Linux STREAMS (LiS)

Linux STREAMS (LiS) Installation and Reference Manual

About This Manual

This is Edition 6, last updated 2007-06-24, of The Linux STREAMS (LiS) Installation and Reference Manual, for Version 2.18 release 6 of the Linux STREAMS (LiS) package.

Preface

Notice

This package is released and distributed under the GPL (see GNU General Public License). Please note, however, that there are different licensing terms for the manual pages and some of the documentation (derived from OpenGroup¹ publications and other sources). Consult the permission notices contained in the documentation for more information.

This manual is released under the FDL (see GNU Free Documentation License) with all sections invariant.

Abstract

This manual provides a Installation and Reference Manual for Linux STREAMS (LiS).

Objective

The objective of this manual is to provide a guide for the STREAMS programmer when developing STREAMS modules, drivers and application programs for Linux STREAMS (LiS).

This guide provides information to developers on the use of the STREAMS mechanism at user and kernel levels.

STREAMS was incorporated in UNIX System V Release 3 to augment the character input/output (I/O) mechanism and to support development of communication services.

STREAMS provides developers with integral functions, a set of utility routines, and facilities that expedite software design and implementation.

Intent

The intent of this manual is to act as an introductory guide to the STREAMS programmer. It is intended to be read alone and is not intended to replace or supplement the Linux STREAMS (LiS) manual pages. For a reference for writing code, the manual pages (see STREAMS(9)) provide a better reference to the programmer. Although this describes the features of the Linux STREAMS (LiS) package, OpenSS7 Corporation is under no obligation to provide any software, system or feature listed herein.

Audience

This manual is intended for a highly technical audience. The reader should already be familiar with Linux kernel programming, the Linux file system, character devices, driver input and output, interrupts, software interrupt handling, scheduling, process contexts, multiprocessor locks, etc.

The guide is intended for network and systems programmers, who use the STREAMS mechanism at user and kernel levels for Linux and UNIX system communication services.

Readers of the guide are expected to possess prior knowledge of the Linux and UNIX system, programming, networking, and data communication.

Revisions

Take care that you are working with a current version of this manual: you will not be notified of updates. To ensure that you are working with a current version, contact the Author, or check The OpenSS7 Project website for a current version.

A current version of this manual is normally distributed with the Linux STREAMS (LiS) package.

Version Control

     
     LiS.texi,v
     Revision 1.1.6.23  2007/02/28 06:30:12  brian
     - updates and corrections, #ifdef instead of #if
     
     Revision 1.1.6.22  2006/09/18 01:06:12  brian
     - updated manuals and release texi docs
     
     Revision 1.1.6.21  2006/08/28 10:46:49  brian
     - correction
     
     Revision 1.1.6.20  2006/08/28 10:32:41  brian
     - updated references
     
     Revision 1.1.6.19  2006/08/27 12:26:26  brian
     - finalizing auto release files
     
     Revision 1.1.6.18  2006/08/26 18:31:29  brian
     - handle long urls
     
     Revision 1.1.6.17  2006/08/26 09:15:54  brian
     - better release file generation
     
     Revision 1.1.6.16  2006/08/23 11:00:19  brian
     - added preface, corrections and updates for release
     
     Revision 1.1.6.14  2006-07-02 06:13:45 -0600  brian
     - release documentation updates
     
     Revision 1.1.6.13  2006-03-22 03:01:55 -0700  brian
     - added makefile target index
     
     Revision 1.1.6.12  2005-09-15 06:59:33 -0600  brian
     - testsuite documentation update
     
     Revision 1.1.6.11  2005-06-24 07:38:56 -0600  brian
     - added troubleshooting section to manuals
     
     Revision 1.1.6.10  2005-06-23 21:35:15 -0600  brian
     - minor updates to documentation
     
     Revision 1.1.6.9  2005-05-14 02:35:10 -0600  brian
     - copyright header correction
     
     Revision 1.1.6.8  2005-04-12 03:28:52 -0600  brian
     - corrections
     
     Revision 1.1.6.7  2005-04-11 14:48:39 -0600  brian
     - documentation updates and corrections
     
     Revision 1.1.6.6  2005-04-04 22:30:00 -0600  brian
     - correct include path
     
     Revision 1.1.6.5  2005-03-15 05:06:37 -0700  brian
     - Updated texinfo documentation.
     
     Revision 1.1.6.4  2005-03-14 17:56:39 -0700  brian
     - Updated version numbering in texinfo files.
     
     Revision 1.1.6.3  2005-03-14 17:51:29 -0700  brian
     - Updated version numbering in texinfo files.
     
     Revision 1.1.6.2  2005-03-13 18:25:36 -0700  brian
     - manual updates for LiS-2.18
     
     Revision 1.1.6.1  2005-03-09 16:14:15 -0700  brian
     - First stab at autoconf'ed 2.18.0.  Results of merge.
     
     Revision 1.1.4.17  2005-02-17 13:00:03 -0700  brian
     - Fixes for texi documentation.
     
     Revision 1.1.4.16  2005-01-24 04:57:52 -0700  brian
     - Updated texinfo headers.
     
     Revision 1.1.4.15  2004-12-19 08:14:19 -0700  brian
     - Corrected include position.
     
     Revision 1.1.4.14  2004-12-16 21:02:34 -0700  brian
     - Improving spec files.
     
     Revision 1.1.4.13  2004-11-09 04:49:12 -0700  brian
     - Manual updates
     
     Revision 1.1.4.12  2004-11-06 02:14:24 -0700  brian
     - Use automatic node features.
     
     Revision 1.1.4.11  2004-10-11 20:26:26 -0600  brian
     - Added texinfo configuration file.
     
     Revision 1.1.4.10  2004-08-21 23:07:16 -0600  brian
     - Converted to common file operation.
     
     Revision 1.1.4.9  2004-08-20 15:15:51 -0600  brian
     - Documentation updates.
     
     Revision 1.1.4.8  2004-08-15 14:01:51 -0600  brian
     - Build system updates.
     
     Revision 1.1.4.7  2004-08-04 13:18:00 -0600  brian
     - Removed references to drivers and modules moved to strxns and strxnet.
     
     Revision 1.1.4.6  2004-08-04 11:24:52 -0600  brian
     - Typographical errors corrected.
     
     Revision 1.1.4.5  2004-05-25 23:25:55 -0600  brian
     - html target more sensitive to syntax
     
     Revision 1.1.4.4  2004-05-25 18:11:10 -0600  brian
     - Updated manual with NexusWare instructions.
     
     Revision 1.1.4.3  2004-05-25 14:03:06 -0600  brian
     - Updating release notes and documentation.
     
     Revision 1.1.4.2  2004-05-13 03:06:35 -0600  brian
     - made tli modules, inet driver and xnet library optional
     - added HTML output to texinfo build
     - passes distcheck
     
     Revision 1.1.4.1  2004-01-12 16:45:36 -0700  brian
     - Updated missing directories.
     
     Revision 1.1.2.5  2004-01-07 04:34:48 -0700  brian
     - Updated copyright dates.
     
     Revision 1.1.2.4  2003-12-22 21:07:31 -0700  brian
     - Updates to manuals.
     
     Revision 1.1.2.3  2003-12-16 17:23:42 -0700  brian
     - Added XTI/TLI package into release.
     
     Revision 1.1.2.2  2003-12-16 05:21:04 -0700  brian
     - Added license files and extra distributions.
     
     Revision 1.1.2.1  2003-12-15 16:38:03 -0700  brian
     - New info documentation.
     
     Revision 1.1  2003-12-15 16:38:03 -0700  brian
     file LiS.texi was initially added on branch LIS-2-16-16-autoconf.
     

ISO 9000 Compliance

Only the TeX, texinfo, or roff source for this manual is controlled. An opaque (printed, postscript or portable document format) version of this manual is an UNCONTROLLED VERSION.

Disclaimer

OpenSS7 Corporation disclaims all warranties with regard to this documentation including all implied warranties of merchantability, fitness for a particular purpose, non-infringement, or title; that the contents of the manual are suitable for any purpose, or that the implementation of such contents will not infringe on any third party patents, copyrights, trademarks or other rights. In no event shall OpenSS7 Corporation be liable for any direct, indirect, special or consequential damages or any damages whatsoever resulting from loss of use, data or profits, whether in an action of contract, negligence or other tortious action, arising out of or in connection with any use of this manual or the performance or implementation of the contents thereof.

OpenSS7 Corporation reserves the right to revise this software and documentation for any reason, including but not limited to, conformity with standards promulgated by various agencies, utilization of advances in the state of the technical arts, or the reflection of changes in the design of any techniques, or procedures embodied, described, or referred to herein. OpenSS7 Corporation is under no obligation to provide any feature listed herein.

U.S. Government Restricted Rights

If you are licensing this Software on behalf of the U.S. Government ("Government"), the following provisions apply to you. If the Software is supplied by the Department of Defense ("DoD"), it is classified as "Commercial Computer Software" under paragraph 252.227-7014 of the DoD Supplement to the Federal Acquisition Regulations ("DFARS") (or any successor regulations) and the Government is acquiring only the license rights granted herein (the license rights customarily provided to non-Government users). If the Software is supplied to any unit or agency of the Government other than DoD, it is classified as "Restricted Computer Software" and the Government's rights in the Software are defined in paragraph 52.227-19 of the Federal Acquisition Regulations ("FAR") (or any successor regulations) or, in the cases of NASA, in paragraph 18.52.227-86 of the NASA Supplement to the FAR (or any successor regulations).

Acknowledgements

As with most open source projects, this project would not have been possible without the valiant efforts and productive software of the Free Software Foundation and the Linux Kernel Community.

Sponsors

Funding for completion of the OpenSS7 Linux STREAMS (LiS) package was provided in part by:

Additional funding for The OpenSS7 Project was provided by:

Contributors

The current maintainer of the OpenSS7 Linux STREAMS (LiS) package is Brian F. G. Bidulock.

Linux STREAMS (LiS) was originally written by:

Packaging was performed by:

Packaging would not be what it is today without the invaluable help of these people:

The primary contributor to The OpenSS7 Project is Brian F. G. Bidulock.

The following is a list of significant contributors to The OpenSS7 Project:

Additional people not named here that I missed putting on the list. We thank them too!

Authors

Linux STREAMS, termed LiS, is an SVR4 compatible STREAMS executive which runs in the Linux Kernel as a loadable module. It is the product of a joint effort among the following authors:

The authors of the OpenSS7 Linux STREAMS (LiS) package include:

See Author Index, for a complete listing and cross-index of authors to sections of this manual.

Maintainer

Brian Bidulock is the principal active maintainer of LiS, so please direct questions to him rather than the others.² Ole Husgaard has contributed to the kerneld support and installation procedures. Jürgen Magin contributed patches for Linux SPARC. G Yeganjaiah added interrupt routine support. John Boyd implemented fattach and STREAMS pipes and FIFOs. Brian Bidulock developed a complete set of manual pages for LiS, converted the build process to autoconf, wrapped the source RPMS, updated this manual for texinfo and currently maintains the package.

The maintainer of the OpenSS7 Linux STREAMS (LiS) package is:

Please send bug reports to bugs@openss7.org using the send-pr script included in the package, only after reading the BUGS file in the release, or See Problem Reports.

Web Resources

The OpenSS7 Project provides a website dedicated to the software packages released by the OpenSS7 Project.

Bug Reports

Please send bug reports to bugs@openss7.org using the send-pr script included in the Linux STREAMS (LiS) package, only after reading the BUGS file in the release, or See Problem Reports. You can access the OpenSS7 GNATS database directly via the web, however, the preferred method for sending new bug reports is via mail with the send-pr script.

Mailing Lists

The OpenSS7 Project provides a number of general discussion Mailing Lists for discussion concerning the OpenSS7 Linux STREAMS (LiS) package as well as other packages released by The OpenSS7 Project.

These are mailman mailing lists and so have convenient web interfaces for subscribers to control their settings. See http://www.openss7.org/mailinglist.html.

The mailing lists are as follows:

openss7

The openss7 mailing list is for general enquiries, information exchange and announcements regarding the OpenSS7 Project. This is our original mailing list and takes the highest amount of traffic.

openss7-announce

The openss7-announce mailing list is for announcements related to the OpenSS7 Project. This list will accept announcements posted by subscribers. Subscribe to this list if you are interested in announcements from the OpenSS7 Project, subscribers and sponsors, related to the OpenSS7 Project or STREAMS, SS7, SIGTRAN or SCTP in general.

openss7-cvs

The openss7-cvs mailing list is for automatic CVS log reporting. You must get permission of the owner to subscribe to this list. Subscribers are not allowed to post to this list, this is merely for distributing notification of changes to the CVS repository.h

openss7-develop

The openss7-develop mailing list is for email exchange related to the development projects under the OpenSS7 Project. This includes development requests, proposals, requests for comment or proposal. Subscribe to this list if you are interested in ongoing development details regarding the OpenSS7 Project.

openss7-test

The openss7-test mailing list is for email exchange related to the testing of code under the OpenSS7 Project. This specifically relates to conformance testing, verification testing, interoperability testing and beta testing. Subscribe to this list if you are interested in participating in and receiving ongoing details of test activities under the OpenSS7 Project.

openss7-bugs

The openss7-bugs mailing list is specifically tailored to bug tracking. The mailing list takes a feed from the OpenSS7 GNATS bug tracking system and accepts posting of responses to bug reports, tracking and resolution. Subscribe to this list if you are interested in receiving detailed OpenSS7 release code bug tracking information. This list is not archived; for historical information on problem reports, see our GNATS databases.

openss7-updates

The openss7-updates mailing list provides updates on OpenSS7 Project code releases and ongoing activities. Subscribers are not allowed to post to this list; this list is for official OpenSS7 Project announcements only. Subscribe to this list if you are interested in receiving updates concerning official releases and activities of the OpenSS7 Project.

openss7-streams

The openss7-streams mailing list is for email exchange related to the STREAMS development projects under the OpenSS7 Project. This includes development requests, proposals, requests for comment or proposal. Subscribe to this list if you are interested in ongoing development details regarding the OpenSS7 Project STREAMS components.

linux-streams

The linux-streams mailing list is for mail exchange related to Linux Fast-STREAMS or Linux STREAMS. This includes patches, development requests, proposals, requests for comment or proposal. Subscribe to this list if you are interested in ongoing development details regarding the STREAMS for Linux components. This is the the new (September 2006) home of the linux-streams list formerly of <gsyc.escet.urjc.es>.

Spam

To avoid spam being sent to the members of the OpenSS7 mailing list(s), we have blocked mail from non-subscribers. Please subscribe to the mailing list before attempting to post to them. (Attempts to post when not subscribed get bounced.)

As an additional measure against spam, subscriber lists for all OpenSS7 mailing lists are not accessible to non-subscribers; for most lists subscriber lists are only accessible to the list administrator. This keeps your mailing address from being picked off our website by bulk mailers.

Acceptable Use Policy

It is acceptable to post professional and courteous messages regarding the OpenSS7 package or any general information or questions concerning STREAMS, SS7, SIGTRAN, SCTP or telecommunications applications in general.

Large Attachments

The mailing list is blocked from messages of greater than 40k. If you have attachments (patches, test programs, etc.) and you mail them to the list, it will bounce to the list administrator. If you are interested in making your patches, test programs, test results or other large attachments available to the members of the mailing list, state in the message that you would like them posted and the list administrator will place them in the mail archives.

