LDP
/
LDP
mirror of https://github.com/tLDP/LDP
0
1
Fork 0
Code Releases Activity

updated

This commit is contained in:
gferg 2002-01-26 20:55:59 +00:00
parent a7d08bef40
commit b996eef2ed
1 changed files with 139 additions and 73 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

212
LDP/howto/docbook/SCSI-2.4-HOWTO/SCSI-2.4-HOWTO.sgml
View File

@ -15,9 +15,18 @@
</affiliation>
</author>
</authorgroup>
<pubdate>2001-12-21</pubdate>
<pubdate>2002-01-26</pubdate>
<revhistory>
<revision>
<revnumber>1.6</revnumber>
<date>2002-01-26</date>
<authorinitials>dpg</authorinitials>
<revremark>
ATAPI cdrom selection
</revremark>
</revision>
<!-- Additional (*earlier*) revision histories go here -->
<revision>
<revnumber>1.5</revnumber>
<date>2001-12-21</date>
@ -64,6 +73,7 @@
<copyright>
<year>2000</year>
<year>2001</year>
<year>2002</year>
<holder>Douglas Gilbert</holder>
</copyright>
@ -146,7 +156,7 @@ version is at <ulink url="http://www.torque.net/scsi/SCSI-2.4-HOWTO.ps">
<literal>www.torque.net/scsi/SCSI-2.4-HOWTO.ps</literal></ulink>).
</para>
<para>
This document was built on 21st December 2001.
This document was built on 26th January 2002.
</para>
</chapter>
@ -869,13 +879,21 @@ to newsgroups or maintainers it is useful to include the output of this
file (e.g. <command>cat /proc/scsi/aic7xxx/0 </command>).
</para>
<para>
The general information on the proc pseudo file system can be found in
the kernel source file:
<filename>/usr/src/linux/Documentation/filesystems/proc.txt</filename>.
The cdrom driver provides information about attached cdrom devices
in the <filename>/proc/sys/dev/cdrom</filename> directory. This will
include both SCSI devices (i.e. those controlled by the sr driver) and
IDE devices (i.e. those controlled by the ide-cd driver).
See <xref linkend="srproc">.
</para>
<para>
The sg driver provides information about hosts and devices in directory
<filename>/proc/scsi/sg</filename> . See <xref linkend="sgproc">.
The sg driver provides information about its state and attached hosts and
devices in the <filename>/proc/scsi/sg</filename> directory.
See <xref linkend="sgproc">.
</para>
<para>
More general information on the proc pseudo file system can be found in
the kernel source file:
<filename>/usr/src/linux/Documentation/filesystems/proc.txt</filename>.
</para>
</chapter>
@ -1252,7 +1270,8 @@ ATA is the modern name for what was previously known as IDE and/or EIDE.
Note that the subsystem that controls ATA devices in Linux is called
the "IDE" subsystem for historical reasons.
</para></footnote>
transport.
transport. [The discussion in this section is also applicable to ATAPI
tape drives and ATAPI floppy drives.]
</para>
<para>
Currently both <command>cdrecord</command> and <command>cdparanoia</command>
@ -1269,48 +1288,55 @@ the ATAPI interface.
</para>
<para>
The default action of the IDE subsystem in Linux is to claim all ATA
devices for its drivers. In the case of an ATAPI cd writer, it will be
claimed by the ide-cd driver. Once this has happened, the SCSI subsystem
is unable to get control over an ATAPI device. The ide-scsi (pseudo
lower level SCSI) driver can only register ATAPI devices in the SCSI
subsystem that have <emphasis>not</emphasis> already been claimed
by IDE subsystem.
devices for its built-in drivers. In the case of an ATAPI cd writer, it
will be claimed by the built-in ide-cd driver. Once this has happened,
the SCSI subsystem is unable to get control over an ATAPI device. The
ide-scsi (pseudo lower level SCSI) driver can only register ATAPI devices
in the SCSI subsystem that have <emphasis>not</emphasis> already been
claimed by IDE subsystem.
</para>
<para>
To change this default action, parameters need to be passed to the IDE
subsystem that instruct it not to register a given device. Let us assume that
a cd writer is connected to the second IDE controller in the slave
position (i.e. <filename>/dev/hdd</filename>). One of the following
two techniques may
<footnote><para>
Another less targeted approach is available when both the ide-cd
and ide-scsi drivers are modules. That is to <command>rmmod ide-scsi;
rmmod ide-cd; modprobe ide-scsi</command>.
</para></footnote>
be used:
<itemizedlist>
<listitem><para>at kernel boot time pass the option: "hdd=scsi"
</para></listitem>
<listitem><para>load the ide-cd module with
<command>modprobe ide-cd ignore=hdd</command>. This can also be
accomplished by placing this line: "options ide-cd ignore=hdd"
in the <filename>/etc/modules.conf</filename> file
</para></listitem>
</itemizedlist>
The first technique can only be done at boot time (and can be automated
using lilo with an "append" in the <filename>/etc/lilo.conf</filename>
file). It assumes that the main IDE subsystem driver is built in, which
is almost always the case. The second technique assumes that the ide-cd
driver is a module which is not normally the case (i.e. most distributions
build it in). After one of these techniques has been applied, loading the
ide-scsi driver should register the cd writer at <filename>/dev/hdd
</filename> as <filename>/dev/sr&lt;n></filename> where the "n" is
the next available number.
