Re: [PATCH v2] mm/huge_memory: don't unpoison huge_zero_folio

[Miaohe Lin]

On 2024/5/16 16:26, Oscar Salvador wrote:
> On Wed, May 15, 2024 at 05:55:39PM +0200, David Hildenbrand wrote:
>>> +	if (is_huge_zero_folio(folio)) {
>>> +		unpoison_pr_info("Unpoison: huge zero page is not supported %#lx\n",
>>> +				 pfn, &unpoison_rs);
>>> +		goto unlock_mutex;
>>> +	}
>>> +
> 
> Sorry for spamming your reply David, but for some unknown reason I am not able
> to find the original patch in my mailbox, in none of the two accountes I am
> subscribed, so I guess I will have to reply here.
> 
> Just two things
> 
> 1) We do not care if someone grabs a refcount for huge_zero_folio,
>    because since it is not supported anyway the outcome will not change.
>    Also, AFAIK, there is no chance we can unpoison that folio.
>    Therefore, I would just lift the check two blocks and place it right after
>    the hw_memory_failure check.
> 
> 2) The whole thing is unsupported, but you will return -EBUSY while you
>    should be returning -EOPNOTSUPP AFAICS.
> 

Thanks for your comment. Do you mean something like below?

diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 16ada4fb02b7..a9fe9eda593f 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -2546,6 +2546,13 @@ int unpoison_memory(unsigned long pfn)
                goto unlock_mutex;
        }

+       if (is_huge_zero_folio(folio)) {
+               unpoison_pr_info("Unpoison: huge zero page is not supported %#lx\n",
+                                pfn, &unpoison_rs);
+               ret = -EOPNOTSUPP;
+               goto unlock_mutex;
+       }
+
        if (!PageHWPoison(p)) {
                unpoison_pr_info("Unpoison: Page was already unpoisoned %#lx\n",
                                 pfn, &unpoison_rs);

Thanks.
.

> with that you can add:
> 
> Reviewed-by: Oscar Salvador <osalvador@xxxxxxx>
> 
>  
>