when using dfsan, i thought i could use the ‘userdata’ field in dfsan_label_info
to maintain input file offsets influencing the labeled variable.
I would label variables as tainted that get input from a file along with the
file offsets at that time (using ‘userdata’).
For dependent variables I would like to use the propagation mechanism of dfsan in
order to see the accumulated input offsets later.
So, when dfsan_union is called, I would like to have a wrapper or callback
function in effect in order to also generate the union of the input file offsets
from userdata of both input labels.
So, do you think the dfsan framework can be used to achieve this?
I am interested in a minimal invasive way to extend the dfsan_union function.
Thanks for any advice and best regards,