[CRIU] [PATCH] dump: do not fail dump when robust_lists are disabled

Filipe Brandenburger filbranden at google.com
Wed Jun 25 07:55:28 PDT 2014


Hi Cyrill,

I thought a little more about it and I wonder if we need to consider
the case where get_robust_list is not available but set_robust_list
is... Such a case might happen from syscall filtering (e.g. using
seccomp-bpf for security and filtering out get_robust_list only) or
possible deprecation of the get_robust_list syscall
(http://marc.info/?l=linux-kernel&m=134399406131605&w=2).

Do you think it would be worth adding code here to assume that the
list is empty only when both get_robust_list and sys_robust_list
return ENOSYS? I can send you an updated patch, I just wanted to know
your opinion before I do.

Cheers,
Filipe


On Tue, Jun 24, 2014 at 10:38 PM, Cyrill Gorcunov <gorcunov at gmail.com> wrote:
> On Tue, Jun 24, 2014 at 04:32:55PM -0700, Filipe Brandenburger wrote:
>> Robust lists may be disabled, for example if the "futex_cmpxchg_enabled"
>> variable in the kernel is unset.
>>
>> Detect that case by checking that the "get_robust_list" syscall returns ENOSYS
>> and assume an empty robust_list in that case.
>>
>> Tested: Successfully ran the zdtm test suite on a kernel where the
>> "get_robust_list" syscall is disabled.
> Acked-by: Cyrill Gorcunov <gorcunov at openvz.org>
>
> Thanks Filipe!


More information about the CRIU mailing list