forked from torvalds/linux
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathldc.c
2381 lines (1884 loc) · 48.8 KB
/
ldc.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/* ldc.c: Logical Domain Channel link-layer protocol driver.
*
* Copyright (C) 2007, 2008 David S. Miller <[email protected]>
*/
#include <linux/kernel.h>
#include <linux/export.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/delay.h>
#include <linux/errno.h>
#include <linux/string.h>
#include <linux/scatterlist.h>
#include <linux/interrupt.h>
#include <linux/list.h>
#include <linux/init.h>
#include <linux/bitmap.h>
#include <asm/hypervisor.h>
#include <asm/iommu.h>
#include <asm/page.h>
#include <asm/ldc.h>
#include <asm/mdesc.h>
#define DRV_MODULE_NAME "ldc"
#define PFX DRV_MODULE_NAME ": "
#define DRV_MODULE_VERSION "1.1"
#define DRV_MODULE_RELDATE "July 22, 2008"
static char version[] =
DRV_MODULE_NAME ".c:v" DRV_MODULE_VERSION " (" DRV_MODULE_RELDATE ")\n";
#define LDC_PACKET_SIZE 64
/* Packet header layout for unreliable and reliable mode frames.
* When in RAW mode, packets are simply straight 64-byte payloads
* with no headers.
*/
struct ldc_packet {
u8 type;
#define LDC_CTRL 0x01
#define LDC_DATA 0x02
#define LDC_ERR 0x10
u8 stype;
#define LDC_INFO 0x01
#define LDC_ACK 0x02
#define LDC_NACK 0x04
u8 ctrl;
#define LDC_VERS 0x01 /* Link Version */
#define LDC_RTS 0x02 /* Request To Send */
#define LDC_RTR 0x03 /* Ready To Receive */
#define LDC_RDX 0x04 /* Ready for Data eXchange */
#define LDC_CTRL_MSK 0x0f
u8 env;
#define LDC_LEN 0x3f
#define LDC_FRAG_MASK 0xc0
#define LDC_START 0x40
#define LDC_STOP 0x80
u32 seqid;
union {
u8 u_data[LDC_PACKET_SIZE - 8];
struct {
u32 pad;
u32 ackid;
u8 r_data[LDC_PACKET_SIZE - 8 - 8];
} r;
} u;
};
struct ldc_version {
u16 major;
u16 minor;
};
/* Ordered from largest major to lowest. */
static struct ldc_version ver_arr[] = {
{ .major = 1, .minor = 0 },
};
#define LDC_DEFAULT_MTU (4 * LDC_PACKET_SIZE)
#define LDC_DEFAULT_NUM_ENTRIES (PAGE_SIZE / LDC_PACKET_SIZE)
struct ldc_channel;
struct ldc_mode_ops {
int (*write)(struct ldc_channel *, const void *, unsigned int);
int (*read)(struct ldc_channel *, void *, unsigned int);
};
static const struct ldc_mode_ops raw_ops;
static const struct ldc_mode_ops nonraw_ops;
static const struct ldc_mode_ops stream_ops;
int ldom_domaining_enabled;
struct ldc_iommu {
/* Protects arena alloc/free. */
spinlock_t lock;
struct iommu_arena arena;
struct ldc_mtable_entry *page_table;
};
struct ldc_channel {
/* Protects all operations that depend upon channel state. */
spinlock_t lock;
unsigned long id;
u8 *mssbuf;
u32 mssbuf_len;
u32 mssbuf_off;
struct ldc_packet *tx_base;
unsigned long tx_head;
unsigned long tx_tail;
unsigned long tx_num_entries;
unsigned long tx_ra;
unsigned long tx_acked;
struct ldc_packet *rx_base;
unsigned long rx_head;
unsigned long rx_tail;
unsigned long rx_num_entries;
unsigned long rx_ra;
u32 rcv_nxt;
u32 snd_nxt;
unsigned long chan_state;
struct ldc_channel_config cfg;
void *event_arg;
const struct ldc_mode_ops *mops;
struct ldc_iommu iommu;
struct ldc_version ver;
u8 hs_state;
#define LDC_HS_CLOSED 0x00
#define LDC_HS_OPEN 0x01
#define LDC_HS_GOTVERS 0x02
#define LDC_HS_SENTRTR 0x03
#define LDC_HS_GOTRTR 0x04
#define LDC_HS_COMPLETE 0x10
u8 flags;
#define LDC_FLAG_ALLOCED_QUEUES 0x01
#define LDC_FLAG_REGISTERED_QUEUES 0x02
