Attention is currently required from: osmith, Hoernchen, neels, laforge, pespin, fixeria. daniel has posted comments on this change. ( https://gerrit.osmocom.org/c/libosmocore/+/30934 )
Change subject: Add osmo_io with initial poll backend ......................................................................
Patch Set 4:
(11 comments)
This change is ready for review.
Patchset:
PS3:
In general I think we need to think about naming / namespace pollution. […]
Are you talking about a map file for all of libosmocore or just a libosmoio map file?
File src/osmo_io.c:
https://gerrit.osmocom.org/c/libosmocore/+/30934/comment/8c9b0167_87a37287 PS3, Line 50: struct iofd_backend_ops g_iofd_ops;
then better name it with an osmo_ prefix.
This should be resolved if we end up introducing a .map file, right?
File src/osmo_io.c:
https://gerrit.osmocom.org/c/libosmocore/+/30934/comment/cdabab1f_b2c19701 PS4, Line 228: msg_pending = iofd_msgb_alloc(iofd); This function now sets pending_out to the remaining msgb and iofd_handle_segmented_read is responsible for setting this. This also saves us one msgb_pending call in the handle_more case below.
https://gerrit.osmocom.org/c/libosmocore/+/30934/comment/0a9f73e5_028dca84 PS4, Line 299: * \returns 0 in case of success; a negative value in case of error Even though almost no code uses it I guess you're right.
https://gerrit.osmocom.org/c/libosmocore/+/30934/comment/1115e0e0_632590fc PS4, Line 356: : struct os Done
https://gerrit.osmocom.org/c/libosmocore/+/30934/comment/b7a0d6ae_03597afb PS4, Line 367: iofd->io_ops = *ioops; I agree, the name should be mandatory.
https://gerrit.osmocom.org/c/libosmocore/+/30934/comment/8471d7a4_6109cd5a PS4, Line 414: * \param[in] headroom the headroom of the msgb when receiving data Ack
File src/osmo_io_internal.h:
https://gerrit.osmocom.org/c/libosmocore/+/30934/comment/8e8e3792_75bbe29d PS3, Line 17: in
I'm wondering if it had any advantage to have only a single function with an argument stating the "o […]
I'll keep it as it is
https://gerrit.osmocom.org/c/libosmocore/+/30934/comment/337a22b0_b3a229c7 PS3, Line 72: struct { : bool read_enabled; : bool read_pending; : bool write_pending; : bool write_enabled; : /* TODO: index into array of registered fd's? */ : } uring;
Ack
Done
File src/osmo_io_poll.c:
https://gerrit.osmocom.org/c/libosmocore/+/30934/comment/66e76cf0_7c9f532a PS4, Line 104: struct osmo_sockaddr saddr; Done
https://gerrit.osmocom.org/c/libosmocore/+/30934/comment/f3a3dc66_965873ff PS4, Line 128: if (rc < 0) { Does the kernel use addrlen for more than just checking if it's long enough for the variant? In any case I guess we only want to copy to the kernel what we really need.
I think should be a helper function in libosmocore like osmo_sockaddr_size() that returns the size depending on the family that way we can reuse it elsewhere.
-
daniel