David Hinds,
v2.118, 06 December 2003
This document describes how to install and use PCMCIA Card Servicesfor Linux, and
answers some frequently asked questions. The latestversion of this document can always
be found at
Table of Contents
General information and hardware requirementsIntroductionCopyright notice and
disclaimerWhat is the latest version, and where can I get it?What systems are
supported?What cards are supported?When will my favorite (unsupported) card become
supported?Mailing lists and other information sourcesCompilation and
installationPrerequisites and kernel setupKernel PCMCIA supportInstallationStartup
optionsSystem resource settingsNotes about specific Linux distributionsResolving installation
and configuration problemsBase PCMCIA kernel modules do not loadSome client driver
modules do not loadISA interrupt scan failuresIO port scan failuresMemory probe
failuresFailure to detect card insertions and removalsInterrupt delivery problemsSystem
resource starvationResource conflict only with two cards insertedDevice configuration does
not completeUsage and featuresTools for configuring and monitoring PCMCIA
devicesOverview of the PCMCIA configuration scriptsPCMCIA network adaptersPCMCIA
serial and modem devicesPCMCIA parallel port devicesPCMCIA SCSI adaptersPCMCIA
memory cardsPCMCIA ATA/IDE card drivesMultifunction cardsAdvanced topicsResource
allocation for PCMCIA devicesPCI interrupt configuration problems and solutionsHow can I
have separate device setups for home and work?Booting from a PCMCIA deviceDealing with
unsupported cardsConfiguring unrecognized cards Adding support for an
NE2000-compatible ethernet cardPCMCIA floppy interface cardsDebugging tips and
programming informationSubmitting useful problem reportsInterpreting kernel trap
reportsLow level PCMCIA debugging aids/proc/bus/pccardWriting Card Services drivers for
new cardsGuidelines for PCMCIA client driver authorsGuidelines for Linux distribution
General information and hardware requirements
Card Services for Linux is a complete PCMCIA or``PC Card'' support package. It includes a
set of loadablekernel modules that implement a version of the Card Servicesapplications
program interface, a set of client drivers for specificcards, and a card manager daemon that
can respond to card insertionand removal events, loading and unloading drivers on demand.
Itsupports ``hot swapping'' of most card types, so cards can be safelyinserted and ejected at
any time.
page 1 of 54 Howto.pdf
This software is a work in progress. It contains bugs, and should beused with caution. I'll do
my best to fix problems that are reportedto me, but if you don't tell me, I may never know. If
you use thiscode, I hope you will send me your experiences, good or bad!
If you have any suggestions for how this document could be improved,please let me know
Copyright notice and disclaimer
Copyright (c) 1998-2002 David A. Hinds
This document may be reproduced or distributed in any form without myprior permission.
Modified versions of this document, includingtranslations into other languages, may be freely
distributed, providedthat they are clearly identified as such, and this copyright isincluded
This document may be included in commercial distributions without myprior consent. While it
is not required, I would like to be informedof such usage. If you intend to incorporate this
document in apublished work, please contact me to make sure you have the latestavailable
This document is provided ``AS IS'', with no express or impliedwarranties.
information in this document at your own risk.
Use the
What is the latest version, and where can I get it?
The current major release of Card Services is version 3.2, and minorupdates or bug fixes are
numbered 3.2.1, 3.2.2, and so on.
Source code for the latest version is available on the web at,
aspcmcia-cs-3.2.?.tar.gz. You may find more than one releasenumber here. It is up to you
to decide which version is moreappropriate, but the CHANGES file will summarize the
mostimportant differences.
Pre-compiled drivers are included with current releases of essentiallyall major Linux
distributions, including Slackware, Debian, Red Hat,Caldera, and SuSE, among others. So
generally there is no need tocompile the drivers from scratch.
What systems are supported?
This package should run on almost Intel-based Linux-capable laptop.It also runs on some
Alpha, PowerPC, ARM, and MIPS platforms. Mostcommon socket controllers are supported.
Card docks for desktopsystems should work as long as they use a supported controller,
andare plugged directly into the ISA or PCI bus, as opposed toSCSI-to-PCMCIA or
IDE-to-PCMCIA adapters. The following controllersare recognized by the supplied socket
* Cirrus Logic (now Basis Communications) PD6710, PD6720, PD6722,PD6729, PD6730,
* ENE Technology CB1211, CB1225, CB1410, CB1420
page 2 of 54 Howto.pdf
* Intel i82365sl B, C, and DF steps, 82092AA
* O2Micro OZ6729, OZ6730, OZ6812, OZ6832, OZ6833, OZ6836, OZ6860,OZ6922,
OZ6933, OZ6912
* Omega Micro 82C365G, 82C092G
* Ricoh RF5C296, RF5C396, RL5C465, RL5C466, RL5C475, RL5C476,RL5C477,
* SMC 34C90
* Texas Instruments PCI1031, PCI1130, PCI1131, PCI1210, PCI1211,PCI1220, PCI1221,
PCI1225, PCI1250A, PCI1251A, PCI1251B, PCI1410,PCI1410A, PCI1420, PCI1450,
PCI1451A, PCI1510, PCI1520, PCI1620,PCI4410, PCI4410A, PCI4450, PCI4451,
PCI4510, PCI4520, PCI7410,PCI7510, PCI7610
* Toshiba ToPIC95, ToPIC97, ToPIC100 (experimental, incomplete)
* Vadem VG465, VG468, VG469
* VLSI Technologies 82C146, VCF94365
* VIA VT83C469
* Databook DB86082, DB86082A, DB86084, DB86084A, DB86072, DB86082B
Other controllers that are register compatible with the Intel i82365slwill generally work, as
Due to the rapid pace of technological change for laptop hardware, newcontrollers appear
frequently, and there may be delays between when anew model appears on the market, and
when driver support becomesavailable.
Support for Toshiba's ToPIC bridges was hindered for a long time by alack of sufficiently
detailed technical documentation. While somedatasheets have been available, a few
idiosyncracies of the ToPICchips were not adequately explained. Toshiba has given some
directtechnical help on some of these issues, and I think the major oneshave been resolved.
However, with the introduction of kernel PCMCIAsupport in 2.4.* and later kernels, some new
Toshiba bugs may havecropped up in the new socket driver code.
The Motorola 6AHC05GA controller used in some Hyundai laptops is notsupported. The
custom host controller in the HP Omnibook 600 isalso unsupported.
What cards are supported?
The current release includes drivers for a variety of ethernet cards,a driver for modem and
serial port cards, several SCSI adapterdrivers, a driver for ATA/IDE drive cards, and memory
card driversthat should support most SRAM cards and some flash cards.
TheSUPPORTED.CARDS file included with each release of Card Serviceslists all cards that
are known to work in at least one actual system.
The likelihood that a card not on the supported list will work dependson the type of card.
Essentially all modems should work with thesupplied driver. Some network cards may work
if they are OEM versionsof supported cards. Other types of IO cards (frame buffers,
soundcards, etc) will not work until someone writes the appropriatedrivers.
page 3 of 54 Howto.pdf
When will my favorite (unsupported) card become supported?
Unfortunately, they usually don't pay me to write device drivers, soif you would like to have a
driver for your favorite card, you areprobably going to have to do at least some of the work.
Ideally, I'dlike to work towards a model like the Linux kernel, where I would beresponsible
mainly for the ``core'' driver code and other authorswould contribute and maintain client
drivers for specific cards. TheSUPPORTED.CARDS file mentions some cards for which
driver work iscurrently in progress. I will try to help where I can, but be warnedthat
debugging kernel device drivers by email is not particularlyeffective.
Mailing lists and other information sources
The Linux PCMCIA information page is at, and has bug
tracking,support and feature requests, and a variety of PCMCIA related messageforums.
Users can request email notification of new responses toparticular questions, or notification
for all new messages in a givencategory. I hope that this will become a useful repository
ofinformation, for questions that go beyond the scope of the HOWTO.
The Linux Laptop Page at http://www.linux-on-laptops.comhas links to a vast number of sites
that have information aboutconfiguring specific types of laptops for Linux. There is also
asearchable database of system configuration information, and pointersto a variety of
laptop-related mailing lists.
Compilation and installation
Prerequisites and kernel setup
Before starting, you should think about whether you really need tocompile the PCMCIA
package yourself. All common Linux distributionscome with pre-compiled driver packages.
Generally, you only need toinstall the drivers from scratch if you need a new feature of
thecurrent drivers, or if you've updated and/or reconfigured your kernelin a way that is
incompatible with the drivers included with yourLinux distribution. While compiling the
package is not technicallydifficult, it does require some general Linux familiarity.
The following things should be installed on your system before youbegin:
* A 2.0, 2.2, 2.4, or 2.6 series kernel source tree.
* An appropriate set of module utilities.
* (Optional) the ``XForms'' X11 user interface toolkit.
You need to have a complete linux source tree for your kernel, notjust an up-to-date kernel
image. The driver modules contain somereferences to kernel source files. While you may
want to build a newkernel to remove unnecessary drivers, installing PCMCIA does notrequire
you to do so.
page 4 of 54 Howto.pdf
Current ``stable'' kernel sources and patches are available
from Currentmodule utilities can be found in the
same locations.
In the Linux kernel source tree, the Documentation/Changesfile describes the versions of all
sorts of other system componentsthat are required for that kernel release. You may want to
checkthrough this and verify that your system is up to date, especially ifyou have updated
your kernel. If you are using a development kernel,be sure that you are using the right
combination of shared librariesand module tools.
On x86 based systems, if you plan to use 16-bit PC Card devices, youshould also enable
CONFIG_ISA, for recent kernels. These cardsbehave much like ISA devices,
and the PCMCIA drivers useCONFIG_ISA to judge whether a platform supports
ISA businterrupts.
When configuring your kernel, if you plan on using a PCMCIA ethernetcard, you should turn
on networking support but turn off the normalLinux network card drivers, including the
``pocket and portableadapters''. The PCMCIA network card drivers are all implemented
asloadable modules.
Any drivers compiled into your kernel will onlywaste space.
If you want to use SLIP, PPP, or PLIP, you do need to either configureyour kernel with these
enabled, or use the loadable module versions ofthese drivers.
In order to use a PCMCIA token ring adapter, your kernel should beconfigured with ``Token
Ring driver support'' (CONFIG_TR)enabled, though you should leave
If you want to use a PCMCIA IDE adapter, your kernel should beconfigured with
enabled, for 2.0.*kernels. Newer kernels do not require a special configurationsetting.
If you will be using a PCMCIA SCSI adapter, then enableCONFIG_SCSI when
configuring your kernel. Also, enable any toplevel drivers (SCSI disk, tape, cdrom, generic)
that you expect touse. All low-level drivers for particular host adapters should bedisabled, as
they will just take up space.
This package includes an X-based card status utility calledcardinfo. This utility is based on a
freely distributed userinterface toolkit called the XForms Library. This library isavailable as a
separate package with most Linux distributions. If youwould like to build cardinfo, you should
install XForms and allthe normal X header files and libraries before configuring the
PCMCIApackage. This tool is completely optional.
Kernel PCMCIA support
PCMCIA driver support is included in the 2.4 and later linux kerneltrees. While it shares most
of the same code with the standalonePCMCIA driver package, there are some important
differences. Thekernel PCMCIA support is also still evolving.
The kernel PCMCIA code has the same functionality as the driver sideof the pcmcia-cs
package. It does not eliminate the need to installthe pcmcia-cs package, since it requires the
same user tools(cardmgr, cardctl, /etc/pcmcia/* files). Thedrivers in pcmcia-cs can still be
built for 2.4 kernels, so you have a choice of using either the in-kernel PCMCIA drivers, or
thedrivers included in pcmcia-cs. With 2.5 and later kernels, thestandalone drivers cannot be
page 5 of 54 Howto.pdf
To use the kernel PCMCIA drivers, configure the kernel
usuallyCONFIG_CARDBUS enabled. On x86 based systems,
CONFIG_ISAshould also be enabled. The drivers can either be built into
thekernel or built as modules. PCMCIA client driver options are listedin their regular driver
categories; thus, PCMCIA network drivers arein a submenu of network drivers, and PCMCIA
serial drivers are in asubmenu of character drivers.
In the standalone pcmcia-cs drivers, the i82365 module supportsboth ISA-to-PCMCIA,
PCI-to-PCMCIA, and PCI-to-CardBus bridges. TheCardBus socket driver in the 2.4 tree is
the yenta_socket driver.It is selected by the CONFIG_CARDBUS
option. In your PCMCIAstartup options, this driver should be specified in place of thei82365
driver. The kernel version of the i82365 driver,selected by CONFIG_I82365,
only supports ISA-to-PCMCIA bridges.PCI-to-PCMCIA bridges that are not CardBus capable,
like the CirrusPD6729, are not supported at all by the kernel PCMCIA drivers.
When compiling the standalone PCMCIA package, the Configure scriptdecides whether or
not to build any kernel modules by looking at thevalue of the CONFIG_PCMCIA
option in your kernel configuration.If CONFIG_PCMCIA is enabled, then by
default, no drivercomponents are built. If CONFIG_PCMCIA is disabled, then all
themodules will be built and installed. It is safe to compile the usertools (cardmgr, cardctl,
etc) in a PCMCIA package whose version numberdiffers from the PCMCIA version number in
the kernel source tree. Thekernel PCMCIA header files take precedence over the ones
included inthe PCMCIA package, if CONFIG_PCMCIA is enabled.
Here is a synopsis of the installation process:
* Unpack pcmcia-cs-3.2.?.tar.gz in /usr/src.
* Run ``make config'' in the new pcmcia-cs-3.2.? directory.
* Run ``make all'', then ``make install''.
* Customize the startup script and the option files in/etc/pcmcia for your site, if needed.
If you plan to install any contributed client drivers not included inthe core PCMCIA
distribution, unpack each of them in the top-leveldirectory of the PCMCIA source tree. Then
follow the normal buildinstructions. The extra drivers will be compiled and
Running ``make config'' prompts for a few configuration options,and checks out your system
to verify that it satisfies allprerequisites for installing PCMCIA support. In most cases, you'll
beable to just accept all the default configuration options. Be sure tocarefully check the
output of this command in case there are problems.The following options are available:
Linux kernel source directory?
This is the location of the source tree for the kernel you want to usewith PCMCIA. Often this
is /usr/src/linux, but the defaultlocation depends on what Linux distribution you're using (or on
whereyou've chosen to place your kernel source tree).
page 6 of 54 Howto.pdf
Build 'trusting' versions of card utilities?
Some of the support utilities (cardctl and cardinfo) can becompiled either in ``safe'' or
``trusting'' forms. The ``safe'' formsprevent non-root users from modifying card configurations.
The``trusting'' forms permit ordinary users to issue commands to suspendand resume cards,
reset cards, and change the current configurationscheme. The default is to build the safe
Include 32-bit (CardBus) card support?
This option must be selected if you wish to use 32-bit CardBus cards.It is not required for
CardBus bridge support, if you only plan to use16-bit PC Cards.
Include PnP BIOS resource checking?
This builds additional code into the PCMCIA core module to communicatewith a system's
PnP BIOS to obtain resource information for built-in``motherboard'' devices (serial and
parallel ports, sound, etc), tohelp avoid resource conflicts. If enabled, some extra resource
fileswill be created under /proc/bus/pccard, and the lspnpand setpnp tools can be used to
view and manipulate PnP BIOSdevices. However, this setting causes problems on some
laptops and isnot turned on by default.
Module install directory?