Notice the <emphasis>built-in</emphasis> qualification in the previous
paragraph. If both the ide-cd and ide-scsi drivers are modules then the
first one loaded will claim the ATAPI cd devices (e.g. cd/dvd readers and
writers). Furthermore you can switch the controlling driver module by
<command>rmmod</command>-ing one and <command>modprobe</command>-ing the
other.
</para>
<para>
To find out if the above worked, the output of
<command>cat /proc/scsi/scsi</command> can be checked. Perhaps a simpler
method is to observe the "drive name:" line of
Probably the most flexible way to instruct the IDE core driver that you
want the cd writer at <filename>/dev/hdd</filename> accessible to
<command>cdrecord</command> is to use the kernel boot option:
"hdd=ide-scsi". This will cause the ide-cd driver to bypass
<filename>/dev/hdd</filename> (irrespective of whether ide-cd driver is
built-in or a module). As long as the ide-scsi driver is built-in or a
module then it will "capture" the cd writer
at <filename>/dev/hdd</filename> (with the IDE core driver loading the
ide-scsi module if required).
</para>
<para>
The ide-cd driver module can be instructed to ignore certain ATA devices
with the following syntax:
<programlisting>
modprobe ide-cd ignore='hdc hdd'
</programlisting>
In this case the ide-cd driver will ignore the devices at
<filename>/dev/hdc</filename> and <filename>/dev/hdd</filename> .
This effect can also be accomplished by placing a line like this:
"options ide-cd ignore=hdd" in the <filename>/etc/modules.conf</filename>
file.
</para>
<para>
A new option added in the lk 2.4 series is of the form "hdd=scsi".
This option seems to have a similar function to the "hdd=ide-scsi"
option discussed above. Furthermore "hdd=scsi" can only be used if
both the SCSI mid-level and the ide-scsi drivers are built into the
kernel (otherwise "BAD OPTION" is reported by the ide_setup function).
</para>
<para>
To find out whether an ATAPI cd device is "owned" by the SCSI subsystem,
the output of <command>cat /proc/scsi/scsi</command> can be checked.
Another technique is to observe the "drive name:" line of
<command>cat /proc/sys/dev/cdrom/info</command> for "sr" entries. The
following output is from my system:
<programlisting>
@ -1342,6 +1368,30 @@ subsystem, then cdroms should be mounted via the "sr" device name
and cd players should also use the "sr" device. Strangely the
<command>hdparm</command> command should still use the <filename>
/dev/hdd</filename> device file.
<footnote><para>
In the linux 2.4 kernel series there has been an increase in problems when
the ide-scsi driver is used so that <command>cdrecord</command> can control
ATAPI (IDE) cd writers. The problem may be related to the aggressive manner
in which the IDE subsystem attempts to optimize the speed of data transfers
on devices it controls. Some people experiencing timeouts and machine lockups
have found that reducing the DMA setting via the <command>hdparm</command>
command has fixed the problem. If the cd writer is connected to
<filename>/dev/hdd</filename> then users have reported success with one of
these two commands:
<programlisting>
hdparm -d0 -c1 /dev/hdd
hdparm -d 1 -X 34 /dev/hdd
</programlisting>
The first one turns off DMA completely while the second one sets it in
"multiword DMA mode 2". Cd writers do not need the types of speeds that
modern disks utilize. Even burning at "x16" implies a sustained transfer
rate of 16 times 150 KB/sec which is approximately 2.4 MB/sec, not
really that fast. There has also been a report that moving a cd writer off
a high speed IDE controller (Promise) and back to the motherboard's lower
speed IDE controllers has fixed a random IDE bus reset problem. Another
report suggests reducing (or turning off) the DMA on the IDE hard disk can
also stop lockups.
</para></footnote>
</para>
</sect2>
</sect1>
@ -1376,6 +1426,11 @@ file name examples in <xref linkend="dnames"> on device names. Any number of
tape drives (up to the overall limit of 32) can be added after the st
driver is loaded.
</para>
<para>
ATAPI tape drives can be controlled by this driver with help from the
ide-scsi pseudo adapter driver. The discussion in <xref linkend="sratapi">
also applies for ATAPI tape drives (and ATAPI floppies).
</para>
<sect2 id="stbparams">
<title>st boot parameters</title>
@ -1464,20 +1519,16 @@ book on the subject of SCSI programming and pass through mechanisms see
reference <link linkend="B3">B3</link>.
</para>
<para>
Currently the sg documentation focuses on the production version of sg found
in the lk 2.2 series. The abridged form is in the file <filename>
scsi-generic.txt</filename> which can also be found in the kernel source at
<filename>/usr/src/linux/Documentation/scsi-generic.txt</filename>.