Quick Start Guide

Linux STREAMS (LiS)

Package LiS-2.18.6 was released under GPLv2 2007-06-24.

The OpenSS7 Linux STREAMS (LiS) package is an OpenSS7 modified version of the LiS-2.18 package formerly from GCOM, and formerly maintained by David Grothe.

Note: The original LiS package from GCOM is no longer actively maintained by either GCOM or the OpenSS7 Project: use the OpenSS7 Linux Fast-STREAMS package <http://www.openss7.org/STREAMS.html> instead.

The following are claims made by its authors and original maintainer:

The OpenSS7 Modified Linux STREAMS (LiS) package is as STREAMS framework that is compatible with SVR 4 STREAMS. It has lots of debugging features not found in other STREAMS packages. Good to do networking and other things. It allows for installation of binary drivers.

Linux STREAMS (LiS) aims to provide SVR 4 compatible STREAMS implementation for Linux and claims to have special debugging facilities; however, the package suffers from the major failings that it is:

• no longer maintained;
• full of bugs;
• incompatible with POSIX;
• implements SVR 4 STREAMS rather than the more popular SVR 4.2 MP/ES STREAMS;
• portions dubiously licensed under the LGPL;
• unsuitable for mainline adoption;
• adapts poorly to production kernels;
• many SMP races and lockups;
• broken 32-bit over 64-bit compatibility;
• no strlog(9) STREAMS logger;

This distribution is only currently applicable to Linux 2.4 and 2.6 kernels and was targeted at ix86, x86_64, ppc and ppc64 architectures, but should build and install for other architectures as well.

Release

This is the LiS-2.18.6 package, released 2007-06-24. This `2.18.6' release, and the latest version, can be obtained from the download area of The OpenSS7 Project website using a command such as:

     $> wget http://www.openss7.org/tarballs/LiS-2.18.6.tar.bz2

The release is available as an autoconf(1) tarball, src.rpm or dsc, or as a set of binary rpms or debs. See the download page for the autoconf(1) tarballs, src.rpms or dscs. See the LiS package page for tarballs, source and binary packages.

Please see the NEWS file for release notes and history of user visible changes for the current version, and the ChangeLog file for a more detailed history of implementation changes. The TODO file lists features not yet implemented and other outstanding items.

Please see the INSTALL, INSTALL-LiS and README-make, files (or see Installation) for installation instructions.

When working from cvs(1) or git(1), please see the README-cvs, file (or see Downloading from CVS). An abbreviated installation procedure that works for most applications appears below.

This release of the package is published strictly under Version 2 of the GNU Public License which can be found in the file COPYING. Package specific licensing terms (if any) can be found in the file LICENSES. Please respect these licensing arrangements. If you are interested in different licensing terms, please contact the copyright holder, or OpenSS7 Corporation <sales@openss7.com>.

See README-alpha (if it exists) for alpha release information.

Prerequisites

The quickest and easiest way to ensure that all prerequisites are met is to download and install this package from within the OpenSS7 Master Package, openss7-0.9.2.F, instead of separately.

Prerequisites for the Linux STREAMS (LiS) package are as follows:

1. Linux distribution, somewhat Linux Standards Base compliant, with a 2.4 or 2.6 kernel and the appropriate tool chain for compiling out-of-tree kernel modules. Most recent Linux distributions are usable out of the box, but some development packages must be installed. For more information, see Compatibility.

   	− A fairly LSB compliant GNU/Linux distribution.3
   	− Linux 2.4 kernel (2.4.10 - 2.4.27), or
   	− Linux 2.6 kernel (2.6.3 - 2.6.21);
   	− glibc2 or better.
   	− GNU info (for info files).
   	− GNU groff (for man pages).4

When configuring and building multiple OpenSS7 Project release packages, place all of the source packages (unpacked tarballs) at the same directory level and all build directories at the same directory level (e.g. all source packages under /usr/src).

When installing packages that install as kernel modules, it is necessary to have the correct kernel development package installed. For the following distributions, use the following commands:

     Ubuntu:  $> apt-get install linux-headers
     Debian:  $> apt-get install kernel-headers
     Fedora:  $> yum install kernel-devel

You also need the same version of gcc(1) compiler with which the kernel was built. If it is not the default, add `CC=kgcc' on the line after `./configure', for example:

     $> ../LiS-2.18.6/configure CC='gcc-3.4'

Installation

The following commands will download, configure, build, check, install, validate, uninstall and remove the package:

     $> wget http://www.openss7.org/tarballs/LiS-2.18.6.tar.bz2
     $> tar -xjvf LiS-2.18.6.tar.bz2
     $> mkdir build
     $> pushd build
     $> ../LiS-2.18.6/configure --enable-autotest
     $> make
     $> make check
     $> sudo make install
     $> sudo make installcheck
     $> sudo make uninstall
     $> popd
     $> sudo rm -rf build
     $> rm -rf LiS-2.18.6
     $> rm -f LiS-2.18.6.tar.bz2

If you have problems, try building with the logging targets instead. If the make of a logging target fails, an automatic problem report will be generated that can be mailed to The OpenSS7 Project.⁵ Installation steps using the logging targets proceed as follows:

     $> wget http://www.openss7.org/tarballs/LiS-2.18.6.tar.bz2
     $> tar -xjvf LiS-2.18.6.tar.bz2
     $> mkdir build
     $> pushd build
     $> ../LiS-2.18.6/configure --enable-autotest
     $> make compile.log
     $> make check.log
     $> sudo make install.log
     $> sudo make installcheck.log
     $> sudo make uninstall.log
     $> popd
     $> sudo rm -rf build
     $> rm -rf LiS-2.18.6
     $> rm -f LiS-2.18.6.tar.bz2

See README-make for additional specialized make targets.

For custom applications, see the INSTALL and INSTALL-LiS files or the see Installation, as listed below. If you encounter troubles, see Troubleshooting, before issuing a bug report.

Brief Installation Instructions

The Linux STREAMS (LiS) package is available from the downloads area of The OpenSS7 Project website using a command such as:

     $> wget http://www.openss7.org/tarballs/LiS-2.18.6.tar.bz2

Unpack the tarball using a command such as:

     $> tar -xjvf LiS-2.18.6.tar.bz2

The tarball will unpack into the relative subdirectory named after the package name: LiS-2.18.6.

The package builds using the GNU autoconf utilities and the configure script. To build the package, we recommend using a separate build directory as follows:

     $> mkdir build
     $> cd build
     $> ../LiS-2.18.6/configure

In general, the package configures and builds without adding any special options to the configure script. For general options to the configure script, see the GNU INSTALL file in the distribution:

     $> less ../LiS-2.18.6/INSTALL

For specific options to the configure script, see the INSTALL-LiS file in the distribution, or simply execute the configure script with the --help option like so:

     $> ../LiS-2.18.6/configure --help

After configuring the package, the package can be compiled simply by issuing the `make' command:

     $> make

Some specialized makefile targets exists, see the README-make file in the distribution or simply invoke the `help' target like so:

     $> make help | less

After successfully building the package, the package can be checked by invoking the `check' make target like so:

     $> make check

After successfully checking the package, the package can be installed by invoking the `install' make target (as root) like so:

     $> sudo make install

The test suites that ship with the package can be invoked after the package has been installed by invoking the `installcheck' target. This target can either be invoked as root, or as a normal user, like so:

     $> make installcheck

(Note: you must add the --enable-autotest flag to configure, above for the test suites to be invoked with `make installcheck'.)

The package can be cleanly removed by invoking the `uninstall' target (as root):

     $> sudo make uninstall

Then the build directory and tarball can be simply removed:

     $> cd ..
     $> rm -rf build
     $> rm -rf LiS-2.18.6
     $> rm -f LiS-2.18.6.tar.bz2

Detailed Installation Instructions

More detailed installation instructions can be found in the Installation, contained in the distribution in `text', `info', `html' and `pdf' formats:

     $> cd ../LiS-2.18.6
     $> less doc/manual/LiS.txt
     $> lynx doc/manual/LiS.html
     $> info doc/manual/LiS.info
     $> xpdf doc/manual/LiS.pdf

The `text' version of the manual is always available in the MANUAL file in the release.

The current manual is also always available online from The OpenSS7 Project website at:

     $> lynx http://www.openss7.org/LiS_manual.html

1 Introduction

This manual documents the design, implementation, installation, operation and future development schedule of the Linux STREAMS (LiS) package.

1.1 Overview

This manual documents the design, implementation, installation, operation and future development of the Linux STREAMS (LiS) package.

LiS is a software package that comprises an implementation of SVR4 compatible STREAMS for Linux. It takes the form of a loadable module for the Linux kernel. LiS installs in any directory on your system, not in the kernel source tree. (see Installation)

LiS-2.12 and beyond utilizes aggressive multi-tasking in multiple CPU SMP environments. For further information concerning this implementation, see LiS SMP Implementation.

WARNING: This autoconf/RPM release of Linux STREAMS is distributed under the terms of the GNU Public License (GPL) and not the GNU Lesser Public License (LGPL).

This means that you cannot link proprietary STREAMS drivers with LiS and load the entirety into the Linux kernel without violating license restrictions. OpenSS7 Corporation can remove this restriction for subscribers and sponsors of the OpenSS7 Project.

1.2 Organization of this Manual

This manual is organized (loosely) into several sections as follows:

1.3 Conventions and Definitions

This manual uses texinfo typographic conventions.

2 Objective

3 Reference

3.1 Files

specfs.o

streams.o

streams-aixcompat.o

streams-hpuxcompat.o

streams-liscompat.o

streams-osfcompat.o

streams-suncompat.o

streams-svr4compat.o

streams-uw7compat.o

3.2 Drivers

The LiS package comes with a number of STREAMS drivers and pushable modules in source code form. A number of these drivers and modules are small entities that are used in the testing of LiS. They are included so as to make it easy for any user to run the LiS tests for themselves.

Other drivers are used to implement STREAMS based pipes and FIFOs.

A driver in STREAMS has a major and minor device number associated with it and an entry in the /dev directory. The driver is opened and closed just like any file. The driver names used in this manual are the declared names that appear in the LiS Config file for the particular driver.

3.2.1 clone-drvr

Device Name

     /dev/clone_drvr

Description

This driver is used to assist LiS in implementing the "clone" open function. It appears under its own name as /dev/clone_drvr. By convention, it is allocated the first major number of all the STREAMS drivers. In order to implement clone opens, one creates a node in the /dev directory for a device whose major number is set to that of the clone driver, and whose minor number is the major number of the driver to which the clone open is to be directed. The clone driver's open routine transfers the open call to the target driver, passing a unique flag that informs the driver that a clone open is being requested. The target driver then allocated a minor device number to uniquely associate with this instance of the open operation. The clone driver synthesizes a new major/minor "device id" to pass back to LiS. LiS recognizes the change of major/minor from the original open and takes steps to allocate control structures unique to this open.

The "clone open" operation is intended to make is easy to open one device from a pool of devices, such as pseudo ttys or logical connections. It saves application programs from having to scan a list of device mnemonics issuing trial opens until one is found that succeeds.

Note that the driver is named /dev/clone_drvr instead of the more traditional SVR4 /dev/clone. This is to avoid a conflict with another driver named /dev/clone on Linux systems.

Author

David Grothe dave@gcom.com

3.2.2 fifo

Device Name

     /dev/fifo (clone device)
     /dev/fifo.0

Description

The fifo pseudo-driver (which is internal to LiS) provides STREAMS-based fifos as single character special files, and STREAMS-based pipes as pairs of character special files which are interconnected (see pipe(3)).

STREAMS-based fifos differ from typical STREAMS-based character special files in that there are not separate stream head and driver queue pair within the STREAMS-based file. Instead, a fifo is created with only a single queue pair for the stream head. Moreover, in a typical driver queue pair, the write queue is not connected to a next queue. In a fifo, the write queue is directed to the read queue of the pair. A pipe comprises a pair of fifos, with the write queue of each pair directed to the read queue of the other. The two fifos comprising a pipe are referred to as peers, and each somewhat represents a driver to the other. As a degenerate case, a fifo is its own peer.

STREAMS modules may be pushed onto fifos and pipes, but should not expect a driver below them; instead, the SAMESTR() function should be used from the write queue of a pair to determine if the module is the lowest in the STREAMS-based file (this is called the midpoint). The structure of a fifo or pipe is preserved when modules are pushed (and popped); i.e., the write queue at the midpoint will always be directed at the read queue of the peer.

Input and output are handled at a fifo stream head as they would normally be handled at a stream head. In LiS, an fifo open() entry point exists to assign minor device numbers to new opens under the fifo major device number, and a close() entry point is used correspondingly to release them. These functions are kept in a streamtab data struc ture (as they would normally be for any STREAMS driver or module) which is private to the LiS implementation.

Application Usage

In the current Linux kernels, character special major numbers are limited to 16 bits, and major and minor device numbers to 8 bits each. This limits a system to 256 total major device numbers and 256 total minor devices per major device number. This is a rather severe limitation where mechanisms like fifos and pipes are concerned.

However, a driver may handle more than one major device number. The fifo pseudo-driver uses this to overcome this limitation, by supporting the automatic allocation and use of multiple major device numbers for fifos and pipes. Specifying more than 256 minor devices is done in the usual manner, i.e., by specifying the number of "units" in the appropriate Config file. Enough major device numbers will be allocated to cover the requested number of minor devices (if available, else an error will occur in strconf(8)). The number allocated will include one minor device per major number to be used as a fifo-specific clone minor device (specifically, minor number 0), which exhibits special behaviour. Normally, when cloning is done via the clone pseudo-driver, the clone major device number is used, along with the desired actual major number as the minor device number. When an open() is performed on such a device, the clone open() routine in turn calls the appropriate driver's open(), with the sflag parameter set to CLONEOPEN. The driver's open() is expected in this case to allocate an unused minor device number, and return it via an entirely new device number in the devp parameter. In this way, a driver can change the device number to be used for a STREAMS-based file. When minor device 0 for a specified for a fifo major device, the driver will also clone a new minor device number. However, LiS opens fifo devices differently; specifically, when an already-opened fifo-specific clone minor device is reopened, the new and subsequent opens will use the already-opened clone. Thus, using minor device 0 for a fifo when creating a file sys tem node will ensure that all concurrent opens of the associated path name will use the same STREAMS-based file; at the same time, opens of different file system nodes via different paths will open their respectively different STREAMS-based files. This is essentially how kernel-based fifos behave -applications and users of STREAMS-based fifos don't have to keep track of minor numbers to achieve this same behaviour when it is desired.

It is in fact recommended that only two forms of file sys tem nodes be used for STREAMS-based fifos: the clone major number as major number with a fifo major number as minor number, to be used when every open of the associated path must clone a new fifo, and a fifo major number as major number with 0 as the minor number, to be used when new opens are to clone a new fifo but subsequent concurrent opens are to use the already opened fifo. These are represented by two device special file paths created when LiS is installed: /dev/fifo for the former, and /dev/fifo.0 for the latter. It is recommended that these be used, possibly along with the equivalent of stat(2) to determine appropriate major device numbers for the clone and fifo pseudo-drivers, which are also determined when LiS is installed. It can be noted that pipes are actually created as instances of the former, after which the write queues are peer-connected. The fifo pseudo-driver allocates minor devices in round-robin fashion; i.e., a list of available minor devices is kept, and once a minor number is finally closed, it is put at the end of this list. Thus, a fifo minor device which is opened and closed will not be immediately reused.

