An off-by-one error in resolv_found() could make an strncat() call
overflow by the terminating null byte.
When building with Clang the following warning was shown:
../../../core/net/ip/resolv.c:1458:17: warning: the value of the
size argument in 'strncat' is too large, might lead to a
buffer overflow [-Wstrncat-size]
sizeof(resolv_hostname) - strlen(resolv_hostname));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../../../core/net/ip/resolv.c:1458:17: note: change the argument to
be the free space in the destination buffer minus the
terminating null byte
sizeof(resolv_hostname) - strlen(resolv_hostname));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
sizeof(resolv_hostname) - strlen(resolv_hostname) - 1
Signed-off-by: Joakim Gebart <joakim.gebart@eistec.se>
The next-hop address did not get updated in the routing table
in case an entry for the destination already existed.
This patch resolves the issue by removing the entry and
having it re-created from scratch.
The issue causes a routing error triggering reconstruction of
the DODAG through version increase.
In case of somewhat frequent downward traffic in not (yet) stabilized DODAG
a vicious circle is formed: unstable topology means churn, downward
routing under churn causes reconstruction of DODAG. In this situation
the network does not have chance to stabilize.
We encountered a constant churn caused by this bug
in a network of 50 nodes and a periodic traffic (a packet every 5
seconds) generated at the root.
More info and a PCAP demonstrating the issue can be found here:
https://github.com/contiki-os/contiki/issues/496
type process_data_t. This was an artifact when the choice was
made to use the void * type for the data parameter in processes.
Changed parameter 'void * data' of process_post_synch to
process_data_t for consistency.
Checked all the uses of process_start() in contiki and fixed casts
of the data parameter.
Currently the SPI driver in core/ sets `SPI_FLUSH()` to be a single read
of the spi RX buffer. This is fine for many microcontrollers, but newer
platforms like the CC2538 have a multi-byte FIFO for the RX buffer. In
that case, a single read is not guaranteed to flush the RX. This commit
allows `SPI_FLUSH()` to be platform dependent if needed, but doesn't
change for platforms where it works.
would create a header with the current packet information.
This allows sicslowpan to calculate the max payload size without
consuming a sequence number or clearing/restoring the packet buffer.
