[PATCH 0/3] changes around fixed rsrc

[PATCH 0/3] changes around fixed rsrc

[Pavel Begunkov]

1-2 put some limits on the fixed file tables sizes, files and
buffers.

3/3 adds compatibility checks for ->splice_fd_in, for all requests
buy rw and some others, see the patch message.

All based on 5.15 and merked stable, looks to me as the best way.

Pavel Begunkov (3):
  io_uring: limit fixed table size by RLIMIT_NOFILE
  io_uring: place fixed tables under memcg limits
  io_uring: add ->splice_fd_in checks

 fs/io_uring.c | 61 ++++++++++++++++++++++++++++++---------------------
 1 file changed, 36 insertions(+), 25 deletions(-)

-- 
2.32.0

[PATCH 1/3] io_uring: limit fixed table size by RLIMIT_NOFILE

[Pavel Begunkov]

Limit the number of files in io_uring fixed tables by RLIMIT_NOFILE,
that's the first and the simpliest restriction that we should impose.

Cc: stable@vger.kernel.org
Suggested-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 30edc329d803..e6301d5d03a8 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -7730,6 +7730,8 @@ static int io_sqe_files_register(struct io_ring_ctx *ctx, void __user *arg,
 		return -EINVAL;
 	if (nr_args > IORING_MAX_FIXED_FILES)
 		return -EMFILE;
+	if (nr_args > rlimit(RLIMIT_NOFILE))
+		return -EMFILE;
 	ret = io_rsrc_node_switch_start(ctx);
 	if (ret)
 		return ret;
-- 
2.32.0

[PATCH 2/3] io_uring: place fixed tables under memcg limits

[Pavel Begunkov]

Fixed tables may be large enough, place all of them together with
allocated tags under memcg limits.

Cc: stable@vger.kernel.org
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index e6301d5d03a8..976fc0509e4b 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -7135,14 +7135,14 @@ static void **io_alloc_page_table(size_t size)
 	size_t init_size = size;
 	void **table;
 
-	table = kcalloc(nr_tables, sizeof(*table), GFP_KERNEL);
+	table = kcalloc(nr_tables, sizeof(*table), GFP_KERNEL_ACCOUNT);
 	if (!table)
 		return NULL;
 
 	for (i = 0; i < nr_tables; i++) {
 		unsigned int this_size = min_t(size_t, size, PAGE_SIZE);
 
-		table[i] = kzalloc(this_size, GFP_KERNEL);
+		table[i] = kzalloc(this_size, GFP_KERNEL_ACCOUNT);
 		if (!table[i]) {
 			io_free_page_table(table, init_size);
 			return NULL;
@@ -7333,7 +7333,8 @@ static int io_rsrc_data_alloc(struct io_ring_ctx *ctx, rsrc_put_fn *do_put,
 
 static bool io_alloc_file_tables(struct io_file_table *table, unsigned nr_files)
 {
-	table->files = kvcalloc(nr_files, sizeof(table->files[0]), GFP_KERNEL);
+	table->files = kvcalloc(nr_files, sizeof(table->files[0]),
+				GFP_KERNEL_ACCOUNT);
 	return !!table->files;
 }
 
-- 
2.32.0

[PATCH 3/3] io_uring: add ->splice_fd_in checks

[Pavel Begunkov]

->splice_fd_in is used only by splice/tee, but no other request checks
it for validity. Add the check for most of request types excluding
reads/writes/sends/recvs, we don't want overhead for them and can leave
them be as is until the field is actually used.

Cc: stable@vger.kernel.org
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 52 +++++++++++++++++++++++++++++----------------------
 1 file changed, 30 insertions(+), 22 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 976fc0509e4b..ff1a8c4e2881 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -3502,7 +3502,7 @@ static int io_renameat_prep(struct io_kiocb *req,
 
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->buf_index)
+	if (sqe->ioprio || sqe->buf_index || sqe->splice_fd_in)
 		return -EINVAL;
 	if (unlikely(req->flags & REQ_F_FIXED_FILE))
 		return -EBADF;
@@ -3553,7 +3553,8 @@ static int io_unlinkat_prep(struct io_kiocb *req,
 
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->off || sqe->len || sqe->buf_index)
+	if (sqe->ioprio || sqe->off || sqe->len || sqe->buf_index ||
+	    sqe->splice_fd_in)
 		return -EINVAL;
 	if (unlikely(req->flags & REQ_F_FIXED_FILE))
 		return -EBADF;
@@ -3599,8 +3600,8 @@ static int io_shutdown_prep(struct io_kiocb *req,
 #if defined(CONFIG_NET)
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->off || sqe->addr || sqe->rw_flags ||
-	    sqe->buf_index)
+	if (unlikely(sqe->ioprio || sqe->off || sqe->addr || sqe->rw_flags ||
+		     sqe->buf_index || sqe->splice_fd_in))
 		return -EINVAL;
 
