Skip to contentSkip to conversation

iio: dac: ad3552r-hs: use sysfs_emit_at() in data source avail show

PATCH In review Babanpreet Singh 2 participants 2 messages Jul 18

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 iio: dac: ad3552r-hs: use sysfs_emit_at() in data source avail show

ad3552r_hs_show_data_source_avail() formats the available data source names into a 128-byte stack buffer, but bounds each scnprintf() with PAGE_SIZE instead of the buffer size, so the bound does not protect the destination at all.

This cannot overflow today - dbgfs_attr_source[] has two entries, "normal" and "ramp-16bit", 18 bytes formatted - but the bound stops protecting the stack the day the table grows.

Found by smatch:

drivers/iio/dac/ad3552r-hs.c:593 ad3552r_hs_show_data_source_avail()
error: scnprintf() 'buf[len]' too small (128 vs 4096)

Instead of shrinking the scnprintf() bound to the buffer size, switch to sysfs_emit_at(), which enforces the PAGE_SIZE bound internally and is the preferred helper for formatting attribute-style output; the IIO core formats its own avail lists the same way in iio_format_list(). sysfs_emit_at() refuses buffers that are not page-aligned, so replace the stack buffer with a kzalloc()'d page - power-of-two kmalloc() allocations are naturally aligned, see Documentation/core-api/memory-allocation.rst.

Fixes b1c5d68ea66e ("iio: dac: ad3552r-hs: add support for internal ramp")
Assisted-by claude-sonnet-5

---
v2:
- Switch to sysfs_emit_at() with a kzalloc()'d page instead of bounding scnprintf() with sizeof(buf), as suggested by Andy Shevchenko. sysfs_emit_at() requires a page-aligned buffer, so the 128-byte stack buffer cannot stay; add the cleanup.h, slab.h and sysfs.h includes the new code needs.
- Retitled; v1 was "iio: dac: ad3552r-hs: fix scnprintf() buffer bound in data source show".

v1: https://lore.kernel.org/r/[email protected]

 drivers/iio/dac/ad3552r-hs.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)
drivers/iio/dac/ad3552r-hs.c +10 -6
index 02a124ac4855..c561fed2ef1a 100644
@@ -7,6 +7,7 @@
  */
 
 #include <linux/bitfield.h>
+#include <linux/cleanup.h>
 #include <linux/debugfs.h>
 #include <linux/delay.h>
 #include <linux/gpio/consumer.h>
@@ -14,6 +15,8 @@
 #include <linux/iio/buffer.h>
 #include <linux/platform_device.h>
 #include <linux/property.h>
+#include <linux/slab.h>
+#include <linux/sysfs.h>
 #include <linux/units.h>
 
 #include "ad3552r.h"
@@ -585,14 +588,15 @@ static ssize_t ad3552r_hs_show_data_source_avail(struct file *f,
 						 char __user *userbuf,
 						 size_t count, loff_t *ppos)
 {
-	ssize_t len = 0;
-	char buf[128];
+	char *buf __free(kfree) = kzalloc(PAGE_SIZE, GFP_KERNEL);
+	int len = 0;
 	int i;
 
-	for (i = 0; i < ARRAY_SIZE(dbgfs_attr_source); i++) {
-		len += scnprintf(buf + len, PAGE_SIZE - len, "%s ",
-				 dbgfs_attr_source[i]);
-	}
+	if (!buf)
+		return -ENOMEM;
+
+	for (i = 0; i < ARRAY_SIZE(dbgfs_attr_source); i++)
+		len += sysfs_emit_at(buf, len, "%s ", dbgfs_attr_source[i]);
 	buf[len - 1] = '\n';
 
 	return simple_read_from_buffer(userbuf, count, ppos, buf, len);

base-commit: fce2dfa773ced15f27dd27cd0b482a7473cdcf2a

Signature · 1 line
2.43.0
#1 lore raw
Quoted 9 lines · Babanpreet Singh

ad3552r_hs_show_data_source_avail() formats the available data source names into a 128-byte stack buffer, but bounds each scnprintf() with PAGE_SIZE instead of the buffer size, so the bound does not protect the destination at all.

This cannot overflow today - dbgfs_attr_source[] has two entries, "normal" and "ramp-16bit", 18 bytes formatted - but the bound stops protecting the stack the day the table grows.

Quoted 7 lines

Found by smatch:

drivers/iio/dac/ad3552r-hs.c:593 ad3552r_hs_show_data_source_avail()
error: scnprintf() 'buf[len]' too small (128 vs 4096)

Fixes b1c5d68ea66e ("iio: dac: ad3552r-hs: add support for internal ramp")

This whole thing is not what I suggested, I only had an idea of using sysfs_emit_at(). BUT...

...

Quoted 2 lines

static ssize_t ad3552r_hs_show_data_source_avail(struct file *f,
char __user *userbuf,

...I haven't paid attention that this is not sysfs attribute handling in the default way, this is customized one and the passed buffer is in user space.

Quoted 1 line

size_t count, loff_t *ppos)

Hence the correct fix is just replace PAGE_SIZE with sizeof(buf) as it was I suppose in v1. Sorry for my mistake.

Signature · 2 lines
With Best Regards,
Andy Shevchenko