The directory that new kernel modules will be installed into.Normally this should be the
subdirectory of /lib/modules thatmatches your kernel version.
How to set kernel-specific options?
There are a few kernel configuration options that affect the PCMCIAtools. The configuration
script can deduce these from the runningkernel (the default and most common case).
Alternatively, if you arecompiling for installation on another machine, it can read
theconfiguration from a kernel source tree, or each option can be setinteractively.
The Configure script can also be executed non-interactively, forautomatic builds or to quickly
reconfigure after a kernel update.Some additional less-frequently-used options can be only
be set fromthe command line. Running ``Configure --help'' lists allavailable options.
Running ``make all'' followed by ``make install'' will buildand then install the kernel modules
and utility programs. Kernelmodules are installed under
/lib/modules/>version>/pcmcia.The cardmgr and cardctl programs are installed in/sbin.
If cardinfo is built, it is installed in/usr/bin/X11.
Configuration files will be installed in the /etc/pcmciadirectory. If you are installing over an
older version, your oldconfig scripts will be backed up before being replaced. The
savedscripts will be given an *.O extension.
If you don't know what kind of host controller your system uses, youcan use the
pcic_probe utility in the cardmgr/subdirectory to determine this. There are
several major types: theDatabook TCIC-2 type and the Intel i82365SL-compatible type. With
thekernel PCMCIA subsystem, Intel compatible controllers are furthersubdivided into ISA-bus
16-bit bridges, and PCI-based CardBus bridges.
In a few cases, the pcic_probe command will be unable to determineyour
controller type automatically. If you have a Halikan NBD 486system, it has a TCIC-2
controller at an unusual location: you'll needto edit rc.pcmcia to load the tcic module, and
also set thePCIC_OPTS parameter to ``tcic_base=0x02c0''.
On some old pre-PCI systems using Cirrus controllers, including theNEC Versa M, the BIOS
puts the controller in a special suspended stateat system startup time. On these systems,
the pcic_probe command willfail to find any known host controller. If this
happens, editrc.pcmcia and set PCIC to i82365, and PCIC_OPTS
page 7 of 54 Howto.pdf
Startup options
The PCMCIA startup script recognizes several groups of startupoptions, set via environment
variables. Multiple options should beseparated by spaces and enclosed in quotes.
Placement of startupoptions depends on the Linux distribution used. They may be
placeddirectly in the startup script, or they may be kept in a separateoption file. See the
Notes about specific Linux distributions
This variable specifies whether PCMCIA support should be started up,or not. If it is set to
anything other than ``yes'', then the startupscript will be disabled.
This identifies the PC Card Interface Controller driver module. Thereare several options:
``tcic'', ``i82365'', and (for the kernel PCMCIAsubsystem) ``yenta_socket''.
Virtually all current controllers are inthe ``i82365'' group for the standalone drivers, and
``yenta_socket''for the kernel drivers. This is the only mandatory option setting.
This specifies options for the PCIC module. Some host controllershave optional features that
may or may not be implemented in aparticular system. In some cases, it is impossible for the
socketdriver to detect if these features are implemented. See thecorresponding man page
for a complete description of the availableoptions.
This specifies options for the pcmcia_core module, whichimplements the core
PC Card driver services.
See ``manpcmcia_core'' for more information.
This specifies options to be passed to the cardmgr daemon. See``man cardmgr'' for more
If set, then the PC Card configuration scheme will be initialized tothis at driver startup time.
See the Overview of the PCMCIA configuration scripts
The low level socket drivers, tcic and i82365, have variousbus timing parameters that may
need to be adjusted for certain systemswith unusual bus clocking. Symptoms of timing
problems can includecard recognition problems, lock-ups under heavy loads, high errorrates,
or poor device performance. Only certain host bridges haveadjustable timing parameters:
check the corresponding man page to seewhat options are available for your controller. Here
is a briefsummary:
* ISA-bus Cirrus controllers have numerous configurable timingparameters. The most
important seems to be the cmd_time flag,which determines the length of
PCMCIA bus cycles. Fast 486 systems(i.e., DX4-100) seem to often benefit from
increasing this from 6 (thedefault) to 12 or 16.
* The Cirrus PD6729 PCI controller has the fast_pci flag, whichshould be set if
the PCI bus speed is greater than 25 MHz.
* For Vadem VG-468 controllers, the async_clock flag changes therelative
clocking of PCMCIA bus and host bus cycles. Setting thisflag adds extra wait states to
some operations.
However, I have yetto hear of a laptop that needs this.
page 8 of 54 Howto.pdf
* The pcmcia_core module has the cis_speed parameter
forchanging the memory speed used for accessing a card's Card InformationStructure
(CIS). On some systems, increasing this parameter (i.e.,slowing down card accesses)
may fix card recognition problems.
* Another pcmcia_core parameter, io_speed, can be used to
slowdown accesses to IO cards. It may help in certain cases with systemsthat have
out-of-spec PCMCIA bus timing.
* This is not a timing issue, but if you have more than one ISA-to-PCMCIAcontroller in your
system or extra sockets in a laptop docking station,the i82365 module should be loaded
with the extra_socketsparameter set to 1. This should not be necessary for
detection ofPCI-to-PCMCIA or PCI-to-CardBus bridges.
Here are some timing settings for a few old systems:
* On the ARM Pentium-90 or Midwest Micro Soundbook Plus,
use``freq_bypass=1 cmd_time=8''.
* On a Compaq Presario 1220, try ``setup_time=1''.
* On a Midwest Micro Soundbook Elite, use ``cmd_time=12''.
* On a Gateway Liberty, try ``cmd_time=16''.
* On a Samsung SENS 810, use ``fast_pci=1''.
Card readers for desktop systems
While almost all PCMCIA card readers and card docks work fine underLinux, some require
special startup options because they do not behaveexactly like laptop PCMCIA bridges. PCI
card readers, in particular,may handle interrupts differently. Some of the following
parametersettings are only available for the i82365 module in thestandalone drivers; the
kernel's yenta_socket driver is notconfigurable.
* The Linksys ProConnect PCMRDWR and Antec DataChute ISA card readersare ``ISA
Plug and Play'' devices. To use these, you must firstactivate them with the Linux isapnp
tools. See the man pages forpnpdump and isapnp for more information.
* For Chase CardPORT and Altec ISA card readers using the Cirrus
PD6722ISA-to-PCMCIA bridge, the i82365 driver should be loaded with
a``has_ring=0'' parameter to prevent irq 15 conflicts.
* For Elan P-series PCI card readers based on the Cirrus PD6729PCI-to-PCMCIA bridge
chip, the i82365 driver requires a``irq_mode=1'' parameter.
* For the Sycard PCChost1200 host adapter, the i82365 driverrequires a ``p2cclk=1''
* For the Alex Electronics PCICBI host adapter based on the TI 1221bridge, the i82365
driver requires ``p2cclk=1 irq_mode=0''as well as PCMCIA driver release
3.1.23 or later.
* With SCM Microsystems SBP series PCI card readers (which are alsobeing distributed
with Lucent WaveLAN IEEE cards), and for theSynchrotech PCM-CR-PC2IF and
PCM-CR-PC2IR, it is necessary tospecify ``irq_mode=0'' for the i82365
module, to force useof PCI interrupts.
page 9 of 54 Howto.pdf
* For the ActionTec PC 750 card reader, and for the Antec Datachute PCIcard reader, the
i82365 driver requires a ``irq_list=0''parameter, to indicate that ISA interrupts
are unavailable.
* The PLX Technologies PCI9052 (also sold as the Linksys WDT11) is not ageneral
purpose PCMCIA card reader at all: it is a PCI interface cardfor use with certain wireless
adapters, that makes them look likeordinary PCI devices. These devices are not
System resource settings
Card Services should automatically avoid allocating IO ports andinterrupts already in use by
other standard devices. It will alsoattempt to detect conflicts with unknown devices, but this
is notcompletely reliable. In some cases, you may need to explicitlyexclude resources for a
device in /etc/pcmcia/config.opts.
Here are some resource settings for specific laptop types. View thislist with suspicion: it may
give useful hints for solving problems,but it is inevitably out of date and certainly contains
mistakes.Corrections and additions are welcome.
* On the AMS SoundPro, exclude irq 10.
* On some AMS TravelPro 5300 models, use memory 0xc8000-0xcffff.
* On the BMX 486DX2-66, exclude irq 5, irq 9.
* On the Chicony NB5, use memory 0xda000-0xdffff.
* On the Compaq Presario 900Z, exclude port 0x3b0-0x3bb.
* On the Compaq Presario 1020, exclude port 0x2f8-0x2ff, irq 3, irq 5.
* On the Compaq Presario 2120EA, exclude irq 10.
* On the Dell Inspiron 7000, exclude irq 3, irq 5.
* On the Dell Inspiron 8000, exclude port 0x800-0x8ff.
* On the Fujitsu C series, exclude port 0x200-0x27f.
* On the HP Omnibook 4000C, exclude port 0x300-0x30f.
* On the HP Omnibook 4100, exclude port 0x220-0x22f.
* On the IBM ThinkPad 380, and maybe the 385 and 600 series, excludeport 0x230-0x233,
and irq 5.
* On IBM ThinkPad 600 and 770 models with internal modems, exclude port0x2f8-0x2ff.
* On the IBM ThinkPad 600E and 770Z, change the high memory window
* On the Micron Millenia Transport, exclude irq 5, irq 9.
* On the NEC Versa M, exclude irq 9, port 0x2e0-2ff.
* On the NEC Versa P/75, exclude irq 5, irq 9.
page 10 of 54 Howto.pdf
On the NEC Versa S, exclude irq 9, irq 12.
* On the NEC Versa 6000 series, exclude port 0x2f8-0x33f, irq 9, irq 10.
* On the NEC Versa SX, exclude port 0x300-0x31f.
* On the ProStar 9200, Altima Virage, and Acquiline Hurricane DX4-100,exclude irq 5, port
0x330-0x35f. Maybe use memory 0xd8000-0xdffff.
* On the Siemens Nixdorf SIMATIC PG 720C, use memory 0xc0000-0xcffff,port
* On the TI TravelMate 5000, use memory 0xd4000-0xdffff.
* On the Toshiba Satellite 4030CDS, exclude irq 9.
* On the Toshiba T4900 CT, exclude irq 5, port 0x2e0-0x2e8, port0x330-0x338.
* On the Toshiba Tecra 8000, exclude irq 3, irq 5, irq 9.
* On the Twinhead 5100, HP 4000, Sharp PC-8700 and PC-8900, excludeirq 9 (sound), irq
* On an MPC 800 Series, exclude irq 5, port 0x300-0x30f for the CD-ROM.
PowerBook specific settings
On PowerPC based PowerBook systems, the default system resources
in/etc/pcmcia/config.opts file are no good at all. Replace allthe IO port and window
definitions with something like:
include port 0x100-0x4ff, port 0x1000-0x17ff
include memory 0x80000000-0x80ffffff
Notes about specific Linux distributions
This section is incomplete. Corrections and additions are welcome.
Debian uses a System V boot script arrangement. The PCMCIA startupscript is installed as
/etc/init.d/pcmcia. New packages use/etc/default/pcmcia for startup options; older versions
used/etc/pcmcia.conf for this purpose. Debian's syslogconfiguration will place kernel
messages in /var/log/messagesand cardmgr messages in /var/log/daemon.log.
Debian distributes the PCMCIA system in two packages: the``pcmcia-cs'' package contains
cardmgr and other tools, manpages, and configuration scripts; and the
``pcmcia-modules''package contains the kernel driver modules.
Starting with 3.1.25, a clean PCMCIA install will identify Debiansystems and create a special
network.opts file that, in theabsence of other network configuration settings, uses
Debian'sifup and ifdown commands to configure a network card basedon settings in
page 11 of 54 Howto.pdf
Red Hat, Caldera, Mandrake
These distributions use a System V boot script organization. ThePCMCIA startup script is
installed as/etc/rc.d/init.d/pcmcia, and boot options are kept in/etc/sysconfig/pcmcia. Beware
that installing the Red Hatpackage may install a default boot option file that has
PCMCIAdisabled. To enable PCMCIA, the ``PCMCIA'' variable should beset to ``yes''. Red
Hat's default syslogd configuration willrecord all interesting messages in /var/log/messages.
Red Hat's PCMCIA package contains a replacement for the network setupscript,
/etc/pcmcia/network, which meshes with the Red Hatlinuxconf configuration system. This is
convenient for the casewhere just one network adapter is used, with one set of
networkparameters, but does not have the full flexibility of the regularPCMCIA network script.
Compiling and installing a clean PCMCIA sourcedistribution will overwrite the network script,
breaking the link tothe Red Hat tools. If you prefer using the Red Hat tools, either useonly
Red Hat RPM's, or replace /etc/pcmcia/network.opts withthe following:
if [ -f /etc/sysconfig/network-scripts/ifcfg-$2 ] ; then
start_fn () {
. /etc/sysconfig/network-scripts/ifcfg-$1
if [ "$ONBOOT" = "yes" ] ; then /sbin/ifup $1 ; fi
stop_fn () {
/sbin/ifdown $1
Starting with the 3.1.22 release, the PCMCIA installation script willautomatically append a
variant of this to the defaultnetwork.opts file, so this problem should no longer be an issue.
If you do use linuxconf (or netconf) to configure yournetwork interface, leave the ``kernel
module'', ``I/O port'', and``irq'' parameters blank. Setting these parameters may interfere
withproper operation of the PCMCIA subsystem.
At boot time, when the Red Hat network subsystem starts up, it may say``Delaying eth0
initialization'' and ``[FAILED]''. This is actuallynot a failure: it means that this
network interface will not beinitialized until after the PCMCIA network device is configured.
Red Hat bundles their slightly modified PCMCIA source distributionwith their kernel sources,
rather than as a separate source package.When preparing to build a new set of PCMCIA
drivers, you willgenerally want to install Red Hat's kernel-source
RPM(kernel-source-*.i386.rpm), and not the kernel SRPM(kernel-*.src.rpm). The SRPM is
tailored for building theirkernel RPM files, which is not exactly what you want. With Red
Hat7.0, the kernel-source RPM also includes a mis-configured PCMCIAsource tree; if you
want to use it, delete their PCMCIA config.outfile and re-do "make config".
Slackware uses a BSD boot script arrangement. The PCMCIA startupscript is installed as
/etc/rc.d/rc.pcmcia, and boot optionsare specified in rc.pcmcia itself. The PCMCIA startup
scriptis invoked from /etc/rc.d/rc.S.
page 12 of 54 Howto.pdf
SuSE uses a System V init script arrangement, with init scripts storedunder /etc/init.d. The
PCMCIA startup script is installed as/etc/init.d/pcmcia, and startup options are kept
in/etc/rc.config. Before release 7.0, init scripts were keptunder /sbin/init.d. In early SuSE
releases (pre-5.3), thePCMCIA startup script was somewhat limited and did not allow
PCMCIAstartup variables to be overridden from the lilo boot prompt.
SuSE 8.0 includes both the standalone PCMCIA modules, and the 2.4kernel PCMCIA
subsystem modules.
A new variable, PCMCIA_SYSTEM,is available in
/etc/sysconfig/pcmcia to choose betweenthese. It can be set to either ``kernel'' or ``external''.
To look up current PCMCIA issues in SuSE's support database, go
Resolving installation and configuration problems
This section describes some of the most common failure modes for thePCMCIA subsystem.
Try to match your symptoms against the examples.This section only describes general
failures that are not specificto a particular client driver or type of card.
