summaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2019-04-20 12:20:58 -0700
committerLinus Torvalds <torvalds@linux-foundation.org>2019-04-20 12:20:58 -0700
commit38a2ca2cac28cd202a364e30f9fd252ac66619be (patch)
treeadeccbd11f5f62a3ddf473075ad04bac27d9cf26 /fs
parent34396bdfd2b7b41b9dcd2126eb1a57e43fe13368 (diff)
parent6bedf00e55e5dd0a4ed1ad3f06131edd6fb56ec8 (diff)
downloadlinux-38a2ca2cac28cd202a364e30f9fd252ac66619be.tar.bz2
Merge tag 'for-linus-20190420' of git://git.kernel.dk/linux-block
Pull block fixes from Jens Axboe: "A set of small fixes that should go into this series. This contains: - Removal of unused queue member (Hou) - Overflow bvec fix (Ming) - Various little io_uring tweaks (me) - kthread parking - Only call cpu_possible() for verified CPU - Drop unused 'file' argument to io_file_put() - io_uring_enter vs io_uring_register deadlock fix - CQ overflow fix - BFQ internal depth update fix (me)" * tag 'for-linus-20190420' of git://git.kernel.dk/linux-block: block: make sure that bvec length can't be overflow block: kill all_q_node in request_queue io_uring: fix CQ overflow condition io_uring: fix possible deadlock between io_uring_{enter,register} io_uring: drop io_file_put() 'file' argument bfq: update internal depth state when queue depth changes io_uring: only test SQPOLL cpu after we've verified it io_uring: park SQPOLL thread if it's percpu
Diffstat (limited to 'fs')
-rw-r--r--fs/io_uring.c41
1 files changed, 30 insertions, 11 deletions
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 89aa8412b5f5..f65f85d89217 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -338,7 +338,7 @@ static struct io_uring_cqe *io_get_cqring(struct io_ring_ctx *ctx)
tail = ctx->cached_cq_tail;
/* See comment at the top of the file */
smp_rmb();
- if (tail + 1 == READ_ONCE(ring->r.head))
+ if (tail - READ_ONCE(ring->r.head) == ring->ring_entries)
return NULL;
ctx->cached_cq_tail++;
@@ -682,11 +682,9 @@ static void io_iopoll_req_issued(struct io_kiocb *req)
list_add_tail(&req->list, &ctx->poll_list);
}
-static void io_file_put(struct io_submit_state *state, struct file *file)
+static void io_file_put(struct io_submit_state *state)
{
- if (!state) {
- fput(file);
- } else if (state->file) {
+ if (state->file) {
int diff = state->has_refs - state->used_refs;
if (diff)
@@ -711,7 +709,7 @@ static struct file *io_file_get(struct io_submit_state *state, int fd)
state->ios_left--;
return state->file;
}
- io_file_put(state, NULL);
+ io_file_put(state);
}
state->file = fget_many(fd, state->ios_left);
if (!state->file)
@@ -1671,7 +1669,7 @@ out:
static void io_submit_state_end(struct io_submit_state *state)
{
blk_finish_plug(&state->plug);
- io_file_put(state, NULL);
+ io_file_put(state);
if (state->free_reqs)
kmem_cache_free_bulk(req_cachep, state->free_reqs,
&state->reqs[state->cur_req]);
@@ -1920,6 +1918,10 @@ static int io_sq_thread(void *data)
unuse_mm(cur_mm);
mmput(cur_mm);
}
+
+ if (kthread_should_park())
+ kthread_parkme();
+
return 0;
}
@@ -2054,6 +2056,7 @@ static void io_sq_thread_stop(struct io_ring_ctx *ctx)
if (ctx->sqo_thread) {
ctx->sqo_stop = 1;
mb();
+ kthread_park(ctx->sqo_thread);
kthread_stop(ctx->sqo_thread);
ctx->sqo_thread = NULL;
}
@@ -2236,10 +2239,6 @@ static int io_sq_offload_start(struct io_ring_ctx *ctx,
mmgrab(current->mm);
ctx->sqo_mm = current->mm;
- ctx->sq_thread_idle = msecs_to_jiffies(p->sq_thread_idle);
- if (!ctx->sq_thread_idle)
- ctx->sq_thread_idle = HZ;
-
ret = -EINVAL;
if (!cpu_possible(p->sq_thread_cpu))
goto err;
@@ -2249,10 +2248,18 @@ static int io_sq_offload_start(struct io_ring_ctx *ctx,
if (!capable(CAP_SYS_ADMIN))
goto err;
+ ctx->sq_thread_idle = msecs_to_jiffies(p->sq_thread_idle);
+ if (!ctx->sq_thread_idle)
+ ctx->sq_thread_idle = HZ;
+
if (p->flags & IORING_SETUP_SQ_AFF) {
int cpu;
cpu = array_index_nospec(p->sq_thread_cpu, NR_CPUS);
+ ret = -EINVAL;
+ if (!cpu_possible(p->sq_thread_cpu))
+ goto err;
+
ctx->sqo_thread = kthread_create_on_cpu(io_sq_thread,
ctx, cpu,
"io_uring-sq");
@@ -2922,11 +2929,23 @@ SYSCALL_DEFINE2(io_uring_setup, u32, entries,
static int __io_uring_register(struct io_ring_ctx *ctx, unsigned opcode,
void __user *arg, unsigned nr_args)
+ __releases(ctx->uring_lock)
+ __acquires(ctx->uring_lock)
{
int ret;
percpu_ref_kill(&ctx->refs);
+
+ /*
+ * Drop uring mutex before waiting for references to exit. If another
+ * thread is currently inside io_uring_enter() it might need to grab
+ * the uring_lock to make progress. If we hold it here across the drain
+ * wait, then we can deadlock. It's safe to drop the mutex here, since
+ * no new references will come in after we've killed the percpu ref.
+ */
+ mutex_unlock(&ctx->uring_lock);
wait_for_completion(&ctx->ctx_done);
+ mutex_lock(&ctx->uring_lock);
switch (opcode) {
case IORING_REGISTER_BUFFERS: