Re: [RFC PATCH] mm: track bad page via kmemleak

[Catalin Marinas]

(sorry, I missed this last month when it was posted)

On Tue, Sep 20, 2022 at 09:11:08PM +0800, zhaoyang.huang wrote:
> From: Zhaoyang Huang <zhaoyang.huang@xxxxxxxxxx>
> 
> Bad pages will failed go back to allocator and leaved as orphan pages, track
> them down via kmemleak.
> 
> Signed-off-by: Zhaoyang Huang <zhaoyang.huang@xxxxxxxxxx>
> ---
>  mm/page_alloc.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index e5486d4..24f682e 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -1408,7 +1408,7 @@ static __always_inline bool free_pages_prepare(struct page *page,
>  			__memcg_kmem_uncharge_page(page, order);
>  		reset_page_owner(page, order);
>  		page_table_check_free(page, order);
> -		return false;
> +		goto err;
>  	}
>  
>  	/*
> @@ -1442,7 +1442,7 @@ static __always_inline bool free_pages_prepare(struct page *page,
>  	if (check_free)
>  		bad += check_free_page(page);
>  	if (bad)
> -		return false;
> +		goto err;
>  
>  	page_cpupid_reset_last(page);
>  	page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
> @@ -1486,6 +1486,10 @@ static __always_inline bool free_pages_prepare(struct page *page,
>  	debug_pagealloc_unmap_pages(page, 1 << order);
>  
>  	return true;
> +err:
> +	kmemleak_alloc(page_address(page), PAGE_SIZE << order, 1, GFP_KERNEL);
> +	return false;
> +
>  }

What's the aim of this tracking? First of all, that's a HWPoison page
and it should note be touched, so the min_count should be -1 otherwise
kmemleak will try to read it for references to other objects.

Now, if all you need is a list of the bad pages, I don't think that's
kmemleak's job. Maybe add them to a list and expose it somewhere else
via debugfs.

-- 
Catalin