Re: [PATCH 2/2] perf/ring_buffer: Fix high-order allocations for AUX space with correct MAX_ORDER limit

[Shuai Xue <xueshuai@linux.alibaba.com>]

On 2023/6/12 17:09, Leo Yan wrote:
> On Mon, Jun 12, 2023 at 09:45:38AM +0100, James Clark wrote:
> 
> [...]
> 
>>> @@ -609,8 +609,8 @@ static struct page *rb_alloc_aux_page(int node, int order)
>>>  {
>>>  	struct page *page;
>>>  
>>> -	if (order > MAX_ORDER)
>>> -		order = MAX_ORDER;
>>> +	if (order >= MAX_ORDER)
>>> +		order = MAX_ORDER - 1;
>>>  
>>>  	do {
>>>  		page = alloc_pages_node(node, PERF_AUX_GFP, order);
>>
>>
>> It seems like this was only just recently changed with this as the
>> commit message (23baf83):
>>
>>    mm, treewide: redefine MAX_ORDER sanely
>>
>>   MAX_ORDER currently defined as number of orders page allocator
>>   supports: user can ask buddy allocator for page order between 0 and
>>   MAX_ORDER-1.
>>
>>   This definition is counter-intuitive and lead to number of bugs all
>>   over the kernel.
>>
>>   Change the definition of MAX_ORDER to be inclusive: the range of
>>   orders user can ask from buddy allocator is 0..MAX_ORDER now.
>>
>> It might be worth referring to this in the commit message or adding a
>> fixes: reference. Or maybe this new change isn't quite right?
> 
> Good point.  If so, we don't need this patch anymore.
> 
> Thanks for reminding, James.
> 
> Leo

Hi, Leo and James,

I tested on the Linus master tree, the mentioned commit 23baf83 ("mm, treewide: redefine MAX_ORDER sanely")
has fix this oops.

I will drop out this patch, thank you :)

Cheers,
Shuai