Before trying to diagnose a problem, you have to know where yoursystem log is kept (see
Notes about specific Linux distributions
In 3.1.15 and later releases, the debug-tools subdirectory of thePCMCIA source tree has a
few scripts to help diagnose some of the mostcommon configuration problems. The
test_setup script checks yourPCMCIA installation for completeness. The
test_network andtest_modem scripts will try to diagnose problems
with PCMCIAnetwork and modem cards. These scripts can be particularly helpful ifyou are
unfamiliar with Linux and are not sure how to approach aproblem.
Try to define your problem as narrowly as possible. If you haveseveral cards, try each card
in isolation, and in differentcombinations. Try cold Linux boots, versus warm boots from
Windows.Compare booting with cards inserted, versus inserting cards afterboot. If you
normally use your laptop docked, try it undocked. Andsometimes, two sockets will behave
For debugging problems in the device configuration scripts, it may beuseful to start cardmgr
with the ``-v'' option. With a3.1.23 or later PCMCIA package, this will cause most important
scriptactions to be recorded in the system log.
It is nearly impossible to debug driver problems encountered whenattempting to install Linux
via a PCMCIA device. Even if you canidentify the problem based on its symptoms,
installation disks aredifficult to modify, especially without access to a running Linuxsystem.
Customization of installation disks is completely dependenton the choice of Linux distribution,
and is beyond the scope of thisdocument. In general, the best course of action is to install
Linuxusing some other means, obtain the latest drivers, and then debug theproblem if it
Base PCMCIA kernel modules do not load
page 13 of 54 Howto.pdf
* Kernel version mismatch errors are reported when the PCMCIAstartup script runs.
* After startup, lsmod does not show any PCMCIA modules.
* cardmgr reports ``no pcmcia driver in/proc/devices'' in the system log.
Kernel modules contain version information that is checked against thecurrent kernel when a
module is loaded. The type of checking dependson the setting of the
CONFIG_MODVERSIONS kernel option. If thisis false, then the kernel version
number is compiled into each module,and insmod checks this for a match with the running
kernel. IfCONFIG_MODVERSIONS is true, then each symbol exported by
thekernel is given a sort of checksum. These codes are all comparedagainst the
corresponding codes compiled into a module. The intentwas for this to make modules less
version-dependent, because thechecksums would only change if a kernel interface changed,
and wouldgenerally stay the same across minor kernel updates. In practice, thechecksums
have turned out to be even more restrictive, because manykernel interfaces depend on
compile-time kernel option settings.Also, the checksums turned out to be an excessively
pessimistic judgeof compatibility.
The practical upshot of this is that kernel modules are closely tiedto both the kernel version,
and the setting of many kernelconfiguration options. Generally, a set of modules compiled
for one2.2.19 kernel will not load against some other 2.2.19 kernel unlessspecial care is
taken to ensure that the two were built with similarconfigurations. This makes distribution of
precompiled kernel modulesa tricky business.
You have several options:
* If you obtained precompiled drivers as part of a Linuxdistribution, verify that you are using
an unmodified kernel assupplied with that distribution.
If you intend to use
precompiledmodules, you generally must stick with the corresponding kernel.
* If you have reconfigured or upgraded your kernel, you willprobably need to compile and
install the PCMCIA package from scratch.This is easily done if you already have the
kernel source treeinstalled. See Compilation and installation
* In some cases, incompatibilities in other system components can prevent correct loading
of kernel modules. If you have upgraded yourown kernel, pay attention to the ``minimal
requirements'' for moduleutilities and binutils listed in the Documentation/Changesfile in
the kernel source code tree.
Some client driver modules do not load
* The base modules (pcmcia_core, ds, i82365) load correctly.
* Inserting a card gives a high beep + low beep pattern.
* cardmgr reports version mismatch errors in the system log.
page 14 of 54 Howto.pdf
Some of the driver modules require kernel services that may or may notbe present,
depending on kernel configuration. For instance, the SCSIcard drivers require that the kernel
be configured with SCSI support,and the network drivers require a networking kernel. If a
kernellacks a necessary feature, insmod may report undefined symbolsand refuse to load a
particular module. Note that insmod errormessages do not distinguish between version
mismatch errors andmissing symbol errors.
* The serial client driver serial_cs requires the kernelserial driver to be
enabled with CONFIG_SERIAL. This driver maybe built as a module.
* Support for multiport serial cards or multifunction cards thatinclude serial or modem
devices requires CONFIG_SERIAL_SHARE_IRQto
be enabled.
* The SCSI client drivers require that CONFIG_SCSI beenabled, along with
the appropriate top level driver
CONFIG_BLK_DEV_SR, etc for 2.2 and
laterkernels). These may be built as modules.
* The network client drivers require that CONFIG_INET isenabled. Kernel
networking support cannot be compiled as a module.
* The token-ring client requires that the kernel be compiled withCONFIG_TR
There are two ways to proceed:
* Rebuild your kernel with the necessary features enabled.
* If the features have been compiled as modules, then modify/etc/pcmcia/config to preload
these modules.
The /etc/pcmcia/config file can specify that additionalmodules need to be loaded for a
particular client. For example, forthe serial driver, one would use:
device "serial_cs"
class "serial" module "misc/serial", "serial_cs"
Module paths are specified relative to the top-level module directoryfor the current kernel
version; if no relative path is given, then thepath defaults to the pcmcia subdirectory.
ISA interrupt scan failures
* The system locks up when the PCMCIA drivers are loaded, evenwith no cards present.
* The system log shows a successful host controller probe justbefore the lock-up, but does
not show interrupt probe results.
page 15 of 54 Howto.pdf
After identifying the host controller type, the socket driver probesfor free ISA bus interrupts.
The probe involves programming thecontroller for each apparently free interrupt, then
generating a``soft'' interrupt, to see if the interrupt can be detected correctly.In some cases,
probing a particular interrupt can interfere withanother system device.
The reason for the probe is to identify interrupts which appear to befree (i.e., are not reserved
by any other Linux device driver), yetare either not physically wired to the host controller, or
areconnected to another device that does not have a driver.
In the system log, a successful probe might look like:
Intel PCIC probe:
TI 1130 CardBus at mem 0x10211000, 2 sockets
ISA irqs (scanned) = 5,7,9,10 status change on irq 10
There are two ways to proceed:
* The ISA interrupt probe can be restricted to a list ofinterrupts using the
irq_list parameter for the socket drivers.For example,
``irq_list=5,9,10'' would limit the scan to threeinterrupts. All 16-bit PCMCIA
devices will be restricted to usingthese interrupts (assuming they pass the probe). You
may need to usetrial and error to find out which interrupts can be safely probed.
* The interrupt probe can be disabled entirely by loading thesocket driver with the
``do_scan=0'' option. In this case, a defaultinterrupt list will be used, which
just avoids interrupts alreadyallocated for other devices.
In either case, the probe options can be specified using thePCIC_OPTS
definition in the PCMCIA startup script, for example:
It should be noted that /proc/interrupts is completelyuseless when it comes to diagnosing
interrupt probe problems. Theprobe is sensible enough to never attempt to use an interrupt
that isalready in use by another Linux driver. So, the PCMCIA drivers arealready using all
the information in /proc/interrupts.Depending on system design, an inactive device can still
occupy aninterrupt and cause trouble if it is probed for PCMCIA.
IO port scan failures
* The system locks up when cardmgr is first started. For3.1.24, the lockup happens even
with no cards present; for 3.1.25, acard must be inserted.
* The system log shows a successful host controller probe,including interrupt probe results,
but does not show IO proberesults.
* In some cases, the IO probe will succeed, but report largenumbers of random exclusions.
page 16 of 54 Howto.pdf
When cardmgr processes IO port ranges listed in/etc/pcmcia/config.opts, the kernel probes
these ranges todetect latent devices that occupy IO space but are not associatedwith a Linux
driver. The probe is read-only, but in rare cases,reading from a device may interfere with an
important system function,resulting in a lock-up.
Your system user's guide may include a map of system devices, showingtheir IO and
memory ranges. These can be explicitly excluded inconfig.opts.
Alternatively, if the probe is unreliable on your system, it can bedisabled by setting
CORE_OPTS to ``probe_io=0''. In thiscase, you should be very
careful to specify only genuinely availableranges of ports in config.opts, instead of using the
Memory probe failures
* The core drivers load correctly when no cards are present, withno errors in the system
* The system freezes and/or reboots as soon as any card isinserted, before any beeps are
Or alternately:
* All card insertions generate a high beep followed by a low beep.
* All cards are identified as ``anonymous memory cards''.
* The system log reports that various memory ranges have beenexcluded.
The core modules perform a memory scan at the time of first 16-bitcard insertion. This scan
can potentially interfere with other memorymapped devices. Also, pre-3.0.0 driver packages
perform a moreaggressive scan than more recent drivers. The memory window isdefined in
/etc/pcmcia/config.opts. The default window islarge, so it may help to restrict the scan to a
narrower range.Reasonable ranges to try include 0xd0000-0xdffff,
0xc0000-0xcffff,0xc8000-0xcffff, or 0xd8000-0xdffff.
If you have DOS or Windows PCMCIA drivers, you may be able to deducewhat memory
region those drivers use. Note that DOS memory addressesare often specified in ``segment''
form, which leaves off the finalhex digit (so an absolute address of 0xd0000 might be given
as0xd000). Be sure to add the extra digit back when making changes toconfig.opts.
Changing BIOS settings affecting how devices are mapped can sometimesbe useful. Try
changing settings for BIOS shadowing, or "Plug andPlay OS support".
In unusual cases, a memory probe failure can indicate a timingregister setup problem with
the host controller. See the Startup options
* cs: warning: no high memory space available!
CardBus bridges can allocate memory windows outside of the 640KB-1MB``memory hole'' in
the ISA bus architecture. It is generally a goodidea to configure CardBus bridges to use high
memory windows, becausethese are unlikely to conflict with other devices.
CardBuscards may require large memory windows, which may be difficult orimpossible to fit
into low memory. Card Services will preferentiallyallocate windows in high memory for
CardBus bridges, if both low andhigh memory windows are defined in config.opts.
page 17 ofThe
defaultconfig.opts includes several candidate high memory windows, oneof which will work in
most cases. Howto.pdf
Failure to detect card insertions and removals
* Cards are detected and configured properly if present at boottime.
* The drivers do not respond to insertion and removal events,either by recording events in
the system log, or by beeping.
In most cases, the socket driver (i82365 or tcic) willautomatically probe and select an
appropriate interrupt to signal cardstatus changes. The automatic interrupt probe doesn't
work on someIntel-compatible controllers, including Cirrus chips and the chipsused in some
IBM ThinkPads. If a device is inactive at probe time,its interrupt may also appear to be
available. In these cases, thesocket driver may pick an interrupt that is used by another
With the i82365 and tcic drivers, the irq_list optioncan be used to limit the
interrupts that will be tested. This listlimits the set of interrupts that can be used by PCMCIA
cards as wellas for monitoring card status changes. The cs_irq option canalso
be used to explicitly set the interrupt to be used for monitoringcard status changes.
If you can't find an interrupt number that works, there is also apolled status mode: both
i82365 and tcic will accept apoll_interval=100 option, to poll for card status
changes onceper second. This option should also be used if your system has ashortage of
interrupts available for use by PCMCIA cards. Especiallyfor systems with more than one
host controller, there is littlepoint in dedicating interrupts for monitoring card status changes.
All these options should be set in the PCIC_OPTS= line in
either/etc/rc.d/rc.pcmcia or /etc/sysconfig/pcmcia,depending on your site setup.
Interrupt delivery problems
* Cards appear to be configured successfully, but don't work.
* Serial and modem cards may respond very sluggishly.
* Network cards may report ``interrupt(s) dropped'', and/ortransmit timeouts.
The most simple interrupt delivery problems are due to conflicts withother system devices.
These can generally be resolved by excludingproblem interrupts in /etc/pcmcia/config.opts.
To test, justexclude interrupts one by one until either the problem is fixed or yourun out of
interrupts. If no interrupts work, then device conflictsare probably not the problem.
For CardBus bridges, a variety of other interrupt delivery issues maycome into play. For a
complete discussion, see PCI interrupt delivery problems
System resource starvation
page 18 of 54 Howto.pdf
* When a card is inserted, it is identified correctly but cannotbe configured (high/low beep
* One of the following messages will appear in the system log:
RequestIO: Resource in use
RequestIRQ: Resource in use
RequestWindow: Resource in use
GetNextTuple: No more items
could not allocate nn IO ports for CardBus socket n
could not allocate nnK memory for CardBus socket n
could not allocate interrupt for CardBus socket n
Interrupt starvation often indicates a problem with the interruptprobe (see ISA interrupt scan
If the interrupt probe is not working properly, the socket driver mayallocate an interrupt for
monitoring card insertions, even wheninterrupts are too scarce for this to be a good idea.
You can switchthe controller to polled mode by setting PCIC_OPTS
to``poll_interval=100'. Or, if you have a CardBus controller andan older version
of the PCMCIA drivers, try ``pci_csc=1'', whichselects a PCI interrupt (if
available) for card status changes.
In some cases, kernel misconfiguration can also produce an apparentinterrupt shortage. On
2.4 and later kernels, if CONFIG_ISAis not enabled, then the PCMCIA drivers
will assume no ISA businterrupts are available.
IO port starvation is fairly uncommon, but sometimes happens withcards that require large,
contiguous, aligned regions of IO portspace, or that only recognize a few specific IO port
positions. Thedefault IO port ranges in /etc/pcmcia/config.opts arenormally sufficient, but
may be extended. If this is the problem,try uncommenting the ``include port 0x1000-0x17ff''
line inconfig.opts. In rare cases, starvation may indicate that the IOport probe failed (see IO
port scan failures
Memory starvation is also uncommon with the default memory windowsettings in config.opts.
CardBus cards may require larger memoryregions than typical 16-bit cards. Since CardBus
memory windows canbe mapped anywhere in the host's PCI address space (rather than
justin the 640K-1MB ``hole'' in PC systems), it is helpful to specifylarge memory windows in
high memory, such as 0xa0000000-0xa0ffffff.
Resource conflict only with two cards inserted
* Two cards each work fine when used separately.
* When both cards are inserted, only one works.
This usually indicates a resource conflict with a system device thatLinux does not know
about. PCMCIA devices are dynamically configured,so, for example, interrupts are allocated
as needed, rather thanspecifically assigned to particular cards or sockets. Given a list
ofresources that appear to be available, cards are assigned resources inthe order they are
configured. In this case, the card configured lastis being assigned a resource that in fact is
not free.
page 19 of 54 Howto.pdf
Check the system log to see what resources are used by the non-workingcard. Exclude
these in /etc/pcmcia/config.opts, and restartthe cardmgr daemon to reload the resource
Device configuration does not complete
* When a card is inserted, exactly one high beep is heard.
* Subsequent card insertions and removals may be ignored.
This indicates that the card was identified successfully, however,cardmgr has been unable to
complete the configuration process forsome reason. The most likely reason is that a step in
the card setupscript has blocked. A good example would be the network scriptblocking if a
network card is inserted with no actual network hookuppresent.
To pinpoint the problem, you can manually run a setup script to seewhere it is blocking. The
scripts are in the /etc/pcmciadirectory. They take two parameters: a device name, and an
action.The cardmgr daemon records the configuration commands in thesystem log. For
example, if the system log shows that the command``./network start eth0'' was the last
command executed bycardmgr, the following command would trace the script:
sh -x /etc/pcmcia/network start eth0
Usage and features
Tools for configuring and monitoring PCMCIA devices
If the modules are all loaded correctly, the output of the lsmodcommand should look like the
following, when no cards are inserted:
Used by
3 [ds i82365]
The system log should also include output from the socket driverdescribing the host
controller(s) found and the number of socketsdetected.