 	req->shutdown.how = READ_ONCE(sqe->len);
@@ -3748,7 +3749,8 @@ static int io_fsync_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 
 	if (unlikely(ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (unlikely(sqe->addr || sqe->ioprio || sqe->buf_index))
+	if (unlikely(sqe->addr || sqe->ioprio || sqe->buf_index ||
+		     sqe->splice_fd_in))
 		return -EINVAL;
 
 	req->sync.flags = READ_ONCE(sqe->fsync_flags);
@@ -3781,7 +3783,8 @@ static int io_fsync(struct io_kiocb *req, unsigned int issue_flags)
 static int io_fallocate_prep(struct io_kiocb *req,
 			     const struct io_uring_sqe *sqe)
 {
-	if (sqe->ioprio || sqe->buf_index || sqe->rw_flags)
+	if (sqe->ioprio || sqe->buf_index || sqe->rw_flags ||
+	    sqe->splice_fd_in)
 		return -EINVAL;
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
@@ -3814,7 +3817,7 @@ static int __io_openat_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe
 
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (unlikely(sqe->ioprio || sqe->buf_index))
+	if (unlikely(sqe->ioprio || sqe->buf_index || sqe->splice_fd_in))
 		return -EINVAL;
 	if (unlikely(req->flags & REQ_F_FIXED_FILE))
 		return -EBADF;
@@ -3933,7 +3936,8 @@ static int io_remove_buffers_prep(struct io_kiocb *req,
 	struct io_provide_buf *p = &req->pbuf;
 	u64 tmp;
 
-	if (sqe->ioprio || sqe->rw_flags || sqe->addr || sqe->len || sqe->off)
+	if (sqe->ioprio || sqe->rw_flags || sqe->addr || sqe->len || sqe->off ||
+	    sqe->splice_fd_in)
 		return -EINVAL;
 
 	tmp = READ_ONCE(sqe->fd);
@@ -4004,7 +4008,7 @@ static int io_provide_buffers_prep(struct io_kiocb *req,
 	struct io_provide_buf *p = &req->pbuf;
 	u64 tmp;
 
-	if (sqe->ioprio || sqe->rw_flags)
+	if (sqe->ioprio || sqe->rw_flags || sqe->splice_fd_in)
 		return -EINVAL;
 
 	tmp = READ_ONCE(sqe->fd);
@@ -4091,7 +4095,7 @@ static int io_epoll_ctl_prep(struct io_kiocb *req,
 			     const struct io_uring_sqe *sqe)
 {
 #if defined(CONFIG_EPOLL)
-	if (sqe->ioprio || sqe->buf_index)
+	if (sqe->ioprio || sqe->buf_index || sqe->splice_fd_in)
 		return -EINVAL;
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
@@ -4137,7 +4141,7 @@ static int io_epoll_ctl(struct io_kiocb *req, unsigned int issue_flags)
 static int io_madvise_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 {
 #if defined(CONFIG_ADVISE_SYSCALLS) && defined(CONFIG_MMU)
-	if (sqe->ioprio || sqe->buf_index || sqe->off)
+	if (sqe->ioprio || sqe->buf_index || sqe->off || sqe->splice_fd_in)
 		return -EINVAL;
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
@@ -4172,7 +4176,7 @@ static int io_madvise(struct io_kiocb *req, unsigned int issue_flags)
 
 static int io_fadvise_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 {
-	if (sqe->ioprio || sqe->buf_index || sqe->addr)
+	if (sqe->ioprio || sqe->buf_index || sqe->addr || sqe->splice_fd_in)
 		return -EINVAL;
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
@@ -4210,7 +4214,7 @@ static int io_statx_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 {
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->buf_index)
+	if (sqe->ioprio || sqe->buf_index || sqe->splice_fd_in)
 		return -EINVAL;
 	if (req->flags & REQ_F_FIXED_FILE)
 		return -EBADF;
@@ -4246,7 +4250,7 @@ static int io_close_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
 	if (sqe->ioprio || sqe->off || sqe->addr || sqe->len ||
-	    sqe->rw_flags || sqe->buf_index)
+	    sqe->rw_flags || sqe->buf_index || sqe->splice_fd_in)
 		return -EINVAL;
 	if (req->flags & REQ_F_FIXED_FILE)
 		return -EBADF;
@@ -4307,7 +4311,8 @@ static int io_sfr_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 
 	if (unlikely(ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (unlikely(sqe->addr || sqe->ioprio || sqe->buf_index))
+	if (unlikely(sqe->addr || sqe->ioprio || sqe->buf_index ||
+		     sqe->splice_fd_in))
 		return -EINVAL;
 
 	req->sync.off = READ_ONCE(sqe->off);
@@ -4734,7 +4739,7 @@ static int io_accept_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->len || sqe->buf_index)
+	if (sqe->ioprio || sqe->len || sqe->buf_index || sqe->splice_fd_in)
 		return -EINVAL;
 
 	accept->addr = u64_to_user_ptr(READ_ONCE(sqe->addr));
@@ -4782,7 +4787,8 @@ static int io_connect_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->len || sqe->buf_index || sqe->rw_flags)
+	if (sqe->ioprio || sqe->len || sqe->buf_index || sqe->rw_flags ||
+	    sqe->splice_fd_in)
 		return -EINVAL;
 
 	conn->addr = u64_to_user_ptr(READ_ONCE(sqe->addr));
@@ -5377,7 +5383,7 @@ static int io_poll_update_prep(struct io_kiocb *req,
 
