Re: [PATCH v2 net-next 2/3] x86: bpf_jit_comp: supportBPF_S_ANC_SECCOMP_LD_W instruction

[Xi Wang]

On Sat, Apr 27, 2013 at 9:21 PM, Eric Dumazet <eric.dumazet@xxxxxxxxx> wrote:
> I would feel more comfortable if you added :
>
>         if (seen & SEEN_DATAREF) {
>                 pr_err_once("SECCOMP_LD_W assertion failed\n"):
>                 goto out;
>         }
>
> This way, if BPF is changed in the future, but not the x86 JIT, we
> can have a working kernel.
>
> Ideally, we should add a SEEN_SKBREF to make sure rdi value can be
> scratched, or you just push %rdi/pop %rdi, its only one byte
> instructions.

Adding SEEN_SKBREF sounds like a good idea. :)

> Or completely optimize the thing and not call seccomp_bpf_load() at all.

This would be cool.

> (current would be loaded once in r9, task_pt_regs() would be loaded once
> in r8)

Both syscall_get_arch() and syscall_get_arguments() need to test for
the TS_COMPAT bit (in task_thread_info(current)->status); we should
load that once, too.

Thanks for the comments.  Will try a v3.

- xi
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/