Skip to contentSkip to conversation

fs: sysfs: don't pass count == 0 to bin file readers

PATCH In review Vladimir Zapolskiy 2 participants 6 messages May 21, 2015 – May 25, 2015 Driver core

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 fs: sysfs: don't pass count == 0 to bin file readers

If count == 0 bytes are requested by a reader, sysfs_kf_bin_read() deliberately returns 0 without passing a potentially harmful value to some externally defined underlying battr->read() function.

However in case of (pos == size && count) the next clause always sets count to 0 and this value is handed over to battr->read().

The change intends to make obsolete (and remove later) a redundant sanity check in battr->read(), if it is present, or add more protection to struct bin_attribute users, who does not care about input arguments.

---
 fs/sysfs/file.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/sysfs/file.c +1 -1
index 7c2867b..6c95628 100644
@@ -90,7 +90,7 @@ static ssize_t sysfs_kf_bin_read(struct kernfs_open_file *of, char *buf,
 		return 0;
 
 	if (size) {
-		if (pos > size)
+		if (pos >= size)
 			return 0;
 		if (pos + count > size)
 			count = size - pos;
Signature · 1 line
1.7.10.4
#1 lore raw

Hello,

Quoted 22 lines · Vladimir Zapolskiy

If count == 0 bytes are requested by a reader, sysfs_kf_bin_read() deliberately returns 0 without passing a potentially harmful value to some externally defined underlying battr->read() function.

However in case of (pos == size && count) the next clause always sets count to 0 and this value is handed over to battr->read().

The change intends to make obsolete (and remove later) a redundant sanity check in battr->read(), if it is present, or add more protection to struct bin_attribute users, who does not care about input arguments.

---
 fs/sysfs/file.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/sysfs/file.c +0 -0
index 7c2867b..6c95628 100644

fs/sysfs/file.c static ssize_t sysfs_kf_bin_read(struct kernfs_open_file *of, char *buf, · line 96

Quoted 8 lines
fs/sysfs/file.c +1 -1
@@ -90,7 +90,7 @@ static ssize_t sysfs_kf_bin_read(struct kernfs_open_file *of, char *buf,
 		return 0;
 
 	if (size) {
-		if (pos > size)
+		if (pos >= size)
 			return 0;
 		if (pos + count > size)
 			count = size - pos;

Hmmm... maybe just move that test upwards?

	if (!count || pos >= size)
		return 0;

	count = min(count, size - pos);
Signature · 1 line
tejun
#2 lore raw

Hello Tejun,

Quoted 36 lines · Tejun Heo

Hello,

Quoted 22 lines · Vladimir Zapolskiy

If count == 0 bytes are requested by a reader, sysfs_kf_bin_read() deliberately returns 0 without passing a potentially harmful value to some externally defined underlying battr->read() function.

However in case of (pos == size && count) the next clause always sets count to 0 and this value is handed over to battr->read().

The change intends to make obsolete (and remove later) a redundant sanity check in battr->read(), if it is present, or add more protection to struct bin_attribute users, who does not care about input arguments.

---
 fs/sysfs/file.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/sysfs/file.c +0 -0
index 7c2867b..6c95628 100644

fs/sysfs/file.c static ssize_t sysfs_kf_bin_read(struct kernfs_open_file *of, char *buf, · line 96

Quoted 8 lines
fs/sysfs/file.c +1 -1
@@ -90,7 +90,7 @@ static ssize_t sysfs_kf_bin_read(struct kernfs_open_file *of, char *buf,
 		return 0;
 
 	if (size) {
-		if (pos > size)
+		if (pos >= size)
 			return 0;
 		if (pos + count > size)
 			count = size - pos;

Hmmm... maybe just move that test upwards?

Quoted 5 lines
	if (!count || pos >= size)
		return 0;

	count = min(count, size - pos);

If the code block stays within if (size && count) { ... }, then !count check is redundant (you may notice that !count check is already present above but not shown in diff's 3 lines context), and I agree that

	if (pos >= size)
		return 0;

	if (pos + count > size)
		count = size - pos;

and

	if (pos >= size)
		return 0;

	count = min(count, size - pos);

are equal.

But "!size" is a special case,

	if (!count || pos >= size)
		return 0;

seems to be incorrect in case of !size ===> (pos >= size) == true.

To the sent change I may add a replacement of "if (pos + count > size) ..." with min_t (ssize_t, count, size - pos), if you wish.

--
With best wishes,
Vladimir

#3 lore raw

Hello, Vladimir.

Quoted 6 lines · Vladimir Zapolskiy

But "!size" is a special case,

	if (!count || pos >= size)
		return 0;

seems to be incorrect in case of !size ===> (pos >= size) == true.

Hmmm... Why is that wrong tho? If size is zero and pos is zero, there's nothing to do, no?

Thanks.

Signature · 1 line
tejun
#4 lore raw

Hello Tejun,

Quoted 5 lines · Tejun Heo

Hello, Vladimir.

Quoted 2 lines · Vladimir Zapolskiy

But "!size" is a special case,

Quoted 7 lines
Quoted 4 lines
	if (!count || pos >= size)
		return 0;

seems to be incorrect in case of !size ===> (pos >= size) == true.

Hmmm... Why is that wrong tho? If size is zero and pos is zero, there's nothing to do, no?

positive size value in the context means the fixed exact length of the file and if size == 0, then it represents some undefined size, often dynamic in runtime. So, if size is zero and pos is zero it stands for reading from the beginning of the file as many bytes as allowed by battr->read() realization. This special case is utilized by quite many bin_attribute users, probably more than half of them set .size to 0.

Quoted 2 lines

Thanks.

--
With best wishes,
Vladimir

Acked-by
#5 lore raw
Quoted 16 lines · Vladimir Zapolskiy

Hello Tejun,

Quoted 5 lines · Tejun Heo

Hello, Vladimir.

Quoted 2 lines · Vladimir Zapolskiy

But "!size" is a special case,

Quoted 7 lines
Quoted 4 lines
	if (!count || pos >= size)
		return 0;

seems to be incorrect in case of !size ===> (pos >= size) == true.

Hmmm... Why is that wrong tho? If size is zero and pos is zero, there's nothing to do, no?

Quoted 6 lines

positive size value in the context means the fixed exact length of the file and if size == 0, then it represents some undefined size, often dynamic in runtime. So, if size is zero and pos is zero it stands for reading from the beginning of the file as many bytes as allowed by battr->read() realization. This special case is utilized by quite many bin_attribute users, probably more than half of them set .size to 0.

Ah, right, I completely forgot about that. Please feel free to add

to the original patch.

Thanks for the patience.

Signature · 1 line
tejun