Internet Info 1997 December/Internet_Info_CD-ROM_Walnut_Creek_December_1997.iso/faqs/comp/answers/386bsd-faq/part3

Internet Message Format | 1997-10-14 | 81KB

Path: senator-bedfellow.mit.edu!bloom-beacon.mit.edu!news-out.internetmci.com!newsfeed.internetmci.com!206.229.87.25!news-peer.sprintlink.net!news-sea-19.sprintlink.net!news-in-west.sprintlink.net!news.sprintlink.net!Sprint!204.248.21.3!neonramp.com!cynjut.neonramp.com!cynjut.neonramp.com!not-for-mail
From: [e-mail] (Dave Burgess)
Newsgroups: comp.unix.bsd.netbsd.announce,comp.unix.bsd.freebsd.announce,comp.answers,news.answers,comp.unix.openbsd.announce
Subject: [comp.unix.bsd] NetBSD, FreeBSD, and OpenBSD FAQ (Part 3 of 10)
Supersedes: <[e-mail]>
Followup-To: comp.unix.bsd.netbsd.misc
Date: 13 Oct 1997 02:00:11 -0500
Organization: Dave's House in Omaha
Lines: 1865
Approved: [e-mail],[e-mail]
Expires: 10/31/97 01:00:03 CDT
Message-ID: <[e-mail]>
References: <[e-mail]>
Reply-To: [e-mail] (BSD FAQ Maintainer)
NNTP-Posting-Host: cynjut.neonramp.com
Keywords: FAQ 386bsd NetBSD FreeBSD !Linux
X-Posting-Frequency: Posted on/about the 13th and the 27th of every month.
Xref: senator-bedfellow.mit.edu comp.unix.bsd.netbsd.announce:563 comp.unix.bsd.freebsd.announce:734 comp.answers:28477 news.answers:114369

Posted-By: auto-faq 3.1.1.2
Archive-name: 386bsd-faq/part3

Section 2.    (Common installation questions)


2.0    Install process

    Once the files are on floppies, thoughts usually turn to
    questions about how to install the boot image on a floppy.  The
    rawrite program (for DOS) is used to write the bootable images
    (dist.fs and fixit.fs) onto floppies.  The same image can used
    for 3 1/2 and 5 1/4 high density diskettes. NetBSD uses the .fs
    file extension for its floppy images.  FreeBSD uses the .flp
    extension.

    Once the bootable images are written onto the floppies, insert
    the dist.fs disk into the A: drive and reboot.  If the system
    does not boot, see section 2.5 below for more information.

    If the disk boots, type install and proceed to use the
    INSTALL.NOTES to get more information.

    Problems with the install are either related to hardware (i.e.
    Do you want to install on your T.V.?) or software.  Of the
    hardware issues, the most common FAQs are usually straight out
    of the installation notes.  Of the software issues, there are
    only two that really concern us.  The first is bad files.

    On some systems, files that are loaded from floppy appear to
    'go bad' when they arrive on the hard disk.  Try some of these
    solutions:

    - You forgot binary.  Don't get insulted.  Those of us that FTP
    for a living forget sometimes.  If so, the distribution will
    come out with all different sizes and install will complain
    about every disk.

    - One or two of the files are no good.  Try getting them again.
    As a precaution, rename the bad files on your hard drive to
    names like foo.1 and bob.23.  Copy the files again from floppy.
    If they are still bad, rename the file, and the one immediately
    before the first bad file (bin01.23 if bin01.24 is bad) and
    copy them again.  If they are still bad, download those files
    again from the distribution site (including the one before and
    after the bad one) and try again.

    The reason for renaming the files is that sometimes, especially
    with drive that do not auto-magically record bad sectors, you
    could copy a distribution file onto a bad spot on the disk.  If
    this happens, you want to isolate the bad spot.  The easiest way
    to do that is just leave the bad file on it.

    For those of you that have received your system on a CD-ROM,
    you will need to find at least three things.  One is this file.
    Since you are reading it, I assume that you got it already. :-)
    If you can't read this file (you got it from the newsgroup, for
    example) there is one thing that you need to know so you don't
    look like a complete idiot on the net.

    There is no such thing as a Unix CD-ROM.  They are all in
    something called the ISO CD-ROM format.  You can read them as
    the D: drive in DOS, or mount them on your Sun or SVR4 system
    or whatever.

    Second, you will need to find the directory with the bootable
    disk images in it.  They will have self explanatory names like:

<pre>
    kerncopy.fs
    base0-9.fs
    fred.fs
    genericaha.fs
    boot-me-first.fs
    this-is-the-file-with-the-fs-extension.fs
</pre>

    You get the idea, right?  Look for the MS-DOS program
    "RAWRITE.EXE".  It should be right near the file system (.fs)
    files.  Another clue for the truly lost will be that the file
    system files will all be 1.2 Meg big.  These files will fit
    onto either a 1.2Meg 5.25 inch diskette, or a 1.44Meg 2.5 inch
    diskette.  Use rawrite to write the fs files to diskettes and
    boot from the diskettes.

    The FreeBSD system uses a system 'pretty much' the same as this,
    except that the filesystem files are 1.2 Meg files and they all
    have a '.flp' extension.  Other than that, the instructions
    apply.

    You did back your system up, right?

    For those of you trying to build installation floppies, you
    will need to verify the media type on the 'dd' and 'disklabel'
    commands in the Makefile.  The default is to build to 1.2 Meg
    disks (being the smallest in terms of room).  Change the 12100
    and floppy5 entries to 14410 and floppy3 (respectivly).


2.0.1    Boot disks  (versions and media formats)

2.0.1.1    I have the base system installed, and now I want to install the
    rest of the system.  Where did the 'extract' program go?

    When installing NetBSD, the 'set_tmp_dir' and 'extract' programs
    are part of the .profile that is booted when you are installing.
    This .profile is overwritten as part of the install process, and
    extract then disappears.  If you need extract again, you can mount
    the install disk and source .profile.  This will recreate these
    two routines.

    There is also an install procedure that FreeBSD uses that does
    the same job.  It is defined as part of the .profile on one of the
    installation floppies.  You can either copy it from there, or use
    the procedure for 'real disk partitioning'.

    Failing that, you can use the following process to extract the
    sources.

        -   First, 'cd' to where your files are.
        -   Assuming you want to extract the kernel sources, use the
        following command to extract the files:

        "cat ksrc* | tar -xvf - -C /"

    This will combine the pieces, feed them to tar, and load the
    files in the 'standard' place.

2.0.1.2a    The floppy booted, but now the hard disk won't boot?
2.0.1.2b    I am trying to reinstall.  I run install and it loops
          asking me if I want to use the whole disk?

    The most likely culprit is your hard disk controller.  If you
    have an IDE or EIDE controller, it is probably doing some type
    of disk translation for you.  If this is the case (assume it
    is) then you will need to find out the real disk controller
    geometry, and rewrite your disk label.  See section 2.6.2, but
    before doing that get the program pfdisk.exe.  This program
    will tell you what the controller geometry is (right before
    it reboots your computer).  Make the disklabel agree with
    this program and your system should boot.  You may have to
    reinstall, but at least your disklabel will be right.  Note
    that this is a nearly required step for all NetBSD and
    FreeBSD installs.  You need to know what the disk geometry
    is before the BIOS messes with it.  If you start having these
    kinds of installation problems, I can virtually assure you that
    it is because your controller geometry and your disklabel
    geometry are different.

    NOTE:  If the hard disk controller does NOT have an option for
    turning off the geometry, you may well be completely out of
    luck.  There are very few controllers that fall into this
    category.  The ones that do full time translation will often
    boot up in translated mode.  pfdisk will help you determine the
    correct geometry for your drive by telling you what the geometry
    looks like when 386bsd boots up.

    But on the other hand, maybe not...

    See section 2.5.5 below for a detailed set of instructions about
    getting NetBSD (and by implication 386BSD and FreeBSD) to work with
    a system that uses full time translation.


2.0.1.4    What are the options on the boot prompt?

    The most amazing thing about the boot process in *BSD is the
    boot up alternatives that are available.  There is little that
    a person can NOT do from the boot prompt.  The boot diskette or
    disk can be selected (fd(1,a) for fd1a (my B: drive is DOS))
    can be the source of my kernel.  In addition, the name of the
    kernel can be chosen (this allows you to boot with a test
    kernel or reboot an older kernel if the new one gets hosed).
    Finally, there are three choices for options that may or may
    not work, depending on the age and proclivities of your boot
    blocks.  These options are documented in 2.5.9 below.


2.0.1.5    I just used the '-s' option on the boot, but I can't write
    anything onto the disk.  What is wrong?  If I use a plain 'mount'
    command it tells me that my root file system is read-only.

    In single-user (system booted with -s or an error in one of
    the processes started by /etc/rc) the root filesystem mounts
    as read-only by default. This was intended so that some range
    of problems would not be made worse by writes to the disk.

    The 'dos' partitions mount as read-only in that there are
    reservations as to how well some of the FreeBSD tools work with
    the pcfs.  The same kind of reservations exist with NetBSD and
    the '-t msdosfs' option.  These options (-r for read-only, -w
    for read-write) can be set in /etc/fstab.

