This code will likely crash if we try to do a zero byte write. The code
looks like this:
/* strip trailing whitespace */
for (i = count - 1; i > 0; i--)
if (isspace(buf[i]))
...
We're writing zero bytes so count = 0. You would think that "count - 1"
would be negative one, but because "i" is unsigned it is a large
positive numer instead. The "i > 0" condition is true and the "buf[i]"
access will be out of bounds.
The fix is to make "i" signed and now everything works as expected. The
upper bound of "count" is capped in __kernel_write() at MAX_RW_COUNT so
we don't have to worry about it being higher than INT_MAX.
Fixes: 02dd95fe3169 ("drm/tinydrm: Add MIPI DBI support")
Signed-off-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
---
drivers/gpu/drm/drm_mipi_dbi.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/drm_mipi_dbi.c b/drivers/gpu/drm/drm_mipi_dbi.c
index 1961f713aaab..c4ee2709a6f3 100644
--- a/drivers/gpu/drm/drm_mipi_dbi.c
+++ b/drivers/gpu/drm/drm_mipi_dbi.c
@@ -1187,8 +1187,7 @@ static ssize_t mipi_dbi_debugfs_command_write(struct file *file,
struct mipi_dbi_dev *dbidev = m->private;
u8 val, cmd = 0, parameters[64];
char *buf, *pos, *token;
- unsigned int i;
- int ret, idx;
+ int i, ret, idx;
if (!drm_dev_enter(&dbidev->drm, &idx))
return -ENODEV;
--
2.20.1
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel
