On Mon, Jan 06, 2020 at 10:15:08AM -0800, Jaegeuk Kim wrote:
> > static struct page *f2fs_read_merkle_tree_page(struct inode *inode,
> > - pgoff_t index)
> > + pgoff_t index,
> > + unsigned long num_ra_pages)
> > {
> > + struct page *page;
> > +
> > index += f2fs_verity_metadata_pos(inode) >> PAGE_SHIFT;
> >
> > - return read_mapping_page(inode->i_mapping, index, NULL);
> > + page = find_get_page(inode->i_mapping, index);
> > + if (!page || !PageUptodate(page)) {
> > + if (page)
> > + put_page(page);
> > + else if (num_ra_pages > 1)
> > + f2fs_merkle_tree_readahead(inode->i_mapping, index,
> > + num_ra_pages);
> > + page = read_mapping_page(inode->i_mapping, index, NULL);
> > + if (IS_ERR(page))
> > + return page;
>
> We don't need to check this, but can use the below return page?
>
Indeed, I'll remove the unnecessary IS_ERR(page) check.
> > + }
>
> mark_page_accessed(page)?
>
> > + return page;
> > }
Good idea, but read_mapping_page() already calls mark_page_accessed(). It's
just find_get_page() that doesn't. So after this patch, mark_page_accessed() is
no longer called in the case where the page is already cached and Uptodate.
I'll change it to use:
find_get_page_flags(inode->i_mapping, index, FGP_ACCESSED);
- Eric
