linux-kernel - Re: [patch] call truncate_inode_pages in the DIO fallback to buffered I/O path

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Wed, 11 Oct 2006 12:48:57 -0400
From:	jmoyer@...hat.com
To:	Andrew Morton <akpm@...l.org>
Cc:	Zach Brown <zach.brown@...cle.com>, linux-kernel@...r.kernel.org
Subject: Re: [patch] call truncate_inode_pages in the DIO fallback to buffered I/O path

==> Regarding Re: [patch] call truncate_inode_pages in the DIO fallback to buffered I/O path; Andrew Morton <akpm@...l.org> adds:

akpm> Patch is below.  The end result looks like:

akpm> 	/* coalesce the iovecs and go direct-to-BIO for O_DIRECT */
akpm> 	if (unlikely(file->f_flags & O_DIRECT)) {
akpm> 		loff_t endbyte;
akpm> 		ssize_t written_buffered;

akpm> 		written = generic_file_direct_write(iocb, iov, &nr_segs, pos,
akpm> 							ppos, count, ocount);
akpm> 		if (written < 0 || written == count)
akpm> 			goto out;
akpm> 		/*
akpm> 		 * direct-io write to a hole: fall through to buffered I/O
akpm> 		 * for completing the rest of the request.
akpm> 		 */
akpm> 		pos += written;
akpm> 		count -= written;
akpm> 		written_buffered = generic_file_buffered_write(iocb, iov,
akpm> 						nr_segs, pos, ppos, count,
akpm> 						written);

akpm> 		/*
akpm> 		 * We need to ensure that the page cache pages are written to
akpm> 		 * disk and invalidated to preserve the expected O_DIRECT
akpm> 		 * semantics.
akpm> 		 */
akpm> 		endbyte = pos + written_buffered - 1;

We probably want to handle the case where generic_file_buffered_write
returns an error or nothing written.

akpm> 		err = do_sync_file_range(file, pos, endbyte,
akpm> 					 SYNC_FILE_RANGE_WAIT_BEFORE|
akpm> 					 SYNC_FILE_RANGE_WRITE|
akpm> 					 SYNC_FILE_RANGE_WAIT_AFTER);
akpm> 		if (err == 0) {
akpm> 			written += written_buffered;
akpm> 			invalidate_mapping_pages(mapping,
akpm> 						 pos >> PAGE_CACHE_SHIFT,
akpm> 						 endbyte >> PAGE_CACHE_SHIFT);

generic_file_buffered_write takes written as an argument, and returns that
amount plus whatever it managed to write.  As such, you don't want to add
written_buffered to written.  Instead, you want written = written_buffered.
The endbyte calculation has to be altered in kind.

Incremental, locally tested patch attached.  Comments are welcome as
always.  Once there is consensus, I'll send this off for testing with
Oracle again.

-Jeff

--- linux-2.6.18.i686/mm/filemap.c.orig	2006-10-11 11:58:29.000000000 -0400
+++ linux-2.6.18.i686/mm/filemap.c	2006-10-11 12:31:11.000000000 -0400
@@ -2419,19 +2419,21 @@ __generic_file_aio_write_nolock(struct k
 		written_buffered = generic_file_buffered_write(iocb, iov,
 						nr_segs, pos, ppos, count,
 						written);
+		if (written_buffered < 0 || written_buffered == written)
+			goto out;

 		/*
 		 * We need to ensure that the page cache pages are written to
 		 * disk and invalidated to preserve the expected O_DIRECT
 		 * semantics.
 		 */
-		endbyte = pos + written_buffered - 1;
+		endbyte = pos + written_buffered - written - 1;
 		err = do_sync_file_range(file, pos, endbyte,
 					 SYNC_FILE_RANGE_WAIT_BEFORE|
 					 SYNC_FILE_RANGE_WRITE|
 					 SYNC_FILE_RANGE_WAIT_AFTER);
 		if (err == 0) {
-			written += written_buffered;
+			written = written_buffered;
 			invalidate_mapping_pages(mapping,
 						 pos >> PAGE_CACHE_SHIFT,
 						 endbyte >> PAGE_CACHE_SHIFT);
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/