Following regression was introduced in commit 5e75160f48 ("blobmsg:
fix attrs iteration in the blobmsg_check_array_len()"):
Thread 1 "test-fuzz" received signal SIGSEGV, Segmentation fault.
in blob_len (attr=0x6020000100d4) at libubox/blob.h:102
102 return (be32_to_cpu(attr->id_len) & BLOB_ATTR_LEN_MASK) - sizeof(struct blob_attr);
blob_len (attr=0x6020000100d4) at /libubox/blob.h:102
blob_raw_len (attr=0x6020000100d4) at /libubox/blob.h:111
blob_pad_len (attr=0x6020000100d4) at /libubox/blob.h:120
blobmsg_check_array_len (attr=0x6020000000d0, type=0, blob_len=10) at /libubox/blobmsg.c:145
fuzz_blobmsg_parse (data=0x6020000000d0 "\001\004", size=10) at /libubox/tests/fuzz/test-fuzz.c:57
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Fix out of bounds read in blobmsg_parse and blobmsg_check_name. The
out of bounds read happens because blob_attr and blobmsg_hdr have
flexible array members, whose size is 0 in the corresponding sizeofs.
For example the __blob_for_each_attr macro checks whether rem >=
sizeof(struct blob_attr). However, what LibFuzzer discovered was,
if the input data was only 4 bytes, the data would be casted to blob_attr,
and later on blob_data(attr) would be called even though attr->data was empty.
The same issue could appear with data larger than 4 bytes, where data
wasn't empty, but contained only the start of the blobmsg_hdr struct,
and blobmsg_hdr name was empty. The bugs were discovered by fuzzing
blobmsg_parse and blobmsg_array_parse with LibFuzzer.
CC: Luka Perkov <luka.perkov@sartura.hr>
Reviewed-by: Jo-Philipp Wich <jo@mein.io>
Signed-off-by: Juraj Vijtiuk <juraj.vijtiuk@sartura.hr>
[refactored some checks, added fuzz inputs, adjusted unit test results]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
