summaryrefslogtreecommitdiff
path: root/drivers/crypto/caam/qi.c
blob: 673f4742c14d25badb85657e26c58e4e43f5079b (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
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
/*
 * CAAM/SEC 4.x QI transport/backend driver
 * Queue Interface backend functionality
 *
 * Copyright 2013 Freescale Semiconductor, Inc.
 */

#include <linux/cpumask.h>
#include <linux/kthread.h>
#include <linux/fsl_qman.h>

#include "regs.h"
#include "qi.h"
#include "desc.h"
#include "intern.h"
#include "desc_constr.h"

#define CAAM_REQ_CGR_THRESHOLD	0x1000000
#define PRE_HDR_LEN		2	/* Length in u32 words */
#define PREHDR_RSLS_SHIFT	31
#define PENDING_JOBS_DEPTH	512
/*
 * The jobs are processed by the driver against a driver context.
 * With every cryptographic context, a driver context is attached.
 * The driver context contains data for private use by driver.
 * For the applications, this is an opaque structure.
 */

struct caam_drv_ctx {
	u32 prehdr[PRE_HDR_LEN];	/* Preheader placed before shrd desc */
	u32 sh_desc[MAX_SDLEN];		/* Shared descriptor */
	dma_addr_t context_a; /* shared descriptor dma address */
	struct qman_fq *req_fq;		/* Request frame queue to caam */
	struct qman_fq *rsp_fq;		/* Response frame queue from caam */
	int cpu;			/* cpu on which to recv caam rsp */
	struct device *qidev;		/* device pointer for QI backend */
} ____cacheline_aligned;

/*
 * percpu private data structure to main list of pending responses expected
 * on each cpu.
 */
struct caam_qi_pcpu_priv {
	struct napi_struct irqtask;	/* IRQ task for QI backend */
	struct net_device net_dev;	/* netdev used by NAPI */
	struct qman_fq rsp_fq;		/* Response FQ from CAAM */
	spinlock_t listlock ____cacheline_aligned; /* for protecting
						    * simultaneous access
						    * to bklog_list */
	struct list_head bklog_list;	/* List of pending responses*/
	atomic_t pending;		/* Number of pending responses
					 * from CAAM on this cpu */
} ____cacheline_aligned;

static DEFINE_PER_CPU(struct caam_qi_pcpu_priv, pcpu_qipriv);

struct caam_qi_priv {
	bool sec_congested;		/* Indicates whether SEC is congested */
	bool cpu_congested;		/* Indicates whether CPU is congested */
	struct qman_cgr rsp_cgr;	/* QMAN response CGR */
	struct qman_cgr req_cgr;	/* QMAN request CGR */
	struct platform_device *qi_pdev; /* Platform device for QI backend */
};

static struct caam_qi_priv qipriv ____cacheline_aligned;

/*
 * CPU from where the module initialised. This is required because
 * QMAN driver requires CGRs to be removed from same CPU from where
 * they were originally allocated
 */
static int mod_init_cpu;

bool caam_drv_ctx_busy(struct caam_drv_ctx *drv_ctx)
{
	int pending;

	if (qipriv.sec_congested)
		return true;

	pending = atomic_read(&per_cpu(pcpu_qipriv.pending, drv_ctx->cpu));
	if (pending >= PENDING_JOBS_DEPTH)
		return true;

	return false;
}
EXPORT_SYMBOL(caam_drv_ctx_busy);

int caam_qi_enqueue(struct device *qidev, struct caam_drv_req *req)
{
	struct qm_fd fd;
	int ret;
	size_t size;
	struct list_head *list;
	int num_retries = 0;
	unsigned long flags;

	fd.cmd = 0;
	fd.format = qm_fd_compound;
	fd.cong_weight = req->fd_sgt[1].length;

	size = 2 * sizeof(struct qm_sg_entry);

	fd.addr = dma_map_single(qidev, req->fd_sgt, size , DMA_BIDIRECTIONAL);
	if (dma_mapping_error(qidev, fd.addr)) {
		dev_err(qidev, "DMA mapping error for QI enqueue request\n");
		return -EIO;
	}

	req->hwaddr = qm_fd_addr(&fd);
	list = &per_cpu(pcpu_qipriv.bklog_list, req->drv_ctx->cpu);

