Re: [PATCH v3 2/3] seccomp_filters: system call filtering using BPF

[Will Drewry <wad@chromium.org>]

On Fri, Jan 13, 2012 at 11:39 AM, Eric Paris <eparis@redhat.com> wrote:
> On Thu, 2012-01-12 at 17:38 -0600, Will Drewry wrote:
>
>> diff --git a/include/linux/seccomp.h b/include/linux/seccomp.h
>> index cc7a4e9..0296871 100644
>> --- a/include/linux/seccomp.h
>> +++ b/include/linux/seccomp.h
>
>> -typedef struct { int mode; } seccomp_t;
>> +struct seccomp_filter;
>> +/**
>> + * struct seccomp_struct - the state of a seccomp'ed process
>> + *
>> + * @mode:
>> + *     if this is 0, seccomp is not in use.
>> + *             is 1, the process is under standard seccomp rules.
>> + *             is 2, the process is only allowed to make system calls where
>> + *                   associated filters evaluate successfully.
>> + * @filter: Metadata for filter if using CONFIG_SECCOMP_FILTER.
>> + *          @filter must only be accessed from the context of current as there
>> + *          is no guard.
>> + */
>> +typedef struct seccomp_struct {
>> +     int mode;
>> +#ifdef CONFIG_SECCOMP_FILTER
>> +     struct seccomp_filter *filter;
>> +#endif
>> +} seccomp_t;
>>
>>  extern void __secure_computing(int);
>>  static inline void secure_computing(int this_syscall)
>
> Can we get rid of all of the typedef stuff?  I know you didn't add it
> but now seems like a good time to follow typical kernel semantics if you
> have to re-rev for some other reason.

Yup - I was hoping to do that separately since it touches extra files.
 I'll make a prereq patch that enacts the change (so it can be picked
up even if the seccomp-bpf is less successful).

cheers!
will