On Mon, Aug 25, 2014 at 2:33 PM, Cong Wang [off-list ref] wrote:
On Sun, Aug 24, 2014 at 1:21 PM, Alexei Starovoitov [off-list ref] wrote:
quoted
'maps' is a generic storage of different types for sharing data between kernel
and userspace.
The maps are accessed from user space via BPF syscall, which has commands:
- create a map with given type and attributes
fd = bpf_map_create(map_type, struct nlattr *attr, int len)
returns fd or negative error
- lookup key in a given map referenced by fd
err = bpf_map_lookup_elem(int fd, void *key, void *value)
returns zero and stores found elem into value or negative error
- create or update key/value pair in a given map
err = bpf_map_update_elem(int fd, void *key, void *value)
returns zero or negative error
- find and delete element by key in a given map
err = bpf_map_delete_elem(int fd, void *key)
- iterate map elements (based on input key return next_key)
err = bpf_map_get_next_key(int fd, void *key, void *next_key)
I think you need to document the bpf() syscall instead of wrappers on it,
from a developer's point of view. You will anyway need to document a new
syscall with a man page as a general rule.
yep. I've mentioned before that man page is on todo list. I'm delaying
writing it, because it's the most difficult part and I don't want to keep
rewriting it when interface changes (like it did from global id to fd).
Once implementation lands, manpage will be the highest priority.
In the changelog I mean something like:
err = bpf(BPF_MAP_LOOKUP_ELEM, ...);
Are you saying instead of:
err = bpf_map_lookup_elem(int fd, void *key, void *value)
write
err = bpf(BPF_MAP_LOOKUP_ELEM, fd, key, value)
in commit log?
I think that style carries less information per line.
For man page I'll document the syscall in a traditional way, but
for commit log I like to have maximum info in the fewest lines.