diff options
author | thegeorg <thegeorg@yandex-team.com> | 2023-07-26 17:26:21 +0300 |
---|---|---|
committer | thegeorg <thegeorg@yandex-team.com> | 2023-07-26 17:26:21 +0300 |
commit | 3785d5f97965bccf048718d8717904cf50f9f8f9 (patch) | |
tree | b7ce8ae67d7eb7fcf7767c54379f0564c281147f /contrib/libs/liburing/src | |
parent | 1f6b57071583f89299bb5abd3863d594f23c5be5 (diff) | |
download | ydb-3785d5f97965bccf048718d8717904cf50f9f8f9.tar.gz |
Update contrib/libs/liburing to 2.4
Diffstat (limited to 'contrib/libs/liburing/src')
-rw-r--r-- | contrib/libs/liburing/src/arch/aarch64/lib.h | 2 | ||||
-rw-r--r-- | contrib/libs/liburing/src/include/liburing.h | 407 | ||||
-rw-r--r-- | contrib/libs/liburing/src/include/liburing/compat.h | 2 | ||||
-rw-r--r-- | contrib/libs/liburing/src/include/liburing/io_uring.h | 70 | ||||
-rw-r--r-- | contrib/libs/liburing/src/include/liburing/io_uring_version.h | 8 | ||||
-rw-r--r-- | contrib/libs/liburing/src/int_flags.h | 1 | ||||
-rw-r--r-- | contrib/libs/liburing/src/lib.h | 21 | ||||
-rw-r--r-- | contrib/libs/liburing/src/nolibc.c | 56 | ||||
-rw-r--r-- | contrib/libs/liburing/src/queue.c | 4 | ||||
-rw-r--r-- | contrib/libs/liburing/src/register.c | 161 | ||||
-rw-r--r-- | contrib/libs/liburing/src/setup.c | 114 | ||||
-rw-r--r-- | contrib/libs/liburing/src/version.c | 22 |
12 files changed, 568 insertions, 300 deletions
diff --git a/contrib/libs/liburing/src/arch/aarch64/lib.h b/contrib/libs/liburing/src/arch/aarch64/lib.h index 5a75c1a6723..3b701b1fb77 100644 --- a/contrib/libs/liburing/src/arch/aarch64/lib.h +++ b/contrib/libs/liburing/src/arch/aarch64/lib.h @@ -21,7 +21,7 @@ static inline long __get_page_size(void) ssize_t x; x = __sys_read(fd, buf, sizeof(buf)); - if (x < sizeof(buf)) + if (x < (long) sizeof(buf)) break; if (buf[0] == AT_PAGESZ) { diff --git a/contrib/libs/liburing/src/include/liburing.h b/contrib/libs/liburing/src/include/liburing.h index 12a703fdd2e..e6dc9a792f0 100644 --- a/contrib/libs/liburing/src/include/liburing.h +++ b/contrib/libs/liburing/src/include/liburing.h @@ -23,6 +23,7 @@ #include <linux/swab.h> #include "liburing/compat.h" #include "liburing/io_uring.h" +#include "liburing/io_uring_version.h" #include "liburing/barrier.h" #ifndef uring_unlikely @@ -33,6 +34,10 @@ #define uring_likely(cond) __builtin_expect(!!(cond), 1) #endif +#ifndef IOURINGINLINE +#define IOURINGINLINE static inline +#endif + #ifdef __alpha__ /* * alpha and mips are the exceptions, all other architectures have @@ -151,7 +156,7 @@ struct io_uring_probe *io_uring_get_probe(void); */ void io_uring_free_probe(struct io_uring_probe *probe); -static inline int io_uring_opcode_supported(const struct io_uring_probe *p, +IOURINGINLINE int io_uring_opcode_supported(const struct io_uring_probe *p, int op) { if (op > p->last_op) @@ -226,6 +231,7 @@ int io_uring_register_iowq_max_workers(struct io_uring *ring, unsigned int *values); int io_uring_register_ring_fd(struct io_uring *ring); int io_uring_unregister_ring_fd(struct io_uring *ring); +int io_uring_close_ring_fd(struct io_uring *ring); int io_uring_register_buf_ring(struct io_uring *ring, struct io_uring_buf_reg *reg, unsigned int flags); int io_uring_unregister_buf_ring(struct io_uring *ring, int bgid); @@ -251,6 +257,16 @@ int io_uring_register(unsigned int fd, unsigned int opcode, const void *arg, unsigned int nr_args); /* + * Mapped buffer ring alloc/register + unregister/free helpers + */ +struct io_uring_buf_ring *io_uring_setup_buf_ring(struct io_uring *ring, + unsigned int nentries, + int bgid, unsigned int flags, + int *ret); +int io_uring_free_buf_ring(struct io_uring *ring, struct io_uring_buf_ring *br, + unsigned int nentries, int bgid); + +/* * Helper for the peek/wait single cqe functions. Exported because of that, * but probably shouldn't be used directly in an application. */ @@ -283,8 +299,7 @@ int __io_uring_get_cqe(struct io_uring *ring, /* * Must be called after io_uring_for_each_cqe() */ -static inline void io_uring_cq_advance(struct io_uring *ring, - unsigned nr) +IOURINGINLINE void io_uring_cq_advance(struct io_uring *ring, unsigned nr) { if (nr) { struct io_uring_cq *cq = &ring->cq; @@ -301,7 +316,7 @@ static inline void io_uring_cq_advance(struct io_uring *ring, * Must be called after io_uring_{peek,wait}_cqe() after the cqe has * been processed by the application. */ -static inline void io_uring_cqe_seen(struct io_uring *ring, +IOURINGINLINE void io_uring_cqe_seen(struct io_uring *ring, struct io_uring_cqe *cqe) { if (cqe) @@ -316,12 +331,12 @@ static inline void io_uring_cqe_seen(struct io_uring *ring, * Associate pointer @data with the sqe, for later retrieval from the cqe * at command completion time with io_uring_cqe_get_data(). */ -static inline void io_uring_sqe_set_data(struct io_uring_sqe *sqe, void *data) +IOURINGINLINE void io_uring_sqe_set_data(struct io_uring_sqe *sqe, void *data) { sqe->user_data = (unsigned long) data; } -static inline void *io_uring_cqe_get_data(const struct io_uring_cqe *cqe) +IOURINGINLINE void *io_uring_cqe_get_data(const struct io_uring_cqe *cqe) { return (void *) (uintptr_t) cqe->user_data; } @@ -331,13 +346,13 @@ static inline void *io_uring_cqe_get_data(const struct io_uring_cqe *cqe) * time with io_uring_cqe_get_data64. Just like the non-64 variants, except * these store a 64-bit type rather than a data pointer. */ -static inline void io_uring_sqe_set_data64(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_sqe_set_data64(struct io_uring_sqe *sqe, __u64 data) { sqe->user_data = data; } -static inline __u64 io_uring_cqe_get_data64(const struct io_uring_cqe *cqe) +IOURINGINLINE __u64 io_uring_cqe_get_data64(const struct io_uring_cqe *cqe) { return cqe->user_data; } @@ -347,20 +362,20 @@ static inline __u64 io_uring_cqe_get_data64(const struct io_uring_cqe *cqe) */ #define LIBURING_HAVE_DATA64 -static inline void io_uring_sqe_set_flags(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_sqe_set_flags(struct io_uring_sqe *sqe, unsigned flags) { sqe->flags = (__u8) flags; } -static inline void __io_uring_set_target_fixed_file(struct io_uring_sqe *sqe, +IOURINGINLINE void __io_uring_set_target_fixed_file(struct io_uring_sqe *sqe, unsigned int file_index) { /* 0 means no fixed files, indexes should be encoded as "index + 1" */ sqe->file_index = file_index + 1; } -static inline void io_uring_prep_rw(int op, struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_rw(int op, struct io_uring_sqe *sqe, int fd, const void *addr, unsigned len, __u64 offset) { @@ -379,29 +394,28 @@ static inline void io_uring_prep_rw(int op, struct io_uring_sqe *sqe, int fd, sqe->__pad2[0] = 0; } -/** - * @pre Either fd_in or fd_out must be a pipe. - * @param off_in If fd_in refers to a pipe, off_in must be (int64_t) -1; - * If fd_in does not refer to a pipe and off_in is (int64_t) -1, - * then bytes are read from fd_in starting from the file offset - * and it is adjust appropriately; - * If fd_in does not refer to a pipe and off_in is not - * (int64_t) -1, then the starting offset of fd_in will be - * off_in. - * @param off_out The description of off_in also applied to off_out. - * @param splice_flags see man splice(2) for description of flags. +/* + * io_uring_prep_splice() - Either @fd_in or @fd_out must be a pipe. + * + * - If @fd_in refers to a pipe, @off_in is ignored and must be set to -1. + * + * - If @fd_in does not refer to a pipe and @off_in is -1, then @nbytes are read + * from @fd_in starting from the file offset, which is incremented by the + * number of bytes read. + * + * - If @fd_in does not refer to a pipe and @off_in is not -1, then the starting + * offset of @fd_in will be @off_in. * * This splice operation can be used to implement sendfile by splicing to an * intermediate pipe first, then splice to the final destination. * In fact, the implementation of sendfile in kernel uses splice internally. * * NOTE that even if fd_in or fd_out refers to a pipe, the splice operation - * can still failed with EINVAL if one of the fd doesn't explicitly support - * splice operation, e.g. reading from terminal is unsupported from kernel 5.7 - * to 5.11. + * can still fail with EINVAL if one of the fd doesn't explicitly support splice + * operation, e.g. reading from terminal is unsupported from kernel 5.7 to 5.11. * Check