On Thu Jun 8 16:03:04 2023 +0000, Alex Henrie wrote:
Whoops. I fixed the title, but I will have to come back to this when I have time to look at the other double frees you mentioned (hopefully later today). Thanks for the feedback.
Thanks for spotting the third double free. You're probably right that it makes sense to zero out the pointers so that alloc_dependency can always be paired with free_dependency. I've pushed a new patch that zeroes them out on read_identity's error path, which I think makes a little more sense than zeroing them out in every call to clear_identity.