From owner-svn-src-stable-11@freebsd.org Thu Feb 8 17:14:57 2018 Return-Path: Delivered-To: svn-src-stable-11@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2610:1c1:1:606c::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id 95F3BF144D2; Thu, 8 Feb 2018 17:14:57 +0000 (UTC) (envelope-from kevans@FreeBSD.org) Received: from mxrelay.nyi.freebsd.org (mxrelay.nyi.freebsd.org [IPv6:2610:1c1:1:606c::19:3]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client CN "mxrelay.nyi.freebsd.org", Issuer "Let's Encrypt Authority X3" (verified OK)) by mx1.freebsd.org (Postfix) with ESMTPS id 46E8171B21; Thu, 8 Feb 2018 17:14:57 +0000 (UTC) (envelope-from kevans@FreeBSD.org) Received: from repo.freebsd.org (repo.freebsd.org [IPv6:2610:1c1:1:6068::e6a:0]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client did not present a certificate) by mxrelay.nyi.freebsd.org (Postfix) with ESMTPS id 41C0D18EC9; Thu, 8 Feb 2018 17:14:57 +0000 (UTC) (envelope-from kevans@FreeBSD.org) Received: from repo.freebsd.org ([127.0.1.37]) by repo.freebsd.org (8.15.2/8.15.2) with ESMTP id w18HEvhB032194; Thu, 8 Feb 2018 17:14:57 GMT (envelope-from kevans@FreeBSD.org) Received: (from kevans@localhost) by repo.freebsd.org (8.15.2/8.15.2/Submit) id w18HEuWI032186; Thu, 8 Feb 2018 17:14:56 GMT (envelope-from kevans@FreeBSD.org) Message-Id: <201802081714.w18HEuWI032186@repo.freebsd.org> X-Authentication-Warning: repo.freebsd.org: kevans set sender to kevans@FreeBSD.org using -f From: Kyle Evans Date: Thu, 8 Feb 2018 17:14:56 +0000 (UTC) To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-stable@freebsd.org, svn-src-stable-11@freebsd.org Subject: svn commit: r329020 - in stable/11/sys: fs/ext2fs kern sys ufs/ffs vm X-SVN-Group: stable-11 X-SVN-Commit-Author: kevans X-SVN-Commit-Paths: in stable/11/sys: fs/ext2fs kern sys ufs/ffs vm X-SVN-Commit-Revision: 329020 X-SVN-Commit-Repository: base MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-BeenThere: svn-src-stable-11@freebsd.org X-Mailman-Version: 2.1.25 Precedence: list List-Id: SVN commit messages for only the 11-stable src tree List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 08 Feb 2018 17:14:57 -0000 Author: kevans Date: Thu Feb 8 17:14:56 2018 New Revision: 329020 URL: https://svnweb.freebsd.org/changeset/base/329020 Log: MFC r309062: Release laundered vnode pages to the head of the inactive queue. The swap pager enqueues laundered pages near the head of the inactive queue to avoid another trip through LRU before reclamation. This change adds support for this behaviour to the vnode pager and makes use of it in UFS and ext2fs. Some ioflag handling is consolidated into a common subroutine so that this support can be easily extended to other filesystems which make use of the buffer cache. No changes are needed for ZFS since its putpages routine always undirties the pages before returning, and the laundry thread requeues the pages appropriately in this case. Modified: stable/11/sys/fs/ext2fs/ext2_vnops.c stable/11/sys/kern/vfs_bio.c stable/11/sys/sys/buf.h stable/11/sys/sys/vnode.h stable/11/sys/ufs/ffs/ffs_vnops.c stable/11/sys/vm/vm_pageout.c stable/11/sys/vm/vm_pager.h stable/11/sys/vm/vnode_pager.c Directory Properties: stable/11/ (props changed) Modified: stable/11/sys/fs/ext2fs/ext2_vnops.c ============================================================================== --- stable/11/sys/fs/ext2fs/ext2_vnops.c Thu Feb 8 17:07:27 2018 (r329019) +++ stable/11/sys/fs/ext2fs/ext2_vnops.c Thu Feb 8 17:14:56 2018 (r329020) @@ -2061,15 +2061,6 @@ ext2_ind_read(struct vop_read_args *ap) } /* - * If IO_DIRECT then set B_DIRECT