issue #291 for more information. */ -static inline void io_uring_prep_splice(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_splice(struct io_uring_sqe *sqe, int fd_in, int64_t off_in, int fd_out, int64_t off_out, unsigned int nbytes, @@ -414,7 +428,7 @@ static inline void io_uring_prep_splice(struct io_uring_sqe *sqe, sqe->splice_flags = splice_flags; } -static inline void io_uring_prep_tee(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_tee(struct io_uring_sqe *sqe, int fd_in, int fd_out, unsigned int nbytes, unsigned int splice_flags) @@ -425,14 +439,14 @@ static inline void io_uring_prep_tee(struct io_uring_sqe *sqe, sqe->splice_flags = splice_flags; } -static inline void io_uring_prep_readv(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_readv(struct io_uring_sqe *sqe, int fd, const struct iovec *iovecs, unsigned nr_vecs, __u64 offset) { io_uring_prep_rw(IORING_OP_READV, sqe, fd, iovecs, nr_vecs, offset); } -static inline void io_uring_prep_readv2(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_readv2(struct io_uring_sqe *sqe, int fd, const struct iovec *iovecs, unsigned nr_vecs, __u64 offset, int flags) @@ -441,7 +455,7 @@ static inline void io_uring_prep_readv2(struct io_uring_sqe *sqe, int fd, sqe->rw_flags = flags; } -static inline void io_uring_prep_read_fixed(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_read_fixed(struct io_uring_sqe *sqe, int fd, void *buf, unsigned nbytes, __u64 offset, int buf_index) { @@ -449,14 +463,14 @@ static inline void io_uring_prep_read_fixed(struct io_uring_sqe *sqe, int fd, sqe->buf_index = (__u16) buf_index; } -static inline void io_uring_prep_writev(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_writev(struct io_uring_sqe *sqe, int fd, const struct iovec *iovecs, unsigned nr_vecs, __u64 offset) { io_uring_prep_rw(IORING_OP_WRITEV, sqe, fd, iovecs, nr_vecs, offset); } -static inline void io_uring_prep_writev2(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_writev2(struct io_uring_sqe *sqe, int fd, const struct iovec *iovecs, unsigned nr_vecs, __u64 offset, int flags) @@ -465,7 +479,7 @@ static inline void io_uring_prep_writev2(struct io_uring_sqe *sqe, int fd, sqe->rw_flags = flags; } -static inline void io_uring_prep_write_fixed(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_write_fixed(struct io_uring_sqe *sqe, int fd, const void *buf, unsigned nbytes, __u64 offset, int buf_index) { @@ -473,21 +487,22 @@ static inline void io_uring_prep_write_fixed(struct io_uring_sqe *sqe, int fd, sqe->buf_index = (__u16) buf_index; } -static inline void io_uring_prep_recvmsg(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_recvmsg(struct io_uring_sqe *sqe, int fd, struct msghdr *msg, unsigned flags) { io_uring_prep_rw(IORING_OP_RECVMSG, sqe, fd, msg, 1, 0); sqe->msg_flags = flags; } -static inline void io_uring_prep_recvmsg_multishot(struct io_uring_sqe *sqe, int fd, - struct msghdr *msg, unsigned flags) +IOURINGINLINE void io_uring_prep_recvmsg_multishot(struct io_uring_sqe *sqe, + int fd, struct msghdr *msg, + unsigned flags) { io_uring_prep_recvmsg(sqe, fd, msg, flags); sqe->ioprio |= IORING_RECV_MULTISHOT; } -static inline void io_uring_prep_sendmsg(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_sendmsg(struct io_uring_sqe *sqe, int fd, const struct msghdr *msg, unsigned flags) { @@ -495,7 +510,7 @@ static inline void io_uring_prep_sendmsg(struct io_uring_sqe *sqe, int fd, sqe->msg_flags = flags; } -static inline unsigned __io_uring_prep_poll_mask(unsigned poll_mask) +IOURINGINLINE unsigned __io_uring_prep_poll_mask(unsigned poll_mask) { #if __BYTE_ORDER == __BIG_ENDIAN poll_mask = __swahw32(poll_mask); @@ -503,28 +518,28 @@ static inline unsigned __io_uring_prep_poll_mask(unsigned poll_mask) return poll_mask; } -static inline void io_uring_prep_poll_add(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_poll_add(struct io_uring_sqe *sqe, int fd, unsigned poll_mask) { io_uring_prep_rw(IORING_OP_POLL_ADD, sqe, fd, NULL, 0, 0); sqe->poll32_events = __io_uring_prep_poll_mask(poll_mask); } -static inline void io_uring_prep_poll_multishot(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_poll_multishot(struct io_uring_sqe *sqe, int fd, unsigned poll_mask) { io_uring_prep_poll_add(sqe, fd, poll_mask); sqe->len = IORING_POLL_ADD_MULTI; } -static inline void io_uring_prep_poll_remove(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_poll_remove(struct io_uring_sqe *sqe, __u64 user_data) { io_uring_prep_rw(IORING_OP_POLL_REMOVE, sqe, -1, NULL, 0, 0); sqe->addr = user_data; } -static inline void io_uring_prep_poll_update(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_poll_update(struct io_uring_sqe *sqe, __u64 old_user_data, __u64 new_user_data, unsigned poll_mask, unsigned flags) @@ -535,19 +550,19 @@ static inline void io_uring_prep_poll_update(struct io_uring_sqe *sqe, sqe->poll32_events = __io_uring_prep_poll_mask(poll_mask); } -static inline void io_uring_prep_fsync(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_fsync(struct io_uring_sqe *sqe, int fd, unsigned fsync_flags) { io_uring_prep_rw(IORING_OP_FSYNC, sqe, fd, NULL, 0, 0); sqe->fsync_flags = fsync_flags; } -static inline void io_uring_prep_nop(struct io_uring_sqe *sqe) +IOURINGINLINE void io_uring_prep_nop(struct io_uring_sqe *sqe) { io_uring_prep_rw(IORING_OP_NOP, sqe, -1, NULL, 0, 0); } -static inline void io_uring_prep_timeout(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_timeout(struct io_uring_sqe *sqe, struct __kernel_timespec *ts, unsigned count, unsigned flags) { @@ -555,7 +570,7 @@ static inline void io_uring_prep_timeout(struct io_uring_sqe *sqe, sqe->timeout_flags = flags; } -static inline void io_uring_prep_timeout_remove(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_timeout_remove(struct io_uring_sqe *sqe, __u64 user_data, unsigned flags) { io_uring_prep_rw(IORING_OP_TIMEOUT_REMOVE, sqe, -1, NULL, 0, 0); @@ -563,7 +578,7 @@ static inline void io_uring_prep_timeout_remove(struct io_uring_sqe *sqe, sqe->timeout_flags = flags; } -static inline void io_uring_prep_timeout_update(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_timeout_update(struct io_uring_sqe *sqe, struct __kernel_timespec *ts, __u64 user_data, unsigned flags) { @@ -573,7 +588,7 @@ static inline void io_uring_prep_timeout_update(struct io_uring_sqe *sqe, sqe->timeout_flags = flags | IORING_TIMEOUT_UPDATE; } -static inline void io_uring_prep_accept(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_accept(struct io_uring_sqe *sqe, int fd, struct sockaddr *addr, socklen_t *addrlen, int flags) { @@ -583,16 +598,19 @@ static inline void io_uring_prep_accept(struct io_uring_sqe *sqe, int fd, } /* accept directly into the fixed file table */ -static inline void io_uring_prep_accept_direct(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_accept_direct(struct io_uring_sqe *sqe, int fd, struct sockaddr *addr, socklen_t *addrlen, int flags, unsigned int file_index) { io_uring_prep_accept(sqe, fd, addr, addrlen, flags); + /* offset by 1 for allocation */ + if (file_index == IORING_FILE_INDEX_ALLOC) + file_index--; __io_uring_set_target_fixed_file(sqe, file_index); } -static inline void io_uring_prep_multishot_accept(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_multishot_accept(struct io_uring_sqe *sqe, int fd, struct sockaddr *addr, socklen_t *addrlen, int flags) { @@ -601,7 +619,7 @@ static inline void io_uring_prep_multishot_accept(struct io_uring_sqe *sqe, } /* multishot accept directly into the fixed file table */ -static inline void io_uring_prep_multishot_accept_direct(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_multishot_accept_direct(struct io_uring_sqe *sqe, int fd, struct sockaddr *addr, socklen_t *addrlen, @@ -611,7 +629,7 @@ static inline void io_uring_prep_multishot_accept_direct(struct io_uring_sqe *sq __io_uring_set_target_fixed_file(sqe, IORING_FILE_INDEX_ALLOC - 1); } -static inline void io_uring_prep_cancel64(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_cancel64(struct io_uring_sqe *sqe, __u64 user_data, int flags) { io_uring_prep_rw(IORING_OP_ASYNC_CANCEL, sqe, -1, NULL, 0, 0); @@ -619,20 +637,20 @@ static inline void io_uring_prep_cancel64(struct io_uring_sqe *sqe, sqe->cancel_flags = (__u32) flags; } -static inline void