Warnings

Because STREAMS-based fifos and pipes are implemented as character special devices, they do not appear as pipe devices when examined with stat(2) or the equivalent (e.g., ls(1)); i.e. the S_IFIFO indication is not set in the mode - S_IFCHR is set instead, and the actual device number is indicated in the st_rdev field of the stat data structure.

Because of the potential use of multiple major numbers, applications should not depend on a fifo or pipe having a specific major device number, nor should an application depend on all fifos and pipes having the same major device number.

See Also

clone(9), connld(9), fifo(4), ls(1), pipe(3), pipemod(9), STREAMS(4), stat(2), strconf(8)

Author

John Boyd, protologos LLC. jaboydjr@netwalk.com

3.2.3 loop-around

Device Name

     /dev/loop_clone (clone device)
     /dev/loop.1
     /dev/loop.2

Description

This driver is used by LiS and the strtst(8) utility to assist in the regression testing of LiS. It connects two streams together in a manner similar to that of a pipe. Messages written into one stream can be read back from the other.

The driver can be operated as a clone device with the two streams being connected via ioctls. A number of ioctls exist that tailor the operation of the driver. The user codes these ioctls as type I_STR and passes a structure of type struct strioctl to the driver. The ic_cmd field of this structure is decoded according to the following table. the ic_dp and ic_len fields delimit an argument structure which is also passed to the driver. The argument structure differs for each type of ic_cmd.

Author

David Grothe dave@gcom.com plus others originally.

3.2.4 mini-mux

Device Name

     /dev/mux_clone (clone device)
     /dev/minimux.1
     /dev/minimux.2

Description

This driver is used by LiS in its testing procedures. It is a small multiplexing driver that allows cascaded multiplexors to be built and torn down. The driver uses a pair of ioctls to establish connectivity between upper streams and lower streams. This allows control over how data flows through the multiplexor.

Both of these ioctls are coded as type I_STR and pass a structure of type struct strioctl to the driver. The ic_cmd field of this structure is decoded according to the following table. the ic_dp and ic_len fields delimit an argument structure which is also passed to the driver. The argument structure may differ for each type of ic_cmd.

Author

David Grothe dave@gcom.com

3.2.5 printk

Device Name

     /dev/printk

Description

This driver accepts messages written to it and prints them from the kernel using the kernel's printk function. It is used by the LiS test software to keep messages from LiS and messages from the test program in sequence.

Author

David Grothe dave@gcom.com

3.2.6 sad

Device Name

     /dev/sad

Description

The STREAMS Administrative Driver manages the autopush function of LiS. Using ioctls the system administrator can provide a list of modules that are to be automatically pushed onto a given device when that device is opened. The controls are specified via the strapush structure which is defined in <sys/sad.h>.

The ioctl used by the user is of the form:

ioctl(fd, command, arg)

Where fd is the file descriptor of the file that is open to the sad driver, command and arg are described in the following table.

The strapush structure used by the SAD_SAP and SAD_GAP ioctls contains the following fields.

unsigned sap_cmd

This is the autopush command to be executed. The values are as follows.

SAP_ONE

Configure one minor device of the driver indicated by sap_major.

SAP_RANGE

Configure a range of minor devices of the driver indicated by sap_major. The range runs from sap_minor to sap_lastminor, inclusively.

SAP_ALL

Configure all minor devices of the driver indicated by sap_major.

SAP_CLEAR

Undo all autopush configuration for the driver indicated by sap_major.

major_t sap_major

The major device number of the driver which is being configured for autopush.

minor_t sap_minor

The minor device being configured, or the first of a range.

minor_t sap_lastminor

The last minor device of a range to be configured.

unsigned sap_npush

Number of modules to be pushed when the indicated device is opened.

char sap_list[MAXAPUSH][FMNAMESZ+1]

List of module names to be pushed, or list of modules names returned to user.

The ioctl function call returns zero upon success or -1 on failure. Upon failure errno(3) is set to the error number describing the failure, usually either EFAULT or EINVAL.

Note that the sad driver is a standard AT&T STREAMS function. More comprehensive documentation for this driver can be found in the [40]SVR4 Programmer's Guide: STREAMS.

Author

Ole Husgaard sparre@login.dknet.dk

3.3 Modules

streams-connld.o

Connld module.

streams-pipemod.o

Pipe module.

streams-sc.o

STREAMS configuration module.

streams-sth.o

Stream Head module.

The LiS package comes with a number of STREAMS drivers and pushable modules in source code form. A number of these drivers and modules are small entities that are used in the testing of LiS. They are included so as to make it easy for any user to run the LiS tests for themselves.

A pushable module in STREAMS is an entity that is added to an existing STREAMS file via the I_PUSH ioctl. These modules are known to LiS by mnemonic name, given as an argument to the I_PUSH ioctl. There are no major and minor device numbers or /dev entries associated with pushable modules.

3.3.1 connld

Module Name

connld

Description

The connld module provides a means to generate multiple unique STREAMS-based pipes from a single existing pipe end. connld may only be pushed (via the STREAMS I_PUSH ioctl) onto a STREAMS-based pipe. When first pushed, connld does nothing; on each subsequent open(2), connld will generate a unique STREAMS-based pipe. One end of each new pipe replaces the original pipe end from the perspective of the open call. The other end of each new pipe is sent, effectively as if by the I_SENDFD ioctl, to the other end of the original pipe, ostensibly to be received by a subsequent I_RECVFD ioctl operation.

Application Usage

The intent of connld is to provide a means to generate unique pipes which separately and independently connect client processes to a server process. The point of access for such clients is expected to be a path name known to all such clients and to which a pipe end may be connected (via fattach(3)) by the server process. The server establishes the original pipe, pushes connld onto the client end, and then listens via I_RECVFD for new connections on the server end. A client wishing to connect to the server will open(2) the path name representing the client end, and can determine via isastream(3) whether or not the server process is active and attached. If it is, the open() call returns one end of a unique new pipe that thus connects the client to the server.

Such a server is responsible both for accepting new connections via I_RECVFD on the original pipe, and for communicating with clients so connected via the received pipe ends. It would also be reasonable for such a server process to invalidate the point of access by calling fdetach(3) before terminating.

It should be noted that the poll(2) primitive may be used to indicate when an M_PASSFP representing a newly passed file is available on the original server pipe end. This is reflected by the POLLIN status setting in the events and revents fields of a pollfd structure. Moreover, any attempt to read an M_PASSFP message via the data-receiving primitives (i.e., read(2), getmsg(3), and getpmsg(3)) will fail with errno(3) returning an EBADMSG indication without discarding the message.

Even so, it should be reasonable to expect only M_PASSFP messages will be received on the original server pipe end, since it is not possible to carry on normal data traffic which has connld on one end, since connld does not support such traffic.

The use of connld can be made entirely free-standing by attaching well-known paths to both ends of the original pipe. The relevant capabilities are implemented in LiS so that the original creator of the pipe can close both ends after attaching paths to them, and the process of passing file descriptors can still be carried out via new open()'s as long as both ends remain attached.

See Also

fattach(3), fattach(8), fdetach(3), fifo(4), fifo(9),

pipe(3), STREAMS(4)

History

Unix System V Release 4 (SVR4)

Author

John Boyd, protologos LLC. jaboydjr@netwalk.com

3.3.2 pipemod

Module Name

pipemod

Description

The pipemod module has the relatively simple task of reversing the sense of the FLUSH flag bits in M_FLUSH messages sent in STREAMS-based fifos and pipes. This must happen at the midpoint of a fifo or pipe, so that FLUSHR becomes FLUSHW, and FLUSHW becomes FLUSHR. pipemod does this, and has no other function.

To be used appropriately, then, pipemod must be the first module pushed onto a pipe end or a fifo, but it is only necessary on one end of a pipe.

pipemod is not needed if flush handling need not be supported, or if its function is supported by other means.

See Also

fifo(9), pipe(3), fifo(4), STREAMS(4)

History

Unix System V Release 4 (SVR4)

Author

John Boyd, protologos LLC. jaboydjr@netwalk.com

3.3.3 relay, relay2

Module Name

relay relay2

Description

These are two names for the same module. All the module does is forward STREAMS messages along on the stream using putnext(9). These modules are used in the testing of LiS but are not otherwise useful. One could use the source code as a starting point for coding a pushable STREAMS module.

Author

David Grothe dave@gcom.com

3.4 Libraries

During the installation process of Linux STREAMS (LiS) a subroutine library is built and installed on your system. Three versions of the library are built and installed. They are as follows.

libLiS.a

Interface routines to LiS in static library form.

libLiS.so

Interface routines to LiS in dynamic library form.

libpLiS.so

Like libLiS.so but omits the "pipe" system call.

These three libraries are copied to the directory /usr/lib when LiS is installed. In addition, the utility program ldconfig is run during the LiS make install. This causes this library to be linked, or searched, ahead of the standard C library. This is necessary because the standard C library contains dummy routines for the STREAMS interface functions, or most of them in the best case. If these dummy routines preempt the LiS versions then STREAMS applications will always perceive error returns from such routines as see getmsg(2) and see putmsg(2).

3.4.1 Library Routines

The following routines are present in the libraries libLiS.a and libLiS.so. The library libpLiS.so omits the "pipe" routine.

The routines in these libraries are standard STREAMS interface routines. As such we do not offer detailed descriptions of the functions of these routines. Instead we refer the reader to the AT&T SVR4 STREAMS documentation.

     int fattach(int fd, const char *path);
     int fdetach(const char *path);
     int getmsg(int fd, void *ctlptr, void *dataptr, int *flagsp);
     int getpmsg(int fd, void *ctlptr, void *dataptr, int *bandp, int *flagsp);
     int isastream(int fd);
     int pipe(int *fd);
     int poll(void *pollfds, long nfds, int timeout);
     int putmsg(int fd, void *ctlptr, void *dataptr, int flags);
     int putpmsg(int fd, void *ctlptr, void *dataptr, int *bandp, int *flagsp);

int fattach(int fd, const char *path);

see fattach(3)

int fdetach(const char *path);

see detach(3)

int getmsg(int fd, void *ctlptr, void *dataptr, int *flagsp);

see getmsg(2)

int getpmsg(int fd, void *ctlptr, void *dataptr, int *bandp, int *flagsp);

see getpmsg(2s)

int isastream(int fd);

see isastream(3)

int pipe(int *fd);

see pipe(2s)

int poll(void *pollfds, long nfds, int timeout);

see poll(2s)

int putmsg(int fd, void *ctlptr, void *dataptr, int flags);

see putmsg(2)

int putpmsg(int fd, void *ctlptr, void *dataptr, int *bandp, int *flagsp);

see putpmsg(2s)

These routines are all very small pieces of code. Most of them simply pass their parameters to LiS via a system call. The see fattach(3) and fdetach see fdetach(3) routines use ioctls to LiS if there is no system call available to call directly.

The poll see poll(2s) routine simply executes the poll system call. It is present for backward compatibility to 2.0 kernels, in which LiS provided the poll system call.

The see pipe(2s) routine has the same semantics as the standard C library routine. It uses STREAMS FIFOs to implement the pipe instead of the standard Linux pipes.

The libpLiS.so library, the one that preempts the standard C library, omits the STREAMS pipe routine so that standard Linux pipes are used unless the user explicitly links in libLiS.

3.4.2 Using the Library

To use one of the LiS libraries you can include the file <sys/stropts.h> in your program source code. On your compiler command line you can add the option `-I/usr/include/LiS' to include the version of stropts.h that is distributed with LiS, or omit the option to include the system standard header file. The two header files are believed to be compatible enough that it does not matter which one you include in your program. When linking your program, or performing a final cc to build your executable, include one of the following options on your command line.

/usr/lib/libLiS.a

Use libLiS.a (static, includes "pipe")

-lLiS

Use libLiS.so (dynamic, includes "pipe")

-lpLiS

Use libpLiS.so (dynamic, omits "pipe")

Omit any options

As of libc-2.2.1 the LiS STREAMS interface routines will be used automatically via libpLiS.so.

3.5 Utilities

The Linux STREAMS (LiS) package contains some user level commands that are used to manage the package and assist the user with STREAMS functions.

These commands are installed in /usr/bin or /usr/sbin. They are referred to a "global commands." A second group is built in the LiS installation directory and left there. This second group is oriented more toward testing of LiS than toward its operation. These commands remain undocumented since they are primarily intended for the use of the authors of the modules that they test.

These are the commands that are installed in /usr/bin or /usr/sbin, and are thus globally accessible to any user with those directories in his/her path.

3.5.1 fattach

     /usr/sbin/fattach [-v] [-m|-u|-M mode] [-p|STREAMS-path] path ...
     /usr/sbin/fattach -?

Description

The fattach program provides a command-line interface to the underlying fattach(3) function. If the -p and/or the -c option is specified, a STREAMS-based pipe is created and its two ends are alternately attached to the path names given. In this mode of usage, at least two path names are required, but there need not be an even number of path names (i.e., the pipe ends need not be attached to the same number of paths).

If the -p and -c options are not specified, the first path name given must identify a STREAMS-based file. That file will be opened, and it will be attached to each of the path names subsequently specified (of which there must be at least one).

Options

-p

Create a STREAMS-based pipe, to which to attach the subsequently specified path names. The first path will be attached to the first pipe end, the second to the second pipe end, the third to the first pipe end, etc., until the list of path names is exhausted.

By default, the umask (see umask(2)) is also applied to each end of the pipe after attaching. (See fattach(3)).

-c

Like -p (both may be given), but additionally pushes the connld module onto the first end of the pipe. This conveniently creates a free-standing pipe-serving pipe (see connld(9), and below).

-m

Apply the mode of the last-specified path(s) to the attached STREAMS-based file(s) after attaching. (See fattach(3).

-u

Apply the umask (see umask(2)) of the STREAMS-based file after attaching. (See fattach(3)). This is done by default when a pipe is created via -p.

-M mode

Apply the given mode to the STREAMS-based file(s) after attaching. (See fattach(3)).

-v

Operate in a "verbose" manner. This causes fattach to report its progress via message output to stdout or stderr.

-?

Provide a usage summary.

Return Value

Upon successful completion, i.e., if all given path names are attached to, fattach returns 0. Upon failure, fattach returns 1. However, the failure of one more attachments does not otherwise affect those that succeed, and the user is responsible for detaching any that may have succeeded if that is the desired behaviour in the event of any failures.

Application Usage

The -p and -c options provide a convenient means for creating free-standing mounted pipes. The openers of the paths attached via -p will share a single pipe, while the openers of the paths attached via -c will have access to a pipe-serving pipe. I.e., each open of the first end (e.g., the client end) will generate a new pipe, one end of which will be given to the opener, and the other end of which will be passed as if by the I_SENDFD ioctl to the path attached to the other end (e.g., the server end). Each opener of the server path could poll(2) for input, receive a new pipe end using the I_RECVFD ioctl, and then close the server path, thereafter using the new pipe end to communicate with the corresponding opener of the client path (note that the sense of client and server will in fact depend on the application - users of the two paths need only be aware of whether or not an I_RECVFD ioctl must be performed).

See Also

connld(9), fattach(3), fdetach(3), fdetach(8), STREAMS(4), umask(2)

History

An fattach function has been provided for various STREAMS implementations based on SVR4 STREAMS. Not all of these have provided a corresponding utility program of this sort.

