Skip to contentSkip to conversation

blktrace: various cleanups and fixes

PATCH In review Li Zefan 5 participants 39 messages Mar 20, 2009 – Mar 23, 2009

Subsystems (2)
lore
Share & tools
lore

Li Zefan

author
#0 lore raw

cover blktrace: various cleanups and fixes

While trying to use blktrace in -tip tree, I encounted kernel NULL dereference, so I looked into the code, and then found some other bugs.

This patchset is part 1, and I have some other pending fixes.

Each patch is small and straightforward, and should be easy to review:

[PATCH 1/7] blktrace: fix possible memory leak
[PATCH 2/7] blktrace: make blk_tracer_enabled a bool flag
[PATCH 3/7] blktrace: remove blk_probe_mutex
[PATCH 4/7] blktrace: don't increase blk_probes_ref if failed to setup blk trace [PATCH 5/7] blktrace: report EBUSY correctly
[PATCH 6/7] blktrace: remove sysfs_blk_trace_enable_show/store() [PATCH 7/7] blktrace: avoid accessing NULL bdev->bd_disk

---
 blktrace.c |  154 +++++++++++++++++--------------------------------------------
 1 file changed, 45 insertions(+), 109 deletions(-)
author
#1 lore raw

PATCH 1/7 blktrace: fix possible memory leak

When we failed to create "block" debugfs dir, we should do come cleanups.

---
 kernel/trace/blktrace.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
kernel/trace/blktrace.c +1 -1
index b171778..fb3bc53 100644
@@ -432,7 +432,7 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	if (!blk_tree_root) {
 		blk_tree_root = debugfs_create_dir("block", NULL);
 		if (!blk_tree_root)
-			return -ENOMEM;
+			goto err;
 	}
 
 	dir = debugfs_create_dir(buts->name, blk_tree_root);
Signature · 1 line
1.5.4.rc3
author
#5 lore raw

PATCH 2/7 blktrace: make blk_tracer_enabled a bool flag

In reply to Li Zefan

It doesn't have to be a counter, and it can be a bool flag instead.

---
 kernel/trace/blktrace.c |   12 +++---------
 1 files changed, 3 insertions(+), 9 deletions(-)
kernel/trace/blktrace.c +3 -9
index fb3bc53..73845b7 100644
@@ -30,7 +30,7 @@
 static unsigned int blktrace_seq __read_mostly = 1;
 
 static struct trace_array *blk_tr;
-static int __read_mostly  blk_tracer_enabled;
+static bool blk_tracer_enabled __read_mostly;
 
 /* Select an alternative, minimalistic output than the original one */
 #define TRACE_BLK_OPT_CLASSIC	0x1
@@ -1111,9 +1111,7 @@ static int blk_tracer_init(struct trace_array *tr)
 {
 	blk_tr = tr;
 	blk_tracer_start(tr);
-	mutex_lock(&blk_probe_mutex);
-	blk_tracer_enabled++;
-	mutex_unlock(&blk_probe_mutex);
+	blk_tracer_enabled = true;
 	return 0;
 }
 
@@ -1131,11 +1129,7 @@ static void blk_tracer_reset(struct trace_array *tr)
 	if (!atomic_read(&blk_probes_ref))
 		return;
 
-	mutex_lock(&blk_probe_mutex);
-	blk_tracer_enabled--;
-	WARN_ON(blk_tracer_enabled < 0);
-	mutex_unlock(&blk_probe_mutex);
-
+	blk_tracer_enabled = false;
 	blk_tracer_stop(tr);
 }
 
Signature · 1 line
1.5.4.rc3
author
#10 lore raw

PATCH 3/7 blktrace: remove blk_probe_mutex

In reply to Li Zefan

blk_register_tracepoints() always returns 0, so make it return void, thus we don't need to use blk_probe_mutex to protect blk_probes_ref.

---
 kernel/trace/blktrace.c |   27 +++++----------------------
 1 files changed, 5 insertions(+), 22 deletions(-)
kernel/trace/blktrace.c +5 -22
index 73845b7..223b92e 100644
@@ -47,10 +47,9 @@ static struct tracer_flags blk_tracer_flags = {
 };
 
 /* Global reference count of probes */
-static DEFINE_MUTEX(blk_probe_mutex);
 static atomic_t blk_probes_ref = ATOMIC_INIT(0);
 
-static int blk_register_tracepoints(void);
+static void blk_register_tracepoints(void);
 static void blk_unregister_tracepoints(void);
 
 /*
@@ -256,10 +255,8 @@ static void blk_trace_cleanup(struct blk_trace *bt)
 	free_percpu(bt->sequence);
 	free_percpu(bt->msg_data);
 	kfree(bt);
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_dec_and_test(&blk_probes_ref))
 		blk_unregister_tracepoints();
-	mutex_unlock(&blk_probe_mutex);
 }
 
 int blk_trace_remove(struct request_queue *q)
@@ -471,13 +468,8 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	bt->pid = buts->pid;
 	bt->trace_state = Blktrace_setup;
 
-	mutex_lock(&blk_probe_mutex);
-	if (atomic_add_return(1, &blk_probes_ref) == 1) {
-		ret = blk_register_tracepoints();
-		if (ret)
-			goto probe_err;
-	}
-	mutex_unlock(&blk_probe_mutex);
+	if (atomic_add_return(1, &blk_probes_ref) == 1)
+		blk_register_tracepoints();
 
 	ret = -EBUSY;
 	old_bt = xchg(&q->blk_trace, bt);
@@ -487,9 +479,6 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	}
 
 	return 0;
-probe_err:
-	atomic_dec(&blk_probes_ref);
-	mutex_unlock(&blk_probe_mutex);
 err:
 	if (bt) {
 		if (bt->msg_file)
@@ -863,7 +852,7 @@ void blk_add_driver_data(struct request_queue *q,
 }
 EXPORT_SYMBOL_GPL(blk_add_driver_data);
 
-static int blk_register_tracepoints(void)
+static void blk_register_tracepoints(void)
 {
 	int ret;
 
@@ -901,7 +890,6 @@ static int blk_register_tracepoints(void)
 	WARN_ON(ret);
 	ret = register_trace_block_remap(blk_add_trace_remap);
 	WARN_ON(ret);
-	return 0;
 }
 
 static void blk_unregister_tracepoints(void)
@@ -1099,11 +1087,8 @@ static void blk_tracer_print_header(struct seq_file *m)
 
 static void blk_tracer_start(struct trace_array *tr)
 {
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_add_return(1, &blk_probes_ref) == 1)
-		if (blk_register_tracepoints())
-			atomic_dec(&blk_probes_ref);
-	mutex_unlock(&blk_probe_mutex);
+		blk_register_tracepoints();
 	trace_flags &= ~TRACE_ITER_CONTEXT_INFO;
 }
 
@@ -1118,10 +1103,8 @@ static int blk_tracer_init(struct trace_array *tr)
 static void blk_tracer_stop(struct trace_array *tr)
 {
 	trace_flags |= TRACE_ITER_CONTEXT_INFO;
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_dec_and_test(&blk_probes_ref))
 		blk_unregister_tracepoints();
-	mutex_unlock(&blk_probe_mutex);
 }
 
 static void blk_tracer_reset(struct trace_array *tr)
Signature · 1 line
1.5.4.rc3
author
#17 lore raw

PATCH 4/7 blktrace: don't increase blk_probes_ref if failed to setup blk trace

In reply to Li Zefan

do_blk_trace_setup() may return EBUSY, but the current code doesn't decrease blk_probes_ref in this case.

---
 kernel/trace/blktrace.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)
kernel/trace/blktrace.c +3 -3
index 223b92e..11e7c8d 100644
@@ -468,9 +468,6 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	bt->pid = buts->pid;
 	bt->trace_state = Blktrace_setup;
 
-	if (atomic_add_return(1, &blk_probes_ref) == 1)
-		blk_register_tracepoints();
-
 	ret = -EBUSY;
 	old_bt = xchg(&q->blk_trace, bt);
 	if (old_bt) {
@@ -478,6 +475,9 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 		goto err;
 	}
 
+	if (atomic_add_return(1, &blk_probes_ref) == 1)
+		blk_register_tracepoints();
+
 	return 0;
 err:
 	if (bt) {
Signature · 1 line
1.5.4.rc3
author
#21 lore raw

PATCH 5/7 blktrace: report EBUSY correctly

In reply to Li Zefan

blk_trace_remove_queue() returns EINVAL if q->blk_trace == NULL, but blk_trace_setup_queue() doesn't return EBUSY if q->blk_trace != NULL.

 # echo 0 > sdaX/trace/enable
 # echo 0 > sdaX/trace/enable
 bash: echo: write error: Invalid argument
 # echo 1 > sdaX/trace/enable
 # echo 1 > sdaX/trace/enable
 (should return EBUSY)
---
 kernel/trace/blktrace.c |    9 +++------
 1 files changed, 3 insertions(+), 6 deletions(-)
kernel/trace/blktrace.c +3 -6
index 11e7c8d..14986af 100644
@@ -1260,12 +1260,10 @@ static int blk_trace_remove_queue(struct request_queue *q)
 static int blk_trace_setup_queue(struct request_queue *q, dev_t dev)
 {
 	struct blk_trace *old_bt, *bt = NULL;
-	int ret;
 
-	ret = -ENOMEM;
 	bt = kzalloc(sizeof(*bt), GFP_KERNEL);
 	if (!bt)
-		goto err;
+		return -ENOMEM;
 
 	bt->dev = dev;
 	bt->act_mask = (u16)-1;
@@ -1276,11 +1274,10 @@ static int blk_trace_setup_queue(struct request_queue *q, dev_t dev)
 	if (old_bt != NULL) {
 		(void)xchg(&q->blk_trace, old_bt);
 		kfree(bt);
-		ret = -EBUSY;
+		return -EBUSY;
 	}
+
 	return 0;
-err:
-	return ret;
 }
 
 /*
Signature · 1 line
1.5.4.rc3
author
#26 lore raw

PATCH 6/7 blktrace: remove sysfs_blk_trace_enable_show/store()

In reply to Li Zefan

sysfs_blk_trace_enable_show()/store() share most of code with sysfs_blk_trace_attr_show()/store().