io_uring_prep_cancel(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_cancel(struct io_uring_sqe *sqe, void *user_data, int flags) { io_uring_prep_cancel64(sqe, (__u64) (uintptr_t) user_data, flags); } -static inline void io_uring_prep_cancel_fd(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_cancel_fd(struct io_uring_sqe *sqe, int fd, unsigned int flags) { io_uring_prep_rw(IORING_OP_ASYNC_CANCEL, sqe, fd, NULL, 0, 0); sqe->cancel_flags = (__u32) flags | IORING_ASYNC_CANCEL_FD; } -static inline void io_uring_prep_link_timeout(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_link_timeout(struct io_uring_sqe *sqe, struct __kernel_timespec *ts, unsigned flags) { @@ -640,14 +658,14 @@ static inline void io_uring_prep_link_timeout(struct io_uring_sqe *sqe, sqe->timeout_flags = flags; } -static inline void io_uring_prep_connect(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_connect(struct io_uring_sqe *sqe, int fd, const struct sockaddr *addr, socklen_t addrlen) { io_uring_prep_rw(IORING_OP_CONNECT, sqe, fd, addr, 0, addrlen); } -static inline void io_uring_prep_files_update(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_files_update(struct io_uring_sqe *sqe, int *fds, unsigned nr_fds, int offset) { @@ -655,15 +673,15 @@ static inline void io_uring_prep_files_update(struct io_uring_sqe *sqe, (__u64) offset); } -static inline void io_uring_prep_fallocate(struct io_uring_sqe *sqe, int fd, - int mode, off_t offset, off_t len) +IOURINGINLINE void io_uring_prep_fallocate(struct io_uring_sqe *sqe, int fd, + int mode, __u64 offset, __u64 len) { io_uring_prep_rw(IORING_OP_FALLOCATE, sqe, fd, 0, (unsigned int) mode, (__u64) offset); sqe->addr = (__u64) len; } -static inline void io_uring_prep_openat(struct io_uring_sqe *sqe, int dfd, +IOURINGINLINE void io_uring_prep_openat(struct io_uring_sqe *sqe, int dfd, const char *path, int flags, mode_t mode) { @@ -672,34 +690,37 @@ static inline void io_uring_prep_openat(struct io_uring_sqe *sqe, int dfd, } /* open directly into the fixed file table */ -static inline void io_uring_prep_openat_direct(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_openat_direct(struct io_uring_sqe *sqe, int dfd, const char *path, int flags, mode_t mode, unsigned file_index) { io_uring_prep_openat(sqe, dfd, path, flags, mode); + /* offset by 1 for allocation */ + if (file_index == IORING_FILE_INDEX_ALLOC) + file_index--; __io_uring_set_target_fixed_file(sqe, file_index); } -static inline void io_uring_prep_close(struct io_uring_sqe *sqe, int fd) +IOURINGINLINE void io_uring_prep_close(struct io_uring_sqe *sqe, int fd) { io_uring_prep_rw(IORING_OP_CLOSE, sqe, fd, NULL, 0, 0); } -static inline void io_uring_prep_close_direct(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_close_direct(struct io_uring_sqe *sqe, unsigned file_index) { io_uring_prep_close(sqe, 0); __io_uring_set_target_fixed_file(sqe, file_index); } -static inline void io_uring_prep_read(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_read(struct io_uring_sqe *sqe, int fd, void *buf, unsigned nbytes, __u64 offset) { io_uring_prep_rw(IORING_OP_READ, sqe, fd, buf, nbytes, offset); } -static inline void io_uring_prep_write(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_write(struct io_uring_sqe *sqe, int fd, const void *buf, unsigned nbytes, __u64 offset) { @@ -707,37 +728,54 @@ static inline void io_uring_prep_write(struct io_uring_sqe *sqe, int fd, } struct statx; -static inline void io_uring_prep_statx(struct io_uring_sqe *sqe, int dfd, - const char *path, int flags, unsigned mask, - struct statx *statxbuf) +IOURINGINLINE void io_uring_prep_statx(struct io_uring_sqe *sqe, int dfd, + const char *path, int flags, + unsigned mask, struct statx *statxbuf) { io_uring_prep_rw(IORING_OP_STATX, sqe, dfd, path, mask, (__u64) (unsigned long) statxbuf); sqe->statx_flags = (__u32) flags; } -static inline void io_uring_prep_fadvise(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_fadvise(struct io_uring_sqe *sqe, int fd, __u64 offset, off_t len, int advice) { io_uring_prep_rw(IORING_OP_FADVISE, sqe, fd, NULL, (__u32) len, offset); sqe->fadvise_advice = (__u32) advice; } -static inline void io_uring_prep_madvise(struct io_uring_sqe *sqe, void *addr, +IOURINGINLINE void io_uring_prep_madvise(struct io_uring_sqe *sqe, void *addr, off_t length, int advice) { io_uring_prep_rw(IORING_OP_MADVISE, sqe, -1, addr, (__u32) length, 0); sqe->fadvise_advice = (__u32) advice; } -static inline void io_uring_prep_send(struct io_uring_sqe *sqe, int sockfd, +IOURINGINLINE void io_uring_prep_send(struct io_uring_sqe *sqe, int sockfd, const void *buf, size_t len, int flags) { io_uring_prep_rw(IORING_OP_SEND, sqe, sockfd, buf, (__u32) len, 0); sqe->msg_flags = (__u32) flags; } -static inline void io_uring_prep_send_zc(struct io_uring_sqe *sqe, int sockfd, +IOURINGINLINE void io_uring_prep_send_set_addr(struct io_uring_sqe *sqe, + const struct sockaddr *dest_addr, + __u16 addr_len) +{ + sqe->addr2 = (unsigned long)(const void *)dest_addr; + sqe->addr_len = addr_len; +} + +IOURINGINLINE void io_uring_prep_sendto(struct io_uring_sqe *sqe, int sockfd, + const void *buf, size_t len, int flags, + const struct sockaddr *addr, + socklen_t addrlen) +{ + io_uring_prep_send(sqe, sockfd, buf, len, flags); + io_uring_prep_send_set_addr(sqe, addr, addrlen); +} + +IOURINGINLINE void io_uring_prep_send_zc(struct io_uring_sqe *sqe, int sockfd, const void *buf, size_t len, int flags, unsigned zc_flags) { @@ -746,7 +784,7 @@ static inline void io_uring_prep_send_zc(struct io_uring_sqe *sqe, int sockfd, sqe->ioprio = zc_flags; } -static inline void io_uring_prep_send_zc_fixed(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_send_zc_fixed(struct io_uring_sqe *sqe, int sockfd, const void *buf, size_t len, int flags, unsigned zc_flags, @@ -757,7 +795,7 @@ static inline void io_uring_prep_send_zc_fixed(struct io_uring_sqe *sqe, sqe->buf_index = buf_index; } -static inline void io_uring_prep_sendmsg_zc(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_sendmsg_zc(struct io_uring_sqe *sqe, int fd, const struct msghdr *msg, unsigned flags) { @@ -765,22 +803,14 @@ static inline void io_uring_prep_sendmsg_zc(struct io_uring_sqe *sqe, int fd, sqe->opcode = IORING_OP_SENDMSG_ZC; } -static inline void io_uring_prep_send_set_addr(struct io_uring_sqe *sqe, - const struct sockaddr *dest_addr, - __u16 addr_len) -{ - sqe->addr2 = (unsigned long)(const void *)dest_addr; - sqe->addr_len = addr_len; -} - -static inline void io_uring_prep_recv(struct io_uring_sqe *sqe, int sockfd, +IOURINGINLINE void io_uring_prep_recv(struct io_uring_sqe *sqe, int sockfd, void *buf, size_t len, int flags) { io_uring_prep_rw(IORING_OP_RECV, sqe, sockfd, buf, (__u32) len, 0); sqe->msg_flags = (__u32) flags; } -static inline void io_uring_prep_recv_multishot(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_recv_multishot(struct io_uring_sqe *sqe, int sockfd, void *buf, size_t len, int flags) { @@ -788,7 +818,7 @@ static inline void io_uring_prep_recv_multishot(struct io_uring_sqe *sqe, sqe->ioprio |= IORING_RECV_MULTISHOT; } -static inline struct io_uring_recvmsg_out * +IOURINGINLINE struct io_uring_recvmsg_out * io_uring_recvmsg_validate(void *buf, int buf_len, struct msghdr *msgh) { unsigned long header = msgh->msg_controllen + msgh->msg_namelen + @@ -798,12 +828,12 @@ io_uring_recvmsg_validate(void *buf, int buf_len, struct msghdr *msgh) return (struct io_uring_recvmsg_out *)buf; } -static inline void *io_uring_recvmsg_name(struct io_uring_recvmsg_out *o) +IOURINGINLINE void *io_uring_recvmsg_name(struct io_uring_recvmsg_out *o) { return (void *) &o[1]; } -static inline struct cmsghdr * +IOURINGINLINE struct cmsghdr * io_uring_recvmsg_cmsg_firsthdr(struct io_uring_recvmsg_out *o, struct msghdr *msgh) { @@ -814,7 +844,7 @@ io_uring_recvmsg_cmsg_firsthdr(struct io_uring_recvmsg_out *o, msgh->msg_namelen); } -static inline struct cmsghdr * +IOURINGINLINE struct cmsghdr * io_uring_recvmsg_cmsg_nexthdr(struct io_uring_recvmsg_out *o, struct msghdr *msgh, struct cmsghdr *cmsg) { @@ -835,14 +865,14 @@ io_uring_recvmsg_cmsg_nexthdr(struct io_uring_recvmsg_out *o, struct msghdr *msg return cmsg; } -static inline void *io_uring_recvmsg_payload(struct io_uring_recvmsg_out *o, +IOURINGINLINE void *io_uring_recvmsg_payload(struct