[src/tls-maxphys]: src/sys/ufs/ufs In the FFS writebehind code (ufs_readwrite...

[tls <tls%NetBSD.org@localhost>]

details:   https://anonhg.NetBSD.org/src/rev/f7d71f411b4a
branches:  tls-maxphys
changeset: 852931:f7d71f411b4a
user:      tls <tls%NetBSD.org@localhost>
date:      Sun Oct 14 14:33:32 2012 +0000

description:
In the FFS writebehind code (ufs_readwrite.c:WRITE()), use division
and multiplication instead of shifts, to accomodate devices with
MAXPHYS that is a multiple of the page size, but not a power of 2.

MegaRAID neatly writes out 192k chunks now.

An open question: is is really a good idea to always writebehind
at the largest size supported by the device?  Likely not, as this
could have a major impact on I/O fairness.  OS X and Solaris both
seem to limit transfers to 128k likely for this reason (the same
problem exists with the readahead code but since it is adaptive,
it will not *always* do huge transfers).

However, simply imposing a smallish limit like 128k here seems
like a bad idea because then we cannot accomodate greedy devices
like RAID, for which you want something like 128k * number of
data components.  Hmmmmmm.

diffstat:

 sys/ufs/ufs/ufs_readwrite.c |  27 ++++++++++++++++++---------
 1 files changed, 18 insertions(+), 9 deletions(-)

diffs (48 lines):

diff -r 934e2a0fe686 -r f7d71f411b4a sys/ufs/ufs/ufs_readwrite.c
--- a/sys/ufs/ufs/ufs_readwrite.c       Thu Oct 11 17:09:56 2012 +0000
+++ b/sys/ufs/ufs/ufs_readwrite.c       Sun Oct 14 14:33:32 2012 +0000
@@ -1,4 +1,4 @@
-/*     $NetBSD: ufs_readwrite.c,v 1.104 2012/04/29 22:54:01 chs Exp $  */
+/*     $NetBSD: ufs_readwrite.c,v 1.104.2.1 2012/10/14 14:33:32 tls Exp $      */
 
 /*-
  * Copyright (c) 1993
@@ -32,7 +32,7 @@
  */
 
 #include <sys/cdefs.h>
-__KERNEL_RCSID(1, "$NetBSD: ufs_readwrite.c,v 1.104 2012/04/29 22:54:01 chs Exp $");
+__KERNEL_RCSID(1, "$NetBSD: ufs_readwrite.c,v 1.104.2.1 2012/10/14 14:33:32 tls Exp $");
 
 #ifdef LFS_READWRITE
 #define        FS                      struct lfs
@@ -414,13 +414,22 @@
                 */
 
 #ifndef LFS_READWRITE
-               if (!async && oldoff >> 16 != uio->uio_offset >> 16) {
-                       mutex_enter(vp->v_interlock);
-                       error = VOP_PUTPAGES(vp, (oldoff >> 16) << 16,
-                           (uio->uio_offset >> 16) << 16,
-                           PGO_CLEANIT | PGO_JOURNALLOCKED | PGO_LAZY);
-                       if (error)
-                               break;
+               {
+                       int maximum = vp->v_mount->mnt_maxphys;
+                       off_t oldchunk, newchunk;
+
+                       oldchunk = (oldoff / maximum) * maximum;
+                       newchunk = (uio->uio_offset / maximum) * maximum;
+
+                       if (!async && oldchunk != newchunk) {
+                               mutex_enter(vp->v_interlock);
+                               error = VOP_PUTPAGES(vp, oldchunk, newchunk,
+                                                    PGO_CLEANIT |
+                                                    PGO_JOURNALLOCKED |
+                                                    PGO_LAZY);
+                               if (error)
+                                       break;
+                       }
                }
 #endif
        }