aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2012-07-03 18:45:10 -0400
committerLinus Torvalds <torvalds@linux-foundation.org>2012-07-03 18:45:10 -0400
commita3da2c6913469ecb2224d891c45470b37b4d67f4 (patch)
treecbedea8d0c8c101cfffccdbaf273e7adf92a4e46
parentff826b2b5b269ad440afa686ede879ccabfda387 (diff)
parent6d9359280753d2955f86d6411047516a9431eb51 (diff)
Merge branch 'for-linus' of git://git.kernel.dk/linux-block
Pull block bits from Jens Axboe: "As vacation is coming up, thought I'd better get rid of my pending changes in my for-linus branch for this iteration. It contains: - Two patches for mtip32xx. Killing a non-compliant sysfs interface and moving it to debugfs, where it belongs. - A few patches from Asias. Two legit bug fixes, and one killing an interface that is no longer in use. - A patch from Jan, making the annoying partition ioctl warning a bit less annoying, by restricting it to !CAP_SYS_RAWIO only. - Three bug fixes for drbd from Lars Ellenberg. - A fix for an old regression for umem, it hasn't really worked since the plugging scheme was changed in 3.0. - A few fixes from Tejun. - A splice fix from Eric Dumazet, fixing an issue with pipe resizing." * 'for-linus' of git://git.kernel.dk/linux-block: scsi: Silence unnecessary warnings about ioctl to partition block: Drop dead function blk_abort_queue() block: Mitigate lock unbalance caused by lock switching block: Avoid missed wakeup in request waitqueue umem: fix up unplugging splice: fix racy pipe->buffers uses drbd: fix null pointer dereference with on-congestion policy when diskless drbd: fix list corruption by failing but already aborted reads drbd: fix access of unallocated pages and kernel panic xen/blkfront: Add WARN to deal with misbehaving backends. blkcg: drop local variable @q from blkg_destroy() mtip32xx: Create debugfs entries for troubleshooting mtip32xx: Remove 'registers' and 'flags' from sysfs blkcg: fix blkg_alloc() failure path block: blkcg_policy_cfq shouldn't be used if !CONFIG_CFQ_GROUP_IOSCHED block: fix return value on cfq_init() failure mtip32xx: Remove version.h header file inclusion xen/blkback: Copy id field when doing BLKIF_DISCARD.
-rw-r--r--Documentation/ABI/testing/sysfs-block-rssd21
-rw-r--r--block/blk-cgroup.c9
-rw-r--r--block/blk-core.c25
-rw-r--r--block/blk-timeout.c41
-rw-r--r--block/cfq-iosched.c30
-rw-r--r--block/scsi_ioctl.c5
-rw-r--r--drivers/block/drbd/drbd_bitmap.c11
-rw-r--r--drivers/block/drbd/drbd_req.c66
-rw-r--r--drivers/block/mtip32xx/mtip32xx.c166
-rw-r--r--drivers/block/mtip32xx/mtip32xx.h5
-rw-r--r--drivers/block/umem.c40
-rw-r--r--drivers/block/xen-blkback/common.h2
-rw-r--r--drivers/block/xen-blkfront.c58
-rw-r--r--fs/splice.c35
-rw-r--r--include/linux/blkdev.h1
-rw-r--r--include/linux/splice.h8
-rw-r--r--kernel/relay.c5
-rw-r--r--kernel/trace/trace.c6
-rw-r--r--mm/shmem.c3
-rw-r--r--net/core/skbuff.c1
20 files changed, 338 insertions, 200 deletions
diff --git a/Documentation/ABI/testing/sysfs-block-rssd b/Documentation/ABI/testing/sysfs-block-rssd
index 679ce3543122..beef30c046b0 100644
--- a/Documentation/ABI/testing/sysfs-block-rssd
+++ b/Documentation/ABI/testing/sysfs-block-rssd
@@ -1,26 +1,5 @@
1What: /sys/block/rssd*/registers
2Date: March 2012
3KernelVersion: 3.3
4Contact: Asai Thambi S P <asamymuthupa@micron.com>
5Description: This is a read-only file. Dumps below driver information and
6 hardware registers.
7 - S ACTive
8 - Command Issue
9 - Completed
10 - PORT IRQ STAT
11 - HOST IRQ STAT
12 - Allocated
13 - Commands in Q
14
15What: /sys/block/rssd*/status 1What: /sys/block/rssd*/status
16Date: April 2012 2Date: April 2012
17KernelVersion: 3.4 3KernelVersion: 3.4
18Contact: Asai Thambi S P <asamymuthupa@micron.com> 4Contact: Asai Thambi S P <asamymuthupa@micron.com>
19Description: This is a read-only file. Indicates the status of the device. 5Description: This is a read-only file. Indicates the status of the device.
20
21What: /sys/block/rssd*/flags
22Date: May 2012
23KernelVersion: 3.5
24Contact: Asai Thambi S P <asamymuthupa@micron.com>
25Description: This is a read-only file. Dumps the flags in port and driver
26 data structure
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 02cf6335e9bd..e7dee617358e 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -125,12 +125,8 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct request_queue *q)
125 125
126 blkg->pd[i] = pd; 126 blkg->pd[i] = pd;
127 pd->blkg = blkg; 127 pd->blkg = blkg;
128 }
129
130 /* invoke per-policy init */
131 for (i = 0; i < BLKCG_MAX_POLS; i++) {
132 struct blkcg_policy *pol = blkcg_policy[i];
133 128
129 /* invoke per-policy init */
134 if (blkcg_policy_enabled(blkg->q, pol)) 130 if (blkcg_policy_enabled(blkg->q, pol))
135 pol->pd_init_fn(blkg); 131 pol->pd_init_fn(blkg);
136 } 132 }
@@ -245,10 +241,9 @@ EXPORT_SYMBOL_GPL(blkg_lookup_create);
245 241
246static void blkg_destroy(struct blkcg_gq *blkg) 242static void blkg_destroy(struct blkcg_gq *blkg)
247{ 243{
248 struct request_queue *q = blkg->q;
249 struct blkcg *blkcg = blkg->blkcg; 244 struct blkcg *blkcg = blkg->blkcg;
250 245
251 lockdep_assert_held(q->queue_lock); 246 lockdep_assert_held(blkg->q->queue_lock);
252 lockdep_assert_held(&blkcg->lock); 247 lockdep_assert_held(&blkcg->lock);
253 248
254 /* Something wrong if we are trying to remove same group twice */ 249 /* Something wrong if we are trying to remove same group twice */
diff --git a/block/blk-core.c b/block/blk-core.c
index 3c923a7aeb56..93eb3e4f88ce 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -361,9 +361,10 @@ EXPORT_SYMBOL(blk_put_queue);
361 */ 361 */
362void blk_drain_queue(struct request_queue *q, bool drain_all) 362void blk_drain_queue(struct request_queue *q, bool drain_all)
363{ 363{
364 int i;
365
364 while (true) { 366 while (true) {
365 bool drain = false; 367 bool drain = false;
366 int i;
367 368
368 spin_lock_irq(q->queue_lock); 369 spin_lock_irq(q->queue_lock);
369 370
@@ -408,6 +409,18 @@ void blk_drain_queue(struct request_queue *q, bool drain_all)
408 break; 409 break;
409 msleep(10); 410 msleep(10);
410 } 411 }
412
413 /*
414 * With queue marked dead, any woken up waiter will fail the
415 * allocation path, so the wakeup chaining is lost and we're
416 * left with hung waiters. We need to wake up those waiters.
