1 files changed, 144 insertions, 47 deletions

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 210b8e726a97..673756468019 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -319,11 +319,12 @@ static void put_probe_ref(void)
 
 static void blk_trace_cleanup(struct blk_trace *bt)
 {
+	synchronize_rcu();
 	blk_trace_free(bt);
 	put_probe_ref();
 }
 
-int blk_trace_remove(struct request_queue *q)
+static int __blk_trace_remove(struct request_queue *q)
 {
 	struct blk_trace *bt;
 
@@ -336,6 +337,17 @@ int blk_trace_remove(struct request_queue *q)
 
 	return 0;
 }
+
+int blk_trace_remove(struct request_queue *q)
+{
+	int ret;
+
+	mutex_lock(&q->blk_trace_mutex);
+	ret = __blk_trace_remove(q);
+	mutex_unlock(&q->blk_trace_mutex);
+
+	return ret;
+}
 EXPORT_SYMBOL_GPL(blk_trace_remove);
 
 static ssize_t blk_dropped_read(struct file *filp, char __user *buffer,
@@ -546,9 +558,8 @@ err:
 	return ret;
 }
 
-int blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
-		    struct block_device *bdev,
-		    char __user *arg)
+static int __blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
+			     struct block_device *bdev, char __user *arg)
 {
 	struct blk_user_trace_setup buts;
 	int ret;
@@ -562,11 +573,24 @@ int blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 		return ret;
 
 	if (copy_to_user(arg, &buts, sizeof(buts))) {
-		blk_trace_remove(q);
+		__blk_trace_remove(q);
 		return -EFAULT;
 	}
 	return 0;
 }
+
+int blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
+		    struct block_device *bdev,
+		    char __user *arg)
+{
+	int ret;
+
+	mutex_lock(&q->blk_trace_mutex);
+	ret = __blk_trace_setup(q, name, dev, bdev, arg);
+	mutex_unlock(&q->blk_trace_mutex);
+
+	return ret;
+}
 EXPORT_SYMBOL_GPL(blk_trace_setup);
 
 #if defined(CONFIG_COMPAT) && defined(CONFIG_X86_64)
@@ -595,7 +619,7 @@ static int compat_blk_trace_setup(struct request_queue *q, char *name,
 		return ret;
 
 	if (copy_to_user(arg, &buts.name, ARRAY_SIZE(buts.name))) {
-		blk_trace_remove(q);
+		__blk_trace_remove(q);
 		return -EFAULT;
 	}
 
@@ -603,11 +627,13 @@ static int compat_blk_trace_setup(struct request_queue *q, char *name,
 }
 #endif
 
-int blk_trace_startstop(struct request_queue *q, int start)
+static int __blk_trace_startstop(struct request_queue *q, int start)
 {
 	int ret;
-	struct blk_trace *bt = q->blk_trace;
+	struct blk_trace *bt;
 
+	bt = rcu_dereference_protected(q->blk_trace,
+				       lockdep_is_held(&q->blk_trace_mutex));
 	if (bt == NULL)
 		return -EINVAL;
 
@@ -642,8 +668,25 @@ int blk_trace_startstop(struct request_queue *q, int start)
 
 	return ret;
 }
+
+int blk_trace_startstop(struct request_queue *q, int start)
+{
+	int ret;
+
+	mutex_lock(&q->blk_trace_mutex);
+	ret = __blk_trace_startstop(q, start);
+	mutex_unlock(&q->blk_trace_mutex);
+
+	return ret;
+}
 EXPORT_SYMBOL_GPL(blk_trace_startstop);
 
+/*
+ * When reading or writing the blktrace sysfs files, the references to the
+ * opened sysfs or device files should prevent the underlying block device
+ * from being removed. So no further delete protection is really needed.
+ */
+
 /**
  * blk_trace_ioctl: - handle the ioctls associated with tracing
  * @bdev:	the block device
@@ -661,12 +704,12 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg)
 	if (!q)
 		return -ENXIO;
 
-	mutex_lock(&bdev->bd_mutex);
+	mutex_lock(&q->blk_trace_mutex);
 
 	switch (cmd) {
 	case BLKTRACESETUP:
 		bdevname(bdev, b);
-		ret = blk_trace_setup(q, b, bdev->bd_dev, bdev, arg);
+		ret = __blk_trace_setup(q, b, bdev->bd_dev, bdev, arg);
 		break;
 #if defined(CONFIG_COMPAT) && defined(CONFIG_X86_64)
 	case BLKTRACESETUP32:
@@ -677,17 +720,17 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg)
 	case BLKTRACESTART:
 		start = 1;
 	case BLKTRACESTOP:
-		ret = blk_trace_startstop(q, start);
+		ret = __blk_trace_startstop(q, start);
 		break;
 	case BLKTRACETEARDOWN:
-		ret = blk_trace_remove(q);
+		ret = __blk_trace_remove(q);
 		break;
 	default:
 		ret = -ENOTTY;
 		break;
 	}
 
