aboutsummaryrefslogtreecommitdiffstats
path: root/drivers/infiniband/hw/ipath/ipath_intr.c
blob: 0bcb428041f32eb7fb2990ee9ae0610390fe201e (plain)
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
/*
 * Copyright (c) 2003, 2004, 2005, 2006 PathScale, Inc. All rights reserved.
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
 *
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
 *
 *      - Redistributions of source code must retain the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer.
 *
 *      - Redistributions in binary form must reproduce the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer in the documentation and/or other materials
 *        provided with the distribution.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */

#include <linux/pci.h>

#include "ipath_kernel.h"
#include "ips_common.h"
#include "ipath_layer.h"

#define E_SUM_PKTERRS \
	(INFINIPATH_E_RHDRLEN | INFINIPATH_E_RBADTID | \
	 INFINIPATH_E_RBADVERSION | INFINIPATH_E_RHDR | \
	 INFINIPATH_E_RLONGPKTLEN | INFINIPATH_E_RSHORTPKTLEN | \
	 INFINIPATH_E_RMAXPKTLEN | INFINIPATH_E_RMINPKTLEN | \
	 INFINIPATH_E_RFORMATERR | INFINIPATH_E_RUNSUPVL | \
	 INFINIPATH_E_RUNEXPCHAR | INFINIPATH_E_REBP)

#define E_SUM_ERRS \
	(INFINIPATH_E_SPIOARMLAUNCH | INFINIPATH_E_SUNEXPERRPKTNUM | \
	 INFINIPATH_E_SDROPPEDDATAPKT | INFINIPATH_E_SDROPPEDSMPPKT | \
	 INFINIPATH_E_SMAXPKTLEN | INFINIPATH_E_SUNSUPVL | \
	 INFINIPATH_E_SMINPKTLEN | INFINIPATH_E_SPKTLEN | \
	 INFINIPATH_E_INVALIDADDR)

static u64 handle_e_sum_errs(struct ipath_devdata *dd, ipath_err_t errs)
{
	unsigned long sbuf[4];
	u64 ignore_this_time = 0;
	u32 piobcnt;

	/* if possible that sendbuffererror could be valid */
	piobcnt = dd->ipath_piobcnt2k + dd->ipath_piobcnt4k;
	/* read these before writing errorclear */
	sbuf[0] = ipath_read_kreg64(
		dd, dd->ipath_kregs->kr_sendbuffererror);
	sbuf[1] = ipath_read_kreg64(
		dd, dd->ipath_kregs->kr_sendbuffererror + 1);
	if (piobcnt > 128) {
		sbuf[2] = ipath_read_kreg64(
			dd, dd->ipath_kregs->kr_sendbuffererror + 2);
		sbuf[3] = ipath_read_kreg64(
			dd, dd->ipath_kregs->kr_sendbuffererror + 3);
	}

	if (sbuf[0] || sbuf[1] || (piobcnt > 128 && (sbuf[2] || sbuf[3]))) {
		int i;

		ipath_cdbg(PKT, "SendbufErrs %lx %lx ", sbuf[0], sbuf[1]);
		if (ipath_debug & __IPATH_PKTDBG && piobcnt > 128)
			printk("%lx %lx ", sbuf[2], sbuf[3]);
		for (i = 0; i < piobcnt; i++) {
			if (test_bit(i, sbuf)) {
				u32 __iomem *piobuf;
				if (i < dd->ipath_piobcnt2k)
					piobuf = (u32 __iomem *)
						(dd->ipath_pio2kbase +
						 i * dd->ipath_palign);
				else
					piobuf = (u32 __iomem *)
						(dd->ipath_pio4kbase +
						 (i - dd->ipath_piobcnt2k) *
						 dd->ipath_4kalign);

				ipath_cdbg(PKT,
					   "PIObuf[%u] @%p pbc is %x; ",
					   i, piobuf, readl(piobuf));

