Re: [PATCH RFC net-next 03/14] bpf: introduce syscall(BPF, ...) and BPF maps

On Fri, Jul 4, 2014 at 8:17 AM, Andy Lutomirski [off-list ref] wrote:

On Wed, Jul 2, 2014 at 7:29 PM, Alexei Starovoitov [off-list ref] wrote:

quoted

non-root API:

ufd = bpf_create_map(local_map_id,… )
bpf_map_update/delete/lookup_elem(ufd,…)
ufd = bpf_prog_load(insns)
close(ufd)

root only API:

global_id = bpf_get_id(ufd) // returns either map or prog global id
bpf_map_delete(global_map_id)
bpf_prog_unload(global_prog_id)

Details:

ufd = bpf_create_map(local_map_id, ...);

local_map_id - process local map_id
(this id is used to access maps from eBPF program loaded by this process)

ufd - process local file descriptor
(used to update/lookup maps from this process)

global_map_id = bpf_get_id(ufd)
this is root only call to get global_ids and pass them to global events
like tracing.

global ids will only be seen by root. There is no way for root or non-root
to influence id ranges.

I think it will be cleaner once I finish fd conversion as a patch.

OK

FWIW, per-process local id maps sound almost equivalent to relocations
-- the latter could be as simple as an extra nlattr giving a list of
pairs of (per-eBPF-program id, fd).

I thought about array of such pairs as well, but it felt cleaner to remember
the (per-eBPF-program id, fd) pair in a kernel, instead of asking user
space to keep track of them. Either way I think it's minor. I'll implement
both and see which way is cleaner.

So far I did 3-way enum split in verifier and addressed Namhyung's
and Chema's comments. Updated in the same place:
git://git.kernel.org/pub/scm/linux/kernel/git/ast/bpf master

I'll be traveling next week.
Once I'm done with fd-style interface, I'll post a v2.

Thanks