size - 870 (wpipe->pipe_buffer.windex - wpipe->pipe_buffer.rindex); 871 cpu_lfence(); 872 if ((space < uio->uio_resid) && (orig_resid <= PIPE_BUF)) 873 space = 0; 874 875 /* 876 * Retest EOF - acquiring a new token can temporarily release 877 * tokens already held. 878 */ 879 if (wpipe->pipe_state & PIPE_WEOF) { 880 lwkt_reltoken(&wpipe->pipe_rlock); 881 error = EPIPE; 882 break; 883 } 884 885 /* 886 * We have no more space and have something to offer, 887 * wake up select/poll/kq. 888 */ 889 if (space == 0) { 890 wpipe->pipe_state |= PIPE_WANTW; 891 ++wpipe->pipe_wantwcnt; 892 pipewakeup(wpipe, 1); 893 if (wpipe->pipe_state & PIPE_WANTW) 894 error = tsleep(wpipe, PCATCH, "pipewr", 0); 895 ++pipe_wblocked_count; 896 } 897 lwkt_reltoken(&wpipe->pipe_rlock); 898 899 /* 900 * Break out if we errored or the read side wants us to go 901 * away. 902 */ 903 if (error) 904 break; 905 if (wpipe->pipe_state & PIPE_WEOF) { 906 error = EPIPE; 907 break; 908 } 909 } 910 pipe_end_uio(wpipe, &wpipe->pipe_wip); 911 912 /* 913 * If we have put any characters in the buffer, we wake up 914 * the reader. 915 * 916 * Both rlock and wlock are required to be able to modify pipe_state. 917 */ 918 if (wpipe->pipe_buffer.windex != wpipe->pipe_buffer.rindex) { 919 if (wpipe->pipe_state & PIPE_WANTR) { 920 lwkt_gettoken(&wpipe->pipe_rlock); 921 if (wpipe->pipe_state & PIPE_WANTR) { 922 wpipe->pipe_state &= ~PIPE_WANTR; 923 lwkt_reltoken(&wpipe->pipe_rlock); 924 wakeup(wpipe); 925 } else { 926 lwkt_reltoken(&wpipe->pipe_rlock); 927 } 928 } 929 lwkt_gettoken(&wpipe->pipe_rlock); 930 pipewakeup(wpipe, 1); 931 lwkt_reltoken(&wpipe->pipe_rlock); 932 } 933 934 /* 935 * Don't return EPIPE if I/O was successful 936 */ 937 if ((wpipe->pipe_buffer.rindex == wpipe->pipe_buffer.windex) && 938 (uio->uio_resid == 0) && 939 (error == EPIPE)) { 940 error = 0; 941 } 942 943 if (error == 0) 944 vfs_timestamp(&wpipe->pipe_mtime); 945 946 /* 947 * We have something to offer, 948 * wake up select/poll/kq. 949 */ 950 /*space = wpipe->pipe_buffer.windex - wpipe->pipe_buffer.rindex;*/ 951 lwkt_reltoken(&wpipe->pipe_wlock); 952 return (error); 953 } 954 955 /* 956 * we implement a very minimal set of ioctls for compatibility with sockets. 957 */ 958 int 959 pipe_ioctl(struct file *fp, u_long cmd, caddr_t data, 960 struct ucred *cred, struct sysmsg *msg) 961 { 962 struct pipe *mpipe; 963 int error; 964 965 mpipe = (struct pipe *)fp->f_data; 966 967 lwkt_gettoken(&mpipe->pipe_rlock); 968 lwkt_gettoken(&mpipe->pipe_wlock); 969 970 switch (cmd) { 971 case FIOASYNC: 972 if (*(int *)data) { 973 mpipe->pipe_state |= PIPE_ASYNC; 974 } else { 975 mpipe->pipe_state &= ~PIPE_ASYNC; 976 } 977 error = 0; 978 break; 979 case FIONREAD: 980 *(int *)data = mpipe->pipe_buffer.windex - 981 mpipe->pipe_buffer.rindex; 982 error = 0; 983 break; 984 case FIOSETOWN: 985 error = fsetown(*(int *)data, &mpipe->pipe_sigio); 986 break; 987 case