				ipath_disarm_piobufs(dd, i, 1);
			}
		}
		if (ipath_debug & __IPATH_PKTDBG)
			printk("\n");
	}
	if ((errs & (INFINIPATH_E_SDROPPEDDATAPKT |
		     INFINIPATH_E_SDROPPEDSMPPKT |
		     INFINIPATH_E_SMINPKTLEN)) &&
	    !(dd->ipath_flags & IPATH_LINKACTIVE)) {
		/*
		 * This can happen when SMA is trying to bring the link
		 * up, but the IB link changes state at the "wrong" time.
		 * The IB logic then complains that the packet isn't
		 * valid.  We don't want to confuse people, so we just
		 * don't print them, except at debug
		 */
		ipath_dbg("Ignoring pktsend errors %llx, because not "
			  "yet active\n", (unsigned long long) errs);
		ignore_this_time = INFINIPATH_E_SDROPPEDDATAPKT |
			INFINIPATH_E_SDROPPEDSMPPKT |
			INFINIPATH_E_SMINPKTLEN;
	}

	return ignore_this_time;
}

/* return the strings for the most common link states */
static char *ib_linkstate(u32 linkstate)
{
	char *ret;

	switch (linkstate) {
	case IPATH_IBSTATE_INIT:
		ret = "Init";
		break;
	case IPATH_IBSTATE_ARM:
		ret = "Arm";
		break;
	case IPATH_IBSTATE_ACTIVE:
		ret = "Active";
		break;
	default:
		ret = "Down";
	}

	return ret;
}

static void handle_e_ibstatuschanged(struct ipath_devdata *dd,
				     ipath_err_t errs, int noprint)
{
	u64 val;
	u32 ltstate, lstate;

	/*
	 * even if diags are enabled, we want to notice LINKINIT, etc.
	 * We just don't want to change the LED state, or
	 * dd->ipath_kregs->kr_ibcctrl
	 */
	val = ipath_read_kreg64(dd, dd->ipath_kregs->kr_ibcstatus);
	lstate = val & IPATH_IBSTATE_MASK;
	if (lstate == IPATH_IBSTATE_INIT || lstate == IPATH_IBSTATE_ARM ||
	    lstate == IPATH_IBSTATE_ACTIVE) {
		/*
		 * only print at SMA if there is a change, debug if not
		 * (sometimes we want to know that, usually not).
		 */
		if (lstate == ((unsigned) dd->ipath_lastibcstat
			       & IPATH_IBSTATE_MASK)) {
			ipath_dbg("Status change intr but no change (%s)\n",
				  ib_linkstate(lstate));
		}
		else
			ipath_cdbg(SMA, "Unit %u link state %s, last "
				   "was %s\n", dd->ipath_unit,
				   ib_linkstate(lstate),
				   ib_linkstate((unsigned)
						dd->ipath_lastibcstat
						& IPATH_IBSTATE_MASK));
	}
	else {
		lstate = dd->ipath_lastibcstat & IPATH_IBSTATE_MASK;
		if (lstate == IPATH_IBSTATE_INIT ||
		    lstate == IPATH_IBSTATE_ARM ||
		    lstate == IPATH_IBSTATE_ACTIVE)
			ipath_cdbg(SMA, "Unit %u link state down"
				   " (state 0x%x), from %s\n",
				   dd->ipath_unit,
				   (u32)val & IPATH_IBSTATE_MASK,
				   ib_linkstate(lstate));
		else
			ipath_cdbg(VERBOSE, "Unit %u link state changed "
				   "to 0x%x from down (%x)\n",
				   dd->ipath_unit, (u32) val, lstate);
	}
	ltstate = (val >> INFINIPATH_IBCS_LINKTRAININGSTATE_SHIFT) &
		INFINIPATH_IBCS_LINKTRAININGSTATE_MASK;
	lstate = (val >> INFINIPATH_IBCS_LINKSTATE_SHIFT) &
		INFINIPATH_IBCS_LINKSTATE_MASK;

	if (ltstate == INFINIPATH_IBCS_LT_STATE_POLLACTIVE ||
	    ltstate == INFINIPATH_IBCS_LT_STATE_POLLQUIET) {
		u32 last_ltstate;