The cardmgr configuration daemon
The >cdx>cardmgr>/cdx> daemon is responsible for monitoringPCMCIA sockets,
loading client drivers when needed, and running user-level scripts inresponse to card
insertions and removals. It records its actions inthe system log, but also uses beeps to signal
card status changes.The tones of the beeps indicate success or failure of
particularconfiguration steps. Two high beeps indicate that a card wasidentified and
configured successfully. A high beep followed by a lowbeep indicates that a card was
identified, but could not be configuredfor some reason. One low beep indicates that a card
could not beidentified.
page 20 of 54 Howto.pdf
The cardmgr daemon configures cards based on a database of knowncard types kept in
>cdx>/etc/pcmcia/config>/cdx>. This filedescribes the various client drivers, then
describes how to identifyvarious cards, and which driver(s) belong with which cards.
Theformat of this file is described in the pcmcia(5) man page.
The socket status file, stab
Cardmgr records device information for each socket
in>cdx>/var/lib/pcmcia/stab>/cdx>. Here is a samplestab listing:
Socket 0: Adaptec APA-1460 SlimSCSI
Socket 1: Serial or Modem Card
For the lines describing devices, the first field is the socket, thesecond is the device class,
the third is the driver name, the fourthis used to number multiple devices associated with the
same driver,the fifth is the device name, and the final two fields are the majorand minor
device numbers for this device (if applicable). See thestab man page for more info.
In 2.4 and later kernels, hot plut PCI drivers for CardBus cards arenot managed by cardmgr;
they are managed by the hotplugsubsystem. See
forinformation about this facility. When cardmgr sees a card that isowned by a hot plug PCI
driver, it will ignore that card. There willbe one beep when these cards are inserted or
ejected, but they will beidentified only as a ``CardBus hotplug device'' in the system log
andstab file.
The cardctl and cardinfo utilities
The >cdx>cardctl>/cdx> command can be used to check the status of asocket, or to
see how it is configured. It can also be used to alterthe configuration status of a card. Here
is an example of theoutput of the ``cardctl config'' command:
Socket 0:
not configured
Socket 1:
Vcc = 5.0, Vpp1 = 0.0, Vpp2 = 0.0
Card type is memory and I/O
IRQ 3 is dynamic shared, level mode, enabled
Speaker output is enabled
Function 0:
Config register base = 0x0800
Option = 0x63, status = 0x08
I/O window 1: 0x0280 to 0x02bf, auto sized
I/O window 2: 0x02f8 to 0x02ff, 8 bit
Or ``cardctl ident'', to get card identification information:
Socket 0:
no product info available
Socket 1:
product info: "LINKSYS", "PCMLM336", "A", "0040052D6400"
manfid: 0x0143, 0xc0ab
function: 0 (multifunction)
page 21 of 54 Howto.pdf
The ``cardctl suspend'' and ``cardctl resume'' commands canbe used to shut down a card
without unloading its associated drivers.The ``cardctl reset'' command attempts to reset and
reconfigure acard. ``cardctl insert'' and ``cardctl eject'' mimic theactions performed when a
card is physically inserted or ejected,including loading or unloading drivers, and configuring
or shuttingdown devices.
If you are running X, the >cdx>cardinfo>/cdx> utility producesa graphical display
showing the current status of all PCMCIA sockets,similar in content to ``cardctl config''. It
also provides agraphical interface to most other cardctl functions.
Inserting and ejecting cards
In theory, you can insert and remove PCMCIA cards at any time.However, it is a good idea
not to eject a card that is currently beingused by an application program. Kernels older than
1.1.77 would oftenlock up when serial/modem cards were ejected, but this should be
Some card types cannot be safely hot ejected. Specifically, ATA/IDEand SCSI interface
cards are not hot-swap-safe. This is unlikely tobe fixed, because a complete solution would
require significantchanges to the Linux block device model. Also, it is generally notsafe to
hot eject CardBus cards of any type. This is likely toimprove gradually as hot swap bugs in
the CardBus drivers are foundand fixed. For these card types (IDE, SCSI, CardBus), it
isrecommended that you always use ``cardctl eject'' beforeejecting.
Card Services and Advanced Power Management
Card Services can be compiled with support for APM(Advanced Power Management) if
you've configured yourkernel with APM support. The APM kernel driver is maintained
byStephen Rothwell ( The apmddaemon is
maintained by Avery Pennarun (, withmore information available
at The PCMCIAmodules will automatically be
configured for APM if a compatibleversion is detected on your system.
Whether or not APM is configured, you can use ``cardctl suspend''before suspending your
laptop, and ``cardctl resume'' afterresuming, to cleanly shut down and restart your PCMCIA
cards. Thiswill not work with a modem that is in use, because the serial driverisn't able to
save and restore the modem operating parameters.
APM seems to be unstable on some systems. If you experience troublewith APM and
PCMCIA on your system, try to narrow down the problem toone package or the other before
reporting a bug.
Some drivers, notably the PCMCIA SCSI drivers, cannot recover from asuspend/resume
cycle. When using a PCMCIA SCSI card, always use``cardctl eject'' prior to suspending the
Shutting down the PCMCIA system
To unload the entire PCMCIA package, invoke rc.pcmcia with:
/etc/rc.d/rc.pcmcia stop
page 22 of 54 Howto.pdf
This script will take several seconds to run, to give all clientdrivers time to shut down
gracefully. If a device is currently inuse, the shutdown will be incomplete, and some kernel
modules may notbe unloaded. To avoid this, use ``cardctl eject'' to shut downall sockets
before invoking rc.pcmcia. The exit status of thecardctl command will indicate if any sockets
could not be shutdown.
Overview of the PCMCIA configuration scripts
The following information applies to cards that are managed bycardmgr. In 2.4 and later
kernels, if the kernel PCMCIAsubsystem is active, then CardBus cards are managed by
thehotplug subsystem and the PCMCIA scripts are not used.
Each PCMCIA device has an associated ``class'' that describes how itshould be configured
and managed. Classes are associated with devicedrivers in /etc/pcmcia/config. There are
currently five IOdevice classes (network, SCSI, cdrom, fixed disk, and serial) andtwo memory
device classes (memory and FTL). For each class,there are twoscripts in
>cdx>/etc/pcmcia>/cdx>: a main configuration script(i.e., /etc/pcmcia/scsi for SCSI
devices), and an optionsscript (i.e., /etc/pcmcia/scsi.opts). The main script for adevice will be
invoked to configure that device when a card isinserted, and to shut down the device when
the card is removed. Forcards with several associated devices, the script will be invoked
foreach device.
The config scripts start by extracting some information about a devicefrom the stab file. Each
script constructs a ``device address'',that uniquely describes the device it has been asked to
configure, inthe ADDRESS shell variable. This is passed to the *.optsscript, which should
return information about how a device at thisaddress should be configured. For some
devices, the device address isjust the socket number. For others, it includes extra
informationthat may be useful in deciding how to configure the device. Forexample, network
devices pass their hardware ethernet address as partof the device address, so the
network.opts script could use thisto select from several different configurations.
The first part of all device addresses is the current PCMCIA``scheme''. This parameter is
used to support multiple sets of deviceconfigurations based on a single external
user-specified variable.One use of schemes would be to have a ``home'' scheme, and a
``work''scheme, which would include different sets of network configurationparameters. The
current scheme is selected using the ``cardctlscheme'' command. The default if no scheme
is set is ``default''.
There are a few additional shell variables that can be used in*.opts files in addition to
These correspond to individual fields from one line in the stabfile. See its man page for
These are equivalent to the output of ``cardctl info'' and givemore detailed card identification
As the *.opts files are just shell scripts, it is not requiredthat they follow the form of the
examples, which just return settingsbased on ADDRESS.
page 23 of 54 Howto.pdf
As a general rule, when configuring Linux for a laptop, PCMCIA devicesshould only be
configured from the PCMCIA device scripts. Do not tryto configure a PCMCIA device the
same way you would configure apermanently attached device. However, some Linux
distributionsprovide PCMCIA packages that are hooked into those distributions' owndevice
configuration tools. In that case, some of the followingsections may not apply; ideally, this
will be documented by thedistribution maintainers.
PCMCIA network adapters
Linux ethernet-type network interfaces normally have names likeeth0, eth1, and so on.
Token-ring adapters are handledsimilarly, however they are named tr0, tr1, and so on.The
ifconfig command is used toview or modify the state of a network interface. A peculiarity
ofLinux is that network interfaces do not have corresponding devicefiles under /dev, so do
not be surprised when you do not findthem.
When an ethernet card is detected, it will be assigned the first freeinterface name, which will
normally be eth0. Cardmgr willrun the /etc/pcmcia/network script to configure theinterface,
which normally reads network settings from/etc/pcmcia/network.opts.
The network
andnetwork.opts scripts will be executed only when your ethernetcard is actually present. If
your system has an automatic networkconfiguration facility, it may or may not be
PCMCIA-aware. Consultthe documentation of your Linux distribution and theNotes about
specific Linux distributions
The device address passed to network.opts consists of fourcomma-separated fields: the
scheme, the socket number, the deviceinstance, and the card's hardware ethernet address.
The deviceinstance is used to number devices for cards that have several network interfaces,
so itwill usually be 0. If you have several network cards used fordifferent purposes, one
option would be to configure the cards basedon socket position, as in:
case "$ADDRESS" in
# definitions for network card in socket 0
# definitions for network card in socket 1
Alternatively, they could be configured using their hardwareaddresses, as in:
case "$ADDRESS" in
# definitions for a D-Link card
# definitions for an IBM card
Network device parameters
page 24 of 54 Howto.pdf
The following parameters can be defined in network.opts:
Specifies the ethernet transceiver type, for certain 16-bit cards thatdo not autodetect. See
``man ifport'' and ``man mii-tool''for more information.
A boolean (y/n) value: indicates if the host's IP address and routinginformation should be
obtained using the BOOTP protocol, withbootpc or pump.
A boolean (y/n) value: indicates if the host's IP address and routinginformation should be
obtained from a DHCP server. The network scriptfirst looks for dhcpcd, then dhclient, then
Specifies a hostname to be passed to dhcpcd or pump, forinclusion in DHCP messages.
The IP address for this interface.
Basic network parameters: see the networking HOWTO for moreinformation.
The IP address of a gateway for this host's subnet. Packets withdestinations outside this
subnet will be routed to this gateway.
The local network domain name for this host, to be used in creating/etc/resolv.conf.
A search list for host name lookup, to be added to/etc/resolv.conf. DOMAIN and SEARCH
are mutuallyexclusive: see ``man resolver'' for more information.
DNS_1, DNS_2, DNS_3
Host names or IP addresses for nameservers for this interface, to beadded to /etc/resolv.conf
A space-separated list of NFS mount points to be mounted for this interface.
For IPX networks: the frame type and network number, passed to
theipx_interface command.
Boolean (y/n) settings for card eject policy. If NO_CHECK isset, then ``cardctl
eject'' will shut down a device even ifthere are open connections. If NO_FUSER
is set, then the scriptwill not check for busy NFS mounts or kill processes using those
For example:
case "$ADDRESS" in
page 25 of 54 Howto.pdf
To automatically mount and unmount NFS filesystems, first add allthese filesystems to
/etc/fstab, but include noautoin the mount options. In network.opts, list the filesystemmount
points in the MOUNTS variable. It is especiallyimportant to use either cardctl or cardinfo to
shut down anetwork card when NFS mounts are active. It is not possible tocleanly unmount
NFS filesystems if a network card is simply ejectedwithout warning.
In addition to the usual network configuration parameters, thenetwork.opts script can specify
extra actions to be taken afteran interface is configured, or before an interface is shut down.
Ifnetwork.opts defines a shell function called start_fn, itwill be invoked by the
network script after the interface isconfigured, and the interface name will be passed to the
function as itsfirst (and only) argument. Similarly, if it is defined, stop_fnwill be
invoked before shutting down an interface.
The transceiver type for some (mostly old) cards must be manually beselected using the
IF_PORT setting. This can either be a numericvalue, or a keyword identifying
the transceiver type. All the networkdrivers default to either autodetect the interface if
possible, or10baseT otherwise. The ifport command can be used to check orset the current
transceiver type. For example:
# ifport eth0 10base2
# ifport eth0
2 (10base2)
Most modern 10/100baseT cards use a ``media independent interface''(MII) transceiver that
automatically selects line speed and duplexsetting. The mii-tool command can be used to
monitor and controlthe behavior of the MII interface.
Comments about specific cards
* With IBM CCAE and Socket EA cards, the transceiver type (10base2,10baseT, AUI)
needs to be set when the network device is configured.Make sure that the transceiver
type reported in the system log matchesyour connection.
* The Farallon EtherWave is actually based on the 3Com 3c589, with aspecial transceiver.
Though the EtherWave uses 10baseT-styleconnections, its transceiver requires that the
3c589 be configured in10base2 mode.
* If you have trouble with an IBM CCAE, NE4100, Thomas Conrad, orKingston adapter, try
increasing the memory accesstime with the mem_speed=# option
to the pcnet_cs module.An example of how to do this is given in the standard
config.optsfile. Try speeds of up to 1000 (in nanoseconds).
* For the New Media Ethernet adapter, on some systems, it may benecessary to increase
the IO port access time with theio_speed=# option when the
pcmcia_core module is loaded.Edit CORE_OPTS in the startup
script to set this option.
* The multicast support in the New Media Ethernet driver is incomplete.The latest driver
will function with multicast kernels, but will ignoremulticast packets. Promiscuous mode
should work properly.
page 26 of 54 Howto.pdf
* The driver used by the IBM and 3Com token ring adaptersseems to behave very badly if
the cards are not connected to a ringwhen they get initialized. Always connect these
cards to the netbefore they are powered up. If ifconfig reports the hardwareaddress as
all 0's, this is likely to be due to a memory windowconfiguration problem.
* Some Linksys, D-Link, and IC-Card 10baseT/10base2 cards have a uniqueway of
selecting the transceiver type that isn't handled by the Linuxdrivers. One workaround is
to boot DOS and use the vendor-suppliedutility to select the transceiver, then warm boot
Linux.Alternatively, a Linux utility to perform this function is availableat
* 16-bit PCMCIA cards have a maximum performance of 1.5-2 MB/sec. Thatmeans that
any 16-bit 100baseT card (i.e., any card that uses thepcnet_cs,
3c574_cs, smc91c92_cs, or xirc2ps_csdriver) will
never achieve full 100baseT throughput. Only CardBusnetwork adapters can fully exploit
100baseT data rates.
* For WaveLAN wireless network adapters, Jean Tourrilhes( has put
together a wireless HOWTO at
Diagnosing problems with network adapters
* In 3.1.15 and later PCMCIA releases, the test_network script inthe
debug-tools subdirectory of the PCMCIA source tree will spotsome common problems.
* Is your card recognized as an ethernet card? Check the system log andmake sure that
cardmgr identifies the card correctly and startsup one of the network drivers. If it doesn't,
your card might stillbe usable if it is compatible with a supported card. This will bemost
easily done if the card claims to be ``NE2000 compatible''.
* Is the card configured properly? If you are using a supported card,and it was recognized
by cardmgr, but still doesn't work, theremight be an interrupt or port conflict with another
device. Find outwhat resources the card is using (from the system log),and try excluding
these in /etc/pcmcia/config.opts to forcethe card to use something different.
* If your card seems to be configured properly, but sometimes locks up,particularly under
high load, you may need to try changing your socketdriver timing parameters. See the
Startup options
* If you get ``Network is unreachable'' messages when you try toaccess the network, then
the routing information specified in/etc/pcmcia/network.opts is incorrect. This exact
message isan absolutely foolproof indication of a routing error. On the otherhand,
mis-configured cards will usually fail silently.
* If you are trying to use DHCP to configure your network interface, trytesting things with a
static IP address instead, to rule out a DHCPconfiguration problem.
* To diagnose problems in /etc/pcmcia/network.opts, start bytrying to ping other systems
on the same subnet using their IPaddresses. Then try to ping your gateway, and then
machines on othersubnets. Ping machines by name only after trying these simpler tests.
* Make sure your problem is really a PCMCIA one. It may help to see seeif the card works
under DOS with the vendor's drivers.
Double checkyour modifications to the
/etc/pcmcia/network.opts script.Make sure your drop cable, ``T'' jack, terminator, etc are
page 27 of 54 Howto.pdf
* Use real network cables. Don't even think about using that old phonecord you found in
your basement. And this means Category 5 cable for100baseT. It really matters.
PCMCIA serial and modem devices
Linux serial devices are accessed via the /dev/ttyS* and/dev/cua* special device files. In
pre-2.2 kernels, thettyS* devices were for incoming connections, such as directlyconnected
terminals, and the cua* devices were for outgoingconnections, such as modems. Use of
cua* devices is deprecatedin current kernels, and ttyS* can be used for all applications.The
configuration of a serial device can be examined and modified withthe setserial command.
When a serial or modem card is detected, it will be assigned to thefirst available serial device
slot. This will usually be/dev/ttyS1 (cua1) or /dev/ttyS2 (cua2),depending on the number of
built-in serial ports. The ttyS*device is the one reported in stab. The defaultserial device
option script, /etc/pcmcia/serial.opts, willlink the device file to /dev/modem as a convenience.
Forpre-2.2 kernels, the link is made to the cua* device.
Do not try to use /etc/rc.d/rc.serial to configure a PCMCIAmodem. This script should only be
used to configure non-removabledevices. Modify /etc/pcmcia/serial.opts if you want to
doanything special to set up your modem. Also, do not try to change theIO port and interrupt
settings of a serial device usingsetserial. This would tell the serial driver to look for thedevice
in a different place, but would not change how the card'shardware is actually configured.
The serial configuration scriptallows you to specify other setserial options, as well as
whethera line should be added to /etc/inittab for this port.
The device address passed to serial.opts has threecomma-separated fields: the first is the
scheme, the second is thesocket number, and the third is the device instance. The
deviceinstance may take several values for cards that support multipleserial ports, but for
single-port cards, it will always be 0. If youcommonly use more than one modem, you may
want to specify differentsettings based on socket position, as in:
case "$ADDRESS" in
# Options for modem in socket 0
# Options for modem in socket 1
If a PCMCIA modem is already configured when Linux boots, it may beincorrectly identified
as an ordinary built-in serial port. This isharmless, however, when the PCMCIA drivers take
control of the modem,it will be assigned a different device slot. It is best to eitherparse stab
or use /dev/modem, rather thanexpecting a PCMCIA modem to always have the same
device assignment.
If you configure your kernel to load the basic Linux serial portdriver as a module, you must
edit /etc/pcmcia/config toindicate that this module must be loaded. Edit the serial deviceentry
to read:
page 28 of 54 Howto.pdf
device "serial_cs"
class "serial" module "misc/serial", "serial_cs"
Serial device parameters
The following parameters can be defined in serial.opts:
Specifies a path for a symbolic link to be created to the ``callout''device (e.g., /dev/cua* for
pre-2.2, or /dev/ttyS*for 2.2 kernels).
Specifies options to be passed to the setserial command.
If specified, this will be used to construct an inittab entry forthe device.
Boolean (y/n) settings for card eject policy. If NO_CHECK istrue, then ``cardctl
eject'' will shut down a device even if itis busy. If NO_FUSER is true, then the
script will not try tokill processes using an ejected device.
For example:
case "$ADDRESS" in
Comments about specific cards
* The Uniden Data 2000 Wireless CDPD card has some special dialingstrings for initiating
SLIP and PPP mode. For SLIP, use ``ATDT2'';for PPP, "ATDT0".
* Socket IO revision H serial port cards have a faster-than-normalclock rate for the UART.
The card's actual baud rate is four timesfaster than the serial driver thinks it is. To work
around theproblem, specify SERIAL_OPTS="baud_base
460800" in/etc/pcmcia/serial.opts.
Diagnosing problems with serial devices
* In 3.1.15 and later PCMCIA releases, the test_modem script in
thedebug-tools subdirectory of the PCMCIA source tree will spot somecommon
* Is your card recognized as a modem? Check the system log andmake sure that cardmgr
identifies the card correctly and starts up theserial_cs driver. If it doesn't,
you may need to add a new entry toyour /etc/pcmcia/config file so that it will be identified
properly.See the Configuring unrecognized cards
page 29 of 54 Howto.pdf
* Is the modem configured successfully by serial_cs? Again, checkthe system
log and look for messages from the serial_cs driver.
Ifyou see
``register_serial() failed'', you may have an I/O port conflictwith another
device. Anothertip-off of a conflict is if the device is reported to be an 8250; mostmodern
modems should be identified as 16550A UART's. If youthink you're seeing a port
conflict, edit /etc/pcmcia/config.optsand exclude the port range that was allocated for the
* Is there an interrupt conflict? If the system log looks good,but the modem just doesn't
seem to work, try using setserial tochange the irq to 0, and see if the modem works. This
causes theserial driver to use a slower polled mode instead of using interrupts.If this
seems to fix the problem, it is likely that some other devicein your system is using the
interrupt selected by serial_cs. Youshould add a line to
/etc/pcmcia/config.opts to exclude thisinterrupt.
* If the modem seems to work only very, very slowly, this is an almostcertain indicator of an
interrupt conflict.
* Make sure your problem is really a PCMCIA one. It may help to seeif the card works
under DOS with the vendor's drivers. Also, don'ttest the card with something complex
like SLIP or PPP until you aresure you can make simple connections. If simple things
work but SLIPdoes not, your problem is most likely with SLIP, not with PCMCIA.
* If you get kernel messages indicating that the serial_cs module cannotbe
loaded, it means that your kernel does not have serial devicesupport. If you have
compiled the serial driver as a module, you mustmodify /etc/pcmcia/config to indicate that
theserial module should be loaded before serial_cs.
PCMCIA parallel port devices
The Linux parallel port driver is layered so that several high-leveldevice types can share use
of the same low level port driver. Printerdevices are accessed via the /dev/lp* special device
files.The configuration of a printer device can be examined and modified withthe tunelp
The parport_cs module depends on the parport andparport_pc
drivers, which may be either compiled into the kernelor compiled as modules. The layered
driver structure means that anytop-level parallel drivers (such as the plip driver, the
printerdriver, etc) must be compiled as modules. These drivers only recognize parallel port
devices at module startup time, so they needto be loaded after any PC Card parallel devices
are configured.
The device address passed to parport.opts has threecomma-separated fields: the first is the
scheme, the second is thesocket number, and the third is the device instance. The
deviceinstance may take several values for cards that support multipleparallel ports, but for
single-port cards, it will always be 0. Ifyou commonly use more than one such card, you may
want to specifydifferent settings based on socket position, as in:
case "$ADDRESS" in
# Options for card in socket 0
# Options for card in socket 1
page 30 of 54 Howto.pdf
Parallel device parameters
The following parameters can be defined in parport.opts:
Specifies a path for a symbolic link to be created to the printerport.
Specifies options to be passed to the tunelp command.
Boolean (y/n) settings for card eject policy. If NO_CHECK istrue, then ``cardctl
eject'' will shut down a device even if itis busy. If NO_FUSER is true, then the
script will not try tokill processes using an ejected device.
For example:
case "$ADDRESS" in
Diagnosing problems with parallel port devices
* Is there an interrupt conflict? If the system log looks good,but the port just doesn't seem
to work, try using tunelp to change the irq to 0, and see if things improve. This switches
thedriver to polling mode. If this seems to fix the problem, it islikely that some other
device in your system is using the interruptselected by parport_cs. You
should add a line to/etc/pcmcia/config.opts to exclude this interrupt.
* If you get kernel messages indicating that the parport_cs modulecannot be
loaded, it means that your kernel does not have paralleldevice support. If you have
compiled the parallel driver as a module,you may need to modify /etc/pcmcia/config to
indicate that theparport and parport_pc modules should be loaded
PCMCIA SCSI adapters
All the currently supported PCMCIA SCSI cards are work-alikes of oneof the following ISA
bus cards: the Qlogic, the Adaptec AHA-152X, orthe Future Domain TMC-16x0. The
PCMCIA drivers are built by linkingsome PCMCIA-specific code (in qlogic_cs.c,
aha152x_cs.c, orfdomain_cs.c) with the normal Linux SCSI driver,
pulled from theLinux kernel source tree. The Adaptec APA1480 CardBus driver is basedon
the kernel aic7xxx PCI driver. Due to limitations in the LinuxSCSI driver model, only one
removable card per driver is supported.
When a new SCSI host adapter is detected, the SCSI drivers will probefor devices. Check
the system log to make sure your devices aredetected properly. New SCSI devices will be
assigned to the firstavailable SCSI device files. The first SCSI disk will be/dev/sda, the first
SCSI tape will be /dev/st0, andthe first CD-ROM will be /dev/scd0.
page 31 of 54 Howto.pdf
A list of SCSI devices connected to this host adapter will be shown instab, and the SCSI
configuration script,/etc/pcmcia/scsi, will be called once for each attacheddevice, to either
configure or shut down that device. The defaultscript does not take any actions to configure
SCSI devices, but willproperly unmount filesystems on SCSI devices when a card is
The device addresses passed to scsi.opts are complicated, becauseof the variety of things
that can be attached to a SCSI adapter.Addresses consist of either six or seven
comma-separated fields: thecurrent scheme, thedevice type, the socket number, the SCSI
channel, ID, and logical unitnumber, and optionally, the partition number. The device type
will be``sd'' for disks, ``st'' for tapes, ``sr'' for CD-ROM devices, and``sg'' for generic SCSI
devices. For most setups, the SCSI channeland logical unit number will be 0. For disk
devices with severalpartitions, scsi.opts will first be called for the whole device,with a
five-field address. The script should set the PARTSvariable to a list of partitions. Then,
scsi.opts will be calledfor each partition, with the longer six-field addresses.
If your kernel does not have a top-level driver (disk, tape, etc) fora particular SCSI device,
then the device will not be configured bythe PCMCIA drivers. As a side effect, the device's
name instab will be something like ``sd#nnnn'' where ``nnnn''is a four-digit hex
number. This happens when cardmgr is unableto translate a SCSI device ID into a
corresponding Linux device name.
It is possible to modularize the top-level SCSI drivers so that theyare loaded on demand. To
do so, you need to edit/etc/pcmcia/config to tell cardmgr which extra modulesneed to be
loaded when your adapter is configured. For example:
device "aha152x_cs"
class "scsi" module "scsi/scsi_mod", "scsi/sd_mod", "aha152x_cs"
would say to load the core SCSI module and the top-level disk drivermodule before loading
the regular PCMCIA driver module.
Always turn on SCSI devices before powering up your laptop, or beforeinserting the adapter
card, so that the SCSI bus is properlyterminated when the adapter is configured. Also be
very careful aboutejecting a SCSI adapter. Be sure that all associated SCSI devices
areunmounted and closed before ejecting the card. The best way to ensurethis is to use
either cardctl or cardinfo to request cardremoval before physically ejecting the card. For
now, all SCSIdevices should be powered up before plugging in a SCSI adapter, andshould
stay connected until after you unplug the adapter and/or powerdown your laptop.
There is a potential complication when using these cards that does notarise with ordinary ISA
bus adapters. The SCSI bus carries a``termination power'' signal that is necessary for proper
operation ofordinary passive SCSI terminators.
PCMCIA SCSI adapters do not
supplytermination power, so if it is required, an external device mustsupply it. Some external
SCSI devices may be configured to supplytermination power. Others, such as the Zip Drive
and the SyquestEZ-Drive, use active terminators that do not depend on it. In somecases, it
may be necessary to use a special terminator block such asthe APS SCSI Sentry 2, which
has an external power supply. Whenconfiguring your SCSI device chain, be aware of
whether or not any ofyour devices require or can provide termination power.
page 32 of 54 Howto.pdf
SCSI device parameters
The following parameters can be defined in scsi.opts:
Specifies a path for a symbolic link to be created to this device.
A boolean (y/n) setting: specifies if an entry should be added to/etc/fstab for this device.
A boolean (y/n) setting: specifies if the filesystem should be checkedbefore being mounted,
with ``fsck -Ta''.
A boolean (y/n) setting: specifies if this device should beautomatically mounted at card
insertion time.
The filesystem type, mount options, and mount point to be used for thefstab entry and/or
mounting the device.
Boolean (y/n) settings for card eject policy. If NO_CHECK istrue, then ``cardctl
eject'' will shut down a device even if itis busy. If NO_FUSER is true, then the
script will not try tokill processes using an ejected device.
For example, here is a script for configuring a disk device at SCSI ID3, with two partitions,
and a CD-ROM at SCSI ID 6:
case "$ADDRESS" in
# This device has two partitions...
PARTS="1 2"
# Options for partition 1:
# update /etc/fstab, and mount an ext2 fs on /usr1
DO_FSTAB="y" ; DO_FSCK="y" ; DO_MOUNT="y"
# Options for partition 2:
# update /etc/fstab, and mount an MS-DOS fs on /usr2
DO_FSTAB="y" ; DO_FSCK="y" ; DO_MOUNT="y"
# Options for CD-ROM at SCSI ID 6
DO_FSTAB="y" ; DO_FSCK="n" ; DO_MOUNT="y"
page 33 of 54 Howto.pdf
Comments about specific cards
* The Adaptec APA-1480 CardBus card needs a large IO port window (256contiguous
ports aligned on a 256-port boundary). It may be necessaryto expand the IO port regions
in /etc/pcmcia/config.opts toguarantee that such a large window can be found.
* The Adaptec APA-460 SlimSCSI adapter is not supported. This card wasoriginally sold
under the Trantor name, and when Adaptec merged withTrantor, they continued to sell
the Trantor card with an Adapteclabel. The APA-460 is not compatible with any existing
Linux driver.
* I have had one report of a bad interaction between the New Media BusToaster and a
UMAX Astra 1200s scanner. Due to the complexity of theSCSI protocol, when
diagnosing problems with SCSI devices, it is worthconsidering that incompatible
combinations like this may exist and maynot be documented.
Diagnosing problems with SCSI adapters
* With the aha152x_cs driver (used by Adaptec, New Media, anda few others),
it seems that SCSI disconnect/reconnect support is afrequent source of trouble with tape
To disable this ``feature,''add the following to /etc/pcmcia/config.opts:
module "aha152x_cs" opts "reconnect=0"
* Also with the aha152x_cs driver, certain devices seem to requirea longer
startup delay, controlled via the reset_delay moduleparameter. The Yamaha
4416S CDR drive is one such device. The resultis the device is identified successfully,
then hangs the system. Insuch cases, try:
module "aha152x_cs" opts "reset_delay=500"
* Another potential source of SCSI device probe problems is probing ofmultiple LUN's. If
you see successful detection of a device, followedby SCSI bus timeouts when LUN 1 for
that device is probed, thendisable the kernel's
* If you have compiled SCSI support as modules (CONFIG_SCSI is``m''), you
may need to modify /etc/pcmcia/config to load theSCSI modules before the appropriate
*_cs driver is loaded.
