On 05.06.23 09:21, Christoph Hellwig wrote:
> free_page takes the virtual address of the memory to be freed, and
> does so as an unsigned long just to make things confusing. Use
> put_page instead, which actually works on the page pointer.
>
> Note that this is a reason why this should have used __bio_add_page
> instead for this impossible to hit case..
>
> Fixes: 2f9848178cfa ("md: raid1: use __bio_add_page for adding single page to bio")
> Signed-off-by: Christoph Hellwig <hch@xxxxxx>
> ---
> drivers/md/raid1.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
> index ff89839455ec11..3570da63969b58 100644
> --- a/drivers/md/raid1.c
> +++ b/drivers/md/raid1.c
> @@ -1148,7 +1148,7 @@ static void alloc_behind_master_bio(struct r1bio *r1_bio,
> goto free_pages;
>
> if (!bio_add_page(behind_bio, page, len, 0)) {
> - free_page(page);
> + put_page(page);
> goto free_pages;
> }
>
That should already be fixed in Jens' tree.
