Re: [PATCH 1/2] mm/memory-failure: add panic_on_unrecoverable_memory_failure sysctl

[Miaohe Lin]

On 2026/3/30 21:45, Breno Leitao wrote:
> On Mon, Mar 30, 2026 at 03:55:00PM +0800, Miaohe Lin wrote:
>> On 2026/3/23 23:29, Breno Leitao wrote:
>>
>>> @@ -1298,6 +1309,10 @@ 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 (sysctl_panic_on_unrecoverable_mf &&
>>> +	    type == MF_MSG_GET_HWPOISON && result == MF_IGNORED)
>>> +		panic("Memory failure: %#lx: unrecoverable page", pfn);
>>
>> MF_MSG_GET_HWPOISON contains some other scenarios. For example, an isolated folio will
>> make get_hwpoison_page return -EIO so we will see MF_MSG_GET_HWPOISON and MF_IGNORED in
>> action_result. But that's recoverable if folio is used by userspace thus panic will be
>> unacceptable.
>> Will it better to check type against MF_MSG_KERNEL_HIGH_ORDER?
> 
> Yes, I was discussing this with akpm, and maybe the better
> approach would be to panic for types MF_MSG_KERNEL_HIGH_ORDER and MF_MSG_KERNEL.
> 
> In both cases, it seems that, the page would not be able to migrate. What do
> you think about a change like this:
> 
> 
> @@ -1298,6 +1309,10 @@ 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 (sysctl_panic_on_unrecoverable_mf && result == MF_IGNORED &&
> +           (type == MF_MSG_KERNEL || type == MF_MSG_KERNEL_HIGH_ORDER))
> +               panic("Memory failure: %#lx: unrecoverable page", pfn);
> +
>         return (result == MF_RECOVERED || result == MF_DELAYED) ? 0 : -EBUSY;
>  }
> 

Maybe MF_MSG_UNKNOWN can also be considered? Kernel can't do anything further
for those folios.
BTW I think current code can't reach to MF_MSG_KERNEL and MF_MSG_UNKNOWN cases
bacause there is always a (PageHuge() || HWPoisonHandlable()) check before calling
identify_page_state.

Thanks.
.