Re: [PATCH 2/3] perf: Do error out on a kernel filter on an exclude_filter event

[Mathieu Poirier]

On Thu, Jan 26, 2017 at 11:40:56AM +0200, Alexander Shishkin wrote:
> It is currently possible to configure a kernel address filter for a
> event that excludes kernel from its traces (attr.exclude_kernel==1).
> 
> While in reality this doesn't make sense, the SET_FILTER ioctl() should
> return a error in such case, currently it does not. Furthermore, it
> will still silently discard the filter and any potentially valid filters
> that came with it.
> 
> This patch makes the SET_FILTER ioctl() error out in such cases.
> 
> Signed-off-by: Alexander Shishkin <alexander.shishkin@xxxxxxxxxxxxxxx>
> Cc: Mathieu Poirier <mathieu.poirier@xxxxxxxxxx>
> ---
>  kernel/events/core.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/kernel/events/core.c b/kernel/events/core.c
> index b422b5feee..36770a13ef 100644
> --- a/kernel/events/core.c
> +++ b/kernel/events/core.c
> @@ -8261,6 +8261,7 @@ perf_event_parse_addr_filter(struct perf_event *event, char *fstr,
>  		 * attribute.
>  		 */
>  		if (state == IF_STATE_END) {
> +			ret = -EINVAL;
>  			if (kernel && event->attr.exclude_kernel)
>  				goto fail;

Yes, I stumbled on that when implementing filters on CS and could swear I had a
patch for it.  It obviously never got sent out.

Reviewed-by: Mathieu Poirier <mathieu.poirier@xxxxxxxxxx>

>  
> -- 
> 2.11.0
>