1 /* 2 * CDDL HEADER START 3 * 4 * The contents of this file are subject to the terms of the 5 * Common Development and Distribution License (the "License"). 6 * You may not use this file except in compliance with the License. 7 * 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9 * or http://www.opensolaris.org/os/licensing. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 22 /* 23 * Copyright 2008 Sun Microsystems, Inc. All rights reserved. 24 * Use is subject to license terms. 25 */ 26 27 #ifndef _LDC_IMPL_H 28 #define _LDC_IMPL_H 29 30 #pragma ident "%Z%%M% %I% %E% SMI" 31 32 #ifdef __cplusplus 33 extern "C" { 34 #endif 35 36 #include <sys/types.h> 37 #include <sys/ddi.h> 38 #include <sys/sunddi.h> 39 #include <sys/ioctl.h> 40 41 /* Memory map table entries */ 42 #define LDC_MTBL_ENTRIES 8192 /* 8 K */ 43 44 /* Define LDC Queue info */ 45 #define LDC_PACKET_SHIFT 6 46 #define LDC_QUEUE_ENTRIES 512 47 #define LDC_MTU_MSGS 4 48 #define LDC_QUEUE_SIZE (LDC_QUEUE_ENTRIES << LDC_PACKET_SHIFT) 49 #define LDC_DEFAULT_MTU (LDC_QUEUE_SIZE / LDC_MTU_MSGS) 50 #define LDC_RXDQ_MULTIPLIER 2 51 52 /* 53 * LDC Reliable mode - initial packet seqid 54 * - If peer initiated handshake, RDX should contain init_seqid + 1 55 * - If this endpoint initiated handshake first data packet should 56 * contain the message init_seqid + 1 57 */ 58 #define LDC_INIT_SEQID 0x0 59 60 /* LDC Message types */ 61 #define LDC_CTRL 0x01 /* Control Pkt */ 62 #define LDC_DATA 0x02 /* Data Pkt */ 63 #define LDC_ERR 0x10 /* Error Pkt */ 64 65 /* LDC Message Subtypes */ 66 #define LDC_INFO 0x01 /* Control/Data/Error info pkt */ 67 #define LDC_ACK 0x02 /* Control/Data ACK */ 68 #define LDC_NACK 0x04 /* Control/Data NACK */ 69 70 /* LDC Control Messages */ 71 #define LDC_VER 0x01 /* Version message */ 72 #define LDC_RTS 0x02 /* Request to Send */ 73 #define LDC_RTR 0x03 /* Ready To Receive */ 74 #define LDC_RDX 0x04 /* Ready for data exchange */ 75 76 #define LDC_CTRL_MASK 0x0f /* Mask to read control bits */ 77 78 /* LDC Channel Transport State (tstate) */ 79 #define TS_TXQ_RDY 0x01 /* allocated TX queue */ 80 #define TS_RXQ_RDY 0x02 /* allocated RX queue */ 81 #define TS_INIT (TS_TXQ_RDY | TS_RXQ_RDY) 82 #define TS_QCONF_RDY 0x04 /* registered queues with HV */ 83 #define TS_CNEX_RDY 0x08 /* registered channel with cnex */ 84 #define TS_OPEN (TS_INIT | TS_QCONF_RDY | TS_CNEX_RDY) 85 #define TS_LINK_READY 0x10 /* both endpts registered Rx queues */ 86 #define TS_READY (TS_OPEN | TS_LINK_READY) 87 #define TS_VER_DONE 0x20 /* negotiated version */ 88 #define TS_VREADY (TS_READY | TS_VER_DONE) 89 #define TS_HSHAKE_DONE 0x40 /* completed handshake */ 90 #define TS_UP (TS_READY | TS_VER_DONE | TS_HSHAKE_DONE) 91 92 #define TS_IN_RESET 0x100 /* channel is in reset state */ 93 94 /* LDC Channel Transport Handshake states */ 95 #define TS_SENT_VER 0x01 /* Sent version */ 96 #define TS_SENT_RTS 0x02 /* Sent RTS */ 97 #define TS_RCVD_RTR 0x04 /* Received RTR */ 98 #define TS_SENT_RDX 0x08 /* Sent RDX */ 99 #define TS_RCVD_VER 0x10 /* Received version */ 100 #define TS_RCVD_RTS 0x20 /* Received RTS */ 101 #define TS_SENT_RTR 0x40 /* Sent RTR */ 102 #define TS_RCVD_RDX 0x80 /* Received RDX */ 103 104 /* LDC Interrupt State */ 105 #define LDC_INTR_NONE 0x00 /* No interrupts */ 106 #define LDC_INTR_ACTIVE 0x01 /* Interrupt being processed */ 107 #define LDC_INTR_PEND 0x02 /* Interrupt pending */ 108 109 /* LDC MSG Envelope */ 110 #define LDC_LEN_MASK 0x3F 111 #define LDC_FRAG_MASK 0xC0 112 113 #define LDC_FRAG_START 0x40 /* frag_info = 0x01 */ 114 #define LDC_FRAG_STOP 0x80 /* frag_info = 0x02 */ 115 #define LDC_FRAG_CONT 0x00 /* frag_info = 0x00 */ 116 117 /* 118 * LDC will retry LDC_MAX_RETRIES times when sending or 119 * receiving data or if the HV returns back EWOULDBLOCK. 