On Wed, Sep 29, 2021 at 03:02:12PM -0700, Kees Cook wrote: > Hi, > > This attempts to solve the issues from the discussion here[1]. Specifically: > > 1) wchan leaking raw addresses since 152c432b128c (v5.12). > > patch 1 fixes this with a revert. > > 2) wchan has been broken under ORC, seen as a failure to stack walk > resulting in _usually_ a 0 value, since ee9f8fce9964 (v4.14). > > patches 2-5 fixes this with Qi Zheng's new get_wchan() and changes to > the /proc code to use the new helper suggested by Peter to do the stack > walk only if the process can be kept blocked: > https://lore.kernel.org/lkml/20210929194026.GA4323@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/ > > Peter, can you take this via -tip? It all looks sane to me. Thanks for cleaning up this mess. - Should we use a similar sched wrapper for /proc/$pid/stack to make its raciness go away? - At the risk of triggering a much larger patch set, I suspect get_wchan() can be made generic ;-) It's just a glorified wrapper around stack_trace_save_tsk(). Regardless: Acked-by: Josh Poimboeuf <jpoimboe@xxxxxxxxxx> -- Josh