Author

John Boyd, protologos LLC jaboydjr@netwalk.com

3.5.2 fdetach

     /usr/sbin/fdetach [-v] path ...
     /usr/sbin/fdetach -a
     /usr/sbin/fdetach -?

Description

The fdetach program provides a command-line interface to the underlying fdetach(3) function.

It is thus intended to provide a convenient means to dismantle so-called mounted STREAMS.

If the -a option is specified, all currently attached STREAMS-based files are detached. If the -a option is not specified, the path names given are taken to identify paths to which STREAMS-based files are currently attached. Those files will be detached from these paths.

Options

-a

Undo all attachments currently in effect.

-v

Operate in a "verbose" manner. This causes fdetach to report its progress via message output to stdout or stderr.

-?

Provide a usage summary.

Return Value

Upon successful completion, i.e., if all given path names identify mounted STREAMS and these are all successfully detached, fdetach returns 0. Upon failure, fdetach returns 1.

Note, however, that a failure indication does not mean that no action is taken; i.e., those detachments that succeed are not affected by those that fail.

Warnings

It should be noted that although the fdetach program implements the -a option, by passing "*" to the fdetach function, this is not at all equivalent to specifying "*" on the command line when executing the program. Normally, "*" specified on the command line will be converted by a shell into a list of all files in the current working directory. By contrast, the -a option causes the fdetach operation to operate not with respect to path names at all, but with respect to STREAMS devices currently active within the STREAMS subsystem. I.e., each active stream head is examined for attachments, and any attachments found are dismantled.

The intended use for the -a option is thus to undo all attachments, e.g., in preparation for unloading the STREAMS subsystem.

See Also

fdetach(3), fattach(8), STREAMS(4)

History

An fdetach function has been provided for various STREAMS implementations based on SVR4 STREAMS. Not all of these have provided a corresponding utility program of this sort.

Author

John Boyd, protologos LLC jaboydjr@netwalk.com

3.5.3 polltst

     /usr/bin/polltst

Description

polltst is a simple test program for the poll system call. Using poll, it reads keystrokes from stdin, writes them to one end of the LiS loopback driver, reads them from the other end and then writes them back to stdout.

While performing this operation it configures stdin for "no echo" mode, so the appearance of "echoed" characters is evidence of the operation of poll involving both a STREAMS and a non-STREAMS file.

Author

David Grothe dave@gcom.com

3.5.4 streams

     /usr/sbin/streams Options

Description

The streams program is used to perform several different management functions for the LiS package, including starting and stopping the LiS subsystem.

Options

`start'

Start the LiS subsystem. This amounts to performing the command "modprobe streams".

`stop'

Stop the LiS subsystem. This amounts to performing the command "modprobe -r streams".

`status'

Reports on the status of the LiS subsystem.

-c Kbytes

Print or set the maximum message memory usage for LiS. The value 0 (default) means unlimited.

-C Kbytes

Print or set the maximum total memory usage for LiS. The value 0 (default) means unlimited.

-d mask

Set the debug mask for LiS. See below for details.

-D mask

Set an additional debug mask for LiS. See below for details.

-s

Print STREAMS memory usage statistics.

-L

Print out lock contention statistics. Use debug bit DEBUG_LOCK_CONTENTION to enable the lock contention statistics gathering.

-m

Print STREAMS memory allocation to the system messages file (from kernel). This option should be used only for debugging and only when LiS is in a quiescent state. Unpredictable results can occur if this option is used while LiS memory allocations are changing dynamically.

-p

Print the LiS lock trace buffer to the system messages file (from kernel). Used in conjunction with the DEBUG_SPL_TRACE debug option.

-q

Print all STREAMS queues to the system messages file (from kernel). This option should be used only for debugging and only when LiS is in a quiescent state. Unpredictable results can occur if this option is used while LiS queue allocations are changing dynamically.

-S

Print out STREAMS queue-runner thread statistics.

-t

Print STREAMS timing statistics. Used in conjunction with the DEBUG_MEAS_TIME debug option.

-T

Print the LiS semaphore latency histogram. Use debug bit DEBUG_SEMTIME to enable the statistics collection.

-h

Print a command synopsis.

-H

Print a command synopsis including the debug mask mnemonics.

Debug Options

The value that is used with the -d option consists of the logical "or" of the following single bit options.

-d Options

          DEBUG_OPEN             0x00000001
          DEBUG_CLOSE            0x00000002
          DEBUG_READ             0x00000004
          DEBUG_WRITE            0x00000008
          DEBUG_IOCTL            0x00000010
          DEBUG_PUTNEXT          0x00000020
          DEBUG_STRRPUT          0x00000040
          DEBUG_SIG              0x00000080
          DEBUG_PUTMSG           0x00000100
          DEBUG_GETMSG           0x00000200
          DEBUG_POLL             0x00000400
          DEBUG_LINK             0x00000800
          DEBUG_MEAS_TIME        0x00001000
          DEBUG_MEM_LEAK         0x00002000
          DEBUG_FLUSH            0x00004000
          DEBUG_FATTACH          0x00008000
          DEBUG_SAFE             0x00010000
          DEBUG_TRCE_MSG         0x00020000
          DEBUG_CLEAN_MSG        0x00040000
          DEBUG_SPL_TRACE        0x00080000
          DEBUG_MP_ALLOC         0x00100000
          DEBUG_MP_FREEMSG       0x00200000
          DEBUG_MALLOC           0x00400000
          DEBUG_MONITOR_MEM      0x00800000
          DEBUG_DMP_QUEUE        0x01000000
          DEBUG_DMP_MBLK         0x02000000
          DEBUG_DMP_DBLK         0x04000000
          DEBUG_DMP_STRHD        0x08000000
          DEBUG_ADDRS            0x80000000
     

-D Options

          DEBUG_SNDFD            0x00000001
          DEBUG_CP	       0x00000002
          DEBUG_CACHE	       0x00000004
          DEBUG_LOCK_CONTENTION  0x00000008
          DEBUG_REFCNTS          0x00000010
          DEBUG_SEMTIME          0x00000020
     

Most of these options are intuitive as to their operation from the mnemonics.

The DEBUG_MEAS_TIME option causes LiS to use a high precision timer to calculate the execution time of several operations within itself. These timings include the time spent in STREAMS drivers. Thus, under controlled circumstances this option can be used to time STREAMS driver code. It is used in conjunction with the -t option to print out the timing statistics.

The DEBUG_SAFE option causes LiS to carefully check for NULL pointers when performing message passing and queueing operations such as putq(9) and putnext(9).

The DEBUG_CLEAN_MSG option causes LiS to clear message data buffers to zero when they are allocated. It is useful for tracking down driver problems relating to using uninitialized areas of messages.

The DEBUG_SPL_TRACE option causes LiS to maintain a trace table of all LiS locking operations. It is used in conjunction with the -p option to print out the lock trace table. The locking operations that are traced include calls on the LiS locking primitives from STREAMS drivers.

The options DEBUG_DMP_QUEUE, DEBUG_DMP_MBLK and DEBUG_DMP_DBLK control the verbosity of the printing out of LiS memory areas via the -m option. With these debug mask bits set, LiS will print out the contents of these structures as well as the headers indicating that such a structure was allocated.

The DEBUG_ADDRS option causes the -m option to print out the addresses of structures as well as their memory tags and/or contents.

The DEBUG_MONITOR_MEM option causes LiS to monitor the guard words surrounding allocated memory areas in an attempt to catch overwriting of these words in a timely fashion. This option comes at a fairly substantial CPU time penalty.

Author

David Grothe dave@gcom.com

3.5.5 strmakenodes

     /usr/sbin/strmakenodes

Description

strmakenodes makes all of the /dev entries that are associated with LiS as a result of the LiS build process. All of the Config files that contributed to the LiS build are scanned for their "node" declarations. strmakenodes performs a mknod system call for each specified "node". This command must be run before LiS can operate correctly after it is installed. This command is run automatically as a result of the "make install" operation of LiS.

This command accepts the option "-r" to mean remove nodes instead of making them. The command is run with this option as a result of the "make uninstall" operation.

The source code for this command is generated automatically as a side-effect of running the strconf utility.

3.5.6 strtst

     /usr/bin/strtst

Description

strtst is a test program which tests the core functionality of LiS. It is a user level program which uses the built-in drivers that are installed by default with LiS. It performs numerous STREAMS operations and checks the results for correctness. It prints out a voluminous log file whose output is routed to the "messages" file (kernel informational messages).

The output of strtst can be compared to earlier "reference" outputs to see if the behaviour of LiS has changed as a result of modifications to the code.

Author

David Grothe dave@gcom.com

3.5.7 timetst

     /usr/bin/timetst [Iterations]

Description

timetst peforms a timing test using the LiS loopback driver. It writes short messages downstream under several different LiS options and measures the round trip time for the messages. The Iterations parameter specified the number of iterations that timetst uses in its timing loop, the default being 100,000.

Author

David Grothe dave@gcom.com

3.6 Development

Linux STREAMS (LiS) provides for an interface between STREAMS drivers and the surrounding kernel environment. This interface has grown over time and is likely to expand in the future.

In the Linux kernel, much of the interface between drivers and other kernel modules and the core kernel services, such as memory allocation and synchronization primitives, is implemented in macros and inline functions declared in kernel header files. This technique was used (probably) out of considerations of efficiency (defined as execution speed) and a consideration that there were no version problems with such constructs because one could always recompile one's drivers in the context of the new kernel. The only "kernel primitives" compatibility that has been attempted from one kernel release to the next is source code compatibility.

The real world of paying customers is quite different. And, as it happens, the world of paying customers seems to impinge upon LiS considerably.

In this world, the customers do not want to rebuild the kernel. They don't want to build the kernel at all. They want to install a distribution with a binary kernel that was configured only at install time. They then want to install add-on binary packages, and they expect these packages to operate correctly with their kernel.

When these add-on packages consist of STREAMS based protocol drivers, LiS is usually the only piece of code that is recompiled from source upon installation into the customer's environment. The STREAMS drivers themselves are typically distributed in binary and linked in with LiS. The resulting module is then typically loaded using "modprobe" or some equivalent command.

In these circumstances it is highly desirable for LiS to "buffer" the interface between the STREAMS drivers and the kernel environment. This allows the STREAMS driver writers to deliver smaller binary packages to their customers and minimizes the number of different versions of those packages that must be maintained by the STREAMS driver writers. Ideally, LiS would be able to present a uniform DKI that would support one version of a user's STREAMS driver across all versions of the Linux kernel.

This ultimate goal is probably not achievable, but it is possible to insulate STREAMS drivers from the Linux kernel to a considerable extent. This is possible in part due to the implied DKI of a STREAMS driver. A STREAMS driver most likely will confine itself to the SVR4 types of DKI calls which have syntax and semantics that do not change over time. The main challenges come from the use of constructs, such as PCI configuration and interrupt service routines, that go outside the SVR4 DKI and must use services of the Linux kernel more-or-less directly.

In general, LiS attempts to replace inline functions and macros with actual subroutine calls to perform kernel operations. This allows the STREAMS driver to be compiled once with references to these routines, with the routines themselves being compiled in the context of the specific kernel version at package installation time. Thus, the STREAMS drivers do not have to be sensitive to differences in kernel versions.

3.6.1 Coding STREAMS Applications

This manual is concerned with the include files and compilation techniques for STREAMS application programs. It is not intended to be a tutorial on the subject of writing STREAMS applications. Additional resources are available [17]here for reference material.

3.6.1.1 Header Files

In your STREAMS application program C language source, use the following line to include LiS header files.

     #include <sys/stropts.h>

This will include all of the STREAMS related information that you need for a user level program.

If your application program uses the poll system call then you need to include one or the other of the following lines depending upon the kernel version that the application is intended to run on. For kernel versions in the 2.0 group, use the following in order to include the poll.h from LiS.

     #include <sys/poll.h>

For kernel versions in the 2.2 group, use the following in order to include poll.h from the kernel's source tree.

     #include <linux/poll.h>

3.6.1.2 Compilation Options

When you compile your STREAMS application, put the following compiler option on the gcc (cc) command line for each C language file that contains any of the above include lines.

     -I/usr/include/LiS

3.6.1.3 Linking Options

When you perform the final link of your application using cc or gcc, add the following to the end of your list of files and libraries to be linked. This links in the system call interface routines for LiS.

     -lLiS

This library includes the STREAMS based version of the pipe system call. If you want to use the standard STREAMS library routines, such as getmsg and putmsg, but you want to use the standard Linux pipe system call, use the following instead.

     -lpLiS

3.6.1.4 Other STREAMS Resources

Click [18]here for a list of other locations that you can consult for general information concerning writing STREAMS applications.

3.6.2 LiS SMP Implementation

Beginning with LiS-2.12, LiS makes aggressive use of multiple CPUs in SMP kernels. It is useful for the STREAMS programmer to have some insight into this design in order to know whether, or which, locking techniques must be used in driver code.

3.6.2.1 CPU Scheduling

LiS starts up a kernel thread for each CPU on the system. In the output of a ps display each thread will show as a process with a name such as "LiS-2.12:0". This notation means that an LiS kernel thread is running and is bound to CPU 0 (":0").

LiS maintains a single global list of queues whose service procedures need to be run. A queue is place into this list by calling the function qenable, whether directly or indirectly. A given queue can only be in this list once. The read queue and write queue of a queue-pair are considered two different queues for scheduling purposes and both can be scheduled simultaneously.

At "certain points in time" LiS performs an operation that makes a decision concerning the manner in which service procedures are to be invoked via the list of scheduled queues. There are several factors which influence this decision.

• If the decision is being made just prior to LiS exiting back to user mode from a system call, if the LiS kernel thread for this CPU is inactive and if there are few enough entries in the list of scheduled queues then LiS calls the routine that processes queues directly without waking up any of its kernel threads.
• If the decision is being made from an interrupt routine then the queue processing routine is not to be called directly.
• If the decision is being made from a call on the routine qenable(9) then the queue processing routine is not to be called directly.
• The number of queues in the scheduling list affects the decision making process.
• The number of LiS kernel threads running affects the decision making process.
• Whether or not LiS is executing a system call affects the decision making process. In this case LiS may defer any queue processing action until the system call is about to exit.

In the case that the queue processing routine does not get called directly, LiS needs to decide whether to wake up a kernel thread process or whether to defer queue processing until an LiS system call is about to exit.

LiS tries to enlist one CPU for every four queues that are scheduled. This number is based on considerations of CPU loading and average queue lengths from queueing theory. If the number of CPUs currently processing queues is not enough to meet this target then the scheduling process seeks to enlist more CPUs until the number of CPUs is sufficient to meet this target value. Of course, sometimes there are simply too may queues schedule for processing for the number of available CPUs. In that case, all available CPUs run their queue processing threads.

When making the decision as to whether or not to wake up a kernel thread, LiS gives precedence to the CPU that it is running on. If the scheduling algorithm is called from a point just prior to executing back to the user, and if the kernel thread for the active CPU is sleeping, then LiS will simply call the queue processing routine without waking up the kernel thread. This saves the wakeup and context switch overhead.