120 * Between each retry it will wait LDC_DELAY usecs. 121 */ 122 #define LDC_MAX_RETRIES 1000 123 #define LDC_DELAY 1 124 125 /* delay(usec) between channel unregister retries in ldc_close() */ 126 #define LDC_CLOSE_DELAY 1 127 128 /* 129 * LDC Version information 130 */ 131 #define LDC_PAYLOAD_VER_OFF 8 /* offset of version in payload */ 132 133 typedef struct ldc_ver { 134 uint16_t major; 135 uint16_t minor; 136 } ldc_ver_t; 137 138 /* 139 * Each guest consists of one or more LDC endpoints represented by a ldc_chan 140 * structure. Each ldc_chan structure points to a ldc_mtbl structure that 141 * contains information about the map table associated with this LDC endpoint. 142 * The map table contains the list of pages being shared by this guest over 143 * this endpoint with the guest at the other end of this endpoint. Each LDC 144 * endpoint also points to a list of memory handles used to bind and export 145 * memory segments from this guest. If a memory segment is bound, it points to 146 * a memory segment structure, which inturn consists of an array of ldc_page 147 * structure for all the pages within that segment. Each ldc_page structure 148 * contains information about the shared page and also points to the 149 * corresponding entry in the map table. 150 * 151 * Each LDC endpoint also points to a list of ldc_dring structures that refer 152 * to both imported and exported descriptor rings. If it is a exported 153 * descriptor ring, it then points to memory handle/memseg corresponding to 154 * the region of memory associated with the descriptor ring. 155 * 156 * +----------+ +----------+ +----------+ 157 * | ldc_chan |-->| ldc_chan |-->| ldc_chan |-->.... 158 * +----------+ +----------+ +----------+ 159 * | | | 160 * | | | 161 * | | | +-----------+ +-----------+ 162 * | | +----->| ldc_dring |---->| ldc_dring |---->...... 163 * | | +-----------+ +-----------+ 164 * | | | 165 * | | +----------------------------+ 166 * | | | 167 * | | v 168 * | | +----------+ +----------+ +----------+ 169 * | +----->| ldc_mhdl |---->| ldc_mhdl |---->| ldc_mhdl |---> .... 170 * | +----------+ +----------+ +----------+ 171 * v | | 172 * +----------+ | +------------+ | +------------+ 173 * | ldc_mtbl |--+ +--->| ldc_memseg |-----+ +--->| ldc_memseg | 174 * +----------+ | +------------+ | +------------+ 175 * | | | | | 176 * v v v | v 177 * +--------------+ +----------+ +--------+ | +--------+ 178 * | ldc_mte_slot |<--------| ldc_page | | cookie | | | cookie | 179 * +--------------+ +----------+ +--------+ | +--------+ 180 * | ldc_mte_slot |<--------| ldc_page | | cookie | v 181 * +--------------+ +----------+ +--------+ +----------+ 182 * | ldc_mte_slot |<-----------------------------------| ldc_page | 183 * +--------------+ +----------+ 184 * | ldc_mte_slot | 185 * +--------------+ 186 * | ...... |/ +------------+ 187 * +--------------+ | entry | 188 * | ldc_mte_slot | +------------+ 189 * +--------------+ | inv_cookie | 190 * \ +------------+ 191 * 192 */ 193 194 /* 195 * Message format of each packet sent over the LDC channel. 