-	mutex_unlock(&bdev->bd_mutex);
+	mutex_unlock(&q->blk_trace_mutex);
 	return ret;
 }
 
@@ -698,10 +741,14 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg)
  **/
 void blk_trace_shutdown(struct request_queue *q)
 {
-	if (q->blk_trace) {
-		blk_trace_startstop(q, 0);
-		blk_trace_remove(q);
+	mutex_lock(&q->blk_trace_mutex);
+	if (rcu_dereference_protected(q->blk_trace,
+				      lockdep_is_held(&q->blk_trace_mutex))) {
+		__blk_trace_startstop(q, 0);
+		__blk_trace_remove(q);
 	}
+
+	mutex_unlock(&q->blk_trace_mutex);
 }
 
 /*
@@ -722,10 +769,14 @@ void blk_trace_shutdown(struct request_queue *q)
 static void blk_add_trace_rq(struct request_queue *q, struct request *rq,
 			     unsigned int nr_bytes, u32 what)
 {
-	struct blk_trace *bt = q->blk_trace;
+	struct blk_trace *bt;
 
-	if (likely(!bt))
+	rcu_read_lock();
+	bt = rcu_dereference(q->blk_trace);
+	if (likely(!bt)) {
+		rcu_read_unlock();
 		return;
+	}
 
 	if (rq->cmd_type == REQ_TYPE_BLOCK_PC) {
 		what |= BLK_TC_ACT(BLK_TC_PC);
@@ -736,6 +787,7 @@ static void blk_add_trace_rq(struct request_queue *q, struct request *rq,
 		__blk_add_trace(bt, blk_rq_pos(rq), nr_bytes,
 				rq->cmd_flags, what, rq->errors, 0, NULL);
 	}
+	rcu_read_unlock();
 }
 
 static void blk_add_trace_rq_abort(void *ignore,
@@ -785,13 +837,18 @@ static void blk_add_trace_rq_complete(void *ignore,
 static void blk_add_trace_bio(struct request_queue *q, struct bio *bio,
 			      u32 what, int error)
 {
-	struct blk_trace *bt = q->blk_trace;
+	struct blk_trace *bt;
 
-	if (likely(!bt))
+	rcu_read_lock();
+	bt = rcu_dereference(q->blk_trace);
+	if (likely(!bt)) {
+		rcu_read_unlock();
 		return;
+	}
 
 	__blk_add_trace(bt, bio->bi_iter.bi_sector, bio->bi_iter.bi_size,
 			bio->bi_rw, what, error, 0, NULL);
+	rcu_read_unlock();
 }
 
 static void blk_add_trace_bio_bounce(void *ignore,
@@ -836,10 +893,13 @@ static void blk_add_trace_getrq(void *ignore,
 	if (bio)
 		blk_add_trace_bio(q, bio, BLK_TA_GETRQ, 0);
 	else {
-		struct blk_trace *bt = q->blk_trace;
+		struct blk_trace *bt;
 
+		rcu_read_lock();
+		bt = rcu_dereference(q->blk_trace);
 		if (bt)
 			__blk_add_trace(bt, 0, 0, rw, BLK_TA_GETRQ, 0, 0, NULL);
+		rcu_read_unlock();
 	}
 }
 
@@ -851,27 +911,35 @@ static void blk_add_trace_sleeprq(void *ignore,
 	if (bio)
 		blk_add_trace_bio(q, bio, BLK_TA_SLEEPRQ, 0);
 	else {
-		struct blk_trace *bt = q->blk_trace;
+		struct blk_trace *bt;
 
+		rcu_read_lock();
+		bt = rcu_dereference(q->blk_trace);
 		if (bt)
 			__blk_add_trace(bt, 0, 0, rw, BLK_TA_SLEEPRQ,
 					0, 0, NULL);
+		rcu_read_unlock();
 	}
 }
 
 static void blk_add_trace_plug(void *ignore, struct request_queue *q)
 {
-	struct blk_trace *bt = q->blk_trace;
+	struct blk_trace *bt;
 