io_uring_recvmsg_out *o, struct msghdr *msgh) { return (void *)((unsigned char *)io_uring_recvmsg_name(o) + msgh->msg_namelen + msgh->msg_controllen); } -static inline unsigned int +IOURINGINLINE unsigned int io_uring_recvmsg_payload_length(struct io_uring_recvmsg_out *o, int buf_len, struct msghdr *msgh) { @@ -853,7 +883,7 @@ io_uring_recvmsg_payload_length(struct io_uring_recvmsg_out *o, return (unsigned int) (payload_end - payload_start); } -static inline void io_uring_prep_openat2(struct io_uring_sqe *sqe, int dfd, +IOURINGINLINE void io_uring_prep_openat2(struct io_uring_sqe *sqe, int dfd, const char *path, struct open_how *how) { io_uring_prep_rw(IORING_OP_OPENAT2, sqe, dfd, path, sizeof(*how), @@ -861,17 +891,20 @@ static inline void io_uring_prep_openat2(struct io_uring_sqe *sqe, int dfd, } /* open directly into the fixed file table */ -static inline void io_uring_prep_openat2_direct(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_openat2_direct(struct io_uring_sqe *sqe, int dfd, const char *path, struct open_how *how, unsigned file_index) { io_uring_prep_openat2(sqe, dfd, path, how); + /* offset by 1 for allocation */ + if (file_index == IORING_FILE_INDEX_ALLOC) + file_index--; __io_uring_set_target_fixed_file(sqe, file_index); } struct epoll_event; -static inline void io_uring_prep_epoll_ctl(struct io_uring_sqe *sqe, int epfd, +IOURINGINLINE void io_uring_prep_epoll_ctl(struct io_uring_sqe *sqe, int epfd, int fd, int op, struct epoll_event *ev) { @@ -879,7 +912,7 @@ static inline void io_uring_prep_epoll_ctl(struct io_uring_sqe *sqe, int epfd, (__u32) op, (__u32) fd); } -static inline void io_uring_prep_provide_buffers(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_provide_buffers(struct io_uring_sqe *sqe, void *addr, int len, int nr, int bgid, int bid) { @@ -888,33 +921,33 @@ static inline void io_uring_prep_provide_buffers(struct io_uring_sqe *sqe, sqe->buf_group = (__u16) bgid; } -static inline void io_uring_prep_remove_buffers(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_remove_buffers(struct io_uring_sqe *sqe, int nr, int bgid) { io_uring_prep_rw(IORING_OP_REMOVE_BUFFERS, sqe, nr, NULL, 0, 0); sqe->buf_group = (__u16) bgid; } -static inline void io_uring_prep_shutdown(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_shutdown(struct io_uring_sqe *sqe, int fd, int how) { io_uring_prep_rw(IORING_OP_SHUTDOWN, sqe, fd, NULL, (__u32) how, 0); } -static inline void io_uring_prep_unlinkat(struct io_uring_sqe *sqe, int dfd, +IOURINGINLINE void io_uring_prep_unlinkat(struct io_uring_sqe *sqe, int dfd, const char *path, int flags) { io_uring_prep_rw(IORING_OP_UNLINKAT, sqe, dfd, path, 0, 0); sqe->unlink_flags = (__u32) flags; } -static inline void io_uring_prep_unlink(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_unlink(struct io_uring_sqe *sqe, const char *path, int flags) { io_uring_prep_unlinkat(sqe, AT_FDCWD, path, flags); } -static inline void io_uring_prep_renameat(struct io_uring_sqe *sqe, int olddfd, +IOURINGINLINE void io_uring_prep_renameat(struct io_uring_sqe *sqe, int olddfd, const char *oldpath, int newdfd, const char *newpath, unsigned int flags) { @@ -924,13 +957,14 @@ static inline void io_uring_prep_renameat(struct io_uring_sqe *sqe, int olddfd, sqe->rename_flags = (__u32) flags; } -static inline void io_uring_prep_rename(struct io_uring_sqe *sqe, - const char *oldpath, const char *newpath) +IOURINGINLINE void io_uring_prep_rename(struct io_uring_sqe *sqe, + const char *oldpath, + const char *newpath) { io_uring_prep_renameat(sqe, AT_FDCWD, oldpath, AT_FDCWD, newpath, 0); } -static inline void io_uring_prep_sync_file_range(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_sync_file_range(struct io_uring_sqe *sqe, int fd, unsigned len, __u64 offset, int flags) { @@ -938,19 +972,19 @@ static inline void io_uring_prep_sync_file_range(struct io_uring_sqe *sqe, sqe->sync_range_flags = (__u32) flags; } -static inline void io_uring_prep_mkdirat(struct io_uring_sqe *sqe, int dfd, +IOURINGINLINE void io_uring_prep_mkdirat(struct io_uring_sqe *sqe, int dfd, const char *path, mode_t mode) { io_uring_prep_rw(IORING_OP_MKDIRAT, sqe, dfd, path, mode, 0); } -static inline void io_uring_prep_mkdir(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_mkdir(struct io_uring_sqe *sqe, const char *path, mode_t mode) { io_uring_prep_mkdirat(sqe, AT_FDCWD, path, mode); } -static inline void io_uring_prep_symlinkat(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_symlinkat(struct io_uring_sqe *sqe, const char *target, int newdirfd, const char *linkpath) { @@ -958,13 +992,14 @@ static inline void io_uring_prep_symlinkat(struct io_uring_sqe *sqe, (uint64_t) (uintptr_t) linkpath); } -static inline void io_uring_prep_symlink(struct io_uring_sqe *sqe, - const char *target, const char *linkpath) +IOURINGINLINE void io_uring_prep_symlink(struct io_uring_sqe *sqe, + const char *target, + const char *linkpath) { io_uring_prep_symlinkat(sqe, target, AT_FDCWD, linkpath); } -static inline void io_uring_prep_linkat(struct io_uring_sqe *sqe, int olddfd, +IOURINGINLINE void io_uring_prep_linkat(struct io_uring_sqe *sqe, int olddfd, const char *oldpath, int newdfd, const char *newpath, int flags) { @@ -973,25 +1008,55 @@ static inline void io_uring_prep_linkat(struct io_uring_sqe *sqe, int olddfd, sqe->hardlink_flags = (__u32) flags; } -static inline void io_uring_prep_link(struct io_uring_sqe *sqe, - const char *oldpath, const char *newpath, int flags) +IOURINGINLINE void io_uring_prep_link(struct io_uring_sqe *sqe, + const char *oldpath, const char *newpath, + int flags) { io_uring_prep_linkat(sqe, AT_FDCWD, oldpath, AT_FDCWD, newpath, flags); } -static inline void io_uring_prep_msg_ring(struct io_uring_sqe *sqe, int fd, +IOURINGINLINE void io_uring_prep_msg_ring_cqe_flags(struct io_uring_sqe *sqe, + int fd, unsigned int len, __u64 data, + unsigned int flags, unsigned int cqe_flags) +{ + io_uring_prep_rw(IORING_OP_MSG_RING, sqe, fd, NULL, len, data); + sqe->msg_ring_flags = IORING_MSG_RING_FLAGS_PASS | flags; + sqe->file_index = cqe_flags; +} + +IOURINGINLINE void io_uring_prep_msg_ring(struct io_uring_sqe *sqe, int fd, unsigned int len, __u64 data, unsigned int flags) { io_uring_prep_rw(IORING_OP_MSG_RING, sqe, fd, NULL, len, data); - sqe->rw_flags = flags; + sqe->msg_ring_flags = flags; } -static inline void io_uring_prep_getxattr(struct io_uring_sqe *sqe, - const char *name, - char *value, - const char *path, - unsigned int len) +IOURINGINLINE void io_uring_prep_msg_ring_fd(struct io_uring_sqe *sqe, int fd, + int source_fd, int target_fd, + __u64 data, unsigned int flags) +{ + io_uring_prep_rw(IORING_OP_MSG_RING, sqe, fd, + (void *) (uintptr_t) IORING_MSG_SEND_FD, 0, data); + sqe->addr3 = source_fd; + /* offset by 1 for allocation */ + if ((unsigned int) target_fd == IORING_FILE_INDEX_ALLOC) + target_fd--; + __io_uring_set_target_fixed_file(sqe, target_fd); + sqe->msg_ring_flags = flags; +} + +IOURINGINLINE void io_uring_prep_msg_ring_fd_alloc(struct io_uring_sqe *sqe, + int fd, int source_fd, + __u64 data, unsigned int flags) +{ + io_uring_prep_msg_ring_fd(sqe, fd, source_fd, IORING_FILE_INDEX_ALLOC, + data, flags); +} + +IOURINGINLINE void io_uring_prep_getxattr(struct io_uring_sqe *sqe, + const char *name, char *value, + const char *path, unsigned int len) { io_uring_prep_rw(IORING_OP_GETXATTR, sqe, 0, name, len, (__u64) (uintptr_t) value); @@ -999,11 +1064,9 @@ static inline void io_uring_prep_getxattr(struct io_uring_sqe *sqe, sqe->xattr_flags = 0; } -static inline void io_uring_prep_setxattr(struct io_uring_sqe *sqe, - const char *name, - const char *value, - const char *path, - int flags, +IOURINGINLINE void io_uring_prep_setxattr(struct io_uring_sqe *sqe, + const char *name, const char *value, + const char *path, int flags, unsigned int len) { io_uring_prep_rw(IORING_OP_SETXATTR, sqe, 0, name, len, @@ -1012,30 +1075,25 @@ static inline void io_uring_prep_setxattr(struct io_uring_sqe *sqe, sqe->xattr_flags = flags; } -static inline void io_uring_prep_fgetxattr(struct io_uring_sqe *sqe, - int fd, - const char *name, - char *value, - unsigned int len) +IOURINGINLINE void io_uring_prep_fgetxattr(struct io_uring_sqe *sqe, + int fd, const char *name, + char *value, unsigned int