196 * Each packet is 64-bytes long. 197 * 198 * Each packet that is sent over LDC can contain either data or acks. 199 * The type will reflect the contents. The len will contain in bytes 200 * the amount of data being sent. In the case of ACKs, the seqid and 201 * data fields will contain the SEQIDs of messages for which ACKs are 202 * being sent. 203 * 204 * Raw pkt format: 205 * 206 * +------------------------------------------------------+ 207 * 0 - 7 | data payload | 208 * +------------------------------------------------------+ 209 * 210 * Unreliable pkt format: 211 * 212 * +------------------------------------------------------+ 213 * 0 | seqid | env | ctrl | stype | type | 214 * +------------------------------------------------------+ 215 * 1 - 7 | data payload | 216 * +------------------------------------------------------+ 217 * 218 * Reliable pkt format: 219 * 220 * +------------------------------------------------------+ 221 * 0 | seqid | env | ctrl | stype | type | 222 * +------------------------------------------------------+ 223 * 1 | ackid | unused | 224 * +------------------------------------------------------+ 225 * 2 - 7 | data payload | 226 * +------------------------------------------------------+ 227 */ 228 229 typedef struct ldc_msg { 230 union { 231 struct { 232 uint8_t _type; /* Message type */ 233 uint8_t _stype; /* Message subtype */ 234 uint8_t _ctrl; /* Control/Error Message */ 235 uint8_t _env; /* Message Envelope */ 236 uint32_t _seqid; /* Sequence ID */ 237 238 union { 239 uint8_t _ud[LDC_PAYLOAD_SIZE_UNRELIABLE]; 240 /* Unreliable data payload */ 241 struct { 242 uint32_t _unused; /* unused */ 243 uint32_t _ackid; /* ACK ID */ 244 uint8_t _rd[LDC_PAYLOAD_SIZE_RELIABLE]; 245 /* Reliable data payload */ 246 } _rl; 247 } _data; 248 } _tpkt; 249 250 uint8_t _raw[LDC_PAYLOAD_SIZE_RAW]; 251 } _pkt; 252 253 } ldc_msg_t; 254 255 #define raw _pkt._raw 256 #define type _pkt._tpkt._type 257 #define stype _pkt._tpkt._stype 258 #define ctrl _pkt._tpkt._ctrl 259 #define env _pkt._tpkt._env 260 #define seqid _pkt._tpkt._seqid 261 #define udata _pkt._tpkt._data._ud 262 #define ackid _pkt._tpkt._data._rl._ackid 263 #define rdata _pkt._tpkt._data._rl._rd 264 265 /* 266 * LDC Map Table Entry (MTE) 267 * 268 * 6 6 1 1 1 269 * |3 0| psz| 3| 1| 0| 9| 8| 7|6|5|4| 0| 270 * +------+--------------------------+----+----+--+--+--+--+-+-+-+-------+ 271 * | rsvd | PFN | 0 | 0 |CW|CR|IW|IR|X|W|R| pgszc | 272 * +------+--------------------------+----+----+--+--+--+--+-+-+-+-------+ 273 * | hv invalidation cookie | 274 * +---------------------------------------------------------------------+ 275 */ 276 typedef union { 277 struct { 278 uint64_t _rsvd2:8, /* <63:56> reserved */ 279 rpfn:43, /* <55:13> real pfn */ 280 _rsvd1:2, /* <12:11> reserved */ 281 cw:1, /* <10> copy write access */ 282 cr:1, /* <9> copy read perm */ 283 iw:1, /* <8> iommu write perm */ 284 ir:1, /* <7> iommu read perm */ 285 x:1, /* <6> execute perm */ 286 w:1, /* <5> write perm */ 287 r:1, /* <4> read perm */ 288 pgszc:4; /* <3:0> pgsz code */ 289 } mte_bit; 290 291 uint64_t ll; 292 293 } ldc_mte_t; 294 295 #define mte_rpfn mte_bit.rpfn 296 #define mte_cw mte_bit.cw 297 #define mte_cr mte_bit.cr 298 #define mte_iw mte_bit.iw 299 #define mte_ir mte_bit.ir 300 #define mte_x mte_bit.x 301 #define mte_w mte_bit.w 302 #define mte_r mte_bit.r 303 #define mte_pgszc mte_bit.pgszc 304 305 #define MTE_BSZS_SHIFT(sz) ((sz) * 3) 306 #define MTEBYTES(sz) (MMU_PAGESIZE << MTE_BSZS_SHIFT(sz)) 307 #define MTEPAGES(sz) (1 << MTE_BSZS_SHIFT(sz)) 308 #define MTE_PAGE_SHIFT(sz) (MMU_PAGESHIFT + MTE_BSZS_SHIFT(sz)) 309 #define MTE_PAGE_OFFSET(sz) (MTEBYTES(sz) - 1) 310 #define MTE_PAGEMASK(sz) (~MTE_PAGE_OFFSET(sz)) 311 #define MTE_PFNMASK(sz) (~(MTE_PAGE_OFFSET(sz) >> MMU_PAGESHIFT)) 312 313 /* 314 * LDC Map Table Slot 315 */ 316 typedef struct ldc_mte_slot { 317 ldc_mte_t entry; 318 uint64_t cookie; 319 } ldc_mte_slot_t; 320 321 /* 322 * LDC Memory Map Table 323 * 324 * Each LDC has a memory map table it uses to list all the pages 325 * it exporting to its peer over the channel. This structure 326 * contains information about the map table and is pointed to 327 * by the ldc_chan structure. 