		/*
		 * Ignore cycling back and forth from Polling.Active
		 * to Polling.Quiet while waiting for the other end of
		 * the link to come up. We will cycle back and forth
		 * between them if no cable is plugged in,
		 * the other device is powered off or disabled, etc.
		 */
		last_ltstate = (dd->ipath_lastibcstat >>
				INFINIPATH_IBCS_LINKTRAININGSTATE_SHIFT)
			& INFINIPATH_IBCS_LINKTRAININGSTATE_MASK;
		if (last_ltstate == INFINIPATH_IBCS_LT_STATE_POLLACTIVE
		    || last_ltstate ==
		    INFINIPATH_IBCS_LT_STATE_POLLQUIET) {
			if (dd->ipath_ibpollcnt > 40) {
				dd->ipath_flags |= IPATH_NOCABLE;
				*dd->ipath_statusp |=
					IPATH_STATUS_IB_NOCABLE;
			} else
				dd->ipath_ibpollcnt++;
			goto skip_ibchange;
		}
	}
	dd->ipath_ibpollcnt = 0;	/* some state other than 2 or 3 */
	ipath_stats.sps_iblink++;
	if (ltstate != INFINIPATH_IBCS_LT_STATE_LINKUP) {
		dd->ipath_flags |= IPATH_LINKDOWN;
		dd->ipath_flags &= ~(IPATH_LINKUNK | IPATH_LINKINIT
				     | IPATH_LINKACTIVE |
				     IPATH_LINKARMED);
		*dd->ipath_statusp &= ~IPATH_STATUS_IB_READY;
		if (!noprint) {
			if (((dd->ipath_lastibcstat >>
			      INFINIPATH_IBCS_LINKSTATE_SHIFT) &
			     INFINIPATH_IBCS_LINKSTATE_MASK)
			    == INFINIPATH_IBCS_L_STATE_ACTIVE)
				/* if from up to down be more vocal */
				ipath_cdbg(SMA,
					   "Unit %u link now down (%s)\n",
					   dd->ipath_unit,
					   ipath_ibcstatus_str[ltstate]);
			else
				ipath_cdbg(VERBOSE, "Unit %u link is "
					   "down (%s)\n", dd->ipath_unit,
					   ipath_ibcstatus_str[ltstate]);
		}

		dd->ipath_f_setextled(dd, lstate, ltstate);
	} else if ((val & IPATH_IBSTATE_MASK) == IPATH_IBSTATE_ACTIVE) {
		dd->ipath_flags |= IPATH_LINKACTIVE;
		dd->ipath_flags &=
			~(IPATH_LINKUNK | IPATH_LINKINIT | IPATH_LINKDOWN |
			  IPATH_LINKARMED | IPATH_NOCABLE);
		*dd->ipath_statusp &= ~IPATH_STATUS_IB_NOCABLE;
		*dd->ipath_statusp |=
			IPATH_STATUS_IB_READY | IPATH_STATUS_IB_CONF;
		dd->ipath_f_setextled(dd, lstate, ltstate);

		__ipath_layer_intr(dd, IPATH_LAYER_INT_IF_UP);
	} else if ((val & IPATH_IBSTATE_MASK) == IPATH_IBSTATE_INIT) {
		/*
		 * set INIT and DOWN.  Down is checked by most of the other
		 * code, but INIT is useful to know in a few places.
		 */
		dd->ipath_flags |= IPATH_LINKINIT | IPATH_LINKDOWN;
		dd->ipath_flags &=
			~(IPATH_LINKUNK | IPATH_LINKACTIVE | IPATH_LINKARMED
			  | IPATH_NOCABLE);
		*dd->ipath_statusp &= ~(IPATH_STATUS_IB_NOCABLE
					| IPATH_STATUS_IB_READY);
		dd->ipath_f_setextled(dd, lstate, ltstate);
	} else if ((val & IPATH_IBSTATE_MASK) == IPATH_IBSTATE_ARM) {
		dd->ipath_flags |= IPATH_LINKARMED;
		dd->ipath_flags &=
			~(IPATH_LINKUNK | IPATH_LINKDOWN | IPATH_LINKINIT |
			  IPATH_LINKACTIVE | IPATH_NOCABLE);
		*dd->ipath_statusp &= ~(IPATH_STATUS_IB_NOCABLE
					| IPATH_STATUS_IB_READY);
		dd->ipath_f_setextled(dd, lstate, ltstate);
	} else {
		if (!noprint)
			ipath_dbg("IBstatuschange unit %u: %s (%x)\n",
				  dd->ipath_unit,
				  ipath_ibcstatus_str[ltstate], ltstate);
	}
skip_ibchange:
	dd->ipath_lastibcstat = val;
}

