2 * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved.
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
17 * - Redistributions in binary form must reproduce the above
18 * copyright notice, this list of conditions and the following
19 * disclaimer in the documentation and/or other materials
20 * provided with the distribution.
22 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
23 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
24 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
25 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
26 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
27 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
28 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
37 #include "t4fw_ri_api.h"
39 #define T4_MAX_NUM_QP (1<<16)
40 #define T4_MAX_NUM_CQ (1<<15)
41 #define T4_MAX_NUM_PD (1<<15)
42 #define T4_EQ_STATUS_ENTRIES (L1_CACHE_BYTES > 64 ? 2 : 1)
43 #define T4_MAX_EQ_SIZE (65520 - T4_EQ_STATUS_ENTRIES)
44 #define T4_MAX_IQ_SIZE (65520 - 1)
45 #define T4_MAX_RQ_SIZE (8192 - T4_EQ_STATUS_ENTRIES)
46 #define T4_MAX_SQ_SIZE (T4_MAX_EQ_SIZE - 1)
47 #define T4_MAX_QP_DEPTH (T4_MAX_RQ_SIZE - 1)
48 #define T4_MAX_CQ_DEPTH (T4_MAX_IQ_SIZE - 1)
49 #define T4_MAX_NUM_STAG (1<<15)
50 #define T4_MAX_MR_SIZE (~0ULL - 1)
51 #define T4_PAGESIZE_MASK 0xffff000 /* 4KB-128MB */
52 #define T4_STAG_UNSET 0xffffffff
54 #define T4_EQ_STATUS_ENTRIES (L1_CACHE_BYTES > 64 ? 2 : 1)
55 #define A_PCIE_MA_SYNC 0x30b4
57 struct t4_status_page
{
58 __be32 rsvd1
; /* flit 0 - hw owns */
63 u8 qp_err
; /* flit 1 - sw owns */
67 #define T4_EQ_ENTRY_SIZE 64
69 #define T4_SQ_NUM_SLOTS 5
70 #define T4_SQ_NUM_BYTES (T4_EQ_ENTRY_SIZE * T4_SQ_NUM_SLOTS)
71 #define T4_MAX_SEND_SGE ((T4_SQ_NUM_BYTES - sizeof(struct fw_ri_send_wr) - \
72 sizeof(struct fw_ri_isgl)) / sizeof(struct fw_ri_sge))
73 #define T4_MAX_SEND_INLINE ((T4_SQ_NUM_BYTES - sizeof(struct fw_ri_send_wr) - \
74 sizeof(struct fw_ri_immd)))
75 #define T4_MAX_WRITE_INLINE ((T4_SQ_NUM_BYTES - \
76 sizeof(struct fw_ri_rdma_write_wr) - \
77 sizeof(struct fw_ri_immd)))
78 #define T4_MAX_WRITE_SGE ((T4_SQ_NUM_BYTES - \
79 sizeof(struct fw_ri_rdma_write_wr) - \
80 sizeof(struct fw_ri_isgl)) / sizeof(struct fw_ri_sge))
81 #define T4_MAX_FR_IMMD ((T4_SQ_NUM_BYTES - sizeof(struct fw_ri_fr_nsmr_wr) - \
82 sizeof(struct fw_ri_immd)) & ~31UL)
83 #define T4_MAX_FR_DEPTH (T4_MAX_FR_IMMD / sizeof(u64))
85 #define T4_RQ_NUM_SLOTS 2
86 #define T4_RQ_NUM_BYTES (T4_EQ_ENTRY_SIZE * T4_RQ_NUM_SLOTS)
87 #define T4_MAX_RECV_SGE 4
90 struct fw_ri_res_wr res
;
92 struct fw_ri_rdma_write_wr write
;
93 struct fw_ri_send_wr send
;
94 struct fw_ri_rdma_read_wr read
;
95 struct fw_ri_bind_mw_wr bind
;
96 struct fw_ri_fr_nsmr_wr fr
;
97 struct fw_ri_inv_lstag_wr inv
;
98 struct t4_status_page status
;
99 __be64 flits
[T4_EQ_ENTRY_SIZE
/ sizeof(__be64
) * T4_SQ_NUM_SLOTS
];
103 struct fw_ri_recv_wr recv
;
104 struct t4_status_page status
;
105 __be64 flits
[T4_EQ_ENTRY_SIZE
/ sizeof(__be64
) * T4_RQ_NUM_SLOTS
];
108 static inline void init_wr_hdr(union t4_wr
*wqe
, u16 wrid
,
109 enum fw_wr_opcodes opcode
, u8 flags
, u8 len16
)
111 wqe
->send
.opcode
= (u8
)opcode
;
112 wqe
->send
.flags
= flags
;
113 wqe
->send
.wrid
= wrid
;
117 wqe
->send
.len16
= len16
;
120 /* CQE/AE status codes */
121 #define T4_ERR_SUCCESS 0x0
122 #define T4_ERR_STAG 0x1 /* STAG invalid: either the */
123 /* STAG is offlimt, being 0, */
124 /* or STAG_key mismatch */
125 #define T4_ERR_PDID 0x2 /* PDID mismatch */
126 #define T4_ERR_QPID 0x3 /* QPID mismatch */
127 #define T4_ERR_ACCESS 0x4 /* Invalid access right */
128 #define T4_ERR_WRAP 0x5 /* Wrap error */
129 #define T4_ERR_BOUND 0x6 /* base and bounds voilation */
130 #define T4_ERR_INVALIDATE_SHARED_MR 0x7 /* attempt to invalidate a */
131 /* shared memory region */
132 #define T4_ERR_INVALIDATE_MR_WITH_MW_BOUND 0x8 /* attempt to invalidate a */
133 /* shared memory region */
134 #define T4_ERR_ECC 0x9 /* ECC error detected */
135 #define T4_ERR_ECC_PSTAG 0xA /* ECC error detected when */
136 /* reading PSTAG for a MW */
138 #define T4_ERR_PBL_ADDR_BOUND 0xB /* pbl addr out of bounds: */
140 #define T4_ERR_SWFLUSH 0xC /* SW FLUSHED */
141 #define T4_ERR_CRC 0x10 /* CRC error */
142 #define T4_ERR_MARKER 0x11 /* Marker error */
143 #define T4_ERR_PDU_LEN_ERR 0x12 /* invalid PDU length */
144 #define T4_ERR_OUT_OF_RQE 0x13 /* out of RQE */
145 #define T4_ERR_DDP_VERSION 0x14 /* wrong DDP version */
146 #define T4_ERR_RDMA_VERSION 0x15 /* wrong RDMA version */
147 #define T4_ERR_OPCODE 0x16 /* invalid rdma opcode */
148 #define T4_ERR_DDP_QUEUE_NUM 0x17 /* invalid ddp queue number */
149 #define T4_ERR_MSN 0x18 /* MSN error */
150 #define T4_ERR_TBIT 0x19 /* tag bit not set correctly */
151 #define T4_ERR_MO 0x1A /* MO not 0 for TERMINATE */
153 #define T4_ERR_MSN_GAP 0x1B
154 #define T4_ERR_MSN_RANGE 0x1C
155 #define T4_ERR_IRD_OVERFLOW 0x1D
156 #define T4_ERR_RQE_ADDR_BOUND 0x1E /* RQE addr out of bounds: */
158 #define T4_ERR_INTERNAL_ERR 0x1F /* internal error (opcode */
185 /* macros for flit 0 of the cqe */
187 #define S_CQE_QPID 12
188 #define M_CQE_QPID 0xFFFFF
189 #define G_CQE_QPID(x) ((((x) >> S_CQE_QPID)) & M_CQE_QPID)
190 #define V_CQE_QPID(x) ((x)<<S_CQE_QPID)
192 #define S_CQE_SWCQE 11
193 #define M_CQE_SWCQE 0x1
194 #define G_CQE_SWCQE(x) ((((x) >> S_CQE_SWCQE)) & M_CQE_SWCQE)
195 #define V_CQE_SWCQE(x) ((x)<<S_CQE_SWCQE)
197 #define S_CQE_STATUS 5
198 #define M_CQE_STATUS 0x1F
199 #define G_CQE_STATUS(x) ((((x) >> S_CQE_STATUS)) & M_CQE_STATUS)
200 #define V_CQE_STATUS(x) ((x)<<S_CQE_STATUS)
203 #define M_CQE_TYPE 0x1
204 #define G_CQE_TYPE(x) ((((x) >> S_CQE_TYPE)) & M_CQE_TYPE)
205 #define V_CQE_TYPE(x) ((x)<<S_CQE_TYPE)
207 #define S_CQE_OPCODE 0
208 #define M_CQE_OPCODE 0xF
209 #define G_CQE_OPCODE(x) ((((x) >> S_CQE_OPCODE)) & M_CQE_OPCODE)
210 #define V_CQE_OPCODE(x) ((x)<<S_CQE_OPCODE)
212 #define SW_CQE(x) (G_CQE_SWCQE(be32_to_cpu((x)->header)))
213 #define CQE_QPID(x) (G_CQE_QPID(be32_to_cpu((x)->header)))
214 #define CQE_TYPE(x) (G_CQE_TYPE(be32_to_cpu((x)->header)))
215 #define SQ_TYPE(x) (CQE_TYPE((x)))
216 #define RQ_TYPE(x) (!CQE_TYPE((x)))
217 #define CQE_STATUS(x) (G_CQE_STATUS(be32_to_cpu((x)->header)))
218 #define CQE_OPCODE(x) (G_CQE_OPCODE(be32_to_cpu((x)->header)))
220 #define CQE_SEND_OPCODE(x)( \
221 (G_CQE_OPCODE(be32_to_cpu((x)->header)) == FW_RI_SEND) || \
222 (G_CQE_OPCODE(be32_to_cpu((x)->header)) == FW_RI_SEND_WITH_SE) || \
223 (G_CQE_OPCODE(be32_to_cpu((x)->header)) == FW_RI_SEND_WITH_INV) || \
224 (G_CQE_OPCODE(be32_to_cpu((x)->header)) == FW_RI_SEND_WITH_SE_INV))
226 #define CQE_LEN(x) (be32_to_cpu((x)->len))
228 /* used for RQ completion processing */
229 #define CQE_WRID_STAG(x) (be32_to_cpu((x)->u.rcqe.stag))
230 #define CQE_WRID_MSN(x) (be32_to_cpu((x)->u.rcqe.msn))
232 /* used for SQ completion processing */
233 #define CQE_WRID_SQ_IDX(x) ((x)->u.scqe.cidx)
235 /* generic accessor macros */
236 #define CQE_WRID_HI(x) ((x)->u.gen.wrid_hi)
237 #define CQE_WRID_LOW(x) ((x)->u.gen.wrid_low)
239 /* macros for flit 3 of the cqe */
240 #define S_CQE_GENBIT 63
241 #define M_CQE_GENBIT 0x1
242 #define G_CQE_GENBIT(x) (((x) >> S_CQE_GENBIT) & M_CQE_GENBIT)
243 #define V_CQE_GENBIT(x) ((x)<<S_CQE_GENBIT)
245 #define S_CQE_OVFBIT 62
246 #define M_CQE_OVFBIT 0x1
247 #define G_CQE_OVFBIT(x) ((((x) >> S_CQE_OVFBIT)) & M_CQE_OVFBIT)
249 #define S_CQE_IQTYPE 60
250 #define M_CQE_IQTYPE 0x3
251 #define G_CQE_IQTYPE(x) ((((x) >> S_CQE_IQTYPE)) & M_CQE_IQTYPE)
253 #define M_CQE_TS 0x0fffffffffffffffULL
254 #define G_CQE_TS(x) ((x) & M_CQE_TS)
256 #define CQE_OVFBIT(x) ((unsigned)G_CQE_OVFBIT(be64_to_cpu((x)->bits_type_ts)))
257 #define CQE_GENBIT(x) ((unsigned)G_CQE_GENBIT(be64_to_cpu((x)->bits_type_ts)))