#define LDC_FLAG_REGISTERED_IRQS 0x04
#define LDC_FLAG_RESET 0x10
u8 mss;
u8 state;
#define LDC_IRQ_NAME_MAX 32
char rx_irq_name[LDC_IRQ_NAME_MAX];
char tx_irq_name[LDC_IRQ_NAME_MAX];
struct hlist_head mh_list;
struct hlist_node list;
};
#define ldcdbg(TYPE, f, a...) \
do { if (lp->cfg.debug & LDC_DEBUG_##TYPE) \
printk(KERN_INFO PFX "ID[%lu] " f, lp->id, ## a); \
} while (0)
static const char *state_to_str(u8 state)
{
switch (state) {
case LDC_STATE_INVALID:
return "INVALID";
case LDC_STATE_INIT:
return "INIT";
case LDC_STATE_BOUND:
return "BOUND";
case LDC_STATE_READY:
return "READY";
case LDC_STATE_CONNECTED:
return "CONNECTED";
default:
return "<UNKNOWN>";
}
}
static void ldc_set_state(struct ldc_channel *lp, u8 state)
{
ldcdbg(STATE, "STATE (%s) --> (%s)\n",
state_to_str(lp->state),
state_to_str(state));
lp->state = state;
}
static unsigned long __advance(unsigned long off, unsigned long num_entries)
{
off += LDC_PACKET_SIZE;
if (off == (num_entries * LDC_PACKET_SIZE))
off = 0;
return off;
}
static unsigned long rx_advance(struct ldc_channel *lp, unsigned long off)
{
return __advance(off, lp->rx_num_entries);
}
static unsigned long tx_advance(struct ldc_channel *lp, unsigned long off)
{
return __advance(off, lp->tx_num_entries);
}
static struct ldc_packet *handshake_get_tx_packet(struct ldc_channel *lp,
unsigned long *new_tail)
{
struct ldc_packet *p;
unsigned long t;
t = tx_advance(lp, lp->tx_tail);
if (t == lp->tx_head)
return NULL;
*new_tail = t;
p = lp->tx_base;
return p + (lp->tx_tail / LDC_PACKET_SIZE);
}
/* When we are in reliable or stream mode, have to track the next packet
* we haven't gotten an ACK for in the TX queue using tx_acked. We have
* to be careful not to stomp over the queue past that point. During
* the handshake, we don't have TX data packets pending in the queue
* and that's why handshake_get_tx_packet() need not be mindful of
* lp->tx_acked.
*/
static unsigned long head_for_data(struct ldc_channel *lp)
{
if (lp->cfg.mode == LDC_MODE_STREAM)
return lp->tx_acked;
return lp->tx_head;
}
static int tx_has_space_for(struct ldc_channel *lp, unsigned int size)
{
unsigned long limit, tail, new_tail, diff;
unsigned int mss;
limit = head_for_data(lp);
tail = lp->tx_tail;
new_tail = tx_advance(lp, tail);
if (new_tail == limit)
return 0;
if (limit > new_tail)
diff = limit - new_tail;
else
diff = (limit +
((lp->tx_num_entries * LDC_PACKET_SIZE) - new_tail));
diff /= LDC_PACKET_SIZE;
mss = lp->mss;
if (diff * mss < size)
return 0;
return 1;
}
static struct ldc_packet *data_get_tx_packet(struct ldc_channel *lp,
unsigned long *new_tail)
{
struct ldc_packet *p;
unsigned long h, t;
h = head_for_data(lp);
t = tx_advance(lp, lp->tx_tail);
if (t == h)
return NULL;
*new_tail = t;
p = lp->tx_base;
return p + (lp->tx_tail / LDC_PACKET_SIZE);
}
static int set_tx_tail(struct ldc_channel *lp, unsigned long tail)
{
unsigned long orig_tail = lp->tx_tail;
int limit = 1000;
lp->tx_tail = tail;
while (limit-- > 0) {
unsigned long err;
err = sun4v_ldc_tx_set_qtail(lp->id, tail);
if (!err)
return 0;
if (err != HV_EWOULDBLOCK) {
lp->tx_tail = orig_tail;
return -EINVAL;
}
udelay(1);
}
lp->tx_tail = orig_tail;
return -EBUSY;
}
/* This just updates the head value in the hypervisor using
* a polling loop with a timeout. The caller takes care of
* upating software state representing the head change, if any.