static void handle_supp_msgs(struct ipath_devdata *dd,
			     unsigned supp_msgs, char msg[512])
{
	/*
	 * Print the message unless it's ibc status change only, which
	 * happens so often we never want to count it.
	 */
	if (dd->ipath_lasterror & ~INFINIPATH_E_IBSTATUSCHANGED) {
		ipath_decode_err(msg, sizeof msg, dd->ipath_lasterror &
				 ~INFINIPATH_E_IBSTATUSCHANGED);
		if (dd->ipath_lasterror &
		    ~(INFINIPATH_E_RRCVEGRFULL | INFINIPATH_E_RRCVHDRFULL))
			ipath_dev_err(dd, "Suppressed %u messages for "
				      "fast-repeating errors (%s) (%llx)\n",
				      supp_msgs, msg,
				      (unsigned long long)
				      dd->ipath_lasterror);
		else {
			/*
			 * rcvegrfull and rcvhdrqfull are "normal", for some
			 * types of processes (mostly benchmarks) that send
			 * huge numbers of messages, while not processing
			 * them. So only complain about these at debug
			 * level.
			 */
			ipath_dbg("Suppressed %u messages for %s\n",
				  supp_msgs, msg);
		}
	}
}

static unsigned handle_frequent_errors(struct ipath_devdata *dd,
				       ipath_err_t errs, char msg[512],
				       int *noprint)
{
	unsigned long nc;
	static unsigned long nextmsg_time;
	static unsigned nmsgs, supp_msgs;

	/*
	 * Throttle back "fast" messages to no more than 10 per 5 seconds.
	 * This isn't perfect, but it's a reasonable heuristic. If we get
	 * more than 10, give a 6x longer delay.
	 */
	nc = jiffies;
	if (nmsgs > 10) {
		if (time_before(nc, nextmsg_time)) {
			*noprint = 1;
			if (!supp_msgs++)
				nextmsg_time = nc + HZ * 3;
		}
		else if (supp_msgs) {
			handle_supp_msgs(dd, supp_msgs, msg);
			supp_msgs = 0;
			nmsgs = 0;
		}
	}
	else if (!nmsgs++ || time_after(nc, nextmsg_time))
		nextmsg_time = nc + HZ / 2;

	return supp_msgs;
}

static void handle_errors(struct ipath_devdata *dd, ipath_err_t errs)
{
	char msg[512];
	u64 ignore_this_time = 0;
	int i;
	int chkerrpkts = 0, noprint = 0;
	unsigned supp_msgs;

	supp_msgs = handle_frequent_errors(dd, errs, msg, &noprint);

	/*
	 * don't report errors that are masked (includes those always
	 * ignored)
	 */
	errs &= ~dd->ipath_maskederrs;

	/* do these first, they are most important */
	if (errs & INFINIPATH_E_HARDWARE) {
		/* reuse same msg buf */
		dd->ipath_f_handle_hwerrors(dd, msg, sizeof msg);
	}

	if (!noprint && (errs & ~infinipath_e_bitsextant))
		ipath_dev_err(dd, "error interrupt with unknown errors "
			      "%llx set\n", (unsigned long long)
			      (errs & ~infinipath_e_bitsextant));

	if (errs & E_SUM_ERRS)
		ignore_this_time = handle_e_sum_errs(dd, errs);

	if (supp_msgs == 250000) {
		/*
		 * It's not entirely reasonable assuming that the errors set
		 * in the last clear period are all responsible for the
		 * problem, but the alternative is to assume it's the only
		 * ones on this particular interrupt, which also isn't great
		 */
		dd->ipath_maskederrs |= dd->ipath_lasterror | errs;
		ipath_write_kreg(dd, dd->ipath_kregs->kr_errormask,
				 ~dd->ipath_maskederrs);
		ipath_decode_err(msg, sizeof msg,
				 (dd->ipath_maskederrs & ~dd->
				  ipath_ignorederrs));