The routine that actually runs the queues removes one element at a time from the list of scheduled queues and calls the service procedure pointed to by the queue. The routine continues until the list of scheduled queues is empty. Thus, when a kernel thread is actively processing queues, and if the number of scheduled queues does not exceed the estimated capacity of the running threads, it is quite efficient to simply add a queue to the list and let the already running threads process them in due course.

3.6.2.2 Queue Locking

The queue_t structure in LiS contains a spin lock that is used by LiS to ensure that service procedures are not reentered for the same queue. This lock is not to be used by driver code.

When the LiS queue running routine removes a queue from the list of scheduled queues it acquires this lock prior to calling the service procedure.

LiS also acquires this lock when calling the put procedure associated with a queue. Thus, execution of the put and service procedure are excluded for the same queue.

In a multi-CPU environment, it can happen that one CPU is calling the put procedure while a second CPU is calling the service procedure for the same queue. In this case, one or the other spins until the first CPU finishes the operation and releases the spin lock.

When LiS is about to call the put procedure of a queue from the put or service procedure of a neighbouring queue (because the driver called the putnext function), it continues to hold the lock for the calling queue while acquiring the lock for the destination queue. The locks are acquired sequentially as the chain of putnext calls traverse the stream. The locks are released in reverse order as the put procedures return. This has the effect of incrementally locking the entire stream as messages are passed from one module to another.

This behaviour is only of interest when modules are I_PUSHed on top of a driver. Otherwise, it is just the stream head write queue and the driver write queue that need to be locked (or other pairwise combinations such as the driver read queue and stream head read queue, or queues involving multiplexers).

The lock that LiS uses has the effect of excluding multiple entries from different threads into the put or service procedure for a given queue. The other queue in the queue pair is unaffected by this locking. Therefore, if there are data structures shared between the read and write put and service procedures of a driver or module, it is up to the driver writer to protect these structures with spin locks.

3.6.2.3 Service Procedure Context

Due to the manner in which service procedures are called, sometimes from the LiS queue runner threads and sometimes from a "borrowed" system call, service procedures may or may not have some user context present when they run. Service procedures should always assume that there is no user context. Even in the cases where there is some user context, the identity of the user process is unpredictable.

LiS does, however, maintain a copy of the credentials of the process that opened the stream when it calls service procedures on the stream. LiS saves the user and group identifiers plus the capability masks (credentials) of the running process in the stream head structure at the time that the STREAMS file is opened. These identifiers are restored to the task structure before calling a service procedure on that stream.

When calling put procedures, however, no such identity restoration occurs. So the credentials in place when a driver or module put procedure is invoked are those of the invoking entity. Because the queue runner threads always begin driver entry with a call to the service procedure, entries into the put procedures of subsequent drivers will have the credentials of the stream whose service procedure was called in the first instance. When a driver's put procedure is entered from a system call the credentials will be that of the user process which issued the system call.

3.6.2.4 Scheduling Statistics

LiS gathers statistics on its queue scheduling algorithm. They can be printed out with the command streams -S. The output looks like the following.

     N-CPUs N-Qrunners N-Running N-Requested
          2          2         0          0
     
     CPU   Qsched-Cnts Qsched-ISR Svc-Q-Cnts Qrun-Cnts Active Thread-PID
       0     540752204  175753842  459587537 239611835      0        857
       1     540683832  175833424  459150290 239672683      0        858

The fields have the following meanings.

`N-CPUs'

Number of CPUs on the system.

`N-Qrunners'

Number of queue runner kernel threads. These are the processes that appear as LiS-2.12:0 in a ps display.

`N-Running'

Number of queue runner threads that are currently active.

`N-Requested'

The number of queues that are in the list of scheduled queues.

`CPU'

The remainder of the statistics are kept on a per-CPU basis.

`Qsched-Cnts'

This is the number times the routine (lis_setqsched) that decides whether or not to wake up a queue runner process or to directly process scheduled queues has been called. This routine is called whenever a queue is added to the scheduling list. The counter reflects which CPU made the call to the routine.

`Qsched-ISR'

The number of times lis_setqsched was called from an interrupt routine and from which CPU.

`Svc-Q-Cnts'

The number of callouts to service procedures on a per-CPU basis.

`Qrun-Cnts'

The number of times the routine (queurun) that removes queues from the schedule list was called. This routine does not return until the queue scheduling list is empty. It can be running on multiple CPUs simultaneously. It is typically called from the queue runner threads but can also be called from an LiS system call either just prior to returning to the user or just prior to sleeping on some event such as the arrival of messages at the stream head.

`Active'

Displays as 0 or 1 depending upon whether there is a queue runner thread running on the particular CPU at the time that the statistics were sampled.

`Thread-PID'

The process id of the queue runner thread asiigned to eachCPU.

3.6.3 Operating System Interface Routines

In the file <sys/osif.h>, LiS provides insulation routines for a number of commonly used kernel functions. These functions are used with their Linux kernel names, but those names are redefined in <sys/osif.h> to be subroutine calls on functions that are actually defined in the file osif.c within LiS. The osif.c file is compiled at LiS installation time and is sensitive to kernel version information.

To use this interface, you include the header files that you would normally include to use the kernel functions, and then include <sys/osif.h> after all of the kernel include files. This allows for the redefinition of the names.

The kernel functions provided via <sys/osif.h> are as follows, grouped by type of function.

3.6.4 PCI BIOS Interface

These are routines that utilize or simulate the original PCI BIOS interface of the 2.0 series of kernels. The names of these routines are changed via defines. Use them as if the prototypes were as follows. You can use these routines on 2.2 kernels even though they represent the 2.0 style of interface.

     #if LINUX_VERSION_CODE < 0x020100       /* 2.0 kernel */
     unsigned long pcibios_init(unsigned long memory_start,
                                unsigned long memory_end);
     #else   /* 2.1 or 2.2 kernel * /
     void pcibios_init(void) ;
     #endif
     int pcibios_find_class(unsigned int class_code, unsigned short index,
                            unsigned char *bus, unsigned char *dev_fn);
     int pcibios_find_device(unsigned short vendor, unsigned short dev_id,
                             unsigned short index, unsigned char *bus,
                             unsigned char *dev_fn);
     int pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn,
                                  unsigned char where, unsigned char *val);
     int pcibios_read_config_word(unsigned char bus, unsigned char dev_fn,
                                  unsigned char where, unsigned short *val);
     int pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn,
                                   unsigned char where, unsigned int *val);
     int pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn,
                                   unsigned char where, unsigned char val);
     int pcibios_write_config_word(unsigned char bus, unsigned char dev_fn,
                                   unsigned char where, unsigned short val);
     int pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn,
                                    unsigned char where, unsigned int val);
     const char *pcibios_strerror(int error) ;

3.6.5 PCI Interface

These routines constitute the PCI interface as implemented in the 2.2 series of kernels. Please note that these are filtered calls to the operating system and still depend directly upon the kernel structure "struct pci_dev". LiS provides a more abstract interface to PCI that does not depend upon the direct definition kernel structures. The [61]LiS PCI interface is to be preferred since it provides more insulation against changes in the kernel.

     struct pci_dev *pci_find_device(unsigned int vendor, unsigned int device,
                                     struct pci_dev *from);
     struct pci_dev *pci_find_class(unsigned int class, struct pci_dev *from);
     struct pci_dev *pci_find_slot(unsigned int bus, unsigned int devfn);
     int pci_read_config_byte(struct pci_dev *dev, u8 where, u8 * val);
     int pci_read_config_word(struct pci_dev *dev, u8 where, u16 * val);
     int pci_read_config_dword(struct pci_dev *dev, u8 where, u32 * val);
     int pci_write_config_byte(struct pci_dev *dev, u8 where, u8 val);
     int pci_write_config_word(struct pci_dev *dev, u8 where, u16 val);
     int pci_write_config_dword(struct pci_dev *dev, u8 where, u32 val);
     void pci_set_master(struct pci_dev *dev);

3.6.6 IRQ Interface

These are the routines that are used to attach and detach interrupt service routines to hardware interrupts.

     int request_irq(unsigned int irq,
     void (*handler) ((int, void *, void *), unsigned long flags, const char *device,
                      void *dev_id);
     void free_irq(unsigned int irq, void *dev_id);
     void disable_irq(unsigned int irq);
     void enable_irq(unsigned int irq);

3.6.7 I/O Memory Mapping

These are the routines that are typically used to map PCI bus or physical addresses to CPU virtual addresses. LiS includes some backward compatibility here to older kernel versions.

     void *ioremap_nocache(unsigned long offset, unsigned long size);
     void iounmap(void *addr);
     void *vremap(unsigned long offset, unsigned long size);
     unsigned long virt_to_phys(volatile void *addr);
     void *phys_to_virt(unsigned long addr);

3.6.8 I/O Port Access

These are the routines that allow a driver to register I/O ports.

     int check_region(unsigned int from, unsigned int extent);
     void request_region(unsigned int from, unsigned int extent, const char *name);
     void release_region(unsigned int from, unsigned int extent);

3.6.9 Memory Allocation

These are the kernel routines that can be used to allocate memory. LiS also has a more insulated abstraction for kernel memory allocation. It is recommended that you use the [66]LiS memory allocator versions rather than the direct kernel versions.

     void *kmalloc(size_t nbytes, int type);
     void kfree(const void *ptr);
     void *vmalloc(unsigned long size);
     void vfree(void *ptr);

3.6.10 DMA Routines

These are the routines that are used to allocate a main-board old-style DMA channel for use by your driver. These are not much used anymore. See below for a more elaborate abstraction of DMA routines.

     int request_dma(unsigned int dma_nr, const char *device_id);
     void free_dma(unsigned int dma_nr);

3.6.11 Delay Routines

This is the routine that simply spins the CPU for a given number of microseconds. LiS also redefines the symbol "jiffies" to a subroutine call to help insulate STREAMS drivers from changes in the way the kernel keeps track of time. Remember, the redefinition is accomplished using C language defines, so the following declarations describe the effective usage of these symbols, not their literal definition.

     void udelay(long micro_secs);
     unsigned long jiffies;

3.6.12 Printing Routines

These are the most commonly used printf-like routines in the kernel. STREAMS drivers would be more portable if they used the cmn_err routine instead of printk.

     int printk(const char *fmt, ...);
     int sprintf(char *bfr, const char *fmt, ...);
     int vsprintf(char *bfr, const char *fmt, va_list args);

3.6.13 Timer Routines

These are the the routines that start and stop kernel timers. STREAMS drivers would be more portable if they used the standard "[71]timeout" routine.

     void add_timer(struct timer_list *timer);
     int del_timer(struct timer_list *timer);

The following routine converts time in micro seconds to system "ticks". The "ticks" value is suitable for use with the timeout routine. Note that if the micro_sec parameter is less than the number of micro seconds in a system tick then the routine returns zero.

     unsigned lis_usectohz(unsigned micro_sec);

The following routine is an LiS abstraction of the C library routine gettimeofday. Note the absence of the time zone parameter.

     void lis_gettimeofday(struct timeval *tv);

The following two kernel routines are called via the LiS osif.c code.

     void do_gettimeofday(struct timeval *tp);
     void do_settimeofday(struct timeval *tp);

3.6.14 Sleep and Wakeup Routines

These are the kernel routines for sleeping using wait queues. STREAMS drivers should not be using these since only "open" and "close" routines are allowed to sleep, and for those cases, [73]LiS semaphores would provide better insulation from the kernel. STREAMS "put" and "service" routines should use [74]LiS spin locks for mutual exclusion.

     void sleep_on(OSIF_WAIT_Q_ARG);
     void interruptible_sleep_on(OSIF_WAIT_Q_ARG);
     void wake_up(OSIF_WAIT_Q_ARG);
     void wake_up_interruptible(OSIF_WAIT_Q_ARG);

3.6.15 Thread Creation

A STREAMS driver in LiS can create kernel threads if it so chooses. The following routine simplifies this task. It consolidates all of the kernel manipulations involved with the creation of a kernel thread into one place, thus removing references to these kernel functions from STREAMS driver code.

3.6.15.1 Prototype

     pid_t lis_thread_start(int (*fcn) (void *), void *arg, const char *name);
     int lis_thread_stop(pid_t pid);

Arguments

fcn

The function that is to be used as the entry point for the thread.

arg

The argument passed to the function.

name

An ASCII name associated with the thread. This name should be less than 16 characters in length. It will be the name of the thread that displays in a ps listing.

3.6.15.2 Operation

lis_thread_start creates a new thread, performs some operations prior to entering the fcn, and then calls fcn which acts as the "main" routine for the thread. The arg parameter is passed to fcn.

Before fcn is entered, the newly created thread will have shed all user space files and mapped memory. Thus, it is a kernel-only thread.

All signals are still enabled. Note that when the kernel goes down for reboot all processes are first sent a SIGTERM. Once those have been processed, all processes are then sent a SIGKILL. It is the implementor's choice which of these it pays attention to in order to exit prior to a reboot.

The fcn is entered with the "big kernel lock" NOT held, just as it would be for calling the "kernel_thread" function directly. On 2.2 kernels, the fcn should get this lock so that it can utilize kernel services safely.

The user's fcn returns a value when it exits and that value is returned to the kernel. It is not clear that anything actually pays any attention to this returned value. It particular, it is not visible to the thread that started the new thread.

lis_thread_start itself returns the process id of the new thread, or a negative error number. This value can be used to kill the thread.

lis_thread_stop kills a thread started by lis_thread_start. It returns 0 for success or a negative error number for failure.

3.6.16 Major/Minor Device Numbering

Please note that LiS-2.17 changed the internal representation of the major and minor device numbers within the 32 bit dev_t structure. The following documents the new format and usage conventions. In STREAMS the dev_t structure is used to combine a major device number and a minor device number into a single integer length quantity. The Linux kernel restricts these numbers to the range 0 to 255 (8-bit values).

LiS provides a typedef for dev_t that results in an unsigned integer quantity. Internal to LiS the high order 12 bits are used for major device number and the low order 20 bits are used for minor device number.

STREAMS drivers include the file <sys/stream.h> that causes the view of dev_t to change from the kernel's 8/8 view to the LiS 12/20 view. To ensure proper operation, STREAMS drivers should use the following functions to manipulate dev_t variables. These functions are SVR4 compatible.

int getmajor(dev_t dev);

Extracts the major device number

int getminor(dev_t dev);

Extracts the minor device number

dev_t makedevice(int maj, int min);

Combines a major and minor device number into a dev_t

int DEV_SAME(dev_t d1, dev_t d2);

True if the two devices are the same

int DEV_TO_INT(dev_t dev);

Converts dev_t to an int

The sample drivers that come with LiS now use these constructs to manipulate device structures and can serve as examples for their usage.

Within a STREAMS driver it is occasionally necessary to make a dev_t value in the external 8/8 format. This is required, for example, when a driver is using the lis_mknod() function to create a device node at driver initialization time. LiS provides the function UMKDEV(major, minor) for this purpose.

3.6.17 LiS Memory Allocation

LiS provides for several different styles of memory allocation, all of them insulated from the Linux kernel. These routines allow your driver to allocate memory in several different ways while still maintaining compatibility with different versions of the Linux kernel, with no driver recompilation required.

To use the LiS memory allocation routines include the file <sys/lismem.h> in your STREAMS driver source code.

3.6.18 LiS malloc and free Equivalents

The first group of memory allocation routines are the routines that play the role of "malloc" and "free." These routines keep a master linked list of all allocated memory areas. This list can be printed out via an ioctl to LiS. Each allocated area is tagged with the file name and line number of the code that caused it to be allocated. Each area contains a guard word at the front and back to enable the allocator to detect "off by one" accesses outside the allocated area.