len) { io_uring_prep_rw(IORING_OP_FGETXATTR, sqe, fd, name, len, (__u64) (uintptr_t) value); sqe->xattr_flags = 0; } -static inline void io_uring_prep_fsetxattr(struct io_uring_sqe *sqe, - int fd, - const char *name, - const char *value, - int flags, - unsigned int len) +IOURINGINLINE void io_uring_prep_fsetxattr(struct io_uring_sqe *sqe, int fd, + const char *name, const char *value, + int flags, unsigned int len) { io_uring_prep_rw(IORING_OP_FSETXATTR, sqe, fd, name, len, (__u64) (uintptr_t) value); sqe->xattr_flags = flags; } -static inline void io_uring_prep_socket(struct io_uring_sqe *sqe, int domain, +IOURINGINLINE void io_uring_prep_socket(struct io_uring_sqe *sqe, int domain, int type, int protocol, unsigned int flags) { @@ -1043,7 +1101,7 @@ static inline void io_uring_prep_socket(struct io_uring_sqe *sqe, int domain, sqe->rw_flags = flags; } -static inline void io_uring_prep_socket_direct(struct io_uring_sqe *sqe, +IOURINGINLINE void io_uring_prep_socket_direct(struct io_uring_sqe *sqe, int domain, int type, int protocol, unsigned file_index, @@ -1051,12 +1109,16 @@ static inline void io_uring_prep_socket_direct(struct io_uring_sqe *sqe, { io_uring_prep_rw(IORING_OP_SOCKET, sqe, domain, NULL, protocol, type); sqe->rw_flags = flags; + /* offset by 1 for allocation */ + if (file_index == IORING_FILE_INDEX_ALLOC) + file_index--; __io_uring_set_target_fixed_file(sqe, file_index); } -static inline void io_uring_prep_socket_direct_alloc(struct io_uring_sqe *sqe, - int domain, int type, int protocol, - unsigned int flags) +IOURINGINLINE void io_uring_prep_socket_direct_alloc(struct io_uring_sqe *sqe, + int domain, int type, + int protocol, + unsigned int flags) { io_uring_prep_rw(IORING_OP_SOCKET, sqe, domain, NULL, protocol, type); sqe->rw_flags = flags; @@ -1067,7 +1129,7 @@ static inline void io_uring_prep_socket_direct_alloc(struct io_uring_sqe *sqe, * Returns number of unconsumed (if SQPOLL) or unsubmitted entries exist in * the SQ ring */ -static inline unsigned io_uring_sq_ready(const struct io_uring *ring) +IOURINGINLINE unsigned io_uring_sq_ready(const struct io_uring *ring) { unsigned khead = *ring->sq.khead; @@ -1086,7 +1148,7 @@ static inline unsigned io_uring_sq_ready(const struct io_uring *ring) /* * Returns how much space is left in the SQ ring. */ -static inline unsigned io_uring_sq_space_left(const struct io_uring *ring) +IOURINGINLINE unsigned io_uring_sq_space_left(const struct io_uring *ring) { return ring->sq.ring_entries - io_uring_sq_ready(ring); } @@ -1098,7 +1160,7 @@ static inline unsigned io_uring_sq_space_left(const struct io_uring *ring) * action is taken. Note: may return -EINVAL if the kernel doesn't support * this feature. */ -static inline int io_uring_sqring_wait(struct io_uring *ring) +IOURINGINLINE int io_uring_sqring_wait(struct io_uring *ring) { if (!(ring->flags & IORING_SETUP_SQPOLL)) return 0; @@ -1111,7 +1173,7 @@ static inline int io_uring_sqring_wait(struct io_uring *ring) /* * Returns how many unconsumed entries are ready in the CQ ring */ -static inline unsigned io_uring_cq_ready(const struct io_uring *ring) +IOURINGINLINE unsigned io_uring_cq_ready(const struct io_uring *ring) { return io_uring_smp_load_acquire(ring->cq.ktail) - *ring->cq.khead; } @@ -1120,7 +1182,7 @@ static inline unsigned io_uring_cq_ready(const struct io_uring *ring) * Returns true if there are overflow entries waiting to be flushed onto * the CQ ring */ -static inline bool io_uring_cq_has_overflow(const struct io_uring *ring) +IOURINGINLINE bool io_uring_cq_has_overflow(const struct io_uring *ring) { return IO_URING_READ_ONCE(*ring->sq.kflags) & IORING_SQ_CQ_OVERFLOW; } @@ -1128,7 +1190,7 @@ static inline bool io_uring_cq_has_overflow(const struct io_uring *ring) /* * Returns true if the eventfd notification is currently enabled */ -static inline bool io_uring_cq_eventfd_enabled(const struct io_uring *ring) +IOURINGINLINE bool io_uring_cq_eventfd_enabled(const struct io_uring *ring) { if (!ring->cq.kflags) return true; @@ -1140,7 +1202,7 @@ static inline bool io_uring_cq_eventfd_enabled(const struct io_uring *ring) * Toggle eventfd notification on or off, if an eventfd is registered with * the ring. */ -static inline int io_uring_cq_eventfd_toggle(struct io_uring *ring, +IOURINGINLINE int io_uring_cq_eventfd_toggle(struct io_uring *ring, bool enabled) { uint32_t flags; @@ -1168,7 +1230,7 @@ static inline int io_uring_cq_eventfd_toggle(struct io_uring *ring, * readily available. Returns 0 with cqe_ptr filled in on success, -errno on * failure. */ -static inline int io_uring_wait_cqe_nr(struct io_uring *ring, +IOURINGINLINE int io_uring_wait_cqe_nr(struct io_uring *ring, struct io_uring_cqe **cqe_ptr, unsigned wait_nr) { @@ -1180,7 +1242,7 @@ static inline int io_uring_wait_cqe_nr(struct io_uring *ring, * "official" versions of this, io_uring_peek_cqe(), io_uring_wait_cqe(), * or io_uring_wait_cqes*(). */ -static inline int __io_uring_peek_cqe(struct io_uring *ring, +IOURINGINLINE int __io_uring_peek_cqe(struct io_uring *ring, struct io_uring_cqe **cqe_ptr, unsigned *nr_available) { @@ -1226,7 +1288,7 @@ static inline int __io_uring_peek_cqe(struct io_uring *ring, * Return an IO completion, if one is readily available. Returns 0 with * cqe_ptr filled in on success, -errno on failure. */ -static inline int io_uring_peek_cqe(struct io_uring *ring, +IOURINGINLINE int io_uring_peek_cqe(struct io_uring *ring, struct io_uring_cqe **cqe_ptr) { if (!__io_uring_peek_cqe(ring, cqe_ptr, NULL) && *cqe_ptr) @@ -1239,7 +1301,7 @@ static inline int io_uring_peek_cqe(struct io_uring *ring, * Return an IO completion, waiting for it if necessary. Returns 0 with * cqe_ptr filled in on success, -errno on failure. */ -static inline int io_uring_wait_cqe(struct io_uring *ring, +IOURINGINLINE int io_uring_wait_cqe(struct io_uring *ring, struct io_uring_cqe **cqe_ptr) { if (!__io_uring_peek_cqe(ring, cqe_ptr, NULL) && *cqe_ptr) @@ -1255,7 +1317,7 @@ static inline int io_uring_wait_cqe(struct io_uring *ring, * * Returns a vacant sqe, or NULL if we're full. */ -static inline struct io_uring_sqe *_io_uring_get_sqe(struct io_uring *ring) +IOURINGINLINE struct io_uring_sqe *_io_uring_get_sqe(struct io_uring *ring) { struct io_uring_sq *sq = &ring->sq; unsigned int head, next = sq->sqe_tail + 1; @@ -1282,12 +1344,12 @@ static inline struct io_uring_sqe *_io_uring_get_sqe(struct io_uring *ring) /* * Return the appropriate mask for a buffer ring of size 'ring_entries' */ -static inline int io_uring_buf_ring_mask(__u32 ring_entries) +IOURINGINLINE int io_uring_buf_ring_mask(__u32 ring_entries) { return ring_entries - 1; } -static inline void io_uring_buf_ring_init(struct io_uring_buf_ring *br) +IOURINGINLINE void io_uring_buf_ring_init(struct io_uring_buf_ring *br) { br->tail = 0; } @@ -1295,7 +1357,7 @@ static inline void io_uring_buf_ring_init(struct io_uring_buf_ring *br) /* * Assign 'buf' with the addr/len/buffer ID supplied */ -static inline void io_uring_buf_ring_add(struct io_uring_buf_ring *br, +IOURINGINLINE void io_uring_buf_ring_add(struct io_uring_buf_ring *br, void *addr, unsigned int len, unsigned short bid, int mask, int buf_offset) @@ -1312,7 +1374,7 @@ static inline void io_uring_buf_ring_add(struct io_uring_buf_ring *br, * io_uring_buf_ring_add() has been called 'count' times to fill in new * buffers. */ -static inline void io_uring_buf_ring_advance(struct io_uring_buf_ring *br, +IOURINGINLINE void io_uring_buf_ring_advance(struct io_uring_buf_ring *br, int count) { unsigned short new_tail = br->tail + count; @@ -1320,6 +1382,14 @@ static inline void io_uring_buf_ring_advance(struct io_uring_buf_ring *br, io_uring_smp_store_release(&br->tail, new_tail); } +IOURINGINLINE void __io_uring_buf_ring_cq_advance(struct io_uring *ring, + struct io_uring_buf_ring *br, + int cq_count, int buf_count) +{ + br->tail += buf_count; + io_uring_cq_advance(ring, cq_count); +} + /* * Make 'count' new buffers visible to the kernel while at the same time * advancing the CQ ring seen entries. This can be used when the application @@ -1327,16 +1397,15 @@ static inline void io_uring_buf_ring_advance(struct io_uring_buf_ring *br, * avoiding an extra atomic when needing to increment both the CQ ring and * the ring buffer index at the same time. */ -static inline void io_uring_buf_ring_cq_advance(struct