    The status of both can be changed with 'mount -wu /{mount.dir}'
    (where {mount-dir} is the name of the directory that the
    offending partition is mounted) to read-write.  Particularly for
    the dos filesystem, the man page for mount should be read in
    detail and the 'noexec' option examined.

    Note that mounting the file systems using the '-a' option will
    mount all of the file systems that are normally mounted with
    their usual read-write bits set normally.  Using this option
    makes your root partition writable, and also mounts the rest of
    the partitions in your /etc/fstab that are normally mounted
    during boot-up.


2.1    Binary distribution
2.1.1    I want to install by NFS but I am having all kinds of problems
    connecting to the Sun server where the files are.

    There is an unusual problem when installing over NFS.  This
    solution may have been corrected in the documentation that comes
    with FreeBSD and NetBSD, but if not, here it is.

    The most common problem seems to be that FreeBSD (and by
    inference NetBSD and all the other 4.4 based systems) do not
    send out NFS requests over privileged ports.  Sun's NFS
    implementation (and others, once again by inference) expect
    precisely the opposite.  These systems will quietly fail if you
    try to NFS to them.

    The usual error message (which may ONLY appear in
    /var/adm/messages) is "nfs_server: weak authentication, source
    IP address=xx.xx.xx.xx"

    SunOS is particularly insidious at this point.  The mount
    succeeds, but then everything else after that fails.  This means
    that your FreeBSD or NetBSD system will return an EACCESS error
    whenever you try to grab a file from the NFS filesystem.  The
    solution (tested in FreeBSD) is to include the 'resvport' flag
    like this:

<pre>
    # mount -o resvport server:/fs /mnt_point

</pre>
    or to use the -P flag (which does the same thing).  See the
    mount and mount_nfs man pages for the details.

    In fact, the -P flag provides a solution to the FreeBSD NFS
    installation problem.  When prompted for server/filesystem, type
    in the flag before the server/filesystem pair:

<pre>
      -P server:/fs
</pre>

    If you are using an 8-bit network card, and want to avoid the
    ring buffer overflow problems that seem to come standard with
    this class of cards, you can also include the "-r4096 -w4096"
    flags between the -P and the server.


2.2    Configuration

    By far, the most common configuration questions are partitioning,
    followed closely by all of the other software in the system.
    Sendmail and named are also problems occasionally, but the
    documentation that comes with them usually gets you through.  If
    you run into a problem, post a question to comp.os.386bsd.questions.

    A less frequently asked question is "Where can I get info on how
    to configure a kernel?"  The answer to this question has been
    provided by Richard Murphey (Email address [e-mail]).

    --------------------------------------------------------------------
    Ready-to-print PostScript files for each section of the net2 system
    maintainer's manual are on nova.cc.purdue.edu in
    pub/386bsd/submissions/bsd.manuals.

    smm.02.config.ps.Z describes kernel configuration for the VAX,
    however some of it is relevant to 386BSD.  There is no freely
    available rewrite for 386BSD that I know of.
    --------------------------------------------------------------------

    Most of these manuals are now included in the standard release of
    NetBSD and FreeBSD in the /usr/share/doc directories.

2.2.1    Partitions

    This section describes many of the questions that people ask about
    hard disk partitioning.

    The first is a brief explanation of the BSD system disk partitions.

2.2.1.1    What is a 'disklabel' and why do I need one?

    The BSD partition table supplements the DOS partition table.  The
    entries in this table are meaningful to BSD.  There are eight
    partitions in the BSD partition table, and they are normally
    lettered from a: to h:.  This supplemental partition table is
    often referred to as the 'disklabel'.

    There have been many good articles in both the mailing lists and
    the newsgroups about disk labeling and partitioning.  I have
    included a few of them here.  NOTE:  This information has not
    really changed since 386BSD 0.1.  Some of the specifics may be
    out of date (the use of the d: partition, for example) but the
    steps and information are still pertinent.

    Phil Nelson ([e-mail]) writes:
    I have installed several disks that have > 1024 cylinders and
    have used both DOS and NetBSD.  What has worked for me EVERY TIME
    is the following:

        a) Tell the BIOS that you have 1023 cylinders and the correct
    geometry for heads and sectors.  (This will limit your DOS part
    of the disk to be LESS than the first 1023 cylinders.) You need
    to have ALL of  your partition A (/dev/wd?a) in the first 1023
    cylinders so that the boot program can read the kernel from
    the root partition using the BIOS routines.  (ed note: You can
    specify the full number of cylinders in some BIOSes and it won't
    make any difference.  The DOS part of the disk will always be less
    than 1023 cylinders.)

    b) With fdisk, partition your 1023 cylinders as you want them.

    c) Use the real geometry in NetBSD.  Once the NetBSD kernel is
    booted, it does not have the 1024 cylinder limit: that is only
    for the BIOS.  NetBSD only looks at the BSD disklabel, not the
    DOS disk label.  The two disk labels (DOS and BSD) may not agree
    on the BSD partition size!  This isn't a problem, since each
    system's idea of the disks geometry is based on different
    information.

    d) Use NetBSD!

    Chris Jones writes:

    I was getting different reports of disk geometry from different
    programs, so I opened up the computer and read the plastic label
    on the drive.  I then instructed the BIOS (which, when using
    auto-detect disagreed) what the disk geometry was.  Then, I
    used pfdisk to create partitions.  The first thing I did with it
    was to tell it what the geometry really was.  It said something
    about a symbolic mapping and dealt with it.  Then I was able to
    specify all partitions in real units instead of virtual ones.
    NetBSD boots fine, and if memory serves, it is the only program
    that has recognized the real disk geometry from the beginning.

    This tutorial is provided by by "Hacksaw" <[e-mail]>
    and provides an  excellent overview of the hard disk partitioning
    procedure from start to finish.

        "Disk Partitioning for the Compleat Idiot"

    There are times, in our trials with our computers, that it becomes
    necessary to mess about with the disklabel. For those not
    knowledgeable of BSD or Unix Systems administration, this somewhat
    simple task can be somewhat daunting. This document is the result of
    my own short experience.

    This does not cover physical installation of the disk. For those who
    are having trouble with that, I direct you to any of the fine manuals
    dealing with hard drives and your hardware.

    It also does not deal with the vagaries of the DOS partition manager.
    It assumes you have done that as well, if need be...

    After the drive is physically installed and is recognized in the BSD
    startup, and it mentions both your drives, in the order you expect
    them... Or perhaps just the one, if you had special problems with
    installation. Now all you have to do is "disklabel" the drive... Well,
    what is *THAT*???

    The disklabel is used by the kernel and other utilities to tell how
    you want or have the drive set up *logically*.

    In a beautiful world, we might have a very free hand at this set-up
    and expect it to work. Unfortunately, the authors of the software
    dealing with the hard drives either decided or were forced by
    circumstance to make certain things about the disklabel inviolate.

    When you let the installation disk set the disklabel for you first
    drive it comes out like this:
<pre>

        The a: partition is the primary partition.
        The b: partition is the swap partition.
        The c: partition is the amount of the disk used by 386bsd
            (swap and data)
        The d: partition is the entire disk (on the PC version only).
</pre>

    Of these, the only one that could be different is a:...

    (Note for those of us who have spent far too much time using DOS: the
    labels a: b: c: d: e: f: g: h: DO NOT refer to DOS drives, but to
    partitions in your 386bsd partition... confusing, eh?  For the sake
    of consistency I will never make a reference to DOS drives except by
    saying something like "DOS drive C:". )

    It's possible to divide up the disk a bit differently, but three
    things MUST be:

    c: must refer to every cylinder you wish 386bsd to use, either
    for your data or the swap space.

    b: Must always refer to a swap partition. Note that on any
    other than the first disk it does not have to, but if you
    enable swapping on that drive, and you are using b: for
    something else, that something else will be killed.

    The reason for this is simple: It's hard coded in.

    "WHY?" you ask? (I did...) Probably time constraints, maybe tradition.
    But if you look at the code in "isofs" and "ufs" in your sys.386bsd
    directory, you will see numerous comments asking some of the same
    questions, which leads me to believe this may change in the future,
    making our lives both more complicated and easier at the same time...

    Getting past the esoteric explanations, here is a method for figuring
    out and "labeling" your disk.

    We'll start with the disklabel from my second disk, in the form most
    understandable by humans... #'s signify the start of a comment.