		if ((dd->ipath_maskederrs & ~dd->ipath_ignorederrs) &
		    ~(INFINIPATH_E_RRCVEGRFULL | INFINIPATH_E_RRCVHDRFULL))
			ipath_dev_err(dd, "Disabling error(s) %llx because "
				      "occuring too frequently (%s)\n",
				      (unsigned long long)
				      (dd->ipath_maskederrs &
				       ~dd->ipath_ignorederrs), msg);
		else {
			/*
			 * rcvegrfull and rcvhdrqfull are "normal",
			 * for some types of processes (mostly benchmarks)
			 * that send huge numbers of messages, while not
			 * processing them.  So only complain about
			 * these at debug level.
			 */
			ipath_dbg("Disabling frequent queue full errors "
				  "(%s)\n", msg);
		}

		/*
		 * Re-enable the masked errors after around 3 minutes.  in
		 * ipath_get_faststats().  If we have a series of fast
		 * repeating but different errors, the interval will keep
		 * stretching out, but that's OK, as that's pretty
		 * catastrophic.
		 */
		dd->ipath_unmasktime = jiffies + HZ * 180;
	}

	ipath_write_kreg(dd, dd->ipath_kregs->kr_errorclear, errs);
	if (ignore_this_time)
		errs &= ~ignore_this_time;
	if (errs & ~dd->ipath_lasterror) {
		errs &= ~dd->ipath_lasterror;
		/* never suppress duplicate hwerrors or ibstatuschange */
		dd->ipath_lasterror |= errs &
			~(INFINIPATH_E_HARDWARE |
			  INFINIPATH_E_IBSTATUSCHANGED);
	}
	if (!errs)
		return;

	if (!noprint)
		/*
		 * the ones we mask off are handled specially below or above
		 */
		ipath_decode_err(msg, sizeof msg,
				 errs & ~(INFINIPATH_E_IBSTATUSCHANGED |
					  INFINIPATH_E_RRCVEGRFULL |
					  INFINIPATH_E_RRCVHDRFULL |
					  INFINIPATH_E_HARDWARE));
	else
		/* so we don't need if (!noprint) at strlcat's below */
		*msg = 0;

	if (errs & E_SUM_PKTERRS) {
		ipath_stats.sps_pkterrs++;
		chkerrpkts = 1;
	}
	if (errs & E_SUM_ERRS)
		ipath_stats.sps_errs++;

	if (errs & (INFINIPATH_E_RICRC | INFINIPATH_E_RVCRC)) {
		ipath_stats.sps_crcerrs++;
		chkerrpkts = 1;
	}

	/*
	 * We don't want to print these two as they happen, or we can make
	 * the situation even worse, because it takes so long to print
	 * messages to serial consoles.  Kernel ports get printed from
	 * fast_stats, no more than every 5 seconds, user ports get printed
	 * on close
	 */
	if (errs & INFINIPATH_E_RRCVHDRFULL) {
		int any;
		u32 hd, tl;
		ipath_stats.sps_hdrqfull++;
		for (any = i = 0; i < dd->ipath_cfgports; i++) {
			struct ipath_portdata *pd = dd->ipath_pd[i];
			if (i == 0) {
				hd = dd->ipath_port0head;
				tl = (u32) le64_to_cpu(
					*dd->ipath_hdrqtailptr);
			} else if (pd && pd->port_cnt &&
				   pd->port_rcvhdrtail_kvaddr) {
				/*
				 * don't report same point multiple times,
				 * except kernel
				 */
				tl = (u32) * pd->port_rcvhdrtail_kvaddr;
				if (tl == dd->ipath_lastrcvhdrqtails[i])
					continue;
				hd = ipath_read_ureg32(dd, ur_rcvhdrhead,
						       i);
			} else
				continue;
			if (hd == (tl + 1) ||
			    (!hd && tl == dd->ipath_hdrqlast)) {
				dd->ipath_lastrcvhdrqtails[i] = tl;
				pd->port_hdrqfull++;
				if (i == 0)
					chkerrpkts = 1;
			}
		}
	}
	if (errs & INFINIPATH_E_RRCVEGRFULL) {
		/*
		 * since this is of less importance and not likely to
		 * happen without also getting hdrfull, only count
		 * occurrences; don't check each port (or even the kernel
		 * vs user)
		 */
		ipath_stats.sps_etidfull++;
		if (dd->ipath_port0head !=
		    (u32) le64_to_cpu(*dd->ipath_hdrqtailptr))
			chkerrpkts = 1;
	}