FIOGETOWN: 988 *(int *)data = fgetown(&mpipe->pipe_sigio); 989 error = 0; 990 break; 991 case TIOCSPGRP: 992 /* This is deprecated, FIOSETOWN should be used instead. */ 993 error = fsetown(-(*(int *)data), &mpipe->pipe_sigio); 994 break; 995 996 case TIOCGPGRP: 997 /* This is deprecated, FIOGETOWN should be used instead. */ 998 *(int *)data = -fgetown(&mpipe->pipe_sigio); 999 error = 0; 1000 break; 1001 default: 1002 error = ENOTTY; 1003 break; 1004 } 1005 lwkt_reltoken(&mpipe->pipe_wlock); 1006 lwkt_reltoken(&mpipe->pipe_rlock); 1007 1008 return (error); 1009 } 1010 1011 /* 1012 * MPSAFE 1013 */ 1014 static int 1015 pipe_stat(struct file *fp, struct stat *ub, struct ucred *cred) 1016 { 1017 struct pipe *pipe; 1018 1019 pipe = (struct pipe *)fp->f_data; 1020 1021 bzero((caddr_t)ub, sizeof(*ub)); 1022 ub->st_mode = S_IFIFO; 1023 ub->st_blksize = pipe->pipe_buffer.size; 1024 ub->st_size = pipe->pipe_buffer.windex - pipe->pipe_buffer.rindex; 1025 ub->st_blocks = (ub->st_size + ub->st_blksize - 1) / ub->st_blksize; 1026 ub->st_atimespec = pipe->pipe_atime; 1027 ub->st_mtimespec = pipe->pipe_mtime; 1028 ub->st_ctimespec = pipe->pipe_ctime; 1029 /* 1030 * Left as 0: st_dev, st_ino, st_nlink, st_uid, st_gid, st_rdev, 1031 * st_flags, st_gen. 1032 * XXX (st_dev, st_ino) should be unique. 1033 */ 1034 return (0); 1035 } 1036 1037 static int 1038 pipe_close(struct file *fp) 1039 { 1040 struct pipe *cpipe; 1041 1042 cpipe = (struct pipe *)fp->f_data; 1043 fp->f_ops = &badfileops; 1044 fp->f_data = NULL; 1045 funsetown(&cpipe->pipe_sigio); 1046 pipeclose(cpipe); 1047 return (0); 1048 } 1049 1050 /* 1051 * Shutdown one or both directions of a full-duplex pipe. 1052 */ 1053 static int 1054 pipe_shutdown(struct file *fp, int how) 1055 { 1056 struct pipe *rpipe; 1057 struct pipe *wpipe; 1058 int error = EPIPE; 1059 1060 rpipe = (struct pipe *)fp->f_data; 1061 wpipe = rpipe->pipe_peer; 1062 1063 /* 1064 * We modify pipe_state on both pipes, which means we need 1065 * all four tokens! 1066 */ 1067 lwkt_gettoken(&rpipe->pipe_rlock); 1068 lwkt_gettoken(&rpipe->pipe_wlock); 1069 lwkt_gettoken(&wpipe->pipe_rlock); 1070 lwkt_gettoken(&wpipe->pipe_wlock); 1071 1072 switch(how) { 1073 case SHUT_RDWR: 1074 case SHUT_RD: 1075 rpipe->pipe_state |= PIPE_REOF; /* my reads */ 1076 rpipe->pipe_state |= PIPE_WEOF; /* peer writes */ 1077 if (rpipe->pipe_state & PIPE_WANTR) { 1078 rpipe->pipe_state &= ~PIPE_WANTR; 1079 wakeup(rpipe); 1080 } 1081 if (rpipe->pipe_state & PIPE_WANTW) { 1082 rpipe->pipe_state &= ~PIPE_WANTW; 1083 wakeup(rpipe); 1084 } 1085 error = 0; 1086 if (how == SHUT_RD) 1087 break; 1088 /* fall through */ 1089 case SHUT_WR: 1090 wpipe->pipe_state |= PIPE_REOF; /* peer reads */ 1091 wpipe->pipe_state |= PIPE_WEOF; /* my writes */ 1092 if (wpipe->pipe_state & PIPE_WANTR) { 1093 wpipe->pipe_state &= ~PIPE_WANTR; 1094 wakeup(wpipe); 1095 } 1096 if (wpipe->pipe_state & PIPE_WANTW) { 