summaryrefslogtreecommitdiff
path: root/io_uring
diff options
context:
space:
mode:
Diffstat (limited to 'io_uring')
-rw-r--r--io_uring/fdinfo.c16
-rw-r--r--io_uring/io-wq.c4
-rw-r--r--io_uring/io_uring.c91
-rw-r--r--io_uring/kbuf.c8
-rw-r--r--io_uring/kbuf.h1
-rw-r--r--io_uring/msg_ring.c1
-rw-r--r--io_uring/net.c46
-rw-r--r--io_uring/refs.h7
-rw-r--r--io_uring/register.c7
-rw-r--r--io_uring/rsrc.c23
-rw-r--r--io_uring/rsrc.h1
-rw-r--r--io_uring/sqpoll.c50
-rw-r--r--io_uring/sqpoll.h8
13 files changed, 171 insertions, 92 deletions
diff --git a/io_uring/fdinfo.c b/io_uring/fdinfo.c
index 6b1247664b35..c6c624eb9866 100644
--- a/io_uring/fdinfo.c
+++ b/io_uring/fdinfo.c
@@ -83,11 +83,11 @@ __cold void io_uring_show_fdinfo(struct seq_file *m, struct file *file)
seq_printf(m, "SqMask:\t0x%x\n", sq_mask);
seq_printf(m, "SqHead:\t%u\n", sq_head);
seq_printf(m, "SqTail:\t%u\n", sq_tail);
- seq_printf(m, "CachedSqHead:\t%u\n", ctx->cached_sq_head);
+ seq_printf(m, "CachedSqHead:\t%u\n", data_race(ctx->cached_sq_head));
seq_printf(m, "CqMask:\t0x%x\n", cq_mask);
seq_printf(m, "CqHead:\t%u\n", cq_head);
seq_printf(m, "CqTail:\t%u\n", cq_tail);
- seq_printf(m, "CachedCqTail:\t%u\n", ctx->cached_cq_tail);
+ seq_printf(m, "CachedCqTail:\t%u\n", data_race(ctx->cached_cq_tail));
seq_printf(m, "SQEs:\t%u\n", sq_tail - sq_head);
sq_entries = min(sq_tail - sq_head, ctx->sq_entries);
for (i = 0; i < sq_entries; i++) {
@@ -146,18 +146,26 @@ __cold void io_uring_show_fdinfo(struct seq_file *m, struct file *file)
if (has_lock && (ctx->flags & IORING_SETUP_SQPOLL)) {
struct io_sq_data *sq = ctx->sq_data;
+ struct task_struct *tsk;
+ rcu_read_lock();
+ tsk = rcu_dereference(sq->thread);
/*
* sq->thread might be NULL if we raced with the sqpoll
* thread termination.
*/
- if (sq->thread) {
+ if (tsk) {
+ get_task_struct(tsk);
+ rcu_read_unlock();
+ getrusage(tsk, RUSAGE_SELF, &sq_usage);
+ put_task_struct(tsk);
sq_pid = sq->task_pid;
sq_cpu = sq->sq_cpu;
- getrusage(sq->thread, RUSAGE_SELF, &sq_usage);
sq_total_time = (sq_usage.ru_stime.tv_sec * 1000000
+ sq_usage.ru_stime.tv_usec);
sq_work_time = sq->work_time;
+ } else {
+ rcu_read_unlock();
}
}
diff --git a/io_uring/io-wq.c b/io_uring/io-wq.c
index a2d577b09930..8f555c1d7185 100644
--- a/io_uring/io-wq.c
+++ b/io_uring/io-wq.c
@@ -1204,8 +1204,10 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
atomic_set(&wq->worker_refs, 1);
init_completion(&wq->worker_done);
ret = cpuhp_state_add_instance_nocalls(io_wq_online, &wq->cpuhp_node);
- if (ret)
+ if (ret) {
+ put_task_struct(wq->task);
goto err;
+ }
return wq;
err:
diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index 19de7129ae0b..52ada466bf98 100644
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -441,24 +441,6 @@ static struct io_kiocb *__io_prep_linked_timeout(struct io_kiocb *req)
return req->link;
}
-static inline struct io_kiocb *io_prep_linked_timeout(struct io_kiocb *req)
