Re: [RFC PATCH v1 02/10] bpf: widen dynptr size/offset to 64 bit

[Eduard Zingerman <eddyz87@gmail.com>]

On Fri, 2025-10-03 at 17:04 +0100, Mykyta Yatsenko wrote:
> From: Mykyta Yatsenko <yatsenko@meta.com>
> 
> Dynptr currently caps size and offset at 24 bits, which isn’t sufficient
> for file-backed use cases; even 32 bits can be limiting. Refactor dynptr
> helpers/kfuncs to use 64-bit size and offset, ensuring consistency
> across the APIs.
> 
> This change does not affect internals of xdp, skb or other dynptrs,
> which continue to behave as before.
> 
> The widening enables large-file access support via dynptr, implemented
> in the next patches.

Maybe add a note here that this change does not break binary
compatibility with BPF programs compiled for older kernels?

> 
> Signed-off-by: Mykyta Yatsenko <yatsenko@meta.com>
> ---

Acked-by: Eduard Zingerman <eddyz87@gmail.com>

[...]

> diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
> index 8f23f5273bab..7cc4f2e05ed2 100644
> --- a/kernel/trace/bpf_trace.c
> +++ b/kernel/trace/bpf_trace.c
> @@ -3372,13 +3372,13 @@ typedef int (*copy_fn_t)(void *dst, const void *src, u32 size, struct task_struc
>   * direct calls into all the specific callback implementations
>   * (copy_user_data_sleepable, copy_user_data_nofault, and so on)
>   */
> -static __always_inline int __bpf_dynptr_copy_str(struct bpf_dynptr *dptr, u32 doff, u32 size,
> +static __always_inline int __bpf_dynptr_copy_str(struct bpf_dynptr *dptr, u64 doff, u64 size,
>  						 const void *unsafe_src,
>  						 copy_fn_t str_copy_fn,

The definition for copy_fn_t looks like:

  typedef int (*copy_fn_t)(void *dst, const void *src, u32 size, struct task_struct *tsk);

should we change it to use u64 as well? Probably does not matter.

>  						 struct task_struct *tsk)
>  {
>  	struct bpf_dynptr_kern *dst;
> -	u32 chunk_sz, off;
> +	u64 chunk_sz, off;
>  	void *dst_slice;
>  	int cnt, err;
>  	char buf[256];

[...]