<pre>
    # /dev/rwd1d:
    type: ESDI
    disk: maxtor7245
    label:
    flags:
    bytes/sector: 512
    sectors/track: 31
    tracks/cylinder: 16
    sectors/cylinder: 496
    cylinders: 967
    rpm: 3600
    interleave: 1
    trackskew: 0
    cylinderskew: 0
    headswitch: 0        # milliseconds
    track-to-track seek: 0    # milliseconds
    drivedata: 0

    5 partitions:
    #      size  offset  fstype [fsize bsize   cpg]
      a:   198400       0  4.2BSD    512  4096    16     # (Cyl.    0 - 399)
      b:    31744  447392    swap                      # (Cyl.  902 - 965)
      c:   479136       0  unused      0     0           # (Cyl.    0 - 965)
      d:   479136       0  unused      0     0           # (Cyl.    0 - 965)
      e:   248992  198400  4.2BSD    512  4096    16    # (Cyl.  400 - 901)
</pre>

    Some math:
    Looking at the comments at the end and the size and offset columns,
    size is a function of (last - first + 1) * sectors per cylinder:
    a: 399 - 0 + 1 = 400 * 496 = 198400
    b: 965 - 902 + 1 = 64 * 496 = 31744
    c: & d: (Since I have no DOS partition, whatsoever)
          965 - 0 + 1 = 966 * 496 = 479136
    e: 901 - 400 + 1= 502 * 496 = 248992

    248992 + 198400 + 31744 = 479136 (all the parts should equal the whole)

    Some things I discovered  (for all you in novice land like me...)

    1. As you can see this disk has 967 cylinders, but I only refer to 966
    of them, 0 - 965... This is because it's good practice to leave the
    "Landing Zone" cylinder out of it... This is usually the last
    cylinder, and it's where the read/write heads hang out when your disk
    is off...

    Note from TSgt Dave:

    Most modern drive heads come to rest on a polished surface inside the
    highest cylinder.  I could be mistaken, of course, and the Hard Drive
    Bible (or other appropriate reference manual) will tell the tale for
    each drive.

    2. a: can be a regular partition, b: should be swap, c: everything
    386bsd will get to use, including swap. d: is the entire disk from
    0 - (cylinder_per_disk - 2)   [leaving out the Landing Zone]

    On the boot drive (The drive that actually contains the kernel), a:
    is the boot partition.  On all other drives, it is a regular partition.
    Regardless of whether you are using DOS or not, the entire a:
    partition must reside completely within the first 1024 sectors.
    This is a limitation of the PC architecture.

    You can then use e - h for your other partitions. I am not sure
    whether you could specify b: as other than a swap partition and not
    run into trouble, but you could surely make it a zero sized one
    starting and stopping on the Landing Zone...

    Note from TSgt Dave:

    This is a good idea.  Another way to accomplish this is to
    simply not specify it in the map.

    3. Stupid human trick: When doing the math don't forget that 400 - 900
    refers to 50*1* cylinders. I did, for a while. No great problem I
    suspect, but why waste a cylinder...

    4. newfs'ing really is that simple if you have the label right:
    "newfs /dev/rwd?x config_template" where the question mark is the
    physical disk, the x is a partition letter, and the config_template
    is the configuration from /etc/disktab for your disk drive.

    * NOTE:  This is a thumbnail sketch; read the man page to verify all
    of the options and be sure about how to proceed...

    5. then fsck the partition:
    fsck /dev/rwd?x

    Don't forget that fsck should be run on the RAW device.

    6. As long as it checks out, you can then mount it and do disk things
    with it...

    7. Add it to the fstab... (follow the man page).  Don't forget
    that your new swap partition won't work if your kernel isn't
    configured for it, but it won't cause you any problem to have
    it there.

    One last note from TSgt Dave:

    And I have yet to figure out a way to determine if it is or
    isn't using the swap partition anyway.  There is a program called
    'swapinfo' and it is part of the NetBSD source tree.  On my system,
    it tells me that I never use the swap area. :)

    A note for those trying to use the CCD: to figure out what the
    disk label should be for your concatenated device, assuming
    your disks are identical, just add up the cylinders (minus the
    ones your reserved for the individual disk labels). I know
    this works for purely concatenated (not striped) IDE disks, I
    am assuming it should work on stripped SCSI disks.

    Commonly used definitions:

    bsize:
    Block Size:  This is the smallest allocatable area on a disk file
    system, sort of.  A file uses the maximum amount of blocks until it
    can not completely fill up a block.

    fsize:
    Fragment Size:  This is the size of the 'leftover' data that didn't
    fit into a full block.  For example, assuming a using an 8K Block
    Size/1K Fragment Size, a 34.5K file, would use up 4-8K Blocks (4 *
    8K = 32K) and 3 1K fragments (3 * 1K = 3K).  There is 512 bytes of
    wasted space, since 32K + 3K = 35K, which is 512 bytes larger than
    34.5K.  If you want to reduce the amount of wasted space, you can
    reduce your fragment size, but you also reduce the amount of data
    you read at one time, so your disk performance decreases also.
    A good setup is 8K/1K for performance, but if you are really
    concerned about wasted space you can consider using a 4K/512byte
    filesystem.

    For further information, find an article that explains the Berkeley
    FFS in more detail.

    cpg:
    Cylinders Per Group, it determines the cylinder group size, which
    in turn determines the number and location of the alternate
    superblocks.


2.2.1.2    What other kinds of information do I need if I really want to
    tune my hard drive's performance in conjunction with a newfs?

    Having taken Aim's suggestion and changed my newfs values, I
    think I've now made some empirical observations that suggest
    that the defaults for newfs should definitely be changed.

    With all the disks I tested with, -n 1 (which isn't even
    *documented!*) provided greatly improved performance, as
    opposed to all other values of -n.  I think that with
    sector-addressed drives with complex physical geometries,
    rotational position optimization is a technique which is no
    longer valid.

    If _anyone_ has _any_ disk larger than 300MB or so (or even
    a small disk) manufactured within the last few years for which
    larger values of -n produce better performance than -n 1, I'm
    very curious to hear about it.  I'd be particularly interested
    in any disk for which the default value produces optimal results.

    Increasing maxcontig seemed to always improve write scores, but
    values of maxcontig above 16 seemed to have a noticeable _negative_
    impact on read performance.  -a 512, for example, on the disk in
    my machine at home, yielded a peak write rate (4MB file, 8K record
    size) of 4.7MB/s, much better than the 4.3MB/s value for -a 64,
    but read performance was reduced from 2.6MB/s to 2.1MB/s.  I do
    not understand why this is the case, and I'd love suggestions.

    I believe that with rotational position optimization turned
    off (-n 1), the value of the -r option is of no consequence.  I
    believe that the fact that with the default value for -n, the
    -r option seemed to have little or no impact on performance
    serves to demonstrate that rotational optimization does not
    work correctly on modern drives.

    The default value of the -d option also produces much worse
    results than -d 0.  I'm probably inexact up above; I believe
    that -n 1 -d 0 is what turns off rotational position optimization
    entirely.  I'm all for it. :-)

    I suggest that the defaults for newfs be changed to:

<pre>
        -n 1 -d 0 -a 16 -r 5400
</pre>

    The -r value just in case someone decides to try playing with
    rotational position optimization for some incomprehensible
    reason.  Though actually, anyone with a disk where said
    optimization is a win might want -r 3600 after all.

    If someone can explain why values of -a above 16 seem to
    negatively impact read performance, I'm all for making -a very
    very large, like 512 or 1024 -- in this case the filesystem
    code will automatically limit maxcontig to the maximum transfer
    size for a given controller/disk, right?  What are some typical
    such sizes?  Why does -a 512 hurt read performance so much, and
    how can it be fixed?  From comments by Larry McVoy, a good
    implementation of UFS with clustering will yield disk speed
    on writes, and about 25% less on reads.

    Right now, on my hardware at least, we seem to _surpass_
    slightly the speed of raw writes to the disk device on writes,
    but on reads we lose big as the maxcontig value goes up, and we
    seem to lose worst on large file/record sizes, where the raw
    device delivers about 5MB/s in my case, but with -a 512 I get
    only about 2.5MB/s under UFS.

    If you can't guess, I'm incredibly curious as to why the value
    of -a affects reads as much as it does, or at all, for that matter.

    Still, we don't do so badly -- with -a 16, we pretty much hit
    Larry's "good" value on reads of 75% efficiency, and we still
    just barely surpass the raw device write figures.  (I am very,
    very, very curious as to how this is possible at all.  Anyone?)


2.2.2    Common Disk Label Problems.

2.2.2.1    Increasing the *BSD partition size.

    There is no easy way to increase this swap partition without
    relabeling the drive.   Unfortunately, relabeling usually involves
    reinstalling.  That involves re-doing just about everything you have
    just finished doing.  The good news is that if all you have done is
    the base installation, you don't have a lot of time and energy
    invested in the system.  Take the time, and make sure that your swap
    space is at least as big as your memory; many people recommend even
    larger.  There is no real limit to the size that this space can
    take.  If you have two disk drives, you can have space space on both.
    Simply follow the instructions above, and you will be all set.
    If your swap space is smaller than your real memory, system core
    dumps will be disabled.

    If you have compiled in the VNODEPAGER option in your config
    file, you can use vnode files for swap space.   The precise
    details are exaplined in the man pages, but the easiest way to
    start is to include the following line in your /etc/fstab:

    /dev/vnd0b    swap    swap    sw

    Defining the file for the vnode is fairly straightforward:

    vnconfig -c /tmp/swapfile /dev/vnd0b

    and actually making it swap is as simple as

    swapon /dev/vnd0b

    From there, the rest of your questions should be answerable from
    the vnconfig manpage.