+	rcu_read_lock();
+	bt = rcu_dereference(q->blk_trace);
 	if (bt)
 		__blk_add_trace(bt, 0, 0, 0, BLK_TA_PLUG, 0, 0, NULL);
+	rcu_read_unlock();
 }
 
 static void blk_add_trace_unplug(void *ignore, struct request_queue *q,
 				    unsigned int depth, bool explicit)
 {
-	struct blk_trace *bt = q->blk_trace;
+	struct blk_trace *bt;
 
+	rcu_read_lock();
+	bt = rcu_dereference(q->blk_trace);
 	if (bt) {
 		__be64 rpdu = cpu_to_be64(depth);
 		u32 what;
@@ -883,14 +951,17 @@ static void blk_add_trace_unplug(void *ignore, struct request_queue *q,
 
 		__blk_add_trace(bt, 0, 0, 0, what, 0, sizeof(rpdu), &rpdu);
 	}
+	rcu_read_unlock();
 }
 
 static void blk_add_trace_split(void *ignore,
 				struct request_queue *q, struct bio *bio,
 				unsigned int pdu)
 {
-	struct blk_trace *bt = q->blk_trace;
+	struct blk_trace *bt;
 
+	rcu_read_lock();
+	bt = rcu_dereference(q->blk_trace);
 	if (bt) {
 		__be64 rpdu = cpu_to_be64(pdu);
 
@@ -898,6 +969,7 @@ static void blk_add_trace_split(void *ignore,
 				bio->bi_iter.bi_size, bio->bi_rw, BLK_TA_SPLIT,
 				bio->bi_error, sizeof(rpdu), &rpdu);
 	}
+	rcu_read_unlock();
 }
 
 /**
@@ -917,11 +989,15 @@ static void blk_add_trace_bio_remap(void *ignore,
 				    struct request_queue *q, struct bio *bio,
 				    dev_t dev, sector_t from)
 {
-	struct blk_trace *bt = q->blk_trace;
+	struct blk_trace *bt;
 	struct blk_io_trace_remap r;
 
-	if (likely(!bt))
+	rcu_read_lock();
+	bt = rcu_dereference(q->blk_trace);
+	if (likely(!bt)) {
+		rcu_read_unlock();
 		return;
+	}
 
 	r.device_from = cpu_to_be32(dev);
 	r.device_to   = cpu_to_be32(bio->bi_bdev->bd_dev);
@@ -930,6 +1006,7 @@ static void blk_add_trace_bio_remap(void *ignore,
 	__blk_add_trace(bt, bio->bi_iter.bi_sector, bio->bi_iter.bi_size,
 			bio->bi_rw, BLK_TA_REMAP, bio->bi_error,
 			sizeof(r), &r);
+	rcu_read_unlock();
 }
 
 /**
@@ -950,11 +1027,15 @@ static void blk_add_trace_rq_remap(void *ignore,
 				   struct request *rq, dev_t dev,
 				   sector_t from)
 {
-	struct blk_trace *bt = q->blk_trace;
+	struct blk_trace *bt;
 	struct blk_io_trace_remap r;
 
-	if (likely(!bt))
+	rcu_read_lock();
+	bt = rcu_dereference(q->blk_trace);
+	if (likely(!bt)) {
+		rcu_read_unlock();
 		return;
+	}
 
 	r.device_from = cpu_to_be32(dev);
 	r.device_to   = cpu_to_be32(disk_devt(rq->rq_disk));
@@ -963,6 +1044,7 @@ static void blk_add_trace_rq_remap(void *ignore,
 	__blk_add_trace(bt, blk_rq_pos(rq), blk_rq_bytes(rq),
 			rq_data_dir(rq), BLK_TA_REMAP, !!rq->errors,
 			sizeof(r), &r);
+	rcu_read_unlock();
 }
 
 /**
@@ -980,10 +1062,14 @@ void blk_add_driver_data(struct request_queue *q,
 			 struct request *rq,
 			 void *data, size_t len)
 {
-	struct blk_trace *bt = q->blk_trace;
+	struct blk_trace *bt;
 
-	if (likely(!bt))
+	rcu_read_lock();
+	bt = rcu_dereference(q->blk_trace);
+	if (likely(!bt)) {
+		rcu_read_unlock();
 		return;
+	}
 
 	if (rq->cmd_type == REQ_TYPE_BLOCK_PC)
 		__blk_add_trace(bt, 0, blk_rq_bytes(rq), 0,
@@ -991,6 +1077,7 @@ void blk_add_driver_data(struct request_queue *q,
 	else
 		__blk_add_trace(bt, blk_rq_pos(rq), blk_rq_bytes(rq), 0,
 				BLK_TA_DRV_DATA, rq->errors, len, data);
+	rcu_read_unlock();
 }
 EXPORT_SYMBOL_GPL(blk_add_driver_data);
 
@@ -1482,6 +1569,7 @@ static int blk_trace_remove_queue(struct request_queue *q)
 		return -EINVAL;
 
 	put_probe_ref();
+	synchronize_rcu();
 	blk_trace_free(bt);
 	return 0;
 }
@@ -1642,6 +1730,7 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 	struct hd_struct *p = dev_to_part(dev);
 	struct request_queue *q;
 	struct block_device *bdev;
+	struct blk_trace *bt;
 	ssize_t ret = -ENXIO;
 
 	bdev = bdget(part_devt(p));
@@ -1652,26 +1741,28 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 	if (q == NULL)
 		goto out_bdput;
 
-	mutex_lock(&bdev->bd_mutex);
+	mutex_lock(&q->blk_trace_mutex);
 
+	bt = rcu_dereference_protected(q->blk_trace,
+				       lockdep_is_held(&q->blk_trace_mutex));
 	if (attr == &dev_attr_enable) {
-		ret = sprintf(buf, "%u\n", !!q->blk_trace);
+		ret = sprintf(buf, "%u\n", !!bt);
 		goto out_unlock_bdev;
 	}
 
-	if (q->blk_trace == NULL)
+	if (bt == NULL)
 		ret = sprintf(buf, "disabled\n");
 	else if (attr == &dev_attr_act_mask)
-		ret = blk_trace_mask2str(buf, q->blk_trace->act_mask);
+		ret = blk_trace_mask2str(buf, bt->act_mask);
 	else if (attr == &dev_attr_pid)
-		ret = sprintf(buf, "%u\n", q->blk_trace->pid);
+		ret = sprintf(buf, "%u\n", bt->pid);
 	else if (attr == &dev_attr_start_lba)
-		ret = sprintf(buf, "%llu\n", q->blk_trace->start_lba);
+		ret = sprintf(buf, "%llu\n", bt->start_lba);
 	else if (attr == &dev_attr_end_lba)
-		ret = sprintf(buf, "%llu\n", q->blk_trace->end_lba);
+		ret = sprintf(buf, "%llu\n", bt->end_lba);
 
 out_unlock_bdev:
-	mutex_unlock(&bdev->bd_mutex);
+	mutex_unlock(&q->blk_trace_mutex);
 out_bdput:
 	bdput(bdev);
 out:
@@ -1685,6 +1776,7 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 	struct block_device *bdev;
 	struct request_queue *q;
 	struct hd_struct *p;
+	struct blk_trace *bt;
 	u64 value;
 	ssize_t ret = -EINVAL;
 
@@ -1713,10 +1805,12 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 	if (q == NULL)
 		goto out_bdput;
 
-	mutex_lock(&bdev->bd_mutex);
+	mutex_lock(&q->blk_trace_mutex);
 
+	bt = rcu_dereference_protected(q->blk_trace,
+				       lockdep_is_held(&q->blk_trace_mutex));
 	if (attr == &dev_attr_enable) {
-		if (!!value == !!q->blk_trace) {
+		if (!!value == !!bt) {
 			ret = 0;
 			goto out_unlock_bdev;
 		}
@@ -1728,22 +1822,25 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 	}
 
 	ret = 0;
-	if (q->blk_trace == NULL)
+	if (bt == NULL) {
 		ret = blk_trace_setup_queue(q, bdev);
+		bt = rcu_dereference_protected(q->blk_trace,
+				lockdep_is_held(&q->blk_trace_mutex));
+	}
 
 	if (ret == 0) {
 		if (attr == &dev_attr_act_mask)
-			q->blk_trace->act_mask = value;
+			bt->act_mask = value;
 		else if (attr == &dev_attr_pid)
-			q->blk_trace->pid = value;
+			bt->pid = value;
 		else if (attr == &dev_attr_start_lba)
-			q->blk_trace->start_lba = value;
+			bt->start_lba = value;
 		else if (attr == &dev_attr_end_lba)
-			q->blk_trace->end_lba = value;
+			bt->end_lba = value;
 	}
 
 out_unlock_bdev:
-	mutex_unlock(&bdev->bd_mutex);
+	mutex_unlock(&q->blk_trace_mutex);
 out_bdput:
 	bdput(bdev);
 out: