On Sat, Oct 07, 2023 at 04:32:34PM +0200, Christophe JAILLET wrote:
> Prepare for the coming implementation by GCC and Clang of the __counted_by
> attribute. Flexible array members annotated with __counted_by can have
> their accesses bounds-checked at run-time checking via CONFIG_UBSAN_BOUNDS
> (for array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
> functions).
>
> Signed-off-by: Christophe JAILLET <christophe.jaillet@xxxxxxxxxx>
> ---
> This patch is part of a work done in parallel of what is currently worked
> on by Kees Cook.
>
> My patches are only related to corner cases that do NOT match the
> semantic of his Coccinelle script[1].
>
> In this case, something similar to struct_size() is implemented in
> handle_arr_calc_size().
I think this should likely lose VMCI_HANDLE_ARRAY_HEADER_SIZE entirely
and the helper to use sizeof() and struct_size() directly, but probably
as a separate patch.
>
> Note that I'm slightly unsure on how things will behave in regards to the
> krealloc() in vmci_handle_arr_append_entry().
It looks correct to me:
new_array = krealloc(array, new_size, GFP_ATOMIC);
...
new_array->capacity += capacity_bump;
i.e. "capacity" is adjusted up before accessing any "entries".
>
> [1] https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci
> ---
> drivers/misc/vmw_vmci/vmci_handle_array.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/misc/vmw_vmci/vmci_handle_array.h b/drivers/misc/vmw_vmci/vmci_handle_array.h
> index 96193f85be5b..b0e6b1956014 100644
> --- a/drivers/misc/vmw_vmci/vmci_handle_array.h
> +++ b/drivers/misc/vmw_vmci/vmci_handle_array.h
> @@ -17,7 +17,7 @@ struct vmci_handle_arr {
> u32 max_capacity;
> u32 size;
> u32 pad;
> - struct vmci_handle entries[];
> + struct vmci_handle entries[] __counted_by(capacity);
> };
>
> #define VMCI_HANDLE_ARRAY_HEADER_SIZE \
> --
> 2.34.1
>
Reviewed-by: Kees Cook <keescook@xxxxxxxxxxxx>
--
Kees Cook