2.2.2.2    I can access the DOS partition on my second disk from Unix but not
    DOS?  Any suggestions?

    One kinky problem that almost got me was when I tried to disklabel
    my second drive in order to use the DOS partition on it, and use
    the rest as swap for BSD (FreeBSD-1.0 Eps, SCSI drive on an
    AHA1542B, to be exact). The DOS partition was visible from UNIX,
    but *not* from DOS.

    What I tried to do:
        Using PFDISK (from DOS), make one big DOS partition at the start
        and use the rest for a BSD partition (type 165).  Something that
        came out like
<pre>
          1    6    0    69 DOSbi # ..
          2    165  70    98 unkno
</pre>
        for a 99 cyl drive.

        Using BSD disklabel generate disk description/label as documented
        in the FAQ. Make only 'c' (total BSD DOS part), 'd' (complete disk)
        and 'b' (intended swap) BSD partitions.

    Problem:
        When writing the label, disklabel would ask about overwriting DOS
        partition table.  Whether I said y or n, the DOS partition table
        was screwed up, as seen from DOS (BSD saw the DOS file system
        very nicely indeed).

    Cause, solution:
        BSD disklabel wants to write the label to the start of the 'a'
        partition; I had *not* defined an 'a' partition (since I was
        only using the disk for swap).  This tells disklabel that the 'a'
        partition is the start of the disk, which means there is no DOS
        partition.  Disklabel then writes the label at the start of the
        drive, which is why it talks about overwriting (aha!); this is
        *bad* for the DOS partition table.  One solution is to have a
        non-empty (e.g. one cylinder) 'a' partition at the start of the
        BSD part of the disk, and resize the 'b' swap partition
        accordingly.  Now everything works just fine.  Note that
        this solution can be used whenever you want the DOS
        partition table to be safe and the DOS partition to be
        mountable.

        One other fly in this ointment.  The disklabel program has
        historically asked "Overwrite disk with DOS-partition [n]: "
        then the normal inclination is to believe the prompt and
        press return for 'no'.  The default answer may or may not be
        'no'.  There are several versions of disklabel where the
        default answer is actually 'yes' even though the prompt
        implies that you can press return and get 'no'.  In this case,
        it might be best to assume that the default answer doesn't
        exist until you have had a chance to actually look at the
        disklabel code.


2.2.2.3    I want to use my entire 2 Gig drive as the root partition.  Why
    doesn't it work?

    The easiest answer is the architecture of the machine has gotten
    you.  Because of the limitations of the BIOS, everything the
    boot process needs must reside in the first 1023 cylinders on
    the disk.  Most really big drives have more 'real' tracks than
    this, so DOS tries to translate the drive so it doesn't.  The
    *BSD systems don't; they rely on the disk geometry being
    correct, or at least the same as the controller thinks it is.
    Once the system is up and running, the BIOS is disabled.  This
    means that the system no longer has that 1023 track limitation.
    What does this mean to you?  Make sure that the root partition
    (the a: partition above) of your boot drive does not extend
    beyond track 1023.  If you have a large DOS partition that
    covers nearly all of that, you may need to make a VERY small
    root partition to make absolutely certain the root does not
    extend past 1023.


2.5.3    How do I set up the system so that I can boot from more than one
    operating system/file-loader without using floppies?

    There are many people that wish to be able to boot DOS or 386bsd
    at will.  There are several programs that allow this.  The
    program "os-bs" is one such program, "BOOTEASY" is another, and
    there are three or four others.  There are problems in some
    configurations using the os/2 boot manager for this, so beware.

    In addition to being able to boot from either of two partitions,
    some people want to operate more than one disk drive (and perhaps
    boot from either as well).  Christoph Robitschko provided one
    description of this.  Since there are virtually limitless
    possibilities for configurations for BSD systems, it will be
    impossible to answer all of the possible questions about these
    features.  Many people operate with multiple disk drives on one
    or more controllers.

    Yu-Han Ting provides this tutorial on partitioning and booting
    multiple systems with a single hard disk.


2.2.3    How do I get the system to boot from the second hard drive?

    Julian Elischer ([e-mail]) adds:

    To make the boot code default to drive 1 look in
    /sys/{arch/}i386/bootboot.c for the following (or similar.  The
    code may have changed a little and may be in a different
    directory:
<pre>

loadstart:

    /***************************************************************\
        * As a default set it to the first partition of the first    *
        * floppy or hard drive                        *
      \***************************************************************/
        part = unit = 0;
        maj = (drive&0x80 ? 0 : 2);             /* a good first bet */
        name = names[currname++];


    and change it to:


loadstart:

    /***************************************************************\
    * As a default set it to the first partition of the SECOND    *
    * floppy or hard drive                        *
    \***************************************************************/
    part = 0;
    unit = (drive & 0x7F);
    maj = (drive&0x80 ? 0 : 2);             /* a good first bet */
    name = names[currname++];

</pre>
        This way, whatever drive the boot blocks are loaded from, it has
    that as default.  In my case, I get wd(0,a) when I have my netbsd
    drive as C:, and wd(1,a) when I have it as D:.  (I've been
    swapping drives left right and center the last day getting dos
    to boot on one drive and netbsd on another).


2.2.4    How do I disklabel my second hard drive?

    The obvious answer is to use 'disklabel -w -r /dev/rwd1d'.
    Unfortunately, this does not always put a real disklabel on the
    drive.  The symptom is that the drive labels and can be used
    until the system is reset, at which point the system tries to
    read the label from the disk.  It was never actually written to
    the disk, so the operation fails.

    There are also reports that the /usr/mdec files are corrupted in
    some of the distributions.  If you have tried everything else, you
    can either load the files from one of the many archive sites that
    keep the /usr/mdec files around, or you can recompile them
    yourself.

    Instead of specifying the entire device path name (i.e. /dev/rsd0c),
    only specify the two letters of the device type and the unit number
    (i.e. "sd0").  Disklabel figures out the rest, and it works.

    For instance, the following line works for me:

      disklabel -w -r sd0 <drive-type>

    assuming of course that the boot block files are in /usr/mdec/ and
    the <drive-type> is in the /etc/disktab.

    This is also a symptom of some of the versions of FreeBSD and
    NetBSD where the disklabel code was 'fixed' to only write a
    disklabel on a drive with a disklabel.  Oops.

    Also, some folks want to mix SCSI and IDE drive together in the
    same system.

    A report about someone with an Austin Tower (486DX/50), AMI BIOS,
    Caviar 2250 IDE, Adaptec 1542CF, and Toshiba SCSI disk (1.2GB)
    posted this set of instructions:

    The BIOS is configured to boot from the IDE drive as type 47
    (user defined).  The IDE drive currently has NetBSD 1.0 BETA on it.

    The 1542CF switches are 1-4 off (open), 5-8 on.  The meaning is as
    follows:

<pre>
        1(off)=Termination software controlled.
        2,3,4(off)=I/O Port x330.
        5(on)=disable floppy.  I use the Austin floppy controller.
        6,7,8(on)=disable Adaptec BIOS.
</pre>

    Note that this means the Adaptec 1542CF on-board setup program is
    also disabled.  If I need to change my SCSI termination, I first
    have to enable the Adapted BIOS (sw 6,7,8), enter 1542CF setup
    and change termination, then change switches again.

    I could not configure the system to boot from the SCSI drive having
    the IDE as a secondary drive.

    (Ed Note: There is more news on this front all of the time.
    Since I personally don't have much interest in doing this (I
    boot from my IDE drives and mount my SCSI drives) I don't see
    the problem. )


2.2.5    NetBSD and FreeBSD cannot handle disk geometry translations,
    but it turns out that my disk geometry is translated.  It has
    five zones, each with a different sec/track!  What kind of
    things can I do about the disk translation my hard disk
    controller uses?

    It turns out that what *BSD cannot handle is not translation, but
    translation that changes during the boot-up process.  For example,
    the configuration above will work just fine IF the translation
    that the controller uses when it powers up is the same one that it
    uses when it boots.  On many PC clones, the BIOS loads a different
    geometry after it boots to make the geometry agree with one that is
    loaded in CMOS.  This is the fatal flaw for *BSD.  Fortunately,
    once the problem has been identified, it is relatively easy to
    handle.  Simply make sure that the BIOS is configured to set the
    controller to the translated geometry that the card powers up
    with.

    There are several ways to get around these problems with disk
    geometry translation.  If you are using a SCSI controller, you can
    specify the geometry such that each 'cylinder' is 1 Meg (64 sectors
    by 32 tracks for example).  Most SCSI controllers will blithely
    ignore what YOU tell it the geometry is and press on using this
    type of 1 Meg cylinder had to get the job done.  NOTE:  If you are
    going to try this, try to ensure that each 'pseudo cylinder' is a
    reasonable size (like 1Meg or 512K).

    An interesting method for dealing with disk geometry comes from
    Alan Barrett ([e-mail]):

    This sort of problem happens when you try to install NetBSD in a
    partition of a disk whose controller does geometry translation.  I
    have not had time to find the bug that causes the problem.  One
    option is to disable the geometry translation:  Use ide_conf to
    find the true geometry, use the CMOS setup program to tell your
    BIOS about the true geometry, and reformat everything.  I
    successfully did that on one of my systems.

    If you are not able to, or do not wish to, disable the geometry
    translation then the following work-around might work for you.
    This requires that the disk have unused space on {cylinder 0,
    head 0}, from sector 2 to sector 16.  Almost all DOS disks that
    I have ever seen satisfy this condition, because they usually
    start the DOS partition in {cylinder 0, head 0, sector 1},
    leaving most of {cylinder 0, head 0} unused apart from the
    partition sector in {cylinder 0, head 0, sector 1}.  However,
    many partitioning programs like to hide this fact from you,
    and pretend that the DOS partition starts at the front of the
    disk; don't believe them until you have checked with a raw
    disk editor.

        0.  Make sure you have adequate backups.

        1.  Use a partition sector editor (fdisk, pfdisk, os-bs,
            booteasy, Norton utilities, whatever) to mark the partition
            that you want for NetBSD as bootable with type 0xA5
            (decimal 165).

        2.  Halt the system.  Boot the NetBSD kernel copy floppy.
        When it asks you to insert the floppy for the root file
        system, switch to the Install-1 floppy and press enter.

        3.  Answer all the installation prompts, using numbers based
        on the translated geometry.  When it asks if you really
        want to label the disk, be brave and say yes.

        4.  Halt the system.  Boot to DOS.  Run a disk editor program,
        such as Norton utilities.

        5.1.  Verify that the partition sector in {cyl 0, head 0,
        sec 1} is undamaged.  Verify that the disklabel program
        run as part of the NetBSD install has written the NetBSD
        primary boot block to {cylinder xx, head 0, sector 1},
        written the disk label to {cyl xx, head 0, sec 2}, and
        written the secondary boot program to {cyl xx, head 0,
        sectors 3 to 16}.  ("xx" represents the translated
        cylinder number you chose for the start of the NetBSD
        partition.  You did choose to start on a cylinder boundary,
        I hope.)

        5.2.  Verify that the space in {cyl 0, head 0, sectors 2 to
        16} is still available.  Copy the fifteen sectors containing
        the NetBSD disk label and secondary boot block from {cyl
        xx, head 0, sectors 2 to 16} to {cyl 0, head 0, sectors 2
        to 16}.

        5.3.  Edit the partition table in {cyl 0, head 0, sec 1}.
        Change the system ID of the NetBSD partition from 0xA5
        (decimal 165) to something else (I use 0xA4, decimal 164),
        but keep it flagged as bootable.  This will let you boot
        to the NetBSD primary boot block.

        5.4.  Edit one of the previously unused partition table entries
        (I hope you have one), to contain the following information:
        {sys id = 0xA5, boot flag = 0, start cylinder/head/sector =
        0/0/1, end cylinder/head/sector = anything, initial
        offset = 0, total size = anything}.  This will tell the
        NetBSD primary boot block, or a NetBSD system booted from
        a floppy, that it should look for the NetBSD disk label
        in {cyl 0, head 0, sec 2}.

        6.  Halt the system.  Boot the NetBSD kernel copy floppy.  When it
            asks you to insert the floppy for the root file system, just
            press enter without changing disks.

        7.  Copy the kernel, and proceed with the rest of the installation
            as per the instructions provided with NetBSD.  It should now
            work because of the trickery with the partition table etc.

2.2.6    I am having trouble installing on the EIDE hard drive.  What are
    some of the things that I need to look into?

    Bradley W Mazurek ([e-mail]) writes:

    First, I had to change the IDE translation mode in my BIOS.
    Rather than using LBA, I used Standard CHS.  When I went in
    to repartition the disk for DOS, DOS reported that the drive
    was only 523Mb (1023cyl, 64h, 63sec/tr), rather than the true
    geometry (2100cyl, 64h, 63sec/tr) but I didn't worry about it.

    Next I created my DOS partition.  I partitioned the disk so that
    cylinders 1-999 were DOS.  That left cylinders 1000-1023 for
    NetBSD.  Lots of room!  :)  Anyway, on a hunch, a friend and I
    were hoping NetBSD didn't look at the ending cylinder entry
    (1023) of the partition table.  Next I calculated the length
    of the partition from 1000-2100, put this into the partition
    table using the disk editor.  The numbers weren't consistent in
    the partition table, but DOS ignored the Non-DOS partition,
    NetBSD was happy...and we've (DOS, NetBSD and my remaining hair)
    all lived happily ever after....

    [Ed.Note.  The partition table needs to correctly identify the
    NetBSD portion of the disk, regardless of whether or not DOS can
    handle it.  See the section on hard drive partitioning for more
    information...]

    My suggestion is to try to find an IDE translation mode in your
    BIOS for which the number of heads and number of sectors per track
    is consistent with the true geometry of your hard drive.  Then
    perhaps this trick will work.

     1. there is _different_ behavior, if one executes
<pre>
        disklabel wd0                           or
        disklabel /dev/wd0c                     or
        disklabel /dev/wd0d
</pre>
        It didn't get quite clear to me, what these differences are exactly.

     2. Any disklabel write will change not only the data on disk, but
        also some data-structures in core. For example, if one tries to
    write a complete different disklabel to a complete different place,
    say /dev/wd0h, there will be strangeness afterwards.  That means,
    writing a disklabel and then reading it back, does not have to
    mean that the write did succeed.  There is an option -r to
    disklabel which is said to access the disk directly, but, as
    I noticed, the core-data is updated thereby, too.

    The following paper explained to me what should happen in sequence
    on boot: /usr/src/sys/arch/i386/boot/README.386BSD. It says (in
    short):

    [...]

    1/ the BIOS loads the first block of the disk (called the Master
    Boot Record or MBR) and if it has the correct magic numbers, jumps
    into it:

    2/ The  MBR code, looks at the Partition table that is embedded
    within it, to determine which is the partition to boot from. If
    you are using the os-bs bootblocks (highly recommended) then it
    will give you a menu to choose from.

    3/ The MBR will load the first record of the selected partition
    and if it has (the same) magic numbers, jumps into it. In 386bsd
    this is the first stage boot, (or boot1) it is represented in
    /usr/mdec by wdboot, asboot and sdboot. If the disk has been set
    up without DOS partitioning then this block will be at block zero,
    and will have been loaded directly by the BIOS.

    4/ Boot1 will look at block0 (which might be itself if there are
    no DOS partitions) and will find the 386bsd partition, and using
    the information regarding the start position of that partition,
    will load the next 13 sectors or so, to around 60000
    (640k - 256k). and will jump into it at the appropriate entry
    point. Since boot1 and boot2 were compiled together as one file
    and then split later, boot1 knows the exact position within
    boot2 of the entry point.

    Boot 1 also contains a compiled in DOS partition table (in case
    it is at block 0), which contains a 386bsd partition starting at
    0. This ensures that the same code can work whether or not boot1
    is at block 0.

    [...]

2.2.7    My disk label is complaining about '256 heads' in the disklabel.
    This is obviously bogus, but it doesn't seem to be hurting anything.
    Is it Okay or should I fix it?

    Steve Gilbert ([e-mail]) provided us with this answer:

        First, If you do a "fdisk wd1" (It may be wd1d, I don't
    remember what it wanted), it will list out the partition table
    for you.  This is something totally different from BSD's idea
    of a partition, mind you.  The last partition (#3) should be BSD.
    All of those figures are correct except for the "ending head" field
    which is set to 255 (thus, 256 heads).

    1. BACK UP EVERYTHING!

    2. fdisk -u wd1

        ...this will prompt you for the stuff you want to change.
        Remember, everything is correct except for the ending
        head.  Accept all the default values it gives you at first.
        You'll have to tell it that you want to explicitly define
        the beginning and ending values.

    3. My 420 MB Conner drive has 16 heads, so I just enter 15 as
       the ending head number.

    4. When you are back out of fdisk, you can do another fdisk wd1
       to make sure the values are correct.  Don't worry if you mess up,
       you can always change it again.  Anything you didn't back up is
       probably gone by now anyway :-)

    5. Reboot and watch NO error message pop up!

    ...remember that all you want to do is fdisk the drive.  You do NOT
    want to run disklabel again or newfs the partitions again.  This will
    write the incorrect 256 crap back.  I did this three times before
    I finally got smart and did it right.


2.2.8    What are the options for the boot up prompt?

    The options are supposed to be as follows:

<pre>
    -s............... boot into single user mode
    -a............... ask the user what device to use as root
              just before mounting it (Not presently supported)
    -d............... once you have the kernel loaded and VM and such up
              and going, drop into the kernel debugger.
              (great for debugging probe code)
</pre>

    A related question concerns the options on the 'reboot' program.
    These flags are as follows:
<pre>
    -a    Ask for a file name to reboot from
    -s    Reboot into single user mode
    -b    Don't reboot, just halt
    -r    Use compiled in Root device
    -c    Invoke the user configuration routines
    -d    Transfer control to the kernel debugger, if available
    -v    Print out all potentially important information
</pre>

    As with so many other things in the systems, each of these may
    (or may not) work for FreeBSD or NetBSD.  Your Mileage May Vary.

    One other note about 'reboot'.  There are some motherboards which
    do not reboot reliably.  Instead of rebooting, they simply hang.
    While this isn't a definitive answer, some folks have noticed
    that have the BIOS relocate option set seems to help them,
    especially with Micronics motherboards.  If you are having
    problems with your system not resetting after a reboot, try
    changing the setting on the BIOS relocation option.


2.2.9    I am having trouble installing WRT 'syslogd: bind: Can't assign
    requested address' errors.  What are some of the things I should
    look at?  I also am having trouble with the network: 'starting
    network ... ifconfig: localhost: badvalue'.

    This is caused by incorrect settings in /etc/netstart and/or
    /etc/hosts.

    In /etc/hosts, you must have a line that says:

<pre>
    127.0.0.1    localhost    localhost.{yourdomainhere}
</pre>

    Errors that will result if you don't do this: ifconfig will not
    be able to figure out what IP address goes with the name
    'localhost' and you'll get 'localhost: bad value.'

    In /etc/netstart, you must do:

<pre>
        ifconfig lo0 localhost
        route add {hostname} localhost
</pre>

    Errors that will result if you don't do this: the loopback
    device will not be properly configured and/or you will have no
    route to it. The result is that programs expecting to have
    networking enabled (including syslog and friends) will get
    horribly confused.

    *AND*, if you're not going to be directly connected to a
    network, you should change /etc/host.conf to say:

<pre>
        hosts
        bind
</pre>

    It's set up the other way around by default. I don't like it
    that way myself.

    Errors that can result if you don't do this: if you don't have
    a nameserver available to you, the resolver will have trouble
    translating hostnames into IP addresses.  Bogosity levels will
    be off the scale. (Note also that if you do have access to a
    nameserver, you need to set up /etc/resolv.conf to point to
    it.)  By changing the order, you'll be telling the resolver to
    check the host files for matches *first*, then roll over to the
    nameserver (if any) if no match is found.

    Make sure that:

    - There are no typos in any of the three files mentioned above.
    - There are no bogus non-ASCII characters in the files
    mentioned above.
    - All three files have their read permission bits set.

    Lastly, be very careful with /etc/hosts.equiv. If you add a
    hostname to it, say 'otherhost.domain,' then root on
    otherhost.domain will be able to rsh/rlogin to your machine
    without a password.

    Once you have everything set correctly, you should be able to
    type 'telnet localhost' and establish a connection to yourself.
    If you get an error such as 'localhost: unknown host' or
    'network unreachable' then you still have work to do.


2.2.10    When I start up my system, it hangs for three or four minutes
    during the 'netstart' program.  Our network nameserver is
    working OK, and I use it all the time; my resolv.conf file says
    to use the network nameserver.  Why would netstart have
    such problems using it?

    When the system is starting, the nameserver hasn't started yet.
    If you are using any names that must be resolved, the system
    will attempt to get the names from the nameserver,  When that
    fails (three timeouts at one minute apiece) the name will be
    resolved from the /etc/hosts file (if the name is available).

    There are essentially two ways to solve the immediate problem.
    The first is to reduce the number of entries you have in your
    /etc/hosts file to the absolute minimum you need for booting and
    change the order for host resolution from 'bind file' to 'file
    bind'.  If you specify a name in any of your start up files and
    the name server isn't available, you will still have the hang,
    but this is only a small annoyance.

    The second (and generally more effective) way to deal with the
    problem is to use only numeric addresses  in your /etc/* files.
    This way, the resolver is never called upon to figure out the
    addresses and your boot-up will always 'just work'.  This is
    sometimes more time intensive to set up, since all of the names
    in the files need to be removed and replaced with numbers.
    "C'est la vie".


2.2.11    I am having trouble getting net aliases to work.  What could
    some of the problems be?

    There are many things which will cause network aliases to not
    work right.  Here are a few:

    -  Use "netmask 0xffffff00" (or whatever is appropriate) for the
       first IP address, and "netmask 0xffffffff" for all aliases that
       happen to be in the same (sub)net as the primary one.  The
       reason this is right (no matter how odd it may seem) is you
       have multiple interfaces referring to the same network.  You
       *have* to chose one of the various interface addresses as the
       "gateway" for outgoing packets into this network, you cannot
       have them going out through a dozen of addresses simultaneously.
       The netmask 0xffffffff prevents the kernel from considering this
       IP address as a valid gateway (since it's not pointing to any
       network at all).

    The correct syntax in /etc/rc.local for declaring a net address
    alias (assuming you are updating the eth0 interface) is:

<pre>
        ifconfig eth0 xx.xx.xx.xx netmask 255.255.255.255 alias
        route add -host xx.xx.xx.xx localhost
        arp -s eth0 yy.yy.yy.yy.yy.yy proxy
</pre>

    Where the xx.xx.xx.xx are the host address for the alias and the
    yy.yy.yy.yy.yy.yy is the interface MAC address (if appropriate).


2.2.12    I'm having trouble with the networking code (specifically the
    PPP stuff to my ISP).  How can I debug NetBSD's networking?

    Bring the PPP connection up again.  Retry whatever-it-is that's
    failing.

    PPP includes a link-level checksum.  Watch the packet counts in
    the netstat -I ppp? output over time. Check carefully to see
    whether the PPP driver is recording input errors (frames whose
    CRC failed.)  Frames with bad checksums are counted in Ierr
    field.  A non-zero count indicates _something_, possibly
    overruns, is in fact garbling your PPP traffic.  If the packets
    are being discarded due to errors at the PPP level, they'll
    never even get as far as IP.


    Also, use netstat (or an SNMP daemon and monitor, if you prefer)
    to watch the rate of change of bad packets at the IP and TCP
    level.  I run "netstat -p ip" "netstat -p tcp".  One has to
    manually compute the rate of change; netstat's -i option means
    something different to, say, vmstat's.   (Adding  periodic
    sampling and rate-of-change to netstat would be a Cool Project.)

    At the IP level, the relevant statistics are
<pre>
            0 bad header checksums
            0 with size smaller than minimum
            0 with data size < data length
            0 with header length < data size
            0 with data length < header length
            0 with bad options
            0 with incorrect version number
        [...]
            0 output packets dropped due to no bufs, etc.
</pre>

    At the TCP level,  look for, e.g.,

<pre>
            0 discarded for bad checksums
            0 discarded for bad header offset fields
            0 discarded because packet too short
</pre>

    Any of these being non-zero would support the hypothesis of a
    bug in the PPP implementation.  Unlikely, but one never knows.

    It could be that a TCP ack got munged or dropped by your PPP
    link; or possibly somewhere else in the Internet.  That's not
    abnormal on busy links.

    What OS is your FTP peer running?  Is it a pre-2.0.0 Linux or an
    older version of a commercial Unix? If so, have you tried turning
    off rfc1323 on your NetBSD machine??


2.2.13    I want to hard wire my SCSI devices to a particular device
    number.  Is that possible?

    You can do the numbering any way you please.  Say I had two
    controllers.  You could number them as:

<pre>
    sd10    at scsibus0 target 0 lun ?
    sd11    at scsibus0 target 1 lun ?
    [...]
    sd20    at scsibus1 target 0 lun ?
    sd21    at scsibus1 target 1 lun ?
    [...]
</pre>

    Of course, you will need to add devices to the /dev/ directory
    for each of them, pointing to their correct major and minor
    numbers.

    You can also hardwire the 'scsibus' numbers, by doing something
    like the following (assuming "whatever" is the SCSI host adapter
    driver's name 8-):

    whatever0    at whateverbus? [whateverbus config info]
    scsibus0    at whatever0

    then

<pre>
    sd0    at scsibus0 target 0 lun 0
</pre>

    etc.


    That syntax won't work on ports which use 'old config,' but I
    believe an appropriate description of how to do it on them has
    already been posted.

    The most common configuration for locked down drive numbers is
    actually:

<pre>
    sd0    at scsibus0 target 0 lun 0
    sd1    at scsibus0 target 1 lun 0
    sd*     at scsibus? target ? lun ?      # SCSI disk drives
</pre>

    You can do the same thing with your tapes, CDs, and other SCSI
    devices as well.

<pre>
    st0    at scsibus0 target 6 lun 0
    st*     at scsibus? target ? lun ?      # SCSI tape drives
    cd0     at scsibus? target 5 lun 0
    cd*     at scsibus? target ? lun ?      # SCSI CD-ROM drives
    etc.
</pre>


2.3    Common installation problems.

    There are many common installation problems.  This section covers
    the most basic and common of these problems.  In addition to this
    section, you should also read through the other sections of the
    FAQ, since many of the less common questions are answered in other
    places in the doc.

2.3.2    Endless reboot cycles.

    Another incarnation of this symptom is that the disk geometry on
    your disk label (as installed by install) is different than the
    geometry your hard drive controller thinks it is using.  This
    will most often manifest itself on controllers that insist on
    operating in some type of translation mode.  Normally the fix is to
    find out what the controller geometry is and make the disk label
    agree.  There are programs available to help with this problem.


2.4    The computer just sits there, or 'that isn't right'.

    This class of problems is sometimes caused by an incorrect FTP of
    the boot disk.  Make sure that the files were grabbed in 'binary'
    mode and that the size reported back is 1244000 bytes.  Use the
    Unix program 'dd' or the DOS program RAWRITE to put these files
    onto the diskette.  In addition, this is the 'miscellaneous'
    section of the FAQ.  These problems are included here because they
    are usually preceded by 'I just finished installing...'