---
 kernel/trace/blktrace.c |   89 +++++++++++------------------------------------
 1 files changed, 21 insertions(+), 68 deletions(-)
kernel/trace/blktrace.c +21 -68
index 14986af..46e0cfe 100644
@@ -1284,72 +1284,6 @@ static int blk_trace_setup_queue(struct request_queue *q, dev_t dev)
  * sysfs interface to enable and configure tracing
  */
 
-static ssize_t sysfs_blk_trace_enable_show(struct device *dev,
-					   struct device_attribute *attr,
-					   char *buf)
-{
-	struct hd_struct *p = dev_to_part(dev);
-	struct block_device *bdev;
-	ssize_t ret = -ENXIO;
-
-	lock_kernel();
-	bdev = bdget(part_devt(p));
-	if (bdev != NULL) {
-		struct request_queue *q = bdev_get_queue(bdev);
-
-		if (q != NULL) {
-			mutex_lock(&bdev->bd_mutex);
-			ret = sprintf(buf, "%u\n", !!q->blk_trace);
-			mutex_unlock(&bdev->bd_mutex);
-		}
-
-		bdput(bdev);
-	}
-
-	unlock_kernel();
-	return ret;
-}
-
-static ssize_t sysfs_blk_trace_enable_store(struct device *dev,
-					    struct device_attribute *attr,
-					    const char *buf, size_t count)
-{
-	struct block_device *bdev;
-	struct request_queue *q;
-	struct hd_struct *p;
-	int value;
-	ssize_t ret = -ENXIO;
-
-	if (count == 0 || sscanf(buf, "%d", &value) != 1)
-		goto out;
-
-	lock_kernel();
-	p = dev_to_part(dev);
-	bdev = bdget(part_devt(p));
-	if (bdev == NULL)
-		goto out_unlock_kernel;
-
-	q = bdev_get_queue(bdev);
-	if (q == NULL)
-		goto out_bdput;
-
-	mutex_lock(&bdev->bd_mutex);
-	if (value)
-		ret = blk_trace_setup_queue(q, bdev->bd_dev);
-	else
-		ret = blk_trace_remove_queue(q);
-	mutex_unlock(&bdev->bd_mutex);
-
-	if (ret == 0)
-		ret = count;
-out_bdput:
-	bdput(bdev);
-out_unlock_kernel:
-	unlock_kernel();
-out:
-	return ret;
-}
-
 static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 					 struct device_attribute *attr,
 					 char *buf);
@@ -1361,8 +1295,7 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 		    sysfs_blk_trace_attr_show, \
 		    sysfs_blk_trace_attr_store)
 
-static DEVICE_ATTR(enable, S_IRUGO | S_IWUSR,
-		   sysfs_blk_trace_enable_show, sysfs_blk_trace_enable_store);
+static BLK_TRACE_DEVICE_ATTR(enable);
 static BLK_TRACE_DEVICE_ATTR(act_mask);
 static BLK_TRACE_DEVICE_ATTR(pid);
 static BLK_TRACE_DEVICE_ATTR(start_lba);
@@ -1447,6 +1380,12 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 	if (q == NULL)
 		goto out_bdput;
 	mutex_lock(&bdev->bd_mutex);
+
+	if (attr == &dev_attr_enable) {
+		ret = sprintf(buf, "%u\n", !!q->blk_trace);
+		goto out_unlock_bdev;
+	}
+
 	if (q->blk_trace == NULL)
 		ret = sprintf(buf, "disabled\n");
 	else if (attr == &dev_attr_act_mask)
@@ -1457,6 +1396,8 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 		ret = sprintf(buf, "%llu\n", q->blk_trace->start_lba);
 	else if (attr == &dev_attr_end_lba)
 		ret = sprintf(buf, "%llu\n", q->blk_trace->end_lba);
+
+out_unlock_bdev:
 	mutex_unlock(&bdev->bd_mutex);
 out_bdput:
 	bdput(bdev);
@@ -1499,6 +1440,15 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 		goto out_bdput;
 
 	mutex_lock(&bdev->bd_mutex);
+
+	if (attr == &dev_attr_enable) {
+		if (value)
+			ret = blk_trace_setup_queue(q, bdev->bd_dev);
+		else
+			ret = blk_trace_remove_queue(q);
+		goto out_unlock_bdev;
+	}
+
 	ret = 0;
 	if (q->blk_trace == NULL)
 		ret = blk_trace_setup_queue(q, bdev->bd_dev);
@@ -1514,6 +1464,8 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 			q->blk_trace->end_lba = value;
 		ret = count;
 	}
+
+out_unlock_bdev:
 	mutex_unlock(&bdev->bd_mutex);
 out_bdput:
 	bdput(bdev);
@@ -1522,3 +1474,4 @@ out_unlock_kernel:
 out:
 	return ret;
 }
+
Signature · 1 line
1.5.4.rc3
author
#31 lore raw

PATCH 7/7 blktrace: avoid accessing NULL bdev->bd_disk

In reply to Li Zefan

bdev->bd_disk can be NULL, if the block device is not opened.

Try this against an unmounted partition, and you'll see NULL dereference:

  # echo 1 > /sys/block/sda/sda5/enable
---
 kernel/trace/blktrace.c |   13 +++++++++++--
 1 files changed, 11 insertions(+), 2 deletions(-)
kernel/trace/blktrace.c +11 -2
index 46e0cfe..37095d9 100644
@@ -1362,6 +1362,14 @@ static int blk_str2act_mask(const char *str)
 	return mask;
 }
 
+static request_queue *blk_trace_get_queue(struct block_device *bdev)
+{
+	if (bdev->bd_disk == NULL)
+		return NULL;
+
+	return bdev_get_queue(bdev);
+}
+
 static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 					 struct device_attribute *attr,
 					 char *buf)
@@ -1376,9 +1384,10 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 	if (bdev == NULL)
 		goto out_unlock_kernel;
 
-	q = bdev_get_queue(bdev);
+	q = blk_trace_get_queue(bdev);
 	if (q == NULL)
 		goto out_bdput;
+
 	mutex_lock(&bdev->bd_mutex);
 
 	if (attr == &dev_attr_enable) {
@@ -1435,7 +1444,7 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 	if (bdev == NULL)
 		goto out_unlock_kernel;
 
-	q = bdev_get_queue(bdev);
+	q = blk_trace_get_queue(bdev);
 	if (q == NULL)
 		goto out_bdput;
 
Signature · 1 line
1.5.4.rc3
author
#32 lore raw

Re: [PATCH 7/7] blktrace: avoid accessing NULL bdev->bd_disk

Review

Quoted 2 lines
(quoted diff) +2 -0
+static request_queue *blk_trace_get_queue(struct block_device *bdev)
+{

Sorry, should be "struct request_queue", I forgot to re-format the patch after I fixed this.

================

From: Li Zefan <[email protected]>
Subject: [PATCH 7/7] blktrace: avoid accessing NULL bdev->bd_disk

bdev->bd_disk can be NULL, if the block device is not opened.

Try this against an unmounted partition, and you'll see NULL dereference:

  # echo 1 > /sys/block/sda/sda5/enable
---
 kernel/trace/blktrace.c |   13 +++++++++++--
 1 files changed, 11 insertions(+), 2 deletions(-)
kernel/trace/blktrace.c +11 -2
index 46e0cfe..d2dd9ed 100644
@@ -1362,6 +1362,14 @@ static int blk_str2act_mask(const char *str)
 	return mask;
 }
 
+static struct request_queue *blk_trace_get_queue(struct block_device *bdev)
+{
+	if (bdev->bd_disk == NULL)
+		return NULL;
+
+	return bdev_get_queue(bdev);
+}
+
 static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 					 struct device_attribute *attr,
 					 char *buf)
@@ -1376,9 +1384,10 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 	if (bdev == NULL)
 		goto out_unlock_kernel;
 
-	q = bdev_get_queue(bdev);
+	q = blk_trace_get_queue(bdev);
 	if (q == NULL)
 		goto out_bdput;
+
 	mutex_lock(&bdev->bd_mutex);
 
 	if (attr == &dev_attr_enable) {
@@ -1435,7 +1444,7 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 	if (bdev == NULL)
 		goto out_unlock_kernel;
 
-	q = bdev_get_queue(bdev);
+	q = blk_trace_get_queue(bdev);
 	if (q == NULL)
 		goto out_bdput;
 
Signature · 1 line
1.5.4.rc3
author
#27 lore raw

Re: [PATCH 6/7] blktrace: remove sysfs_blk_trace_enable_show/store()

In reply to Li Zefan

Review

Quoted 5 lines
(quoted diff) +5 -0
+	if (attr == &dev_attr_enable) {
+		if (value)
+			ret = blk_trace_setup_queue(q, bdev->bd_dev);
+		else
+			ret = blk_trace_remove_queue(q);

(ret = count) should be returned if ret == 0.. patch updated below

Review

Quoted 2 lines
(quoted diff) +2 -0
+		goto out_unlock_bdev;
+	}

===================

From: Li Zefan <[email protected]>
Subject: [PATCH 6/7] blktrace: sysfs_blk_trace_enable_show/store()

sysfs_blk_trace_enable_show()/store() share most of code with sysfs_blk_trace_attr_show()/store().

---
 kernel/trace/blktrace.c |   92 +++++++++++-----------------------------------
 1 files changed, 22 insertions(+), 70 deletions(-)
kernel/trace/blktrace.c +22 -70
index 14986af..dfee6f9 100644
@@ -1284,72 +1284,6 @@ static int blk_trace_setup_queue(struct request_queue *q, dev_t dev)
  * sysfs interface to enable and configure tracing
  */
 
-static ssize_t sysfs_blk_trace_enable_show(struct device *dev,
-					   struct device_attribute *attr,
-					   char *buf)
-{
-	struct hd_struct *p = dev_to_part(dev);
-	struct block_device *bdev;
-	ssize_t ret = -ENXIO;
-
-	lock_kernel();
-	bdev = bdget(part_devt(p));
-	if (bdev != NULL) {
-		struct request_queue *q = bdev_get_queue(bdev);
-
-		if (q != NULL) {
-			mutex_lock(&bdev->bd_mutex);
-			ret = sprintf(buf, "%u\n", !!q->blk_trace);
-			mutex_unlock(&bdev->bd_mutex);
-		}
-
-		bdput(bdev);
-	}
-
-	unlock_kernel();
-	return ret;
-}
-
-static ssize_t sysfs_blk_trace_enable_store(struct device *dev,
-					    struct device_attribute *attr,
-					    const char *buf, size_t count)
-{
-	struct block_device *bdev;
-	struct request_queue *q;
-	struct hd_struct *p;
-	int value;
-	ssize_t ret = -ENXIO;
-
-	if (count == 0 || sscanf(buf, "%d", &value) != 1)
-		goto out;
-
-	lock_kernel();
-	p = dev_to_part(dev);
-	bdev = bdget(part_devt(p));
-	if (bdev == NULL)
-		goto out_unlock_kernel;
-
-	q = bdev_get_queue(bdev);
-	if (q == NULL)
-		goto out_bdput;
-
-	mutex_lock(&bdev->bd_mutex);
-	if (value)
-		ret = blk_trace_setup_queue(q, bdev->bd_dev);
-	else
-		ret = blk_trace_remove_queue(q);
-	mutex_unlock(&bdev->bd_mutex);
-
-	if (ret == 0)
-		ret = count;
-out_bdput:
-	bdput(bdev);
-out_unlock_kernel:
-	unlock_kernel();
-out:
-	return ret;
-}
-
 static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 					 struct device_attribute *attr,
 					 char *buf);
@@ -1361,8 +1295,7 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 		    sysfs_blk_trace_attr_show, \
 		    sysfs_blk_trace_attr_store)
 