-{
- if (likely(!(req->flags & REQ_F_ARM_LTIMEOUT)))
- return NULL;
- return __io_prep_linked_timeout(req);
-}
-
-static noinline void __io_arm_ltimeout(struct io_kiocb *req)
-{
- io_queue_linked_timeout(__io_prep_linked_timeout(req));
-}
-
-static inline void io_arm_ltimeout(struct io_kiocb *req)
-{
- if (unlikely(req->flags & REQ_F_ARM_LTIMEOUT))
- __io_arm_ltimeout(req);
-}
-
static void io_prep_async_work(struct io_kiocb *req)
{
const struct io_issue_def *def = &io_issue_defs[req->opcode];
@@ -511,7 +493,6 @@ static void io_prep_async_link(struct io_kiocb *req)
static void io_queue_iowq(struct io_kiocb *req)
{
- struct io_kiocb *link = io_prep_linked_timeout(req);
struct io_uring_task *tctx = req->task->io_uring;
BUG_ON(!tctx);
@@ -536,8 +517,6 @@ static void io_queue_iowq(struct io_kiocb *req)
trace_io_uring_queue_async_work(req, io_wq_is_hashed(&req->work));
io_wq_enqueue(tctx->io_wq, &req->work);
- if (link)
- io_queue_linked_timeout(link);
}
static void io_req_queue_iowq_tw(struct io_kiocb *req, struct io_tw_state *ts)
@@ -651,6 +630,7 @@ static void __io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool dying)
* to care for a non-real case.
*/
if (need_resched()) {
+ ctx->cqe_sentinel = ctx->cqe_cached;
io_cq_unlock_post(ctx);
mutex_unlock(&ctx->uring_lock);
cond_resched();
@@ -884,13 +864,26 @@ bool io_req_post_cqe(struct io_kiocb *req, s32 res, u32 cflags)
struct io_ring_ctx *ctx = req->ctx;
bool posted;
+ /*
+ * If multishot has already posted deferred completions, ensure that
+ * those are flushed first before posting this one. If not, CQEs
+ * could get reordered.
+ */
+ if (!wq_list_empty(&ctx->submit_state.compl_reqs))
+ __io_submit_flush_completions(ctx);
+
lockdep_assert(!io_wq_current_is_worker());
lockdep_assert_held(&ctx->uring_lock);
- __io_cq_lock(ctx);
- posted = io_fill_cqe_aux(ctx, req->cqe.user_data, res, cflags);
+ if (!ctx->lockless_cq) {
+ spin_lock(&ctx->completion_lock);
+ posted = io_fill_cqe_aux(ctx, req->cqe.user_data, res, cflags);
+ spin_unlock(&ctx->completion_lock);
+ } else {
+ posted = io_fill_cqe_aux(ctx, req->cqe.user_data, res, cflags);
+ }
+
ctx->submit_state.cq_flush = true;
- __io_cq_unlock_post(ctx);
return posted;
}
@@ -1103,21 +1096,22 @@ static __cold void io_fallback_tw(struct io_uring_task *tctx, bool sync)
while (node) {
req = container_of(node, struct io_kiocb, io_task_work.node);
node = node->next;
- if (sync && last_ctx != req->ctx) {
+ if (last_ctx != req->ctx) {
if (last_ctx) {
- flush_delayed_work(&last_ctx->fallback_work);
+ if (sync)
+ flush_delayed_work(&last_ctx->fallback_work);
percpu_ref_put(&last_ctx->refs);
}
last_ctx = req->ctx;
percpu_ref_get(&last_ctx->refs);
}
- if (llist_add(&req->io_task_work.node,
- &req->ctx->fallback_llist))
- schedule_delayed_work(&req->ctx->fallback_work, 1);
+ if (llist_add(&req->io_task_work.node, &last_ctx->fallback_llist))
