Merge in the unified buffer cache code as found in NetBSD 2001/03/10. The

code is written mostly by Chuck Silvers <chuq@chuq.com>/<chs@netbsd.org>.

Tested for the past few weeks by many developers, should be in a pretty stable
state, but will require optimizations and additional cleanups.

1 files changed, 58 insertions, 63 deletions

diff --git a/sys/ufs/ffs/ffs_inode.c b/sys/ufs/ffs/ffs_inode.c
index c81c795b2ac..cddf6a368ca 100644
--- a/sys/ufs/ffs/ffs_inode.c
+++ b/sys/ufs/ffs/ffs_inode.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: ffs_inode.c,v 1.25 2001/11/21 21:23:56 csapuntz Exp $	*/
+/*	$OpenBSD: ffs_inode.c,v 1.26 2001/11/27 05:27:12 art Exp $	*/
 /*	$NetBSD: ffs_inode.c,v 1.10 1996/05/11 18:27:19 mycroft Exp $	*/
 
 /*
@@ -150,14 +150,14 @@ ffs_truncate(struct inode *oip, off_t length, int flags, struct ucred *cred)
 {
 	struct vnode *ovp;
 	daddr_t lastblock;
-	daddr_t bn, lbn, lastiblock[NIADDR], indir_lbn[NIADDR];
+	daddr_t bn, lastiblock[NIADDR], indir_lbn[NIADDR];
 	daddr_t oldblks[NDADDR + NIADDR], newblks[NDADDR + NIADDR];
 	struct fs *fs;
-	struct buf *bp;
+	struct proc *p = curproc;
 	int offset, size, level;
 	long count, nblocks, vflags, blocksreleased = 0;
 	register int i;
-	int aflags, error, allerror;
+	int error, allerror;
 	off_t osize;
 
 	if (length < 0)
@@ -188,10 +188,55 @@ ffs_truncate(struct inode *oip, off_t length, int flags, struct ucred *cred)
 	if ((error = getinoquota(oip)) != 0)
 		return (error);
 
-	uvm_vnp_setsize(ovp, length);
+	fs = oip->i_fs;
+	if (length > fs->fs_maxfilesize)
+		return (EFBIG);
+	osize = oip->i_ffs_size; 
 	oip->i_ci.ci_lasta = oip->i_ci.ci_clen 
 	    = oip->i_ci.ci_cstart = oip->i_ci.ci_lastw = 0;
 
+	/*
+	 * Lengthen the size of the file. We must ensure that the
+	 * last byte of the file is allocated. Since the smallest
+	 * value of osize is 0, length will be at least 1.
+	 */
+
+	if (osize < length) {
+		ufs_balloc_range(ovp, length - 1, 1, cred,
+		    flags & IO_SYNC ? B_SYNC : 0);
+		oip->i_flag |= IN_CHANGE | IN_UPDATE;
+		return (UFS_UPDATE(oip, 1));
+	}
+
+	/*
+	 * When truncating a regular file down to a non-block-aligned size,
+	 * we must zero the part of last block which is past the new EOF.
+	 * We must synchronously flush the zeroed pages to disk
+	 * since the new pages will be invalidated as soon as we
+	 * inform the VM system of the new, smaller size.
+	 * We must to this before acquiring the GLOCK, since fetching
+	 * the pages will acquire the GLOCK internally.
+	 * So there is a window where another thread could see a whole
+	 * zeroed page past EOF, but that's life.
+	 */
+
+	offset = blkoff(fs, length);
+	if (ovp->v_type == VREG && length < osize && offset != 0) {
+		struct uvm_object *uobj;
+		voff_t eoz;
+
+		size = blksize(fs, oip, lblkno(fs, length));
+		eoz = min(lblktosize(fs, lblkno(fs, length)) + size, osize);
+		uvm_vnp_zerorange(ovp, length, eoz - length);
+		uobj = &ovp->v_uvm.u_obj;
+		simple_lock(&uobj->vmobjlock);
+		uobj->pgops->pgo_flush(uobj, length, eoz,
+		    PGO_CLEANIT|PGO_DEACTIVATE|PGO_SYNCIO);
+		simple_unlock(&ovp->v_uvm.u_obj.vmobjlock);
+	}
+
+	lockmgr(&ovp->v_glock, LK_EXCLUSIVE, NULL, p);
+
 	if (DOINGSOFTDEP(ovp)) {
 		if (length > 0 || softdep_slowdown(ovp)) {
 			/*
@@ -204,80 +249,29 @@ ffs_truncate(struct inode *oip, off_t length, int flags, struct ucred *cred)
 			 * so that it will have no data structures left.
 			 */
 			if ((error = VOP_FSYNC(ovp, cred, MNT_WAIT,
-					       curproc)) != 0)
+			    curproc)) != 0) {
+				lockmgr(&ovp->v_glock, LK_RELEASE, NULL, p);
 				return (error);
+			}
 		} else {
+			uvm_vnp_setsize(ovp, length);
 			(void)ufs_quota_free_blocks(oip, oip->i_ffs_blocks, 
 			    NOCRED);
 			softdep_setup_freeblocks(oip, length);
 			(void) vinvalbuf(ovp, 0, cred, curproc, 0, 0);
+			lockmgr(&ovp->v_glock, LK_RELEASE, NULL, p);
 			oip->i_flag |= IN_CHANGE | IN_UPDATE;
 			return (UFS_UPDATE(oip, 0));
 		}
 	}
 
