Re: [PATCH v2 2/3] mm/memory-failure: add panic_on_unrecoverable_memory_failure sysctl

[Miaohe Lin]

On 2026/4/10 22:17, Breno Leitao wrote:
> On Tue, Apr 07, 2026 at 10:57:36AM +0800, Miaohe Lin wrote:
>> On 2026/3/31 19:00, Breno Leitao wrote:
>>> +	if (sysctl_panic_on_unrecoverable_mf && result == MF_IGNORED &&
>>> +	    (type == MF_MSG_KERNEL || type == MF_MSG_KERNEL_HIGH_ORDER ||
>>> +	     type == MF_MSG_UNKNOWN))
>>> +		panic("Memory failure: %#lx: unrecoverable page", pfn);
>>
>> Will it be better to add a helper here?
> 
> Yes, a helper would make things easier to read and digest. Thanks for
> the feedback. This is what I have in mind:
> 
> commit 36d5b3cbbe6d6abfe3296b7b21135a5f01e743eb
> Author: Breno Leitao <leitao@xxxxxxxxxx>
> Date:   Mon Mar 23 08:00:29 2026 -0700
> 
>     mm/memory-failure: add panic_on_unrecoverable_memory_failure sysctl
>     
>     Add a sysctl that allows the system to panic when an unrecoverable
>     memory failure is detected. This covers kernel pages, high-order
>     kernel pages, and unknown page types that cannot be recovered.
>     
>     Signed-off-by: Breno Leitao <leitao@xxxxxxxxxx>
> 
> diff --git a/mm/memory-failure.c b/mm/memory-failure.c
> index 6ff80e01b91a4..a29b6688fe2d3 100644
> --- a/mm/memory-failure.c
> +++ b/mm/memory-failure.c
> @@ -74,6 +74,8 @@ static int sysctl_memory_failure_recovery __read_mostly = 1;
>  
>  static int sysctl_enable_soft_offline __read_mostly = 1;
>  
> +static int sysctl_panic_on_unrecoverable_mf __read_mostly;
> +
>  atomic_long_t num_poisoned_pages __read_mostly = ATOMIC_LONG_INIT(0);
>  
>  static bool hw_memory_failure __read_mostly = false;
> @@ -155,6 +157,15 @@ static const struct ctl_table memory_failure_table[] = {
>  		.proc_handler	= proc_dointvec_minmax,
>  		.extra1		= SYSCTL_ZERO,
>  		.extra2		= SYSCTL_ONE,
> +	},
> +	{
> +		.procname	= "panic_on_unrecoverable_memory_failure",
> +		.data		= &sysctl_panic_on_unrecoverable_mf,
> +		.maxlen		= sizeof(sysctl_panic_on_unrecoverable_mf),
> +		.mode		= 0644,
> +		.proc_handler	= proc_dointvec_minmax,
> +		.extra1		= SYSCTL_ZERO,
> +		.extra2		= SYSCTL_ONE,
>  	}
>  };
>  
> @@ -1281,6 +1292,16 @@ static void update_per_node_mf_stats(unsigned long pfn,
>  	++mf_stats->total;
>  }
>  
> +static bool is_unrecoverable_memory_failure(enum mf_action_page_type type,
> +					    enum mf_result result)

Thanks for your update.

> +{
> +	return sysctl_panic_on_unrecoverable_mf &&
> +	       result == MF_IGNORED &&
> +	       (type == MF_MSG_KERNEL ||
> +		type == MF_MSG_KERNEL_HIGH_ORDER ||
> +		type == MF_MSG_UNKNOWN);
> +}
> +
>  /*
>   * "Dirty/Clean" indication is not 100% accurate due to the possibility of
>   * setting PG_dirty outside page lock. See also comment above set_page_dirty().
> @@ -1298,6 +1319,9 @@ static int action_result(unsigned long pfn, enum mf_action_page_type type,
>  	pr_err("%#lx: recovery action for %s: %s\n",
>  		pfn, action_page_types[type], action_name[result]);
>  
> +	if (is_unrecoverable_memory_failure(type, result))

Would it be better to name it as panic_on_unrecoverable_mf() or something like it?
This function determines whether panic on the specified memory error.

Thanks.
.