+ schedule_delayed_work(&last_ctx->fallback_work, 1);
}
if (last_ctx) {
- flush_delayed_work(&last_ctx->fallback_work);
+ if (sync)
+ flush_delayed_work(&last_ctx->fallback_work);
percpu_ref_put(&last_ctx->refs);
}
}
@@ -1687,7 +1681,7 @@ queue:
spin_unlock(&ctx->completion_lock);
io_prep_async_link(req);
- de = kmalloc(sizeof(*de), GFP_KERNEL);
+ de = kmalloc(sizeof(*de), GFP_KERNEL_ACCOUNT);
if (!de) {
ret = -ENOMEM;
io_req_defer_failed(req, ret);
@@ -1722,17 +1716,24 @@ static bool io_assign_file(struct io_kiocb *req, const struct io_issue_def *def,
return !!req->file;
}
+#define REQ_ISSUE_SLOW_FLAGS (REQ_F_CREDS | REQ_F_ARM_LTIMEOUT)
+
static int io_issue_sqe(struct io_kiocb *req, unsigned int issue_flags)
{
const struct io_issue_def *def = &io_issue_defs[req->opcode];
const struct cred *creds = NULL;
+ struct io_kiocb *link = NULL;
int ret;
if (unlikely(!io_assign_file(req, def, issue_flags)))
return -EBADF;
- if (unlikely((req->flags & REQ_F_CREDS) && req->creds != current_cred()))
- creds = override_creds(req->creds);
+ if (unlikely(req->flags & REQ_ISSUE_SLOW_FLAGS)) {
+ if ((req->flags & REQ_F_CREDS) && req->creds != current_cred())
+ creds = override_creds(req->creds);
+ if (req->flags & REQ_F_ARM_LTIMEOUT)
+ link = __io_prep_linked_timeout(req);
+ }
if (!def->audit_skip)
audit_uring_entry(req->opcode);
@@ -1742,8 +1743,12 @@ static int io_issue_sqe(struct io_kiocb *req, unsigned int issue_flags)
if (!def->audit_skip)
audit_uring_exit(!ret, ret);
- if (creds)
- revert_creds(creds);
+ if (unlikely(creds || link)) {
+ if (creds)
+ revert_creds(creds);
+ if (link)
+ io_queue_linked_timeout(link);
+ }
if (ret == IOU_OK) {
if (issue_flags & IO_URING_F_COMPLETE_DEFER)
@@ -1756,7 +1761,6 @@ static int io_issue_sqe(struct io_kiocb *req, unsigned int issue_flags)
if (ret == IOU_ISSUE_SKIP_COMPLETE) {
ret = 0;
- io_arm_ltimeout(req);
/* If the op doesn't have a file, we're not polling for it */
if ((req->ctx->flags & IORING_SETUP_IOPOLL) && def->iopoll_queue)
@@ -1777,7 +1781,7 @@ struct io_wq_work *io_wq_free_work(struct io_wq_work *work)
struct io_kiocb *req = container_of(work, struct io_kiocb, work);
struct io_kiocb *nxt = NULL;
- if (req_ref_put_and_test(req)) {
+ if (req_ref_put_and_test_atomic(req)) {
if (req->flags & IO_REQ_LINK_FLAGS)
nxt = io_req_find_next(req);
io_free_req(req);
@@ -1799,8 +1803,6 @@ void io_wq_submit_work(struct io_wq_work *work)
else
req_ref_get(req);
- io_arm_ltimeout(req);
-
/* either cancelled or io-wq is dying, so don't touch tctx->iowq */
if (atomic_read(&work->flags) & IO_WQ_WORK_CANCEL) {
fail:
@@ -1920,15 +1922,11 @@ struct file *io_file_get_normal(struct io_kiocb *req, int fd)
static void io_queue_async(struct io_kiocb *req, int ret)
__must_hold(&req->ctx->uring_lock)
{
- struct io_kiocb *linked_timeout;
-
if (ret != -EAGAIN || (req->flags & REQ_F_NOWAIT)) {
io_req_defer_failed(req, ret);
return;