for the buffer. This - * will cause us to attempt to release the buffer later on - * and will cause the buffer cache to attempt to free the - * underlying pages. - */ - if (ioflag & IO_DIRECT) - bp->b_flags |= B_DIRECT; - - /* * We should only get non-zero b_resid when an I/O error * has occurred, which should cause us to break above. * However, if the short read did not cause an error, @@ -2086,25 +2077,7 @@ ext2_ind_read(struct vop_read_args *ap) (int)xfersize, uio); if (error) break; - - if (ioflag & (IO_VMIO|IO_DIRECT)) { - /* - * If it's VMIO or direct I/O, then we don't - * need the buf, mark it available for - * freeing. If it's non-direct VMIO, the VM has - * the data. - */ - bp->b_flags |= B_RELBUF; - brelse(bp); - } else { - /* - * Otherwise let whoever - * made the request take care of - * freeing it. We just queue - * it onto another list. - */ - bqrelse(bp); - } + vfs_bio_brelse(bp, ioflag); } /* @@ -2113,14 +2086,8 @@ ext2_ind_read(struct vop_read_args *ap) * completion has not set a new value into it. so it must have come * from a 'break' statement */ - if (bp != NULL) { - if (ioflag & (IO_VMIO|IO_DIRECT)) { - bp->b_flags |= B_RELBUF; - brelse(bp); - } else { - bqrelse(bp); - } - } + if (bp != NULL) + vfs_bio_brelse(bp, ioflag); if ((error == 0 || uio->uio_resid != orig_resid) && (vp->v_mount->mnt_flag & (MNT_NOATIME | MNT_RDONLY)) == 0) @@ -2378,9 +2345,8 @@ ext2_write(struct vop_write_args *ap) if (error != 0 && (bp->b_flags & B_CACHE) == 0 && fs->e2fs_bsize == xfersize) vfs_bio_clrbuf(bp); - if (ioflag & (IO_VMIO|IO_DIRECT)) { - bp->b_flags |= B_RELBUF; - } + + vfs_bio_set_flags(bp, ioflag); /* * If IO_SYNC each buffer is written synchronously. Otherwise Modified: stable/11/sys/kern/vfs_bio.c ============================================================================== --- stable/11/sys/kern/vfs_bio.c Thu Feb 8 17:07:27 2018 (r329019) +++ stable/11/sys/kern/vfs_bio.c Thu Feb 8 17:14:56 2018 (r329020) @@ -4452,6 +4452,45 @@ vfs_bio_bzero_buf(struct buf *bp, int base, int size) } /* + * Update buffer flags based on I/O request parameters, optionally releasing the + * buffer. If it's VMIO or direct I/O, the buffer pages are released to the VM, + * where they may be placed on a page queue (VMIO) or freed immediately (direct + * I/O). Otherwise the buffer is released to the cache. + */ +static void +b_io_dismiss(struct buf *bp, int ioflag, bool release) +{ + + KASSERT((ioflag & IO_NOREUSE) == 0 || (ioflag & IO_VMIO) != 0, + ("buf %p non-VMIO noreuse", bp)); + + if ((ioflag & IO_DIRECT) != 0) + bp->b_flags |= B_DIRECT; + if ((ioflag & (IO_VMIO | IO_DIRECT)) != 0 && LIST_EMPTY(&bp->b_dep)) { + bp->b_flags |= B_RELBUF; + if ((ioflag & IO_NOREUSE) != 0) + bp->b_flags |= B_NOREUSE; + if (release) + brelse(bp); + } else if (release) + bqrelse(bp); +} + +void +vfs_bio_brelse(struct buf *bp, int ioflag) +{ + + b_io_dismiss(bp, ioflag, true); +} + +void +vfs_bio_set_flags(struct buf *bp, int ioflag) +{ + + b_io_dismiss(bp, ioflag, false); +} + +/* * vm_hold_load_pages and vm_hold_free_pages get pages into * a buffers address space. The pages are anonymous and are * not associated with a file object. Modified: stable/11/sys/sys/buf.h ============================================================================== --- stable/11/sys/sys/buf.h Thu Feb 8 17:07:27 2018 (r329019) +++ stable/11/sys/sys/buf.h Thu Feb 8 17:14:56 2018 (r329020) @@ -526,9 +526,11 @@ int cluster_read(struct vnode *, u_quad_t, daddr_t, lo struct ucred *, long, int, int, struct buf **); int cluster_wbuild(struct vnode *, long, daddr_t, int, int); void cluster_write(struct vnode *, struct buf *, u_quad_t, int, int); +void vfs_bio_brelse(struct buf *bp, int ioflags); void vfs_bio_bzero_buf(struct buf *bp, int base, int size); -void vfs_bio_set_valid(struct buf *, int base, int size); void vfs_bio_clrbuf(struct buf *); +void vfs_bio_set_flags(struct buf *bp, int ioflags); +void vfs_bio_set_valid(struct buf *, int base, int size); void vfs_busy_pages(struct buf *, int clear_modify); void vfs_unbusy_pages(struct buf *); int vmapbuf(struct buf *, int); Modified: stable/11/sys/sys/vnode.h ============================================================================== --- stable/11/sys/sys/vnode.h Thu Feb 8 17:07:27 2018 (r329019) +++ stable/11/sys/sys/vnode.h Thu Feb 8 17:14:56 2018 (r329020) @@ -303,6 +303,7 @@ struct vattr { #define IO_INVAL 0x0040 /* invalidate after I/O */ #define IO_SYNC 0x0080 /* do I/O synchronously */ #define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */ +#define IO_NOREUSE 0x0200 /* VMIO data won't be reused */ #define IO_EXT 0x0400 /* operate on external attributes */ #define IO_NORMAL 0x0800 /* operate on regular data */ #define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */ Modified: stable/11/sys/ufs/ffs/ffs_vnops.c ============================================================================== --- stable/11/sys/ufs/ffs/ffs_vnops.c Thu Feb 8 17:07:27 2018 (r329019) +++ stable/11/sys/ufs/ffs/ffs_vnops.c Thu Feb 8 17:14:56 2018 (r329020) @@ -601,15 +601,6 @@ ffs_read(ap) } /* - * If IO_DIRECT then set B_DIRECT for the buffer. This - * will cause us to attempt to release the buffer later on - * and will cause the buffer cache to attempt to free the - * underlying pages. - */ - if (ioflag & IO_DIRECT) - bp->b_flags |= B_DIRECT; - - /* * We should only get non-zero b_resid when an I/O error * has occurred, which should cause us to break above. * However, if the short read did not cause an error, @@ -633,25 +624,7 @@ ffs_read(ap) if (error) break; - if ((ioflag & (IO_VMIO|IO_DIRECT)) && - (LIST_EMPTY(&bp->b_dep))) { - /* - * If there are no dependencies, and it's VMIO, - * then we don't need the buf, mark it available - * for freeing. For non-direct VMIO reads, the VM - * has the data. - */ - bp->b_flags |= B_RELBUF; - brelse(bp); - } else { - /* - * Otherwise let whoever - * made the request take care of - * freeing it. We just queue - * it onto another list. - */ - bqrelse(bp); - } + vfs_bio_brelse(bp, ioflag); } /* @@ -660,15 +633,8 @@ ffs_read(ap) * and on normal completion has not set a new value into it. * so it must have come from a 'break' statement */ - if (bp != NULL) { - if ((ioflag & (IO_VMIO|IO_DIRECT)) && - (LIST_EMPTY(&bp->b_dep))) { - bp->b_flags |= B_RELBUF; - brelse(bp); - } else { - bqrelse(bp); - } - } + if (bp != NULL) + vfs_bio_brelse(bp, ioflag); if ((error == 0 || uio->uio_resid != orig_resid) && (vp->v_mount->mnt_flag & (MNT_NOATIME | MNT_RDONLY)) == 0 && @@ -786,8 +752,6 @@ ffs_write(ap) vnode_pager_setsize(vp, ip->i_size); break; } - if (ioflag & IO_DIRECT) - bp->b_flags |= B_DIRECT; if ((ioflag & (IO_SYNC|IO_INVAL)) == (IO_SYNC|IO_INVAL)) bp->b_flags |= B_NOCACHE; @@ -827,11 +791,9 @@ ffs_write(ap) if (error != 0 && (bp->b_flags & B_CACHE) == 0 && fs->fs_bsize == xfersize) vfs_bio_clrbuf(bp); - if ((ioflag & (IO_VMIO|IO_DIRECT)) && - (LIST_EMPTY(&bp->b_dep))) { - bp->b_flags |= B_RELBUF; - } + vfs_bio_set_flags(bp, ioflag); + /* * If IO_SYNC each buffer is written synchronously. Otherwise * if we have a severe page deficiency write the buffer @@ -978,15 +940,6 @@ ffs_extread(struct vnode *vp, struct uio *uio, int iof } /* - * If IO_DIRECT then set B_DIRECT for the buffer. This - * will cause us