417 */
418 if (q->request_fn) {
419 spin_lock_irq(q->queue_lock);
420 for (i = 0; i < ARRAY_SIZE(q->rq.wait); i++)
421 wake_up_all(&q->rq.wait[i]);
422 spin_unlock_irq(q->queue_lock);
423 }
411} 424}
412 425
413/** 426/**
@@ -467,7 +480,6 @@ void blk_cleanup_queue(struct request_queue *q)
467 /* mark @q DEAD, no new request or merges will be allowed afterwards */ 480 /* mark @q DEAD, no new request or merges will be allowed afterwards */
468 mutex_lock(&q->sysfs_lock); 481 mutex_lock(&q->sysfs_lock);
469 queue_flag_set_unlocked(QUEUE_FLAG_DEAD, q); 482 queue_flag_set_unlocked(QUEUE_FLAG_DEAD, q);
470
471 spin_lock_irq(lock); 483 spin_lock_irq(lock);
472 484
473 /* 485 /*
@@ -485,10 +497,6 @@ void blk_cleanup_queue(struct request_queue *q)
485 queue_flag_set(QUEUE_FLAG_NOMERGES, q); 497 queue_flag_set(QUEUE_FLAG_NOMERGES, q);
486 queue_flag_set(QUEUE_FLAG_NOXMERGES, q); 498 queue_flag_set(QUEUE_FLAG_NOXMERGES, q);
487 queue_flag_set(QUEUE_FLAG_DEAD, q); 499 queue_flag_set(QUEUE_FLAG_DEAD, q);
488
489 if (q->queue_lock != &q->__queue_lock)
490 q->queue_lock = &q->__queue_lock;
491
492 spin_unlock_irq(lock); 500 spin_unlock_irq(lock);
493 mutex_unlock(&q->sysfs_lock); 501 mutex_unlock(&q->sysfs_lock);
494 502
@@ -499,6 +507,11 @@ void blk_cleanup_queue(struct request_queue *q)
499 del_timer_sync(&q->backing_dev_info.laptop_mode_wb_timer); 507 del_timer_sync(&q->backing_dev_info.laptop_mode_wb_timer);
500 blk_sync_queue(q); 508 blk_sync_queue(q);
501 509
510 spin_lock_irq(lock);
511 if (q->queue_lock != &q->__queue_lock)
512 q->queue_lock = &q->__queue_lock;
513 spin_unlock_irq(lock);
514
502 /* @q is and will stay empty, shutdown and put */ 515 /* @q is and will stay empty, shutdown and put */
503 blk_put_queue(q); 516 blk_put_queue(q);
504} 517}
diff --git a/block/blk-timeout.c b/block/blk-timeout.c
index 780354888958..6e4744cbfb56 100644
--- a/block/blk-timeout.c
+++ b/block/blk-timeout.c
@@ -197,44 +197,3 @@ void blk_add_timer(struct request *req)
197 mod_timer(&q->timeout, expiry); 197 mod_timer(&q->timeout, expiry);
198} 198}
199 199
200/**
201 * blk_abort_queue -- Abort all request on given queue
202 * @queue: pointer to queue
203 *
204 */
205void blk_abort_queue(struct request_queue *q)
206{
207 unsigned long flags;
208 struct request *rq, *tmp;
209 LIST_HEAD(list);
210
211 /*
212 * Not a request based block device, nothing to abort
213 */
214 if (!q->request_fn)
215 return;
216
217 spin_lock_irqsave(q->queue_lock, flags);
218
219 elv_abort_queue(q);
220
221 /*
222 * Splice entries to local list, to avoid deadlocking if entries
223 * get readded to the timeout list by error handling
224 */
225 list_splice_init(&q->timeout_list, &list);
226
227 list_for_each_entry_safe(rq, tmp, &list, timeout_list)
228 blk_abort_request(rq);
229
230 /*
231 * Occasionally, blk_abort_request() will return without
232 * deleting the element from the list. Make sure we add those back
233 * instead of leaving them on the local stack list.
234 */
235 list_splice(&list, &q->timeout_list);
236
237 spin_unlock_irqrestore(q->queue_lock, flags);
238
239}
240EXPORT_SYMBOL_GPL(blk_abort_queue);
diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 673c977cc2bf..fb52df9744f5 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -17,8 +17,6 @@
17#include "blk.h" 17#include "blk.h"
18#include "blk-cgroup.h" 18#include "blk-cgroup.h"
19 19
20static struct blkcg_policy blkcg_policy_cfq __maybe_unused;
21
22/* 20/*
23 * tunables 21 * tunables
24 */ 22 */
@@ -418,11 +416,6 @@ static inline struct cfq_group *pd_to_cfqg(struct blkg_policy_data *pd)
418 return pd ? container_of(pd, struct cfq_group, pd) : NULL; 416 return pd ? container_of(pd, struct cfq_group, pd) : NULL;
419} 417}
420 418
421static inline struct cfq_group *blkg_to_cfqg(struct blkcg_gq *blkg)
422{
423 return pd_to_cfqg(blkg_to_pd(blkg, &blkcg_policy_cfq));
424}
425
426static inline struct blkcg_gq *cfqg_to_blkg(struct cfq_group *cfqg) 419static inline struct blkcg_gq *cfqg_to_blkg(struct cfq_group *cfqg)
427{ 420{
428 return pd_to_blkg(&cfqg->pd); 421 return pd_to_blkg(&cfqg->pd);
@@ -572,6 +565,13 @@ static inline void cfqg_stats_update_avg_queue_size(struct cfq_group *cfqg) { }
572 565
573#ifdef CONFIG_CFQ_GROUP_IOSCHED 566#ifdef CONFIG_CFQ_GROUP_IOSCHED
574 567
568static struct blkcg_policy blkcg_policy_cfq;
569
570static inline struct cfq_group *blkg_to_cfqg(struct blkcg_gq *blkg)
571{
572 return pd_to_cfqg(blkg_to_pd(blkg, &blkcg_policy_cfq));
573}
574
575static inline void cfqg_get(struct cfq_group *cfqg) 575static inline void cfqg_get(struct cfq_group *cfqg)
576{ 576{
577 return blkg_get(cfqg_to_blkg(cfqg)); 577 return blkg_get(cfqg_to_blkg(cfqg));
@@ -3951,10 +3951,11 @@ static void cfq_exit_queue(struct elevator_queue *e)
3951 3951
3952 cfq_shutdown_timer_wq(cfqd); 3952 cfq_shutdown_timer_wq(cfqd);
3953 3953
3954#ifndef CONFIG_CFQ_GROUP_IOSCHED 3954#ifdef CONFIG_CFQ_GROUP_IOSCHED
3955 blkcg_deactivate_policy(q, &blkcg_policy_cfq);
3956#else
3955 kfree(cfqd->root_group); 3957 kfree(cfqd->root_group);
3956#endif 3958#endif
3957 blkcg_deactivate_policy(q, &blkcg_policy_cfq);
3958 kfree(cfqd); 3959 kfree(cfqd);
3959} 3960}
3960 3961
@@ -4194,14 +4195,15 @@ static int __init cfq_init(void)
4194#ifdef CONFIG_CFQ_GROUP_IOSCHED 4195#ifdef CONFIG_CFQ_GROUP_IOSCHED
4195 if (!cfq_group_idle) 4196 if (!cfq_group_idle)
4196 cfq_group_idle = 1; 4197 cfq_group_idle = 1;
4197#else
4198 cfq_group_idle = 0;
4199#endif
4200 4198
4201 ret = blkcg_policy_register(&blkcg_policy_cfq); 4199 ret = blkcg_policy_register(&blkcg_policy_cfq);
4202 if (ret) 4200 if (ret)
4203 return ret; 4201 return ret;
4202#else
4203 cfq_group_idle = 0;
4204#endif
4204 4205
4206 ret = -ENOMEM;
4205 cfq_pool = KMEM_CACHE(cfq_queue, 0); 4207 cfq_pool = KMEM_CACHE(cfq_queue, 0);
4206 if (!cfq_pool) 4208 if (!cfq_pool)
4207 goto err_pol_unreg; 4209 goto err_pol_unreg;
@@ -4215,13 +4217,17 @@ static int __init cfq_init(void)
4215err_free_pool: 4217err_free_pool:
4216 kmem_cache_destroy(cfq_pool); 4218 kmem_cache_destroy(cfq_pool);
4217err_pol_unreg: 4219err_pol_unreg:
4220#ifdef CONFIG_CFQ_GROUP_IOSCHED
4218 blkcg_policy_unregister(&blkcg_policy_cfq); 4221 blkcg_policy_unregister(&blkcg_policy_cfq);
4222#endif
4219 return ret; 4223 return ret;
4220} 4224}
4221 4225
4222static void __exit cfq_exit(void) 4226static void __exit cfq_exit(void)
4223{ 4227{
4228#ifdef CONFIG_CFQ_GROUP_IOSCHED
4224 blkcg_policy_unregister(&blkcg_policy_cfq); 4229 blkcg_policy_unregister(&blkcg_policy_cfq);
4230#endif
4225 elv_unregister(&iosched_cfq); 4231 elv_unregister(&iosched_cfq);
4226 kmem_cache_destroy(cfq_pool); 4232 kmem_cache_destroy(cfq_pool);
4227} 4233}
diff --git a/block/scsi_ioctl.c b/block/scsi_ioctl.c
index 260fa80ef575..9a87daa6f4fb 100644
--- a/block/scsi_ioctl.c
+++ b/block/scsi_ioctl.c
@@ -721,11 +721,14 @@ int scsi_verify_blk_ioctl(struct block_device *bd, unsigned int cmd)
721 break; 721 break;
722 } 722 }
723 723
724 if (capable(CAP_SYS_RAWIO))
725 return 0;
726
724 /* In particular, rule out all resets and host-specific ioctls. */ 727 /* In particular, rule out all resets and host-specific ioctls. */
725 printk_ratelimited(KERN_WARNING 728 printk_ratelimited(KERN_WARNING
726 "%s: sending ioctl %x to a partition!\n", current->comm, cmd); 729 "%s: sending ioctl %x to a partition!\n", current->comm, cmd);
727 730
728 return capable(CAP_SYS_RAWIO) ? 0 : -ENOIOCTLCMD; 731 return -ENOIOCTLCMD;
729} 732}
730EXPORT_SYMBOL(scsi_verify_blk_ioctl); 733EXPORT_SYMBOL(scsi_verify_blk_ioctl);
731 734
diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c
index b5c5ff53cb57..fcb956bb4b4c 100644
--- a/drivers/block/drbd/drbd_bitmap.c
+++ b/drivers/block/drbd/drbd_bitmap.c
@@ -1475,10 +1475,17 @@ void _drbd_bm_set_bits(struct drbd_conf *mdev, const unsigned long s, const unsi
1475 first_word = 0; 1475 first_word = 0;
1476 spin_lock_irq(&b->bm_lock); 1476 spin_lock_irq(&b->bm_lock);
1477 } 1477 }
1478
1479 /* last page (respectively only page, for first page == last page) */ 1478 /* last page (respectively only page, for first page == last page) */
1480 last_word = MLPP(el >> LN2_BPL); 1479 last_word = MLPP(el >> LN2_BPL);
1481 bm_set_full_words_within_one_page(mdev->bitmap, last_page, first_word, last_word); 1480
1481 /* consider bitmap->bm_bits = 32768, bitmap->bm_number_of_pages = 1. (or multiples).