	spin_lock_irqsave(&per_cpu(pcpu_qipriv.listlock, req->drv_ctx->cpu),
			  flags);
	list_add_tail(&req->hdr__, list);
	spin_unlock_irqrestore(&per_cpu(pcpu_qipriv.listlock,
					req->drv_ctx->cpu), flags);
	atomic_inc(&per_cpu(pcpu_qipriv.pending, req->drv_ctx->cpu));

	do {
		ret = qman_enqueue(req->drv_ctx->req_fq, &fd, 0);
		if (likely(!ret))
			return 0;

		if (-EBUSY != ret)
			break;
		num_retries++;
	} while (num_retries < 10000);

	dev_err(qidev, "qman_enqueue failed: %d\n", ret);

	spin_lock_irqsave(&per_cpu(pcpu_qipriv.listlock, req->drv_ctx->cpu),
			  flags);
	list_del(&req->hdr__);
	spin_unlock_irqrestore(&per_cpu(pcpu_qipriv.listlock,
					req->drv_ctx->cpu), flags);
	atomic_dec(&per_cpu(pcpu_qipriv.pending, req->drv_ctx->cpu));

	dma_unmap_single(qidev, fd.addr, size, DMA_BIDIRECTIONAL);
	return ret;
}
EXPORT_SYMBOL(caam_qi_enqueue);

struct caam_drv_req *lookup_drv_req(const struct qm_fd *fd, int cpu)
{
	struct list_head *pos, *list, *n;
	struct caam_drv_req *req;
	unsigned long flags;

	list = &per_cpu(pcpu_qipriv.bklog_list, cpu);
	list_for_each_safe(pos, n, list) {
		req = container_of(pos, struct caam_drv_req, hdr__);

		if (req->hwaddr == qm_fd_addr(fd)) {
			BUG_ON(req->drv_ctx->cpu != cpu);

			spin_lock_irqsave(&per_cpu(pcpu_qipriv.listlock,
						   req->drv_ctx->cpu), flags);
			list_del(&req->hdr__);
			spin_unlock_irqrestore(&per_cpu(pcpu_qipriv.listlock,
							req->drv_ctx->cpu),
					       flags);
			atomic_dec(&per_cpu(pcpu_qipriv.pending,
					    req->drv_ctx->cpu));
			return req;
		}
	}

	return NULL;
}


static struct caam_drv_req *fd_to_drv_req(const struct qm_fd *fd)
{
	struct caam_drv_req *req;
	const cpumask_t *cpus = qman_affine_cpus();
	int i;

	/*
	 * First check on this_cpu since this is likely case of normal caam
	 * response path.
	 */
	req = lookup_drv_req(fd, smp_processor_id());
	if (likely(req))
		return req;
	/*
	 * If drv_req is not found on this_cpu, then try searching on other
	 * portal owning cpus. This is required to handle ERN callbacks and
	 * volatile dequeues. These may be issued on a CPU which is different
	 * than the one associated with the drv_req's drv_ctx.
	 */
	for_each_cpu(i, cpus) {
		if (i == smp_processor_id())
			continue;	/* Already checked */
		req = lookup_drv_req(fd, i);

		if (req)
			return req;
	}

	return NULL;
}

static void caam_fq_ern_cb(struct qman_portal *qm, struct qman_fq *fq,
			   const struct qm_mr_entry *msg)
{
	const struct qm_fd *fd;
	struct caam_drv_req *drv_req;
	size_t size;
	struct device *qidev = &per_cpu(pcpu_qipriv.net_dev,
					smp_processor_id()).dev;

	fd = &msg->ern.fd;

	if (qm_fd_compound != fd->format) {
		dev_err(qidev, "Non compound FD from CAAM\n");
		return;
	}

	drv_req = fd_to_drv_req(fd);
	if (!drv_req) {
		dev_err(qidev,
			"Can't find original request for caam response\n");
		return;
	}

	size = 2 * sizeof(struct qm_sg_entry);
	dma_unmap_single(drv_req->drv_ctx->qidev, fd->addr,
			 size, DMA_BIDIRECTIONAL);

