From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id BC63B2590 for ; Sun, 22 Jan 2023 15:23:40 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id 4134FC433EF; Sun, 22 Jan 2023 15:23:40 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linuxfoundation.org; s=korg; t=1674401020; bh=1KtQN+MgJAvaFzVqinQwmmmEZxWYgFL40cvB5Mu8f3U=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=v3PhlrexscdcRI+lqU1MdP8hyLo8YozsRRb6YBB/XlJzxNpPjyrk/SA6ddPUukMzM PBUh/KoutQhFG93a2MyXUgg5XX1G51E3pZdjxvVU1o8o5+cIy32h5Ww6p+v9ZqKKhd 0AHf+FrrRTJ0T2fo/GbMEQM3DSym09Op0BiKZlA8= From: Greg Kroah-Hartman To: stable@vger.kernel.org Cc: Greg Kroah-Hartman , patches@lists.linux.dev, Zach OKeefe , Hugh Dickins , Yang Shi , Andrew Morton Subject: [PATCH 6.1 080/193] mm/MADV_COLLAPSE: dont expand collapse when vm_end is past requested end Date: Sun, 22 Jan 2023 16:03:29 +0100 Message-Id: <20230122150250.021789003@linuxfoundation.org> X-Mailer: git-send-email 2.39.1 In-Reply-To: <20230122150246.321043584@linuxfoundation.org> References: <20230122150246.321043584@linuxfoundation.org> User-Agent: quilt/0.67 Precedence: bulk X-Mailing-List: patches@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit From: Zach O'Keefe commit 52dc031088f00e323140ece4004e70c33153c6dd upstream. MADV_COLLAPSE acts on one hugepage-aligned/sized region at a time, until it has collapsed all eligible memory contained within the bounds supplied by the user. At the top of each hugepage iteration we (re)lock mmap_lock and (re)validate the VMA for eligibility and update variables that might have changed while mmap_lock was dropped. One thing that might occur is that the VMA could be resized, and as such, we refetch vma->vm_end to make sure we don't collapse past the end of the VMA's new end. However, it's possible that when refetching vma->vm_end that we expand the region acted on by MADV_COLLAPSE if vma->vm_end is greater than size+len supplied by the user. The consequence here is that we may attempt to collapse more memory than requested, possibly yielding either "too much success" or "false failure" user-visible results. An example of the former is if we MADV_COLLAPSE the first 4MiB of a 2TiB mmap()'d file, the incorrect refetch would cause the operation to block for much longer than anticipated as we attempt to collapse the entire TiB region. An example of the latter is that applying MADV_COLLPSE to a 4MiB file mapped to the start of a 6MiB VMA will successfully collapse the first 4MiB, then incorrectly attempt to collapse the last hugepage-aligned/sized region -- fail (since readahead/page cache lookup will fail) -- and report a failure to the user. I don't believe there is a kernel stability concern here as we always (re)validate the VMA / region accordingly. Also as Hugh mentions, the user-visible effects are: we try to collapse more memory than requested by the user, and/or failing an operation that should have otherwise succeeded. An example is trying to collapse a 4MiB file contained within a 12MiB VMA. Don't expand the acted-on region when refetching vma->vm_end. Link: https://lkml.kernel.org/r/20221224082035.3197140-1-zokeefe@google.com Fixes: 4d24de9425f7 ("mm: MADV_COLLAPSE: refetch vm_end after reacquiring mmap_lock") Signed-off-by: Zach O'Keefe Reported-by: Hugh Dickins Cc: Yang Shi Cc: Signed-off-by: Andrew Morton Signed-off-by: Greg Kroah-Hartman --- mm/khugepaged.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) --- a/mm/khugepaged.c +++ b/mm/khugepaged.c @@ -2644,7 +2644,7 @@ int madvise_collapse(struct vm_area_stru goto out_nolock; } - hend = vma->vm_end & HPAGE_PMD_MASK; + hend = min(hend, vma->vm_end & HPAGE_PMD_MASK); } mmap_assert_locked(mm); memset(cc->node_load, 0, sizeof(cc->node_load));