* If you get ``aborting command due to timeout'' messages when the SCSIbus is probed,
you almost certainly have an interrupt conflict.
* If the host driver reports ``no SCSI devices found'', verify that yourkernel was compiled
with the appropriate top-level SCSI drivers foryour devices (i.e., disk, tape, CD-ROM,
and/or generic). If atop-level driver is missing, devices of that type will be ignored.
page 34 of 54 Howto.pdf
PCMCIA memory cards
The memory_cs driver handles all types of memory cards, as wellas providing
direct access to the PCMCIA memory address space forcards that have other functions.
When loaded, it creates acombination of character and block devices. See the man page for
themodule for a complete description of the device naming scheme. Blockdevices are used
for disk-like access (creating and mountingfilesystems, etc). The character devices are for
"raw" unbufferedreads and writes at arbitrary locations.
The device address passed to memory.opts consists of two fields:the scheme, and the
socket number. The options are applied to thefirst common memory partition on the
corresponding memory card.
Some flash memory cards, and most simple static RAM cards, lack a``Card Information
Structure'' (CIS), which is the system PCMCIA cardsuse to identify themselves. Normally,
cardmgr will assume thatany card that lacks a CIS is a simple memory card, and load
thememory_cs driver. Thus, a common side effect of a general
cardidentification problem is that other types of cards may be misdetectedas memory cards.
There is another issue to consider when handling memory cards that donot have CIS
information. At startup time, the PCMCIA package triesto use the first detected card to
determine what memory regions areusable for PCMCIA. The memory scan can be fooled if
that card is asimple memory card. If you plan to use memory cards often, it is bestto limit the
memory windows in /etc/pcmcia/config.opts toknown-good regions.
The memory_cs driver uses a heuristic to guess the capacity ofthese cards. The
heuristic does not work for write protected cards,and may make mistakes in some other
cases as well. If a card ismisdetected, its size should then be explicitly specified when
usingcommands such as dd or mkfs. The memory_cs module alsohas a
parameter for overriding the size detection. See the man page.
Memory device parameters
The following parameters can be specified in memory.opts:
A boolean (y/n) setting: specifies if an entry should be added to/etc/fstab for this device.
A boolean (y/n) setting: specifies if the filesystem should be checkedbefore being mounted,
with ``fsck -Ta''.
A boolean (y/n) setting: specifies if this device should beautomatically mounted at card
insertion time.
The filesystem type, mount options, and mount point to be used for thefstab entry and/or
mounting the device.
Boolean (y/n) settings for card eject policy. If NO_CHECK istrue, then ``cardctl
eject'' will shut down a device even if itis busy. If NO_FUSER is true, then the
script will not try tokill processes using an ejected device.
page 35 of 54 Howto.pdf
Here is an example of a script that will automatically mount memorycards based on which
socket they are inserted into:
case "$ADDRESS" in
# Mount filesystem, but don't update /etc/fstab
DO_FSTAB="n" ; DO_FSCK="y" ; DO_MOUNT="y"
FSTYPE="ext2" ; OPTS=""
# Mount filesystem, but don't update /etc/fstab
DO_FSTAB="n" ; DO_FSCK="y" ; DO_MOUNT="y"
FSTYPE="ext2" ; OPTS=""
Using linear flash memory cards
The following information applies only to so-called ``linear flash''memory cards. Many flash
cards, including all SmartMedia andCompactFlash cards, actually include circuitry to emulate
an IDE diskdevice. Those cards are thus handled as IDE devices, not memorycards.
There are two major formats for flash memory cards: the FTLor ``flash translation layer'' style,
and the MicrosoftFlash File System. The FTL format is generally moreflexible because it
allows any ordinary high-level filesystem (ext2,ms-dos, etc) to be used on a flash card as if it
were an ordinary diskdevice. The FFS is a completely different filesystem type. Linuxcannot
currently handle cards formated with FFS.
To use a flash memory card as an ordinary disk-like block device,first create an FTL partition
on the device with the>cdx>ftl_format>/cdx> command. This layer
hides thedevice-specific details of flash memory programming and make the cardlook like a
simple block device. For example:
ftl_format -i /dev/mem0c0c
Note that this command accesses the card through the ``raw'' memorycard interface. Once
formatted, the card can be accessed as anordinary block device via the ftl_cs
driver. For example:
mke2fs /dev/ftl0c0
mount -t ext2 /dev/ftl0c0 /mnt
Device naming for FTL devices is tricky. Minor device numbers havethree parts: the card
number, the region number on that card, andoptionally, the partition within that region. A
region can either betreated as a single block device with no partition table (like afloppy), or it
can be partitioned like a hard disk device. The``ftl0c0'' device is card 0, common memory
region 0, the entireregion. The ``ftl0c0p1'' through ``ftl0c0p4'' devices are primarypartitions 1
through 4 if the region has been partitioned.
page 36 of 54 Howto.pdf
Configuration options for FTL partitions can be given inftl.opts, which is similar in structure to
memory.opts.The device address passed to ftl.opts consists of three or fourfields: the
scheme, the socket number, the region number, andoptionally, the partition number. Most
flash cards have just oneflash memory region, so the region number will generally always
Intel Series 100 flash cards use the first 128K flash block to storethe cards' configuration
information. To prevent accidental erasureof this information, ftl_format will
automatically detect thisand skip the first block when creating an FTL partition.
PCMCIA ATA/IDE card drives
ATA/IDE drive support is based on the regular kernel IDE driver. Thisincludes SmartMedia
and CompactFlash devices: these flash memory cardsare set up so that they emulate an IDE
interface. The PCMCIA-specificpart of the driver is ide_cs. Be sure to use
cardctl orcardinfo to shut down an ATA/IDE card before ejecting it, as thedriver has not been
made ``hot-swap-proof''.
The device addresses passed to ide.opts consist of either threeor four fields: the current
scheme, the socket number, the drive'sserial number, and an optional partition number. The
ide_infocommand can be used to obtain an IDE device's serial number. As
withSCSI devices, ide.opts is first called for the entire device. Ifide.opts returns a list of
partitions in the PARTSvariable, the script will then be called for each partition.
ATA/IDE fixed-disk device parameters
The following parameters can be specified in ide.opts:
A boolean (y/n) setting: specifies if an entry should be added to/etc/fstab for this device.
A boolean (y/n) setting: specifies if the filesystem should be checkedbefore being mounted,
with ``fsck -Ta''.
A boolean (y/n) setting: specifies if this device should beautomatically mounted at card
insertion time.
The filesystem type, mount options, and mount point to be used for thefstab entry and/or
mounting the device.
Boolean (y/n) settings for card eject policy. If NO_CHECK istrue, then ``cardctl
eject'' will shut down a device even if itis busy. If NO_FUSER is true, then the
script will not try tokill processes using an ejected device.
Here is an example ide.opts file to mount the first partitionof any ATA/IDE card on /mnt.
case "$ADDRESS" in
DO_FSTAB="y" ; DO_FSCK="y" ; DO_MOUNT="y"
page 37 of 54 Howto.pdf
Diagnosing problems with ATA/IDE adapters
* An IO port conflict may cause the IDE driver to misdetect the drivegeometry and report
``INVALID GEOMETRY: 0 PHYSICAL HEADS?''. Tofix, try excluding the selected IO port
range in/etc/pcmcia/config.opts.
* Some IDE drives violate the PCMCIA specification by requiring moretime to spin up than
the maximum allowed card setup time. This mayresult in ``timed out during reset''
messages at card detect time.Adjust the unreset_delay and/or
unreset_limit parameters forthe pcmcia_core module to give a
drive more time to spin up; seethe pcmcia_core man page for parameter
details. For example:
* To use an ATA/IDE CD-ROM device, your kernel must be compiled
withCONFIG_BLK_DEV_IDECD enabled. This will
normally be the case forstandard kernels, however it is something to be aware of if
youcompile a custom kernel.
* A common error when using IDE drives is try to mount the wrong devicefile. Generally,
you want to mount a partition of the device, not theentire device (i.e., /dev/hde1, not
* The Linux IDE driver may have trouble configuring certainremovable-media drives if no
media is present at insertion time. TheIBM Portable DriveBay has this problem.
* Some kernels will report a pair of ``drive_cmd'' errors at insertiontime. These
errors can be ignored: they pop up when a removable IDEdevice does not accept the IDE
``door lock'' command.
Multifunction cards
A single interrupt can be shared by several drivers, such as theserial driver and an ethernet
driver: in fact, the PCMCIAspecification requires all card functions to share the same
interrupt.Normally, all card functions are available without having to swapdrivers. All Linux
kernels support this kind of interrupt sharing.
Simultaneous use of two card functions is ``tricky'' and varioushardware vendors have
implemented interrupt sharing in their ownincompatible (and sometimes proprietary) ways.
The drivers for somecards (Ositech Jack of Diamonds, 3Com 3c562 and related cards,
Linksyscards) properly support simultaneous access, but others (olderMegahertz cards in
particular) do not. If you have trouble using acard with both functions active, try using each
function in isolation.That may require explicitly doing an ``ifconfig down'' to shutdown a
network interface and use a modem on the same card.
Advanced topics
page 38 of 54 Howto.pdf
Resource allocation for PCMCIA devices
In theory, it should not really matter which interrupt is allocated towhich device, as long as
two devices are not configured to use thesame interrupt. In /etc/pcmcia/config.opts you'll
finda place for excluding interrupts that are used by non-PCMCIA devices.
Similarly, there is no way to directly specify the I/O addresses for acard to use. The
/etc/pcmcia/config.opts file allowsyou to specify ranges of ports available for use by any card,
or toexclude ranges that conflict with other devices.
After modifying /etc/pcmcia/config.opts, you can reinitializecardmgr with ``kill -HUP''.
The interrupt used to monitor card status changes is chosenby the low-level socket driver
module (i82365 or tcic)before cardmgr parses /etc/pcmcia/config, so it is notaffected by
changes to this file. To set this interrupt, use thecs_irq= option when the socket
driver is loaded, by settingthe PCIC_OPTS variable in /etc/rc.d/rc.pcmcia.
All the client card drivers have a parameter called irq_list forspecifying which
interrupts they may try to allocate.These driver options should be set inyour
/etc/pcmcia/config file. For example:
device "serial_cs"
module "serial_cs" opts "irq_list=8,12"
would specify that the serial driver should only use irq 8 or irq 12.Regardless of
irq_list settings, Card Services will neverallocate an interrupt that is already in
use by another device, or aninterrupt that is excluded in the config file.
PCI interrupt configuration problems and solutions
Most of the following discussion applies to 2.2 and earlier kernels.With 2.4 and later kernels,
the PCI subsystem has more completeresponsibility for PCI interrupt management. The
following tips mayhelp diagnose a problem, though some workarounds described here
maynot be available.
An overview of PCI interrupt routing issues
Each PCI slot has four PCI interrupt pins, INTA through INTD. Singlefunction devices will
only use the INTA pin; multifunction devices mayuse multiple INT pins. On the processor
side, on x86 single processorsystems, incoming hardware interrupts are directed to
interruptrequests (irq's) numbered 0..15. The PCI interrupt router, usuallypart of the
PCI-to-ISA host bridge, determines how incoming PCIinterrupts are mapped to CPU irq
numbers. Most modern bridge chipshave several PCI interrupt inputs, known as PIRQ1,
PIRQ2, etc, each ofwhich can be routed to any CPU irq number. So we might have
PCI slot 1 INTA --> router PIRQ1 --> CPU irq 9
PCI slot 1 INTB --> router PIRQ2 --> CPU irq 10
PCI slot 2 INTA --> router PIRQ2 --> CPU irq 10
PCI slot 2 INTB --> router PIRQ1 --> CPU irq 9
page 39 of 54 Howto.pdf
Multiple INT pins are often connected to the same PIRQ pin. Usually,the connections from
INT pins to PIRQ pins are arranged to spreadinstalled devices out as much as possible, to
give the OS the mostflexibility for choosing how interrupts are shared. The mapping
frombridge PIRQ pins to CPU irq numbers can be obtained by readingregisters in the
interrupt router. The mapping from INT pins to therouter's PIRQ pins, however, depends on
how the board designer decidedto connect things up, and cannot be directly determined by
For most PCI devices, the OS does not need to understand the interruptrouter details. Each
PCI device has a configuration register, the PCIInterrupt Line Register, that the BIOS
initializes with theappropriate CPU irq number for that device.
Unfortunately, the
BIOSgenerally will not configure PCI interrupts for CardBus bridge devices.
The PCI BIOS's Interrupt Routing Table is a data structure thatcontains information about the
mapping from PCI INT pins to the PIRQpins on the PCI interrupt router. The routing
information in thetable is stored in a somewhat unhelpful form, however. For eachdevice's
INT pins, the table specifies a ``link value''. Allinterrupts with the same link value are wired to
the same PIRQ pin;however, the meaning of the link values is defined by the chipsetvendor.
Several tools are available for examining PCI interrupt routinginformation:
lspci, /proc/pci
These will show you resource information (including interruptassignments, where they are
known) for all your PCI devices.
This is in the debug-tools directory of the PCMCIA sourcedistribution. It dumps the contents
of your PCI interrupt routingtable, if available. It also scans for known interrupt routers
anddumps their current interrupt steering settings.
Several PCMCIA module parameters affect PCI interrupt routing:
pcmcia_core module: cb_pci_irq=n
This option specifies one interrupt number to be used to program thePCI interrupt router for
all CardBus sockets that do not already havean interrupt assignment. It only has any effect
on systems that have aPCI irq routing table, and a known interrupt router.
i82365 module: irq_mode=n
Most CardBus bridges offer several methods for delivering interruptsto the host. The i82365
module by default assumes that a bridge candeliver both PCI and ISA interrupts, since this is
normal for laptops.A setting of ``irq_mode=0'' can be used to force a bridge to
use onlyPCI interrupts. See the man page for the i82365 module for adescription of what
other values mean for different bridge types.
i82365 module: irq_list=n,n,...
This parameter lists which ISA interrupt(s) can be used for PCMCIA.If no ISA interrupts are
available, specify ``irq_list=0''.
Notethat ``irq_mode=0'' implies
i82365 module: pci_irq_list=n,n,...
This option specifies a list of PCI interrupt numbers to use forCardBus sockets. It differs from
cb_pci_irq, because it does notactually program the PCI interrupt
router; it can be used when youknow the PCI interrupts are already set up a certain way,
even if youdo not know how the router works.
page 40 of 54 Howto.pdf
If you are having problems that you think may be related to PCIinterrupt configuration, you
should first verify that you have areasonably current PCMCIA driver package. Also carefully
look at thestartup messages when the PCMCIA kernel modules are loaded. Youshould see
something like:
Linux PCMCIA Card Services 3.1.18
kernel build: 2.2.14-5.0 #1 Tue May 9 10:44:24 PDT 2000
options: [pci] [cardbus] [apm] [pnp]
PCI routing table version 1.0 at 0xfdf30
Intel PCIC probe:
TI 1125 rev 02 PCI-to-CardBus at slot 00:07, mem 0x20000000
host opts [0]: [ring] [serial pci & irq] [pci irq 11] ...
host opts [0]: [ring] [serial pci & irq] [pci irq 11] ...
