[PATCH 0/3] small cleanups for 5.12

[PATCH 0/3] small cleanups for 5.12

[Pavel Begunkov]

Final easy 5.12 cleanups.

Pavel Begunkov (3):
  io_uring: don't check PF_EXITING from syscall
  io_uring: clean io_req_find_next() fast check
  io_uring: optimise io_init_req() flags setting

 fs/io_uring.c | 24 ++++++++++--------------
 1 file changed, 10 insertions(+), 14 deletions(-)

-- 
2.24.0

[PATCH 1/3] io_uring: don't check PF_EXITING from syscall

[Pavel Begunkov]

io_sq_thread_acquire_mm_files() can find a PF_EXITING task only when
it's called from task_work context. Don't check it in all other cases,
that are when we're in io_uring_enter().

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 9c58be0579f3..66bbb0dc50af 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -1213,8 +1213,6 @@ static int __io_sq_thread_acquire_mm_files(struct io_ring_ctx *ctx,
 static inline int io_sq_thread_acquire_mm_files(struct io_ring_ctx *ctx,
 						struct io_kiocb *req)
 {
-	if (unlikely(current->flags & PF_EXITING))
-		return -EFAULT;
 	if (!(ctx->flags & IORING_SETUP_SQPOLL))
 		return 0;
 	return __io_sq_thread_acquire_mm_files(ctx, req);
@@ -2338,7 +2336,8 @@ static void __io_req_task_submit(struct io_kiocb *req)
 
 	/* ctx stays valid until unlock, even if we drop all ours ctx->refs */
 	mutex_lock(&ctx->uring_lock);
-	if (!ctx->sqo_dead && !io_sq_thread_acquire_mm_files(ctx, req))
+	if (!ctx->sqo_dead && !(current->flags & PF_EXITING) &&
+	    !io_sq_thread_acquire_mm_files(ctx, req))
 		__io_queue_sqe(req);
 	else
 		__io_req_task_cancel(req, -EFAULT);
-- 
2.24.0

[PATCH 2/3] io_uring: clean io_req_find_next() fast check

[Pavel Begunkov]

Indirectly io_req_find_next() is called for every request, optimise the
check by testing flags as it was long before -- __io_req_find_next()
tolerates false-positives well (i.e. link==NULL), and those should be
really rare.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 66bbb0dc50af..776531f6e18b 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -2172,7 +2172,7 @@ static struct io_kiocb *__io_req_find_next(struct io_kiocb *req)
 
 static inline struct io_kiocb *io_req_find_next(struct io_kiocb *req)
 {
-	if (likely(!(req->link) && !(req->flags & REQ_F_LINK_TIMEOUT)))
+	if (likely(!(req->flags & (REQ_F_LINK|REQ_F_HARDLINK))))
 		return NULL;
 	return __io_req_find_next(req);
 }
-- 
2.24.0

[PATCH 3/3] io_uring: optimise io_init_req() flags setting

[Pavel Begunkov]

Invalid req->flags are tolerated by free/put well, avoid this dancing
needlessly presetting it to zero, and then not even resetting but
modifying it, i.e. "|=".

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 17 +++++++----------
 1 file changed, 7 insertions(+), 10 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 776531f6e18b..2e8cb739c835 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -6806,14 +6806,15 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req,
 {
 	struct io_submit_state *state;
 	unsigned int sqe_flags;
-	int id, ret;
+	int id, ret = 0;
 
 	req->opcode = READ_ONCE(sqe->opcode);
+	/* same numerical values with corresponding REQ_F_*, safe to copy */
+	req->flags = sqe_flags = READ_ONCE(sqe->flags);
 	req->user_data = READ_ONCE(sqe->user_data);
 	req->async_data = NULL;
 	req->file = NULL;
 	req->ctx = ctx;
-	req->flags = 0;
 	req->link = NULL;
 	req->fixed_rsrc_refs = NULL;
 	/* one is dropped after submission, the other at completion */
@@ -6821,17 +6822,16 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req,
 	req->task = current;
 	req->result = 0;
 
+	/* enforce forwards compatibility on users */
+	if (unlikely(sqe_flags & ~SQE_VALID_FLAGS))
+		return -EINVAL;
+
 	if (unlikely(req->opcode >= IORING_OP_LAST))
 		return -EINVAL;
 
 	if (unlikely(io_sq_thread_acquire_mm_files(ctx, req)))
 		return -EFAULT;
 
-	sqe_flags = READ_ONCE(sqe->flags);
-	/* enforce forwards compatibility on users */
-	if (unlikely(sqe_flags & ~SQE_VALID_FLAGS))
-		return -EINVAL;
-
 	if (unlikely(!io_check_restriction(ctx, req, sqe_flags)))
 		return -EACCES;
 
@@ -6854,8 +6854,6 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req,
 		req->work.flags |= IO_WQ_WORK_CREDS;
 	}
 
-	/* same numerical values with corresponding REQ_F_*, safe to copy */
-	req->flags |= sqe_flags;
 	state = &ctx->submit_state;
 
 	/*
@@ -6868,7 +6866,6 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req,
 		state->plug_started = true;
 	}
 
-	ret = 0;
 	if (io_op_defs[req->opcode].needs_file) {
 		bool fixed = req->flags & REQ_F_FIXED_FILE;
 
-- 
2.24.0

Re: [PATCH 0/3] small cleanups for 5.12

[Jens Axboe]

On 2/12/21 11:41 AM, Pavel Begunkov wrote:
> Final easy 5.12 cleanups.
> 
> Pavel Begunkov (3):
>   io_uring: don't check PF_EXITING from syscall
>   io_uring: clean io_req_find_next() fast check
>   io_uring: optimise io_init_req() flags setting
> 
>  fs/io_uring.c | 24 ++++++++++--------------
>  1 file changed, 10 insertions(+), 14 deletions(-)

Nice little cleanups, applied.

-- 
Jens Axboe