The web site also contains a longer form called <filename>
scsi-generic_long.txt</filename>. This documentation describes
what is termed as "version 2" sg.
</para>
<para>
The sg driver in lk 2.4 has the "version 3" sg driver which adds an
additional interface and some new ioctl()s. The most interesting new ioctl()
is SG_IO which sends a SCSI command and waits for its response. The
additions and differences in the version 3 sg driver are documented on
the web site in the file <filename>scsi-generic_v3.txt</filename>.
The sg driver in lk 2.4 is "version 3" which adds an additional interface
structure and some new ioctl()s. The most interesting new ioctl()
is SG_IO which sends a SCSI command and waits for its response.
See the Linux Documentation Project site:
<ulink url="http://www.linuxdoc.org/HOWTO/SCSI-Generic-HOWTO/">
<literal>www.linuxdoc.org/HOWTO/SCSI-Generic-HOWTO/</literal></ulink>
for a full description of the sg driver.
A (possibly later) version of this document can be found at
<ulink url="http://www.torque.net/sg/p/sg_v3_ho.html">
<literal>www.torque.net/sg/p/sg_v3_ho.html</literal></ulink>.
</para>
<para>
The abbreviation "sg" is used within the kernel to refer both to the
@ -2249,16 +2300,19 @@ most recent ATA technology is outlined in the next paragraph.
</para>
</formalpara>
<formalpara><title>ATA 100</title>
<formalpara><title>ATA 133</title>
<para>
The ATA standards used by IDE devices have also been marching through
the adjectives (e.g. fast and ultra) and the numbers (e.g. 2, 33,
66 and 100). The most recent addition is ATA 100 which supports
burst rates of 100 MB/sec and up to 4 (was 2) devices per bus. [PCs often
have 2 ATA buses.] Both ATA 66 and 100 need a special cable. ATA
cables are relatively short precluding IDE devices being external to
66, 100 and 133). The most recent addition is ATA 133 which supports
burst rates of 133 MB/sec and up to 2 devices per bus. [PCs typically
have 2 and often 4 ATA buses.] ATA 66, 100 and 133 need a special cable.
ATA cables are relatively short precluding IDE devices being external to
the computer. Cable lengths have previously been limited to 18 inches
although 1 metre long cables have now appeared.
although 1 metre long cables have now appeared. Coincidently 133 MB/sec
in also the maximum throughput of the normal PCI bus found in most
PCs. The are higher speed (and wider) versions of PCI but they are
relatively rare.
</para>
</formalpara>
@ -2364,10 +2418,8 @@ No sr changes reported. As a related matter, the way in which the
IDE subsystem is informed that an ATA device (e.g. a cd writer) is
to be ignored has changed. [Instructing an ATA device to be ignored
by the IDE subsystem makes it a candidate to be recognized by the
ide-scsi pseudo driver.] Previously a kernel boot time option such as
"hdb=ide-scsi" would instruct the IDE subsystem to ignore the
device at <filename>/dev/hdb</filename>. In lk 2.4 this has become
"hdb=scsi".
ide-scsi pseudo driver.] In lk 2.4 the "hdb=scsi" kernel boot option
has been added. See <xref linkend="sratapi"> for more details.
</para>
</sect1>
@ -2503,6 +2555,17 @@ SCSI subsystem (and into block subsystem),
<listitem><para>
[2.4.15] lk 2.5.0 forks off so lk2.4.15==lk2.5.0 .
</para></listitem>
<listitem><para>
[2.4.17] add generic_unplug_device() call to scsi_wait_req(). This stops
long waits in SCSI_IOCTL_SEND_COMMAND.
</para></listitem>
<listitem><para>
[2.4.17] fix device scanning bug where, in some cases, the scsi_level
(i.e. SCSI standard adherence) was misplaced.
</para></listitem>
<listitem><para>
[2.4.17] major sg driver update, add mmap()-ed IO
</para></listitem>
</itemizedlist>
</para>
</sect1>
@ -3178,10 +3241,13 @@ if any of these links become stale.
<anchor id="W4">
[W4] The author's scsi generic (sg) site is:
<ulink url="http://www.torque.net/sg">
<literal>www.torque.net/sg</literal></ulink>. The documentation for
version 3 of the sg driver is also at this site:
<literal>www.torque.net/sg</literal></ulink>.
The Linux Documentation Project's site includes the
<ulink url="http://www.linuxdoc.org/HOWTO/SCSI-Generic-HOWTO/">
<literal>www.linuxdoc.org/HOWTO/SCSI-Generic-HOWTO/</literal></ulink> .
A (possibly later) version of that document can be found at
<ulink url="http://www.torque.net/sg/p/sg_v3_ho">
<literal>www.torque.net/sg/p/sg_v3_ho</literal></ulink>.
<literal>www.torque.net/sg/p/sg_v3_ho</literal></ulink> .
The sg_utils and sg3_utils packages, as tarballs and as binary and source
rpms can also be found on this page. These packages and others available
for the sg driver are discussed at