[5/5] io_uring: add IORING_OP_WAITID support

Message ID	20230711204352.214086-6-axboe@kernel.dk (mailing list archive)
State	New
Headers	show Return-Path: <io-uring-owner@vger.kernel.org> From: Jens Axboe <axboe@kernel.dk> To: io-uring@vger.kernel.org, linux-kernel@vger.kernel.org Cc: brauner@kernel.org, arnd@arndb.de, Jens Axboe <axboe@kernel.dk> Subject: [PATCH 5/5] io_uring: add IORING_OP_WAITID support Date: Tue, 11 Jul 2023 14:43:52 -0600 Message-Id: <20230711204352.214086-6-axboe@kernel.dk> In-Reply-To: <20230711204352.214086-1-axboe@kernel.dk> References: <20230711204352.214086-1-axboe@kernel.dk> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Precedence: bulk
Series	Add io_uring support for waitid \| expand [PATCHSET,0/5] Add io_uring support for waitid [1/5] exit: abtract out should_wake helper for child_wait_callback() [2/5] exit: move core of do_wait() into helper [3/5] exit: add kernel_waitid_prepare() helper [4/5] exit: add internal include file with helpers [5/5] io_uring: add IORING_OP_WAITID support

diff --git a/include/linux/io_uring_types.h b/include/linux/io_uring_types.h index a7f03d8d879f..598553877fc2 100644 --- a/include/linux/io_uring_types.h +++ b/include/linux/io_uring_types.h @@ -276,6 +276,8 @@ struct io_ring_ctx { struct hlist_head futex_list; struct io_alloc_cache futex_cache; + struct hlist_head waitid_list; + const struct cred *sq_creds; /* cred used for __io_sq_thread() */ struct io_sq_data *sq_data; /* if using sq thread polling */ diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h index 420f38675769..8fca2cffc343 100644 --- a/include/uapi/linux/io_uring.h +++ b/include/uapi/linux/io_uring.h @@ -66,6 +66,7 @@ struct io_uring_sqe { __u32 msg_ring_flags; __u32 uring_cmd_flags; __u32 futex_flags; + __u32 waitid_flags; }; __u64 user_data; /* data to be passed back at completion time */ /* pack this to avoid bogus arm OABI complaints */ @@ -239,6 +240,7 @@ enum io_uring_op { IORING_OP_FUTEX_WAIT, IORING_OP_FUTEX_WAKE, IORING_OP_FUTEX_WAITV, + IORING_OP_WAITID, /* this goes last, obviously */ IORING_OP_LAST, diff --git a/io_uring/Makefile b/io_uring/Makefile index 2e4779bc550c..e5be47e4fc3b 100644 --- a/io_uring/Makefile +++ b/io_uring/Makefile @@ -8,6 +8,6 @@ obj-$(CONFIG_IO_URING) += io_uring.o xattr.o nop.o fs.o splice.o \ statx.o net.o msg_ring.o timeout.o \ sqpoll.o fdinfo.o tctx.o poll.o \ cancel.o kbuf.o rsrc.o rw.o opdef.o \ - notif.o + notif.o waitid.o obj-$(CONFIG_IO_WQ) += io-wq.o obj-$(CONFIG_FUTEX) += futex.o diff --git a/io_uring/cancel.c b/io_uring/cancel.c index 3dba8ccb1cd8..a01f3f41012b 100644 --- a/io_uring/cancel.c +++ b/io_uring/cancel.c @@ -16,6 +16,7 @@ #include "poll.h" #include "timeout.h" #include "futex.h" +#include "waitid.h" #include "cancel.h" struct io_cancel { @@ -124,6 +125,10 @@ int io_try_cancel(struct io_uring_task *tctx, struct io_cancel_data *cd, if (ret != -ENOENT) return ret; + ret = io_waitid_cancel(ctx, cd, issue_flags); + if (ret != -ENOENT) + return ret; + spin_lock(&ctx->completion_lock); if (!