xref: /dragonfly/share/man/man9/buf.9 (revision 6bd457ed)
1.\" Copyright (c) 1998
2.\"	The Regents of the University of California.  All rights reserved.
3.\"
4.\" Redistribution and use in source and binary forms, with or without
5.\" modification, are permitted provided that the following conditions
6.\" are met:
7.\" 1. Redistributions of source code must retain the above copyright
8.\"    notice, this list of conditions and the following disclaimer.
9.\" 2. Redistributions in binary form must reproduce the above copyright
10.\"    notice, this list of conditions and the following disclaimer in the
11.\"    documentation and/or other materials provided with the distribution.
12.\" 3. All advertising materials mentioning features or use of this software
13.\"    must display the following acknowledgement:
14.\"	This product includes software developed by the University of
15.\"	California, Berkeley and its contributors.
16.\" 4. Neither the name of the University nor the names of its contributors
17.\"    may be used to endorse or promote products derived from this software
18.\"    without specific prior written permission.
19.\"
20.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30.\" SUCH DAMAGE.
31.\"
32.\" $FreeBSD: src/share/man/man9/buf.9,v 1.5.2.5 2001/12/17 11:30:18 ru Exp $
33.\" $DragonFly: src/share/man/man9/buf.9,v 1.3 2004/12/22 05:53:09 hmp Exp $
34.\"
35.Dd December 21, 2004
36.Dt BUF 9
37.Os
38.Sh NAME
39.Nm buf
40.Nd "kernel buffer I/O scheme used in FreeBSD VM system"
41.Sh DESCRIPTION
42The kernel implements a KVM abstraction of the buffer cache which allows it
43to map potentially disparate vm_page's into contiguous KVM for use by
44(mainly filesystem) devices and device I/O.  This abstraction supports
45block sizes from DEV_BSIZE (usually 512) to upwards of several pages or more.
46It also supports a relatively primitive byte-granular valid range and dirty
47range currently hardcoded for use by NFS.
48.Pp
49The code implementing the VM Buffer abstraction is mostly concentrated in
50.Pa /usr/src/sys/kern/vfs_bio.c and
51.Pa /usr/src/sys/sys/buf.h .
52.Pp
53One of the most important things to remember when dealing with buffer pointers
54(struct buf) is that the underlying pages are mapped directly from the buffer
55cache.  No data copying occurs in the scheme proper, though some filesystems
56such as UFS do have to copy a little when dealing with file fragments.  The
57second most important thing to remember is that due to the underlying page
58mapping, the b_data base pointer in a buf is always *page* aligned, not
59*block* aligned.  When you have a VM buffer representing some b_offset and
60b_size, the actual start of the buffer is (b_data + (b_offset & PAGE_MASK))
61and not just b_data.  Finally, the VM system's core buffer cache supports
62valid and dirty bits (m->valid, m->dirty) for pages in DEV_BSIZE chunks.  Thus
63a platform with a hardware page size of 4096 bytes has 8 valid and 8 dirty
64bits.  These bits are generally set and cleared in groups based on the device
65block size of the device backing the page.  Complete page's worth are often
66referred to using the VM_PAGE_BITS_ALL bitmask (i.e. 0xFF if the hardware page
67size is 4096).
68.Pp
69VM buffers also keep track of a byte-granular dirty range and valid range.
70This feature is normally only used by the NFS subsystem.  I'm not sure why it
71is used at all, actually, since we have DEV_BSIZE valid/dirty granularity
72within the VM buffer.  If a buffer dirty operation creates a 'hole',
73the dirty range will extend to cover the hole.  If a buffer validation
74operation creates a 'hole' the byte-granular valid range is left alone and
75will not take into account the new extension.  Thus the whole byte-granular
76abstraction is considered a bad hack and it would be nice if we could get rid
77of it completely.
78.Pp
79A VM buffer is capable of mapping the underlying VM cache pages into KVM in
80order to allow the kernel to directly manipulate the data associated with
81the (vnode,b_offset,b_size).  The kernel typically unmaps VM buffers the moment
82they are no longer needed but often keeps the 'struct buf' structure
83instantiated and even bp->b_pages array instantiated despite having unmapped
84them from KVM.  If a page making up a VM buffer is about to undergo I/O, the
85system typically unmaps it from KVM and replaces the page in the b_pages[]
86array with a placemarker called bogus_page.  The placemarker forces any kernel
87subsystems referencing the associated struct buf to re-lookup the associated
88page.  I believe the placemarker hack is used to allow sophisticated devices
89such as filesystem devices to remap underlying pages in order to deal with,
90for example, remapping a file fragment into a file block.
91.Pp
92VM buffers are used to track I/O operations within the kernel.  Unfortunately,
93the I/O implementation is also somewhat of a hack because the kernel wants
94to clear the dirty bit on the underlying pages the moment it queues the I/O
95to the VFS device, not when the physical I/O is actually initiated.  This
96can create confusion within filesystem devices that use delayed-writes because
97you wind up with pages marked clean that are actually still dirty.  If not
98treated carefully, these pages could be thrown away!  Indeed, a number of
99serious bugs related to this hack were not fixed until the 2.2.8/3.0 release.
100The kernel uses an instantiated VM buffer (i.e. struct buf) to placemark pages
101in this special state.  The buffer is typically flagged B_DELWRI.  When a
102device no longer needs a buffer it typically flags it as B_RELBUF.  Due to
103the underlying pages being marked clean, the B_DELWRI|B_RELBUF combination must
104be interpreted to mean that the buffer is still actually dirty and must be
105written to its backing store before it can actually be released.  In the case
106where B_DELWRI is not set, the underlying dirty pages are still properly
107marked as dirty and the buffer can be completely freed without losing that
108clean/dirty state information.  ( XXX do we have to check other flags in
109regards to this situation ??? ).
110.Pp
111The kernel reserves a portion of its KVM space to hold VM Buffer's data
112maps.  Even though this is virtual space (since the buffers are mapped
113from the buffer cache), we cannot make it arbitrarily large because
114instantiated VM Buffers (struct buf's) prevent their underlying pages in the
115buffer cache from being freed.  This can complicate the life of the paging
116system.
117.Pp
118.\" .Sh SEE ALSO
119.\" .Xr <fillmein> 9
120.Sh HISTORY
121The
122.Nm
123manual page was originally written by
124.An Matthew Dillon
125and first appeared in
126.Fx 3.1 ,
127December 1998.
128