io_uring *ring, +IOURINGINLINE void io_uring_buf_ring_cq_advance(struct io_uring *ring, struct io_uring_buf_ring *br, int count) { - br->tail += count; - io_uring_cq_advance(ring, count); + __io_uring_buf_ring_cq_advance(ring, br, count, count); } #ifndef LIBURING_INTERNAL -static inline struct io_uring_sqe *io_uring_get_sqe(struct io_uring *ring) +IOURINGINLINE struct io_uring_sqe *io_uring_get_sqe(struct io_uring *ring) { return _io_uring_get_sqe(ring); } @@ -1347,8 +1416,30 @@ struct io_uring_sqe *io_uring_get_sqe(struct io_uring *ring); ssize_t io_uring_mlock_size(unsigned entries, unsigned flags); ssize_t io_uring_mlock_size_params(unsigned entries, struct io_uring_params *p); +/* + * Versioning information for liburing. + * + * Use IO_URING_CHECK_VERSION() for compile time checks including from + * preprocessor directives. + * + * Use io_uring_check_version() for runtime checks of the version of + * liburing that was loaded by the dynamic linker. + */ +int io_uring_major_version(void); +int io_uring_minor_version(void); +bool io_uring_check_version(int major, int minor); + +#define IO_URING_CHECK_VERSION(major,minor) \ + (major > IO_URING_VERSION_MAJOR || \ + (major == IO_URING_VERSION_MAJOR && \ + minor >= IO_URING_VERSION_MINOR)) + #ifdef __cplusplus } #endif +#ifdef IOURINGINLINE +#undef IOURINGINLINE +#endif + #endif diff --git a/contrib/libs/liburing/src/include/liburing/compat.h b/contrib/libs/liburing/src/include/liburing/compat.h index e5794395781..717f81cb7fa 100644 --- a/contrib/libs/liburing/src/include/liburing/compat.h +++ b/contrib/libs/liburing/src/include/liburing/compat.h @@ -3,6 +3,8 @@ #define LIBURING_COMPAT_H #include <linux/time_types.h> +/* <linux/time_types.h> is included above and not needed again */ +#define UAPI_LINUX_IO_URING_H_SKIP_LINUX_TIME_TYPES_H 1 #include <linux/openat2.h> diff --git a/contrib/libs/liburing/src/include/liburing/io_uring.h b/contrib/libs/liburing/src/include/liburing/io_uring.h index a3e09208df7..84c33b5e841 100644 --- a/contrib/libs/liburing/src/include/liburing/io_uring.h +++ b/contrib/libs/liburing/src/include/liburing/io_uring.h @@ -12,12 +12,11 @@ #include <linux/types.h> /* * this file is shared with liburing and that has to autodetect - * if linux/time_types.h is available + * if linux/time_types.h is available or not, it can + * define UAPI_LINUX_IO_URING_H_SKIP_LINUX_TIME_TYPES_H + * if linux/time_types.h is not available */ -#ifdef __KERNEL__ -#define HAVE_LINUX_TIME_TYPES_H 1 -#endif -#ifdef HAVE_LINUX_TIME_TYPES_H +#ifndef UAPI_LINUX_IO_URING_H_SKIP_LINUX_TIME_TYPES_H #include <linux/time_types.h> #endif @@ -231,7 +230,7 @@ enum io_uring_op { /* * sqe->uring_cmd_flags - * IORING_URING_CMD_FIXED use registered buffer; pass thig flag + * IORING_URING_CMD_FIXED use registered buffer; pass this flag * along with setting sqe->buf_index. */ #define IORING_URING_CMD_FIXED (1U << 0) @@ -251,6 +250,7 @@ enum io_uring_op { #define IORING_TIMEOUT_REALTIME (1U << 3) #define IORING_LINK_TIMEOUT_UPDATE (1U << 4) #define IORING_TIMEOUT_ETIME_SUCCESS (1U << 5) +#define IORING_TIMEOUT_MULTISHOT (1U << 6) #define IORING_TIMEOUT_CLOCK_MASK (IORING_TIMEOUT_BOOTTIME | IORING_TIMEOUT_REALTIME) #define IORING_TIMEOUT_UPDATE_MASK (IORING_TIMEOUT_UPDATE | IORING_LINK_TIMEOUT_UPDATE) /* @@ -305,10 +305,28 @@ enum io_uring_op { * * IORING_RECVSEND_FIXED_BUF Use registered buffers, the index is stored in * the buf_index field. + * + * IORING_SEND_ZC_REPORT_USAGE + * If set, SEND[MSG]_ZC should report + * the zerocopy usage in cqe.res + * for the IORING_CQE_F_NOTIF cqe. + * 0 is reported if zerocopy was actually possible. + * IORING_NOTIF_USAGE_ZC_COPIED if data was copied + * (at least partially). */ #define IORING_RECVSEND_POLL_FIRST (1U << 0) #define IORING_RECV_MULTISHOT (1U << 1) #define IORING_RECVSEND_FIXED_BUF (1U << 2) +#define IORING_SEND_ZC_REPORT_USAGE (1U << 3) + +/* + * cqe.res for IORING_CQE_F_NOTIF if + * IORING_SEND_ZC_REPORT_USAGE was requested + * + * It should be treated as a flag, all other + * bits of cqe.res should be treated as reserved! + */ +#define IORING_NOTIF_USAGE_ZC_COPIED (1U << 31) /* * accept flags stored in sqe->ioprio @@ -330,6 +348,8 @@ enum { * applicable for IORING_MSG_DATA, obviously. */ #define IORING_MSG_RING_CQE_SKIP (1U << 0) +/* Pass through the flags from sqe->file_index to cqe->flags */ +#define IORING_MSG_RING_FLAGS_PASS (1U << 1) /* * IO completion data structure (Completion Queue Entry) @@ -370,6 +390,9 @@ enum { #define IORING_OFF_SQ_RING 0ULL #define IORING_OFF_CQ_RING 0x8000000ULL #define IORING_OFF_SQES 0x10000000ULL +#define IORING_OFF_PBUF_RING 0x80000000ULL +#define IORING_OFF_PBUF_SHIFT 16 +#define IORING_OFF_MMAP_MASK 0xf8000000ULL /* * Filled with the offset for mmap(2) @@ -453,6 +476,7 @@ struct io_uring_params { #define IORING_FEAT_RSRC_TAGS (1U << 10) #define IORING_FEAT_CQE_SKIP (1U << 11) #define IORING_FEAT_LINKED_FILE (1U << 12) +#define IORING_FEAT_REG_REG_RING (1U << 13) /* * io_uring_register(2) opcodes and arguments @@ -500,7 +524,10 @@ enum { IORING_REGISTER_FILE_ALLOC_RANGE = 25, /* this goes last */ - IORING_REGISTER_LAST + IORING_REGISTER_LAST, + + /* flag added to the opcode to use a registered ring fd */ + IORING_REGISTER_USE_REGISTERED_RING = 1U << 31 }; /* io-wq worker categories */ @@ -545,19 +572,6 @@ struct io_uring_rsrc_update2 { __u32 resv2; }; -struct io_uring_notification_slot { - __u64 tag; - __u64 resv[3]; -}; - -struct io_uring_notification_register { - __u32 nr_slots; - __u32 resv; - __u64 resv2; - __u64 data; - __u64 resv3; -}; - /* Skip updating fd indexes set to this value in the fd table */ #define IORING_REGISTER_FILES_SKIP (-2) @@ -612,12 +626,26 @@ struct io_uring_buf_ring { }; }; +/* + * Flags for IORING_REGISTER_PBUF_RING. + * + * IOU_PBUF_RING_MMAP: If set, kernel will allocate the memory for the ring. + * The application must not set a ring_addr in struct + * io_uring_buf_reg, instead it must subsequently call + * mmap(2) with the offset set as: + * IORING_OFF_PBUF_RING | (bgid << IORING_OFF_PBUF_SHIFT) + * to get a virtual mapping for the ring. + */ +enum { + IOU_PBUF_RING_MMAP = 1, +}; + /* argument for IORING_(UN)REGISTER_PBUF_RING */ struct io_uring_buf_reg { __u64 ring_addr; __u32 ring_entries; __u16 bgid; - __u16 pad; + __u16 flags; __u64 resv[3]; }; diff --git a/contrib/libs/liburing/src/include/liburing/io_uring_version.h b/contrib/libs/liburing/src/include/liburing/io_uring_version.h new file mode 100644 index 00000000000..8029e041f9d --- /dev/null +++ b/contrib/libs/liburing/src/include/liburing/io_uring_version.h @@ -0,0 +1,8 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef LIBURING_VERSION_H +#define LIBURING_VERSION_H + +#define IO_URING_VERSION_MAJOR 2 +#define IO_URING_VERSION_MINOR 4 + +#endif diff --git a/contrib/libs/liburing/src/int_flags.h b/contrib/libs/liburing/src/int_flags.h index 90505ec340b..71774fbca53 100644 --- a/contrib/libs/liburing/src/int_flags.h +++ b/contrib/libs/liburing/src/int_flags.h @@ -4,6 +4,7 @@ enum { INT_FLAG_REG_RING = 1, + INT_FLAG_REG_REG_RING = 2, }; #endif diff --git a/contrib/libs/liburing/src/lib.h b/contrib/libs/liburing/src/lib.h index 82c0ab10d3f..635a30ece51 100644 --- a/contrib/libs/liburing/src/lib.h +++ b/contrib/libs/liburing/src/lib.h @@ -37,25 +37,14 @@ #define __hot __attribute__((__hot__)) #define __cold __attribute__((__cold__)) +#ifdef CONFIG_NOLIBC +void *__uring_memset(void *s, int c, size_t n); void *__uring_malloc(size_t len); void __uring_free(void *p); -static inline void *uring_malloc(size_t len) -{ -#ifdef CONFIG_NOLIBC - return __uring_malloc(len); -#else - return malloc(len); -#endif -} - -static inline void uring_free(void *ptr) -{ -#ifdef CONFIG_NOLIBC - __uring_free(ptr); -#else - free(ptr); +#define malloc(LEN) __uring_malloc(LEN) +#define free(PTR) __uring_free(PTR) +#define memset(PTR, C, LEN) __uring_memset(PTR, C, LEN) #endif -} #endif /* #ifndef LIBURING_LIB_H */ diff --git a/contrib/libs/liburing/src/nolibc.c b/contrib/libs/liburing/src/nolibc.c new file mode 100644 index 00000000000..22d23ec50df --- /dev/null +++ b/contrib/libs/liburing/src/nolibc.c @@ -0,0 +1,56 @@ +#include "../config-host.h" +/* SPDX-License-Identifier: MIT */ + +#ifndef CONFIG_NOLIBC +#error "This file should only be compiled for no libc