	/*
	 * do this before IBSTATUSCHANGED, in case both bits set in a single
	 * interrupt; we want the STATUSCHANGE to "win", so we do our
	 * internal copy of state machine correctly
	 */
	if (errs & INFINIPATH_E_RIBLOSTLINK) {
		/*
		 * force through block below
		 */
		errs |= INFINIPATH_E_IBSTATUSCHANGED;
		ipath_stats.sps_iblink++;
		dd->ipath_flags |= IPATH_LINKDOWN;
		dd->ipath_flags &= ~(IPATH_LINKUNK | IPATH_LINKINIT
				     | IPATH_LINKARMED | IPATH_LINKACTIVE);
		*dd->ipath_statusp &= ~IPATH_STATUS_IB_READY;
		if (!noprint) {
			u64 st = ipath_read_kreg64(
				dd, dd->ipath_kregs->kr_ibcstatus);

			ipath_dbg("Lost link, link now down (%s)\n",
				  ipath_ibcstatus_str[st & 0xf]);
		}
	}
	if (errs & INFINIPATH_E_IBSTATUSCHANGED)
		handle_e_ibstatuschanged(dd, errs, noprint);

	if (errs & INFINIPATH_E_RESET) {
		if (!noprint)
			ipath_dev_err(dd, "Got reset, requires re-init "
				      "(unload and reload driver)\n");
		dd->ipath_flags &= ~IPATH_INITTED;	/* needs re-init */
		/* mark as having had error */
		*dd->ipath_statusp |= IPATH_STATUS_HWERROR;
		*dd->ipath_statusp &= ~IPATH_STATUS_IB_CONF;
	}

	if (!noprint && *msg)
		ipath_dev_err(dd, "%s error\n", msg);
	if (dd->ipath_sma_state_wanted & dd->ipath_flags) {
		ipath_cdbg(VERBOSE, "sma wanted state %x, iflags now %x, "
			   "waking\n", dd->ipath_sma_state_wanted,
			   dd->ipath_flags);
		wake_up_interruptible(&ipath_sma_state_wait);
	}

	if (chkerrpkts)
		/* process possible error packets in hdrq */
		ipath_kreceive(dd);
}

/* this is separate to allow for better optimization of ipath_intr() */

static void ipath_bad_intr(struct ipath_devdata *dd, u32 * unexpectp)
{
	/*
	 * sometimes happen during driver init and unload, don't want
	 * to process any interrupts at that point
	 */

	/* this is just a bandaid, not a fix, if something goes badly
	 * wrong */
	if (++*unexpectp > 100) {
		if (++*unexpectp > 105) {
			/*
			 * ok, we must be taking somebody else's interrupts,
			 * due to a messed up mptable and/or PIRQ table, so
			 * unregister the interrupt.  We've seen this during
			 * linuxbios development work, and it may happen in
			 * the future again.
			 */
			if (dd->pcidev && dd->pcidev->irq) {
				ipath_dev_err(dd, "Now %u unexpected "
					      "interrupts, unregistering "
					      "interrupt handler\n",
					      *unexpectp);
				ipath_dbg("free_irq of irq %x\n",
					  dd->pcidev->irq);
				free_irq(dd->pcidev->irq, dd);
			}
		}
		if (ipath_read_kreg32(dd, dd->ipath_kregs->kr_intmask)) {
			ipath_dev_err(dd, "%u unexpected interrupts, "
				      "disabling interrupts completely\n",
				      *unexpectp);
			/*
			 * disable all interrupts, something is very wrong
			 */
			ipath_write_kreg(dd, dd->ipath_kregs->kr_intmask,
					 0ULL);
		}
	} else if (*unexpectp > 1)
		ipath_dbg("Interrupt when not ready, should not happen, "
			  "ignoring\n");
}

static void ipath_bad_regread(struct ipath_devdata *dd)
{
	static int allbits;

