> On Feb 23, 2021, at 3:06 AM, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote: > > On Mon, Feb 22, 2021 at 05:20:10PM -0800, Song Liu wrote: >> BPF helpers bpf_task_storage_[get|delete] could hold two locks: >> bpf_local_storage_map_bucket->lock and bpf_local_storage->lock. Calling >> these helpers from fentry/fexit programs on functions in bpf_*_storage.c >> may cause deadlock on either locks. >> >> Prevent such deadlock with a per cpu counter, bpf_task_storage_busy, which >> is similar to bpf_prog_active. We need this counter to be global, because > > So bpf_prog_active is one of the biggest turds around, and now you're > making it worse ?! bpf_prog_active is a distraction here. We are trying to enable task local storage for fentry/fext programs, which do not use bpf_prog_active. bpf_task_storage_busy counter is introduced to protect against a specific pattern of deadlocks (attaching fentry/fexit on bpf_task_storage_[get|delete] helpers, then let the programs call these two helpers again). Thanks, Song