1482 * ==> e = 32767, el = 32768, last_page = 2,
1483 * and now last_word = 0.
1484 * We do not want to touch last_page in this case,
1485 * as we did not allocate it, it is not present in bitmap->bm_pages.
1486 */
1487 if (last_word)
1488 bm_set_full_words_within_one_page(mdev->bitmap, last_page, first_word, last_word);
1482 1489
1483 /* possibly trailing bits. 1490 /* possibly trailing bits.
1484 * example: (e & 63) == 63, el will be e+1. 1491 * example: (e & 63) == 63, el will be e+1.
diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c
index 9c5c84946b05..8e93a6ac9bb6 100644
--- a/drivers/block/drbd/drbd_req.c
+++ b/drivers/block/drbd/drbd_req.c
@@ -472,12 +472,17 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what,
472 req->rq_state |= RQ_LOCAL_COMPLETED; 472 req->rq_state |= RQ_LOCAL_COMPLETED;
473 req->rq_state &= ~RQ_LOCAL_PENDING; 473 req->rq_state &= ~RQ_LOCAL_PENDING;
474 474
475 D_ASSERT(!(req->rq_state & RQ_NET_MASK)); 475 if (req->rq_state & RQ_LOCAL_ABORTED) {
476 _req_may_be_done(req, m);
477 break;
478 }
476 479
477 __drbd_chk_io_error(mdev, false); 480 __drbd_chk_io_error(mdev, false);
478 481
479 goto_queue_for_net_read: 482 goto_queue_for_net_read:
480 483
484 D_ASSERT(!(req->rq_state & RQ_NET_MASK));
485
481 /* no point in retrying if there is no good remote data, 486 /* no point in retrying if there is no good remote data,
482 * or we have no connection. */ 487 * or we have no connection. */
483 if (mdev->state.pdsk != D_UP_TO_DATE) { 488 if (mdev->state.pdsk != D_UP_TO_DATE) {
@@ -765,6 +770,40 @@ static int drbd_may_do_local_read(struct drbd_conf *mdev, sector_t sector, int s
765 return 0 == drbd_bm_count_bits(mdev, sbnr, ebnr); 770 return 0 == drbd_bm_count_bits(mdev, sbnr, ebnr);
766} 771}
767 772
773static void maybe_pull_ahead(struct drbd_conf *mdev)
774{
775 int congested = 0;
776
777 /* If I don't even have good local storage, we can not reasonably try
778 * to pull ahead of the peer. We also need the local reference to make
779 * sure mdev->act_log is there.
780 * Note: caller has to make sure that net_conf is there.
781 */
782 if (!get_ldev_if_state(mdev, D_UP_TO_DATE))
783 return;
784
785 if (mdev->net_conf->cong_fill &&
786 atomic_read(&mdev->ap_in_flight) >= mdev->net_conf->cong_fill) {
787 dev_info(DEV, "Congestion-fill threshold reached\n");
788 congested = 1;
789 }
790
791 if (mdev->act_log->used >= mdev->net_conf->cong_extents) {
792 dev_info(DEV, "Congestion-extents threshold reached\n");
793 congested = 1;
794 }
795
796 if (congested) {
797 queue_barrier(mdev); /* last barrier, after mirrored writes */
798
799 if (mdev->net_conf->on_congestion == OC_PULL_AHEAD)
800 _drbd_set_state(_NS(mdev, conn, C_AHEAD), 0, NULL);
801 else /*mdev->net_conf->on_congestion == OC_DISCONNECT */
802 _drbd_set_state(_NS(mdev, conn, C_DISCONNECTING), 0, NULL);
803 }
804 put_ldev(mdev);
805}
806
768static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio, unsigned long start_time) 807static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio, unsigned long start_time)
769{ 808{
770 const int rw = bio_rw(bio); 809 const int rw = bio_rw(bio);
@@ -972,29 +1011,8 @@ allocate_barrier:
972 _req_mod(req, queue_for_send_oos); 1011 _req_mod(req, queue_for_send_oos);
973 1012
974 if (remote && 1013 if (remote &&
975 mdev->net_conf->on_congestion != OC_BLOCK && mdev->agreed_pro_version >= 96) { 1014 mdev->net_conf->on_congestion != OC_BLOCK && mdev->agreed_pro_version >= 96)
976 int congested = 0; 1015 maybe_pull_ahead(mdev);
977
978 if (mdev->net_conf->cong_fill &&
979 atomic_read(&mdev->ap_in_flight) >= mdev->net_conf->cong_fill) {
980 dev_info(DEV, "Congestion-fill threshold reached\n");
981 congested = 1;
982 }
983
984 if (mdev->act_log->used >= mdev->net_conf->cong_extents) {
985 dev_info(DEV, "Congestion-extents threshold reached\n");
986 congested = 1;
987 }
988
989 if (congested) {
990 queue_barrier(mdev); /* last barrier, after mirrored writes */
991
992 if (mdev->net_conf->on_congestion == OC_PULL_AHEAD)
993 _drbd_set_state(_NS(mdev, conn, C_AHEAD), 0, NULL);
994 else /*mdev->net_conf->on_congestion == OC_DISCONNECT */
995 _drbd_set_state(_NS(mdev, conn, C_DISCONNECTING), 0, NULL);
996 }
997 }
998 1016
999 spin_unlock_irq(&mdev->req_lock); 1017 spin_unlock_irq(&mdev->req_lock);
1000 kfree(b); /* if someone else has beaten us to it... */ 1018 kfree(b); /* if someone else has beaten us to it... */
diff --git a/drivers/block/mtip32xx/mtip32xx.c b/drivers/block/mtip32xx/mtip32xx.c
index 264bc77dcb91..a8fddeb3d638 100644
--- a/drivers/block/mtip32xx/mtip32xx.c
+++ b/drivers/block/mtip32xx/mtip32xx.c
@@ -37,6 +37,7 @@
37#include <linux/kthread.h> 37#include <linux/kthread.h>
38#include <../drivers/ata/ahci.h> 38#include <../drivers/ata/ahci.h>
39#include <linux/export.h> 39#include <linux/export.h>
40#include <linux/debugfs.h>
40#include "mtip32xx.h" 41#include "mtip32xx.h"
41 42
42#define HW_CMD_SLOT_SZ (MTIP_MAX_COMMAND_SLOTS * 32) 43#define HW_CMD_SLOT_SZ (MTIP_MAX_COMMAND_SLOTS * 32)
@@ -85,6 +86,7 @@ static int instance;
85 * allocated in mtip_init(). 86 * allocated in mtip_init().
86 */ 87 */
87static int mtip_major; 88static int mtip_major;
89static struct dentry *dfs_parent;
88 90
89static DEFINE_SPINLOCK(rssd_index_lock); 91static DEFINE_SPINLOCK(rssd_index_lock);
90static DEFINE_IDA(rssd_index_ida); 92static DEFINE_IDA(rssd_index_ida);
@@ -2546,7 +2548,7 @@ static struct scatterlist *mtip_hw_get_scatterlist(struct driver_data *dd,
2546} 2548}
2547 2549
2548/* 2550/*
2549 * Sysfs register/status dump. 2551 * Sysfs status dump.