    Another incarnation of this problem is that, sometimes, the major
    or minor device numbers for a particular device may not get made
    correctly in the install (or upgrade) procedure.  If you have a
    problem where you can install and everything seems to go well
    until you try to boot onto the hard drive, try running the
    MAKEDEV script that resides in /dev.  More the file to see what
    kind of options you should include (if the sd0a drive needs to
    be fixed, for example, the command './MAKEDEV sd0' should get
    your devices back on the road.  If that doesn't work, try one of
    the many things below.  It could be any (or all) of them....


2.4.1    The boot disk works all right on one computer but not another.

    This could be a problem with many different pieces, some of which
    are:

    - Misconfigured hardware.  The iomem, IRQ, and other board
    settings must match the ones listed in the INSTALL.NOTES.
    Unfortunately, the INSTALL.NOTES are on the disk that will
    not boot.  You can grab them via FTP from many archive sites.
    This installation file may have many names.  Look for something
    kind of obvious (like 'install.notes' or 'readme' or
    'configuration guide') and you should find it.  Finally, there
    have been many reports (particularly with the BusLogic SCSI
    cards (specifically reported was the BT445C VLB host adapter)
    where the system seems to boot up, but starts getting
    'stray interrupt c' messages.  This is usually caused by people
    having there SCSI card set up on some IRQ other than the one
    that the kernel expects.  The factory default for this card
    seems to be IRQ 12, but the kernel wants the card at IRQ 11.
    Setting the card (using the configuration program supplied)
    changes the setting so that it matches the kernel and the card
    then works.

    - Unsupported hardware.  There are several SCSI controllers on the
    market that are not fully supported by 386bsd.  This is due in
    large part to the way these controllers work.  Instead of using
    a standard interface and command set for the controllers, most
    manufacturers make up their own controller interface language,
    which is then translated into SCSI commands which are
    interpreted by the drives.

    - One or more of the devices in the /dev directory on the
    intended root partition was either not created correctly or was
    not created at all.  Run the program MAKEDEV in the /dev/ directory
    to ensure that all of the correct devices are built.


2.4.2    Really strange errors in the various *BSD flavors.

2.4.2.2    Using the new code in NetBSD, I get a "panic: pdti 206067" in
    pmap_enter().  What should I do?

    Increase NKPDE in /sys/i386/include/pmap.h. The largest it should
    be 31; see question 3.2.8.1 for other useful values.  Be sure to
    keep the changes around as a patch file, since this is one of the
    files that will get overwritten during a source code update.
    Note that in the versions of NetBSD newer than 1.2.1, this value
    is computed, so you won't need to change it.


2.4.3a    I get the error "isr 15 and error: isr 17" on an NE2000 card.
2.4.3b    I have some card on IRQ2 and it doesn't work; why?
2.4.3c    I am getting lousy performance out of my network card.  What are
        some of the other possibilities?

    The description of this problem is that one of the cards in your
    system (most likely the VGA card) is either generating interrupts
    or is causing the IRQ 2 to be actively disabled.  Older VGA card
    used IRQ 2 during vertical retrace to prevent sparklies.

    One solution would be to plan on not using your Ethernet card
    (or any other card you want on IRQ 2) until you have rebuilt
    the kernel so that it expects it at an interrupt other than
    IRQ 2 or 9, re-jumper or reconfigure the card to match the IRQ
    you have selected, and enable it.

    From time to time, this problem will manifest itself as a general
    tendency of the network card to transfer either very sporadically
    or very slowly. It is precisely the same problem.

    James Van Artsdalen ([e-mail]) has offered at
    least one solution:

        If this is the problem, you can use Scotch tape to cover
        the IRQ 2 signal on the VGA's ISA connector.

    There has been some discussion as to whether scotch tape is really
    appropriate inside a card slot.  My answer would be "yes".  This is
    because the alternate solution of cutting the trace on the video
    board seems, to my mind, to reduce the value of the board.  It is
    possible that, in the future, with a bi-partite driver, you would
    want to catch the retrace interrupt to get rid of "sparklies" or to
    implement a driver for a very high resolution monitor for X.  If
    this happens, given a choice between alcohol and solder, I vote for
    alcohol.

    One other thing to look for (if your video card seems to be the
    culprit) is a jumper which enables or disables the card's IRQ 2.
    Newer cards may have a jumper of switch which does this, so take
    the time and look for it before you get the razor blade out.

    Either way, you will probably find that your VGA card uses IRQ 2
    strictly for compatibility with older cards.  With the advent of
    dual-ported memory for video cards, virtually all of these types
    of problems have disappeared.


2.4.4    What is the difference between IRQ2 and IRQ9?  Are they really
    the same, or are they really different?

    On the XT, there was one interrupt controller, an Intel 8259, which
    handled 8 interrupts numbered IRQ0 through IRQ7.  IRQs 2 through 7
    were accessible via bus lines IRQ2 through IRQ7.

    The AT had two interrupt controllers.  Due to the design of the
    8259, one has to be the master and the rest (up to 8) must be
    slaves.  Each slave controller output its interrupt request to
    and input on the master controller.  In the case of the AT, the
    master controller handles IRQ0 through IRQ7.  The slave handles
    IRQ8 through IRQ15.  The interrupt request from the slave to the
    master goes through IRQ2, which is termed the cascase input.
    IRQ2 was chosen to allow future compatabilty with the old XT
    hardware; it was the first IRQ that was 'available'.

    This means. of course, that the bus line for IRQ2 could no longer
    be used for external interrupts.  Instead, the bus line that WAS
    IRQ2 in the XT became IRQ9 on the AT.  This whole issue is
    confused further by the fact that some vendors refer to this
    external interrupt as IRQ2, while others refer to it as IRQ9.  In
    either case, if you are talking about an external interrupt, it
    means the same thing.

    BTW, IRQ8 is used for the Real Time Clock, and does not have an
    external interrupt.  Here is a map, in case anyone still needs it:

<pre>
        Internal    External    Function
        IRQ0        n/a        Refresh/Timer
        IRQ1        n/a        Keyboard
        IRQ2        n/a (AT only)    Cascade Input to Master
        IRQ3        IRQ3        Free (Com port)
        IRQ4        IRQ4        Free (Com port)
        IRQ5        IRQ5        Free
        IRQ6        IRQ6        Floppy Controller
        IRQ7        IRQ7        Free (Printer/Sound Card*)
        IRQ8        n/a        Real Time Clock
        IRQ9        IRQ2        Free (Network card)
        IRQ10        IRQ10        Free
</pre>

        etc.
    * NOTE:  The IRQ7 entry is spooky.  If you use the Interruptless
    printer driver (either from 386bsd, NetBSD, or FreeBSD) then you
    can still have an interrupting device (like a sound card) on
    interrupt 7.  Basically, you can as many devices on each IRQ as
    you want, but only one of them can be 'actively' interrupting.


2.4.5    Some of my SCSI devices (like a tape drive) don't work; why?

    Even with the newer systems, you run the risk of having a
    problem with a SCSI device from time to time.  There are some
    cards (like the new Adaptec 27* series) that software drivers
    are either not in the works or the documentation is simply
    unavailable.  Another culprit here is that some machines are
    very touchy about the quality and length of cables, as well
    as SCSI IDs.  There was one report of a older hard drive that
    took a little longer to spin up than the rest of the drives
    in the chain.  Whenever this drive was put early in the ID
    string (like 1 or 2) it would be 'not found' but if it was
    placed near the end (like after the tape drive) it would have
    spun up and been found.


2.4.6    I want to use the Adaptec 1542C SCSI controller.  What are the
    problems/tricks you need to know to get it working?

    The first thing to check when trying to use the 1542C is the setting
    of 'Enable Disconnection' under the 'SCSI Device Configuration'
    menu.  It should be set to YES for all devices, as the manual warns
    you.

    Matthias Urlichs ([e-mail]) has provided this
    description of the types of things that can cause problems for the
    controller and devices attached to it.

    The problem is that the Adaptec 1542C has (a) rather powerful line
    drivers, and (b) is sensitive to transient signals which can be
    induced by them via either a bad cable or a bad external terminator.

    A bad cable is almost any cable which doesn't meet SCSI-2 specs.

    A bad external terminator is one which doesn't adequately buffer
    its resistor network.

    So...

    - Remove the internal terminator from the last drive in your chain.
      Replace with an active SCSI-2 external terminator.  Side
      improvement: active terminators consume a bit less power.

    - Check cables.  Specifically, some cables carry less than the
      nominal 50 signal wires. Manufacturers sometimes think they can
      get away with this because almost all odd-numbered pins are GROUND
      anyway. So, if pins 1 and 3 or 3 and 5 are connected, you're
      likely to have a marginal cable.

    - Make sure that the terminator power is supplied by all devices
      and that the power pin is actually connected on your cable. The
      problem here is that some idiot device manufacturers save on
      2-cent diodes, which means that the thing will pull terminator
      power to ground if it's not plugged in.  (Two of these on one
      bus are even worse.)