*/
static int __set_rx_head(struct ldc_channel *lp, unsigned long head)
{
int limit = 1000;
while (limit-- > 0) {
unsigned long err;
err = sun4v_ldc_rx_set_qhead(lp->id, head);
if (!err)
return 0;
if (err != HV_EWOULDBLOCK)
return -EINVAL;
udelay(1);
}
return -EBUSY;
}
static int send_tx_packet(struct ldc_channel *lp,
struct ldc_packet *p,
unsigned long new_tail)
{
BUG_ON(p != (lp->tx_base + (lp->tx_tail / LDC_PACKET_SIZE)));
return set_tx_tail(lp, new_tail);
}
static struct ldc_packet *handshake_compose_ctrl(struct ldc_channel *lp,
u8 stype, u8 ctrl,
void *data, int dlen,
unsigned long *new_tail)
{
struct ldc_packet *p = handshake_get_tx_packet(lp, new_tail);
if (p) {
memset(p, 0, sizeof(*p));
p->type = LDC_CTRL;
p->stype = stype;
p->ctrl = ctrl;
if (data)
memcpy(p->u.u_data, data, dlen);
}
return p;
}
static int start_handshake(struct ldc_channel *lp)
{
struct ldc_packet *p;
struct ldc_version *ver;
unsigned long new_tail;
ver = &ver_arr[0];
ldcdbg(HS, "SEND VER INFO maj[%u] min[%u]\n",
ver->major, ver->minor);
p = handshake_compose_ctrl(lp, LDC_INFO, LDC_VERS,
ver, sizeof(*ver), &new_tail);
if (p) {
int err = send_tx_packet(lp, p, new_tail);
if (!err)
lp->flags &= ~LDC_FLAG_RESET;
return err;
}
return -EBUSY;
}
static int send_version_nack(struct ldc_channel *lp,
u16 major, u16 minor)
{
struct ldc_packet *p;
struct ldc_version ver;
unsigned long new_tail;
ver.major = major;
ver.minor = minor;
p = handshake_compose_ctrl(lp, LDC_NACK, LDC_VERS,
&ver, sizeof(ver), &new_tail);
if (p) {
ldcdbg(HS, "SEND VER NACK maj[%u] min[%u]\n",
ver.major, ver.minor);
return send_tx_packet(lp, p, new_tail);
}
return -EBUSY;
}
static int send_version_ack(struct ldc_channel *lp,
struct ldc_version *vp)
{
struct ldc_packet *p;
unsigned long new_tail;
p = handshake_compose_ctrl(lp, LDC_ACK, LDC_VERS,
vp, sizeof(*vp), &new_tail);
if (p) {
ldcdbg(HS, "SEND VER ACK maj[%u] min[%u]\n",
vp->major, vp->minor);
return send_tx_packet(lp, p, new_tail);
}
return -EBUSY;
}
static int send_rts(struct ldc_channel *lp)
{
struct ldc_packet *p;
unsigned long new_tail;
p = handshake_compose_ctrl(lp, LDC_INFO, LDC_RTS, NULL, 0,
&new_tail);
if (p) {
p->env = lp->cfg.mode;
p->seqid = 0;
lp->rcv_nxt = 0;
ldcdbg(HS, "SEND RTS env[0x%x] seqid[0x%x]\n",
p->env, p->seqid);
