On 8/20/20 2:38 PM, Mat Martineau wrote:
Commit 17f79c855e61 added a variable 'r' to
that holds either a new dhcp6_state or an error (negative value). 'r'
was not set in several client states, but it appears that the unit tests
only cover cases where it is set.
It appears that the new code at the end of dhcp6_client_rx_message() is
only intended for the cases where 'r' was set, so initialize 'r' to the
current client->state and rely on the existing state check to preserve
Ah thanks for this. I still have some pending changes that masked it on my end.
I need to be more careful.
Denis, I don't have enough background knowledge about DHCPv6 to be 100%
sure this patch does the right thing. Is my guess correct about the
intended use of the new code at the end of dhcp6_client_rx_message()?
The fix is fine for now. I think eventually all cases will be handled such that
r is always set, but I'm not there yet.
ell/dhcp6.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)