-static DEVICE_ATTR(enable, S_IRUGO | S_IWUSR,
-		   sysfs_blk_trace_enable_show, sysfs_blk_trace_enable_store);
+static BLK_TRACE_DEVICE_ATTR(enable);
 static BLK_TRACE_DEVICE_ATTR(act_mask);
 static BLK_TRACE_DEVICE_ATTR(pid);
 static BLK_TRACE_DEVICE_ATTR(start_lba);
@@ -1447,6 +1380,12 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 	if (q == NULL)
 		goto out_bdput;
 	mutex_lock(&bdev->bd_mutex);
+
+	if (attr == &dev_attr_enable) {
+		ret = sprintf(buf, "%u\n", !!q->blk_trace);
+		goto out_unlock_bdev;
+	}
+
 	if (q->blk_trace == NULL)
 		ret = sprintf(buf, "disabled\n");
 	else if (attr == &dev_attr_act_mask)
@@ -1457,6 +1396,8 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 		ret = sprintf(buf, "%llu\n", q->blk_trace->start_lba);
 	else if (attr == &dev_attr_end_lba)
 		ret = sprintf(buf, "%llu\n", q->blk_trace->end_lba);
+
+out_unlock_bdev:
 	mutex_unlock(&bdev->bd_mutex);
 out_bdput:
 	bdput(bdev);
@@ -1499,6 +1440,15 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 		goto out_bdput;
 
 	mutex_lock(&bdev->bd_mutex);
+
+	if (attr == &dev_attr_enable) {
+		if (value)
+			ret = blk_trace_setup_queue(q, bdev->bd_dev);
+		else
+			ret = blk_trace_remove_queue(q);
+		goto out_unlock_bdev;
+	}
+
 	ret = 0;
 	if (q->blk_trace == NULL)
 		ret = blk_trace_setup_queue(q, bdev->bd_dev);
@@ -1512,13 +1462,15 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 			q->blk_trace->start_lba = value;
 		else if (attr == &dev_attr_end_lba)
 			q->blk_trace->end_lba = value;
-		ret = count;
 	}
+
+out_unlock_bdev:
 	mutex_unlock(&bdev->bd_mutex);
 out_bdput:
 	bdput(bdev);
 out_unlock_kernel:
 	unlock_kernel();
 out:
-	return ret;
+	return ret ? ret : count;
 }
+
Signature · 1 line
1.5.4.rc3
#6 lore raw

Re: [PATCH 2/7] blktrace: make blk_tracer_enabled a bool flag

In reply to Li Zefan
Quoted 46 lines · Li Zefan

It doesn't have to be a counter, and it can be a bool flag instead.

---
 kernel/trace/blktrace.c |   12 +++---------
 1 files changed, 3 insertions(+), 9 deletions(-)
kernel/trace/blktrace.c +3 -9
index fb3bc53..73845b7 100644
@@ -30,7 +30,7 @@
 static unsigned int blktrace_seq __read_mostly = 1;
 
 static struct trace_array *blk_tr;
-static int __read_mostly  blk_tracer_enabled;
+static bool blk_tracer_enabled __read_mostly;
 
 /* Select an alternative, minimalistic output than the original one */
 #define TRACE_BLK_OPT_CLASSIC	0x1
@@ -1111,9 +1111,7 @@ static int blk_tracer_init(struct trace_array *tr)
 {
 	blk_tr = tr;
 	blk_tracer_start(tr);
-	mutex_lock(&blk_probe_mutex);
-	blk_tracer_enabled++;
-	mutex_unlock(&blk_probe_mutex);
+	blk_tracer_enabled = true;
 	return 0;
 }
 
@@ -1131,11 +1129,7 @@ static void blk_tracer_reset(struct trace_array *tr)
 	if (!atomic_read(&blk_probes_ref))
 		return;
 
-	mutex_lock(&blk_probe_mutex);
-	blk_tracer_enabled--;
-	WARN_ON(blk_tracer_enabled < 0);
-	mutex_unlock(&blk_probe_mutex);
-
+	blk_tracer_enabled = false;
 	blk_tracer_stop(tr);
 }
 
Signature · 1 line
1.5.4.rc3

Looks good.
Indeed the init() and reset() callbacks are not supposed to be called nested.

#11 lore raw

Re: [PATCH 3/7] blktrace: remove blk_probe_mutex

In reply to Li Zefan
Quoted 37 lines · Li Zefan

blk_register_tracepoints() always returns 0, so make it return void, thus we don't need to use blk_probe_mutex to protect blk_probes_ref.

---
 kernel/trace/blktrace.c |   27 +++++----------------------
 1 files changed, 5 insertions(+), 22 deletions(-)
kernel/trace/blktrace.c +1 -4
index 73845b7..223b92e 100644
@@ -47,10 +47,9 @@ static struct tracer_flags blk_tracer_flags = {
 };
 
 /* Global reference count of probes */
-static DEFINE_MUTEX(blk_probe_mutex);
 static atomic_t blk_probes_ref = ATOMIC_INIT(0);
 
-static int blk_register_tracepoints(void);
+static void blk_register_tracepoints(void);
 static void blk_unregister_tracepoints(void);
 
 /*
@@ -256,10 +255,8 @@ static void blk_trace_cleanup(struct blk_trace *bt)
 	free_percpu(bt->sequence);
 	free_percpu(bt->msg_data);
 	kfree(bt);
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_dec_and_test(&blk_probes_ref))
 		blk_unregister_tracepoints();
-	mutex_unlock(&blk_probe_mutex);
 }
 
 int blk_trace_remove(struct request_queue *q)
@@ -471,13 +468,8 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	bt->pid = buts->pid;

kernel/trace/blktrace.c int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev, · line 471

Quoted 10 lines
kernel/trace/blktrace.c +1 -7
 	bt->trace_state = Blktrace_setup;
 
-	mutex_lock(&blk_probe_mutex);
-	if (atomic_add_return(1, &blk_probes_ref) == 1) {
-		ret = blk_register_tracepoints();
-		if (ret)
-			goto probe_err;
-	}
-	mutex_unlock(&blk_probe_mutex);
+	if (atomic_add_return(1, &blk_probes_ref) == 1)

atomic_inc_return() is a bit more simple.

Quoted 58 lines
kernel/trace/blktrace.c +3 -11
+		blk_register_tracepoints();
 
 	ret = -EBUSY;
 	old_bt = xchg(&q->blk_trace, bt);
@@ -487,9 +479,6 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	}
 
 	return 0;
-probe_err:
-	atomic_dec(&blk_probes_ref);
-	mutex_unlock(&blk_probe_mutex);
 err:
 	if (bt) {
 		if (bt->msg_file)
@@ -863,7 +852,7 @@ void blk_add_driver_data(struct request_queue *q,
 }
 EXPORT_SYMBOL_GPL(blk_add_driver_data);
 
-static int blk_register_tracepoints(void)
+static void blk_register_tracepoints(void)
 {
 	int ret;
 
@@ -901,7 +890,6 @@ static int blk_register_tracepoints(void)
 	WARN_ON(ret);
 	ret = register_trace_block_remap(blk_add_trace_remap);
 	WARN_ON(ret);
-	return 0;
 }
 
 static void blk_unregister_tracepoints(void)
@@ -1099,11 +1087,8 @@ static void blk_tracer_print_header(struct seq_file *m)
 
 static void blk_tracer_start(struct trace_array *tr)
 {
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_add_return(1, &blk_probes_ref) == 1)
-		if (blk_register_tracepoints())
-			atomic_dec(&blk_probes_ref);
-	mutex_unlock(&blk_probe_mutex);
+		blk_register_tracepoints();
 	trace_flags &= ~TRACE_ITER_CONTEXT_INFO;
 }
 
@@ -1118,10 +1103,8 @@ static int blk_tracer_init(struct trace_array *tr)
 static void blk_tracer_stop(struct trace_array *tr)
 {
 	trace_flags |= TRACE_ITER_CONTEXT_INFO;
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_dec_and_test(&blk_probes_ref))
 		blk_unregister_tracepoints();
-	mutex_unlock(&blk_probe_mutex);
 }
 
 static void blk_tracer_reset(struct trace_array *tr)
Signature · 1 line
1.5.4.rc3

Looks good.

#36 lore raw
In reply to Li Zefan
Quoted 15 lines · Li Zefan

While trying to use blktrace in -tip tree, I encounted kernel NULL dereference, so I looked into the code, and then found some other bugs.

This patchset is part 1, and I have some other pending fixes.

Each patch is small and straightforward, and should be easy to review:

[PATCH 1/7] blktrace: fix possible memory leak
[PATCH 2/7] blktrace: make blk_tracer_enabled a bool flag
[PATCH 3/7] blktrace: remove blk_probe_mutex
[PATCH 4/7] blktrace: don't increase blk_probes_ref if failed to setup blk trace [PATCH 5/7] blktrace: report EBUSY correctly
[PATCH 6/7] blktrace: remove sysfs_blk_trace_enable_show/store() [PATCH 7/7] blktrace: avoid accessing NULL bdev->bd_disk

Quoted 5 lines
---
 blktrace.c |  154 +++++++++++++++++--------------------------------------------
 1 file changed, 45 insertions(+), 109 deletions(-)

Nice fixes. Applied to tip:tracing/blktrace, thanks!

Ingo

author
#2 lore raw

[tip:tracing/blktrace] blktrace: fix possible memory leak

In reply to Li Zefan
Commit-ID:  86d7c7acc6b5ec500f319120f7e72ea62fb6252d
Gitweb:     http://git.kernel.org/tip/86d7c7acc6b5ec500f319120f7e72ea62fb6252d
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 09:47:30 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Fri, 20 Mar 2009 11:20:04 +0100

blktrace: fix possible memory leak

When we failed to create "block" debugfs dir, we should do some cleanups.

Cc Jens Axboe <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
kernel/trace/blktrace.c +1 -1
index b171778..fb3bc53 100644
@@ -432,7 +432,7 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	if (!blk_tree_root) {
 		blk_tree_root = debugfs_create_dir("block", NULL);
 		if (!blk_tree_root)
-			return -ENOMEM;
+			goto err;
 	}
 
 	dir = debugfs_create_dir(buts->name, blk_tree_root);
author
#12 lore raw

[tip:tracing/blktrace] blktrace: remove blk_probe_mutex

In reply to Li Zefan
Commit-ID:  a2df44c805026d40fae556b04f9addbce486bbe2
Gitweb:     http://git.kernel.org/tip/a2df44c805026d40fae556b04f9addbce486bbe2
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 09:48:26 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Fri, 20 Mar 2009 11:20:05 +0100

blktrace: remove blk_probe_mutex

blk_register_tracepoints() always returns 0, so make it return void, thus we don't need to use blk_probe_mutex to protect blk_probes_ref.