return send_tx_packet(lp, p, new_tail);
}
return -EBUSY;
}
static int send_rtr(struct ldc_channel *lp)
{
struct ldc_packet *p;
unsigned long new_tail;
p = handshake_compose_ctrl(lp, LDC_INFO, LDC_RTR, NULL, 0,
&new_tail);
if (p) {
p->env = lp->cfg.mode;
p->seqid = 0;
ldcdbg(HS, "SEND RTR env[0x%x] seqid[0x%x]\n",
p->env, p->seqid);
return send_tx_packet(lp, p, new_tail);
}
return -EBUSY;
}
static int send_rdx(struct ldc_channel *lp)
{
struct ldc_packet *p;
unsigned long new_tail;
p = handshake_compose_ctrl(lp, LDC_INFO, LDC_RDX, NULL, 0,
&new_tail);
if (p) {
p->env = 0;
p->seqid = ++lp->snd_nxt;
p->u.r.ackid = lp->rcv_nxt;
ldcdbg(HS, "SEND RDX env[0x%x] seqid[0x%x] ackid[0x%x]\n",
p->env, p->seqid, p->u.r.ackid);
return send_tx_packet(lp, p, new_tail);
}
return -EBUSY;
}
static int send_data_nack(struct ldc_channel *lp, struct ldc_packet *data_pkt)
{
struct ldc_packet *p;
unsigned long new_tail;
int err;
p = data_get_tx_packet(lp, &new_tail);
if (!p)
return -EBUSY;
memset(p, 0, sizeof(*p));
p->type = data_pkt->type;
p->stype = LDC_NACK;
p->ctrl = data_pkt->ctrl & LDC_CTRL_MSK;
p->seqid = lp->snd_nxt + 1;
p->u.r.ackid = lp->rcv_nxt;
ldcdbg(HS, "SEND DATA NACK type[0x%x] ctl[0x%x] seq[0x%x] ack[0x%x]\n",
p->type, p->ctrl, p->seqid, p->u.r.ackid);
err = send_tx_packet(lp, p, new_tail);
if (!err)
lp->snd_nxt++;
return err;
}
static int ldc_abort(struct ldc_channel *lp)
{
unsigned long hv_err;
ldcdbg(STATE, "ABORT\n");
/* We report but do not act upon the hypervisor errors because
* there really isn't much we can do if they fail at this point.
*/
hv_err = sun4v_ldc_tx_qconf(lp->id, lp->tx_ra, lp->tx_num_entries);
if (hv_err)
printk(KERN_ERR PFX "ldc_abort: "
"sun4v_ldc_tx_qconf(%lx,%lx,%lx) failed, err=%lu\n",
lp->id, lp->tx_ra, lp->tx_num_entries, hv_err);
hv_err = sun4v_ldc_tx_get_state(lp->id,
&lp->tx_head,
&lp->tx_tail,
&lp->chan_state);
if (hv_err)
printk(KERN_ERR PFX "ldc_abort: "
"sun4v_ldc_tx_get_state(%lx,...) failed, err=%lu\n",
lp->id, hv_err);
hv_err = sun4v_ldc_rx_qconf(lp->id, lp->rx_ra, lp->rx_num_entries);
if (hv_err)
printk(KERN_ERR PFX "ldc_abort: "
"sun4v_ldc_rx_qconf(%lx,%lx,%lx) failed, err=%lu\n",
lp->id, lp->rx_ra, lp->rx_num_entries, hv_err);
/* Refetch the RX queue state as well, because we could be invoked
* here in the queue processing context.
