Re: [PATCH] mm/vmalloc: move 'area->pages' after if statement

[Vlastimil Babka]

On 8/30/19 10:50 PM, Andrew Morton wrote:
> On Fri, 30 Aug 2019 12:57:16 +0900 Austin Kim <austindh.kim@xxxxxxxxx> wrote:
> 
>> If !area->pages statement is true where memory allocation fails, 
>> area is freed.
>>
>> In this case 'area->pages = pages' should not executed.

That read like a use after free fix and made me wonder about stable,
fixes etc... luckily it's not.

>> So move 'area->pages = pages' after if statement.
>>
>> ...
>>
>> --- a/mm/vmalloc.c
>> +++ b/mm/vmalloc.c
>> @@ -2416,13 +2416,15 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
>>  	} else {
>>  		pages = kmalloc_node(array_size, nested_gfp, node);
>>  	}
>> -	area->pages = pages;
>> -	if (!area->pages) {
>> +
>> +	if (!pages) {
>>  		remove_vm_area(area->addr);
>>  		kfree(area);
>>  		return NULL;
>>  	}
>>  
>> +	area->pages = pages;
>> +
>>  	for (i = 0; i < area->nr_pages; i++) {
>>  		struct page *page;
>>  
> 
> Fair enough.  But we can/should also do this?

Agreed, same thing.

Acked-by: Vlastimil Babka <vbabka@xxxxxxx>

> --- a/mm/vmalloc.c~mm-vmalloc-move-area-pages-after-if-statement-fix
> +++ a/mm/vmalloc.c
> @@ -2409,7 +2409,6 @@ static void *__vmalloc_area_node(struct
>  	nr_pages = get_vm_area_size(area) >> PAGE_SHIFT;
>  	array_size = (nr_pages * sizeof(struct page *));
>  
> -	area->nr_pages = nr_pages;
>  	/* Please note that the recursion is strictly bounded. */
>  	if (array_size > PAGE_SIZE) {
>  		pages = __vmalloc_node(array_size, 1, nested_gfp|highmem_mask,
> @@ -2425,6 +2424,7 @@ static void *__vmalloc_area_node(struct
>  	}
>  
>  	area->pages = pages;
> +	area->nr_pages = nr_pages;
>  
>  	for (i = 0; i < area->nr_pages; i++) {
>  		struct page *page;
> _
> 
>