(cd->flags & IORING_ASYNC_CANCEL_FD)) ret = io_timeout_cancel(ctx, cd); diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c index 67ff148bc394..6d99d51b84e6 100644 --- a/io_uring/io_uring.c +++ b/io_uring/io_uring.c @@ -93,6 +93,7 @@ #include "net.h" #include "notif.h" #include "futex.h" +#include "waitid.h" #include "timeout.h" #include "poll.h" @@ -336,6 +337,7 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p) ctx->submit_state.free_list.next = NULL; INIT_WQ_LIST(&ctx->locked_free_list); INIT_HLIST_HEAD(&ctx->futex_list); + INIT_HLIST_HEAD(&ctx->waitid_list); INIT_DELAYED_WORK(&ctx->fallback_work, io_fallback_req_func); INIT_WQ_LIST(&ctx->submit_state.compl_reqs); return ctx; @@ -3259,6 +3261,7 @@ static __cold bool io_uring_try_cancel_requests(struct io_ring_ctx *ctx, mutex_lock(&ctx->uring_lock); ret |= io_poll_remove_all(ctx, task, cancel_all); ret |= io_futex_remove_all(ctx, task, cancel_all); + ret |= io_waitid_remove_all(ctx, task, cancel_all); mutex_unlock(&ctx->uring_lock); ret |= io_kill_timeouts(ctx, task, cancel_all); if (task) diff --git a/io_uring/opdef.c b/io_uring/opdef.c index 2034acfe10d0..2fbdf6a6c24a 100644 --- a/io_uring/opdef.c +++ b/io_uring/opdef.c @@ -34,6 +34,7 @@ #include "cancel.h" #include "rw.h" #include "futex.h" +#include "waitid.h" static int io_no_issue(struct io_kiocb *req, unsigned int issue_flags) { @@ -453,6 +454,10 @@ const struct io_issue_def io_issue_defs[] = { .prep = io_eopnotsupp_prep, #endif }, + [IORING_OP_WAITID] = { + .prep = io_waitid_prep, + .issue = io_waitid, + }, }; const struct io_cold_def io_cold_defs[] = { @@ -681,6 +686,10 @@ const struct io_cold_def io_cold_defs[] = { [IORING_OP_FUTEX_WAITV] = { .name = "FUTEX_WAITV", }, + [IORING_OP_WAITID] = { + .name = "WAITID", + .async_size = sizeof(struct io_waitid_async), + }, }; const char *io_uring_get_opcode(u8 opcode) diff --git a/io_uring/waitid.c b/io_uring/waitid.c new file mode 100644 index 000000000000..8d6ac22113dd --- /dev/null +++ b/io_uring/waitid.c @@ -0,0 +1,271 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Support for async notification of waitid + */ +#include <linux/kernel.h> +#include <linux/errno.h> +#include <linux/fs.h> +#include <linux/file.h> +#include <linux/io_uring.h> + +#include <uapi/linux/io_uring.h> + +#include "io_uring.h" +#include "cancel.h" +#include "waitid.h" +#include "../kernel/exit.h" + +struct io_waitid { + struct file *file; + int which; + pid_t upid; + int options; + struct wait_queue_head *head; + struct siginfo __user *infop; + struct waitid_info info; +}; + +static void io_waitid_free(struct io_kiocb *req) +{ + struct io_waitid_async *iwa = req->async_data; + + put_pid(iwa->wo.wo_pid); + kfree(req->async_data); + req->async_data = NULL; + req->flags &= ~REQ_F_ASYNC_DATA; +} + +static int io_waitid_finish(struct io_kiocb *req, int ret) +{ + struct io_waitid *iw = io_kiocb_to_cmd(req, struct io_waitid); + int signo = 0; + + if (ret > 0) { + signo = SIGCHLD; + ret = 0; + } + if (!iw->infop) + goto done; + + if (!user_write_access_begin(iw->infop, sizeof(*iw->infop))) { + ret = -EFAULT; + goto