Re: [PATCH] mm: usercopy: move the virt_addr_valid() below the is_vmalloc_addr()

[Andrew Morton]

Matthew & Kees,

On Thu, 5 May 2022 07:10:37 +0000 Yuanzheng Song <songyuanzheng@xxxxxxxxxx> wrote:

> The is_kmap_addr() and the is_vmalloc_addr() in the check_heap_object()
> will not work, because the virt_addr_valid() will exclude the kmap and
> vmalloc regions. So let's move the virt_addr_valid() below
> the is_vmalloc_addr().

The author,

> Signed-off-by: Yuanzheng Song <songyuanzheng@xxxxxxxxxx>

Tells me off-list that this fix:

> --- a/mm/usercopy.c
> +++ b/mm/usercopy.c
> @@ -163,9 +163,6 @@ static inline void check_heap_object(const void *ptr, unsigned long n,
>  {
>  	struct folio *folio;
>  
> -	if (!virt_addr_valid(ptr))
> -		return;
> -
>  	if (is_kmap_addr(ptr)) {
>  		unsigned long page_end = (unsigned long)ptr | (PAGE_SIZE - 1);
>  
> @@ -190,6 +187,9 @@ static inline void check_heap_object(const void *ptr, unsigned long n,
>  		return;
>  	}
>  
> +	if (!virt_addr_valid(ptr))
> +		return;
> +
>  	folio = virt_to_folio(ptr);
>  
>  	if (folio_test_slab(folio)) {

is required to fix patches "mm/usercopy: Check kmap addresses properly"
and "mm/usercopy: Detect vmalloc overruns".