LiS uses this allocator internally for allocating queues, messages and other internal data structures. This would be the allocator of choice for STREAMS drivers to use to allocate instance structures.

Memory allocated in this manner is ultimately allocated by the kernel routine "kmalloc". As such, it is not guaranteed to be DMA-able (in the old style), or to occupy physically contiguous memory locations. [78]See below for routines that can be used to allocate these types of memory areas.

The routines are as follows:

     void *ALLOC(int nbytes);
     void *ALLOCF(int nbytes, char *tag);
     void FREE(void *ptr);

The ALLOC and FREE routines are analogous to "malloc" and "free". The ALLOCF routine includes a character string which is prepended to the file name stored as the location from which the allocation occurred. It can serve as a tag for the type of memory being allocated.

Usage examples:

     ptr = ALLOC(456);
     FREE(ptr);
     ptr = ALLOCF(578, "Instance: ");
     FREE(ptr);

3.6.19 LiS Kernel Memory Allocators

These routines use the LiS malloc/free internal routines to allow for more flexibility in the options used when calling the kernel allocator. These routines all lead to a call on "kmalloc" with appropriate options. It is worth noting that the numerical value of the constants used in calling the kernel's "kmalloc" routine changed between the 2.2 and 2.4 versions of the kernel. Thus, drivers which called the kernel's "kmalloc" directly have to be recompiled to run in a 2.4 kernel. STREAMS drivers using the memory allocation interface defined here could run without modification and without a recompilation on both kernels, assuming that the drivers otherwise did not use any direct kernel functions.

     void *lis_alloc_atomic(int nbytes);
     void *lis_alloc_kernel(int nbytes);
     void *lis_alloc_dma(int nbytes);
     void *lis_free_mem(void *mem_area);

These routines pass the allocation options GFP_ATOMIC, GFP_KERNEL, and GFP_DMA, respectively, to "kmalloc" when allocating the memory. LiS takes care of passing the proper values to the kernel routine so that driver code can remain portable.

The routine lis_free_mem returns a NULL pointer for the convenience of the caller.

The kernel's kmalloc is restricted as to the number of bytes that it will allocate. The LiS routines do not have this restriction. If the number of requested bytes is larger than 16K the LiS allocation routines will call the page allocator to allocate the memory. The lis_free_mem routine knows whether to free pages or to use the kernel's kfree routine.

Usage Examples:

     ptr = lis_alloc_kernel(sizeof(structure));
     ptr = lis_free_mem(ptr);        /* returns NULL pointer */

3.6.20 LiS Page Allocator

These routines allow a STREAMS driver to allocate memory directly from the kernel's page allocator. Memory allocated in this manner occupies physically contiguous locations and is suitable for use with bus master DMA PCI devices.

Unlike the kernel's page allocator, the size that is specified when calling the LiS page allocator is in bytes, not "order", or other encoding of page size. LiS calculates the number of pages based upon the requested size.

Also, LiS does not require you to pass the size of the area when freeing the page.

The routines are as follows:

     void *lis_get_free_pages(int nbytes);
     void *lis_free_pages(void *ptr);

The lis_free_pages routine returns a NULL pointer for the convenience of the caller.

Usage Examples:

     ptr = lis_get_free_pages(1024 * kbytes);
     ptr = lis_free_pages(ptr);

3.6.21 LiS PCI Interface

To assist in the portability of STREAMS drivers across different versions of the Linux kernel, LiS provides an abstraction of the PCI configuration interface. It defines a data structure that is used to describe a PCI device and a set of routines that perform operations on PCI configuration space.

Using these abstractions, a STREAMS driver can be portable from the 2.2 kernel to the 2.4 kernel with no recompilation required. The LiS structures completely hide the kernel data structures and PCI configuration space operations from the STREAMS driver.

To use this interface include the file <sys/lispci.h> in your STREAMS driver source code.

3.6.22 The LiS PCI Device Structure

This structure is distinct from a similar structure which is defined by the Linux kernel, but which differs significantly between the 2.2 and 2.4 kernels. The LiS version of this structure is oriented toward providing just enough information to allow a driver to operate the PCI device, without being concerned about the details of PCI bus topology.

This structure is used to return information to the STREAMS driver concerning devices that meet certain criteria, such as device class or manufacturer device identification.

     #define LIS_PCI_MEM_CNT 12      /* # mem addrs */
     typedef struct lis_pci_dev {
             unsigned bus;                   /* bus number */
             unsigned dev_fcn;               /* device/function code */
             unsigned vendor;                /* vendor id */
             unsigned device;                /* device id */
             unsigned class;                 /* class type */
             unsigned hdr_type;              /* PCI header type */
             unsigned irq;                   /* IRQ number */
             unsigned long mem_addrs[LIS_PCI_MEM_CNT];
             void *user_ptr;                 /* private for user */
     } lis_pci_dev_t;

The bus field contains the bus number on which the device is located. LiS obtains this information from the kernel.

The dev_fcn field contains an encoding of the device number on the bus and the function number within the device that this particular structure pertains to. The pair bus and dev_fcn uniquely identifies a device in the PCI subsystem. Devices can be searched for on the PCI bus by bus number and dev_fcn value (see below).

Given a dev_fcn value, a pair of macros will extract the "device" portion and the "function number" portion from it.

#define LIS_PCI_DEV(devfcn)

Extracts the "device" portion

#define LIS_PCI_FCN(devfcn)

Extracts the "function number" portion

#define LIS_MK_DEV_FCN(dev,fcn)

Put dev and fcn together

Given a device number and a function number, this macro will synthesize a dev_fcn value suitable for use in searching the bus.

The vendor and device fields contain the vendor id (manufacturer code) and the vendor's device identifier for the device. Devices can be searched for on the PCI bus by vendor and device identifier (see below).

The class field contains the class code associated with the device. Devices can be searched for on the PCI bus by class code (see below).

The hdr_type field gives the type information for the PCI configuration space header.

The irq field gives the IRQ number that is assigned to this device. This is the number that is used to attach an interrupt service routine to the device.

The mem_addrs field contains a list of addresses associated with the device. These are raw PCI bus addresses and are not mapped into the address space of the processor. Empty slots contain the value zero.

3.6.23 LiS PCI Search Routines

These routines allow the STREAMS driver to find devices on the PCI bus and obtain a pointer to the lis_pci_dev_t structure for the device.

3.6.23.1 lis_pci_dev_t *lis_pci_find_device(unsigned vendor, unsigned device, lis_pci_dev_t *previous_struct);

Find the device by vendor identification and vendor device identification. By passing in the pointer to the previous structure returned it is possible to find all devices of a given type.

The routine returns NULL if there are no (more) devices for the given vendor and device identifiers.

Usage example:

     lis_pci_dev_t *pcip = NULL;
     while ((pcip = lis_pci_find_device(0x109e, 0x8474, pcip)) != NULL) {
             pcip points to a unique device from this vendor
     }

3.6.23.2 lis_pci_dev_t *lis_pci_find_class(unsigned class, lis_pci_dev_t *previous_struct);

Find the device by class. The usage is similar to lis_pci_find_device in that you can use a pointer to loop through all devices of a given class.

The function returns NULL if there are no (more) devices of the given class.

3.6.23.3 lis_pci_dev_t *lis_pci_find_slot(unsigned bus, unsigned dev_fcn);

Find the device by slot number. If you know the bus number (zero for most simple Intel PC systems) and the dev_fcn, you can obtain the PCI configuration information for that particular "slot". Use the LIS_MK_DEV_FCN macro to synthesize the dev_fcn value from the "device" (slot) number and the function number.

The function returns NULL if there is no device in that slot.

Note that this routine only returns one structure since it is not meaningful to process a list of devices for the same slot.

3.6.24 LiS PCI Configuration Space Routines

The following routines are used to read and write PCI configuration space for a particular device. Configuration space can be accessed by byte, word (16 bit) or dword (32 bit).

Each routine takes a pointer to an lis_pci_dev_t structure as an argument. It also takes an index value which is the byte offset from the base of the configuration space for the device at which the given byte/word/dword is to be read or written.

Care should be exercised when writing to configuration space since many of these values are determined by the PCI BIOS at system boot time.

The lis_pci_set_master routine sets the "bus master DMA" bit for the given device. This is used for devices that perform bus master DMA.

The routines are as follows:

     int lis_pci_read_config_byte(lis_pci_dev_t *dev, unsigned index,
                                  unsigned char *rtn_val);
     int lis_pci_read_config_word(lis_pci_dev_t *dev, unsigned index,
                                  unsigned short *rtn_val);
     int lis_pci_read_config_dword(lis_pci_dev_t *dev, unsigned index,
                                   unsigned long *rtn_val);
     int lis_pci_write_config_byte(lis_pci_dev_t *dev, unsigned index,
                                   unsigned char val);
     int lis_pci_write_config_word(lis_pci_dev_t *dev, unsigned index,
                                   unsigned short val);
     int lis_pci_write_config_dword(lis_pci_dev_t *dev, unsigned index,
                                    unsigned long val);
     void lis_pci_set_master(lis_pci_dev_t *dev);

3.6.25 LiS PCI DMA Routines

These routines are used to allocate memory suitable for use with PCI bus master DMA devices or to map page-allocated memory for those purposes.

To understand what these routines do, please refer to the file /usr/src/linux/Documentation/DMA-mapping.txt in a fairly recent 2.4 kernel source tree. The kernel provides more functionality than is provided in LiS, so there are more routines documented there than are found in this interface. You can use these routines in 2.2 kernels but the functions performed are simply approximations of the 2.4 semantics and may not work in all cases.

Note that the LiS routines have simplified the kernel interface involving "DMA handles" in such a way as to make these constructs easier to use and less error prone.

The following routines are used to allocate memory which the hardware keeps consistent between CPU access and DMA access.

     void *lis_pci_alloc_consistent(lis_pci_dev_t *dev, size_t size,
                                    lis_dma_addr_t * dma_handle);
     void *lis_pci_free_consistent(lis_dma_addr_t * dma_handle);

The following routines are used to obtain a DMA address from a returned DMA handle. You need to know whether or not your hardware environment is using 32-bit or 64-bit DMA addresses.

     u32 lis_pci_dma_handle_to_32(lis_dma_addr_t * dma_handle);
     u64 lis_pci_dma_handle_to_64(lis_dma_addr_t * dma_handle);

The following routines are usd to map page-allocated memory for DMA purposes. The direction indicator of LIS_SYNC_FOR_CPU means that you intend to use the memory for DMA transfers into memory. The direction indicator of LIS_SYNC_FOR_DMA means that you intend to use the memory for DMA transfers out of memory. If the DMA operation goes both ways then use LIS_SYNC_FOR_BOTH.

     void lis_pci_map_single(lis_pci_dev_t *dev, void *ptr, size_t size,
                             lis_dma_addr_t * dma_handle, int direction);
     void *lis_pci_unmap_single(lis_dma_addr_t * dma_handle);
     int lis_osif_pci_map_sg(struct pci_dev *hwdev, struct scatterlist *sg,
     			int nents, int direction);
     void lis_osif_pci_unmap_sg(struct pci_dev *hwdev, struct scatterlist *sg,
     			int nents, int direction);

The direction indicators are as follows:

     LIS_SYNC_FOR_CPU
     LIS_SYNC_FOR_DMA
     LIS_SYNC_FOR_BOTH

With mapped memory, i.e., non-consistent memory, you need to synchronize the memory whenever the CPU writes into it and the DMA needs to read it, or when the DMA has written into it and the CPU needs to read it. The following routine is used for that purpose.

     void lis_pci_dma_sync_single(lis_dma_addr_t * dma_handle, size_t size,
                                  int direction);
     void lis_osif_pci_dma_sync_sg(struct pci_dev *hwdev,
     			      struct scatterlist *sg,
     			      int nelems, int direction);

The following routines can be used at driver initialization time to discover and control the addressing boundary restrictions of a device.

     int lis_pci_dma_supported(lis_pci_dev_t *dev, u64 mask);
     int lis_pci_set_dma_mask(lis_pci_dev_t *dev, u64 mask);

Please consult the file <sys/osif.h> for additional routines that may be present for DMA support.

3.6.26 LiS Atomic Functions

LiS provides for atomic integers implemented in a portable fashion. To declare an LiS portable atomic integer use the following declaration syntax:

     lis_atomic_t myatom;

LiS then provides the following operations on variables of this type.

     void lis_atomic_set(lis_atomic_t *atomic_addr, int valu);
     int lis_atomic_read(lis_atomic_t *atomic_addr);
     void lis_atomic_add(lis_atomic_t *atomic_addr, int amt);
     void lis_atomic_sub(lis_atomic_t *atomic_addr, int amt);
     void lis_atomic_inc(lis_atomic_t *atomic_addr);
     void lis_atomic_dec(lis_atomic_t *atomic_addr);
     int lis_atomic_dec_and_test(lis_atomic_t *atomic_addr);

Of these, only lis_atomic_dec_and_test needs any explanation. This routine performs an atomic_dec on the variable and returns true if the counter reached zero via that decrement operation. Note that by the time the routine returns some other CPU with access to the same variable may have changed its value. So the return reports only on the instantaneous value of the variable.

3.6.27 LiS Locks

LiS provides an abstraction and an insulated interface to the Linux kernel for spin locks, interrupt disabling and semaphores. If you use this interface in your STREAMS driver you can utilize these kernel services on different versions of the Linux kernel without the necessity of recompiling your driver for each version of the kernel.

The LiS locks are especially useful in consideration of Linux kernels compiled with and without the SMP option set. The spin locks and semaphores of the Linux kernel are implemented using external inline functions. These functions are coded in assembly language and generate different sequences of instructions depending upon the compile time setting of the SMP option. Spin locks and semaphores compiled with SMP reset will not function properly on a multi-CPU system running an SMP kernel.

The LiS locks mechanism solves this problem by abstracting the locking primitives into actual subroutines, not inlines, defined within LiS. Since LiS is compiled from source code when it is installed the subroutines in LiS have the correct setting of SMP for the locking primitives. This allows the STREAMS driver code to be compiled once and the object code reused for multiple installations with varying options.

The following sections document the spin locks, interrupt disabling and semaphore mechanisms offered by LiS. To use these mechanisms include the file <sys/lislocks.h> in your STREAMS driver source code.

In choosing the appropriate type of lock to use, one must bear in mind that STREAMS drivers are not allowed to "sleep" in "put" and "service" procedures, only in "open" and "close" routines. That means that spin locks are the mutual exclusion mechanism of choice for "put" and "service" procedures. It is reasonable to use sleeping semaphores in "open" and "close" routines.

The simple interrupt exclusion mechanism can be used to exclude only interrupt routine execution for a section of code. However, this mechanism does not exclude other "put" or "service" procedures that may be executed on other CPUs. This may not be much of a consideration since LiS acquires a lock in the queue structure before executing the "put" or "service" procedure pointed to by that queue.

However, it could happen that the "read put/service" and "write put/service" procedures get executed simultaneously since there are two different locks in the STREAMS queues, one in the read queue and one in the write queue. In this case, the STREAMS driver code would need to use spin locks to protect data structures shared between the read and write "put" or "service" procedures. See the `qlock' option for strconf for more information about LiS implicit use of locks to protect put and service procedure entries.