}
- linked_timeout = io_prep_linked_timeout(req);
-
switch (io_arm_poll_handler(req, 0)) {
case IO_APOLL_READY:
io_kbuf_recycle(req, 0);
@@ -1941,9 +1939,6 @@ static void io_queue_async(struct io_kiocb *req, int ret)
case IO_APOLL_OK:
break;
}
-
- if (linked_timeout)
- io_queue_linked_timeout(linked_timeout);
}
static inline void io_queue_sqe(struct io_kiocb *req)
@@ -2921,7 +2916,7 @@ static __cold void io_ring_exit_work(struct work_struct *work)
struct task_struct *tsk;
io_sq_thread_park(sqd);
- tsk = sqd->thread;
+ tsk = sqpoll_task_locked(sqd);
if (tsk && tsk->io_uring && tsk->io_uring->io_wq)
io_wq_cancel_cb(tsk->io_uring->io_wq,
io_cancel_ctx_cb, ctx, true);
@@ -3158,7 +3153,7 @@ __cold void io_uring_cancel_generic(bool cancel_all, struct io_sq_data *sqd)
s64 inflight;
DEFINE_WAIT(wait);
- WARN_ON_ONCE(sqd && sqd->thread != current);
+ WARN_ON_ONCE(sqd && sqpoll_task_locked(sqd) != current);
if (!current->io_uring)
return;
diff --git a/io_uring/kbuf.c b/io_uring/kbuf.c
index 7a8c3a004800..9bd27deeee6f 100644
--- a/io_uring/kbuf.c
+++ b/io_uring/kbuf.c
@@ -262,8 +262,12 @@ static int io_ring_buffers_peek(struct io_kiocb *req, struct buf_sel_arg *arg,
/* truncate end piece, if needed, for non partial buffers */
if (len > arg->max_len) {
len = arg->max_len;
- if (!(bl->flags & IOBL_INC))
+ if (!(bl->flags & IOBL_INC)) {
+ arg->partial_map = 1;
+ if (iov != arg->iovs)
+ break;
buf->len = len;
+ }
}
iov->iov_base = u64_to_user_ptr(buf->addr);
@@ -728,7 +732,7 @@ int io_register_pbuf_ring(struct io_ring_ctx *ctx, void __user *arg)
io_destroy_bl(ctx, bl);
}
- free_bl = bl = kzalloc(sizeof(*bl), GFP_KERNEL);
+ free_bl = bl = kzalloc(sizeof(*bl), GFP_KERNEL_ACCOUNT);
if (!bl)
return -ENOMEM;
diff --git a/io_uring/kbuf.h b/io_uring/kbuf.h
index 36aadfe5ac00..2586a292dfb9 100644
--- a/io_uring/kbuf.h
+++ b/io_uring/kbuf.h
@@ -61,6 +61,7 @@ struct buf_sel_arg {
size_t max_len;
unsigned short nr_iovs;
unsigned short mode;
+ unsigned short partial_map;
};
void __user *io_buffer_select(struct io_kiocb *req, size_t *len,
diff --git a/io_uring/msg_ring.c b/io_uring/msg_ring.c
index 7fd9badcfaf8..35b1b585e9cb 100644
--- a/io_uring/msg_ring.c
+++ b/io_uring/msg_ring.c
@@ -94,6 +94,7 @@ static int io_msg_remote_post(struct io_ring_ctx *ctx, struct io_kiocb *req,
kmem_cache_free(req_cachep, req);
return -EOWNERDEAD;
}
+ req->opcode = IORING_OP_NOP;
req->cqe.user_data = user_data;
io_req_set_res(req, res, cflags);
percpu_ref_get(&ctx->refs);
diff --git a/io_uring/net.c b/io_uring/net.c
index 384915d931b7..0116cfaec848 100644
--- a/io_uring/net.c
+++ b/io_uring/net.c
@@ -76,12 +76,18 @@ struct io_sr_msg {
/* initialised and used only by !msg send variants */
u16 addr_len;
u16 buf_group;
+ unsigned short retry_flags;
void __user *addr;
void __user *msg_control;
/* used only for send zerocopy */
struct io_kiocb *notif;
};
+enum sr_retry_flags {
+ IO_SR_MSG_RETRY = 1,