-	fs = oip->i_fs;
-	osize = oip->i_ffs_size; 
 	/*
-	 * Lengthen the size of the file. We must ensure that the
-	 * last byte of the file is allocated. Since the smallest
-	 * value of osize is 0, length will be at least 1.
+	 * Reduce the size of the file.
 	 */
-	if (osize < length) {
-		if (length > fs->fs_maxfilesize)
-			return (EFBIG);
-		aflags = B_CLRBUF;
-		if (flags & IO_SYNC)
-			aflags |= B_SYNC;
-		error = UFS_BUF_ALLOC(oip, length - 1, 1, 
-				   cred, aflags, &bp);
-		if (error)
-			return (error);
-		oip->i_ffs_size = length;
-		uvm_vnp_setsize(ovp, length);
-		(void) uvm_vnp_uncache(ovp);
-		if (aflags & B_SYNC)
-			bwrite(bp);
-		else
-			bawrite(bp);
-		oip->i_flag |= IN_CHANGE | IN_UPDATE;
-		return (UFS_UPDATE(oip, MNT_WAIT));
-	}
+	oip->i_ffs_size = length;
 	uvm_vnp_setsize(ovp, length);
 
 	/*
-	 * Shorten the size of the file. If the file is not being
-	 * truncated to a block boundary, the contents of the
-	 * partial block following the end of the file must be
-	 * zero'ed in case it ever becomes accessible again because
-	 * of subsequent file growth. Directories however are not
-	 * zero'ed as they should grow back initialized to empty.
-	 */
-	offset = blkoff(fs, length);
-	if (offset == 0) {
-		oip->i_ffs_size = length;
-	} else {
-		lbn = lblkno(fs, length);
-		aflags = B_CLRBUF;
-		if (flags & IO_SYNC)
-			aflags |= B_SYNC;
-		error = UFS_BUF_ALLOC(oip, length - 1, 1,
-				   cred, aflags, &bp);
-		if (error)
-			return (error);
-		oip->i_ffs_size = length;
-		size = blksize(fs, oip, lbn);
-		(void) uvm_vnp_uncache(ovp);
-		if (ovp->v_type != VDIR)
-			bzero((char *)bp->b_data + offset,
-			      (u_int)(size - offset));
-		allocbuf(bp, size);
-		if (aflags & B_SYNC)
-			bwrite(bp);
-		else
-			bawrite(bp);
-	}
-	/*
 	 * Calculate index into inode's block list of
 	 * last direct and indirect blocks (if any)
 	 * which we want to keep.  Lastblock is -1 when
@@ -402,6 +396,7 @@ done:
 	oip->i_ffs_blocks -= blocksreleased;
 	if (oip->i_ffs_blocks < 0)			/* sanity */
 		oip->i_ffs_blocks = 0;
+	lockmgr(&ovp->v_glock, LK_RELEASE, NULL, p);
 	oip->i_flag |= IN_CHANGE;
 	(void)ufs_quota_free_blocks(oip, blocksreleased, NOCRED);
 	return (allerror);