2550 * 2552 *
2551 * @dev Pointer to the device structure, passed by the kernrel. 2553 * @dev Pointer to the device structure, passed by the kernrel.
2552 * @attr Pointer to the device_attribute structure passed by the kernel. 2554 * @attr Pointer to the device_attribute structure passed by the kernel.
@@ -2555,45 +2557,68 @@ static struct scatterlist *mtip_hw_get_scatterlist(struct driver_data *dd,
2555 * return value 2557 * return value
2556 * The size, in bytes, of the data copied into buf. 2558 * The size, in bytes, of the data copied into buf.
2557 */ 2559 */
2558static ssize_t mtip_hw_show_registers(struct device *dev, 2560static ssize_t mtip_hw_show_status(struct device *dev,
2559 struct device_attribute *attr, 2561 struct device_attribute *attr,
2560 char *buf) 2562 char *buf)
2561{ 2563{
2562 u32 group_allocated;
2563 struct driver_data *dd = dev_to_disk(dev)->private_data; 2564 struct driver_data *dd = dev_to_disk(dev)->private_data;
2564 int size = 0; 2565 int size = 0;
2566
2567 if (test_bit(MTIP_DDF_OVER_TEMP_BIT, &dd->dd_flag))
2568 size += sprintf(buf, "%s", "thermal_shutdown\n");
2569 else if (test_bit(MTIP_DDF_WRITE_PROTECT_BIT, &dd->dd_flag))
2570 size += sprintf(buf, "%s", "write_protect\n");
2571 else
2572 size += sprintf(buf, "%s", "online\n");
2573
2574 return size;
2575}
2576
2577static DEVICE_ATTR(status, S_IRUGO, mtip_hw_show_status, NULL);
2578
2579static ssize_t mtip_hw_read_registers(struct file *f, char __user *ubuf,
2580 size_t len, loff_t *offset)
2581{
2582 struct driver_data *dd = (struct driver_data *)f->private_data;
2583 char buf[MTIP_DFS_MAX_BUF_SIZE];
2584 u32 group_allocated;
2585 int size = *offset;
2565 int n; 2586 int n;
2566 2587
2567 size += sprintf(&buf[size], "Hardware\n--------\n"); 2588 if (!len || size)
2568 size += sprintf(&buf[size], "S ACTive : [ 0x"); 2589 return 0;
2590
2591 if (size < 0)
2592 return -EINVAL;
2593
2594 size += sprintf(&buf[size], "H/ S ACTive : [ 0x");
2569 2595
2570 for (n = dd->slot_groups-1; n >= 0; n--) 2596 for (n = dd->slot_groups-1; n >= 0; n--)
2571 size += sprintf(&buf[size], "%08X ", 2597 size += sprintf(&buf[size], "%08X ",
2572 readl(dd->port->s_active[n])); 2598 readl(dd->port->s_active[n]));
2573 2599
2574 size += sprintf(&buf[size], "]\n"); 2600 size += sprintf(&buf[size], "]\n");
2575 size += sprintf(&buf[size], "Command Issue : [ 0x"); 2601 size += sprintf(&buf[size], "H/ Command Issue : [ 0x");
2576 2602
2577 for (n = dd->slot_groups-1; n >= 0; n--) 2603 for (n = dd->slot_groups-1; n >= 0; n--)
2578 size += sprintf(&buf[size], "%08X ", 2604 size += sprintf(&buf[size], "%08X ",
2579 readl(dd->port->cmd_issue[n])); 2605 readl(dd->port->cmd_issue[n]));
2580 2606
2581 size += sprintf(&buf[size], "]\n"); 2607 size += sprintf(&buf[size], "]\n");
2582 size += sprintf(&buf[size], "Completed : [ 0x"); 2608 size += sprintf(&buf[size], "H/ Completed : [ 0x");
2583 2609
2584 for (n = dd->slot_groups-1; n >= 0; n--) 2610 for (n = dd->slot_groups-1; n >= 0; n--)
2585 size += sprintf(&buf[size], "%08X ", 2611 size += sprintf(&buf[size], "%08X ",
2586 readl(dd->port->completed[n])); 2612 readl(dd->port->completed[n]));
2587 2613
2588 size += sprintf(&buf[size], "]\n"); 2614 size += sprintf(&buf[size], "]\n");
2589 size += sprintf(&buf[size], "PORT IRQ STAT : [ 0x%08X ]\n", 2615 size += sprintf(&buf[size], "H/ PORT IRQ STAT : [ 0x%08X ]\n",
2590 readl(dd->port->mmio + PORT_IRQ_STAT)); 2616 readl(dd->port->mmio + PORT_IRQ_STAT));
2591 size += sprintf(&buf[size], "HOST IRQ STAT : [ 0x%08X ]\n", 2617 size += sprintf(&buf[size], "H/ HOST IRQ STAT : [ 0x%08X ]\n",
2592 readl(dd->mmio + HOST_IRQ_STAT)); 2618 readl(dd->mmio + HOST_IRQ_STAT));
2593 size += sprintf(&buf[size], "\n"); 2619 size += sprintf(&buf[size], "\n");
2594 2620
2595 size += sprintf(&buf[size], "Local\n-----\n"); 2621 size += sprintf(&buf[size], "L/ Allocated : [ 0x");
2596 size += sprintf(&buf[size], "Allocated : [ 0x");
2597 2622
2598 for (n = dd->slot_groups-1; n >= 0; n--) { 2623 for (n = dd->slot_groups-1; n >= 0; n--) {
2599 if (sizeof(long) > sizeof(u32)) 2624 if (sizeof(long) > sizeof(u32))
@@ -2605,7 +2630,7 @@ static ssize_t mtip_hw_show_registers(struct device *dev,
2605 } 2630 }
2606 size += sprintf(&buf[size], "]\n"); 2631 size += sprintf(&buf[size], "]\n");
2607 2632
2608 size += sprintf(&buf[size], "Commands in Q: [ 0x"); 2633 size += sprintf(&buf[size], "L/ Commands in Q : [ 0x");
2609 2634
2610 for (n = dd->slot_groups-1; n >= 0; n--) { 2635 for (n = dd->slot_groups-1; n >= 0; n--) {
2611 if (sizeof(long) > sizeof(u32)) 2636 if (sizeof(long) > sizeof(u32))
@@ -2617,44 +2642,53 @@ static ssize_t mtip_hw_show_registers(struct device *dev,
2617 } 2642 }
2618 size += sprintf(&buf[size], "]\n"); 2643 size += sprintf(&buf[size], "]\n");
2619 2644
2620 return size; 2645 *offset = size <= len ? size : len;
2646 size = copy_to_user(ubuf, buf, *offset);
2647 if (size)
2648 return -EFAULT;
2649
2650 return *offset;
2621} 2651}
2622 2652
2623static ssize_t mtip_hw_show_status(struct device *dev, 2653static ssize_t mtip_hw_read_flags(struct file *f, char __user *ubuf,
2624 struct device_attribute *attr, 2654 size_t len, loff_t *offset)
2625 char *buf)
2626{ 2655{
2627 struct driver_data *dd = dev_to_disk(dev)->private_data; 2656 struct driver_data *dd = (struct driver_data *)f->private_data;
2628 int size = 0; 2657 char buf[MTIP_DFS_MAX_BUF_SIZE];
2658 int size = *offset;
2629 2659
2630 if (test_bit(MTIP_DDF_OVER_TEMP_BIT, &dd->dd_flag)) 2660 if (!len || size)
2631 size += sprintf(buf, "%s", "thermal_shutdown\n"); 2661 return 0;
2632 else if (test_bit(MTIP_DDF_WRITE_PROTECT_BIT, &dd->dd_flag))
2633 size += sprintf(buf, "%s", "write_protect\n");
2634 else
2635 size += sprintf(buf, "%s", "online\n");
2636
2637 return size;
2638}
2639 2662
2640static ssize_t mtip_hw_show_flags(struct device *dev, 2663 if (size < 0)
2641 struct device_attribute *attr, 2664 return -EINVAL;
2642 char *buf)
2643{
2644 struct driver_data *dd = dev_to_disk(dev)->private_data;
2645 int size = 0;
2646 2665
2647 size += sprintf(&buf[size], "Flag in port struct : [ %08lX ]\n", 2666 size += sprintf(&buf[size], "Flag-port : [ %08lX ]\n",
2648 dd->port->flags); 2667 dd->port->flags);
2649 size += sprintf(&buf[size], "Flag in dd struct : [ %08lX ]\n", 2668 size += sprintf(&buf[size], "Flag-dd : [ %08lX ]\n",
2650 dd->dd_flag); 2669 dd->dd_flag);
2651 2670
2652 return size; 2671 *offset = size <= len ? size : len;
2672 size = copy_to_user(ubuf, buf, *offset);
2673 if (size)
2674 return -EFAULT;
2675
2676 return *offset;
2653} 2677}
2654 2678
2655static DEVICE_ATTR(registers, S_IRUGO, mtip_hw_show_registers, NULL); 2679static const struct file_operations mtip_regs_fops = {
2656static DEVICE_ATTR(status, S_IRUGO, mtip_hw_show_status, NULL); 2680 .owner = THIS_MODULE,
2657static DEVICE_ATTR(flags, S_IRUGO, mtip_hw_show_flags, NULL); 2681 .open = simple_open,
2682 .read = mtip_hw_read_registers,
2683 .llseek = no_llseek,
2684};
2685
2686static const struct file_operations mtip_flags_fops = {
2687 .owner = THIS_MODULE,
2688 .open = simple_open,
2689 .read = mtip_hw_read_flags,
2690 .llseek = no_llseek,
2691};
2658 2692
2659/* 2693/*
2660 * Create the sysfs related attributes. 2694 * Create the sysfs related attributes.