	drv_req->cbk(drv_req, -EIO);
}

static enum qman_cb_dqrr_result	caam_req_fq_dqrr_cb(struct qman_portal *p,
					struct qman_fq *req_fq,
					const struct qm_dqrr_entry *dqrr)
{
	struct caam_drv_req *drv_req;
	const struct qm_fd *fd;
	size_t size;
	struct device *qidev = &per_cpu(pcpu_qipriv.net_dev,
					smp_processor_id()).dev;

	fd = &dqrr->fd;

	drv_req = fd_to_drv_req(fd);
	if (!drv_req) {
		dev_err(qidev,
			"Can't find original request for caam response\n");
		return qman_cb_dqrr_consume;
	}

	size = 2 * sizeof(struct qm_sg_entry);
	dma_unmap_single(drv_req->drv_ctx->qidev, fd->addr,
			 size, DMA_BIDIRECTIONAL);

	drv_req->cbk(drv_req, -EIO);

	return qman_cb_dqrr_consume;
}

static struct qman_fq *create_caam_req_fq(struct device *qidev,
					  struct qman_fq *rsp_fq,
					  dma_addr_t hwdesc,
					  int fq_sched_flag)
{
	int ret, flags;
	struct qman_fq *req_fq;
	struct qm_mcc_initfq opts;

	req_fq = kzalloc(sizeof(*req_fq), GFP_ATOMIC);
	if (!req_fq) {
		dev_err(qidev, "Mem alloc for CAAM req FQ failed\n");
		return ERR_PTR(-ENOMEM);
	}

	req_fq->cb.dqrr = caam_req_fq_dqrr_cb;
	req_fq->cb.ern = caam_fq_ern_cb;
	req_fq->cb.fqs = NULL;

	flags = QMAN_FQ_FLAG_DYNAMIC_FQID |
		QMAN_FQ_FLAG_TO_DCPORTAL |
		QMAN_FQ_FLAG_LOCKED;

	ret = qman_create_fq(0, flags, req_fq);
	if (ret) {
		dev_err(qidev, "Failed to create session REQ FQ\n");
		goto create_req_fq_fail;
	}

	flags = fq_sched_flag;
	opts.we_mask = QM_INITFQ_WE_FQCTRL | QM_INITFQ_WE_DESTWQ |
			QM_INITFQ_WE_CONTEXTB | QM_INITFQ_WE_CONTEXTA |
			QM_INITFQ_WE_CGID;

	opts.fqd.fq_ctrl = QM_FQCTRL_CPCSTASH | QM_FQCTRL_CGE;
	opts.fqd.dest.channel = qm_channel_caam;
	opts.fqd.dest.wq = 3;
	opts.fqd.cgid = qipriv.req_cgr.cgrid;
	opts.fqd.context_b = qman_fq_fqid(rsp_fq);
	opts.fqd.context_a.hi = upper_32_bits(hwdesc);
	opts.fqd.context_a.lo = lower_32_bits(hwdesc);

	ret = qman_init_fq(req_fq, flags, &opts);
	if (ret) {
		dev_err(qidev, "Failed to init session req FQ\n");
		goto init_req_fq_fail;
	}

	return req_fq;

init_req_fq_fail:
	qman_destroy_fq(req_fq, 0);

create_req_fq_fail:
	kfree(req_fq);
	return ERR_PTR(ret);
}

static int empty_retired_fq(struct device *qidev, struct qman_fq *fq)
{
	int ret;
	enum qman_fq_state state;

	u32 flags = QMAN_VOLATILE_FLAG_WAIT_INT | QMAN_VOLATILE_FLAG_FINISH;
	u32 vdqcr = QM_VDQCR_PRECEDENCE_VDQCR | QM_VDQCR_NUMFRAMES_TILLEMPTY;

	ret = qman_volatile_dequeue(fq, flags, vdqcr);
	if (ret) {
		dev_err(qidev, "Volatile dequeue fail for FQ: %u\n", fq->fqid);
		return ret;
	}

	do {
		qman_poll_dqrr(16);
		qman_fq_state(fq, &state, &flags);
	} while (flags & QMAN_FQ_STATE_NE);

	return 0;
}

static int kill_fq(struct device *qidev, struct qman_fq *fq)
{
	enum qman_fq_state state;
	u32 flags;
	int ret;

	ret = qman_retire_fq(fq, &flags);
	if (ret < 0) {
		dev_err(qidev, "qman_retire_fq failed\n");
		return ret;
	}

	if (!ret)
		goto empty_fq;

