diff options
author | Jens Axboe <axboe@kernel.dk> | 2021-11-06 02:20:54 +0300 |
---|---|---|
committer | Jens Axboe <axboe@kernel.dk> | 2023-05-16 17:04:55 +0300 |
commit | 03d89a2de25bbc5c77e61a0cf77663978c4b6ea7 (patch) | |
tree | 67e8857880d7d111f2595b895e2efda5e70940ab /include | |
parent | 9c189eee73af1825ea9c895fafad469de5f82641 (diff) | |
download | linux-03d89a2de25bbc5c77e61a0cf77663978c4b6ea7.tar.xz |
io_uring: support for user allocated memory for rings/sqes
Currently io_uring applications must call mmap(2) twice to map the rings
themselves, and the sqes array. This works fine, but it does not support
using huge pages to back the rings/sqes.
Provide a way for the application to pass in pre-allocated memory for
the rings/sqes, which can then suitably be allocated from shmfs or
via mmap to get huge page support.
Particularly for larger rings, this reduces the TLBs needed.
If an application wishes to take advantage of that, it must pre-allocate
the memory needed for the sq/cq ring, and the sqes. The former must
be passed in via the io_uring_params->cq_off.user_data field, while the
latter is passed in via the io_uring_params->sq_off.user_data field. Then
it must set IORING_SETUP_NO_MMAP in the io_uring_params->flags field,
and io_uring will then map the existing memory into the kernel for shared
use. The application must not call mmap(2) to map rings as it otherwise
would have, that will now fail with -EINVAL if this setup flag was used.
The pages used for the rings and sqes must be contigious. The intent here
is clearly that huge pages should be used, otherwise the normal setup
procedure works fine as-is. The application may use one huge page for
both the rings and sqes.
Outside of those initialization changes, everything works like it did
before.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Diffstat (limited to 'include')
-rw-r--r-- | include/linux/io_uring_types.h | 10 | ||||
-rw-r--r-- | include/uapi/linux/io_uring.h | 9 |
2 files changed, 17 insertions, 2 deletions
diff --git a/include/linux/io_uring_types.h b/include/linux/io_uring_types.h index 1b2a20a42413..f04ce513fadb 100644 --- a/include/linux/io_uring_types.h +++ b/include/linux/io_uring_types.h @@ -211,6 +211,16 @@ struct io_ring_ctx { unsigned int compat: 1; enum task_work_notify_mode notify_method; + + /* + * If IORING_SETUP_NO_MMAP is used, then the below holds + * the gup'ed pages for the two rings, and the sqes. + */ + unsigned short n_ring_pages; + unsigned short n_sqe_pages; + struct page **ring_pages; + struct page **sqe_pages; + struct io_rings *rings; struct task_struct *submitter_task; struct percpu_ref refs; diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h index 0716cb17e436..2edba9a274de 100644 --- a/include/uapi/linux/io_uring.h +++ b/include/uapi/linux/io_uring.h @@ -173,6 +173,11 @@ enum { */ #define IORING_SETUP_DEFER_TASKRUN (1U << 13) +/* + * Application provides the memory for the rings + */ +#define IORING_SETUP_NO_MMAP (1U << 14) + enum io_uring_op { IORING_OP_NOP, IORING_OP_READV, @@ -406,7 +411,7 @@ struct io_sqring_offsets { __u32 dropped; __u32 array; __u32 resv1; - __u64 resv2; + __u64 user_addr; }; /* @@ -425,7 +430,7 @@ struct io_cqring_offsets { __u32 cqes; __u32 flags; __u32 resv1; - __u64 resv2; + __u64 user_addr; }; /* |