258 #define CQE_TS(x) (G_CQE_TS(be64_to_cpu((x)->bits_type_ts)))
270 static inline pgprot_t
t4_pgprot_wc(pgprot_t prot
)
272 #if defined(__i386__) || defined(__x86_64__) || defined(CONFIG_PPC64)
273 return pgprot_writecombine(prot
);
275 return pgprot_noncached(prot
);
279 static inline int t4_ocqp_supported(void)
281 #if defined(__i386__) || defined(__x86_64__) || defined(CONFIG_PPC64)
289 T4_SQ_ONCHIP
= (1<<0),
295 DEFINE_DMA_UNMAP_ADDR(mapping
);
296 unsigned long phys_addr
;
297 struct t4_swsqe
*sw_sq
;
298 struct t4_swsqe
*oldest_read
;
315 union t4_recv_wr
*queue
;
317 DEFINE_DMA_UNMAP_ADDR(mapping
);
318 struct t4_swrqe
*sw_rq
;
337 struct c4iw_rdev
*rdev
;
340 static inline int t4_rqes_posted(struct t4_wq
*wq
)
342 return wq
->rq
.in_use
;
345 static inline int t4_rq_empty(struct t4_wq
*wq
)
347 return wq
->rq
.in_use
== 0;
350 static inline int t4_rq_full(struct t4_wq
*wq
)
352 return wq
->rq
.in_use
== (wq
->rq
.size
- 1);
355 static inline u32
t4_rq_avail(struct t4_wq
*wq
)
357 return wq
->rq
.size
- 1 - wq
->rq
.in_use
;
360 static inline void t4_rq_produce(struct t4_wq
*wq
, u8 len16
)
363 if (++wq
->rq
.pidx
== wq
->rq
.size
)
365 wq
->rq
.wq_pidx
+= DIV_ROUND_UP(len16
*16, T4_EQ_ENTRY_SIZE
);
366 if (wq
->rq
.wq_pidx
>= wq
->rq
.size
* T4_RQ_NUM_SLOTS
)
367 wq
->rq
.wq_pidx
%= wq
->rq
.size
* T4_RQ_NUM_SLOTS
;
370 static inline void t4_rq_consume(struct t4_wq
*wq
)
374 if (++wq
->rq
.cidx
== wq
->rq
.size
)
378 static inline int t4_sq_onchip(struct t4_sq
*sq
)
380 return sq
->flags
& T4_SQ_ONCHIP
;
383 static inline int t4_sq_empty(struct t4_wq
*wq
)
385 return wq
->sq
.in_use
== 0;
388 static inline int t4_sq_full(struct t4_wq
*wq
)
390 return wq
->sq
.in_use
== (wq
->sq
.size
- 1);
393 static inline u32
t4_sq_avail(struct t4_wq
*wq
)
395 return wq
->sq
.size
- 1 - wq
->sq
.in_use
;
398 static inline void t4_sq_produce(struct t4_wq
*wq
, u8 len16
)
401 if (++wq
->sq
.pidx
== wq
->sq
.size
)
403 wq
->sq
.wq_pidx
+= DIV_ROUND_UP(len16
*16, T4_EQ_ENTRY_SIZE
);
404 if (wq
->sq
.wq_pidx
>= wq
->sq
.size
* T4_SQ_NUM_SLOTS
)
405 wq
->sq
.wq_pidx
%= wq
->sq
.size
* T4_SQ_NUM_SLOTS
;
408 static inline void t4_sq_consume(struct t4_wq
*wq
)
411 if (++wq
->sq
.cidx
== wq
->sq
.size
)
415 static inline void t4_ring_sq_db(struct t4_wq
*wq
, u16 inc
)
418 writel(QID(wq
->sq
.qid
) | PIDX(inc
), wq
->db
);
421 static inline void t4_ring_rq_db(struct t4_wq
*wq
, u16 inc
)
424 writel(QID(wq
->rq
.qid
) | PIDX(inc
), wq
->db
);
427 static inline int t4_wq_in_error(struct t4_wq
*wq
)
429 return wq
->rq
.queue
[wq
->rq
.size
].status