	/* Async FQ retirement condition */
	if (1 == ret) {
		/* Retry till FQ gets in retired state */
		do {
			msleep(20);
			qman_fq_state(fq, &state, &flags);
		} while (qman_fq_state_retired != state);

		WARN_ON(flags & QMAN_FQ_STATE_BLOCKOOS);
		WARN_ON(flags & QMAN_FQ_STATE_ORL);
	}

empty_fq:
	if (flags & QMAN_FQ_STATE_NE) {
		ret = empty_retired_fq(qidev, fq);
		if (ret) {
			dev_err(qidev, "empty_retired_fq fail for FQ: %u\n",
				fq->fqid);
			return ret;
		}
	}

	ret = qman_oos_fq(fq);
	if (ret)
		dev_err(qidev, "OOS of FQID: %u failed\n", fq->fqid);

	qman_destroy_fq(fq, 0);

	return ret;
}

/*
 * TODO: This CAAM FQ empty logic can be improved. We can enqueue a NULL
 * job descriptor to the FQ. This must be the last enqueue request to the
 * FQ. When the response of this job comes back, the FQ is empty. Also
 * holding tanks are guaranteed to be not holding any jobs from this FQ.
 */
static int empty_caam_fq(struct qman_fq *fq)
{
	int ret;
	struct qm_mcr_queryfq_np np;

	/* Wait till the older CAAM FQ get empty */
	do {
		ret = qman_query_fq_np(fq, &np);
		if (ret)
			return ret;

		if (!np.frm_cnt)
			break;

		msleep(20);
	} while (1);

	/*
	 * Give extra time for pending jobs from this FQ in holding tanks
	 * to get processed
	 */
	msleep(20);
	return 0;
}

int caam_drv_ctx_update(struct caam_drv_ctx *drv_ctx, u32 *sh_desc)
{
	size_t size;
	u32 num_words;
	int ret;
	struct qman_fq *new_fq, *old_fq;
	struct device *qidev = drv_ctx->qidev;

	/* Check the size of new shared descriptor */
	num_words = desc_len(sh_desc);
	if (num_words > MAX_SDLEN) {
		dev_err(qidev, "Invalid descriptor len: %d words\n",
			num_words);
		return -EINVAL;
	}

	/* Note down older req FQ */
	old_fq = drv_ctx->req_fq;

	/* Create a new req FQ in parked state */
	new_fq = create_caam_req_fq(drv_ctx->qidev, drv_ctx->rsp_fq,
				    drv_ctx->context_a, 0);
	if (unlikely(IS_ERR_OR_NULL(new_fq))) {
		dev_err(qidev, "FQ allocation for shdesc update failed\n");
		return PTR_ERR(new_fq);
	}

	/* Hook up new FQ to context so that new requests keep queueing */
	drv_ctx->req_fq = new_fq;

	/* Empty and remove the older FQ */
	ret = empty_caam_fq(old_fq);
	if (ret) {
		dev_err(qidev, "Old SEC FQ empty failed\n");

		/* We can revert to older FQ */
		drv_ctx->req_fq = old_fq;

		if (kill_fq(qidev, new_fq)) {
			dev_warn(qidev, "New SEC FQ: %u kill failed\n",
				 new_fq->fqid);
		}

		return ret;
	}

	/*
	 * Now update the shared descriptor for driver context.
	 * Re-initialise pre-header. Set RSLS and SDLEN
	 */
	drv_ctx->prehdr[0] = (1 << PREHDR_RSLS_SHIFT) | num_words;

	/* Copy the new shared descriptor now */
	memcpy(drv_ctx->sh_desc, sh_desc, desc_bytes(sh_desc));

	size = sizeof(drv_ctx->sh_desc) + sizeof(drv_ctx->prehdr);
	dma_sync_single_for_device(qidev, drv_ctx->context_a,
				   size, DMA_BIDIRECTIONAL);

	/* Put the new FQ in scheduled state */
	ret = qman_schedule_fq(new_fq);
	if (ret) {
		dev_err(qidev, "Fail to sched new SEC FQ, ecode = %d\n", ret);