1097 wpipe->pipe_state &= ~PIPE_WANTW; 1098 wakeup(wpipe); 1099 } 1100 error = 0; 1101 break; 1102 } 1103 pipewakeup(rpipe, 1); 1104 pipewakeup(wpipe, 1); 1105 1106 lwkt_reltoken(&wpipe->pipe_wlock); 1107 lwkt_reltoken(&wpipe->pipe_rlock); 1108 lwkt_reltoken(&rpipe->pipe_wlock); 1109 lwkt_reltoken(&rpipe->pipe_rlock); 1110 1111 return (error); 1112 } 1113 1114 static void 1115 pipe_free_kmem(struct pipe *cpipe) 1116 { 1117 if (cpipe->pipe_buffer.buffer != NULL) { 1118 if (cpipe->pipe_buffer.size > PIPE_SIZE) 1119 atomic_subtract_int(&pipe_nbig, 1); 1120 kmem_free(&kernel_map, 1121 (vm_offset_t)cpipe->pipe_buffer.buffer, 1122 cpipe->pipe_buffer.size); 1123 cpipe->pipe_buffer.buffer = NULL; 1124 cpipe->pipe_buffer.object = NULL; 1125 } 1126 } 1127 1128 /* 1129 * Close the pipe. The slock must be held to interlock against simultanious 1130 * closes. The rlock and wlock must be held to adjust the pipe_state. 1131 */ 1132 static void 1133 pipeclose(struct pipe *cpipe) 1134 { 1135 globaldata_t gd; 1136 struct pipe *ppipe; 1137 1138 if (cpipe == NULL) 1139 return; 1140 1141 /* 1142 * The slock may not have been allocated yet (close during 1143 * initialization) 1144 * 1145 * We need both the read and write tokens to modify pipe_state. 1146 */ 1147 if (cpipe->pipe_slock) 1148 lockmgr(cpipe->pipe_slock, LK_EXCLUSIVE); 1149 lwkt_gettoken(&cpipe->pipe_rlock); 1150 lwkt_gettoken(&cpipe->pipe_wlock); 1151 1152 /* 1153 * Set our state, wakeup anyone waiting in select/poll/kq, and 1154 * wakeup anyone blocked on our pipe. 1155 */ 1156 cpipe->pipe_state |= PIPE_CLOSED | PIPE_REOF | PIPE_WEOF; 1157 pipewakeup(cpipe, 1); 1158 if (cpipe->pipe_state & (PIPE_WANTR | PIPE_WANTW)) { 1159 cpipe->pipe_state &= ~(PIPE_WANTR | PIPE_WANTW); 1160 wakeup(cpipe); 1161 } 1162 1163 /* 1164 * Disconnect from peer. 1165 */ 1166 if ((ppipe = cpipe->pipe_peer) != NULL) { 1167 lwkt_gettoken(&ppipe->pipe_rlock); 1168 lwkt_gettoken(&ppipe->pipe_wlock); 1169 ppipe->pipe_state |= PIPE_REOF | PIPE_WEOF; 1170 pipewakeup(ppipe, 1); 1171 if (ppipe->pipe_state & (PIPE_WANTR | PIPE_WANTW)) { 1172 ppipe->pipe_state &= ~(PIPE_WANTR | PIPE_WANTW); 1173 wakeup(ppipe); 1174 } 1175 if (SLIST_FIRST(&ppipe->pipe_kq.ki_note)) 1176 KNOTE(&ppipe->pipe_kq.ki_note, 0); 1177 lwkt_reltoken(&ppipe->pipe_wlock); 1178 lwkt_reltoken(&ppipe->pipe_rlock); 1179 } 1180 1181 /* 1182 * If the peer is also closed we can free resources for both 1183 * sides, otherwise we leave our side intact to deal with any 1184 * races (since we only have the slock). 