Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Wed, 6 Jul 2022 04:21:12 +0100
From:   Matthew Wilcox <willy@infradead.org>
To:     Liu Shixin <liushixin2@huawei.com>
Cc:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Jan Kara <jack@suse.cz>,
        William Kucharski <william.kucharski@oracle.com>,
        Christoph Hellwig <hch@lst.de>, linux-kernel@vger.kernel.org,
        stable@vger.kernel.org
Subject: Re: [PATCH 5.15] mm/filemap: fix UAF in find_lock_entries
Message-ID: <YsT/qAPruIimH3+R@casper.infradead.org>
References: <20220706032434.579610-1-liushixin2@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220706032434.579610-1-liushixin2@huawei.com>
Precedence: bulk

On Wed, Jul 06, 2022 at 11:24:34AM +0800, Liu Shixin wrote:
> Release refcount after xas_set to fix UAF which may cause panic like this:

I think we can do better.  How about this?

diff --git a/mm/filemap.c b/mm/filemap.c
index 00e391e75880..11ae38cc4fd3 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -2090,7 +2090,9 @@ unsigned find_lock_entries(struct address_space *mapping, pgoff_t start,
 
 	rcu_read_lock();
 	while ((page = find_get_entry(&xas, end, XA_PRESENT))) {
+		unsigned long next_idx = xas.xa_index + 1;
 		if (!xa_is_value(page)) {
+			next_idx = page->index + thp_nr_pages(page);
 			if (page->index < start)
 				goto put;
 			if (page->index + thp_nr_pages(page) - 1 > end)
@@ -2111,14 +2113,11 @@ unsigned find_lock_entries(struct address_space *mapping, pgoff_t start,
 put:
 		put_page(page);
 next:
-		if (!xa_is_value(page) && PageTransHuge(page)) {
-			unsigned int nr_pages = thp_nr_pages(page);
-
-			/* Final THP may cross MAX_LFS_FILESIZE on 32-bit */
-			xas_set(&xas, page->index + nr_pages);
-			if (xas.xa_index < nr_pages)
-				break;
-		}
+		/* Final THP may cross MAX_LFS_FILESIZE on 32-bit */
+		if (next_idx < xas.xa_index)
+			break;
+		if (next_idx != xas.xa_index + 1)
+			xas_set(&xas, next_idx);
 	}
 	rcu_read_unlock();