		/*
		 * We can kill new FQ and revert to old FQ.
		 * Since the desc is already modified, it is success case
		 */

		drv_ctx->req_fq = old_fq;

		if (kill_fq(qidev, new_fq)) {
			dev_warn(qidev, "New SEC FQ: %u kill failed\n",
				 new_fq->fqid);
		}
	} else {
		/* Remove older FQ */
		if (kill_fq(qidev, old_fq)) {
			dev_warn(qidev, "Old SEC FQ: %u kill failed\n",
				 old_fq->fqid);
		}
	}

	return 0;
}
EXPORT_SYMBOL(caam_drv_ctx_update);



struct caam_drv_ctx *caam_drv_ctx_init(struct device *qidev,
				       int *cpu,
				       u32 *sh_desc)
{
	size_t size;
	u32 num_words;
	dma_addr_t hwdesc;
	struct qman_fq *rsp_fq;
	struct caam_drv_ctx *drv_ctx;
	const cpumask_t *cpus = qman_affine_cpus();
	static DEFINE_PER_CPU(int, last_cpu);

	num_words = desc_len(sh_desc);
	if (num_words > MAX_SDLEN) {
		dev_err(qidev, "Invalid descriptor len: %d words\n",
			num_words);
		return ERR_PTR(-EINVAL);
	}

	drv_ctx = kzalloc(sizeof(*drv_ctx), GFP_ATOMIC);
	if (!drv_ctx) {
		dev_err(qidev, "Mem alloc for driver context failed\n");
		return ERR_PTR(-ENOMEM);
	}

	/* Initialise pre-header. Set RSLS and SDLEN */
	drv_ctx->prehdr[0] = (1 << PREHDR_RSLS_SHIFT) | num_words;

	/* Copy the shared descriptor now */
	memcpy(drv_ctx->sh_desc, sh_desc, desc_bytes(sh_desc));

	/* Map address for pre-header + descriptor */
	size = sizeof(drv_ctx->prehdr) + sizeof(drv_ctx->sh_desc);
	hwdesc = dma_map_single(qidev, drv_ctx->prehdr,
				size, DMA_BIDIRECTIONAL);
	if (dma_mapping_error(qidev, hwdesc)) {
		dev_err(qidev, "DMA map error for preheader+shdesc\n");
		kfree(drv_ctx);
		return ERR_PTR(-ENOMEM);
	}

	drv_ctx->context_a = hwdesc;

	/*
	 * If the given CPU does not own the portal, choose another
	 * one with a portal.
	 */
	if (!cpumask_test_cpu(*cpu, cpus)) {
		last_cpu = cpumask_next(last_cpu, cpus);
		if (last_cpu >= nr_cpu_ids)
			last_cpu = cpumask_first(cpus);
		 *cpu = last_cpu;
	}

	drv_ctx->cpu = *cpu;

	/* Find response FQ hooked with this CPU*/
	rsp_fq = &per_cpu(pcpu_qipriv.rsp_fq, drv_ctx->cpu);
	drv_ctx->rsp_fq = rsp_fq;

	/*Attach request FQ*/
	drv_ctx->req_fq = create_caam_req_fq(qidev, rsp_fq,
					     hwdesc, QMAN_INITFQ_FLAG_SCHED);
	if (unlikely(IS_ERR_OR_NULL(drv_ctx->req_fq))) {
		dev_err(qidev, "create_caam_req_fq failed\n");
		dma_unmap_single(qidev, hwdesc, size, DMA_BIDIRECTIONAL);
		kfree(drv_ctx);
		return ERR_PTR(-ENOMEM);
	}

	drv_ctx->qidev = qidev;
	return drv_ctx;
}
EXPORT_SYMBOL(caam_drv_ctx_init);

static int caam_qi_poll(struct napi_struct *napi, int budget)
{
	int cleaned = qman_poll_dqrr(budget);

	if (cleaned < budget) {
		napi_complete(napi);
		qman_irqsource_add(QM_PIRQ_DQRI);
	}

	return cleaned;
}


void caam_drv_ctx_rel(struct caam_drv_ctx *drv_ctx)
{
	size_t size;

	if (IS_ERR_OR_NULL(drv_ctx))
		return;

	size = sizeof(drv_ctx->sh_desc) + sizeof(drv_ctx->prehdr);