@@ -2671,15 +2705,9 @@ static int mtip_hw_sysfs_init(struct driver_data *dd, struct kobject *kobj)
2671 if (!kobj || !dd) 2705 if (!kobj || !dd)
2672 return -EINVAL; 2706 return -EINVAL;
2673 2707
2674 if (sysfs_create_file(kobj, &dev_attr_registers.attr))
2675 dev_warn(&dd->pdev->dev,
2676 "Error creating 'registers' sysfs entry\n");
2677 if (sysfs_create_file(kobj, &dev_attr_status.attr)) 2708 if (sysfs_create_file(kobj, &dev_attr_status.attr))
2678 dev_warn(&dd->pdev->dev, 2709 dev_warn(&dd->pdev->dev,
2679 "Error creating 'status' sysfs entry\n"); 2710 "Error creating 'status' sysfs entry\n");
2680 if (sysfs_create_file(kobj, &dev_attr_flags.attr))
2681 dev_warn(&dd->pdev->dev,
2682 "Error creating 'flags' sysfs entry\n");
2683 return 0; 2711 return 0;
2684} 2712}
2685 2713
@@ -2698,13 +2726,39 @@ static int mtip_hw_sysfs_exit(struct driver_data *dd, struct kobject *kobj)
2698 if (!kobj || !dd) 2726 if (!kobj || !dd)
2699 return -EINVAL; 2727 return -EINVAL;
2700 2728
2701 sysfs_remove_file(kobj, &dev_attr_registers.attr);
2702 sysfs_remove_file(kobj, &dev_attr_status.attr); 2729 sysfs_remove_file(kobj, &dev_attr_status.attr);
2703 sysfs_remove_file(kobj, &dev_attr_flags.attr);
2704 2730
2705 return 0; 2731 return 0;
2706} 2732}
2707 2733
2734static int mtip_hw_debugfs_init(struct driver_data *dd)
2735{
2736 if (!dfs_parent)
2737 return -1;
2738
2739 dd->dfs_node = debugfs_create_dir(dd->disk->disk_name, dfs_parent);
2740 if (IS_ERR_OR_NULL(dd->dfs_node)) {
2741 dev_warn(&dd->pdev->dev,
2742 "Error creating node %s under debugfs\n",
2743 dd->disk->disk_name);
2744 dd->dfs_node = NULL;
2745 return -1;
2746 }
2747
2748 debugfs_create_file("flags", S_IRUGO, dd->dfs_node, dd,
2749 &mtip_flags_fops);
2750 debugfs_create_file("registers", S_IRUGO, dd->dfs_node, dd,
2751 &mtip_regs_fops);
2752
2753 return 0;
2754}
2755
2756static void mtip_hw_debugfs_exit(struct driver_data *dd)
2757{
2758 debugfs_remove_recursive(dd->dfs_node);
2759}
2760
2761
2708/* 2762/*
2709 * Perform any init/resume time hardware setup 2763 * Perform any init/resume time hardware setup
2710 * 2764 *
@@ -3730,6 +3784,7 @@ skip_create_disk:
3730 mtip_hw_sysfs_init(dd, kobj); 3784 mtip_hw_sysfs_init(dd, kobj);
3731 kobject_put(kobj); 3785 kobject_put(kobj);
3732 } 3786 }
3787 mtip_hw_debugfs_init(dd);
3733 3788
3734 if (dd->mtip_svc_handler) { 3789 if (dd->mtip_svc_handler) {
3735 set_bit(MTIP_DDF_INIT_DONE_BIT, &dd->dd_flag); 3790 set_bit(MTIP_DDF_INIT_DONE_BIT, &dd->dd_flag);
@@ -3755,6 +3810,8 @@ start_service_thread:
3755 return rv; 3810 return rv;
3756 3811
3757kthread_run_error: 3812kthread_run_error:
3813 mtip_hw_debugfs_exit(dd);
3814
3758 /* Delete our gendisk. This also removes the device from /dev */ 3815 /* Delete our gendisk. This also removes the device from /dev */
3759 del_gendisk(dd->disk); 3816 del_gendisk(dd->disk);
3760 3817
@@ -3805,6 +3862,7 @@ static int mtip_block_remove(struct driver_data *dd)
3805 kobject_put(kobj); 3862 kobject_put(kobj);
3806 } 3863 }
3807 } 3864 }
3865 mtip_hw_debugfs_exit(dd);
3808 3866
3809 /* 3867 /*
3810 * Delete our gendisk structure. This also removes the device 3868 * Delete our gendisk structure. This also removes the device
@@ -4152,10 +4210,20 @@ static int __init mtip_init(void)
4152 } 4210 }
4153 mtip_major = error; 4211 mtip_major = error;
4154 4212
4213 if (!dfs_parent) {
4214 dfs_parent = debugfs_create_dir("rssd", NULL);
4215 if (IS_ERR_OR_NULL(dfs_parent)) {
4216 printk(KERN_WARNING "Error creating debugfs parent\n");
4217 dfs_parent = NULL;
4218 }
4219 }
4220
4155 /* Register our PCI operations. */ 4221 /* Register our PCI operations. */
4156 error = pci_register_driver(&mtip_pci_driver); 4222 error = pci_register_driver(&mtip_pci_driver);
4157 if (error) 4223 if (error) {
4224 debugfs_remove(dfs_parent);
4158 unregister_blkdev(mtip_major, MTIP_DRV_NAME); 4225 unregister_blkdev(mtip_major, MTIP_DRV_NAME);
4226 }
4159 4227
4160 return error; 4228 return error;
4161} 4229}
@@ -4172,6 +4240,8 @@ static int __init mtip_init(void)
4172 */ 4240 */
4173static void __exit mtip_exit(void) 4241static void __exit mtip_exit(void)
4174{ 4242{
4243 debugfs_remove_recursive(dfs_parent);
4244
4175 /* Release the allocated major block device number. */ 4245 /* Release the allocated major block device number. */
4176 unregister_blkdev(mtip_major, MTIP_DRV_NAME); 4246 unregister_blkdev(mtip_major, MTIP_DRV_NAME);
4177 4247
diff --git a/drivers/block/mtip32xx/mtip32xx.h b/drivers/block/mtip32xx/mtip32xx.h
index b2c88da26b2a..f51fc23d17bb 100644
--- a/drivers/block/mtip32xx/mtip32xx.h
+++ b/drivers/block/mtip32xx/mtip32xx.h
@@ -26,7 +26,6 @@
26#include <linux/ata.h> 26#include <linux/ata.h>
27#include <linux/interrupt.h> 27#include <linux/interrupt.h>
28#include <linux/genhd.h> 28#include <linux/genhd.h>
29#include <linux/version.h>
30 29
31/* Offset of Subsystem Device ID in pci confoguration space */ 30/* Offset of Subsystem Device ID in pci confoguration space */
32#define PCI_SUBSYSTEM_DEVICEID 0x2E 31#define PCI_SUBSYSTEM_DEVICEID 0x2E
@@ -111,6 +110,8 @@
111 #define dbg_printk(format, arg...) 110 #define dbg_printk(format, arg...)