*/
hv_err = sun4v_ldc_rx_get_state(lp->id,
&lp->rx_head,
&lp->rx_tail,
&lp->chan_state);
if (hv_err)
printk(KERN_ERR PFX "ldc_abort: "
"sun4v_ldc_rx_get_state(%lx,...) failed, err=%lu\n",
lp->id, hv_err);
return -ECONNRESET;
}
static struct ldc_version *find_by_major(u16 major)
{
struct ldc_version *ret = NULL;
int i;
for (i = 0; i < ARRAY_SIZE(ver_arr); i++) {
struct ldc_version *v = &ver_arr[i];
if (v->major <= major) {
ret = v;
break;
}
}
return ret;
}
static int process_ver_info(struct ldc_channel *lp, struct ldc_version *vp)
{
struct ldc_version *vap;
int err;
ldcdbg(HS, "GOT VERSION INFO major[%x] minor[%x]\n",
vp->major, vp->minor);
if (lp->hs_state == LDC_HS_GOTVERS) {
lp->hs_state = LDC_HS_OPEN;
memset(&lp->ver, 0, sizeof(lp->ver));
}
vap = find_by_major(vp->major);
if (!vap) {
err = send_version_nack(lp, 0, 0);
} else if (vap->major != vp->major) {
err = send_version_nack(lp, vap->major, vap->minor);
} else {
struct ldc_version ver = *vp;
if (ver.minor > vap->minor)
ver.minor = vap->minor;
err = send_version_ack(lp, &ver);
if (!err) {
lp->ver = ver;
lp->hs_state = LDC_HS_GOTVERS;
}
}
if (err)
return ldc_abort(lp);
return 0;
}
static int process_ver_ack(struct ldc_channel *lp, struct ldc_version *vp)
{
ldcdbg(HS, "GOT VERSION ACK major[%x] minor[%x]\n",
vp->major, vp->minor);
if (lp->hs_state == LDC_HS_GOTVERS) {
if (lp->ver.major != vp->major ||
lp->ver.minor != vp->minor)
return ldc_abort(lp);
} else {
lp->ver = *vp;
lp->hs_state = LDC_HS_GOTVERS;
}
if (send_rts(lp))
return ldc_abort(lp);
return 0;
}
static int process_ver_nack(struct ldc_channel *lp, struct ldc_version *vp)
{
struct ldc_version *vap;
struct ldc_packet *p;
unsigned long new_tail;
if (vp->major == 0 && vp->minor == 0)
return ldc_abort(lp);
vap = find_by_major(vp->major);
if (!vap)
return ldc_abort(lp);
p = handshake_compose_ctrl(lp, LDC_INFO, LDC_VERS,
vap, sizeof(*vap),
&new_tail);
if (!p)
return ldc_abort(lp);
return send_tx_packet(lp, p, new_tail);
}
static int process_version(struct ldc_channel *lp,
struct ldc_packet *p)
{
struct ldc_version *vp;
vp = (struct ldc_version *) p->u.u_data;
switch (p->stype) {
case LDC_INFO:
return process_ver_info(lp, vp);
case LDC_ACK:
return process_ver_ack(lp, vp);
case LDC_NACK:
return process_ver_nack(lp, vp);
default:
return ldc_abort(lp);
}
}
static int process_rts(struct ldc_channel *lp,
struct ldc_packet *p)
{
ldcdbg(HS, "GOT RTS stype[%x] seqid[%x] env[%x]\n",
p->stype, p->seqid, p->env);
if (p->stype != LDC_INFO ||
lp->hs_state != LDC_HS_GOTVERS ||
p->env != lp->cfg.mode)
return ldc_abort(lp);
lp->snd_nxt = p->seqid;
lp->rcv_nxt = p->seqid;
lp->hs_state = LDC_HS_SENTRTR;
if (send_rtr(lp))
return ldc_abort(lp);
return 0;
}
static int process_rtr(struct ldc_channel *lp,
struct ldc_packet *p)
{
ldcdbg(HS, "GOT RTR stype[%x] seqid[%x] env[%x]\n",