328 */ 329 typedef struct ldc_mtbl { 330 kmutex_t lock; /* Table lock */ 331 size_t size; /* Table size (in bytes) */ 332 uint64_t next_entry; /* Next entry to use */ 333 uint64_t num_entries; /* Num entries in table */ 334 uint64_t num_avail; /* Num of available entries */ 335 boolean_t contigmem; /* TRUE=Contig mem alloc'd */ 336 ldc_mte_slot_t *table; /* The table itself */ 337 } ldc_mtbl_t; 338 339 /* 340 * LDC page and memory segment information 341 */ 342 typedef struct ldc_page { 343 uintptr_t raddr; /* Exported page RA */ 344 uint64_t index; /* Index in map table */ 345 ldc_mte_slot_t *mte; /* Map table entry */ 346 } ldc_page_t; 347 348 typedef struct ldc_memseg { 349 caddr_t vaddr; /* Exported segment VA */ 350 uintptr_t raddr; /* Exported segment VA */ 351 size_t size; /* Exported segment size */ 352 uint64_t npages; /* Number of pages */ 353 ldc_page_t *pages; /* Array of exported pages */ 354 uint32_t ncookies; /* Number of cookies */ 355 ldc_mem_cookie_t *cookies; 356 uint64_t next_cookie; /* Index to next cookie */ 357 } ldc_memseg_t; 358 359 /* 360 * LDC Cookie address format 361 * 362 * 6 6 m+n 363 * |3| 0| | m| 0| 364 * +-+-------+----------+-------------------+-------------------+ 365 * |X| pgszc | rsvd | table_idx | page_offset | 366 * +-+-------+----------+-------------------+-------------------+ 367 */ 368 #define LDC_COOKIE_PGSZC_MASK 0x7 369 #define LDC_COOKIE_PGSZC_SHIFT 60 370 371 /* 372 * LDC Memory handle 373 */ 374 typedef struct ldc_chan ldc_chan_t; 375 376 typedef struct ldc_mhdl { 377 kmutex_t lock; /* Mutex for memory handle */ 378 ldc_mstatus_t status; /* Memory map status */ 379 380 uint8_t mtype; /* Type of sharing */ 381 uint8_t perm; /* Access permissions */ 382 boolean_t myshadow; /* TRUE=alloc'd shadow mem */ 383 384 ldc_chan_t *ldcp; /* Pointer to channel struct */ 385 ldc_memseg_t *memseg; /* Bound memory segment */ 386 struct ldc_mhdl *next; /* Next memory handle */ 387 } ldc_mhdl_t; 388 389 /* 390 * LDC Descriptor rings 391 */ 392 393 typedef struct ldc_dring { 394 kmutex_t lock; /* Desc ring lock */ 395 ldc_mstatus_t status; /* Desc ring status */ 396 397 uint32_t dsize; /* Descriptor size */ 398 uint32_t length; /* Descriptor ring length */ 399 uint64_t size; /* Desc ring size (in bytes) */ 400 caddr_t base; /* Descriptor ring base addr */ 401 402 ldc_chan_t *ldcp; /* Pointer to bound channel */ 403 ldc_mem_handle_t mhdl; /* Mem handle to desc ring */ 404 405 struct ldc_dring *ch_next; /* Next dring in channel */ 406 struct ldc_dring *next; /* Next dring overall */ 407 408 } ldc_dring_t; 409 410 411 /* 412 * Channel specific information is kept in a separate 413 * structure. These are then stored on a array indexed 414 * by the channel number. 