112#endif 111#endif
113 112
113#define MTIP_DFS_MAX_BUF_SIZE 1024
114
114#define __force_bit2int (unsigned int __force) 115#define __force_bit2int (unsigned int __force)
115 116
116enum { 117enum {
@@ -447,6 +448,8 @@ struct driver_data {
447 unsigned long dd_flag; /* NOTE: use atomic bit operations on this */ 448 unsigned long dd_flag; /* NOTE: use atomic bit operations on this */
448 449
449 struct task_struct *mtip_svc_handler; /* task_struct of svc thd */ 450 struct task_struct *mtip_svc_handler; /* task_struct of svc thd */
451
452 struct dentry *dfs_node;
450}; 453};
451 454
452#endif 455#endif
diff --git a/drivers/block/umem.c b/drivers/block/umem.c
index aa2712060bfb..9a72277a31df 100644
--- a/drivers/block/umem.c
+++ b/drivers/block/umem.c
@@ -513,6 +513,44 @@ static void process_page(unsigned long data)
513 } 513 }
514} 514}
515 515
516struct mm_plug_cb {
517 struct blk_plug_cb cb;
518 struct cardinfo *card;
519};
520
521static void mm_unplug(struct blk_plug_cb *cb)
522{
523 struct mm_plug_cb *mmcb = container_of(cb, struct mm_plug_cb, cb);
524
525 spin_lock_irq(&mmcb->card->lock);
526 activate(mmcb->card);
527 spin_unlock_irq(&mmcb->card->lock);
528 kfree(mmcb);
529}
530
531static int mm_check_plugged(struct cardinfo *card)
532{
533 struct blk_plug *plug = current->plug;
534 struct mm_plug_cb *mmcb;
535
536 if (!plug)
537 return 0;
538
539 list_for_each_entry(mmcb, &plug->cb_list, cb.list) {
540 if (mmcb->cb.callback == mm_unplug && mmcb->card == card)
541 return 1;
542 }
543 /* Not currently on the callback list */
544 mmcb = kmalloc(sizeof(*mmcb), GFP_ATOMIC);
545 if (!mmcb)
546 return 0;
547
548 mmcb->card = card;
549 mmcb->cb.callback = mm_unplug;
550 list_add(&mmcb->cb.list, &plug->cb_list);
551 return 1;
552}
553
516static void mm_make_request(struct request_queue *q, struct bio *bio) 554static void mm_make_request(struct request_queue *q, struct bio *bio)
517{ 555{
518 struct cardinfo *card = q->queuedata; 556 struct cardinfo *card = q->queuedata;
@@ -523,6 +561,8 @@ static void mm_make_request(struct request_queue *q, struct bio *bio)
523 *card->biotail = bio; 561 *card->biotail = bio;
524 bio->bi_next = NULL; 562 bio->bi_next = NULL;
525 card->biotail = &bio->bi_next; 563 card->biotail = &bio->bi_next;
564 if (bio->bi_rw & REQ_SYNC || !mm_check_plugged(card))
565 activate(card);
526 spin_unlock_irq(&card->lock); 566 spin_unlock_irq(&card->lock);
527 567
528 return; 568 return;
diff --git a/drivers/block/xen-blkback/common.h b/drivers/block/xen-blkback/common.h
index 773cf27dc23f..9ad3b5ec1dc1 100644
--- a/drivers/block/xen-blkback/common.h
+++ b/drivers/block/xen-blkback/common.h
@@ -257,6 +257,7 @@ static inline void blkif_get_x86_32_req(struct blkif_request *dst,
257 break; 257 break;
258 case BLKIF_OP_DISCARD: 258 case BLKIF_OP_DISCARD:
259 dst->u.discard.flag = src->u.discard.flag; 259 dst->u.discard.flag = src->u.discard.flag;
260 dst->u.discard.id = src->u.discard.id;
260 dst->u.discard.sector_number = src->u.discard.sector_number; 261 dst->u.discard.sector_number = src->u.discard.sector_number;
261 dst->u.discard.nr_sectors = src->u.discard.nr_sectors; 262 dst->u.discard.nr_sectors = src->u.discard.nr_sectors;
262 break; 263 break;
@@ -287,6 +288,7 @@ static inline void blkif_get_x86_64_req(struct blkif_request *dst,
287 break; 288 break;
288 case BLKIF_OP_DISCARD: 289 case BLKIF_OP_DISCARD:
289 dst->u.discard.flag = src->u.discard.flag; 290 dst->u.discard.flag = src->u.discard.flag;
291 dst->u.discard.id = src->u.discard.id;
290 dst->u.discard.sector_number = src->u.discard.sector_number; 292 dst->u.discard.sector_number = src->u.discard.sector_number;
291 dst->u.discard.nr_sectors = src->u.discard.nr_sectors; 293 dst->u.discard.nr_sectors = src->u.discard.nr_sectors;
292 break; 294 break;
diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
index 60eed4bdd2e4..e4fb3374dcd2 100644
--- a/drivers/block/xen-blkfront.c
+++ b/drivers/block/xen-blkfront.c
@@ -141,14 +141,36 @@ static int get_id_from_freelist(struct blkfront_info *info)
141 return free; 141 return free;
142} 142}
143 143
144static void add_id_to_freelist(struct blkfront_info *info, 144static int add_id_to_freelist(struct blkfront_info *info,
145 unsigned long id) 145 unsigned long id)
146{ 146{
147 if (info->shadow[id].req.u.rw.id != id)
148 return -EINVAL;
149 if (info->shadow[id].request == NULL)
150 return -EINVAL;
147 info->shadow[id].req.u.rw.id = info->shadow_free; 151 info->shadow[id].req.u.rw.id = info->shadow_free;
148 info->shadow[id].request = NULL; 152 info->shadow[id].request = NULL;
149 info->shadow_free = id; 153 info->shadow_free = id;
154 return 0;
150} 155}
151 156
157static const char *op_name(int op)
158{
159 static const char *const names[] = {
160 [BLKIF_OP_READ] = "read",
161 [BLKIF_OP_WRITE] = "write",
162 [BLKIF_OP_WRITE_BARRIER] = "barrier",
163 [BLKIF_OP_FLUSH_DISKCACHE] = "flush",
164 [BLKIF_OP_DISCARD] = "discard" };
165
166 if (op < 0 || op >= ARRAY_SIZE(names))
167 return "unknown";
168
169 if (!names[op])
170 return "reserved";
171
172 return names[op];
173}
152static int xlbd_reserve_minors(unsigned int minor, unsigned int nr) 174static int xlbd_reserve_minors(unsigned int minor, unsigned int nr)
153{ 175{
154 unsigned int end = minor + nr; 176 unsigned int end = minor + nr;
@@ -746,20 +768,36 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id)
746 768
747 bret = RING_GET_RESPONSE(&info->ring, i); 769 bret = RING_GET_RESPONSE(&info->ring, i);
748 id = bret->id; 770 id = bret->id;
771 /*
772 * The backend has messed up and given us an id that we would
773 * never have given to it (we stamp it up to BLK_RING_SIZE -
774 * look in get_id_from_freelist.
775 */
776 if (id >= BLK_RING_SIZE) {
777 WARN(1, "%s: response to %s has incorrect id (%ld)\n",
778 info->gd->disk_name, op_name(bret->operation), id);
779 /* We can't safely get the 'struct request' as
780 * the id is busted. */
781 continue;
782 }
749 req = info->shadow[id].request; 783 req = info->shadow[id].request;
750 784
751 if (bret->operation != BLKIF_OP_DISCARD) 785 if (bret->operation != BLKIF_OP_DISCARD)
752 blkif_completion(&info->shadow[id]); 786 blkif_completion(&info->shadow[id]);
753 787
754 add_id_to_freelist(info, id); 788 if (add_id_to_freelist(info, id)) {
789 WARN(1, "%s: response to %s (id %ld) couldn't be recycled!\n",
790 info->gd->disk_name, op_name(bret->operation), id);
791 continue;
792 }
755 793
756 error = (bret->status == BLKIF_RSP_OKAY) ? 0 : -EIO; 794 error = (bret->status == BLKIF_RSP_OKAY) ? 0 : -EIO;
757 switch (bret->operation) { 795 switch (bret->operation) {
758 case BLKIF_OP_DISCARD: 796 case BLKIF_OP_DISCARD:
759 if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) { 797 if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
760 struct request_queue *rq = info->rq; 798 struct request_queue *rq = info->rq;
761 printk(KERN_WARNING "blkfront: %s: discard op failed\n", 799 printk(KERN_WARNING "blkfront: %s: %s op failed\n",
762 info->gd->disk_name); 800 info->gd->disk_name, op_name(bret->operation));
763 error = -EOPNOTSUPP; 801 error = -EOPNOTSUPP;
764 info->feature_discard = 0; 802 info->feature_discard = 0;
765 info->feature_secdiscard = 0; 803 info->feature_secdiscard = 0;
@@ -771,18 +809,14 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id)
771 case BLKIF_OP_FLUSH_DISKCACHE: 809 case BLKIF_OP_FLUSH_DISKCACHE:
772 case BLKIF_OP_WRITE_BARRIER: 810 case BLKIF_OP_WRITE_BARRIER:
773 if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) { 811 if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
774 printk(KERN_WARNING "blkfront: %s: write %s op failed\n", 812 printk(KERN_WARNING "blkfront: %s: %s op failed\n",
775 info->flush_op == BLKIF_OP_WRITE_BARRIER ? 813 info->gd->disk_name, op_name(bret->operation));
776 "barrier" : "flush disk cache",
777 info->gd->disk_name);
778 error = -EOPNOTSUPP; 814 error = -EOPNOTSUPP;
779 } 815 }
780 if (unlikely(bret->status == BLKIF_RSP_ERROR && 816 if (unlikely(bret->status == BLKIF_RSP_ERROR &&
781 info->shadow[id].req.u.rw.nr_segments == 0)) { 817 info->shadow[id].req.u.rw.nr_segments == 0)) {
782 printk(KERN_WARNING "blkfront: %s: empty write %s op failed\n", 818 printk(KERN_WARNING "blkfront: %s: empty %s op failed\n",
783 info->flush_op == BLKIF_OP_WRITE_BARRIER ? 819 info->gd->disk_name, op_name(bret->operation));
784 "barrier" : "flush disk cache",
785 info->gd->disk_name);
786 error = -EOPNOTSUPP; 820 error = -EOPNOTSUPP;
787 } 821 }
788 if (unlikely(error)) { 822 if (unlikely(error)) {
diff --git a/fs/splice.c b/fs/splice.c
index c9f1318a3b82..7bf08fa22ec9 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -273,13 +273,16 @@ void spd_release_page(struct splice_pipe_desc *spd, unsigned int i)
273 * Check if we need to grow the arrays holding pages and partial page 273 * Check if we need to grow the arrays holding pages and partial page
274 * descriptions. 274 * descriptions.