p->stype, p->seqid, p->env);
if (p->stype != LDC_INFO ||
p->env != lp->cfg.mode)
return ldc_abort(lp);
lp->snd_nxt = p->seqid;
lp->hs_state = LDC_HS_COMPLETE;
ldc_set_state(lp, LDC_STATE_CONNECTED);
send_rdx(lp);
return LDC_EVENT_UP;
}
static int rx_seq_ok(struct ldc_channel *lp, u32 seqid)
{
return lp->rcv_nxt + 1 == seqid;
}
static int process_rdx(struct ldc_channel *lp,
struct ldc_packet *p)
{
ldcdbg(HS, "GOT RDX stype[%x] seqid[%x] env[%x] ackid[%x]\n",
p->stype, p->seqid, p->env, p->u.r.ackid);
if (p->stype != LDC_INFO ||
!(rx_seq_ok(lp, p->seqid)))
return ldc_abort(lp);
lp->rcv_nxt = p->seqid;
lp->hs_state = LDC_HS_COMPLETE;
ldc_set_state(lp, LDC_STATE_CONNECTED);
return LDC_EVENT_UP;
}
static int process_control_frame(struct ldc_channel *lp,
struct ldc_packet *p)
{
switch (p->ctrl) {
case LDC_VERS:
return process_version(lp, p);
case LDC_RTS:
return process_rts(lp, p);
case LDC_RTR:
return process_rtr(lp, p);
case LDC_RDX:
return process_rdx(lp, p);
default:
return ldc_abort(lp);
}
}
static int process_error_frame(struct ldc_channel *lp,
struct ldc_packet *p)
{
return ldc_abort(lp);
}
static int process_data_ack(struct ldc_channel *lp,
struct ldc_packet *ack)
{
unsigned long head = lp->tx_acked;
u32 ackid = ack->u.r.ackid;
while (1) {
struct ldc_packet *p = lp->tx_base + (head / LDC_PACKET_SIZE);
head = tx_advance(lp, head);
if (p->seqid == ackid) {
lp->tx_acked = head;
return 0;
}
if (head == lp->tx_tail)
return ldc_abort(lp);
}
return 0;
}
static void send_events(struct ldc_channel *lp, unsigned int event_mask)
{
if (event_mask & LDC_EVENT_RESET)
lp->cfg.event(lp->event_arg, LDC_EVENT_RESET);
if (event_mask & LDC_EVENT_UP)
lp->cfg.event(lp->event_arg, LDC_EVENT_UP);
if (event_mask & LDC_EVENT_DATA_READY)
lp->cfg.event(lp->event_arg, LDC_EVENT_DATA_READY);
}
static irqreturn_t ldc_rx(int irq, void *dev_id)
{
struct ldc_channel *lp = dev_id;
unsigned long orig_state, flags;
unsigned int event_mask;
spin_lock_irqsave(&lp->lock, flags);
orig_state = lp->chan_state;
/* We should probably check for hypervisor errors here and
* reset the LDC channel if we get one.
*/
sun4v_ldc_rx_get_state(lp->id,
&lp->rx_head,
&lp->rx_tail,
&lp->chan_state);
ldcdbg(RX, "RX state[0x%02lx:0x%02lx] head[0x%04lx] tail[0x%04lx]\n",
orig_state, lp->chan_state, lp->rx_head, lp->rx_tail);
event_mask = 0;
if (lp->cfg.mode == LDC_MODE_RAW &&
lp->chan_state == LDC_CHANNEL_UP) {
lp->hs_state = LDC_HS_COMPLETE;
ldc_set_state(lp, LDC_STATE_CONNECTED);
event_mask |= LDC_EVENT_UP;
orig_state = lp->chan_state;
}
/* If we are in reset state, flush the RX queue and ignore
* everything.
*/
if (lp->flags & LDC_FLAG_RESET) {
(void) __set_rx_head(lp, lp->rx_tail);
goto out;
}
/* Once we finish the handshake, we let the ldc_read()
* paths do all of the control frame and state management.
* Just trigger the callback.