.qp_err
;
432 static inline void t4_set_wq_in_error(struct t4_wq
*wq
)
434 wq
->rq
.queue
[wq
->rq
.size
].status
.qp_err
= 1;
437 static inline void t4_disable_wq_db(struct t4_wq
*wq
)
439 wq
->rq
.queue
[wq
->rq
.size
].status
.db_off
= 1;
442 static inline void t4_enable_wq_db(struct t4_wq
*wq
)
444 wq
->rq
.queue
[wq
->rq
.size
].status
.db_off
= 0;
447 static inline int t4_wq_db_enabled(struct t4_wq
*wq
)
449 return !wq
->rq
.queue
[wq
->rq
.size
].status
.db_off
;
453 struct t4_cqe
*queue
;
455 DEFINE_DMA_UNMAP_ADDR(mapping
);
456 struct t4_cqe
*sw_queue
;
458 struct c4iw_rdev
*rdev
;
463 u16 size
; /* including status page */
473 static inline int t4_arm_cq(struct t4_cq
*cq
, int se
)
477 while (cq
->cidx_inc
> CIDXINC_MASK
) {
478 val
= SEINTARM(0) | CIDXINC(CIDXINC_MASK
) | TIMERREG(7) |
479 INGRESSQID(cq
->cqid
);
480 writel(val
, cq
->gts
);
481 cq
->cidx_inc
-= CIDXINC_MASK
;
483 val
= SEINTARM(se
) | CIDXINC(cq
->cidx_inc
) | TIMERREG(6) |
484 INGRESSQID(cq
->cqid
);
485 writel(val
, cq
->gts
);
490 static inline void t4_swcq_produce(struct t4_cq
*cq
)
493 if (++cq
->sw_pidx
== cq
->size
)
497 static inline void t4_swcq_consume(struct t4_cq
*cq
)
500 if (++cq
->sw_cidx
== cq
->size
)
504 static inline void t4_hwcq_consume(struct t4_cq
*cq
)
506 cq
->bits_type_ts
= cq
->queue
[cq
->cidx
].bits_type_ts
;
507 if (++cq
->cidx_inc
== (cq
->size
>> 4)) {
510 val
= SEINTARM(0) | CIDXINC(cq
->cidx_inc
) | TIMERREG(7) |
511 INGRESSQID(cq
->cqid
);
512 writel(val
, cq
->gts
);
515 if (++cq
->cidx
== cq
->size
) {
521 static inline int t4_valid_cqe(struct t4_cq
*cq
, struct t4_cqe
*cqe
)
523 return (CQE_GENBIT(cqe
) == cq
->gen
);
526 static inline int t4_next_hw_cqe(struct t4_cq
*cq
, struct t4_cqe
**cqe
)
532 prev_cidx
= cq
->size
- 1;
534 prev_cidx
= cq
->cidx
- 1;
536 if (cq
->queue
[prev_cidx
].bits_type_ts
!= cq
->bits_type_ts
) {
539 printk(KERN_ERR MOD
"cq overflow cqid %u\n", cq
->cqid
);
540 } else if (t4_valid_cqe(cq
, &cq
->queue
[cq
->cidx
])) {
541 *cqe
= &cq
->queue
[cq
->cidx
];
548 static inline struct t4_cqe
*t4_next_sw_cqe(struct t4_cq
*cq
)
551 return &cq
->sw_queue
[cq
->sw_cidx
];
555 static inline int t4_next_cqe(struct t4_cq
*cq
, struct t4_cqe
**cqe
)
561 else if (cq
->sw_in_use
)
562 *cqe
= &cq
->sw_queue
[cq
->sw_cidx
];
564 ret
= t4_next_hw_cqe(cq
, cqe
);
568 static inline int t4_cq_in_error(struct t4_cq
*cq
)
570 return ((struct t4_status_page
*)&cq
->queue
[cq
->size
])->qp_err
;
573 static inline void t4_set_cq_in_error(struct t4_cq
*cq
)
575 ((struct t4_status_page
*)&cq
->queue
[cq
->size
])->qp_err
= 1;