On Tue, Jul 1, 2025 at 5:34 PM Emil Tsalapatis <emil@xxxxxxxxxxxxxxx> wrote:
>
> Add a new BPF arena kfunc for reserving a range of arena virtual
> addresses without backing them with pages. This prevents the range from
> being populated using bpf_arena_alloc_pages().
>
> Signed-off-by: Emil Tsalapatis <emil@xxxxxxxxxxxxxxx>
> ---
> kernel/bpf/arena.c | 43 +++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 43 insertions(+)
>
> diff --git a/kernel/bpf/arena.c b/kernel/bpf/arena.c
> index 0d56cea71602..18db8b826836 100644
> --- a/kernel/bpf/arena.c
> +++ b/kernel/bpf/arena.c
> @@ -550,6 +550,34 @@ static void arena_free_pages(struct bpf_arena *arena, long uaddr, long page_cnt)
> }
> }
>
> +/*
> + * Reserve an arena virtual address range without populating it. This call stops
> + * bpf_arena_alloc_pages from adding pages to this range.
> + */
> +static int arena_reserve_pages(struct bpf_arena *arena, long uaddr, u32 page_cnt)
> +{
> + long page_cnt_max = (arena->user_vm_end - arena->user_vm_start) >> PAGE_SHIFT;
> + long pgoff;
> + int ret;
> +
> + if (uaddr & ~PAGE_MASK)
> + return 0;
> +
> + pgoff = compute_pgoff(arena, uaddr);
> + if (pgoff + page_cnt > page_cnt_max)
> + return -EINVAL;
> +
> + guard(mutex)(&arena->lock);
> +
> + /* Cannot guard already allocated pages. */
> + ret = is_range_tree_set(&arena->rt, pgoff, page_cnt);
> + if (ret)
> + return -EALREADY;
Overall looks good, but this error code gives me a pause.
We never used it in a bpf subsystem.
We typically return EBUSY in such cases.
Unless there is a strong reason to use this new error code
my preference is to stick to old code.
--
pw-bot: cr