3.6.28 LiS Spin Locks

LiS provides an implementation of spin locks that utilizes the Linux kernel's spin lock mechanism to perform the actual locking functions. The LiS implementation adds features to the kernel spin locks such as the following:

• LiS spin locks are nestable. The same thread can acquire the same lock and release it in nested fashion.
• LiS spin locks are more debuggable. The LiS lock structure contains an ASCII name for the lock which makes it easier to identify in debugging situations.
• LiS maintains a lock trace table. A debugging option for LiS causes it to log all spin lock operations to a trace table which can be printed out via an option to the streams command.
• LiS spin locks are portable. A STREAMS driver can utilize the same LiS lock mechanism across different versions of the Linux kernel. This pushes the kernel differences into LiS and out of the STREAMS driver code.
• LiS spin locks are documented. You don't have to read the kernel source code to figure out how to use them.

For these reasons I highly recommend that STREAMS drivers use the LiS spin lock implementation in place of the direct kernel spin locks. The portability aspect of LiS spin locks cannot be overemphasized. Different Linux kernel compile-time options can lead to a proliferation of STREAMS driver code versions, or the necessity of always compiling the driver from source when it is installed. LiS spin locks allow a STREAMS driver to be compiled independently of kernel options with only the binary needed at driver installation time.

To declare a spin lock, use the typedef lis_spin_lock_t, as in the following:

     lis_spin_lock_t mylock;

LiS spin locks must be initialized before they are used. There is one initialization routine no matter which style of locking you intend to use.

void lis_spin_lock_init(lis_spin_lock_t *lock, const char *name) ;

This routine initializes the spin lock and associates an ASCII string name with it. The pointer name is saved in the lock structure for later use in printing out the lock trace table. It is the caller's responsibility to ensure that the name resides in memory that will persist for the duration of the existence of the lock.

You can also use dynamically allocated spin locks. This technique allows your STREAMS driver to be completely immune from changes in kernel version regarding the size of a spin lock since your driver only has to store a pointer to the allocated lock. The allocation and deallocation routines are as follows.

     lis_spin_lock_t *lis_spin_lock_alloc(const char *name);
     lis_spin_lock_t *lis_spin_lock_free(lis_spin_lock_t *lock, const char *name);

The allocation function returns a pointer to the spin lock, or NULL if the memory could not be allocated. The free function returns a NULL pointer for the convenience of the caller.

For further information on spin locks, see the section on [89]debugging spin locks.

To lock and unlock a spinlock, use any of the following pairs of routines. If you use the first routine to lock the spin lock then be sure to use its companion unlock routine. For nesting considerations, [91]see below.

     void lis_spin_lock(lis_spin_lock_t *lock);
     void lis_spin_unlock(lis_spin_lock_t *lock);
     int lis_spin_trylock(lis_spin_lock_t *lock);

These routines are to be called only from background processing to lock and unlock a spin lock. The trylock routine locks the spin lock if it is available, returning "true", or leaves it unlocked if it is unavailable, returning "false".

Background processing means any STREAMS driver processing that does not occur at interrupt time. These routines lock the lock but do not exclude interrupt routines from execution. Thus, your interrupt service routine can still be called whether or not your driver is holding a spin lock that was locked with one of these routines.

You can nest pairs of calls to these routines from the same thread of execution. [92]See below for more information on lock nesting.

Usage example:

     lis_spin_lock(&mylock);
     ...
     lis_spin_unlock(&mylock);

     void lis_spin_lock_irq(lis_spin_lock_t *lock);
     void lis_spin_unlock_irq(lis_spin_lock_t *lock);

This pair of routines locks the spin lock with interrupts disabled for the duration of the holding of the lock. The routine lis_spin_lock_irq re-enables interrupts after unlocking the lock.

You can use this technique to exclude interrupt routine execution. However, it is not advisable for interrupt routines themselves, or any routines called from an interrupt routine, to use this mechanism since the unlock primitive unconditionally enables interrupts, which may not be desirable from inside an interrupt routine.

These routines may be used in nested fashion. Only the outermost unlock routine will actually enable interrupts. [94]See below for more information about lock nesting.

Usage example:

     lis_spin_lock_irq(&mylock);
     ...
     lis_spin_unlock_irq(&mylock);

     void lis_spin_lock_irqsave(lis_spin_lock_t *lock, int *flags);
     void lis_spin_unlock_irqrestore(lis_spin_lock_t *lock, int *flags);

This pair of routines is similar to the "spin_lock_irq" routines in that the locking routine disables interrupts. However, it saves the interrupt state in the integer argument whose pointer is passed to the locking routine. The unlock routine then restores the interrupt state after unlocking the lock.

These routines are suitable for use by routines that are called both from interrupt level and from background. They also have the effect, when used in an interrupt routine, of excluding multiple execution of an interrupt routine on multiple CPUs in an SMP system.

These routines may be used in nested fashion. Only the outermost unlock routine will actually restore the interrupt state. [96]See below for more information about lock nesting.

Usage example:

     lis_spin_lock_t mylock;
     int flags;
     lis_spin_lock_irqsave(&mylock, &flags);
     ...
     lis_spin_unlock_irqrestore(&mylock, &flags);

Note that the unlock routine is passed the address of the flags just as in calling the lock routine.

3.6.29 Lock Nesting

LiS spin locks can be locked and unlocked in nested fashion. When doing so, it is always best to use the same pair of lock and unlock routines at all levels of nesting for the same lock. Mixing different types of locking can lead to unexpected results and non-portable behaviour.

LiS allows a single thread to lock spin locks in nested fashion. That is, the second and subsequent calls to the lock routine from a single thread will not spin on the lock because of finding it in a locked state from the first call. Also, every unlock call except the last one, the one that balances the first locking call, does not unlock the lock. Only the outermost unlock call causes the lock to be unlocked.

If the nesting is via lis_spin_lock_irq, then only the outermost unlock call enables interrupts. If the nesting is via lis_spin_lock_irqsave, then only the outermost unlock call restores the interrupt state.

When two or more threads attempt to lock a spin lock "simultaneously" only one thread is allowed to proceed at a time. The other threads "spin", that is, the CPUs executing the other threads are executing a loop that tests the lock repeatedly until it becomes available. Consequently, it is advisable to use locks to protect the execution of fairly short pieces of code if there is any likelihood of contention for the lock. While one thread is holding the lock, other CPUs may be idling waiting for it.

In the context of locking, "simultaneously" means any time from the moment of the first thread locking the spin lock until that thread unlocks the lock. If another thread attempts to lock the spin lock at any point in that interval then it will "spin."

When multiple threads use multiple spin locks to protect multiple resources, it is always a good idea if all threads execute "lock" operations on the multiple spin locks in the same order. It is also highly recommended that they execute "unlock" operations in the exact reverse order as the "lock" operations. This avoids so-called "deadly embrace" situations in which process A acquires spin lock A, process B acquires spin lock B, and then process A waits on B while process B waits on A.

3.6.30 LiS Read/Write Locks

LiS offers an abstraction of the kernel's read/write locks. The LiS abstractions allow STREAMS drivers to use these locks without concern for changes that occur from one version of the kernel to the next.

A read/write lock is declared as a special data object of type lis_rw_lock_t. There are two types of routines to manipulate these locks. One set operates on the lock as a "read" lock. The other set operates on the lock as a "write" lock.

There can be multiple threads owning the lock in read mode. There can only be one thread that owns the lock in write mode. Furthermore, in order to acquire the lock in write mode, all the owners of the read mode lock must give it up.

The locks are used in the obvious way. If you only need to read the protected structure you use the read lock routine. If you need to change the structure you use the write lock routine.

Note that once you have a read lock you must give it up in order to get the same lock as a write lock.

The lock manipulation routines also allow for "regular", "irq" and "irqsave" manipulations of the read/write locks, just as with spin locks.

You must initialize your lock before using it, just as with spin locks. And in parallel to spin locks LiS provides two initialization routines. One operates directly on the read/write lock, and the other allocates memory dynamically for the lock. You can deallocate the dynamically allocated lock by calling the "free" routine.

The following is a listing of the read/write lock routines in LiS. The prototypes are in the file <sys/lislocks.h>.

     void lis_rw_read_lock(lis_rw_lock_t *lock);
     void lis_rw_write_lock(lis_rw_lock_t *lock);
     void lis_rw_read_unlock(lis_rw_lock_t *lock);
     void lis_rw_write_unlock(lis_rw_lock_t *lock);
     
     void lis_rw_read_lock_irq(lis_rw_lock_t *lock);
     void lis_rw_write_lock_irq(lis_rw_lock_t *lock);
     void lis_rw_read_unlock_irq(lis_rw_lock_t *lock);
     void lis_rw_write_unlock_irq(lis_rw_lock_t *lock);
     
     void lis_rw_read_lock_irqsave(lis_rw_lock_t *lock, int *flags);
     void lis_rw_write_lock_irqsave(lis_rw_lock_t *lock, int *flags);
     void lis_rw_read_unlock_irqrestore(lis_rw_lock_t *lock, int *flags);
     void lis_rw_write_unlock_irqrestore(lis_rw_lock_t *lock, int *flags);
     
     void lis_rw_lock_init(lis_rw_lock_t *lock, const char *name);
     lis_rw_lock_t *lis_rw_lock_alloc(const char *name);
     lis_rw_lock_t *lis_rw_lock_free(lis_rw_lock_t *lock, const char *name);

3.6.31 LiS Interrupt Enable/Disable

LiS provides primitives for enabling and disabling interrupts modelled after the SVR4 SPL mechanism. There is one routine that is used to disable interrupts and another one for enabling interrupts. The routines are as follows:

     int lis_splstr(void);
     void lis_splx(int x);

The lis_splstr routine is used to disable interrupts. It returns a value that must be passed to lis_splx when it it desired to restore the interrupt level to its previous state. These two routines are implemented using the primitives lis_spin_lock_irqsave and lis_spin_unlock_irqrestore.

These routines can be used from background code ("put" and "service" procedures, or "open" and "close" routines), or from interrupt level. LiS itself uses these routines to protect STREAMS structures from ill-timed modification by interrupt routines. Many LiS utility routines, such as putq, getq and qenable, call these routines within themselves.

It is safe, and occurs frequently, to use these routines in a nested fashion. When using these routines in a nested fashion be sure that the value returned by the call to lis_splstr at level n is the value passed back to lis_splx at level n. The nesting rules for these routines are otherwise the same as for the pair lis_spin_lock_irqsave and lis_spin_unlock_irqrestore.

Usage examples:

     int x, y;
     x = lis_splstr();
     ...
     y = lis_splstr();
     ...
     lis_splx(y);
     ...
     lis_splx(x);

For further information on these routines see the section on [100]debugging spin locks.

3.6.32 LiS Semaphores

LiS provides an implementation of semaphores that is built upon the Linux kernel's semaphores. The LiS implementation adds features to the kernel semaphores such as the following:

• LiS semaphores are more debuggable. The LiS semaphore structure contains fields that save the file name and line number of the semaphore owner. This makes it easier to debug drivers which utilize semaphores.
• LiS semaphores retain error information. When a "down" operation fails, LiS saves the error number in the semaphore structure for post mortem analysis.
• LiS semaphores are portable. A STREAMS driver can utilize the same LiS semaphore mechanism across different versions of the Linux kernel. This pushes the kernel differences into LiS and out of the STREAMS driver code.
• LiS semaphores can be easily allocated dynamically so your driver is completely immune from Linux kernel version considerations.
• LiS semaphores are documented. You don't have to read the kernel source code to figure out how to use them.

For these reasons I highly recommend that STREAMS drivers use the LiS semaphore implementation in place of the direct kernel semaphores. The portability aspect of LiS semaphores cannot be overemphasized. Different Linux kernel compile-time options can lead to a proliferation of STREAMS driver code versions, or the necessity of always compiling the driver from source when it is installed. LiS semaphores allow a STREAMS driver to be compiled independently of kernel options with only the binary needed at driver installation time.

To declare an LiS semaphore, use a declaration similar to the following:

     lis_semaphore_t mysem;

LiS semaphores must be initialized before they are used. Use the following routine to initialize a declared semaphore.

     void lis_sem_init(lis_semaphore_t *, int);

If you initialize the semaphore to 0, then the first "down" operation on the semaphore will wait. If you initialize it to 1, then the first "down" operation will not wait. If you initialize it to n, then the first n "down" operations will not wait.

You can also allocate semaphores dynamically using the following routine.

     lis_semaphore_t *lis_sem_alloc(int);

This routine uses the kernel's memory allocator to allocate space for the semaphore. The lis_sem_destroy routine will deallocate it for you. The advantage of using this routine is that your STREAMS driver only has to have a pointer to the semaphore, not a semaphore structure itself. This adds an extra level of protection of your driver from kernel version considerations.

You can use the semaphore value to manage a pool of resources by initializing a semaphore to the number of items in the resource and having a driver open routine perform a "down" operation on the semaphore. This causes the open operations to be queued until the resource is available.

LiS semaphores should be explicitly destroyed when they are no longer needed, typically from your STREAMS driver close routine. This operation is accomplished via the following routine.

     lis_semaphore_t *lis_sem_destroy(lis_semaphore_t *,int);

This routine returns a NULL pointer for the convenience of the caller.

For further information on semaphores, see the section on [102]debugging semaphores.

The following two routines are used to acquire and release a semaphore.

     int lis_down(lis_semaphore_t *sem);
     void lis_down_nosig(lis_semaphore_t *lsem);
     void lis_up(lis_semaphore_t *sem);

The routine lis_down returns 0 for success and a negative error code for failure. The caller has not acquired the semaphore unless the routine returns zero.

One reason for a negative return could be that the calling task was signalled while waiting for the semaphore to become available. If this has occurred the return code will be set to -EINTR.

The function lis_down_nosig waits for the semaphore with signals blocked. It is useful in driver close routines that must use a semaphore to control access to the structures that need to be deallocated. It is common for the driver close routine to be called from a process that has been signalled – for example a process that was killed with a <Ctrl-C> from the keyboard. In this case, lis_down will return immediately with -EINTR, an undesirable situation. using lis_down_nosig in this situation blocks signals so that the close routine can wait on the semaphore even if the process has been signalled.

Semaphores cannot be used in nested fashion. Care must be exercised that a single thread only performs one "down" operation on a given semaphore.

When multiple threads use multiple semaphores to protect multiple resources, it is always a good idea if all threads execute "down" operations on the multiple semaphores in the same order. It is also highly recommended that they execute "up" operations in the exact reverse order as the "down" operations. This avoids so-called "deadly embrace" situations in which process A acquires semaphore A, process B acquires semaphore B, and then process A waits on B while process B waits on A.

Semaphores should be used only in STREAMS driver "open" and "close" routines. STREAMS driver "put" and "service" procedures are not allowed to sleep. They should use spin locks instead of semaphores.

Usage example:

     if (lis_down(&mysem) == 0) {
             ...
             lis_up(&mysem);
     }

3.6.33 Debugging Spin Locks

LiS spin lock structures contain fields that assist in the debugging of spin-lock related problems. The LiS spin lock structure contains the following fields.