	/* Remove request FQ*/
	if (kill_fq(drv_ctx->qidev, drv_ctx->req_fq))
		dev_err(drv_ctx->qidev, "Crypto session Req FQ kill failed\n");

	dma_unmap_single(drv_ctx->qidev, drv_ctx->context_a,
			 size, DMA_BIDIRECTIONAL);

	kfree(drv_ctx);
}
EXPORT_SYMBOL(caam_drv_ctx_rel);

int caam_qi_shutdown(struct device *qidev)
{
	struct caam_qi_priv *priv = dev_get_drvdata(qidev);
	int i, ret;

	const cpumask_t *cpus = qman_affine_cpus();
	struct cpumask old_cpumask = *tsk_cpus_allowed(current);

	for_each_cpu(i, cpus) {
		napi_disable(&per_cpu(pcpu_qipriv.irqtask, i));
		netif_napi_del(&per_cpu(pcpu_qipriv.irqtask, i));
		if (kill_fq(qidev, &per_cpu(pcpu_qipriv.rsp_fq, i)))
			dev_err(qidev, "Rsp FQ kill failed, cpu: %d\n", i);
	}

	/*
	 * QMAN driver requires CGRs to be deleted from same CPU from where
	 * they were instantiated. Hence we get the module removal execute
	 * from the same CPU from where it was originally inserted.
	 */
	set_cpus_allowed_ptr(current, get_cpu_mask(mod_init_cpu));

	ret = qman_delete_cgr(&priv->req_cgr);
	if (ret)
		dev_err(qidev, "Delete request CGR failed: %d\n", ret);
	else
		qman_release_cgrid(priv->req_cgr.cgrid);

	ret = qman_delete_cgr(&priv->rsp_cgr);
	if (ret)
		dev_err(qidev, "Delete response CGR failed: %d\n", ret);
	else
		qman_release_cgrid(priv->rsp_cgr.cgrid);

	/* Now that we're done with the CGRs, restore the cpus allowed mask */
	set_cpus_allowed_ptr(current, &old_cpumask);

	platform_device_unregister(priv->qi_pdev);
	return ret;
}

static void rsp_cgr_cb(struct qman_portal *qm, struct qman_cgr *cgr,
			int congested)
{
	struct device *qidev = &per_cpu(pcpu_qipriv.net_dev,
					smp_processor_id()).dev;

	qipriv.cpu_congested = congested;

	if (congested)
		dev_warn(qidev, "CAAM rsp path congested\n");
	else
		dev_info(qidev, "CAAM rsp path congestion state exit\n");
}

static void req_cgr_cb(struct qman_portal *qm, struct qman_cgr *cgr,
			int congested)
{
	struct device *qidev = &per_cpu(pcpu_qipriv.net_dev,
					smp_processor_id()).dev;

	qipriv.sec_congested = congested;

	if (congested)
		dev_warn(qidev, "CAAM req path congested\n");
	else
		dev_info(qidev, "CAAM req path congestion state exit\n");
}

static int caam_qi_napi_schedule(struct napi_struct *napi)
{
	/*
	 * In case of threaded ISR for RT enable kernel,
	 * in_irq() does not return appropriate value, so use
	 * in_serving_softirq to distinguish softirq or irq context.
	 */
	if (unlikely(in_irq() || !in_serving_softirq())) {
		/* Disable QMan IRQ and invoke NAPI */
		int ret = qman_irqsource_remove(QM_PIRQ_DQRI);
		if (likely(!ret)) {
			napi_schedule(napi);
			return 1;
		}
	}
	return 0;
}

static enum qman_cb_dqrr_result caam_rsp_fq_dqrr_cb(struct qman_portal *p,
					struct qman_fq *rsp_fq,
					const struct qm_dqrr_entry *dqrr)
{
	struct napi_struct *napi;
	struct caam_drv_req *drv_req;
	const struct qm_fd *fd;
	size_t size;
	struct device *qidev = &per_cpu(pcpu_qipriv.net_dev,
					smp_processor_id()).dev;

	napi = &per_cpu(pcpu_qipriv.irqtask, smp_processor_id());
	if (caam_qi_napi_schedule(napi))
		return qman_cb_dqrr_stop;

	fd = &dqrr->fd;
	if (unlikely(fd->status))
		dev_err(qidev, "Error: %#x in CAAM response FD\n", fd->status);

	if (qm_fd_compound != fd->format) {
		dev_err(qidev, "Non compound FD from CAAM\n");
		return qman_cb_dqrr_consume;
	}