+ IO_SR_MSG_PARTIAL_MAP = 2,
+};
+
/*
* Number of times we'll try and do receives if there's more data. If we
* exceed this limit, then add us to the back of the queue and retry from
@@ -203,6 +209,7 @@ static inline void io_mshot_prep_retry(struct io_kiocb *req,
req->flags &= ~REQ_F_BL_EMPTY;
sr->done_io = 0;
+ sr->retry_flags = 0;
sr->len = 0; /* get from the provided buffer */
req->buf_index = sr->buf_group;
}
@@ -409,6 +416,7 @@ int io_sendmsg_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
struct io_sr_msg *sr = io_kiocb_to_cmd(req, struct io_sr_msg);
sr->done_io = 0;
+ sr->retry_flags = 0;
if (req->opcode == IORING_OP_SEND) {
if (READ_ONCE(sqe->__pad3[0]))
@@ -780,6 +788,7 @@ int io_recvmsg_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
struct io_sr_msg *sr = io_kiocb_to_cmd(req, struct io_sr_msg);
sr->done_io = 0;
+ sr->retry_flags = 0;
if (unlikely(sqe->file_index || sqe->addr2))
return -EINVAL;
@@ -828,6 +837,9 @@ int io_recvmsg_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
return io_recvmsg_prep_setup(req);
}
+/* bits to clear in old and inherit in new cflags on bundle retry */
+#define CQE_F_MASK (IORING_CQE_F_SOCK_NONEMPTY|IORING_CQE_F_MORE)
+
/*
* Finishes io_recv and io_recvmsg.
*
@@ -845,11 +857,27 @@ static inline bool io_recv_finish(struct io_kiocb *req, int *ret,
cflags |= IORING_CQE_F_SOCK_NONEMPTY;
if (sr->flags & IORING_RECVSEND_BUNDLE) {
- cflags |= io_put_kbufs(req, *ret, io_bundle_nbufs(kmsg, *ret),
+ size_t this_ret = *ret - sr->done_io;
+
+ cflags |= io_put_kbufs(req, this_ret, io_bundle_nbufs(kmsg, this_ret),
issue_flags);
+ if (sr->retry_flags & IO_SR_MSG_RETRY)
+ cflags = req->cqe.flags | (cflags & CQE_F_MASK);
/* bundle with no more immediate buffers, we're done */
if (req->flags & REQ_F_BL_EMPTY)
goto finish;
+ /*
+ * If more is available AND it was a full transfer, retry and
+ * append to this one
+ */
+ if (!sr->retry_flags && kmsg->msg.msg_inq > 1 && this_ret > 0 &&
+ !iov_iter_count(&kmsg->msg.msg_iter)) {
+ req->cqe.flags = cflags & ~CQE_F_MASK;
+ sr->len = kmsg->msg.msg_inq;
+ sr->done_io += this_ret;
+ sr->retry_flags |= IO_SR_MSG_RETRY;
+ return false;
+ }
} else {
cflags |= io_put_kbuf(req, *ret, issue_flags);
}
@@ -1088,13 +1116,21 @@ static int io_recv_buf_select(struct io_kiocb *req, struct io_async_msghdr *kmsg
arg.mode |= KBUF_MODE_FREE;
}
- if (kmsg->msg.msg_inq > 0)
+ if (kmsg->msg.msg_inq > 1)
arg.max_len = min_not_zero(sr->len, kmsg->msg.msg_inq);
ret = io_buffers_peek(req, &arg);
if (unlikely(ret < 0))
return ret;
+ if (arg.iovs != &kmsg->fast_iov && arg.iovs != kmsg->free_iov) {
+ kmsg->free_iov_nr = ret;
+ kmsg->free_iov = arg.iovs;
+ req->flags |= REQ_F_NEED_CLEANUP;
+ }
+ if (arg.partial_map)
+ sr->retry_flags |= IO_SR_MSG_PARTIAL_MAP;
+
/* special case 1 vec, can be a fast path */
if (ret == 1) {
sr->buf = arg.iovs[0].iov_base;