	/* separate routine, for better optimization of ipath_intr() */

	/*
	 * We print the message and disable interrupts, in hope of
	 * having a better chance of debugging the problem.
	 */
	ipath_dev_err(dd,
		      "Read of interrupt status failed (all bits set)\n");
	if (allbits++) {
		/* disable all interrupts, something is very wrong */
		ipath_write_kreg(dd, dd->ipath_kregs->kr_intmask, 0ULL);
		if (allbits == 2) {
			ipath_dev_err(dd, "Still bad interrupt status, "
				      "unregistering interrupt\n");
			free_irq(dd->pcidev->irq, dd);
		} else if (allbits > 2) {
			if ((allbits % 10000) == 0)
				printk(".");
		} else
			ipath_dev_err(dd, "Disabling interrupts, "
				      "multiple errors\n");
	}
}

static void handle_port_pioavail(struct ipath_devdata *dd)
{
	u32 i;
	/*
	 * start from port 1, since for now port 0  is never using
	 * wait_event for PIO
	 */
	for (i = 1; dd->ipath_portpiowait && i < dd->ipath_cfgports; i++) {
		struct ipath_portdata *pd = dd->ipath_pd[i];

		if (pd && pd->port_cnt &&
		    dd->ipath_portpiowait & (1U << i)) {
			clear_bit(i, &dd->ipath_portpiowait);
			if (test_bit(IPATH_PORT_WAITING_PIO,
				     &pd->port_flag)) {
				clear_bit(IPATH_PORT_WAITING_PIO,
					  &pd->port_flag);
				wake_up_interruptible(&pd->port_wait);
			}
		}
	}
}

static void handle_layer_pioavail(struct ipath_devdata *dd)
{
	int ret;

	ret = __ipath_layer_intr(dd, IPATH_LAYER_INT_SEND_CONTINUE);
	if (ret > 0)
		goto clear;

	ret = __ipath_verbs_piobufavail(dd);
	if (ret > 0)
		goto clear;

	return;
clear:
	set_bit(IPATH_S_PIOINTBUFAVAIL, &dd->ipath_sendctrl);
	ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl,
			 dd->ipath_sendctrl);
}

static void handle_rcv(struct ipath_devdata *dd, u32 istat)
{
	u64 portr;
	int i;
	int rcvdint = 0;

	portr = ((istat >> INFINIPATH_I_RCVAVAIL_SHIFT) &
		 infinipath_i_rcvavail_mask)
		| ((istat >> INFINIPATH_I_RCVURG_SHIFT) &
		   infinipath_i_rcvurg_mask);
	for (i = 0; i < dd->ipath_cfgports; i++) {
		struct ipath_portdata *pd = dd->ipath_pd[i];
		if (portr & (1 << i) && pd &&
		    pd->port_cnt) {
			if (i == 0)
				ipath_kreceive(dd);
			else if (test_bit(IPATH_PORT_WAITING_RCV,
					  &pd->port_flag)) {
				int rcbit;
				clear_bit(IPATH_PORT_WAITING_RCV,
					  &pd->port_flag);
				rcbit = i + INFINIPATH_R_INTRAVAIL_SHIFT;
				clear_bit(1UL << rcbit, &dd->ipath_rcvctrl);
				wake_up_interruptible(&pd->port_wait);
				rcvdint = 1;
			}
		}
	}
	if (rcvdint) {
		/* only want to take one interrupt, so turn off the rcv
		 * interrupt for all the ports that we did the wakeup on
		 * (but never for kernel port)
		 */
		ipath_write_kreg(dd, dd->ipath_kregs->kr_rcvctrl,
				 dd->ipath_rcvctrl);
	}
}

irqreturn_t ipath_intr(int irq, void *data, struct pt_regs *regs)
{
	struct ipath_devdata *dd = data;
	u32 istat = ipath_read_kreg32(dd, dd->ipath_kregs->kr_intstatus);
	ipath_err_t estat = 0;
	static unsigned unexpected = 0;
	irqreturn_t ret;

	if (unlikely(!istat)) {
		ipath_stats.sps_nullintr++;
		ret = IRQ_NONE; /* not our interrupt, or already handled */
		goto bail;
	}
	if (unlikely(istat == -1)) {
		ipath_bad_regread(dd);
		/* don't know if it was our interrupt or not */
		ret = IRQ_NONE;
		goto bail;
	}

	ipath_stats.sps_ints++;