	drv_req = fd_to_drv_req(fd);
	if (!drv_req) {
		dev_err(qidev,
			"Can't find original request for caam response\n");
		return qman_cb_dqrr_consume;
	}

	size = 2 * sizeof(struct qm_sg_entry);
	dma_unmap_single(drv_req->drv_ctx->qidev, fd->addr,
			 size, DMA_BIDIRECTIONAL);

	drv_req->cbk(drv_req, fd->status);

	return qman_cb_dqrr_consume;
}

static int alloc_rsp_fq_cpu(struct device *qidev, unsigned int cpu)
{
	struct qm_mcc_initfq opts;
	struct qman_fq *fq;
	int ret;
	u32 flags;

	fq = &per_cpu(pcpu_qipriv.rsp_fq, cpu);

	fq->cb.dqrr = caam_rsp_fq_dqrr_cb;

	flags = QMAN_FQ_FLAG_NO_ENQUEUE |
		QMAN_FQ_FLAG_DYNAMIC_FQID;

	ret = qman_create_fq(0, flags, fq);
	if (ret) {
		dev_err(qidev, "Rsp FQ create failed\n");
		return -ENODEV;
	}

	flags = QMAN_INITFQ_FLAG_SCHED;

	opts.we_mask = QM_INITFQ_WE_FQCTRL | QM_INITFQ_WE_DESTWQ |
		QM_INITFQ_WE_CONTEXTB | QM_INITFQ_WE_CONTEXTA |
		QM_INITFQ_WE_CGID;

	opts.fqd.fq_ctrl = QM_FQCTRL_CTXASTASHING |
			   QM_FQCTRL_CPCSTASH |
			   QM_FQCTRL_CGE;

	opts.fqd.dest.channel = qman_affine_channel(cpu);
	opts.fqd.cgid = qipriv.rsp_cgr.cgrid;
	opts.fqd.dest.wq = 1;
	opts.fqd.context_a.stashing.exclusive =
					QM_STASHING_EXCL_CTX |
					QM_STASHING_EXCL_DATA;

	opts.fqd.context_a.stashing.data_cl = 1;
	opts.fqd.context_a.stashing.context_cl = 1;

	ret = qman_init_fq(fq, flags, &opts);
	if (ret) {
		dev_err(qidev, "Rsp FQ init failed\n");
		return -ENODEV;
	}

	return 0;
}

static int alloc_cgrs(struct device *qidev)
{
	struct qm_mcc_initcgr opts;
	int ret;

	/*Allocate response CGR*/
	ret = qman_alloc_cgrid(&qipriv.rsp_cgr.cgrid);
	if (ret) {
		dev_err(qidev, "CGR alloc failed for rsp FQs");
		return ret;
	}

	qipriv.rsp_cgr.cb = rsp_cgr_cb;
	memset(&opts, 0, sizeof(opts));
	opts.we_mask = QM_CGR_WE_CSCN_EN | QM_CGR_WE_CS_THRES |
			QM_CGR_WE_MODE;
	opts.cgr.cscn_en = QM_CGR_EN;
	opts.cgr.mode = QMAN_CGR_MODE_FRAME;
	qm_cgr_cs_thres_set64(&opts.cgr.cs_thres, 0x400 , 1);

	ret = qman_create_cgr(&qipriv.rsp_cgr,
				QMAN_CGR_FLAG_USE_INIT, &opts);
	if (ret) {
		dev_err(qidev, "Error %d creating CAAM rsp CGRID: %u\n",
			ret, qipriv.rsp_cgr.cgrid);
		goto create_rsp_cgr_fail;
	}

	/*Allocate request CGR*/
	ret = qman_alloc_cgrid(&qipriv.req_cgr.cgrid);
	if (ret) {
		dev_err(qidev, "CGR alloc failed for req FQs");
		goto alloc_req_cgrid_fail;
	}

	qipriv.req_cgr.cb = req_cgr_cb;
	memset(&opts, 0, sizeof(opts));
	opts.we_mask = QM_CGR_WE_CSCN_EN | QM_CGR_WE_CS_THRES;
	opts.cgr.cscn_en = QM_CGR_EN;
	qm_cgr_cs_thres_set64(&opts.cgr.cs_thres, CAAM_REQ_CGR_THRESHOLD , 1);

