CVE-2024-50036

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: do not delay dst_entries_add() in dst_release()<br /> <br /> dst_entries_add() uses per-cpu data that might be freed at netns<br /> dismantle from ip6_route_net_exit() calling dst_entries_destroy()<br /> <br /> Before ip6_route_net_exit() can be called, we release all<br /> the dsts associated with this netns, via calls to dst_release(),<br /> which waits an rcu grace period before calling dst_destroy()<br /> <br /> dst_entries_add() use in dst_destroy() is racy, because<br /> dst_entries_destroy() could have been called already.<br /> <br /> Decrementing the number of dsts must happen sooner.<br /> <br /> Notes:<br /> <br /> 1) in CONFIG_XFRM case, dst_destroy() can call<br /> dst_release_immediate(child), this might also cause UAF<br /> if the child does not have DST_NOCOUNT set.<br /> IPSEC maintainers might take a look and see how to address this.<br /> <br /> 2) There is also discussion about removing this count of dst,<br /> which might happen in future kernels.