Re: [PATCH v2 1/2] mm: vmscan: add cgroup IDs to vmscan tracepoints

[Steven Rostedt <rostedt@goodmis.org>]

On Tue, 16 Dec 2025 06:02:51 -0800
Thomas Ballasi <tballasi@linux.microsoft.com> wrote:

> ---
>  include/trace/events/vmscan.h | 65 +++++++++++++++++++++--------------
>  mm/vmscan.c                   | 17 ++++-----
>  2 files changed, 48 insertions(+), 34 deletions(-)
> 
> diff --git a/include/trace/events/vmscan.h b/include/trace/events/vmscan.h
> index d2123dd960d59..afc9f80d03f34 100644
> --- a/include/trace/events/vmscan.h
> +++ b/include/trace/events/vmscan.h
> @@ -114,85 +114,92 @@ TRACE_EVENT(mm_vmscan_wakeup_kswapd,
>  
>  DECLARE_EVENT_CLASS(mm_vmscan_direct_reclaim_begin_template,
>  
> -	TP_PROTO(int order, gfp_t gfp_flags),
> +	TP_PROTO(int order, gfp_t gfp_flags, unsigned short memcg_id),
>  
> -	TP_ARGS(order, gfp_flags),
> +	TP_ARGS(order, gfp_flags, memcg_id),
>  
>  	TP_STRUCT__entry(
>  		__field(	int,	order		)
>  		__field(	unsigned long,	gfp_flags	)
> +		__field(	unsigned short,	memcg_id	)
>  	),

Hmm, the above adds some holes. Note, events are at a minimum, 4 bytes
aligend. On 64bit, they can be 8 byte aligned. Still, above is the same as:

	struct {
		int		order;
		unsigned long	gfp_flags;
		unsigned short	memcg_id;
	};

See the issue? Perhaps it may be better to add the memcg_id in between the
order and gfp_flags?

-- Steve