1185 */ 1186 if (ppipe && (ppipe->pipe_state & PIPE_CLOSED)) { 1187 cpipe->pipe_peer = NULL; 1188 ppipe->pipe_peer = NULL; 1189 ppipe->pipe_slock = NULL; /* we will free the slock */ 1190 pipeclose(ppipe); 1191 ppipe = NULL; 1192 } 1193 1194 lwkt_reltoken(&cpipe->pipe_wlock); 1195 lwkt_reltoken(&cpipe->pipe_rlock); 1196 if (cpipe->pipe_slock) 1197 lockmgr(cpipe->pipe_slock, LK_RELEASE); 1198 1199 /* 1200 * If we disassociated from our peer we can free resources 1201 */ 1202 if (ppipe == NULL) { 1203 gd = mycpu; 1204 if (cpipe->pipe_slock) { 1205 kfree(cpipe->pipe_slock, M_PIPE); 1206 cpipe->pipe_slock = NULL; 1207 } 1208 if (gd->gd_pipeqcount >= pipe_maxcache || 1209 cpipe->pipe_buffer.size != PIPE_SIZE 1210 ) { 1211 pipe_free_kmem(cpipe); 1212 kfree(cpipe, M_PIPE); 1213 } else { 1214 cpipe->pipe_state = 0; 1215 cpipe->pipe_peer = gd->gd_pipeq; 1216 gd->gd_pipeq = cpipe; 1217 ++gd->gd_pipeqcount; 1218 } 1219 } 1220 } 1221 1222 static int 1223 pipe_kqfilter(struct file *fp, struct knote *kn) 1224 { 1225 struct pipe *cpipe; 1226 1227 cpipe = (struct pipe *)kn->kn_fp->f_data; 1228 1229 switch (kn->kn_filter) { 1230 case EVFILT_READ: 1231 kn->kn_fop = &pipe_rfiltops; 1232 break; 1233 case EVFILT_WRITE: 1234 kn->kn_fop = &pipe_wfiltops; 1235 if (cpipe->pipe_peer == NULL) { 1236 /* other end of pipe has been closed */ 1237 return (EPIPE); 1238 } 1239 break; 1240 default: 1241 return (EOPNOTSUPP); 1242 } 1243 kn->kn_hook = (caddr_t)cpipe; 1244 1245 knote_insert(&cpipe->pipe_kq.ki_note, kn); 1246 1247 return (0); 1248 } 1249 1250 static void 1251 filt_pipedetach(struct knote *kn) 1252 { 1253 struct pipe *cpipe = (struct pipe *)kn->kn_hook; 1254 1255 knote_remove(&cpipe->pipe_kq.ki_note, kn); 1256 } 1257 1258 /*ARGSUSED*/ 1259 static int 1260 filt_piperead(struct knote *kn, long hint) 1261 { 1262 struct pipe *rpipe = (struct pipe *)kn->kn_fp->f_data; 1263 int ready = 0; 1264 1265 lwkt_gettoken(&rpipe->pipe_rlock); 1266 lwkt_gettoken(&rpipe->pipe_wlock); 1267 1268 kn->kn_data = rpipe->pipe_buffer.windex - rpipe->pipe_buffer.rindex; 1269 1270 if (rpipe->pipe_state & PIPE_REOF) { 1271 /* 1272 * Only set NODATA if all data has been exhausted 1273 */ 1274 if (kn->kn_data == 0) 1275 kn->kn_flags |= EV_NODATA; 1276 kn->kn_flags |= EV_EOF; 1277 ready = 1; 1278 } 1279 1280 lwkt_reltoken(&rpipe->pipe_wlock); 1281 lwkt_reltoken(&rpipe->pipe_rlock); 1282 1283 if (!ready) 1284 ready = kn->kn_data > 0; 1285 1286 return (ready); 1287 } 1288 1289 /*ARGSUSED*/ 1290 static int 1291 filt_pipewrite(struct knote *kn, long hint) 1292 { 1293 struct pipe *rpipe = (struct pipe *)kn->kn_fp->f_data; 1294 struct pipe *wpipe = rpipe->pipe_peer; 1295 int ready = 0; 1296 1297 kn->kn_data = 0; 1298 if (wpipe == NULL) { 1299 kn->kn_flags |= (EV_EOF | EV_NODATA); 1300 return (1); 1301 } 1302 1303 lwkt_gettoken(&wpipe->pipe_rlock); 1304 lwkt_gettoken(&wpipe->pipe_wlock); 1305 1306 if (wpipe->pipe_state & PIPE_WEOF) { 1307 kn->kn_flags |= (EV_EOF | EV_NODATA); 1308 ready = 1; 1309 } 1310 1311 if (!ready) 1312 kn->kn_data = wpipe->pipe_buffer.size - 1313 (wpipe->pipe_buffer.windex - 1314 wpipe->pipe_buffer.rindex); 1315 1316 lwkt_reltoken(&wpipe->pipe_wlock); 1317 lwkt_reltoken(&wpipe->pipe_rlock); 1318 1319 if (!ready) 1320 ready = kn->kn_data >= PIPE_BUF; 1321 1322 return (ready); 1323 } 1324