@@ -1103,11 +1139,6 @@ static int io_recv_buf_select(struct io_kiocb *req, struct io_async_msghdr *kmsg
}
iov_iter_init(&kmsg->msg.msg_iter, ITER_DEST, arg.iovs, ret,
arg.out_len);
- if (arg.iovs != &kmsg->fast_iov && arg.iovs != kmsg->free_iov) {
- kmsg->free_iov_nr = ret;
- kmsg->free_iov = arg.iovs;
- req->flags |= REQ_F_NEED_CLEANUP;
- }
} else {
void __user *buf;
@@ -1228,6 +1259,7 @@ int io_send_zc_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
struct io_kiocb *notif;
zc->done_io = 0;
+ zc->retry_flags = 0;
req->flags |= REQ_F_POLL_NO_LAZY;
if (unlikely(READ_ONCE(sqe->__pad2[0]) || READ_ONCE(sqe->addr3)))
diff --git a/io_uring/refs.h b/io_uring/refs.h
index 63982ead9f7d..0d928d87c4ed 100644
--- a/io_uring/refs.h
+++ b/io_uring/refs.h
@@ -17,6 +17,13 @@ static inline bool req_ref_inc_not_zero(struct io_kiocb *req)
return atomic_inc_not_zero(&req->refs);
}
+static inline bool req_ref_put_and_test_atomic(struct io_kiocb *req)
+{
+ WARN_ON_ONCE(!(data_race(req->flags) & REQ_F_REFCOUNT));
+ WARN_ON_ONCE(req_ref_zero_or_close_to_overflow(req));
+ return atomic_dec_and_test(&req->refs);
+}
+
static inline bool req_ref_put_and_test(struct io_kiocb *req)
{
if (likely(!(req->flags & REQ_F_REFCOUNT)))
diff --git a/io_uring/register.c b/io_uring/register.c
index eca26d4884d9..a325b493ae12 100644
--- a/io_uring/register.c
+++ b/io_uring/register.c
@@ -268,6 +268,8 @@ static __cold int io_register_iowq_max_workers(struct io_ring_ctx *ctx,
if (ctx->flags & IORING_SETUP_SQPOLL) {
sqd = ctx->sq_data;
if (sqd) {
+ struct task_struct *tsk;
+
/*
* Observe the correct sqd->lock -> ctx->uring_lock
* ordering. Fine to drop uring_lock here, we hold
@@ -277,8 +279,9 @@ static __cold int io_register_iowq_max_workers(struct io_ring_ctx *ctx,
mutex_unlock(&ctx->uring_lock);
mutex_lock(&sqd->lock);
mutex_lock(&ctx->uring_lock);
- if (sqd->thread)
- tctx = sqd->thread->io_uring;
+ tsk = sqpoll_task_locked(sqd);
+ if (tsk)
+ tctx = tsk->io_uring;
}
} else {
tctx = current->io_uring;
diff --git a/io_uring/rsrc.c b/io_uring/rsrc.c
index a67bae350416..1687e35e21c9 100644
--- a/io_uring/rsrc.c
+++ b/io_uring/rsrc.c
@@ -119,8 +119,11 @@ static void io_buffer_unmap(struct io_ring_ctx *ctx, struct io_mapped_ubuf **slo
if (imu != &dummy_ubuf) {
if (!refcount_dec_and_test(&imu->refs))
return;
- for (i = 0; i < imu->nr_bvecs; i++)
- unpin_user_page(imu->bvec[i].bv_page);
+ for (i = 0; i < imu->nr_bvecs; i++) {
+ struct folio *folio = page_folio(imu->bvec[i].bv_page);
+
+ unpin_user_folio(folio, 1);
+ }
if (imu->acct_pages)
io_unaccount_mem(ctx, imu->acct_pages);
kvfree(imu);
@@ -915,6 +918,7 @@ static bool io_try_coalesce_buffer(struct page ***pages, int *nr_pages,
return false;
data->folio_shift = folio_shift(folio);
+ data->first_folio_page_idx = folio_page_idx(folio, page_array[0]);
/*
* Check if pages are contiguous inside a folio, and all folios have
* the same page count except for the head and tail.