Cc Jens Axboe <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |   27 +++++----------------------
 1 files changed, 5 insertions(+), 22 deletions(-)
kernel/trace/blktrace.c +5 -22
index 73845b7..223b92e 100644
@@ -47,10 +47,9 @@ static struct tracer_flags blk_tracer_flags = {
 };
 
 /* Global reference count of probes */
-static DEFINE_MUTEX(blk_probe_mutex);
 static atomic_t blk_probes_ref = ATOMIC_INIT(0);
 
-static int blk_register_tracepoints(void);
+static void blk_register_tracepoints(void);
 static void blk_unregister_tracepoints(void);
 
 /*
@@ -256,10 +255,8 @@ static void blk_trace_cleanup(struct blk_trace *bt)
 	free_percpu(bt->sequence);
 	free_percpu(bt->msg_data);
 	kfree(bt);
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_dec_and_test(&blk_probes_ref))
 		blk_unregister_tracepoints();
-	mutex_unlock(&blk_probe_mutex);
 }
 
 int blk_trace_remove(struct request_queue *q)
@@ -471,13 +468,8 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	bt->pid = buts->pid;
 	bt->trace_state = Blktrace_setup;
 
-	mutex_lock(&blk_probe_mutex);
-	if (atomic_add_return(1, &blk_probes_ref) == 1) {
-		ret = blk_register_tracepoints();
-		if (ret)
-			goto probe_err;
-	}
-	mutex_unlock(&blk_probe_mutex);
+	if (atomic_add_return(1, &blk_probes_ref) == 1)
+		blk_register_tracepoints();
 
 	ret = -EBUSY;
 	old_bt = xchg(&q->blk_trace, bt);
@@ -487,9 +479,6 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	}
 
 	return 0;
-probe_err:
-	atomic_dec(&blk_probes_ref);
-	mutex_unlock(&blk_probe_mutex);
 err:
 	if (bt) {
 		if (bt->msg_file)
@@ -863,7 +852,7 @@ void blk_add_driver_data(struct request_queue *q,
 }
 EXPORT_SYMBOL_GPL(blk_add_driver_data);
 
-static int blk_register_tracepoints(void)
+static void blk_register_tracepoints(void)
 {
 	int ret;
 
@@ -901,7 +890,6 @@ static int blk_register_tracepoints(void)
 	WARN_ON(ret);
 	ret = register_trace_block_remap(blk_add_trace_remap);
 	WARN_ON(ret);
-	return 0;
 }
 
 static void blk_unregister_tracepoints(void)
@@ -1099,11 +1087,8 @@ static void blk_tracer_print_header(struct seq_file *m)
 
 static void blk_tracer_start(struct trace_array *tr)
 {
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_add_return(1, &blk_probes_ref) == 1)
-		if (blk_register_tracepoints())
-			atomic_dec(&blk_probes_ref);
-	mutex_unlock(&blk_probe_mutex);
+		blk_register_tracepoints();
 	trace_flags &= ~TRACE_ITER_CONTEXT_INFO;
 }
 
@@ -1118,10 +1103,8 @@ static int blk_tracer_init(struct trace_array *tr)
 static void blk_tracer_stop(struct trace_array *tr)
 {
 	trace_flags |= TRACE_ITER_CONTEXT_INFO;
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_dec_and_test(&blk_probes_ref))
 		blk_unregister_tracepoints();
-	mutex_unlock(&blk_probe_mutex);
 }
 
 static void blk_tracer_reset(struct trace_array *tr)
author
#7 lore raw

[tip:tracing/blktrace] blktrace: make blk_tracer_enabled a bool flag

In reply to Li Zefan
Commit-ID:  e73fe657c9a1edba1611b4642b2d226634c1a652
Gitweb:     http://git.kernel.org/tip/e73fe657c9a1edba1611b4642b2d226634c1a652
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 09:48:03 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Fri, 20 Mar 2009 11:20:04 +0100

blktrace: make blk_tracer_enabled a bool flag

It doesn't have to be a counter, and it can be a bool flag instead.

Cc Jens Axboe <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |   12 +++---------
 1 files changed, 3 insertions(+), 9 deletions(-)
kernel/trace/blktrace.c +3 -9
index fb3bc53..73845b7 100644
@@ -30,7 +30,7 @@
 static unsigned int blktrace_seq __read_mostly = 1;
 
 static struct trace_array *blk_tr;
-static int __read_mostly  blk_tracer_enabled;
+static bool blk_tracer_enabled __read_mostly;
 
 /* Select an alternative, minimalistic output than the original one */
 #define TRACE_BLK_OPT_CLASSIC	0x1
@@ -1111,9 +1111,7 @@ static int blk_tracer_init(struct trace_array *tr)
 {
 	blk_tr = tr;
 	blk_tracer_start(tr);
-	mutex_lock(&blk_probe_mutex);
-	blk_tracer_enabled++;
-	mutex_unlock(&blk_probe_mutex);
+	blk_tracer_enabled = true;
 	return 0;
 }
 
@@ -1131,11 +1129,7 @@ static void blk_tracer_reset(struct trace_array *tr)
 	if (!atomic_read(&blk_probes_ref))
 		return;
 
-	mutex_lock(&blk_probe_mutex);
-	blk_tracer_enabled--;
-	WARN_ON(blk_tracer_enabled < 0);
-	mutex_unlock(&blk_probe_mutex);
-
+	blk_tracer_enabled = false;
 	blk_tracer_stop(tr);
 }
 
author
#22 lore raw

[tip:tracing/blktrace] blktrace: report EBUSY correctly

In reply to Li Zefan
Commit-ID:  55feae54fd7ccfcbdf198e34e05ff61d9e975ed3
Gitweb:     http://git.kernel.org/tip/55feae54fd7ccfcbdf198e34e05ff61d9e975ed3
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 09:49:08 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Fri, 20 Mar 2009 11:20:06 +0100

blktrace: report EBUSY correctly

blk_trace_remove_queue() returns EINVAL if q->blk_trace == NULL, but blk_trace_setup_queue() doesn't return EBUSY if q->blk_trace != NULL.

 # echo 0 > sdaX/trace/enable
 # echo 0 > sdaX/trace/enable
 bash: echo: write error: Invalid argument
 # echo 1 > sdaX/trace/enable
 # echo 1 > sdaX/trace/enable
 (should return EBUSY)
Cc Jens Axboe <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |    9 +++------
 1 files changed, 3 insertions(+), 6 deletions(-)
kernel/trace/blktrace.c +3 -6
index 11e7c8d..14986af 100644
@@ -1260,12 +1260,10 @@ static int blk_trace_remove_queue(struct request_queue *q)
 static int blk_trace_setup_queue(struct request_queue *q, dev_t dev)
 {
 	struct blk_trace *old_bt, *bt = NULL;
-	int ret;
 
-	ret = -ENOMEM;
 	bt = kzalloc(sizeof(*bt), GFP_KERNEL);
 	if (!bt)
-		goto err;
+		return -ENOMEM;
 
 	bt->dev = dev;
 	bt->act_mask = (u16)-1;
@@ -1276,11 +1274,10 @@ static int blk_trace_setup_queue(struct request_queue *q, dev_t dev)
 	if (old_bt != NULL) {
 		(void)xchg(&q->blk_trace, old_bt);
 		kfree(bt);
-		ret = -EBUSY;
+		return -EBUSY;
 	}
+
 	return 0;
-err:
-	return ret;
 }
 
 /*
author
#28 lore raw

[tip:tracing/blktrace] blktrace: remove sysfs_blk_trace_enable_show/store()

In reply to Li Zefan
Commit-ID:  9626d7ee4addbf97245c129e24ba9a9b765bd5ef
Gitweb:     http://git.kernel.org/tip/9626d7ee4addbf97245c129e24ba9a9b765bd5ef
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 11:33:55 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Fri, 20 Mar 2009 11:20:06 +0100

blktrace: remove sysfs_blk_trace_enable_show/store()

sysfs_blk_trace_enable_show()/store() share most of code with sysfs_blk_trace_attr_show()/store().

Cc Jens Axboe <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |   92 +++++++++++-----------------------------------
 1 files changed, 22 insertions(+), 70 deletions(-)
kernel/trace/blktrace.c +22 -70
index 14986af..dfee6f9 100644
@@ -1284,72 +1284,6 @@ static int blk_trace_setup_queue(struct request_queue *q, dev_t dev)
  * sysfs interface to enable and configure tracing
  */
 
-static ssize_t sysfs_blk_trace_enable_show(struct device *dev,
-					   struct device_attribute *attr,
-					   char *buf)
-{
-	struct hd_struct *p = dev_to_part(dev);
-	struct block_device *bdev;
-	ssize_t ret = -ENXIO;
-
-	lock_kernel();
-	bdev = bdget(part_devt(p));
-	if (bdev != NULL) {
-		struct request_queue *q = bdev_get_queue(bdev);
-
-		if (q != NULL) {
-			mutex_lock(&bdev->bd_mutex);
-			ret = sprintf(buf, "%u\n", !!q->blk_trace);
-			mutex_unlock(&bdev->bd_mutex);
-		}
-
-		bdput(bdev);
-	}
-
-	unlock_kernel();
-	return ret;
-}
-
-static ssize_t sysfs_blk_trace_enable_store(struct device *dev,
-					    struct device_attribute *attr,
-					    const char *buf, size_t count)
-{
-	struct block_device *bdev;
-	struct request_queue *q;
-	struct hd_struct *p;
-	int value;
-	ssize_t ret = -ENXIO;
-
-	if (count == 0 || sscanf(buf, "%d", &value) != 1)
-		goto out;
-
-	lock_kernel();
-	p = dev_to_part(dev);
-	bdev = bdget(part_devt(p));
-	if (bdev == NULL)
-		goto out_unlock_kernel;
-
-	q = bdev_get_queue(bdev);
-	if (q == NULL)
-		goto out_bdput;
-
-	mutex_lock(&bdev->bd_mutex);
-	if (value)
-		ret = blk_trace_setup_queue(q, bdev->bd_dev);
-	else
-		ret = blk_trace_remove_queue(q);
-	mutex_unlock(&bdev->bd_mutex);
-
-	if (ret == 0)
-		ret = count;
-out_bdput:
-	bdput(bdev);
-out_unlock_kernel:
-	unlock_kernel();
-out:
-	return ret;
-}
-
 static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 					 struct device_attribute *attr,
 					 char *buf);
@@ -1361,8 +1295,7 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 		    sysfs_blk_trace_attr_show, \
 		    sysfs_blk_trace_attr_store)
 