to attempt to release the buffer later on - * and will cause the buffer cache to attempt to free the - * underlying pages. - */ - if (ioflag & IO_DIRECT) - bp->b_flags |= B_DIRECT; - - /* * We should only get non-zero b_resid when an I/O error * has occurred, which should cause us to break above. * However, if the short read did not cause an error, @@ -1004,26 +957,7 @@ ffs_extread(struct vnode *vp, struct uio *uio, int iof (int)xfersize, uio); if (error) break; - - if ((ioflag & (IO_VMIO|IO_DIRECT)) && - (LIST_EMPTY(&bp->b_dep))) { - /* - * If there are no dependencies, and it's VMIO, - * then we don't need the buf, mark it available - * for freeing. For non-direct VMIO reads, the VM - * has the data. - */ - bp->b_flags |= B_RELBUF; - brelse(bp); - } else { - /* - * Otherwise let whoever - * made the request take care of - * freeing it. We just queue - * it onto another list. - */ - bqrelse(bp); - } + vfs_bio_brelse(bp, ioflag); } /* @@ -1032,15 +966,8 @@ ffs_extread(struct vnode *vp, struct uio *uio, int iof * and on normal completion has not set a new value into it. * so it must have come from a 'break' statement */ - if (bp != NULL) { - if ((ioflag & (IO_VMIO|IO_DIRECT)) && - (LIST_EMPTY(&bp->b_dep))) { - bp->b_flags |= B_RELBUF; - brelse(bp); - } else { - bqrelse(bp); - } - } + if (bp != NULL) + vfs_bio_brelse(bp, ioflag); return (error); } @@ -1109,8 +1036,6 @@ ffs_extwrite(struct vnode *vp, struct uio *uio, int io */ if ((bp->b_flags & B_CACHE) == 0 && fs->fs_bsize <= xfersize) vfs_bio_clrbuf(bp); - if (ioflag & IO_DIRECT) - bp->b_flags |= B_DIRECT; if (uio->uio_offset + xfersize > dp->di_extsize) dp->di_extsize = uio->uio_offset + xfersize; @@ -1121,10 +1046,8 @@ ffs_extwrite(struct vnode *vp, struct uio *uio, int io error = uiomove((char *)bp->b_data + blkoffset, (int)xfersize, uio); - if ((ioflag & (IO_VMIO|IO_DIRECT)) && - (LIST_EMPTY(&bp->b_dep))) { - bp->b_flags |= B_RELBUF; - } + + vfs_bio_set_flags(bp, ioflag); /* * If IO_SYNC each buffer is written synchronously. Otherwise Modified: stable/11/sys/vm/vm_pageout.c ============================================================================== --- stable/11/sys/vm/vm_pageout.c Thu Feb 8 17:07:27 2018 (r329019) +++ stable/11/sys/vm/vm_pageout.c Thu Feb 8 17:14:56 2018 (r329020) @@ -441,8 +441,8 @@ more: if (ib != 0 && pageout_count < vm_pageout_page_count) goto more; - return (vm_pageout_flush(&mc[page_base], pageout_count, 0, 0, NULL, - NULL)); + return (vm_pageout_flush(&mc[page_base], pageout_count, + VM_PAGER_PUT_NOREUSE, 0, NULL, NULL)); } /* Modified: stable/11/sys/vm/vm_pager.h ============================================================================== --- stable/11/sys/vm/vm_pager.h Thu Feb 8 17:07:27 2018 (r329019) +++ stable/11/sys/vm/vm_pager.h Thu Feb 8 17:14:56 2018 (r329020) @@ -98,6 +98,7 @@ extern struct pagerops mgtdevicepagerops; #define VM_PAGER_PUT_SYNC 0x0001 #define VM_PAGER_PUT_INVAL 0x0002 +#define VM_PAGER_PUT_NOREUSE 0x0004 #define VM_PAGER_CLUSTER_OK 0x0008 #ifdef _KERNEL Modified: stable/11/sys/vm/vnode_pager.c ============================================================================== --- stable/11/sys/vm/vnode_pager.c Thu Feb 8 17:07:27 2018 (r329019) +++ stable/11/sys/vm/vnode_pager.c Thu Feb 8 17:14:56 2018 (r329020) @@ -1366,9 +1366,7 @@ vnode_pager_putpages_ioflags(int pager_flags) else if ((pager_flags & VM_PAGER_CLUSTER_OK) == 0) ioflags |= IO_ASYNC; ioflags |= (pager_flags & VM_PAGER_PUT_INVAL) != 0 ? IO_INVAL: 0; -#ifdef notyet ioflags |= (pager_flags & VM_PAGER_PUT_NOREUSE) != 0 ? IO_NOREUSE : 0; -#endif ioflags |= IO_SEQMAX << IO_SEQSHIFT; return (ioflags); }