*/
if (lp->hs_state == LDC_HS_COMPLETE) {
handshake_complete:
if (lp->chan_state != orig_state) {
unsigned int event = LDC_EVENT_RESET;
if (lp->chan_state == LDC_CHANNEL_UP)
event = LDC_EVENT_UP;
event_mask |= event;
}
if (lp->rx_head != lp->rx_tail)
event_mask |= LDC_EVENT_DATA_READY;
goto out;
}
if (lp->chan_state != orig_state)
goto out;
while (lp->rx_head != lp->rx_tail) {
struct ldc_packet *p;
unsigned long new;
int err;
p = lp->rx_base + (lp->rx_head / LDC_PACKET_SIZE);
switch (p->type) {
case LDC_CTRL:
err = process_control_frame(lp, p);
if (err > 0)
event_mask |= err;
break;
case LDC_DATA:
event_mask |= LDC_EVENT_DATA_READY;
err = 0;
break;
case LDC_ERR:
err = process_error_frame(lp, p);
break;
default:
err = ldc_abort(lp);
break;
}
if (err < 0)
break;
new = lp->rx_head;
new += LDC_PACKET_SIZE;
if (new == (lp->rx_num_entries * LDC_PACKET_SIZE))
new = 0;
lp->rx_head = new;
err = __set_rx_head(lp, new);
if (err < 0) {
(void) ldc_abort(lp);
break;
}
if (lp->hs_state == LDC_HS_COMPLETE)
goto handshake_complete;
}
out:
spin_unlock_irqrestore(&lp->lock, flags);
send_events(lp, event_mask);
return IRQ_HANDLED;
}
static irqreturn_t ldc_tx(int irq, void *dev_id)
{
struct ldc_channel *lp = dev_id;
unsigned long flags, orig_state;
unsigned int event_mask = 0;
spin_lock_irqsave(&lp->lock, flags);
orig_state = lp->chan_state;
/* We should probably check for hypervisor errors here and
* reset the LDC channel if we get one.
*/
sun4v_ldc_tx_get_state(lp->id,
&lp->tx_head,
&lp->tx_tail,
&lp->chan_state);
ldcdbg(TX, " TX state[0x%02lx:0x%02lx] head[0x%04lx] tail[0x%04lx]\n",
orig_state, lp->chan_state, lp->tx_head, lp->tx_tail);
if (lp->cfg.mode == LDC_MODE_RAW &&
lp->chan_state == LDC_CHANNEL_UP) {
lp->hs_state = LDC_HS_COMPLETE;
ldc_set_state(lp, LDC_STATE_CONNECTED);
event_mask |= LDC_EVENT_UP;
}
spin_unlock_irqrestore(&lp->lock, flags);
send_events(lp, event_mask);
return IRQ_HANDLED;
}
/* XXX ldc_alloc() and ldc_free() needs to run under a mutex so
* XXX that addition and removal from the ldc_channel_list has
* XXX atomicity, otherwise the __ldc_channel_exists() check is
* XXX totally pointless as another thread can slip into ldc_alloc()
* XXX and add a channel with the same ID. There also needs to be
* XXX a spinlock for ldc_channel_list.
*/
static HLIST_HEAD(ldc_channel_list);
static int __ldc_channel_exists(unsigned long id)
{
struct ldc_channel *lp;
hlist_for_each_entry(lp, &ldc_channel_list, list) {
if (lp->id == id)
return 1;
}
return 0;
}
static int alloc_queue(const char *name, unsigned long num_entries,
struct ldc_packet **base, unsigned long *ra)
{
unsigned long size, order;
void *q;
size = num_entries * LDC_PACKET_SIZE;
order = get_order(size);
q = (void *) __get_free_pages(GFP_KERNEL, order);
if (!q) {
printk(KERN_ERR PFX "Alloc of %s queue failed with "
"size=%lu order=%lu\n", name, size, order);
return -ENOMEM;
}
memset(q, 0, PAGE_SIZE << order);
*base = q;
*ra = __pa(q);
return 0;
}
static void free_queue(unsigned long num_entries, struct ldc_packet *q)
{
unsigned long size, order;
if (!q)
return;
size = num_entries * LDC_PACKET_SIZE;
order = get_order(size);
free_pages((unsigned long)q, order);
}