io_uring_queue_init - setup io_uring submission and completion queues
int io_uring_queue_init(unsigned entries, struct io_uring *ring, unsigned flags);
int io_uring_queue_init_params(unsigned entries, struct io_uring *ring, struct io_uring_params *params);
int io_uring_queue_init_mem(unsigned entries, struct io_uring *ring, struct io_uring_params *params, void *buf, size_t buf_size);
The io_uring_queue_init(3) function executes the io_uring_setup(2) system call to initialize the submission and completion queues in the kernel with at least entries entries in the submission queue and then maps the resulting file descriptor to memory shared between the application and the kernel.
By default, the CQ ring will have twice the number of entries as specified by entries for the SQ ring. This is adequate for regular file or storage workloads, but may be too small for networked workloads. The SQ ring entries do not impose a limit on the number of in-flight requests that the ring can support, it merely limits the number that can be submitted to the kernel in one go (batch). if the CQ ring overflows, e.g. more entries are generated than fits in the ring before the application can reap them, then if the kernel supports IORING_FEAT_NODROP the ring enters a CQ ring overflow state. Otherwise it drops the CQEs and increments cq.koverflow in struct io_uring with the number of CQEs dropped. The overflow state is indicated by IORING_SQ_CQ_OVERFLOW being set in the SQ ring flags. Unless the kernel runs out of available memory, entries are not dropped, but it is a much slower completion path and will slow down request processing. For that reason it should be avoided and the CQ ring sized appropriately for the workload. Setting cq_entries in struct io_uring_params will tell the kernel to allocate this many entries for the CQ ring, independent of the SQ ring size in given in entries. If the value isn't a power of 2, it will be rounded up to the nearest power of 2.
On success, io_uring_queue_init(3) returns 0 and ring will point to the shared memory containing the io_uring queues. On failure -errno is returned.
flags will be passed through to the io_uring_setup syscall (see io_uring_setup(2)).
The io_uring_queue_init_mem(3) variant uses the provided buf with associated size buf_size as the memory for the ring, using the IORING_SETUP_NO_MMAP flag to io_uring_setup(2). The buffer passed to io_uring_queue_init_mem(3) must already be zeroed. Typically, the caller should allocate a huge page and pass that in to io_uring_queue_init_mem(3). Pages allocated by mmap are already zeroed. io_uring_queue_init_mem(3) returns the number of bytes used from the provided buffer, so that the app can reuse the buffer with the returned offset to put more rings in the same huge page.
On success, the resources held by ring should be released via a corresponding call to io_uring_queue_exit(3).
io_uring_queue_init_mem(3) returns the number of bytes used from the provided buffer on success, and -errno on failure.
|July 10, 2020