275 */ 275 */
276int splice_grow_spd(struct pipe_inode_info *pipe, struct splice_pipe_desc *spd) 276int splice_grow_spd(const struct pipe_inode_info *pipe, struct splice_pipe_desc *spd)
277{ 277{
278 if (pipe->buffers <= PIPE_DEF_BUFFERS) 278 unsigned int buffers = ACCESS_ONCE(pipe->buffers);
279
280 spd->nr_pages_max = buffers;
281 if (buffers <= PIPE_DEF_BUFFERS)
279 return 0; 282 return 0;
280 283
281 spd->pages = kmalloc(pipe->buffers * sizeof(struct page *), GFP_KERNEL); 284 spd->pages = kmalloc(buffers * sizeof(struct page *), GFP_KERNEL);
282 spd->partial = kmalloc(pipe->buffers * sizeof(struct partial_page), GFP_KERNEL); 285 spd->partial = kmalloc(buffers * sizeof(struct partial_page), GFP_KERNEL);
283 286
284 if (spd->pages && spd->partial) 287 if (spd->pages && spd->partial)
285 return 0; 288 return 0;
@@ -289,10 +292,9 @@ int splice_grow_spd(struct pipe_inode_info *pipe, struct splice_pipe_desc *spd)
289 return -ENOMEM; 292 return -ENOMEM;
290} 293}
291 294
292void splice_shrink_spd(struct pipe_inode_info *pipe, 295void splice_shrink_spd(struct splice_pipe_desc *spd)
293 struct splice_pipe_desc *spd)
294{ 296{
295 if (pipe->buffers <= PIPE_DEF_BUFFERS) 297 if (spd->nr_pages_max <= PIPE_DEF_BUFFERS)
296 return; 298 return;
297 299
298 kfree(spd->pages); 300 kfree(spd->pages);
@@ -315,6 +317,7 @@ __generic_file_splice_read(struct file *in, loff_t *ppos,
315 struct splice_pipe_desc spd = { 317 struct splice_pipe_desc spd = {
316 .pages = pages, 318 .pages = pages,
317 .partial = partial, 319 .partial = partial,
320 .nr_pages_max = PIPE_DEF_BUFFERS,
318 .flags = flags, 321 .flags = flags,
319 .ops = &page_cache_pipe_buf_ops, 322 .ops = &page_cache_pipe_buf_ops,
320 .spd_release = spd_release_page, 323 .spd_release = spd_release_page,
@@ -326,7 +329,7 @@ __generic_file_splice_read(struct file *in, loff_t *ppos,
326 index = *ppos >> PAGE_CACHE_SHIFT; 329 index = *ppos >> PAGE_CACHE_SHIFT;
327 loff = *ppos & ~PAGE_CACHE_MASK; 330 loff = *ppos & ~PAGE_CACHE_MASK;
328 req_pages = (len + loff + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; 331 req_pages = (len + loff + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
329 nr_pages = min(req_pages, pipe->buffers); 332 nr_pages = min(req_pages, spd.nr_pages_max);
330 333
331 /* 334 /*
332 * Lookup the (hopefully) full range of pages we need. 335 * Lookup the (hopefully) full range of pages we need.
@@ -497,7 +500,7 @@ fill_it:
497 if (spd.nr_pages) 500 if (spd.nr_pages)
498 error = splice_to_pipe(pipe, &spd); 501 error = splice_to_pipe(pipe, &spd);
499 502
500 splice_shrink_spd(pipe, &spd); 503 splice_shrink_spd(&spd);
501 return error; 504 return error;
502} 505}
503 506
@@ -598,6 +601,7 @@ ssize_t default_file_splice_read(struct file *in, loff_t *ppos,
598 struct splice_pipe_desc spd = { 601 struct splice_pipe_desc spd = {
599 .pages = pages, 602 .pages = pages,
600 .partial = partial, 603 .partial = partial,
604 .nr_pages_max = PIPE_DEF_BUFFERS,
601 .flags = flags, 605 .flags = flags,
602 .ops = &default_pipe_buf_ops, 606 .ops = &default_pipe_buf_ops,
603 .spd_release = spd_release_page, 607 .spd_release = spd_release_page,
@@ -608,8 +612,8 @@ ssize_t default_file_splice_read(struct file *in, loff_t *ppos,
608 612
609 res = -ENOMEM; 613 res = -ENOMEM;
610 vec = __vec; 614 vec = __vec;
611 if (pipe->buffers > PIPE_DEF_BUFFERS) { 615 if (spd.nr_pages_max > PIPE_DEF_BUFFERS) {
612 vec = kmalloc(pipe->buffers * sizeof(struct iovec), GFP_KERNEL); 616 vec = kmalloc(spd.nr_pages_max * sizeof(struct iovec), GFP_KERNEL);
613 if (!vec) 617 if (!vec)
614 goto shrink_ret; 618 goto shrink_ret;
615 } 619 }
@@ -617,7 +621,7 @@ ssize_t default_file_splice_read(struct file *in, loff_t *ppos,
617 offset = *ppos & ~PAGE_CACHE_MASK; 621 offset = *ppos & ~PAGE_CACHE_MASK;
618 nr_pages = (len + offset + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; 622 nr_pages = (len + offset + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
619 623
620 for (i = 0; i < nr_pages && i < pipe->buffers && len; i++) { 624 for (i = 0; i < nr_pages && i < spd.nr_pages_max && len; i++) {
621 struct page *page; 625 struct page *page;
622 626
623 page = alloc_page(GFP_USER); 627 page = alloc_page(GFP_USER);
@@ -665,7 +669,7 @@ ssize_t default_file_splice_read(struct file *in, loff_t *ppos,
665shrink_ret: 669shrink_ret:
666 if (vec != __vec) 670 if (vec != __vec)
667 kfree(vec); 671 kfree(vec);
668 splice_shrink_spd(pipe, &spd); 672 splice_shrink_spd(&spd);
669 return res; 673 return res;
670 674
671err: 675err:
@@ -1614,6 +1618,7 @@ static long vmsplice_to_pipe(struct file *file, const struct iovec __user *iov,
1614 struct splice_pipe_desc spd = { 1618 struct splice_pipe_desc spd = {
1615 .pages = pages, 1619 .pages = pages,
1616 .partial = partial, 1620 .partial = partial,
1621 .nr_pages_max = PIPE_DEF_BUFFERS,
1617 .flags = flags, 1622 .flags = flags,
1618 .ops = &user_page_pipe_buf_ops, 1623 .ops = &user_page_pipe_buf_ops,
1619 .spd_release = spd_release_page, 1624 .spd_release = spd_release_page,
@@ -1629,13 +1634,13 @@ static long vmsplice_to_pipe(struct file *file, const struct iovec __user *iov,
1629 1634
1630 spd.nr_pages = get_iovec_page_array(iov, nr_segs, spd.pages, 1635 spd.nr_pages = get_iovec_page_array(iov, nr_segs, spd.pages,
1631 spd.partial, false, 1636 spd.partial, false,
1632 pipe->buffers); 1637 spd.nr_pages_max);
1633 if (spd.nr_pages <= 0) 1638 if (spd.nr_pages <= 0)
1634 ret = spd.nr_pages; 1639 ret = spd.nr_pages;
1635 else 1640 else
1636 ret = splice_to_pipe(pipe, &spd); 1641 ret = splice_to_pipe(pipe, &spd);
1637 1642
1638 splice_shrink_spd(pipe, &spd); 1643 splice_shrink_spd(&spd);
1639 return ret; 1644 return ret;
1640} 1645}
1641 1646
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index ba43f408baa3..07954b05b86c 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -827,7 +827,6 @@ extern bool __blk_end_request_err(struct request *rq, int error);
827extern void blk_complete_request(struct request *); 827extern void blk_complete_request(struct request *);
828extern void __blk_complete_request(struct request *); 828extern void __blk_complete_request(struct request *);
829extern void blk_abort_request(struct request *); 829extern void blk_abort_request(struct request *);