	ret = qman_create_cgr(&qipriv.req_cgr,
				QMAN_CGR_FLAG_USE_INIT, &opts);
	if (ret) {
		dev_err(qidev, "Error %d creating CAAM req CGRID: %u\n",
			ret, qipriv.req_cgr.cgrid);
		goto create_req_cgr_fail;
	}
	return 0;

create_req_cgr_fail:
	qman_release_cgrid(qipriv.req_cgr.cgrid);

alloc_req_cgrid_fail:
	qman_delete_cgr(&qipriv.rsp_cgr);

create_rsp_cgr_fail:
	qman_release_cgrid(qipriv.rsp_cgr.cgrid);

	return ret;
}

static int alloc_rsp_fqs(struct device *qidev)
{
	const cpumask_t *cpus = qman_affine_cpus();
	int ret, i;

	/*Now create response FQs*/
	for_each_cpu(i, cpus) {
		ret = alloc_rsp_fq_cpu(qidev, i);
		if (ret) {
			dev_err(qidev, "CAAM rsp FQ alloc failed, cpu: %u", i);
			return ret;
		}
	}

	return 0;
}

int caam_qi_init(struct platform_device *caam_pdev, struct device_node *np)
{
	struct platform_device *qi_pdev;
	struct device *ctrldev, *qidev;
	struct caam_drv_private *ctrlpriv;
	int err, i;
	const cpumask_t *cpus = qman_affine_cpus();
	struct cpumask old_cpumask = *tsk_cpus_allowed(current);

	/*
	 * QMAN requires that CGR must be removed from same CPU+portal from
	 * where it was originally allocated. Hence we need to note down
	 * the initialisation CPU and use the same CPU for module exit.
	 * We select the first CPU to from the list of portal owning
	 * CPUs. Then we pin module init to this CPU.
	 */
	mod_init_cpu = cpumask_first(cpus);
	set_cpus_allowed_ptr(current, get_cpu_mask(mod_init_cpu));

	qi_pdev = platform_device_register_simple("caam_qi", 0, NULL, 0);
	if (IS_ERR(qi_pdev))
		return PTR_ERR(qi_pdev);

	ctrldev = &caam_pdev->dev;
	ctrlpriv = dev_get_drvdata(ctrldev);
	qidev = &qi_pdev->dev;

	qipriv.qi_pdev = qi_pdev;
	dev_set_drvdata(qidev, &qipriv);

	/* Copy dma mask from controlling device */
	err = dma_set_mask(qidev, dma_get_mask(ctrldev));
	if (err) {
		platform_device_unregister(qi_pdev);
		return -ENODEV;
	}

	/* Initialise the CGRs congestion detection */
	err = alloc_cgrs(qidev);
	if (err) {
		dev_err(qidev, "Can't allocate CGRs\n");
		platform_device_unregister(qi_pdev);
		return err;
	}

	/* Initialise response FQs */
	err = alloc_rsp_fqs(qidev);
	if (err) {
		dev_err(qidev, "Can't allocate SEC response FQs\n");
		platform_device_unregister(qi_pdev);
		return err;
	}

	/*
	 * Enable the NAPI contexts on each of the core which has a affine
	 * portal.
	 */
	for_each_cpu(i, cpus) {
		per_cpu(pcpu_qipriv.net_dev, i).dev = *qidev;

		spin_lock_init(&per_cpu(pcpu_qipriv.listlock, i));
		INIT_LIST_HEAD(&per_cpu(pcpu_qipriv.bklog_list, i));

		INIT_LIST_HEAD(&per_cpu(pcpu_qipriv.net_dev, i).napi_list);

		netif_napi_add(&per_cpu(pcpu_qipriv.net_dev, i),
			       &per_cpu(pcpu_qipriv.irqtask, i),
			       caam_qi_poll, CAAM_NAPI_WEIGHT);

		napi_enable(&per_cpu(pcpu_qipriv.irqtask, i));
	}

	/* Hook up QI device to parent controlling caam device */
	ctrlpriv->qidev = qidev;

	/* Done with the CGRs; restore the cpus allowed mask */
	set_cpus_allowed_ptr(current, &old_cpumask);

	dev_info(qidev, "Linux CAAM Queue I/F driver initialised\n");

	return 0;
}