done; + } + + unsafe_put_user(signo, &iw->infop->si_signo, Efault); + unsafe_put_user(0, &iw->infop->si_errno, Efault); + unsafe_put_user(iw->info.cause, &iw->infop->si_code, Efault); + unsafe_put_user(iw->info.pid, &iw->infop->si_pid, Efault); + unsafe_put_user(iw->info.uid, &iw->infop->si_uid, Efault); + unsafe_put_user(iw->info.status, &iw->infop->si_status, Efault); +done: + user_write_access_end(); + io_waitid_free(req); + return ret; +Efault: + ret = -EFAULT; + goto done; +} + +static void io_waitid_complete(struct io_kiocb *req, int ret) +{ + struct io_tw_state ts = { .locked = true }; + + lockdep_assert_held(&req->ctx->uring_lock); + + /* + * Did cancel find it meanwhile? + */ + if (hlist_unhashed(&req->hash_node)) + return; + + hlist_del_init(&req->hash_node); + + ret = io_waitid_finish(req, ret); + if (ret < 0) + req_set_fail(req); + io_req_set_res(req, ret, 0); + io_req_task_complete(req, &ts); +} + +static bool __io_waitid_cancel(struct io_ring_ctx *ctx, struct io_kiocb *req) +{ + struct io_waitid *iw = io_kiocb_to_cmd(req, struct io_waitid); + struct wait_queue_head *head; + + head = READ_ONCE(iw->head); + if (head) { + struct io_waitid_async *iwa = req->async_data; + + spin_lock_irq(&head->lock); + list_del_init(&iwa->wo.child_wait.entry); + iw->head = NULL; + spin_unlock_irq(&head->lock); + io_waitid_complete(req, -ECANCELED); + return true; + } + + return false; +} + +int io_waitid_cancel(struct io_ring_ctx *ctx, struct io_cancel_data *cd, + unsigned int issue_flags) +{ + struct hlist_node *tmp; + struct io_kiocb *req; + int nr = 0; + + if (cd->flags & (IORING_ASYNC_CANCEL_FD|IORING_ASYNC_CANCEL_FD_FIXED)) + return -ENOENT; + + io_ring_submit_lock(ctx, issue_flags); + hlist_for_each_entry_safe(req, tmp, &ctx->waitid_list, hash_node) { + if (req->cqe.user_data != cd->data && + !(cd->flags & IORING_ASYNC_CANCEL_ANY)) + continue; + if (__io_waitid_cancel(ctx, req)) + nr++; + if (!(cd->flags & IORING_ASYNC_CANCEL_ALL)) + break; + } + io_ring_submit_unlock(ctx, issue_flags); + + if (nr) + return nr; + + return -ENOENT; +} + +bool io_waitid_remove_all(struct io_ring_ctx *ctx, struct task_struct *task, + bool cancel_all) +{ + struct hlist_node *tmp; + struct io_kiocb *req; + bool found = false; + + lockdep_assert_held(&ctx->uring_lock); + + hlist_for_each_entry_safe(req, tmp, &ctx->waitid_list, hash_node) { + if (!io_match_task_safe(req, task, cancel_all)) + continue; + __io_waitid_cancel(ctx, req); + found = true; + } + + return found; +} + +static void io_waitid_cb(struct io_kiocb *req, struct io_tw_state *ts) +{ + struct io_waitid_async *iwa = req->async_data; + struct io_ring_ctx *ctx = req->ctx; + int ret; + + /* + * If we get -ERESTARTSYS here, we need to re-arm and check again + * to ensure we get another callback. If the retry works, then we can + * just remove ourselves from the waitqueue again and finish the + * request. + */ + ret = __do_wait(&iwa->wo); + if (unlikely(ret == -ERESTARTSYS)) { + struct io_waitid *iw = io_kiocb_to_cmd(req, struct io_waitid); + + io_tw_lock(ctx, ts); + iw->head = &current->signal->wait_chldexit; + add_wait_queue(iw->head, &iwa->wo.child_wait); + ret = __do_wait(&iwa->wo); + if (ret == -ERESTARTSYS) + return; + + remove_wait_queue(iw->head, &iwa->wo.child_wait); + iw->head = NULL; + } + + io_tw_lock(ctx, ts); + io_waitid_complete(req, ret); +} + +static int io_waitid_wait(struct wait_queue_entry *wait, unsigned mode, + int sync, void *key) +{ + struct wait_opts *wo = container_of(wait, struct wait_opts, child_wait); + struct io_waitid_async *iwa = container_of(wo, struct io_waitid_async, wo); + struct io_kiocb *req = iwa->req; + struct io_waitid *iw = io_kiocb_to_cmd(req, struct io_waitid); + struct task_struct *p = key; + + if (!pid_child_should_wake(wo, p)) + return 0; + + req->io_task_work.func = io_waitid_cb; + io_req_task_work_add(req); + iw->head = NULL; + list_del_init(&wait->entry); + return 1; +} + +int io_waitid_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) +{ + struct io_waitid *iw = io_kiocb_to_cmd(req, struct io_waitid); + + if (sqe->addr || sqe->buf_index || sqe->addr3 || sqe->waitid_flags) + return -EINVAL; + + iw->which = READ_ONCE(sqe->len); + iw->options = READ_ONCE(sqe->file_index); + iw->upid = READ_ONCE(sqe->fd); + iw->infop = u64_to_user_ptr(READ_ONCE(sqe->addr2)); + iw->head = NULL; + return 0; +} + +int io_waitid(struct io_kiocb *req, unsigned int issue_flags) +{ + struct io_waitid *iw = io_kiocb_to_cmd(req, struct io_waitid); + struct io_ring_ctx *ctx = req->ctx; + struct io_waitid_async *iwa; + unsigned int f_flags = 0; + int ret; + + if (io_alloc_async_data(req)) + return -ENOMEM; + + iwa = req->async_data; + iwa->req = req; + + ret = kernel_waitid_prepare(&iwa->wo, iw->which, iw->upid, &iw->info, + iw->options, NULL, &f_flags); + if (ret) + goto done; + + /* + * Arm our callback and add us to the waitqueue, in case no events + * are available. + */ + init_waitqueue_func_entry(&iwa->wo.child_wait, io_waitid_wait); + iwa->wo.child_wait.private = req->task; + iw->head = &current->signal->wait_chldexit; + add_wait_queue(iw->head, &iwa->wo.child_wait); + + io_ring_submit_lock(ctx, issue_flags); + hlist_add_head(&req->hash_node, &ctx->waitid_list); + + ret = __do_wait(&iwa->wo); + if (ret == -ERESTARTSYS) { + io_ring_submit_unlock(ctx, issue_flags); + return IOU_ISSUE_SKIP_COMPLETE; + } + + hlist_del_init(&req->hash_node); + remove_wait_queue(iw->head, &iwa->wo.child_wait); + iw->head = NULL; + ret = io_waitid_finish(req, ret); + + io_ring_submit_unlock(ctx, issue_flags); +done: + if (ret < 0) + req_set_fail(req); + io_req_set_res(req, ret, 0); + return IOU_OK; +} diff --git a/io_uring/waitid.h b/io_uring/waitid.h new file mode 100644 index 000000000000..956a8adafe8c --- /dev/null +++ b/io_uring/waitid.h @@ -0,0 +1,15 @@ +// SPDX-License-Identifier: GPL-2.0 + +#include "../kernel/exit.h" + +struct io_waitid_async { + struct io_kiocb *req; + struct wait_opts wo; +}; + +int io_waitid_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe); +int io_waitid(struct io_kiocb *req, unsigned int issue_flags); +int io_waitid_cancel(struct io_ring_ctx *ctx, struct io_cancel_data *cd, + unsigned int issue_flags); +bool io_waitid_remove_all(struct io_ring_ctx *ctx, struct task_struct *task, + bool cancel_all);

[5/5] io_uring: add IORING_OP_WAITID support

Commit Message

Comments

Patch