Re: [PATCH bpf-next] bpftool: Restore support for BPF offload-enabled feature probing

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Quentin,

On 2022-03-08 21:43:07 +0000, Quentin Monnet wrote:

> So I looked a bit more into it. I couldn't remember the reason why
> we'd probe all maps but not all prog types, but the commit log has a
> hint:
> 
>     Among the program types, only the ones that can be offloaded are probed.
>     All map types are probed, as there is no specific rule telling which one
>     could or could not be supported by a device in the future. All helpers
>     are probed (but only for offload-able program types).
> 
> I think at the time, I was referring, for program types, to this check
> in bpf_prog_offload_init() in kernel/bpf/offload.c.
> 
>     if (attr->prog_type != BPF_PROG_TYPE_SCHED_CLS &&
>         attr->prog_type != BPF_PROG_TYPE_XDP)
>         return -EINVAL;
> 
> This makes it impossible to have other types offloaded; but there's no
> equivalent for maps, so we could theoretically have any map type
> supported. So maybe not such a bad thing, filtering out non-relevant
> program types but allowing out-of-kernel drivers to probe all map
> types. Maybe keep this part unchanged for v2, after all?

Thanks for following up with this information. With this background I 
agree, lets keep it as it was for v2.

-- 
Regards,
Niklas Söderlund



[Index of Archives]     [Linux Samsung SoC]     [Linux Rockchip SoC]     [Linux Actions SoC]     [Linux for Synopsys ARC Processors]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]


  Powered by Linux