[PATCH v2] firewire: cdev: check write quadlet request length toavoid buffer overflow

[Stefan Richter]

Check that the data length of a write quadlet request actually is large
enough for a quadlet.  Otherwise, fw_fill_request could access the four
bytes after the end of the outbound_transaction_event structure.

Reported-by: Clemens Ladisch <clemens@xxxxxxxxxx>

Since struct outbound_transaction_event *e is slab-allocated, such an
access may hit unallocated memory only if sizeof(*e) == 256 or 512 or
any other power of 2 above 2**2.  In kernel 2.6.34, sizeof(*e) is > 128
and < 256 on 32bit architectures, and > 256 and < 512 on 64bit
architectures.  Thus the only problem is that a bogus write quadlet
request with user-specified length of < 3 will put 1...4 random bytes
into the packet payload; but this is the users's problem then, not the
kernel's.

Hence the corner case handling can be optimized by a !(sizeof(*e) & 255)
check as a weaker form of a sizeof(*e) == 256 or 512 check.  This is a
constant expression and will cause the whole check to be omitted by the
compiler's dead code elimination.

Of course this relies on certain behavior of the slab allocator.

Signed-off-by: Stefan Richter <stefanr@xxxxxxxxxxxxxxxxx>
---
 drivers/firewire/core-cdev.c |    5 +++++
 1 file changed, 5 insertions(+)

Index: b/drivers/firewire/core-cdev.c
===================================================================
--- a/drivers/firewire/core-cdev.c
+++ b/drivers/firewire/core-cdev.c
@@ -564,6 +564,11 @@ static int init_request(struct client *c
 	    (request->length > 4096 || request->length > 512 << speed))
 		return -EIO;
 
+	/* Corner case: Access past the end of *e in fw_fill_request() */
+	if (request->tcode == TCODE_WRITE_QUADLET_REQUEST &&
+	    request->length < 4 && (sizeof(*e) & 255) == 0)
+		return -EINVAL;
+
 	e = kmalloc(sizeof(*e) + request->length, GFP_KERNEL);
 	if (e == NULL)
 		return -ENOMEM;

-- 
Stefan Richter
-=====-==-=- -=== --===
http://arcgraph.de/sr/

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/