-static DEVICE_ATTR(enable, S_IRUGO | S_IWUSR,
-		   sysfs_blk_trace_enable_show, sysfs_blk_trace_enable_store);
+static BLK_TRACE_DEVICE_ATTR(enable);
 static BLK_TRACE_DEVICE_ATTR(act_mask);
 static BLK_TRACE_DEVICE_ATTR(pid);
 static BLK_TRACE_DEVICE_ATTR(start_lba);
@@ -1447,6 +1380,12 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 	if (q == NULL)
 		goto out_bdput;
 	mutex_lock(&bdev->bd_mutex);
+
+	if (attr == &dev_attr_enable) {
+		ret = sprintf(buf, "%u\n", !!q->blk_trace);
+		goto out_unlock_bdev;
+	}
+
 	if (q->blk_trace == NULL)
 		ret = sprintf(buf, "disabled\n");
 	else if (attr == &dev_attr_act_mask)
@@ -1457,6 +1396,8 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 		ret = sprintf(buf, "%llu\n", q->blk_trace->start_lba);
 	else if (attr == &dev_attr_end_lba)
 		ret = sprintf(buf, "%llu\n", q->blk_trace->end_lba);
+
+out_unlock_bdev:
 	mutex_unlock(&bdev->bd_mutex);
 out_bdput:
 	bdput(bdev);
@@ -1499,6 +1440,15 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 		goto out_bdput;
 
 	mutex_lock(&bdev->bd_mutex);
+
+	if (attr == &dev_attr_enable) {
+		if (value)
+			ret = blk_trace_setup_queue(q, bdev->bd_dev);
+		else
+			ret = blk_trace_remove_queue(q);
+		goto out_unlock_bdev;
+	}
+
 	ret = 0;
 	if (q->blk_trace == NULL)
 		ret = blk_trace_setup_queue(q, bdev->bd_dev);
@@ -1512,13 +1462,15 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 			q->blk_trace->start_lba = value;
 		else if (attr == &dev_attr_end_lba)
 			q->blk_trace->end_lba = value;
-		ret = count;
 	}
+
+out_unlock_bdev:
 	mutex_unlock(&bdev->bd_mutex);
 out_bdput:
 	bdput(bdev);
 out_unlock_kernel:
 	unlock_kernel();
 out:
-	return ret;
+	return ret ? ret : count;
 }
+
author
#18 lore raw

[tip:tracing/blktrace] blktrace: don't increase blk_probes_ref if failed to setup blk trace

In reply to Li Zefan
Commit-ID:  64ddedc03c9ecbe97b0700e85c5f11d2f5a4565c
Gitweb:     http://git.kernel.org/tip/64ddedc03c9ecbe97b0700e85c5f11d2f5a4565c
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 09:48:47 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Fri, 20 Mar 2009 11:20:05 +0100

blktrace: don't increase blk_probes_ref if failed to setup blk trace

do_blk_trace_setup() may return EBUSY, but the current code doesn't decrease blk_probes_ref in this case.

Cc Jens Axboe <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)
kernel/trace/blktrace.c +3 -3
index 223b92e..11e7c8d 100644
@@ -468,9 +468,6 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	bt->pid = buts->pid;
 	bt->trace_state = Blktrace_setup;
 
-	if (atomic_add_return(1, &blk_probes_ref) == 1)
-		blk_register_tracepoints();
-
 	ret = -EBUSY;
 	old_bt = xchg(&q->blk_trace, bt);
 	if (old_bt) {
@@ -478,6 +475,9 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 		goto err;
 	}
 
+	if (atomic_add_return(1, &blk_probes_ref) == 1)
+		blk_register_tracepoints();
+
 	return 0;
 err:
 	if (bt) {
author
#33 lore raw

[tip:tracing/blktrace] blktrace: avoid accessing NULL bdev->bd_disk

In reply to Li Zefan
Commit-ID:  afaebb6720cc62df1b03e36c2f699b00cfee784e
Gitweb:     http://git.kernel.org/tip/afaebb6720cc62df1b03e36c2f699b00cfee784e
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 10:34:00 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Fri, 20 Mar 2009 11:20:07 +0100

blktrace: avoid accessing NULL bdev->bd_disk

bdev->bd_disk can be NULL, if the block device is not opened.

Try this against an unmounted partition, and you'll see NULL dereference:

  # echo 1 > /sys/block/sda/sda5/enable
Cc Jens Axboe <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Steven Rostedt <[email protected]>

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |   13 +++++++++++--
 1 files changed, 11 insertions(+), 2 deletions(-)
kernel/trace/blktrace.c +11 -2
index dfee6f9..108f4f7 100644
@@ -1362,6 +1362,14 @@ static int blk_str2act_mask(const char *str)
 	return mask;
 }
 
+static struct request_queue *blk_trace_get_queue(struct block_device *bdev)
+{
+	if (bdev->bd_disk == NULL)
+		return NULL;
+
+	return bdev_get_queue(bdev);
+}
+
 static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 					 struct device_attribute *attr,
 					 char *buf)
@@ -1376,9 +1384,10 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 	if (bdev == NULL)
 		goto out_unlock_kernel;
 
-	q = bdev_get_queue(bdev);
+	q = blk_trace_get_queue(bdev);
 	if (q == NULL)
 		goto out_bdput;
+
 	mutex_lock(&bdev->bd_mutex);
 
 	if (attr == &dev_attr_enable) {
@@ -1435,7 +1444,7 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 	if (bdev == NULL)
 		goto out_unlock_kernel;
 
-	q = bdev_get_queue(bdev);
+	q = blk_trace_get_queue(bdev);
 	if (q == NULL)
 		goto out_bdput;
 
#37 lore raw
In reply to Ingo Molnar
Quoted 22 lines · Ingo Molnar
Quoted 19 lines · Li Zefan

While trying to use blktrace in -tip tree, I encounted kernel NULL dereference, so I looked into the code, and then found some other bugs.

This patchset is part 1, and I have some other pending fixes.

Each patch is small and straightforward, and should be easy to review:

[PATCH 1/7] blktrace: fix possible memory leak
[PATCH 2/7] blktrace: make blk_tracer_enabled a bool flag
[PATCH 3/7] blktrace: remove blk_probe_mutex
[PATCH 4/7] blktrace: don't increase blk_probes_ref if failed to setup blk trace [PATCH 5/7] blktrace: report EBUSY correctly
[PATCH 6/7] blktrace: remove sysfs_blk_trace_enable_show/store() [PATCH 7/7] blktrace: avoid accessing NULL bdev->bd_disk

---
 blktrace.c |  154 +++++++++++++++++--------------------------------------------
 1 file changed, 45 insertions(+), 109 deletions(-)
Quoted 3 lines
Quoted 1 line

Nice fixes. Applied to tip:tracing/blktrace, thanks!

Which would go upstream in the .30 merge window (unless Jens or Steve objects of course).

Ingo

Acked-by on v1 1/7
#3 lore raw

Re: [PATCH 1/7] blktrace: fix possible memory leak

In reply to Li Zefan

Em Fri, Mar 20, 2009 at 09:47:30AM +0800, Li Zefan escreveu:

Quoted 12 lines

When we failed to create "block" debugfs dir, we should do come cleanups.

---
 kernel/trace/blktrace.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
kernel/trace/blktrace.c +0 -0
index b171778..fb3bc53 100644

kernel/trace/blktrace.c int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev, · line 438

Quoted 8 lines
kernel/trace/blktrace.c +1 -1
@@ -432,7 +432,7 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	if (!blk_tree_root) {
 		blk_tree_root = debugfs_create_dir("block", NULL);
 		if (!blk_tree_root)
-			return -ENOMEM;
+			goto err;
 	}
 
 	dir = debugfs_create_dir(buts->name, blk_tree_root);

Nice catch,

Acked-by on v1 2/7
#8 lore raw

Re: [PATCH 2/7] blktrace: make blk_tracer_enabled a bool flag

In reply to Li Zefan

Em Fri, Mar 20, 2009 at 09:48:03AM +0800, Li Zefan escreveu:

Quoted 33 lines

It doesn't have to be a counter, and it can be a bool flag instead.

---
 kernel/trace/blktrace.c |   12 +++---------
 1 files changed, 3 insertions(+), 9 deletions(-)
kernel/trace/blktrace.c +2 -4
index fb3bc53..73845b7 100644
@@ -30,7 +30,7 @@
 static unsigned int blktrace_seq __read_mostly = 1;
 
 static struct trace_array *blk_tr;
-static int __read_mostly  blk_tracer_enabled;
+static bool blk_tracer_enabled __read_mostly;
 
 /* Select an alternative, minimalistic output than the original one */
 #define TRACE_BLK_OPT_CLASSIC	0x1
@@ -1111,9 +1111,7 @@ static int blk_tracer_init(struct trace_array *tr)
 {
 	blk_tr = tr;
 	blk_tracer_start(tr);
-	mutex_lock(&blk_probe_mutex);
-	blk_tracer_enabled++;
-	mutex_unlock(&blk_probe_mutex);
+	blk_tracer_enabled = true;
 	return 0;
 }
 
@@ -1131,11 +1129,7 @@ static void blk_tracer_reset(struct trace_array *tr)
 	if (!atomic_read(&blk_probes_ref))

kernel/trace/blktrace.c static void blk_tracer_reset(struct trace_array *tr) · line 1134

Quoted 10 lines
kernel/trace/blktrace.c +1 -5
 		return;
 
-	mutex_lock(&blk_probe_mutex);
-	blk_tracer_enabled--;
-	WARN_ON(blk_tracer_enabled < 0);
-	mutex_unlock(&blk_probe_mutex);
-
+	blk_tracer_enabled = false;
 	blk_tracer_stop(tr);
 }

Copy'n'pasted code, some other tracer still have this style:

enable_branch_tracing, enable_mmiotrace,
tracing_start_sched_switch_record

Anyway, thanks,

- Arnaldo

#13 lore raw

Re: [PATCH 3/7] blktrace: remove blk_probe_mutex

In reply to Li Zefan

Em Fri, Mar 20, 2009 at 09:48:26AM +0800, Li Zefan escreveu:

Quoted 4 lines

blk_register_tracepoints() always returns 0, so make it return void, thus we don't need to use blk_probe_mutex to protect blk_probes_ref.

Historic reasons, when I first worked on this I found all those WARN_ONs in blk_register_tracepoints a bogosity that should be later fixed, but to reduce the patch size I never got around to submit the proper patch, so I don't think that the fix is what you suggests, here it is what I had before I removed the non-essential parts of the patch:

+static void blk_tracer_error(const char *name)
+{
+ pr_info("blk trace: Couldn't activate tracepoint "
+ "probe to block_%s\n", name);
+}
+
+#define register_trace_block(tpoint) \
+ ret = register_trace_block_##tpoint(blk_add_trace_##tpoint); \
+ if (ret) { \
+ blk_tracer_error(#tpoint); \
+ goto *exit_point; \
+ } else \
+ exit_point = &&fail_deprobe_##tpoint;
+
+
+#define fail_trace_block(tpoint) \
+ fail_deprobe_##tpoint: \
+ unregister_trace_block_##tpoint(blk_add_trace_##tpoint)
+
+static int tracing_blk_register(void)
+{
+ int ret;
+ void *exit_point = &&error;
+
+ register_trace_block(rq_abort);
+ register_trace_block(rq_insert);
+ register_trace_block(rq_issue);
+ register_trace_block(rq_requeue);
+ register_trace_block(rq_complete);
+ register_trace_block(bio_bounce);
+ register_trace_block(bio_backmerge);
+ register_trace_block(bio_frontmerge);
+ register_trace_block(bio_queue);
+ register_trace_block(getrq);
+ register_trace_block(sleeprq);
+ register_trace_block(plug);
+ register_trace_block(unplug_timer);
+ register_trace_block(unplug_io);
+ register_trace_block(split);
+ register_trace_block(remap);
+
+ return 0;
+
+ fail_trace_block(remap);
+ fail_trace_block(split);
+ fail_trace_block(unplug_io);
+ fail_trace_block(unplug_timer);
+ fail_trace_block(plug);
+ fail_trace_block(sleeprq);
+ fail_trace_block(getrq);
+ fail_trace_block(bio_queue);
+ fail_trace_block(bio_frontmerge);
+ fail_trace_block(bio_backmerge);
+ fail_trace_block(bio_bounce);
+ fail_trace_block(rq_complete);
+ fail_trace_block(rq_requeue);
+ fail_trace_block(rq_issue);
+ fail_trace_block(rq_insert);
+ fail_trace_block(rq_abort);
+error:
+ return ret;
+}
+
+static void tracing_blk_unregister(void)
+{
+ unregister_trace_block_remap(blk_add_trace_remap);
+ unregister_trace_block_split(blk_add_trace_split);
+ unregister_trace_block_unplug_io(blk_add_trace_unplug_io);
+ unregister_trace_block_unplug_timer(blk_add_trace_unplug_timer);
+ unregister_trace_block_plug(blk_add_trace_plug);
+ unregister_trace_block_sleeprq(blk_add_trace_sleeprq);
+ unregister_trace_block_getrq(blk_add_trace_getrq);
+ unregister_trace_block_bio_queue(blk_add_trace_bio_queue);
+ unregister_trace_block_bio_frontmerge(blk_add_trace_bio_frontmerge);
+ unregister_trace_block_bio_backmerge(blk_add_trace_bio_backmerge);
+ unregister_trace_block_bio_complete(blk_add_trace_bio_complete);
+ unregister_trace_block_bio_bounce(blk_add_trace_bio_bounce);
+ unregister_trace_block_rq_complete(blk_add_trace_rq_complete);
+ unregister_trace_block_rq_requeue(blk_add_trace_rq_requeue);
+ unregister_trace_block_rq_issue(blk_add_trace_rq_issue);
+ unregister_trace_block_rq_insert(blk_add_trace_rq_insert);
+ unregister_trace_block_rq_abort(blk_add_trace_rq_abort);
+ tracepoint_synchronize_unregister();
+}

Regards,

- Arnaldo

Acked-by on v1 7/7
#35 lore raw

Re: [PATCH 7/7] blktrace: avoid accessing NULL bdev->bd_disk

In reply to Li Zefan

Em Fri, Mar 20, 2009 at 09:49:48AM +0800, Li Zefan escreveu:

Quoted 6 lines

bdev->bd_disk can be NULL, if the block device is not opened.

Try this against an unmounted partition, and you'll see NULL dereference:

  # echo 1 > /sys/block/sda/sda5/enable

I guess this can be discounted on my block layer newbeeness 8)

But as an extra check in non hot-paths are usually a good thing...

- Arnaldo

Acked-by on v1 5/7
#23 lore raw

Re: [PATCH 5/7] blktrace: report EBUSY correctly

In reply to Li Zefan

Em Fri, Mar 20, 2009 at 09:49:08AM +0800, Li Zefan escreveu:

Quoted 11 lines

blk_trace_remove_queue() returns EINVAL if q->blk_trace == NULL, but blk_trace_setup_queue() doesn't return EBUSY if q->blk_trace != NULL.

 # echo 0 > sdaX/trace/enable
 # echo 0 > sdaX/trace/enable
 bash: echo: write error: Invalid argument
 # echo 1 > sdaX/trace/enable
 # echo 1 > sdaX/trace/enable
 (should return EBUSY)

Thanks for all the testing, its being really appreciated,

- Arnaldo

#24 lore raw

Re: [PATCH 5/7] blktrace: report EBUSY correctly

Quoted 11 lines · Arnaldo Carvalho de Melo

Em Fri, Mar 20, 2009 at 09:49:08AM +0800, Li Zefan escreveu:

Quoted 10 lines

blk_trace_remove_queue() returns EINVAL if q->blk_trace == NULL, but blk_trace_setup_queue() doesn't return EBUSY if q->blk_trace != NULL.

 # echo 0 > sdaX/trace/enable
 # echo 0 > sdaX/trace/enable
 bash: echo: write error: Invalid argument
 # echo 1 > sdaX/trace/enable
 # echo 1 > sdaX/trace/enable
 (should return EBUSY)
Quoted 5 lines
Quoted 1 line

Thanks for all the testing, its being really appreciated,

thanks Arnaldo - i've added your acks to tip:tracing/blktrace.

Jens, are the changes fine with you too, can i merge them into tracing/core, for v2.6.30?

Ingo

author
#4 lore raw

[tip:tracing/blktrace] blktrace: fix possible memory leak

In reply to Li Zefan
Commit-ID:  1a17662ea033674a58bad3603531b0b5d42572f6
Gitweb:     http://git.kernel.org/tip/1a17662ea033674a58bad3603531b0b5d42572f6
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 09:47:30 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Sat, 21 Mar 2009 16:15:47 +0100

blktrace: fix possible memory leak

When we failed to create "block" debugfs dir, we should do some cleanups.

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
kernel/trace/blktrace.c +1 -1
index b171778..fb3bc53 100644
@@ -432,7 +432,7 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	if (!blk_tree_root) {
 		blk_tree_root = debugfs_create_dir("block", NULL);
 		if (!blk_tree_root)
-			return -ENOMEM;
+			goto err;
 	}
 
 	dir = debugfs_create_dir(buts->name, blk_tree_root);
author
#20 lore raw

[tip:tracing/blktrace] blktrace: don't increase blk_probes_ref if failed to setup blk trace

In reply to Li Zefan
Commit-ID:  cbe28296eb1ac441b35cf45804d0ae808add7dd1
Gitweb:     http://git.kernel.org/tip/cbe28296eb1ac441b35cf45804d0ae808add7dd1
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 09:48:47 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Sat, 21 Mar 2009 16:16:37 +0100

blktrace: don't increase blk_probes_ref if failed to setup blk trace

do_blk_trace_setup() may return EBUSY, but the current code doesn't decrease blk_probes_ref in this case.

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)
kernel/trace/blktrace.c +3 -3
index 223b92e..11e7c8d 100644
@@ -468,9 +468,6 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	bt->pid = buts->pid;
 	bt->trace_state = Blktrace_setup;
 
-	if (atomic_add_return(1, &blk_probes_ref) == 1)
-		blk_register_tracepoints();
-
 	ret = -EBUSY;
 	old_bt = xchg(&q->blk_trace, bt);
 	if (old_bt) {
@@ -478,6 +475,9 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 		goto err;
 	}
 
+	if (atomic_add_return(1, &blk_probes_ref) == 1)
+		blk_register_tracepoints();
+
 	return 0;
 err:
 	if (bt) {
author
#9 lore raw

[tip:tracing/blktrace] blktrace: make blk_tracer_enabled a bool flag

In reply to Li Zefan
Commit-ID:  5006ea73f38caef6065d1136808413813271633f
Gitweb:     http://git.kernel.org/tip/5006ea73f38caef6065d1136808413813271633f
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 09:48:03 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Sat, 21 Mar 2009 16:16:13 +0100

blktrace: make blk_tracer_enabled a bool flag

It doesn't have to be a counter, and it can be a bool flag instead.

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |   12 +++---------
 1 files changed, 3 insertions(+), 9 deletions(-)
kernel/trace/blktrace.c +3 -9
index fb3bc53..73845b7 100644
@@ -30,7 +30,7 @@
 static unsigned int blktrace_seq __read_mostly = 1;
 
 static struct trace_array *blk_tr;
-static int __read_mostly  blk_tracer_enabled;
+static bool blk_tracer_enabled __read_mostly;
 
 /* Select an alternative, minimalistic output than the original one */
 #define TRACE_BLK_OPT_CLASSIC	0x1
@@ -1111,9 +1111,7 @@ static int blk_tracer_init(struct trace_array *tr)
 {
 	blk_tr = tr;
 	blk_tracer_start(tr);
-	mutex_lock(&blk_probe_mutex);
-	blk_tracer_enabled++;
-	mutex_unlock(&blk_probe_mutex);
+	blk_tracer_enabled = true;
 	return 0;
 }
 
@@ -1131,11 +1129,7 @@ static void blk_tracer_reset(struct trace_array *tr)
 	if (!atomic_read(&blk_probes_ref))
 		return;
 
-	mutex_lock(&blk_probe_mutex);
-	blk_tracer_enabled--;
-	WARN_ON(blk_tracer_enabled < 0);
-	mutex_unlock(&blk_probe_mutex);
-
+	blk_tracer_enabled = false;
 	blk_tracer_stop(tr);
 }
 
author
#16 lore raw

[tip:tracing/blktrace] blktrace: remove blk_probe_mutex

In reply to Li Zefan
Commit-ID:  3c289ba7c320560ee74979a8895141c829046a2d
Gitweb:     http://git.kernel.org/tip/3c289ba7c320560ee74979a8895141c829046a2d
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 09:48:26 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Sat, 21 Mar 2009 16:16:25 +0100

blktrace: remove blk_probe_mutex

blk_register_tracepoints() always returns 0, so make it return void, thus we don't need to use blk_probe_mutex to protect blk_probes_ref.

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |   27 +++++----------------------
 1 files changed, 5 insertions(+), 22 deletions(-)
kernel/trace/blktrace.c +5 -22
index 73845b7..223b92e 100644
@@ -47,10 +47,9 @@ static struct tracer_flags blk_tracer_flags = {
 };
 
 /* Global reference count of probes */
-static DEFINE_MUTEX(blk_probe_mutex);
 static atomic_t blk_probes_ref = ATOMIC_INIT(0);
 
-static int blk_register_tracepoints(void);
+static void blk_register_tracepoints(void);
 static void blk_unregister_tracepoints(void);
 
 /*
@@ -256,10 +255,8 @@ static void blk_trace_cleanup(struct blk_trace *bt)
 	free_percpu(bt->sequence);
 	free_percpu(bt->msg_data);
 	kfree(bt);
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_dec_and_test(&blk_probes_ref))
 		blk_unregister_tracepoints();
-	mutex_unlock(&blk_probe_mutex);
 }
 
 int blk_trace_remove(struct request_queue *q)
@@ -471,13 +468,8 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	bt->pid = buts->pid;
 	bt->trace_state = Blktrace_setup;
 
-	mutex_lock(&blk_probe_mutex);
-	if (atomic_add_return(1, &blk_probes_ref) == 1) {
-		ret = blk_register_tracepoints();
-		if (ret)
-			goto probe_err;
-	}
-	mutex_unlock(&blk_probe_mutex);
+	if (atomic_add_return(1, &blk_probes_ref) == 1)
+		blk_register_tracepoints();
 
 	ret = -EBUSY;
 	old_bt = xchg(&q->blk_trace, bt);
@@ -487,9 +479,6 @@ int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	}
 
 	return 0;
-probe_err:
-	atomic_dec(&blk_probes_ref);
-	mutex_unlock(&blk_probe_mutex);
 err:
 	if (bt) {
 		if (bt->msg_file)
@@ -863,7 +852,7 @@ void blk_add_driver_data(struct request_queue *q,
 }
 EXPORT_SYMBOL_GPL(blk_add_driver_data);
 
-static int blk_register_tracepoints(void)
+static void blk_register_tracepoints(void)
 {
 	int ret;
 
@@ -901,7 +890,6 @@ static int blk_register_tracepoints(void)
 	WARN_ON(ret);
 	ret = register_trace_block_remap(blk_add_trace_remap);
 	WARN_ON(ret);
-	return 0;
 }
 
 static void blk_unregister_tracepoints(void)
@@ -1099,11 +1087,8 @@ static void blk_tracer_print_header(struct seq_file *m)
 
 static void blk_tracer_start(struct trace_array *tr)
 {
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_add_return(1, &blk_probes_ref) == 1)
-		if (blk_register_tracepoints())
-			atomic_dec(&blk_probes_ref);
-	mutex_unlock(&blk_probe_mutex);
+		blk_register_tracepoints();
 	trace_flags &= ~TRACE_ITER_CONTEXT_INFO;
 }
 
@@ -1118,10 +1103,8 @@ static int blk_tracer_init(struct trace_array *tr)
 static void blk_tracer_stop(struct trace_array *tr)
 {
 	trace_flags |= TRACE_ITER_CONTEXT_INFO;
-	mutex_lock(&blk_probe_mutex);
 	if (atomic_dec_and_test(&blk_probes_ref))
 		blk_unregister_tracepoints();
-	mutex_unlock(&blk_probe_mutex);
 }
 
 static void blk_tracer_reset(struct trace_array *tr)
author
#25 lore raw

[tip:tracing/blktrace] blktrace: report EBUSY correctly

In reply to Li Zefan
Commit-ID:  15152e448b693fa41de40f1e40ffbe717a3aab88
Gitweb:     http://git.kernel.org/tip/15152e448b693fa41de40f1e40ffbe717a3aab88
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 09:49:08 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Sat, 21 Mar 2009 16:16:54 +0100

blktrace: report EBUSY correctly

blk_trace_remove_queue() returns EINVAL if q->blk_trace == NULL, but blk_trace_setup_queue() doesn't return EBUSY if q->blk_trace != NULL.

 # echo 0 > sdaX/trace/enable
 # echo 0 > sdaX/trace/enable
 bash: echo: write error: Invalid argument
 # echo 1 > sdaX/trace/enable
 # echo 1 > sdaX/trace/enable
 (should return EBUSY)

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |    9 +++------
 1 files changed, 3 insertions(+), 6 deletions(-)
kernel/trace/blktrace.c +3 -6
index 11e7c8d..14986af 100644
@@ -1260,12 +1260,10 @@ static int blk_trace_remove_queue(struct request_queue *q)
 static int blk_trace_setup_queue(struct request_queue *q, dev_t dev)
 {
 	struct blk_trace *old_bt, *bt = NULL;
-	int ret;
 
-	ret = -ENOMEM;
 	bt = kzalloc(sizeof(*bt), GFP_KERNEL);
 	if (!bt)
-		goto err;
+		return -ENOMEM;
 
 	bt->dev = dev;
 	bt->act_mask = (u16)-1;
@@ -1276,11 +1274,10 @@ static int blk_trace_setup_queue(struct request_queue *q, dev_t dev)
 	if (old_bt != NULL) {
 		(void)xchg(&q->blk_trace, old_bt);
 		kfree(bt);
-		ret = -EBUSY;
+		return -EBUSY;
 	}
+
 	return 0;
-err:
-	return ret;
 }
 
 /*
author
#34 lore raw

[tip:tracing/blktrace] blktrace: avoid accessing NULL bdev->bd_disk

In reply to Li Zefan
Commit-ID:  b125130b22d67f249beba10b71a254558b5279d0
Gitweb:     http://git.kernel.org/tip/b125130b22d67f249beba10b71a254558b5279d0
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 10:34:00 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Sat, 21 Mar 2009 16:17:24 +0100

blktrace: avoid accessing NULL bdev->bd_disk

bdev->bd_disk can be NULL, if the block device is not opened.

Try this against an unmounted partition, and you'll see NULL dereference:

  # echo 1 > /sys/block/sda/sda5/enable

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |   13 +++++++++++--
 1 files changed, 11 insertions(+), 2 deletions(-)
kernel/trace/blktrace.c +11 -2
index dfee6f9..108f4f7 100644
@@ -1362,6 +1362,14 @@ static int blk_str2act_mask(const char *str)
 	return mask;
 }
 
+static struct request_queue *blk_trace_get_queue(struct block_device *bdev)
+{
+	if (bdev->bd_disk == NULL)
+		return NULL;
+
+	return bdev_get_queue(bdev);
+}
+
 static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 					 struct device_attribute *attr,
 					 char *buf)
@@ -1376,9 +1384,10 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 	if (bdev == NULL)
 		goto out_unlock_kernel;
 
-	q = bdev_get_queue(bdev);
+	q = blk_trace_get_queue(bdev);
 	if (q == NULL)
 		goto out_bdput;
+
 	mutex_lock(&bdev->bd_mutex);
 
 	if (attr == &dev_attr_enable) {
@@ -1435,7 +1444,7 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 	if (bdev == NULL)
 		goto out_unlock_kernel;
 
-	q = bdev_get_queue(bdev);
+	q = blk_trace_get_queue(bdev);
 	if (q == NULL)
 		goto out_bdput;
 
author
#29 lore raw

[tip:tracing/blktrace] blktrace: remove sysfs_blk_trace_enable_show/store()

In reply to Li Zefan
Commit-ID:  cd649b8bb830d65c57c3c8b98d57b5402256d8bd
Gitweb:     http://git.kernel.org/tip/cd649b8bb830d65c57c3c8b98d57b5402256d8bd
Author:     Li Zefan <[email protected]>
AuthorDate: Fri, 20 Mar 2009 11:33:55 +0800
Committer:  Ingo Molnar <[email protected]>
CommitDate: Sat, 21 Mar 2009 16:17:08 +0100

blktrace: remove sysfs_blk_trace_enable_show/store()

sysfs_blk_trace_enable_show()/store() share most of code with sysfs_blk_trace_attr_show()/store().

LKML-Reference: <[email protected]>

---
 kernel/trace/blktrace.c |   92 +++++++++++-----------------------------------
 1 files changed, 22 insertions(+), 70 deletions(-)
kernel/trace/blktrace.c +22 -70
index 14986af..dfee6f9 100644
@@ -1284,72 +1284,6 @@ static int blk_trace_setup_queue(struct request_queue *q, dev_t dev)
  * sysfs interface to enable and configure tracing
  */
 
-static ssize_t sysfs_blk_trace_enable_show(struct device *dev,
-					   struct device_attribute *attr,
-					   char *buf)
-{
-	struct hd_struct *p = dev_to_part(dev);
-	struct block_device *bdev;
-	ssize_t ret = -ENXIO;
-
-	lock_kernel();
-	bdev = bdget(part_devt(p));
-	if (bdev != NULL) {
-		struct request_queue *q = bdev_get_queue(bdev);
-
-		if (q != NULL) {
-			mutex_lock(&bdev->bd_mutex);
-			ret = sprintf(buf, "%u\n", !!q->blk_trace);
-			mutex_unlock(&bdev->bd_mutex);
-		}
-
-		bdput(bdev);
-	}
-
-	unlock_kernel();
-	return ret;
-}
-
-static ssize_t sysfs_blk_trace_enable_store(struct device *dev,
-					    struct device_attribute *attr,
-					    const char *buf, size_t count)
-{
-	struct block_device *bdev;
-	struct request_queue *q;
-	struct hd_struct *p;
-	int value;
-	ssize_t ret = -ENXIO;
-
-	if (count == 0 || sscanf(buf, "%d", &value) != 1)
-		goto out;
-
-	lock_kernel();
-	p = dev_to_part(dev);
-	bdev = bdget(part_devt(p));
-	if (bdev == NULL)
-		goto out_unlock_kernel;
-
-	q = bdev_get_queue(bdev);
-	if (q == NULL)
-		goto out_bdput;
-
-	mutex_lock(&bdev->bd_mutex);
-	if (value)
-		ret = blk_trace_setup_queue(q, bdev->bd_dev);
-	else
-		ret = blk_trace_remove_queue(q);
-	mutex_unlock(&bdev->bd_mutex);
-
-	if (ret == 0)
-		ret = count;
-out_bdput:
-	bdput(bdev);
-out_unlock_kernel:
-	unlock_kernel();
-out:
-	return ret;
-}
-
 static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 					 struct device_attribute *attr,
 					 char *buf);
@@ -1361,8 +1295,7 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 		    sysfs_blk_trace_attr_show, \
 		    sysfs_blk_trace_attr_store)
 
-static DEVICE_ATTR(enable, S_IRUGO | S_IWUSR,
-		   sysfs_blk_trace_enable_show, sysfs_blk_trace_enable_store);
+static BLK_TRACE_DEVICE_ATTR(enable);
 static BLK_TRACE_DEVICE_ATTR(act_mask);
 static BLK_TRACE_DEVICE_ATTR(pid);
 static BLK_TRACE_DEVICE_ATTR(start_lba);
@@ -1447,6 +1380,12 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 	if (q == NULL)
 		goto out_bdput;
 	mutex_lock(&bdev->bd_mutex);
+
+	if (attr == &dev_attr_enable) {
+		ret = sprintf(buf, "%u\n", !!q->blk_trace);
+		goto out_unlock_bdev;
+	}
+
 	if (q->blk_trace == NULL)
 		ret = sprintf(buf, "disabled\n");
 	else if (attr == &dev_attr_act_mask)
@@ -1457,6 +1396,8 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
 		ret = sprintf(buf, "%llu\n", q->blk_trace->start_lba);
 	else if (attr == &dev_attr_end_lba)
 		ret = sprintf(buf, "%llu\n", q->blk_trace->end_lba);
+
+out_unlock_bdev:
 	mutex_unlock(&bdev->bd_mutex);
 out_bdput:
 	bdput(bdev);
@@ -1499,6 +1440,15 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 		goto out_bdput;
 
 	mutex_lock(&bdev->bd_mutex);
+
+	if (attr == &dev_attr_enable) {
+		if (value)
+			ret = blk_trace_setup_queue(q, bdev->bd_dev);
+		else
+			ret = blk_trace_remove_queue(q);
+		goto out_unlock_bdev;
+	}
+
 	ret = 0;
 	if (q->blk_trace == NULL)
 		ret = blk_trace_setup_queue(q, bdev->bd_dev);
@@ -1512,13 +1462,15 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
 			q->blk_trace->start_lba = value;
 		else if (attr == &dev_attr_end_lba)
 			q->blk_trace->end_lba = value;
-		ret = count;
 	}
+
+out_unlock_bdev:
 	mutex_unlock(&bdev->bd_mutex);
 out_bdput:
 	bdput(bdev);
 out_unlock_kernel:
 	unlock_kernel();
 out:
-	return ret;
+	return ret ? ret : count;
 }
+
author
#14 lore raw

Re: [PATCH 3/7] blktrace: remove blk_probe_mutex

In reply to Arnaldo Carvalho de Melo
Quoted 4 lines · Arnaldo Carvalho de Melo

Em Fri, Mar 20, 2009 at 09:48:26AM +0800, Li Zefan escreveu:

Quoted 3 lines

blk_register_tracepoints() always returns 0, so make it return void, thus we don't need to use blk_probe_mutex to protect blk_probes_ref.

