Re: [PATCH bpf-next v1 1/8] bpf: Add bpf_dynptr_data_rdonly

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, Sep 8, 2022 at 1:10 AM Joanne Koong <joannelkoong@xxxxxxxxx> wrote:
>
> Add a new helper bpf_dynptr_data_rdonly
>
> void *bpf_dynptr_data_rdonly(struct bpf_dynptr *ptr, u32 offset, u32 len);
>
> which gets a read-only pointer to the underlying dynptr data.
>
> This is equivalent to bpf_dynptr_data(), except the pointer returned is
> read-only, which allows this to support both read-write and read-only
> dynptrs.
>
> One example where this will be useful is for skb dynptrs where the
> program type only allows read-only access to packet data. This API will
> provide a way to obtain a data slice that can be used for direct reads.
>
> Signed-off-by: Joanne Koong <joannelkoong@xxxxxxxxx>
> ---
>  include/uapi/linux/bpf.h       | 15 +++++++++++++++
>  kernel/bpf/helpers.c           | 32 ++++++++++++++++++++++++++------
>  kernel/bpf/verifier.c          |  7 +++++--
>  tools/include/uapi/linux/bpf.h | 15 +++++++++++++++
>  4 files changed, 61 insertions(+), 8 deletions(-)
>
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index c55c23f25c0f..cce3356765fc 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -5439,6 +5439,20 @@ union bpf_attr {
>   *             *flags* is currently unused, it must be 0 for now.
>   *     Return
>   *             0 on success, -EINVAL if flags is not 0.
> + *
> + * void *bpf_dynptr_data_rdonly(struct bpf_dynptr *ptr, u32 offset, u32 len)
> + *     Description
> + *             Get a read-only pointer to the underlying dynptr data.
> + *
> + *             This is equivalent to **bpf_dynptr_data**\ () except the
> + *             pointer returned is read-only, which allows this to support
> + *             both read-write and read-only dynptrs. For more details on using
> + *             the API, please refer to **bpf_dynptr_data**\ ().
> + *     Return
> + *             Read-only pointer to the underlying dynptr data, NULL if the
> + *             dynptr is invalid or if the offset and length is out of bounds
> + *             or in a paged buffer for skb-type dynptrs or across fragments
> + *             for xdp-type dynptrs.
>   */
>  #define __BPF_FUNC_MAPPER(FN)          \
>         FN(unspec),                     \
> @@ -5652,6 +5666,7 @@ union bpf_attr {
>         FN(ktime_get_tai_ns),           \
>         FN(dynptr_from_skb),            \
>         FN(dynptr_from_xdp),            \
> +       FN(dynptr_data_rdonly),         \
>         /* */
>
>  /* integer value in 'imm' field of BPF_CALL instruction selects which helper
> diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> index befafae34a63..30a59c9e5df3 100644
> --- a/kernel/bpf/helpers.c
> +++ b/kernel/bpf/helpers.c
> @@ -1572,7 +1572,7 @@ static const struct bpf_func_proto bpf_dynptr_write_proto = {
>         .arg5_type      = ARG_ANYTHING,
>  };
>
> -BPF_CALL_3(bpf_dynptr_data, struct bpf_dynptr_kern *, ptr, u32, offset, u32, len)
> +void *__bpf_dynptr_data(struct bpf_dynptr_kern *ptr, u32 offset, u32 len, bool writable)
>  {
>         enum bpf_dynptr_type type;
>         void *data;
> @@ -1585,7 +1585,7 @@ BPF_CALL_3(bpf_dynptr_data, struct bpf_dynptr_kern *, ptr, u32, offset, u32, len
>         if (err)
>                 return 0;

Let's return NULL for void* type.

>
> -       if (bpf_dynptr_is_rdonly(ptr))
> +       if (writable && bpf_dynptr_is_rdonly(ptr))
>                 return 0;
ditto
>
>         type = bpf_dynptr_get_type(ptr);
> @@ -1610,13 +1610,31 @@ BPF_CALL_3(bpf_dynptr_data, struct bpf_dynptr_kern *, ptr, u32, offset, u32, len
>                 /* if the requested data in across fragments, then it cannot
>                  * be accessed directly - bpf_xdp_pointer will return NULL
>                  */
> -               return (unsigned long)bpf_xdp_pointer(ptr->data,
> -                                                     ptr->offset + offset, len);
> +               return bpf_xdp_pointer(ptr->data, ptr->offset + offset, len);
>         default:
> -               WARN_ONCE(true, "bpf_dynptr_data: unknown dynptr type %d\n", type);
> +               WARN_ONCE(true, "__bpf_dynptr_data: unknown dynptr type %d\n", type);

Let's use __func__ so we don't have to change this again.

WARN_ONCE(true, "%s: unknown dynptr type %d\n", __func__, type);

Thanks,
Song

[...]



[Index of Archives]     [Linux Samsung SoC]     [Linux Rockchip SoC]     [Linux Actions SoC]     [Linux for Synopsys ARC Processors]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]


  Powered by Linux