On Tue, Jun 23, 2020 at 08:23 AM CEST, Andrii Nakryiko wrote: > On Mon, Jun 22, 2020 at 9:04 AM Jakub Sitnicki <jakub@xxxxxxxxxxxxxx> wrote: >> >> Prepare for having multi-prog attachments for new netns attach types by >> storing programs to run in a bpf_prog_array, which is well suited for >> iterating over programs and running them in sequence. >> >> Because bpf_prog_array is dynamically resized, after this change a >> potentially blocking memory allocation in bpf(PROG_QUERY) callback can >> happen, in order to collect program IDs before copying the values to >> user-space supplied buffer. This forces us to adapt how we protect access >> to the attached program in the callback. As bpf_prog_array_copy_to_user() >> helper can sleep, we switch from an RCU read lock to holding a mutex that >> serializes updaters. >> >> To handle bpf(PROG_ATTACH) scenario when we are replacing an already >> attached program, we introduce a new bpf_prog_array helper called >> bpf_prog_array_replace_item that will exchange the old program with a new >> one. bpf-cgroup does away with such helper by computing an index into the >> array based on program position in an external list of attached >> programs/links. Such approach seems fragile, however, when dummy progs can >> be left in the array after a memory allocation failure on link release. > > bpf-cgroup can have the same BPF program present multiple times in the > effective prog array due to inheritance. It also has strict > guarantee/requirement about relative order of programs in parent > cgroup vs child cgroups. For such cases, replacing a BPF program based > on its pointer is not going to work correctly. Thanks for the explanation. That did not occur to me. Incorporated it into the description in v2. > > We do need to make sure that cgroup detachment never fails by falling > back to replacing BPF prog with dummy prog, though. If you are > interested in a challenge, you are very welcome to do that! :) I keep a list of tasks for a slow day. [...]