ISA irqs (scanned) = 3,4,7 PCI status changes
The ``PCI routing table'' message indicates that a valid routingtable was found. The ``host
opts'' lines indicate the interruptdelivery mode and whether or not a PCI interrupt could be
determinedfor each socket. And the final line indicates the results of the scanfor available
CardBus bridge is not detected by the PCI BIOS
* Intel PCIC probe: not found.
* The bridge does not show up in lspci or in /proc/pci.
The Lucent/SCM PCI-to-CardBus adapters seem to confuse the PCI BIOS onsome older
systems. Lucent says that this card is only supportedon systems that have a BIOS that
supports the PCI 2.2 specification,or are PC99 compliant. Some older systems will not
detect the Lucentcard at all, and if the system can't detect it, the Linux driverscannot use it.
The only possible resolutions are a BIOS upgrade, orusing a different motherboard or
CardBus adapter.
PCI interrupt delivery problems
* Cards seem to be configured correctly, but do not work.
* /proc/interrupts shows a count of 0 for interruptsassigned to PCMCIA drivers.
CardBus bridges usually support two types of interrupts, PCI and ISA.Partly for historical
reasons, it has become conventional to use PCIinterrupts for signaling card insertion and
removal events, and forCardBus card interrupts; and ISA interrupts for 16-bit cards.
Sinceversion 3.1.9, this is the scheme that the Linux PCMCIA system willuse by default.
Most CardBus bridges support multiple methods fordelivering interrupts to the host CPU.
Methods include ``parallel''interrupts, where each supported irq has a dedicated pin on
thebridge; various serial interrupt protocols, where one or two pins areused to communicate
with an interrupt controller; and hybrids, wherePCI interrupts might be signalled using
dedicated pins, while ISAinterrupts are delivered via a serial controller.
page 41 of 54 Howto.pdf
In general, it is the responsibility of the BIOS to program a bridgefor the appropriate interrupt
delivery method. However, there aresystems that do this incorrectly, and in some cases,
there is no wayfor software to safely detect the correct delivery method. Thei82365 module
reports the bridge mode at startup time, and has aparameter, irq_mode, that can
be used to reconfigure it. Not allbridges support this parameter, and the meaning of
irq_modedepends on the bridge type. See the i82365 man page for adescription
of what values are supported by your bridge. In somecases, a bridge may function correctly
in more than one interruptmode.
Most PCMCIA card readers that fit in a PCI bus slot only provide PCIinterrupt routing. The
Linux drivers assume that all bridges have ISAinterrupt capability, since that is generally
correct on laptops.With a card reader, it will generally be necessary to use
theirq_mode parameter to specify a ``PCI only'' interrupt deliverymode; the value
of the parameter depends on the bridge type, so checkthe i82365 man page. A few PCI card
readers require anirq_mode that permits ISA interrupts, but those interrupts
arenot actually connected; in that case, use ``irq_list=0''.
Check the system log and verify that the CardBus bridge has a PCIinterrupt assignment. If it
does not, then resolve that problemfirst, then return here if the symptoms persist. Next,
experimentwith different values for the irq_mode parameter.
No PCI interrupt assignment; valid routing table
* The Intel PCIC probe reports ``no pci irq'' for each socket.
* There is a routing table, and the router type is supported.
When a routing table is present, the pcmcia_core module will tryto automatically
configure the PCI interrupt router, but only does sowhen it has a safe and unambiguous
choice for what PCI interrupt touse. If there are several valid choices, then you must use
the``cb_pci_irq=...'' option to specify which interrupt to assign.Your
best bet is to pick the most lightly used interrupt that isalready assigned to another PCI
Moving the card to another slot sometimes offers a quick solution. Ifthat slot shares its
interrupt with an already-configured device, thenthe PCMCIA drivers will have no trouble
figuring out the assignment.
No PCI interrupt assignment; unknown interrupt router
* The Intel PCIC probe reports ``no pci irq'' for each socket.
* There is a routing table, but the router is an unknown type.
Adding support for a new interrupt router is tricky but not a bigjob. First determine, from a
datasheet, how your interrupt routersteers PCI interrupts. Then, see if you can guess the
meaning of thelink values from the output of dump_pirq.
Usually this
isreasonably obvious. Most routers have four PIRQ pins, and the linkvalues might be
something like 1,2,3,4, or 0x10,0x18,0x20,0x28, or0x60,0x61,0x62,0x63. The values are
usually chosen so that they canbe easily converted to the location of the appropriate
interruptsteering register. Finally, add small functions tomodules/pci_fixup.c
toof 54
page 42
get/set the interrupt steeringinformation for this router, using the other routers as examples. Howto.pdf
No PCI interrupt assignment; no routing table
* The Intel PCIC probe reports ``no pci irq'' for each socket.
* No interrupt routing table is found.
Without an interrupt routing table, we cannot tell how interrupts fromthe CardBus bridge are
directed to CPU irq numbers. All hope is notlost: you may be able to guess the PCI interrupt
assignment and usethe ``pci_irq_list=...'' option to pass this
information to thei82365 module. Good guesses might include the interrupt(s)assigned to
other PCI devices, the interrupt(s) used under Windows, orany other interrupts that are
unaccounted for.
You may also want to experiment with putting the adapter in differentPCI slots, for each
pci_irq_list you try. You are trying to finda slot that shares its
interrupt with an already-configured device,and might need to try several slots to find one.
How can I have separate device setups for home and work?
This is fairly easy using ``scheme'' support.Use two configuration schemes, called ``home''
and ``work''. Here isan example of a network.opts script with scheme-specificsettings:
case "$ADDRESS" in
# definitions for network card in work scheme
# definitions for network card in home scheme
The first part of a device address is always the configurationscheme. In this example, the
second ``case'' clause will select forboth the ``home'' and ``default'' schemes. So, if the
scheme is unsetfor any reason, it will default to the ``home'' setup.
Now, to select between the two sets of settings, run either:
cardctl scheme home
cardctl scheme work
The cardctl command does the equivalent of shutting down all yourcards and restarting them.
The command can be safely executed whetheror not the PCMCIA system is loaded, but the
command may fail if youare using other PCMCIA devices at the time (even if
theirconfigurations are not explicitly dependant on the scheme setting).
page 43 of 54 Howto.pdf
To find out the current scheme setting, run:
cardctl scheme
By default, the scheme setting is persistent across boots. This canhave undesirable effects if
networking is initialized for the wrongenvironment. Optionally, you can set the initial scheme
value withthe SCHEME startup option (see Startup options
To save even more keystrokes, schemes can be specified in lilo'sconfiguration file. For
instance, you could have:
root = /dev/hda1
image = /boot/vmlinuz
label = home
append = "SCHEME=home"
image = /boot/vmlinuz
label = work
append = "SCHEME=work"
Typing ``home'' or ``work'' at the boot prompt would then boot intothe appropriate scheme.
Booting from a PCMCIA device
Having the root filesystem on a PCMCIA device is tricky because theLinux PCMCIA system
is not designed to be linked into the kernel. Itscore components, the loadable kernel
modules and the user mode cardmgrdaemon, depend on an already running system. The
kernel's``initrd'' facility works around this requirement byallowing Linux to boot using a
temporary ram disk as a minimal rootimage, load drivers, and then re-mount a different root
filesystem.The temporary root can configure PCMCIA devices and then re-mount aPCMCIA
device as root.
The initrd image absolutely must reside on a bootable device: thisgenerally cannot be put on
a PCMCIA device. This is a BIOSlimitation, not a kernel limitation. It is useful here to
distinguishbetween ``boot-able'' devices (i.e., devices that can be booted), and``root-able''
devices (i.e., devices that can be mounted as root).``Boot-able'' devices are determined by
the BIOS, and are generallylimited to internal floppy and hard disk drives.
``Root-able''devices are any block devices that the kernel supports once it hasbeen loaded.
The initrd facility makes more devices ``root-able'',not ``boot-able''.
Some Linux distributions will allow installation to a device connectedto a PCMCIA SCSI
adapter, as an unintended side-effect of theirsupport for installs from PCMCIA SCSI
CD-ROM devices. However, atpresent, no Linux installation tools support configuring
anappropriate ``initrd'' to boot Linux with a PCMCIA root filesystem.Setting up a system with a
PCMCIA root thus requires that you useanother Linux system to create the ``initrd'' image. If
another Linuxsystem is not available, another option would be to temporarilyinstall a minimal
Linux setup on a non-PCMCIA drive, create an initrdimage, and then reinstall to the PCMCIA
The Linux Bootdisk-HOWTO has some general information about setting upboot disks but
nothing specific to initrd. The main initrd documentis included with recent kernel source code
distributions, inlinux/Documentation/initrd.txt.
Before beginning, you shouldread this
document. A familiarity with lilo is also helpful.Using initrd also requires that you have a
kernel compiled withCONFIG_BLK_DEV_RAM and page 44 of 54
CONFIG_BLK_DEV_INITRD enabled. Howto.pdf
This is an advanced configuration technique, and requires a high levelof familiarity with Linux
and the PCMCIA system. Be sure to read allthe relevant documentation before starting. The
following cookbookinstructions should work, but deviations from the examples willquickly put
you in uncharted and ``unsupported'' territory, and youwill be on your own.
This method absolutely requires that you use a PCMCIA driver releaseof 2.9.5 or later. Older
PCMCIA packages or individual componentswill not work in the initrd context. Do not mix
components fromdifferent releases.
The pcinitrd helper script
The pcinitrd script creates a basic initrd image for booting witha PCMCIA root partition. The
image includes a minimal directoryheirarchy, a handful of device files, a few binaries,
sharedlibraries, and a set of PCMCIA driver modules. When invokingpcinitrd, you specify the
driver modules that you want to beincluded in the image. The core PCMCIA components,
pcmcia_coreand ds, are automatically included.
As an example, say that your laptop uses an i82365-compatiblehost controller, and you want
to boot Linux with the root filesystemon a hard drive attached to an Adaptec SlimSCSI
adapter. You couldcreate an appropriate initrd image with:
pcinitrd -v initrd pcmcia/i82365.o pcmcia/aha152x_cs.o
To customize the initrd startup sequence, you could mount the imageusing the ``loopback''
device with a command like:
mount -o loop -t ext2 initrd /mnt
and then edit the linuxrc script. The configuration fileswill be installed under /etc in the
image, and can also becustomized. See the man page for pcinitrd for more information.
Creating an initrd boot floppy
After creating an image with pcinitrd, you can create a bootfloppy by copying the kernel, the
compressed initrd image, and a fewsupport files for lilo to a clean floppy. In the
followingexample, we assume that the desired PCMCIA root device is/dev/sda1:
mke2fs /dev/fd0
mount /dev/fd0 /mnt
mkdir /mnt/etc /mnt/boot /mnt/dev
cp -a /dev/fd0 /dev/sda1 /mnt/dev
cp [kernel-image] /mnt/vmlinuz
cp /boot/boot.b /mnt/boot/boot.b
gzip > [initrd-image] > /mnt/initrd
Create /mnt/etc/lilo.conf with the contents:
page 45 of 54 Howto.pdf
Finally, invoke lilo with:
lilo -r /mnt
When lilo is invoked with -r, it performs all actionsrelative to the specified alternate root
directory. The reason forcreating the device files under /mnt/dev was that lilowill not be able
to use the files in /dev when it is runningin this alternate-root mode.
Installing an initrd image on a non-Linux drive
One common use of the initrd facility would be on systems where theinternal hard drive is
dedicated to another operating system. TheLinux kernel and initrd image can be placed in a
non-Linux partition,and lilo or LOADLIN can be set up to boot Linux from theseimages.
Assuming that you have a kernel has been configured for theappropriate root device, and an
initrd image created on anothersystem, the easiest way to get started is to boot Linux
usingLOADLIN, as:
LOADLIN >kernel> initrd=>initrd-image>
Once you can boot Linux on your target machine, you could theninstall lilo to allow booting
Linux directly.For example, say that /dev/hda1 is the non-Linux targetpartition and /mnt can
be used as a mount point. First,create a subdirectory on the target for the Linux files:
mount /dev/hda1 /mnt
mkdir /mnt/linux
cp [kernel-image] /mnt/linux/vmlinuz
cp [initrd-image] /mnt/linux/initrd
In this example, say that /dev/sda1 is the desired Linux rootpartition, a SCSI hard drive
mounted via a PCMCIA SCSI adapter. Toinstall lilo, create a lilo.conf file with the contents:
The boot= line says to install the boot loader in the master bootrecord of the specified device.
The root= line identifies thedesired root filesystem to be used after loading the initrd image,
andmay be unnecessary if the kernel image is already configured this way.The other=
section is used to describe the other operating systeminstalled on /dev/hda1.
To install lilo in this case, use:
page 46 of 54 Howto.pdf
lilo -C lilo.conf
Note that in this case, the lilo.conf file uses absolute pathsthat include /mnt. I did this in the
example because the targetfilesystem may not support the creation of Linux device files for
theboot= and root= options.
Dealing with unsupported cards
Configuring unrecognized cards
Assuming that your card is supported by an existing driver, allthat needs to be done is to add
an entry to/etc/pcmcia/config to tell cardmgr how to identify the card,and which driver(s) need
to be linked up to this card. Check the manpage for pcmcia for more information about the
config file format.If you insert an unknown card, cardmgr will normally record
someidentification information in the system log that can beused to construct the config entry.
This information can also bedisplayed with the ``cardctl ident'' command.
Here is an example of how cardmgr will report an unsupported card inthe system log:
cardmgr[460]: unsupported card in socket 1
cardmgr[460]: product info: "MEGAHERTZ", "XJ2288", "V.34 PCMCIA MODEM"
cardmgr[460]: manfid: 0x0101, 0x1234 function: 2 (serial)
The corresponding entry in /etc/pcmcia/config would be:
card "Megahertz XJ2288 V.34 Fax Modem"
version "MEGAHERTZ", "XJ2288", "V.34 PCMCIA MODEM"
bind "serial_cs"
or using the more compact product ID codes:
card "Megahertz XJ2288 V.34 Fax Modem"
manfid 0x0101, 0x1234
bind "serial_cs"
You can use ``*'' to match strings that don't need to match exactly,like version numbers.
When making new config entries, be carefulto copy the strings exactly, preserving case and
blank spaces.Also be sure that the config entry has the same number of strings asare
reported in the log file.
Beware that you can specify just about any driver for a card, but ifyou're just shooting in the
dark, there is not much reason to expectthis to be productive. You may get lucky and find
that your card issupported by an existing driver. However, the most likely outcome isthat the
driver won't work, and may have unfortunate side effectslike locking up your system. Unlike
most ordinary device drivers,which probe for an appropriate card, the probe for a PCMCIA
device isdone by cardmgr, and the driver itself may not do much validationbefore attempting
to communicate with the device.
page 47 of 54 Howto.pdf
After editing /etc/pcmcia/config, you can signal cardmgrto reload the file with:
kill -HUP `cat /var/run/`
If you do set up an entry for a new card, please send me a copy sothat I can include it in the
standard config file.
Adding support for an NE2000-compatible ethernet card
Before you begin: this procedure will only work for simple 16-bitethernet cards. Multifunction
cards (i.e., ethernet/modem combocards) have an extra layer of complexity regarding how
the twofunctions are integrated, and generally cannot be supported withoutobtaining some
configuration information from the card vendor. Usingthe following procedure for a
multifunction card will not beproductive.