@@ -983,10 +987,8 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, struct iovec *iov,
goto done;
ret = io_buffer_account_pin(ctx, pages, nr_pages, imu, last_hpage);
- if (ret) {
- unpin_user_pages(pages, nr_pages);
+ if (ret)
goto done;
- }
size = iov->iov_len;
/* store original address for later verification */
@@ -997,7 +999,9 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, struct iovec *iov,
if (coalesced)
imu->folio_shift = data.folio_shift;
refcount_set(&imu->refs, 1);
- off = (unsigned long) iov->iov_base & ((1UL << imu->folio_shift) - 1);
+ off = (unsigned long)iov->iov_base & ~PAGE_MASK;
+ if (coalesced)
+ off += data.first_folio_page_idx << PAGE_SHIFT;
*pimu = imu;
ret = 0;
@@ -1010,8 +1014,13 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, struct iovec *iov,
size -= vec_len;
}
done:
- if (ret)
+ if (ret) {
kvfree(imu);
+ if (pages) {
+ for (i = 0; i < nr_pages; i++)
+ unpin_user_folio(page_folio(pages[i]), 1);
+ }
+ }
kvfree(pages);
return ret;
}
diff --git a/io_uring/rsrc.h b/io_uring/rsrc.h
index 8ed588036210..459cf4c6e856 100644
--- a/io_uring/rsrc.h
+++ b/io_uring/rsrc.h
@@ -56,6 +56,7 @@ struct io_imu_folio_data {
/* For non-head/tail folios, has to be fully included */
unsigned int nr_pages_mid;
unsigned int folio_shift;
+ unsigned long first_folio_page_idx;
};
void io_rsrc_node_ref_zero(struct io_rsrc_node *node);
diff --git a/io_uring/sqpoll.c b/io_uring/sqpoll.c
index 5bc54c6df20f..2faa3058b2d0 100644
--- a/io_uring/sqpoll.c
+++ b/io_uring/sqpoll.c
@@ -20,7 +20,7 @@
#include "sqpoll.h"
#define IORING_SQPOLL_CAP_ENTRIES_VALUE 8
-#define IORING_TW_CAP_ENTRIES_VALUE 8
+#define IORING_TW_CAP_ENTRIES_VALUE 32
enum {
IO_SQ_THREAD_SHOULD_STOP = 0,
@@ -30,7 +30,7 @@ enum {
void io_sq_thread_unpark(struct io_sq_data *sqd)
__releases(&sqd->lock)
{
- WARN_ON_ONCE(sqd->thread == current);
+ WARN_ON_ONCE(sqpoll_task_locked(sqd) == current);
/*
* Do the dance but not conditional clear_bit() because it'd race with
@@ -45,24 +45,32 @@ void io_sq_thread_unpark(struct io_sq_data *sqd)
void io_sq_thread_park(struct io_sq_data *sqd)
__acquires(&sqd->lock)
{
- WARN_ON_ONCE(data_race(sqd->thread) == current);
+ struct task_struct *tsk;
atomic_inc(&sqd->park_pending);
set_bit(IO_SQ_THREAD_SHOULD_PARK, &sqd->state);
mutex_lock(&sqd->lock);
- if (sqd->thread)
- wake_up_process(sqd->thread);
+
+ tsk = sqpoll_task_locked(sqd);
+ if (tsk) {
+ WARN_ON_ONCE(tsk == current);
+ wake_up_process(tsk);
+ }
}
void io_sq_thread_stop(struct io_sq_data *sqd)
{
- WARN_ON_ONCE(sqd->thread == current);
+ struct task_struct *tsk;
+
WARN_ON_ONCE(test_bit(IO_SQ_THREAD_SHOULD_STOP, &sqd->state));
set_bit(IO_SQ_THREAD_SHOULD_STOP, &sqd->state);
mutex_lock(&sqd->lock);
- if (sqd->thread)
- wake_up_process(sqd->thread);
+ tsk = sqpoll_task_locked(sqd);
+ if (tsk) {
+ WARN_ON_ONCE(tsk == current);
