Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 29 Sep 2022 07:57:52 +0530
From:   Pavan Kondeti <quic_pkondeti@quicinc.com>
To:     Kassey Li <quic_yingangl@quicinc.com>
CC:     <mingo@redhat.com>, <peterz@infradead.org>,
        <linux-kernel@vger.kernel.org>, <quic_namajain@quicinc.com>
Subject: Re: [PATCH v2] kernel/hung_task: add option to ignore task
Message-ID: <20220929022752.GA28684@hu-pkondeti-hyd.qualcomm.com>
References: <20220928074841.22545-1-quic_yingangl@quicinc.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Disposition: inline
In-Reply-To: <20220928074841.22545-1-quic_yingangl@quicinc.com>
User-Agent: Mutt/1.5.24 (2015-08-30)
Precedence: bulk

On Wed, Sep 28, 2022 at 03:48:41PM +0800, Kassey Li wrote:
> By default, hung_task will iterate the tasklist and check
> state in TASK_UNINTERRUPTIBLE with a given timeout value.
> 
> Here we add an option for task_struct so it can be ignored.
> Set this flag to default true, it do not break the origin design.
> 
> This is useful when we set timeout value to 5s, where we just want
> to detect some tasks interested.
> 
> Suggested-by: Naman Jain <quic_namajain@quicinc.com>
> Signed-off-by: Kassey Li <quic_yingangl@quicinc.com>
> ---
>  include/linux/sched.h | 1 +
>  kernel/fork.c         | 1 +
>  kernel/hung_task.c    | 3 ++-
>  3 files changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/include/linux/sched.h b/include/linux/sched.h
> index e7b2f8a5c711..7c8596fea1f6 100644
> --- a/include/linux/sched.h
> +++ b/include/linux/sched.h
> @@ -1071,6 +1071,7 @@ struct task_struct {
>  #ifdef CONFIG_DETECT_HUNG_TASK
>  	unsigned long			last_switch_count;
>  	unsigned long			last_switch_time;
> +	bool			hung_task_detect;
>  #endif
>  	/* Filesystem information: */
>  	struct fs_struct		*fs;
> diff --git a/kernel/fork.c b/kernel/fork.c
> index 90c85b17bf69..5c461a37a26e 100644
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -1552,6 +1552,7 @@ static int copy_mm(unsigned long clone_flags, struct task_struct *tsk)
>  #ifdef CONFIG_DETECT_HUNG_TASK
>  	tsk->last_switch_count = tsk->nvcsw + tsk->nivcsw;
>  	tsk->last_switch_time = 0;
> +	tsk->hung_task_detect = 1;
>  #endif
>  
>  	tsk->mm = NULL;
> diff --git a/kernel/hung_task.c b/kernel/hung_task.c
> index bb2354f73ded..74bf4cef857f 100644
> --- a/kernel/hung_task.c
> +++ b/kernel/hung_task.c
> @@ -119,7 +119,8 @@ static void check_hung_task(struct task_struct *t, unsigned long timeout)
>  	if (sysctl_hung_task_panic) {
>  		console_verbose();
>  		hung_task_show_lock = true;
> -		hung_task_call_panic = true;
> +		if (t->hung_task_detect)
> +			hung_task_call_panic = true;
>  	}
>  
>  	/*
> -- 
> 2.17.1
> 

This patch does not seems to be complete. Do you plan to provide an interface
to set/clear task_struct::hung_task_detect? Please explain the motivation and
the problems solved by this interface.

Thanks,
Pavan