First, see if the card is already recognized by cardmgr. Somecards not listed in
SUPPORTED.CARDS are actually OEM versions ofcards that are supported. If you find a
card like this, let me knowso I can add it to the list.
If your card is not recognized, follow the instructions in theConfiguring unrecognized cards
If the pcnet_cs driver says that it is unable to determine yourcard's hardware
ethernet address, then edit your new config entry tobind the card to the memory card driver,
memory_cs.Restart cardmgr to use the new updated config file.You will need to
know your card's hardware ethernet address. Thisaddress is a series of six two-digit hex
numbers, often printed on thecard itself. If it is not printed on the card, you may be able to
usea DOS driver to display the address.
In any case, once you know it,run:
dd if=/dev/mem0a count=20 | od -Ax -t x1
and search the output for your address. Only the even bytes aredefined, so ignore the odd
bytes in the dump. Record the hex offset of thefirst byte of the address. Now, edit
clients/pcnet_cs.c andfind the hw_info structure. You'll need to
create a new entryfor your card. The first field is the memory offset. Thenext three fields are
the first three bytes of the hardware address.The final field contains some flags for specific
card features; tostart, try setting it to 0.
After editing pcnet_cs.c, compile and install the new module.Edit
/etc/pcmcia/config again, and change the card bindingfrom memory_cs to
pcnet_cs. Follow the instructions forreloading the config file, and you should be
all set. Please send mecopies of your new hw_info and config entries.
If you can't find your card's hardware address in the hex dump, as amethod of last resort, it is
possible to ``hard-wire'' the address whenthe pcnet_cs module is initialized.
Edit/etc/pcmcia/config.opts and add a hw_addr= option, likeso:
module "pcnet_cs" opts "hw_addr=0x00,0x80,0xc8,0x01,0x02,0x03"
Substitute your own card's hardware address in the appropriate spot,of course. Beware that
if you've gotten this far, it is very unlikelythat your card is genuinely NE2000 compatible. In
fact, I'm not sureif there are any cards that are not handled by one of the firsttwo methods.
page 48 of 54 Howto.pdf
PCMCIA floppy interface cards
The PCMCIA floppy interface used in the Compaq Aero and a few otherlaptops is not yet
supported by this package. The snag in supportingthe Aero floppy is that the Aero seems to
use a customizedPCMCIA controller to support DMA to the floppy. Withoutknowing exactly
how this is done, there isn't any way to implementsupport under Linux.
If the floppy adapter card is present when an Aero is booted, the AeroBIOS will configure the
card, and Linux will identify it as a normalfloppy drive. When the Linux PCMCIA drivers are
loaded, they willnotice that the card is already configured and attached to a Linuxdriver, and
this socket will be left alone. So, the drive can be usedif it is present at boot time, but the
card is not hot swappable.
Debugging tips and programming information
Submitting useful problem reports
The best way to submit reports is to use the online pcmcia-cs forumsor the bug tracker at
SourceForge. That way, other people can seecurrent problems (and fixes or workarounds, if
available). Here aresome things that should be included in all bug reports:
* Your system brand and model.
* All PCMCIA card(s) you are using.
* Your Linux kernel version (i.e., ``uname -rv''), and PCMCIAdriver version (i.e., ``cardctl
* Output of 'lspci -v'
* Any changes you have made to the startup files in/etc/pcmcia, or to the PCMCIA startup
* All PCMCIA-related messages in your system log file. Thatincludes startup messages,
and messages generated when yourcards are configured.
All the PCMCIA modules and the cardmgr daemon send statusmessages to the system log.
These will usually end up somewhere like/var/log/messages or /var/log/daemon.log.
Thesefiles should be the first place to look when tracking down a problem.When submitting a
bug report, always include the relevant contents ofthese files. If you are having trouble
finding your system messages,check /etc/syslog.conf to see how different classes
ofmessages are handled.
Before submitting a bug report, please check to make sure that you areusing an up-to-date
copy of the driver package. While it is somewhatgratifying to read bug reports for things I've
already fixed, it isn'ta particularly constructive use of my time.
If you do not have web access, bug reports can be sent to me However, I prefer thatbug reports be posted to the
pcmcia-cs SourceForge site, so that theycan be seen by others.
page 49 of 54 Howto.pdf
Interpreting kernel trap reports
If your problem involves a kernel fault, the register dump from thefault is only useful if you
can translate the fault address, EIP, tosomething meaningful. Recent versions of klogd
attempt totranslate fault addresses based on the current kernel symbol map, butthis may not
work if the fault is in a module, or if the problem issevere enough that klogd cannot finish
writing the faultinformation to the system log.
If a fault is in the main kernel, the fault address can be looked upin the file. This
may be installed in/ or /boot/ If a fault is in amodule, the nm
command gives the same information, however, thefault address has to be adjusted based
on the module's load address.Let's say that you have the following kernel fault:
Unable to handle kernel NULL pointer dereference
current->tss.cr3 = 014c9000, %cr3 = 014c9000
*pde = 00000000
Oops: 0002
EFLAGS: 00010282
The fault address is 0xc2026081. Looking at, wesee that this is past the end of
the kernel, i.e., is in a kernelmodule. To determine which module, check the output of``ksyms
-m | sort'':
Defined by
So, 0xc2026081 is in the 3c574_cs module, and is at an offset of0x0081 from
the start of the module. We cannot look up this offset in3c574_cs.o yet: when
the kernel loads a module, it inserts aheader at the module load address, so the real start of
the module isoffset from the address shown in ksyms. The size of the headervaries with
kernel version: to find out the size for your kernel,check a module that exports symbols (like
pcmcia_core above), andcompare a symbol address with nm output for that
symbol. In thisexample, register_ss_entry is loaded at an offset of
0xc200d10c -0xc200d000 = 0x010c, while ``nm pcmcia_core.o'' shows the
offsetas 0x00c0, so the header size is 0x010c - 0x00c0 = 0x004c bytes.
Back to 3c574_cs, our fault offset is 0x0081, and subtracting the0x004c header,
the real module offset is 0x0035. Now looking at``nm 3c574_cs.o |
sort'', we see:
page 50 of 54 Howto.pdf
So, the fault is located in tc574_attach().
In this example, the fault did not cause a total system lockup, soksyms could be executed
after the fault happened. In othercases, you may have to infer the module load addresses
indirectly.The same sequence of events will normally load modules in the sameorder and at
the same addresses. If a fault happens when a certaincard is inserted, get the ksyms output
before inserting the card,or with a different card inserted. You can also manually load
thecard's driver modules with insmod and run ksyms beforeinserting the card.
For more background, see ``man insmod'', ``man ksyms'', and``man klogd''. In the kernel
source tree,Documentation/oops-tracing.txt is also relevant. Here are afew more kernel
debugging hints:
* Depending on the fault, it may also be useful to translateaddresses in the ``Call Trace'',
using the same procedure as for themain fault address.
* To diagnose a silent lock-up, try to provoke the problem with Xdisabled, since kernel
messages sent to the text console will not bevisible under X.
* If you kill klogd, most kernel messages will be echoeddirectly on the text console, which
may be helpful if the problemprevents klogd from writing to the system log.
* To cause all kernel messages to be sent to the console, for 2.2and later kernels, if
/proc/sys/kernel/printk exists, do:
echo 8 > /proc/sys/kernel/printk
* The key combination >RightAlt>>ScrLk> prints aregister dump on the text
console. This may work even if the systemis otherwise completely unresponsive, and the
EIP address can beinterpreted as for a kernel fault.
* For 2.2 and later kernels configured
withCONFIG_MAGIC_SYSRQ enabled, various emergency
functions areavailable via special >Alt>>SysRq> key combinations,documented
in Documentation/sysrq.txt in the kernel sourcetree.
Low level PCMCIA debugging aids
The PCMCIA modules contain a lot of conditionally-compiled debuggingcode. Most of this
code is under control of the PCMCIA_DEBUGpreprocessor define. If this is
undefined, debugging code willnot be compiled. If set to 0, the code is compiled but
inactive.Larger numbers specify increasing levels of verbosity. Each modulebuilt with
PCMCIA_DEBUG defined will have an integer
parameter,pc_debug, that controls the verbosity of its output. Thiscan be
adjusted when the module is loaded, so output can be controlledon a per-module basis
without recompiling.
Your default configuration for syslogd may discard kerneldebugging messages. To ensure
that they are recorded, edit/etc/syslog.conf to ensure that ``kern.debug'' messagesare
recorded somewhere. See ``man syslog.conf'' for details.
page 51 of 54 Howto.pdf
There are a few register-level debugging tools in thedebug_tools/ subdirectory of
the PCMCIA distribution. Thedump_tcic and dump_i365 utilities
generate register dumpsfor ISA-to-PCMCIA controllers. In 3.1.15 and later
releases,dump_i365 is replaced by dump_exca, which is similar
butalso works for PCI-to-CardBus bridges. Also new in 3.1.15 for CardBusbridges is the
dump_cardbus tool, which interprets theCardBus-specific registers. These are
all most useful if you haveaccess to a datasheet for the corresponding controller chip.
Thedump_cis utility (dump_tuples in pre-3.0.2 distributions)lists the
contents of a card's CIS (Card Information Structure), anddecodes most of the important bits.
And the dump_cisreg utilitydisplays a card's local configuration registers.
The memory_cs memory card driver is also sometimes useful fordebugging
problems with 16-bit PC Cards. It can be bound to any card,and does not interfere with other
drivers. It can be used to directlyaccess any card's attribute memory or common memory.
Similarly forCardBus cards, the memory_cb driver can be bound to any
32-bitcard, to give direct access to that card's address spaces. See theman pages for more
On 2.2 and later kernels, the PCMCIA package will create a treeof status information under
>cdx>/proc/bus/pccard>/cdx>.Much of the information can only be interpreted using
the data sheetsfor the PCMCIA host controller. Its contents may depend on how thedrivers
were configured, but may include all or some of the following:
If present, these files contain resource allocation information tosupplement the normal kernel
resource tables. Recent versions ofthe PCMCIA system may obtain additional resource
information fromthe Plug and Play BIOS if configured to do so.
In recent releases, this lists all currently loaded PCMCIA clientdrivers. Unlike /proc/modules,
it also lists drivers thatmay be statically linked into the kernel.
For each socket, describes that socket's host controller and itscapabilities.
This contains a dump of a controller's ``ExCA'' Inteli82365sl-compatible register set.
For CardBus bridges, a dump of the bridge's PCI configuration space,and a dump of the
bridge's CardBus configuration registers.
Writing Card Services drivers for new cards
The Linux PCMCIA Programmer's Guide is the best documentation for theclient driver
interface. The latest version is always available in
/pub/pcmcia-cs/doc, or onthe web at
For devices that are close relatives of normal ISA devices, you willprobably be able to use
parts of existing Linux drivers. In somecases, the biggest stumbling block will be modifying
an existingdriver so that it can handle adding and removing devices after boottime. Of the
current drivers, the memory card driver is the only``self-contained'' driver that does not
page 52 of 54
depend on other parts of theLinux kernel to do most of the dirty work. Howto.pdf
In many cases, the largest barrier to supporting a new card type isobtaining technical
information from the manufacturer. It may bedifficult to figure out who to ask, or to explain
exactly whatinformation is needed. However, with a few exceptions, it is verydifficult if not
impossible to implement a driver for a card withouttechnical information from the
I have written a dummy driver with lots of comments that explains alot of how a driver
communicates with Card Services; you will findthis in the PCMCIA source distribution in
Guidelines for PCMCIA client driver authors
I have decided that it is not really feasible for me to distribute allPCMCIA client drivers as part
of the PCMCIA package. Each new drivermakes the main package incrementally harder to
maintain, and includinga driver inevitably transfers some of the maintenance work from
thedriver author to me. Instead, I will decide on a case by case basiswhether or not to
include contributed drivers, based on user demand aswell as maintainability. For drivers not
included in the corepackage, I suggest that driver authors adopt the following scheme
forpackaging their drivers for distribution.
Driver files should be arranged in the same directory scheme used inthe PCMCIA source
distribution, so that the driver can be unpacked ontop of a complete PCMCIA source tree. A
driver should include sourcefiles (in ./modules/), a man page (in ./man/), andconfiguration
files (in ./etc/). The top level directoryshould also include a README file.
The top-level directory should include a makefile, set up sothat ``make -f ... all'' and ``make -f
... install'' compile thedriver and install all appropriate files. If this makefile is givenan
extension of .mk, then it will automatically be invoked by thetop-level Makefile for the all and
install targets.Here is an example of how such a makefile could be constructed:
# Sample Makefile for contributed client driver
FILES = README.sample_cs \
modules/sample_cs.c modules/sample_cs.h \
etc/sample.conf etc/sample etc/sample.opts \
$(MAKE) -C modules MODULES=sample_cs.o
$(MAKE) -C modules install-modules MODULES=sample_cs.o
$(MAKE) -C etc install-clients CLIENTS=sample
$(MAKE) -C man install-man4 MAN4=sample_cs.4
tar czvf sample_cs.tar.gz $(FILES)
This makefile uses install targets defined in 2.9.10 and laterversions of the PCMCIA
package. This makefile also includes a``dist'' target for the convenience of the driver author.
You wouldprobably want to add a version number to the final package filename(for example,
sample_cs-1.5.tar.gz). A complete distributioncould look like:
page 53 of 54 Howto.pdf
With this arrangement, when the contributed driver is unpacked, itbecomes essentially part of
the PCMCIA source tree. It can make useof the PCMCIA header files, as well as the
machinery for checking theuser's system configuration, and automatic dependency checking,
justlike a ``normal'' client driver.
In this example, etc/sample and etc/sample.optswould be the new driver's configuration
scripts (if needed), andetc/sample.conf would contain any additions to the PCMCIAcard
configuration file. Starting with the 3.1.6 release,cardmgr will automatically process any
*.conf filesinstalled in /etc/pcmcia, so installation of contributeddrivers should no longer
require hand editing configuration files.
I will accept client drivers prepared according to this specificationand place them in the
/pub/pcmcia-cs/contrib directory The README in this directory
willdescribe how to unpack a contributed driver.
The client driver interface has not changed much over time, and hasalmost always preserved
backwards compatibility. A client driver willnot normally need to be updated for minor
revisions in the mainpackage. I will try to notify authors of contributed drivers ofchanges that
require updates to their drivers.
Guidelines for Linux distribution maintainers
If your distribution has system configuration tools that you wouldlike to be PCMCIA-aware,
please use the *.opts files in/etc/pcmcia for your ``hooks.'' These files will not bemodified if a
user compiles and installs a new release of the PCMCIApackage. If you modify the main
configuration scripts, then a freshinstall will silently overwrite your custom scripts and break
theconnection with your configuration tools. Contact me if you are notsure how to write an
appropriate option script, or if you needadditional capabilities.
It is helpful for users (and for me) if you can document how yourdistribution deviates from the
PCMCIA package as described in thisdocument. In particular, please document changes to
the startupscript and configuration scripts. If you send me the appropriateinformation, I will
include it in the Notes about specific Linux distributions
When building PCMCIA for distribution, consider including contributeddrivers that are not part
of the main PCMCIA package. For reasons ofmaintainability, I am trying to limit the core
package size, by onlyadding new drivers if I think they are of particularly broad interest.Other
drivers will be distributed separately, as described in theprevious section. The split between
integral and separate drivers issomewhat arbitrary and partly historical, and should not imply
adifference in quality.
rate this article:
current rating:
Your rating:
Support this site
page 54 of 54