    - Consider creating your own cabling. Take a 50-wire flat ribbon
      and press the appropriate connectors onto it in precisely the
      right places. (Move your devices as to minimize cable length.)
      Be aware that if a device has two external connectors, you must
      take the SCSI bus in at one connector and out at the other
      -- don't leave the other connector dangling; this isn't within
      the SCSI specs because the cable usually is too long.

    - Better but more expensive: use 2-twisted cable. (I.e., wires 1&2
      are twisted around each other, wire 3&4, ...) This will improve
      reliability because the wires are twisted at different rates.
      These cables have short non-twisted segments every 50 cm (1.5')
      so that you can press on your connectors instead of heating up
      that soldering iron.

    - While you're rebuilding your system anyway...: If you have more
      than one drive per power supply, check if these drives have
      adequate condensors to buffer their power.  I have two 80-MB
      Seagates which refused to work more than a few hours without
      glitches -- then I soldered two 10-uF Tantals onto their power
      connector and they've been flawless ever since.

    The terminator power is pin 26. Be aware that SCSI counts pins as
    they appear on a ribbon cable, not as they're sometimes numbered
    on the connectors.  Pin 25 is supposed to be disconnected.


2.4.7    Is there a SCSI utility which works to fix up the random
    problems I sometimes have with my drives?

    That depends on the problem.  One of the first things you can
    try is Ian Dell's ([e-mail]) SCSI Disk
    Doctor (sdd) package.  There are NetBSD i386 and Sparc
    executables on ftp://ftp.mono.org/pub/sdd.  FreeBSD uses a
    couple of utilities which come with the system (scsi and
    scsireprobe) to accomplish some of the same operations.  Try one
    of those (obviously based on your system type) and see if they
    don't fix your problem.  If they don't, then the prospects are
    pretty grim for your drive.


2.4.8    My system boots OK off the floppy, but once I try to boot from
    the hard drive, the message "changing root device to sd0a"
    appears and the system hangs.  What is the most likely thing
    that I have done wrong?

    A common cause for this is when all of the right devices aren't
    created on the root partition.  Since you say you can boot off
    of a floppy, do so and check to make sure everything in /dev
    exists.  You might consider running "MAKEDEV all" to be sure
    everything is created.

    (Ed.Note: I find that whenever I create a new kernel, it isn't a
    'bad' idea to run a precautionary MAKEDEV to make sure that the
    devices are created correctly.  Since I only build a new kernel
    about once a month, it isn't a very costly insurance policy.)

    Also, there are known problems with IRQ configurations and the
    PCI bus.  The system hanging right after the changing root device
    message usually indicates a misconfigured IRQ for the controller.
    The initial probes by most (all?) drivers are done in polled mode,
    only when mounting the disk for real does the kernel begin to do
    interrupt driven I/O and DMA.

    Is this system a PCI system?  Is the SCSI controller a PCI board?
    If so, make sure the IRQ configured in the PCI BIOS matches the
    IRQ configured for the card.

    Also, with PCI, forgetting to enable the slot for "master" in the
    BIOS setup or motherboard jumpers or putting a bus mastering card
    in a slave only slot will give similar symptoms.  The system may not
    have problems under DOS because some SCSI BIOS or device drivers
    don't actually use the DMA or bus mastering features of the
    card... {sigh}, they run in PIO mode under DOS.


2.5    Other common problems that are attributed to the installation
    process but are caused other places.

2.5.1    I want to use more than 16 Megabytes of memory.  Will any of the
    BSD based systems support it?

    When using NetBSD and FreeBSD, there is no SOFTWARE limitation on
    more than 16Meg of memory.  There are still hardware limitations.
    The limit is caused by DMA controllers which copy memory images
    around the system.  Many cards which people use in VESA and EISA
    machines either emulate ISA cards (in order to work with *BSD) or
    are really ISA cards.  There are reports of people having trouble
    with more than 64Meg of memory, but anyone rich enough to have
    that kind of memory should be paying for his OS. :-)

    Recently some folks have been reporting that they are getting
    warnings like these:

<pre>
        hostname /netbsd: sd0: not queued
        hostname /netbsd: aha0: DMA beyond end of ISA
        hostname /netbsd: sd0: not queued aha0: DMA beyond end of ISA
</pre>

    This error is caused when the buffer for I/O is beyond the address
    range that the ISA bus can reach.  With 16M you should be okay,
    however, some motherboards do reclaim all or part of the "lost"
    384K (from the I/O "hole" from A0000-FFFFF) and put it just beyond
    the end of the rest of the memory, so you actually get 16M plus a
    little bit.

    One fix is bounce buffers.  FreeBSD has implemented this, and NetBSD
    will as soon as they come up with a method that is compatible with
    all of the machines that NetBSD supports.

    Another fix is to either turn off the reclaiming of the extra memory
    (most motherboards that do this allow you to disable it), hack
    machdep.c to force the physical memory used to 16M, or use a 32 bit
    bus (EISA, VLB, or PCI) controller.

    Jordan K Hubbard ([e-mail]) has provided this
    explanation of the distinction:

    Just so long as you're using a DMA-using disk controller in EISA
    mode, rather than ISA mode, you can use more than 16 Meg of memory.

    For those who may find such a distinction confusing, let me explain:

    You can use an ISA controller (such as an Adaptec 1542) in an EISA
    machine, but as it will still think it's in an ISA box and refuse to
    use the extra address lines, this is no different than having an
    ISA machine as far as >16MB is concerned.

    You can use an EISA controller in "ISA mode", meaning it uses the
    older protocols for compatibility reasons (examples being Adaptec
    1742 in "standard" mode, DTC 3290 in "Adaptec" mode, etc.) and
    again, does not use the extra address lines.

    The only way to get full EISA, 32MB-of-memory-and-everything, mode
    is to use an EISA controller in full EISA mode (for Adaptec 1742,
    this is "enhanced" mode, for DTC 3290 it's "DTC" mode, the
    Ultrastor 24F in EISA (rather than IDE emulation) mode, etc.).

    - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

    In addition, several other types of ISA controllers which do NOT
    use DMA will not cause problems.  IDE, ESDI, and RLL controllers
    are examples of this type of card.  The discussion above also applies
    to VESA and VLB cards.

    So, the bottom line is that you are limited to the amount of memory
    that your DMA equipped devices can access.  Once again, the weakest
    link is the strength of your machine.


2.5.2    I tried to use a device in my computer that should be there.  When
    I did, I got a "Device not configured error."  What do I do now?

    Garrett A. Wollman ([e-mail]) provides us with this
    brief discussion in answer to a specific question.  It wears well
    as a generic answer as well.

    When any program tells you ``Device not configured'', it's trying
    to tell you something very important about what you tried to do:
    namely, that the device you tried to access is not configured
    into the running operating system.  This is the error message
    corresponding to ENXIO.

    There are three major causes for this error:

    1) The kind of device you requested was not configured into the
       system.  This is R.W.'s problem; the generic kernels
       are not distributed with the Berkeley Packet Filter enabled by
       default.  To correct this, you must add the appropriate device or
       pseudo-device to your kernel configuration file and recompile.
       (In this particular case, `pseudo-device bpfilter
       number-of-network-interfaces'.)

    2) The kind of device you requested was configured into the system,
       but either the device you requested would use more than the
       maximum you configured into the system, or if a physical device,
       was not found during autoconfiguration.  To solve this, either
       change your configuration file, or change the I/O settings on the
       device to match what the file says.

    3) The major or minor device number specified by the device's
       entry(ies) in /dev is incorrect.  To solve this, re-MAKEDEV the
       device (read the MAKEDEV script for more details).  Hopefully
       whatever change caused the kernel's internal device tables to get
       changed also updated your MAKEDEV script; otherwise, you will have
       to grovel through the kernel to see what is going on.

    4) A special case:  Although the 'c' drive on most BSD disks is
       the entire disk, in many NetBSD and FreeBSD systems, the
       entire drive is the 'd' disk.  This special case is wired
       into many programs, and is recognized by the kernel.  From
       time to time, folks will try and access the 'c' partition on
       their harddrive, only to be rebuffed with a 'device not
       configured' error.  Mostly, the 'c' partition is unavailable
       simply because the partition type is 'unused' even though it
       is allocated and has space set aside for it.

2.6    Customizing the system to meet my needs.
2.6.1    How do I get the system to not display the machine name, but
    display our company name?

    Modify the /etc/gettytab file so the default profile uses this:

<pre>
    :im=\r\n  Company Name (%t)\r\n\n:\
</pre>


2.6.2    I have a program that, under normal circumstances, starts once a
    second.  This regularly causes inetd to terminate the program
    with a 'server failing (looping), service terminated' error.
    How do I fix this?

    The inetd program has a 40 start per minute limit for all
    programs started out of inetd.conf.  You need to add a 'max
    starts' option on the end of your 'wait' or 'nowait' option.
    For example, try 'nowait.100' if you expect the program to start
    90 times a minute.

--
Dave Burgess                   Network Engineer - Nebraska On-Ramp, Inc.
*bsd FAQ Maintainer / SysAdmin for the NetBSD system in my spare bedroom
"Just because something is stupid doesn't mean there isn't someone that
doesn't want to do it...."