Re: [PATCH v15 1/2] Reorganize the oom report in dump_header

[Michal Hocko]

On Sat 29-09-18 21:06:26, ufo19890607@xxxxxxxxx wrote:
[...]
> Changes since v14:
> - add the dump_oom_summary for the single line output of oom context.
> - fix the null pointer in the dump_header.

I do not remember details about this null ptr but the fix you seemed to
have done is
[...]
> +static void dump_oom_summary(struct oom_control *oc, struct task_struct *victim)
> +{
> +	/* one line summary of the oom killer context. */
> +	pr_info("oom-kill:constraint=%s,nodemask=%*pbl",
> +			oom_constraint_text[oc->constraint],
> +			nodemask_pr_args(oc->nodemask));
> +	cpuset_print_current_mems_allowed();
> +	pr_cont(",task=%s,pid=%d,uid=%d\n", victim->comm, victim->pid,
> +		from_kuid(&init_user_ns, task_uid(victim)));
> +}
> +
>  /*
>   * Number of OOM victims in flight
>   */
> @@ -951,6 +960,8 @@ static void oom_kill_process(struct oom_control *oc, const char *message)
>  
>  	if (__ratelimit(&oom_rs))
>  		dump_header(oc, p);
> +	if (oc)
> +		dump_oom_summary(oc, victim);
>  

this? If yes then this is bogus because oc is never NULL. Besides that,
you used to have this one line summary in dump_header which looks much
better fit to me than oom_kill_process.

-- 
Michal Hocko
SUSE Labs