Attention is currently required from: daniel.
10 comments:
File src/osmo_io.c:
Patch Set #3, Line 389: g_backend == OSMO_IO_BACKEND_POLL ? "poll" : "uring"
there might be more in the future, it might be worth having a separate value_string for that rather than explicitly hard-coding the names on the caller side here. Also, it doesn't hurt to log the actaul fd number while we're logging something anyway.
c++ style comment
Patch Set #3, Line 430: osmo_iofd_get_priv
please rename to osmo_iofd_get_priv_nr(). "priv" is also sometimes used as alias to "data" in kernel and osmocom.
If we remove the argument from the setup function, then we also need a set_priv_nr().
File src/osmo_io_internal.h:
I'm wondering if it had any advantage to have only a single function with an argument stating the "operation"? The prototype is almost identical (just int/void return).
I'm not sure it would have any advantage. Probably none.
struct {
bool read_enabled;
bool read_pending;
bool write_pending;
bool write_enabled;
/* TODO: index into array of registered fd's? */
} uring;
should this maybe introduced only in the actual io_oring patch?
File src/osmo_io_poll.c:
maybe still look into this before merging it?
extraneous empty line
coding style
same as above
should be static? same as all of the iofd_poll_ functions which are exposed as callback via iofd_poll_ops and don't need to be exported as global functions
To view, visit change 30934. To unsubscribe, or for help writing mail filters, visit settings.