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->buf_index)
+	if (sqe->ioprio || sqe->buf_index || sqe->splice_fd_in)
 		return -EINVAL;
 	flags = READ_ONCE(sqe->len);
 	if (flags & ~(IORING_POLL_UPDATE_EVENTS | IORING_POLL_UPDATE_USER_DATA |
@@ -5616,7 +5622,7 @@ static int io_timeout_remove_prep(struct io_kiocb *req,
 		return -EINVAL;
 	if (unlikely(req->flags & (REQ_F_FIXED_FILE | REQ_F_BUFFER_SELECT)))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->buf_index || sqe->len)
+	if (sqe->ioprio || sqe->buf_index || sqe->len || sqe->splice_fd_in)
 		return -EINVAL;
 
 	tr->addr = READ_ONCE(sqe->addr);
@@ -5677,7 +5683,8 @@ static int io_timeout_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe,
 
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->buf_index || sqe->len != 1)
+	if (sqe->ioprio || sqe->buf_index || sqe->len != 1 ||
+	    sqe->splice_fd_in)
 		return -EINVAL;
 	if (off && is_timeout_link)
 		return -EINVAL;
@@ -5833,7 +5840,8 @@ static int io_async_cancel_prep(struct io_kiocb *req,
 		return -EINVAL;
 	if (unlikely(req->flags & (REQ_F_FIXED_FILE | REQ_F_BUFFER_SELECT)))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->off || sqe->len || sqe->cancel_flags)
+	if (sqe->ioprio || sqe->off || sqe->len || sqe->cancel_flags ||
+	    sqe->splice_fd_in)
 		return -EINVAL;
 
 	req->cancel.addr = READ_ONCE(sqe->addr);
@@ -5874,7 +5882,7 @@ static int io_rsrc_update_prep(struct io_kiocb *req,
 {
 	if (unlikely(req->flags & (REQ_F_FIXED_FILE | REQ_F_BUFFER_SELECT)))
 		return -EINVAL;
-	if (sqe->ioprio || sqe->rw_flags)
+	if (sqe->ioprio || sqe->rw_flags || sqe->splice_fd_in)
 		return -EINVAL;
 
 	req->rsrc_update.offset = READ_ONCE(sqe->off);
-- 
2.32.0

Re: [PATCH 0/3] changes around fixed rsrc

[Jens Axboe]

On 8/20/21 3:36 AM, Pavel Begunkov wrote:
> 1-2 put some limits on the fixed file tables sizes, files and
> buffers.
> 
> 3/3 adds compatibility checks for ->splice_fd_in, for all requests
> buy rw and some others, see the patch message.
> 
> All based on 5.15 and merked stable, looks to me as the best way.
> 
> Pavel Begunkov (3):
>   io_uring: limit fixed table size by RLIMIT_NOFILE
>   io_uring: place fixed tables under memcg limits
>   io_uring: add ->splice_fd_in checks
> 
>  fs/io_uring.c | 61 ++++++++++++++++++++++++++++++---------------------
>  1 file changed, 36 insertions(+), 25 deletions(-)

Applied - especially 3/3 will be a bit of a stable pain. Nothing difficult,
just needs attention for each version...

-- 
Jens Axboe

Re: [PATCH 0/3] changes around fixed rsrc

[Pavel Begunkov]

On 8/21/21 2:18 PM, Jens Axboe wrote:
> On 8/20/21 3:36 AM, Pavel Begunkov wrote:
>> 1-2 put some limits on the fixed file tables sizes, files and
>> buffers.
>>
>> 3/3 adds compatibility checks for ->splice_fd_in, for all requests
>> buy rw and some others, see the patch message.
>>
>> All based on 5.15 and merked stable, looks to me as the best way.
>>
>> Pavel Begunkov (3):
>>   io_uring: limit fixed table size by RLIMIT_NOFILE
>>   io_uring: place fixed tables under memcg limits
>>   io_uring: add ->splice_fd_in checks
>>
>>  fs/io_uring.c | 61 ++++++++++++++++++++++++++++++---------------------
>>  1 file changed, 36 insertions(+), 25 deletions(-)
> 
> Applied - especially 3/3 will be a bit of a stable pain. Nothing difficult,
> just needs attention for each version...

Yep. Thanks, Jens

-- 
Pavel Begunkov