+ wake_up_process(tsk);
+ }
mutex_unlock(&sqd->lock);
wait_for_completion(&sqd->exited);
}
@@ -277,7 +285,8 @@ static int io_sq_thread(void *data)
/* offload context creation failed, just exit */
if (!current->io_uring) {
mutex_lock(&sqd->lock);
- sqd->thread = NULL;
+ rcu_assign_pointer(sqd->thread, NULL);
+ put_task_struct(current);
mutex_unlock(&sqd->lock);
goto err_out;
}
@@ -386,7 +395,8 @@ static int io_sq_thread(void *data)
io_sq_tw(&retry_list, UINT_MAX);
io_uring_cancel_generic(true, sqd);
- sqd->thread = NULL;
+ rcu_assign_pointer(sqd->thread, NULL);
+ put_task_struct(current);
list_for_each_entry(ctx, &sqd->ctx_list, sqd_list)
atomic_or(IORING_SQ_NEED_WAKEUP, &ctx->rings->sq_flags);
io_run_task_work();
@@ -416,7 +426,6 @@ void io_sqpoll_wait_sq(struct io_ring_ctx *ctx)
__cold int io_sq_offload_create(struct io_ring_ctx *ctx,
struct io_uring_params *p)
{
- struct task_struct *task_to_put = NULL;
int ret;
/* Retain compatibility with failing for an invalid attach attempt */
@@ -496,8 +505,11 @@ __cold int io_sq_offload_create(struct io_ring_ctx *ctx,
goto err_sqpoll;
}
- sqd->thread = tsk;
- task_to_put = get_task_struct(tsk);
+ mutex_lock(&sqd->lock);
+ rcu_assign_pointer(sqd->thread, tsk);
+ mutex_unlock(&sqd->lock);
+
+ get_task_struct(tsk);
ret = io_uring_alloc_task_context(tsk, ctx);
wake_up_new_task(tsk);
if (ret)
@@ -507,16 +519,11 @@ __cold int io_sq_offload_create(struct io_ring_ctx *ctx,
ret = -EINVAL;
goto err;
}
-
- if (task_to_put)
- put_task_struct(task_to_put);
return 0;
err_sqpoll:
complete(&ctx->sq_data->exited);
err:
io_sq_thread_finish(ctx);
- if (task_to_put)
- put_task_struct(task_to_put);
return ret;
}
@@ -527,10 +534,13 @@ __cold int io_sqpoll_wq_cpu_affinity(struct io_ring_ctx *ctx,
int ret = -EINVAL;
if (sqd) {
+ struct task_struct *tsk;
+
io_sq_thread_park(sqd);
/* Don't set affinity for a dying thread */
- if (sqd->thread)
- ret = io_wq_cpu_affinity(sqd->thread->io_uring, mask);
+ tsk = sqpoll_task_locked(sqd);
+ if (tsk)
+ ret = io_wq_cpu_affinity(tsk->io_uring, mask);
io_sq_thread_unpark(sqd);
}
diff --git a/io_uring/sqpoll.h b/io_uring/sqpoll.h
index 4171666b1cf4..b83dcdec9765 100644
--- a/io_uring/sqpoll.h
+++ b/io_uring/sqpoll.h
@@ -8,7 +8,7 @@ struct io_sq_data {
/* ctx's that are using this sqd */
struct list_head ctx_list;
- struct task_struct *thread;
+ struct task_struct __rcu *thread;
struct wait_queue_head wait;
unsigned sq_thread_idle;
@@ -29,3 +29,9 @@ void io_sq_thread_unpark(struct io_sq_data *sqd);
void io_put_sq_data(struct io_sq_data *sqd);
void io_sqpoll_wait_sq(struct io_ring_ctx *ctx);
int io_sqpoll_wq_cpu_affinity(struct io_ring_ctx *ctx, cpumask_var_t mask);
+
+static inline struct task_struct *sqpoll_task_locked(struct io_sq_data *sqd)
+{
+ return rcu_dereference_protected(sqd->thread,
+ lockdep_is_held(&sqd->lock));
+}