build" +#endif + +#include "lib.h" +#include "syscall.h" + +void *__uring_memset(void *s, int c, size_t n) +{ + size_t i; + unsigned char *p = s; + + for (i = 0; i < n; i++) { + p[i] = (unsigned char) c; + + /* + * An empty inline ASM to avoid auto-vectorization + * because it's too bloated for liburing. + */ + __asm__ volatile (""); + } + + return s; +} + +struct uring_heap { + size_t len; + char user_p[] __attribute__((__aligned__)); +}; + +void *__uring_malloc(size_t len) +{ + struct uring_heap *heap; + + heap = __sys_mmap(NULL, sizeof(*heap) + len, PROT_READ | PROT_WRITE, + MAP_ANONYMOUS | MAP_PRIVATE, -1, 0); + if (IS_ERR(heap)) + return NULL; + + heap->len = sizeof(*heap) + len; + return heap->user_p; +} + +void __uring_free(void *p) +{ + struct uring_heap *heap; + + if (uring_unlikely(!p)) + return; + + heap = container_of(p, struct uring_heap, user_p); + __sys_munmap(heap, heap->len); +} diff --git a/contrib/libs/liburing/src/queue.c b/contrib/libs/liburing/src/queue.c index 4b4220782d6..9fca31fe349 100644 --- a/contrib/libs/liburing/src/queue.c +++ b/contrib/libs/liburing/src/queue.c @@ -82,7 +82,7 @@ static int _io_uring_get_cqe(struct io_uring *ring, } if (!cqe && !data->wait_nr && !data->submit) { /* - * If we already looped once, we already entererd + * If we already looped once, we already entered * the kernel. Since there's nothing to submit or * wait for, don't keep retrying. */ @@ -202,7 +202,7 @@ again: * Sync internal state with kernel ring state on the SQ side. Returns the * number of pending items in the SQ ring, for the shared ring. */ -unsigned __io_uring_flush_sq(struct io_uring *ring) +static unsigned __io_uring_flush_sq(struct io_uring *ring) { struct io_uring_sq *sq = &ring->sq; unsigned tail = sq->sqe_tail; diff --git a/contrib/libs/liburing/src/register.c b/contrib/libs/liburing/src/register.c index 13e80a5811e..5563db2c4ac 100644 --- a/contrib/libs/liburing/src/register.c +++ b/contrib/libs/liburing/src/register.c @@ -9,6 +9,15 @@ #include "liburing/compat.h" #include "liburing/io_uring.h" +static inline int do_register(struct io_uring *ring, unsigned int opcode, + const void *arg, unsigned int nr_args) +{ + if (ring->int_flags & INT_FLAG_REG_REG_RING) + opcode |= IORING_REGISTER_USE_REGISTERED_RING; + + return __sys_io_uring_register(ring->enter_ring_fd, opcode, arg, nr_args); +} + int io_uring_register_buffers_update_tag(struct io_uring *ring, unsigned off, const struct iovec *iovecs, const __u64 *tags, @@ -21,8 +30,7 @@ int io_uring_register_buffers_update_tag(struct io_uring *ring, unsigned off, .nr = nr, }; - return __sys_io_uring_register(ring->ring_fd,IORING_REGISTER_BUFFERS_UPDATE, &up, - sizeof(up)); + return do_register(ring, IORING_REGISTER_BUFFERS_UPDATE, &up, sizeof(up)); } int io_uring_register_buffers_tags(struct io_uring *ring, @@ -36,9 +44,7 @@ int io_uring_register_buffers_tags(struct io_uring *ring, .tags = (unsigned long)tags, }; - return __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_BUFFERS2, ®, - sizeof(reg)); + return do_register(ring, IORING_REGISTER_BUFFERS2, ®, sizeof(reg)); } int io_uring_register_buffers_sparse(struct io_uring *ring, unsigned nr) @@ -48,27 +54,18 @@ int io_uring_register_buffers_sparse(struct io_uring *ring, unsigned nr) .nr = nr, }; - return __sys_io_uring_register(ring->ring_fd, IORING_REGISTER_BUFFERS2, - ®, sizeof(reg)); + return do_register(ring, IORING_REGISTER_BUFFERS2, ®, sizeof(reg)); } int io_uring_register_buffers(struct io_uring *ring, const struct iovec *iovecs, unsigned nr_iovecs) { - int ret; - - ret = __sys_io_uring_register(ring->ring_fd, IORING_REGISTER_BUFFERS, - iovecs, nr_iovecs); - return (ret < 0) ? ret : 0; + return do_register(ring, IORING_REGISTER_BUFFERS, iovecs, nr_iovecs); } int io_uring_unregister_buffers(struct io_uring *ring) { - int ret; - - ret = __sys_io_uring_register(ring->ring_fd, IORING_UNREGISTER_BUFFERS, - NULL, 0); - return (ret < 0) ? ret : 0; + return do_register(ring, IORING_UNREGISTER_BUFFERS, NULL, 0); } int io_uring_register_files_update_tag(struct io_uring *ring, unsigned off, @@ -82,9 +79,7 @@ int io_uring_register_files_update_tag(struct io_uring *ring, unsigned off, .nr = nr_files, }; - return __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_FILES_UPDATE2, &up, - sizeof(up)); + return do_register(ring, IORING_REGISTER_FILES_UPDATE2, &up, sizeof(up)); } /* @@ -102,9 +97,7 @@ int io_uring_register_files_update(struct io_uring *ring, unsigned off, .fds = (unsigned long) files, }; - return __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_FILES_UPDATE, &up, - nr_files); + return do_register(ring, IORING_REGISTER_FILES_UPDATE, &up, nr_files); } static int increase_rlimit_nofile(unsigned nr) @@ -133,9 +126,7 @@ int io_uring_register_files_sparse(struct io_uring *ring, unsigned nr) int ret, did_increase = 0; do { - ret = __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_FILES2, ®, - sizeof(reg)); + ret = do_register(ring, IORING_REGISTER_FILES2, ®, sizeof(reg)); if (ret >= 0) break; if (ret == -EMFILE && !did_increase) { @@ -160,9 +151,7 @@ int io_uring_register_files_tags(struct io_uring *ring, const int *files, int ret, did_increase = 0; do { - ret = __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_FILES2, ®, - sizeof(reg)); + ret = do_register(ring, IORING_REGISTER_FILES2, ®, sizeof(reg)); if (ret >= 0) break; if (ret == -EMFILE && !did_increase) { @@ -182,9 +171,7 @@ int io_uring_register_files(struct io_uring *ring, const int *files, int ret, did_increase = 0; do { - ret = __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_FILES, files, - nr_files); + ret = do_register(ring, IORING_REGISTER_FILES, files, nr_files); if (ret >= 0) break; if (ret == -EMFILE && !did_increase) { @@ -200,79 +187,50 @@ int io_uring_register_files(struct io_uring *ring, const int *files, int io_uring_unregister_files(struct io_uring *ring) { - int ret; - - ret = __sys_io_uring_register(ring->ring_fd, IORING_UNREGISTER_FILES, - NULL, 0); - return (ret < 0) ? ret : 0; + return do_register(ring, IORING_UNREGISTER_FILES, NULL, 0); } int io_uring_register_eventfd(struct io_uring *ring, int event_fd) { - int ret; - - ret = __sys_io_uring_register(ring->ring_fd, IORING_REGISTER_EVENTFD, - &event_fd, 1); - return (ret < 0) ? ret : 0; + return do_register(ring, IORING_REGISTER_EVENTFD, &event_fd, 1); } int io_uring_unregister_eventfd(struct io_uring *ring) { - int ret; - - ret = __sys_io_uring_register(ring->ring_fd, IORING_UNREGISTER_EVENTFD, - NULL, 0); - return (ret < 0) ? ret : 0; + return do_register(ring, IORING_UNREGISTER_EVENTFD, NULL, 0); } int io_uring_register_eventfd_async(struct io_uring *ring, int event_fd) { - int ret; - - ret = __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_EVENTFD_ASYNC, &event_fd, - 1); - return (ret < 0) ? ret : 0; + return do_register(ring, IORING_REGISTER_EVENTFD_ASYNC, &event_fd, 1); } int io_uring_register_probe(struct io_uring *ring, struct io_uring_probe *p, unsigned int nr_ops) { - int ret; - - ret = __sys_io_uring_register(ring->ring_fd, IORING_REGISTER_PROBE, p, - nr_ops); - return (ret < 0) ? ret : 0; + return do_register(ring, IORING_REGISTER_PROBE, p, nr_ops); } int io_uring_register_personality(struct io_uring *ring) { - return __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_PERSONALITY, NULL, 0); + return do_register(ring, IORING_REGISTER_PERSONALITY, NULL, 0); } int io_uring_unregister_personality(struct io_uring *ring, int id) { - return __sys_io_uring_register(ring->ring_fd, - IORING_UNREGISTER_PERSONALITY, NULL, id); + return do_register(ring, IORING_UNREGISTER_PERSONALITY, NULL, id); } int io_uring_register_restrictions(struct io_uring *ring, struct io_uring_restriction *res, unsigned int nr_res) { - int ret; - - ret = __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_RESTRICTIONS, res, - nr_res); - return (ret < 0) ? ret : 0; + return do_register(ring, IORING_REGISTER_RESTRICTIONS, res, nr_res); } int io_uring_enable_rings(struct io_uring *ring) { - return __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_ENABLE_RINGS, NULL, 0); + return do_register(ring, IORING_REGISTER_ENABLE_RINGS, NULL, 0); } int io_uring_register_iowq_aff(struct io_uring *ring, size_t cpusz, @@ -281,21 +239,17 @@ int io_uring_register_iowq_aff(struct io_uring *ring, size_t