On Thu, Jul 9, 2020 at 12:13 PM Guenter Roeck <li...@roeck-us.net> wrote: > > On 7/9/20 11:51 AM, Cong Wang wrote: > > On Thu, Jul 9, 2020 at 10:10 AM Guenter Roeck <li...@roeck-us.net> wrote: > >> > >> Something seems fishy with the use of skcd->val on big endian systems. > >> > >> Some debug output: > >> > >> [ 22.643703] sock: ##### sk_alloc(sk=000000001be28100): Calling > >> cgroup_sk_alloc(000000001be28550) > >> [ 22.643807] cgroup: ##### cgroup_sk_alloc(skcd=000000001be28550): > >> cgroup_sk_alloc_disabled=0, in_interrupt: 0 > >> [ 22.643886] cgroup: #### cgroup_sk_alloc(skcd=000000001be28550): > >> cset->dfl_cgrp=0000000001224040, skcd->val=0x1224040 > >> [ 22.643957] cgroup: ###### cgroup_bpf_get(cgrp=0000000001224040) > >> [ 22.646451] sock: ##### sk_prot_free(sk=000000001be28100): Calling > >> cgroup_sk_free(000000001be28550) > >> [ 22.646607] cgroup: #### sock_cgroup_ptr(skcd=000000001be28550) -> > >> 0000000000014040 [v=14040, skcd->val=14040] > >> [ 22.646632] cgroup: ####### cgroup_sk_free(): skcd=000000001be28550, > >> cgrp=0000000000014040 > >> [ 22.646739] cgroup: ####### cgroup_sk_free(): skcd->no_refcnt=0 > >> [ 22.646814] cgroup: ####### cgroup_sk_free(): Calling > >> cgroup_bpf_put(cgrp=0000000000014040) > >> [ 22.646886] cgroup: ###### cgroup_bpf_put(cgrp=0000000000014040) > > > > Excellent debugging! I thought it was a double put, but it seems to > > be an endian issue. I didn't realize the bit endian machine actually > > packs bitfields in a big endian way too... > > > > Does the attached patch address this? > > > > Partially. I don't see the crash anymore, but something is still odd - some > of my > tests require a retry with this patch applied, which previously never > happened. > I don't know if this is another problem with this patch, or a different > problem. > Unfortunately, I'll be unable to debug this further until next Tuesday.
Make sure you test the second patch I sent, not the first one. The first one is still incomplete and ugly too. The two bits must be the last two, so correcting the if test is not sufficient, we have to fix the whole bitfield packing. Thanks!