830extern void blk_abort_queue(struct request_queue *);
831extern void blk_unprep_request(struct request *); 830extern void blk_unprep_request(struct request *);
832 831
833/* 832/*
diff --git a/include/linux/splice.h b/include/linux/splice.h
index 26e5b613deda..09a545a7dfa3 100644
--- a/include/linux/splice.h
+++ b/include/linux/splice.h
@@ -51,7 +51,8 @@ struct partial_page {
51struct splice_pipe_desc { 51struct splice_pipe_desc {
52 struct page **pages; /* page map */ 52 struct page **pages; /* page map */
53 struct partial_page *partial; /* pages[] may not be contig */ 53 struct partial_page *partial; /* pages[] may not be contig */
54 int nr_pages; /* number of pages in map */ 54 int nr_pages; /* number of populated pages in map */
55 unsigned int nr_pages_max; /* pages[] & partial[] arrays size */
55 unsigned int flags; /* splice flags */ 56 unsigned int flags; /* splice flags */
56 const struct pipe_buf_operations *ops;/* ops associated with output pipe */ 57 const struct pipe_buf_operations *ops;/* ops associated with output pipe */
57 void (*spd_release)(struct splice_pipe_desc *, unsigned int); 58 void (*spd_release)(struct splice_pipe_desc *, unsigned int);
@@ -85,9 +86,8 @@ extern ssize_t splice_direct_to_actor(struct file *, struct splice_desc *,
85/* 86/*
86 * for dynamic pipe sizing 87 * for dynamic pipe sizing
87 */ 88 */
88extern int splice_grow_spd(struct pipe_inode_info *, struct splice_pipe_desc *); 89extern int splice_grow_spd(const struct pipe_inode_info *, struct splice_pipe_desc *);
89extern void splice_shrink_spd(struct pipe_inode_info *, 90extern void splice_shrink_spd(struct splice_pipe_desc *);
90 struct splice_pipe_desc *);
91extern void spd_release_page(struct splice_pipe_desc *, unsigned int); 91extern void spd_release_page(struct splice_pipe_desc *, unsigned int);
92 92
93extern const struct pipe_buf_operations page_cache_pipe_buf_ops; 93extern const struct pipe_buf_operations page_cache_pipe_buf_ops;
diff --git a/kernel/relay.c b/kernel/relay.c
index ab56a1764d4d..e8cd2027abbd 100644
--- a/kernel/relay.c
+++ b/kernel/relay.c
@@ -1235,6 +1235,7 @@ static ssize_t subbuf_splice_actor(struct file *in,
1235 struct splice_pipe_desc spd = { 1235 struct splice_pipe_desc spd = {
1236 .pages = pages, 1236 .pages = pages,
1237 .nr_pages = 0, 1237 .nr_pages = 0,
1238 .nr_pages_max = PIPE_DEF_BUFFERS,
1238 .partial = partial, 1239 .partial = partial,
1239 .flags = flags, 1240 .flags = flags,
1240 .ops = &relay_pipe_buf_ops, 1241 .ops = &relay_pipe_buf_ops,
@@ -1302,8 +1303,8 @@ static ssize_t subbuf_splice_actor(struct file *in,
1302 ret += padding; 1303 ret += padding;
1303 1304
1304out: 1305out:
1305 splice_shrink_spd(pipe, &spd); 1306 splice_shrink_spd(&spd);
1306 return ret; 1307 return ret;
1307} 1308}
1308 1309
1309static ssize_t relay_file_splice_read(struct file *in, 1310static ssize_t relay_file_splice_read(struct file *in,
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 49249c28690d..a7fa0702be1c 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -3609,6 +3609,7 @@ static ssize_t tracing_splice_read_pipe(struct file *filp,
3609 .pages = pages_def, 3609 .pages = pages_def,
3610 .partial = partial_def, 3610 .partial = partial_def,
3611 .nr_pages = 0, /* This gets updated below. */ 3611 .nr_pages = 0, /* This gets updated below. */
3612 .nr_pages_max = PIPE_DEF_BUFFERS,
3612 .flags = flags, 3613 .flags = flags,
3613 .ops = &tracing_pipe_buf_ops, 3614 .ops = &tracing_pipe_buf_ops,
3614 .spd_release = tracing_spd_release_pipe, 3615 .spd_release = tracing_spd_release_pipe,
@@ -3680,7 +3681,7 @@ static ssize_t tracing_splice_read_pipe(struct file *filp,
3680 3681
3681 ret = splice_to_pipe(pipe, &spd); 3682 ret = splice_to_pipe(pipe, &spd);
3682out: 3683out:
3683 splice_shrink_spd(pipe, &spd); 3684 splice_shrink_spd(&spd);
3684 return ret; 3685 return ret;
3685 3686
3686out_err: 3687out_err:
@@ -4231,6 +4232,7 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
4231 struct splice_pipe_desc spd = { 4232 struct splice_pipe_desc spd = {
4232 .pages = pages_def, 4233 .pages = pages_def,
4233 .partial = partial_def, 4234 .partial = partial_def,
4235 .nr_pages_max = PIPE_DEF_BUFFERS,
4234 .flags = flags, 4236 .flags = flags,
4235 .ops = &buffer_pipe_buf_ops, 4237 .ops = &buffer_pipe_buf_ops,
4236 .spd_release = buffer_spd_release, 4238 .spd_release = buffer_spd_release,
@@ -4318,7 +4320,7 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
4318 } 4320 }
4319 4321
4320 ret = splice_to_pipe(pipe, &spd); 4322 ret = splice_to_pipe(pipe, &spd);
4321 splice_shrink_spd(pipe, &spd); 4323 splice_shrink_spd(&spd);
4322out: 4324out:
4323 return ret; 4325 return ret;
4324} 4326}
diff --git a/mm/shmem.c b/mm/shmem.c
index a15a466d0d1d..4ce02e0673db 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1594,6 +1594,7 @@ static ssize_t shmem_file_splice_read(struct file *in, loff_t *ppos,
1594 struct splice_pipe_desc spd = { 1594 struct splice_pipe_desc spd = {
1595 .pages = pages, 1595 .pages = pages,
1596 .partial = partial, 1596 .partial = partial,
1597 .nr_pages_max = PIPE_DEF_BUFFERS,
1597 .flags = flags, 1598 .flags = flags,
1598 .ops = &page_cache_pipe_buf_ops, 1599 .ops = &page_cache_pipe_buf_ops,
1599 .spd_release = spd_release_page, 1600 .spd_release = spd_release_page,
@@ -1682,7 +1683,7 @@ static ssize_t shmem_file_splice_read(struct file *in, loff_t *ppos,
1682 if (spd.nr_pages) 1683 if (spd.nr_pages)
1683 error = splice_to_pipe(pipe, &spd); 1684 error = splice_to_pipe(pipe, &spd);
1684 1685
1685 splice_shrink_spd(pipe, &spd); 1686 splice_shrink_spd(&spd);
1686 1687
1687 if (error > 0) { 1688 if (error > 0) {
1688 *ppos += error; 1689 *ppos += error;
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index d78671e9d545..46a3d23d259e 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -1755,6 +1755,7 @@ int skb_splice_bits(struct sk_buff *skb, unsigned int offset,
1755 struct splice_pipe_desc spd = { 1755 struct splice_pipe_desc spd = {
1756 .pages = pages, 1756 .pages = pages,
1757 .partial = partial, 1757 .partial = partial,
1758 .nr_pages_max = MAX_SKB_FRAGS,
1758 .flags = flags, 1759 .flags = flags,
1759 .ops = &sock_pipe_buf_ops, 1760 .ops = &sock_pipe_buf_ops,
1760 .spd_release = sock_spd_release, 1761 .spd_release = sock_spd_release,