Re: [PATCH] perf/core: Fix mmap failure caused by bad size check

[Changbin Du]

Hi Peter, Ingo and Arnaldo,
Could you help review this and have this fix merged in next RC? The perf
is totaly broken now! Thanks.

On Tue, Feb 12, 2019 at 11:52:21PM +0800, Changbin Du wrote:
> Perf tool now is broken by commit 9dff0aa95a32 ("perf/core: Don't WARN()
> for impossible ring-buffer sizes"). The get_order() should be used but not
> the order_base_2(). This mistake makes mmap() on perf fd always fail.
> 
> Fixes: 9dff0aa95a32 ("perf/core: Don't WARN() for impossible ring-buffer sizes")
> Signed-off-by: Changbin Du <changbin.du@xxxxxxxxx>
> Cc: Mark Rutland <mark.rutland@xxxxxxx>
> Cc: Julien Thierry <julien.thierry@xxxxxxx>
> Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> Cc: Alexander Shishkin <alexander.shishkin@xxxxxxxxxxxxxxx>
> Cc: <stable@xxxxxxxxxxxxxxx>
> ---
>  kernel/events/ring_buffer.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c
> index 309ef5a64af5..55faafca4a6e 100644
> --- a/kernel/events/ring_buffer.c
> +++ b/kernel/events/ring_buffer.c
> @@ -734,7 +734,7 @@ struct ring_buffer *rb_alloc(int nr_pages, long watermark, int cpu, int flags)
>  	size = sizeof(struct ring_buffer);
>  	size += nr_pages * sizeof(void *);
>  
> -	if (order_base_2(size) >= MAX_ORDER)
> +	if (get_order(size) >= MAX_ORDER)
>  		goto fail;
>  
>  	rb = kzalloc(size, GFP_KERNEL);
> -- 
> 2.19.1
> 

-- 
Cheers,
Changbin Du