Quoted 8 lines
Quoted 1 line

Historic reasons, when I first worked on this I found all those WARN_ONs in blk_register_tracepoints a bogosity that should be later fixed, but to reduce the patch size I never got around to submit the proper patch, so I don't think that the fix is what you suggests, here it is what I had before I removed the non-essential parts of the patch:

I though register_trace_xxx() will fail only when there is a name collision, in this case WARN_ON() and return 0 should be reasonable. But I just dig into the code and found it may allocate memory and thus might return ENOMEM, so I guess it's better to check the return value, though it will hardly happen in real-life..

Quoted 90 lines
(quoted diff) +84 -0
+static void blk_tracer_error(const char *name)
+{
+	pr_info("blk trace: Couldn't activate tracepoint "
+		"probe to block_%s\n", name);
+}
+
+#define register_trace_block(tpoint)				     \
+	ret = register_trace_block_##tpoint(blk_add_trace_##tpoint); \
+	if (ret) { 						     \
+		blk_tracer_error(#tpoint);			     \
+		goto *exit_point;		  		     \
+	} else							     \
+		exit_point = &&fail_deprobe_##tpoint;
+	
+
+#define fail_trace_block(tpoint)	\
+	fail_deprobe_##tpoint:		\
+		unregister_trace_block_##tpoint(blk_add_trace_##tpoint)
+
+static int tracing_blk_register(void)
+{
+	int ret;
+	void *exit_point = &&error;
+
+	register_trace_block(rq_abort);
+	register_trace_block(rq_insert);
+	register_trace_block(rq_issue);
+	register_trace_block(rq_requeue);
+	register_trace_block(rq_complete);
+	register_trace_block(bio_bounce);
+	register_trace_block(bio_backmerge);
+	register_trace_block(bio_frontmerge);
+	register_trace_block(bio_queue);
+	register_trace_block(getrq);
+	register_trace_block(sleeprq);
+	register_trace_block(plug);
+	register_trace_block(unplug_timer);
+	register_trace_block(unplug_io);
+	register_trace_block(split);
+	register_trace_block(remap);
+
+	return 0;
+
+	fail_trace_block(remap);
+	fail_trace_block(split);
+	fail_trace_block(unplug_io);
+	fail_trace_block(unplug_timer);
+	fail_trace_block(plug);
+	fail_trace_block(sleeprq);
+	fail_trace_block(getrq);
+	fail_trace_block(bio_queue);
+	fail_trace_block(bio_frontmerge);
+	fail_trace_block(bio_backmerge);
+	fail_trace_block(bio_bounce);
+	fail_trace_block(rq_complete);
+	fail_trace_block(rq_requeue);
+	fail_trace_block(rq_issue);
+	fail_trace_block(rq_insert);
+	fail_trace_block(rq_abort);
+error:
+	return ret;
+}
+
+static void tracing_blk_unregister(void)
+{
+	unregister_trace_block_remap(blk_add_trace_remap);
+	unregister_trace_block_split(blk_add_trace_split);
+	unregister_trace_block_unplug_io(blk_add_trace_unplug_io);
+	unregister_trace_block_unplug_timer(blk_add_trace_unplug_timer);
+	unregister_trace_block_plug(blk_add_trace_plug);
+	unregister_trace_block_sleeprq(blk_add_trace_sleeprq);
+	unregister_trace_block_getrq(blk_add_trace_getrq);
+	unregister_trace_block_bio_queue(blk_add_trace_bio_queue);
+	unregister_trace_block_bio_frontmerge(blk_add_trace_bio_frontmerge);
+	unregister_trace_block_bio_backmerge(blk_add_trace_bio_backmerge);
+	unregister_trace_block_bio_complete(blk_add_trace_bio_complete);
+	unregister_trace_block_bio_bounce(blk_add_trace_bio_bounce);
+	unregister_trace_block_rq_complete(blk_add_trace_rq_complete);
+	unregister_trace_block_rq_requeue(blk_add_trace_rq_requeue);
+	unregister_trace_block_rq_issue(blk_add_trace_rq_issue);
+	unregister_trace_block_rq_insert(blk_add_trace_rq_insert);
+	unregister_trace_block_rq_abort(blk_add_trace_rq_abort);
+	tracepoint_synchronize_unregister();
+}

Regards,

- Arnaldo

#15 lore raw

Re: [PATCH 3/7] blktrace: remove blk_probe_mutex

Em Sun, Mar 22, 2009 at 02:04:14PM +0800, Li Zefan escreveu:

Quoted 14 lines
Quoted 4 lines · Arnaldo Carvalho de Melo

Em Fri, Mar 20, 2009 at 09:48:26AM +0800, Li Zefan escreveu:

Quoted 3 lines

blk_register_tracepoints() always returns 0, so make it return void, thus we don't need to use blk_probe_mutex to protect blk_probes_ref.

Quoted 8 lines
Quoted 1 line

Historic reasons, when I first worked on this I found all those WARN_ONs in blk_register_tracepoints a bogosity that should be later fixed, but to reduce the patch size I never got around to submit the proper patch, so I don't think that the fix is what you suggests, here it is what I had before I removed the non-essential parts of the patch:

Quoted 5 lines

I though register_trace_xxx() will fail only when there is a name collision, in this case WARN_ON() and return 0 should be reasonable. But I just dig into the code and found it may allocate memory and thus might return ENOMEM, so I guess it's better to check the return value, though it will hardly happen in real-life..

Well, I disagree with that, if it can fail, it will fail someday, if failing is such a disastrous thing, register_trace_xxx() should return void and do the WARN_ON directly :-)

- Arnaldo

Quoted 90 lines
Quoted 90 lines
(quoted diff) +84 -0
+static void blk_tracer_error(const char *name)
+{
+	pr_info("blk trace: Couldn't activate tracepoint "
+		"probe to block_%s\n", name);
+}
+
+#define register_trace_block(tpoint)				     \
+	ret = register_trace_block_##tpoint(blk_add_trace_##tpoint); \
+	if (ret) { 						     \
+		blk_tracer_error(#tpoint);			     \
+		goto *exit_point;		  		     \
+	} else							     \
+		exit_point = &&fail_deprobe_##tpoint;
+	
+
+#define fail_trace_block(tpoint)	\
+	fail_deprobe_##tpoint:		\
+		unregister_trace_block_##tpoint(blk_add_trace_##tpoint)
+
+static int tracing_blk_register(void)
+{
+	int ret;
+	void *exit_point = &&error;
+
+	register_trace_block(rq_abort);
+	register_trace_block(rq_insert);
+	register_trace_block(rq_issue);
+	register_trace_block(rq_requeue);
+	register_trace_block(rq_complete);
+	register_trace_block(bio_bounce);
+	register_trace_block(bio_backmerge);
+	register_trace_block(bio_frontmerge);
+	register_trace_block(bio_queue);
+	register_trace_block(getrq);
+	register_trace_block(sleeprq);
+	register_trace_block(plug);
+	register_trace_block(unplug_timer);
+	register_trace_block(unplug_io);
+	register_trace_block(split);
+	register_trace_block(remap);
+
+	return 0;
+
+	fail_trace_block(remap);
+	fail_trace_block(split);
+	fail_trace_block(unplug_io);
+	fail_trace_block(unplug_timer);
+	fail_trace_block(plug);
+	fail_trace_block(sleeprq);
+	fail_trace_block(getrq);
+	fail_trace_block(bio_queue);
+	fail_trace_block(bio_frontmerge);
+	fail_trace_block(bio_backmerge);
+	fail_trace_block(bio_bounce);
+	fail_trace_block(rq_complete);
+	fail_trace_block(rq_requeue);
+	fail_trace_block(rq_issue);
+	fail_trace_block(rq_insert);
+	fail_trace_block(rq_abort);
+error:
+	return ret;
+}
+
+static void tracing_blk_unregister(void)
+{
+	unregister_trace_block_remap(blk_add_trace_remap);
+	unregister_trace_block_split(blk_add_trace_split);
+	unregister_trace_block_unplug_io(blk_add_trace_unplug_io);
+	unregister_trace_block_unplug_timer(blk_add_trace_unplug_timer);
+	unregister_trace_block_plug(blk_add_trace_plug);
+	unregister_trace_block_sleeprq(blk_add_trace_sleeprq);
+	unregister_trace_block_getrq(blk_add_trace_getrq);
+	unregister_trace_block_bio_queue(blk_add_trace_bio_queue);
+	unregister_trace_block_bio_frontmerge(blk_add_trace_bio_frontmerge);
+	unregister_trace_block_bio_backmerge(blk_add_trace_bio_backmerge);
+	unregister_trace_block_bio_complete(blk_add_trace_bio_complete);
+	unregister_trace_block_bio_bounce(blk_add_trace_bio_bounce);
+	unregister_trace_block_rq_complete(blk_add_trace_rq_complete);
+	unregister_trace_block_rq_requeue(blk_add_trace_rq_requeue);
+	unregister_trace_block_rq_issue(blk_add_trace_rq_issue);
+	unregister_trace_block_rq_insert(blk_add_trace_rq_insert);
+	unregister_trace_block_rq_abort(blk_add_trace_rq_abort);
+	tracepoint_synchronize_unregister();
+}

Regards,

- Arnaldo

maintainer Acked-by
#38 lore raw
In reply to Ingo Molnar
Quoted 25 lines · Ingo Molnar
Quoted 22 lines · Ingo Molnar
Quoted 19 lines · Li Zefan

While trying to use blktrace in -tip tree, I encounted kernel NULL dereference, so I looked into the code, and then found some other bugs.

This patchset is part 1, and I have some other pending fixes.

Each patch is small and straightforward, and should be easy to review:

[PATCH 1/7] blktrace: fix possible memory leak
[PATCH 2/7] blktrace: make blk_tracer_enabled a bool flag
[PATCH 3/7] blktrace: remove blk_probe_mutex
[PATCH 4/7] blktrace: don't increase blk_probes_ref if failed to setup blk trace [PATCH 5/7] blktrace: report EBUSY correctly
[PATCH 6/7] blktrace: remove sysfs_blk_trace_enable_show/store() [PATCH 7/7] blktrace: avoid accessing NULL bdev->bd_disk

---
 blktrace.c |  154 +++++++++++++++++--------------------------------------------
 1 file changed, 45 insertions(+), 109 deletions(-)
Quoted 6 lines
Quoted 3 lines
Quoted 1 line

Nice fixes. Applied to tip:tracing/blktrace, thanks!

Which would go upstream in the .30 merge window (unless Jens or Steve objects of course).

I didn't see anything wrong with these patches. (sorry for the late reply, I'm still going through emails).

-- Steve