cpusz, if (cpusz >= (1U << 31)) return -EINVAL; - return __sys_io_uring_register(ring->ring_fd, IORING_REGISTER_IOWQ_AFF, - mask, (int) cpusz); + return do_register(ring, IORING_REGISTER_IOWQ_AFF, mask, (int) cpusz); } int io_uring_unregister_iowq_aff(struct io_uring *ring) { - return __sys_io_uring_register(ring->ring_fd, - IORING_UNREGISTER_IOWQ_AFF, NULL, 0); + return do_register(ring, IORING_UNREGISTER_IOWQ_AFF, NULL, 0); } int io_uring_register_iowq_max_workers(struct io_uring *ring, unsigned int *val) { - return __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_IOWQ_MAX_WORKERS, val, - 2); + return do_register(ring, IORING_REGISTER_IOWQ_MAX_WORKERS, val, 2); } int io_uring_register_ring_fd(struct io_uring *ring) @@ -306,11 +260,16 @@ int io_uring_register_ring_fd(struct io_uring *ring) }; int ret; - ret = __sys_io_uring_register(ring->ring_fd, IORING_REGISTER_RING_FDS, - &up, 1); + if (ring->int_flags & INT_FLAG_REG_RING) + return -EEXIST; + + ret = do_register(ring, IORING_REGISTER_RING_FDS, &up, 1); if (ret == 1) { ring->enter_ring_fd = up.offset; ring->int_flags |= INT_FLAG_REG_RING; + if (ring->features & IORING_FEAT_REG_REG_RING) { + ring->int_flags |= INT_FLAG_REG_REG_RING; + } } return ret; } @@ -323,48 +282,58 @@ int io_uring_unregister_ring_fd(struct io_uring *ring) }; int ret; - ret = __sys_io_uring_register(ring->ring_fd, IORING_UNREGISTER_RING_FDS, - &up, 1); + if (!(ring->int_flags & INT_FLAG_REG_RING)) + return -EINVAL; + + ret = do_register(ring, IORING_UNREGISTER_RING_FDS, &up, 1); if (ret == 1) { ring->enter_ring_fd = ring->ring_fd; - ring->int_flags &= ~INT_FLAG_REG_RING; + ring->int_flags &= ~(INT_FLAG_REG_RING | INT_FLAG_REG_REG_RING); } return ret; } +int io_uring_close_ring_fd(struct io_uring *ring) +{ + if (!(ring->features & IORING_FEAT_REG_REG_RING)) + return -EOPNOTSUPP; + if (!(ring->int_flags & INT_FLAG_REG_RING)) + return -EINVAL; + if (ring->ring_fd == -1) + return -EBADF; + + __sys_close(ring->ring_fd); + ring->ring_fd = -1; + return 1; +} + int io_uring_register_buf_ring(struct io_uring *ring, struct io_uring_buf_reg *reg, unsigned int __maybe_unused flags) { - return __sys_io_uring_register(ring->ring_fd, IORING_REGISTER_PBUF_RING, - reg, 1); + return do_register(ring, IORING_REGISTER_PBUF_RING, reg, 1); } int io_uring_unregister_buf_ring(struct io_uring *ring, int bgid) { struct io_uring_buf_reg reg = { .bgid = bgid }; - return __sys_io_uring_register(ring->ring_fd, - IORING_UNREGISTER_PBUF_RING, ®, 1); + return do_register(ring, IORING_UNREGISTER_PBUF_RING, ®, 1); } int io_uring_register_sync_cancel(struct io_uring *ring, struct io_uring_sync_cancel_reg *reg) { - return __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_SYNC_CANCEL, reg, 1); + return do_register(ring, IORING_REGISTER_SYNC_CANCEL, reg, 1); } int io_uring_register_file_alloc_range(struct io_uring *ring, unsigned off, unsigned len) { - struct io_uring_file_index_range range; - - memset(&range, 0, sizeof(range)); - range.off = off; - range.len = len; + struct io_uring_file_index_range range = { + .off = off, + .len = len + }; - return __sys_io_uring_register(ring->ring_fd, - IORING_REGISTER_FILE_ALLOC_RANGE, &range, - 0); + return do_register(ring, IORING_REGISTER_FILE_ALLOC_RANGE, &range, 0); } diff --git a/contrib/libs/liburing/src/setup.c b/contrib/libs/liburing/src/setup.c index 0c15e75c1c4..db2f3dfe15e 100644 --- a/contrib/libs/liburing/src/setup.c +++ b/contrib/libs/liburing/src/setup.c @@ -106,6 +106,7 @@ __cold int io_uring_queue_mmap(int fd, struct io_uring_params *p, ring->flags = p->flags; ring->ring_fd = ring->enter_ring_fd = fd; ring->int_flags = 0; + return 0; } return ret; } @@ -206,7 +207,8 @@ __cold void io_uring_queue_exit(struct io_uring *ring) */ if (ring->int_flags & INT_FLAG_REG_RING) io_uring_unregister_ring_fd(ring); - __sys_close(ring->ring_fd); + if (ring->ring_fd != -1) + __sys_close(ring->ring_fd); } __cold struct io_uring_probe *io_uring_get_probe_ring(struct io_uring *ring) @@ -216,7 +218,7 @@ __cold struct io_uring_probe *io_uring_get_probe_ring(struct io_uring *ring) int r; len = sizeof(*probe) + 256 * sizeof(struct io_uring_probe_op); - probe = uring_malloc(len); + probe = malloc(len); if (!probe) return NULL; memset(probe, 0, len); @@ -225,7 +227,7 @@ __cold struct io_uring_probe *io_uring_get_probe_ring(struct io_uring *ring) if (r >= 0) return probe; - uring_free(probe); + free(probe); return NULL; } @@ -246,7 +248,7 @@ __cold struct io_uring_probe *io_uring_get_probe(void) __cold void io_uring_free_probe(struct io_uring_probe *probe) { - uring_free(probe); + free(probe); } static inline int __fls(unsigned long x) @@ -305,12 +307,14 @@ static size_t rings_size(struct io_uring_params *p, unsigned entries, __cold ssize_t io_uring_mlock_size_params(unsigned entries, struct io_uring_params *p) { - struct io_uring_params lp = { }; + struct io_uring_params lp; struct io_uring ring; unsigned cq_entries; long page_size; ssize_t ret; + memset(&lp, 0, sizeof(lp)); + /* * We only really use this inited ring to see if the kernel is newer * or not. Newer kernels don't require memlocked memory. If we fail, @@ -364,7 +368,105 @@ __cold ssize_t io_uring_mlock_size_params(unsigned entries, */ __cold ssize_t io_uring_mlock_size(unsigned entries, unsigned flags) { - struct io_uring_params p = { .flags = flags, }; + struct io_uring_params p; + memset(&p, 0, sizeof(p)); + p.flags = flags; return io_uring_mlock_size_params(entries, &p); } + +#if defined(__hppa__) +static struct io_uring_buf_ring *br_setup(struct io_uring *ring, + unsigned int nentries, int bgid, + unsigned int flags, int *ret) +{ + struct io_uring_buf_ring *br; + struct io_uring_buf_reg reg; + size_t ring_size; + off_t off; + int lret; + + memset(®, 0, sizeof(reg)); + reg.ring_entries = nentries; + reg.bgid = bgid; + reg.flags = IOU_PBUF_RING_MMAP; + + *ret = 0; + lret = io_uring_register_buf_ring(ring, ®, flags); + if (lret) { + *ret = lret; + return NULL; + } + + off = IORING_OFF_PBUF_RING | (unsigned long long) bgid << IORING_OFF_PBUF_SHIFT; + ring_size = nentries * sizeof(struct io_uring_buf); + br = __sys_mmap(NULL, ring_size, PROT_READ | PROT_WRITE, + MAP_SHARED | MAP_POPULATE, ring->ring_fd, off); + if (IS_ERR(br)) { + *ret = PTR_ERR(br); + return NULL; + } + + return br; +} +#else +static struct io_uring_buf_ring *br_setup(struct io_uring *ring, + unsigned int nentries, int bgid, + unsigned int flags, int *ret) +{ + struct io_uring_buf_ring *br; + struct io_uring_buf_reg reg; + size_t ring_size; + int lret; + + memset(®, 0, sizeof(reg)); + ring_size = nentries * sizeof(struct io_uring_buf); + br = __sys_mmap(NULL, ring_size, PROT_READ | PROT_WRITE, + MAP_ANONYMOUS | MAP_PRIVATE, -1, 0); + if (IS_ERR(br)) { + *ret = PTR_ERR(br); + return NULL; + } + + reg.ring_addr = (unsigned long) (uintptr_t) br; + reg.ring_entries = nentries; + reg.bgid = bgid; + + *ret = 0; + lret = io_uring_register_buf_ring(ring, ®, flags); + if (lret) { + __sys_munmap(br, ring_size); + *ret = lret; + br = NULL; + } + + return br; +} +#endif + +struct io_uring_buf_ring *io_uring_setup_buf_ring(struct io_uring *ring, + unsigned int nentries, + int bgid, unsigned int flags, + int *ret) +{ + struct io_uring_buf_ring *br; + + br = br_setup(ring, nentries, bgid, flags, ret); + if (br) + io_uring_buf_ring_init(br); + + return br; +} + +int io_uring_free_buf_ring(struct io_uring *ring, struct io_uring_buf_ring *br, + unsigned int nentries, int bgid) +{ + int ret; + + ret = io_uring_unregister_buf_ring(ring, bgid); + if (ret) + return ret; + + __sys_munmap(br, nentries * sizeof(struct io_uring_buf)); + return 0; +} diff --git a/contrib/libs/liburing/src/version.c b/contrib/libs/liburing/src/version.c new file mode 100644 index 00000000000..e1a01229be5 --- /dev/null +++ b/contrib/libs/liburing/src/version.c @@ -0,0 +1,22 @@ +#include "../config-host.h" +/* SPDX-License-Identifier: MIT */ + +#include "liburing.h" +#include "liburing/io_uring_version.h" + +int io_uring_major_version(void) +{ + return IO_URING_VERSION_MAJOR; +} + +int io_uring_minor_version(void) +{ + return IO_URING_VERSION_MINOR; +} + +bool io_uring_check_version(int major, int minor) +{ + return major > io_uring_major_version() || + (major == io_uring_major_version() && + minor >= io_uring_minor_version()); +} |