	/*
	 * this needs to be flags&initted, not statusp, so we keep
	 * taking interrupts even after link goes down, etc.
	 * Also, we *must* clear the interrupt at some point, or we won't
	 * take it again, which can be real bad for errors, etc...
	 */

	if (!(dd->ipath_flags & IPATH_INITTED)) {
		ipath_bad_intr(dd, &unexpected);
		ret = IRQ_NONE;
		goto bail;
	}
	if (unexpected)
		unexpected = 0;

	ipath_cdbg(VERBOSE, "intr stat=0x%x\n", istat);

	if (istat & ~infinipath_i_bitsextant)
		ipath_dev_err(dd,
			      "interrupt with unknown interrupts %x set\n",
			      istat & (u32) ~ infinipath_i_bitsextant);

	if (istat & INFINIPATH_I_ERROR) {
		ipath_stats.sps_errints++;
		estat = ipath_read_kreg64(dd,
					  dd->ipath_kregs->kr_errorstatus);
		if (!estat)
			dev_info(&dd->pcidev->dev, "error interrupt (%x), "
				 "but no error bits set!\n", istat);
		else if (estat == -1LL)
			/*
			 * should we try clearing all, or hope next read
			 * works?
			 */
			ipath_dev_err(dd, "Read of error status failed "
				      "(all bits set); ignoring\n");
		else
			handle_errors(dd, estat);
	}

	if (istat & INFINIPATH_I_GPIO) {
		if (unlikely(!(dd->ipath_flags & IPATH_GPIO_INTR))) {
			u32 gpiostatus;
			gpiostatus = ipath_read_kreg32(
				dd, dd->ipath_kregs->kr_gpio_status);
			ipath_dbg("Unexpected GPIO interrupt bits %x\n",
				  gpiostatus);
			ipath_write_kreg(dd, dd->ipath_kregs->kr_gpio_clear,
					 gpiostatus);
		}
		else {
			/* Clear GPIO status bit 2 */
			ipath_write_kreg(dd, dd->ipath_kregs->kr_gpio_clear,
					 (u64) (1 << 2));

			/*
			 * Packets are available in the port 0 rcv queue.
			 * Eventually this needs to be generalized to check
			 * IPATH_GPIO_INTR, and the specific GPIO bit, if
			 * GPIO interrupts are used for anything else.
			 */
			ipath_kreceive(dd);
		}
	}

	/*
	 * clear the ones we will deal with on this round
	 * We clear it early, mostly for receive interrupts, so we
	 * know the chip will have seen this by the time we process
	 * the queue, and will re-interrupt if necessary.  The processor
	 * itself won't take the interrupt again until we return.
	 */
	ipath_write_kreg(dd, dd->ipath_kregs->kr_intclear, istat);

	if (istat & INFINIPATH_I_SPIOBUFAVAIL) {
		clear_bit(IPATH_S_PIOINTBUFAVAIL, &dd->ipath_sendctrl);
		ipath_write_kreg(dd, dd->ipath_kregs->kr_sendctrl,
				 dd->ipath_sendctrl);

		if (dd->ipath_portpiowait)
			handle_port_pioavail(dd);

		handle_layer_pioavail(dd);
	}

	/*
	 * we check for both transition from empty to non-empty, and urgent
	 * packets (those with the interrupt bit set in the header)
	 */

	if (istat & ((infinipath_i_rcvavail_mask <<
		      INFINIPATH_I_RCVAVAIL_SHIFT)
		     | (infinipath_i_rcvurg_mask <<
			INFINIPATH_I_RCVURG_SHIFT)))
		handle_rcv(dd, istat);

	ret = IRQ_HANDLED;

bail:
	return ret;
}