415 */ 416 struct ldc_chan { 417 ldc_chan_t *next; /* Next channel */ 418 419 kmutex_t lock; /* Channel lock */ 420 uint64_t id; /* Channel ID */ 421 ldc_status_t status; /* Channel status */ 422 uint32_t tstate; /* Channel transport state */ 423 uint32_t hstate; /* Channel transport handshake state */ 424 425 ldc_dev_t devclass; /* Associated device class */ 426 uint64_t devinst; /* Associated device instance */ 427 ldc_mode_t mode; /* Channel mode */ 428 429 uint64_t mtu; /* Max TU size */ 430 431 ldc_ver_t version; /* Channel version */ 432 uint32_t next_vidx; /* Next version to match */ 433 434 uint_t (*cb)(uint64_t event, caddr_t arg); 435 caddr_t cb_arg; /* Channel callback and arg */ 436 boolean_t cb_inprogress; /* Channel callback in progress */ 437 boolean_t cb_enabled; /* Channel callbacks are enabled */ 438 439 uint8_t tx_intr_state; /* Tx interrupt state */ 440 uint8_t rx_intr_state; /* Rx interrupt state */ 441 442 kmutex_t tx_lock; /* Transmit lock */ 443 uint64_t tx_q_entries; /* Num entries in transmit queue */ 444 uint64_t tx_q_va; /* Virtual addr of transmit queue */ 445 uint64_t tx_q_ra; /* Real addr of transmit queue */ 446 uint64_t tx_head; /* Tx queue head */ 447 uint64_t tx_ackd_head; /* Tx queue ACKd head (Reliable) */ 448 uint64_t tx_tail; /* Tx queue tail */ 449 450 uint64_t rx_q_entries; /* Num entries in receive queue */ 451 uint64_t rx_q_va; /* Virtual addr of receive queue */ 452 uint64_t rx_q_ra; /* Real addr of receive queue */ 453 454 uint64_t rx_dq_entries; /* Num entries in the data queue */ 455 uint64_t rx_dq_va; /* Virtual addr of the data queue */ 456 uint64_t rx_dq_head; /* Receive data queue head */ 457 uint64_t rx_dq_tail; /* Receive data queue tail */ 458 uint64_t rx_ack_head; /* Receive data ACK peek head ptr */ 459 460 uint64_t link_state; /* Underlying HV channel state */ 461 462 ldc_mtbl_t *mtbl; /* Memory table used by channel */ 463 ldc_mhdl_t *mhdl_list; /* List of memory handles */ 464 kmutex_t mlist_lock; /* Mem handle list lock */ 465 466 ldc_dring_t *exp_dring_list; /* Exported desc ring list */ 467 kmutex_t exp_dlist_lock; /* Lock for exported desc ring list */ 468 ldc_dring_t *imp_dring_list; /* Imported desc ring list */ 469 kmutex_t imp_dlist_lock; /* Lock for imported desc ring list */ 470 471 uint8_t pkt_payload; /* Size of packet payload */ 472 473 uint32_t last_msg_snt; /* Seqid of last packet sent */ 474 uint32_t last_ack_rcd; /* Seqid of last ACK recd */ 475 uint32_t last_msg_rcd; /* Seqid of last packet received */ 476 477 uint32_t stream_remains; /* Number of bytes in stream */ 478 /* packet buffer */ 479 uint32_t stream_offset; /* Offset into packet buffer for */ 480 /* next read */ 481 uint8_t *stream_bufferp; /* Stream packet buffer */ 482 483 int (*read_p)(ldc_chan_t *ldcp, caddr_t bufferp, 484 size_t *sizep); 485 int (*write_p)(ldc_chan_t *ldcp, caddr_t bufferp, 486 size_t *sizep); 487 488 uint64_t (*readq_get_state)(ldc_chan_t *ldcp, uint64_t *head, 489 uint64_t *tail, uint64_t *link_state); 490 491 int (*readq_set_head)(ldc_chan_t *ldcp, uint64_t head); 492 }; 493 494 495 /* 496 * LDC module soft state structure 497 */ 498 typedef struct ldc_soft_state { 499 kmutex_t lock; /* Protects ldc_soft_state_t */ 500 ldc_cnex_t cinfo; /* channel nexus info */ 501 uint64_t channel_count; /* Number of channels */ 502 uint64_t channels_open; /* Number of open channels */ 503 ldc_chan_t *chan_list; /* List of LDC endpoints */ 504 ldc_dring_t *dring_list; /* Descriptor rings (for export) */ 505 506 kmem_cache_t *memhdl_cache; /* Memory handle cache */ 507 kmem_cache_t *memseg_cache; /* Memory segment cache */ 508 } ldc_soft_state_t; 509 510 511 /* 512 * Debugging Utilities 513 */ 514 #define DBG_ALL_LDCS -1 515 #ifdef DEBUG 516 #define D1 \ 517 if (ldcdbg & 0x01) \ 518 ldcdebug 519 #define D2 \ 520 if (ldcdbg & 0x02) \ 521 ldcdebug 522 #define DWARN \ 523 if (ldcdbg & 0x04) \ 524 ldcdebug 525 #else 526 #define D1 527 #define D2 528 #define DWARN 529 #endif 530 531 #ifdef __cplusplus 532 } 533 #endif 534 535 #endif /* _LDC_IMPL_H */ 536