Field Description

spin_lock_mem An opaque memory area that contains the kernel's spin lock structure.

name

Pointer to an ASCII name for the lock. This allows one to readily identify the function of the lock (assuming that it is aptly named).

taskp

A (void *) which is really a (struct task_struct *) pointer. It points to the task that originally acquired the lock, or is NULL if no task has acquired the lock. spinner_file, spinner_line

File and line number of the most recent call to one of the lis_spin_lock functions. This tells which line of code most recently tried to get the lock.

owner_file, owner_line

File and line number of the call to one of the lis_spin_lock functions that first acquired the lock. These fields are set at the same time as the taskp field.

unlocker_file, unlocker_line

File and line number of the call to one of the lis_spin_unlock functions that performed the final unlock on the lock, thus making it available for another thread. These fields are set at the same time as the taskp field is set to NULL.

If a thread owns the lock then its value of the current task pointer will be in taskp. If there is no other thread spinning on the lock, and if the lock has not been acquired in a nested fashion, then the spinner and owner fields will indicate the same file and line number.

If the spinner and owner fields are different and if the taskp is non-NULL then if the thread that most recently called one of the lis_spin_lock routines is different from the task that owns the lock, then that other task is spinning on the lock. By examination of the lock you can see which task owns the lock and where in the code it was acquired. This is often enough information to figure out why a deadlock is occurring.

A "deadly embrace" occurs when two threads each need to acquire two spin locks but they acquire them in the opposite order from each other. Under circumstances of contention each process owns the lock that the other is spinning on and will not release the lock until it acquires the other lock. Thus, both threads spin forever.

Note that the LiS splstr and splx functions are written in terms of LiS spin locks. LiS does not use these routines internally. They are provided to the user for backward compatibility. However, it is important to know that these routines are spin locks in disguise. This means that the order of use of these functions mixed in with explicit spin lock manipulations may also lead to deadly embraces.

An effective technique for troubleshooting these kinds of problems is to use the two-machine kernel debugger, [105]kgdb. With this setup you can break into the target machine and look at memory using high level debugging techniques, including printing out of structures. Using kgdb you can find out where each CPU is executing, look at the corresponding source code lines, observe the locks that are involved, and then print out the lis_spin_lock_t structures for the specific locks. Oftentimes the information contained in the two locks will immediately reveal the nature of the deadly embrace.

It is also possible to have LiS trace all lock and semaphore operations. One of the LiS debug bits enables this function. To set this debug bit use the following command.

     streams -d0x0x80000

This causes LiS to make entries in a global trace buffer named lis_spl_track. The global pointer lis_spl_track_ptr indicates the next location in the table into which an entry is to be placed, which means that it points to the oldest entry in the buffer. Entries in the buffer are of type spl_track_t.

The fields of this structure are as follows. Field Description type The type of entry as follows.

Value

Meaning

1

splstr

2

splx

3

spin lock

4

spin unlock

5

semaphore down

6

semaphore up

cpu

The cpu number of the processor which made this entry. addr The address of the spin lock or semaphore involved in the operation.

tskp The task pointer for the task that made this entry. state Nesting value for spin locks, count field of the semaphore.

file, line

File and line number of the call to the LiS locking or semaphore routine that caused this entry to be made.

The trace buffer contains 4096 of these entries, maintained in a circular fashion. By printing out these entries you can see the history of lock manipulation within LiS. The command streams -p causes LiS to print out this table from within the kernel. The resulting output can be found in /var/log/messages (typically). However, in practise the system is usually hung when you need this information so you end up printing it from within the debugger.

3.6.34 Lock Semaphore and Queue Contention

The streams command can be used to enable the tracking of contention for locks, semaphores and STREAMS queues. Use the command `streams -D0x08' to enable the contention tracking. The command `streams -L' then causes the contention tables to be printed out.

Locks and semaphores are in contention when a thread goes to spin on a lock or perform a down function on a semaphore, and the thread has to wait because the lock or semaphore is owned by another thread. LiS counts such occurrences on a per-lock basis and reports the results with the `streams -L' command.

Queues are in contention when the semaphore that controls access to the queue is in contention. However, there are options that affect which semaphore is used to control access to a queue and these options will also have an effect on the reporting of queue contention.

3.6.35 Debugging Semaphores

LiS semaphore structures contain fields that assist in the debugging of semaphore related problems. The LiS semaphore structure contains the following fields.

Field Description

sem_mem An opaque memory area that contains the kernel's semaphore structure.

taskp

A (void *) which is really a (struct task_struct *) pointer. It points to the task that most recently acquired the semaphore, or is NULL if no task has acquired the semaphore. The taskp is set to NULL just prior to calling the kernel's up routine on the semaphore. Thus it stays NULL if no other task is pending on the semaphore. downer_file, downer_line File and line number of the most recent call to the lis_down function. This tells which line of code most recently tried to get the semaphore.

owner_file, owner_line File and line number of the call to the lis_down function that acquired the semaphore. These fields are set at the same time as the taskp field.

upper_file, upper_line File and line number of the call to the lis_up function. These fields are set at the same time as the taskp field is set to NULL.

If the taskp field is non-NULL then the semaphore is owned by the task so indicated. If it is NULL then the semaphore is unowned. The upper fields show where the semaphore was last released.

If the downer and owner fields both indicate the same file and line number then that is an indication that the semaphore was acquired at that location in the program. If they are different, and if the taskp is non-NULL, that is an indication that there is a task waiting on the semaphore at the downer location. The owner fields show where the semaphore was acquired.

Bear in mind that semaphore acquisitions do not nest as is the case with spin locks. Therefore, if the same thread calls lis_down without calling lis_up on the same semaphore then the thread will be deadlocked. The downer and owner fields will usually offer a clue to this type of deadlock.

You can also use the LiS [107]lock trace buffer mechanism to assist in debugging semaphore usage.

3.6.36 STREAMS Utility Routines

The following routines are available to LiS STREAMS drivers. These are standard AT&T SVR4 utility routines. They (hopefully) have the same semantics in LiS as they do in SVR4 STREAMS.

These routines are presented here in alphabetical order with no description. Please refer to the [109]AT&T SVR4 STREAMS documentation for the descriptions of these routines.

3.6.37 Freezing Streams

There are two sets of routines that can be used to `freeze' a stream. They are used in slightly different ways and have slightly different semantics. One set uses the routines freezestr() and unfreezestr(); the other set uses the routines qprocesoff() and qprocson().

3.6.37.1 Freezestr and Unfreezestr

     void freezestr(queue_t *q);
     void unfreezestr(queue_t *q);

These routines operate on the entire stream of which the queue is a member. The stream is found by traversing the chain of queues in both directions until encountering a queue that is not linked to another queue. As a simple example it includes all queues from the stream head down through any pushed modules to the driver queue in which one of those queues is the one passed as the parameter to either of these routines.

The process of freezing the stream is to place it into a state such that messages will not flow up and down the stream. That is put and service procedures will not be called. It putnext() is called on a queue within a frozen stream the passed message is placed into a special deferred message list. Messages are removed from this list and passed to the put procedure when the stream is later unfrozen.

Drivers that have frozen a stream should refrain from performing queueing operations on queues within the stream, such as getq and putq. LiS does not enforce this so one must exercise some care when using these routines.

SVR4 STREAMS specification says that the driver's close routine will not be called if the stream is frozen. LiS does not implement this rule and will close a frozen stream.

LiS uses these routines internally at stream close time to stop message flow when the stream is being dismantled. It also uses them during I_PUSH and I_POP processing to inhibit message flow while replumbing the stream.

Drivers should use these routines with some caution. Because the stream is frozen the driver cannot receive any messages from above or below, including M_IOCTL. This may make it tricky deciding when to unfreeze a stream.

3.6.37.2 Qprocsoff and Qprocson

     void qprocson(queue_t *rdq);
     void qprocsoff(queue_t *rdq);

These routines are conventionally used in a driver open (qprocson) and close (qprocsoff) routine. In some STREAMS implementations qprocson must be called to enable messages to flow into the queue once open processing has completed. This is not necessary in LiS.

In LiS it does no harm to call qprocson in the driver open routine and qprocsoff in the driver close routine, though it is not necessary to do so.

The effect of qprocsoff is similar to that of freezestr except that it applies just to the single queue rather than to the entire stream. Once significant difference is that if a pushable modules is in a `qprocsoff' condition and a message flows into the module, STREAMS will route the message to the next module or driver in the chain of queues, looking for one that is enabled. If no such module or driver exists, the message will be placed into the deferred message list of the queue at the far end of the chain of queues. The messages will be presented to the driver put routine when qprocon is called.

It is best to use these routines only at open and close time since that seems to have been the intent of the STREAMS designers.

3.6.38 Flushing Queue Bands

A special note on flushing queue bands is in order. The rules for flushing queues are a bit complex, so we wish to review them here in some detail.

First some definitions and some things that affect all queue flushing. The term "data message" in the context of queue flushing means messages of type M_DATA, M_PROTO, M_PCPROTO or M_DELAY. All other message types are considered "non-data messages". You may find it less than intuitive that M_PCPROTO is considered a "data message".

The term "ordinary message" in the context of queue flushing means messages of type M_DATA, M_PROTO, M_BREAK, M_CTL, M_DELAY, M_IOCTL, M_PASSFP, M_RSE, M_SETOPTS or M_SIG. Please note that M_PCPROTO is not on this list.

The flag argument of FLUSHDATA means that only "data messages" are to be flushed. The flag argument of FLUSHALL means that "all" messages are to be flushed. As we shall see, in flushing queue bands whether a message gets flushed or not depends upon what the meaning of the word "all" is.

First, let's take the case of the routine flushq(q,flag). If flag is set to FLUSHDATA then all "data messages" in the entire queue, including all queue bands, are flushed. If the flag is set to FLUSHALL then the entire queue is flushed.

The case of the routine flushband(q,band,flag) is more complicated.

If the band argument is zero then special rules apply. In this case, only "ordinary" messages are flushed from the queue. The value of the flag parameter does not influence the operation. In Solaris STREAMS this behaviour does not occur. They flush either "data messages" or "all" messages on band zero. Comments in the Solaris 8 source code indicate that the author of the flush code was somewhat confused on this point.

If the band argument is non-zero then the specific band of the queue is flushed in a manner similar to that of flushq. That is, the flag argument of FLUSHDATA means just flush "data messages" and the value of FLUSHALL means flush "all" messages from the specific band.

One further item needs some attention. Whenever an M_PCPROTO (or other "high priority") message is inserted into a STREAMS queue it is queued ahead of all messages in any queue band. This means that an M_PCPROTO cannot be directed to a queue band. It also means that flushband can never flush an M_PCPROTO, or any other "high priority" message from the queue. In order to flush M_PCPROTOs you must call flushq and flush the entire queue of either "data messages" or "all" messages.

3.6.39 Utility Prototypes

     int adjmsg(mblk_t *mp, int length);
     struct msgb *allocb(int size, unsigned int priority);
     __________________________________________
     
     queue_t *backq(queue_t *q);
     int bcanput(queue_t *q, unsigned char band);
     int bcanputnext(queue_t *q, unsigned char band);
     void bcopy(void *src, void *dst, int nbytes);
     int bufcall(unsigned size, int priority, void (*function) (long), long arg);
     void bzero(void *addr, int nbytes);
     __________________________________________
     
     int canput(queue_t *q);
     int canputnext(queue_t *q);
     void cmn_err(int err_lvl, char *fmt, ...);
     mblk_t *copyb(mblk_t *mp);
     mblk_t *copymsg(mblk_t *mp);
     __________________________________________
     
     #define datamsg(type) -- true if msg->b_datap->db_type is data
     mblk_t *dupb(mblk_t *mp);
     mblk_t *dupmsg(mblk_t *mp);
     __________________________________________
     
     void enableok(queue_t *q);
     mblk_t *esballoc(unsigned char *base, int size, int priority, frtn_t *freeinfo);
     int esbbcall(int priority, void (*function) (long), long arg);
     __________________________________________
     
     void flushband(queue_t *q, unsigned char band, int flag);
     void flushq(queue_t *q, int flag);
     void freeb(mblk_t *bp);
     void freemsg(mblk_t *mp);
     void freezestr(queue_t *q);
     void unfreezestr(queue_t *q);
     __________________________________________
     
     int getmajor(dev_t dev);
     int getminor(dev_t dev);
     mblk_t *getq(queue_t *q);
     __________________________________________
     
     int insq(queue_t *q, mblk_t *emp, mblk_t *mp);
     __________________________________________
     
     void *kmem_alloc(int siz, int wait_code);
     void *kmem_zalloc(int siz, int wait_code);
     void kmem_free(void *ptr, int siz);
     __________________________________________
     
     void linkb(mblk_t *mp1, mblk_t *mp2);
     __________________________________________
     
     int msgdsize(mblk_t *mp);
     mblk_t *msgpullup(mblk_t *mp, int length);
     int msgsize(mblk_t *mp);
     __________________________________________
     
     void noenable(queue_t *q);
     __________________________________________
     
     queue_t *OTHERQ(queue_t *q);
     __________________________________________
     
     int pullupmsg(mblk_t *mp, int length);
     int putbq(queue_t *q, mblk_t *mp);
     int putctl(queue_t *q, int type);
     int putctl1(queue_t *q, int type, int param);
     void putnext(queue_t *q, mblk_t *mp);
     int putnextctl(queue_t *q, int type);
     int putnextctl1(queue_t *q, int type, int param);
     int putq(queue_t *q, mblk_t *mp);
     __________________________________________
     
     void qenable(queue_t *q);
     void qreply(queue_t *q, mblk_t *mp);
     int qsize(queue_t *q);
     void qprocsoff(queue_t *rdq);
     void qprocson(queue_t *rdq);
     __________________________________________
     
     queue_t *RD(queue_t *q);
     queue_t *WR(queue_t *q);
     queue_t *OTHERQ(queue_t *q);
     mblk_t *rmvb(mblk_t *mp, mblk_t *bp);
     void rmvq(queue_t *q, mblk_t *mp);
     __________________________________________
     
     int SAMESTR(queue_t *q);
     int strqget(queue_t *q, qfields_t what, unsigned char band, long *val);
     int strqset(queue_t *q, qfields_t what, unsigned char band, long val);
     __________________________________________
     
     int testb(int size, unsigned int priority);
     __________________________________________
     
     #define HZ -- ticks per second
     typedef void timo_fcn_t (caddr_t arg);
     toid_t timeout(timo_fcn_t *timo_fcn, caddr_t arg, long ticks);
     toid_t lis_untimeout(toid_t id);
     __________________________________________
     
     void unbufcall(int bcid);
     mblk_t *unlinkb(mblk_t *mp);
     int untimeout(int id);
     __________________________________________
     
     queue_t *WR(queue_t *q);
     __________________________________________
     
     int xmsgsize(mblk_t *mp);

3.6.40 System Calls from within the Kernel

LiS provides STREAMS drivers with a few system calls that can be made from within the kernel. These calls are intended to allow STREAMS drivers to manage their device special files through which the drivers are accessed. For example, by using the lis_mknod function a dynamically loaded driver can register itself with LiS, obtain a major device number and make its "/dev" entries at module load time. Using the lis_unlink function it can remove these "/dev" entries when the module unloads.

The semantics of the following routines are exactly the same as the user level routines of the same names without the "lis_" prefix. This is so because these routines are really just wrappers on a kernel system call. We list the function prototypes here but leave the detailed documentation to "man pages" and other documentation.

The following function prototypes exist in the file <sys/dki.h>.

     int lis_mknod(char *name, int mode, dev_t dev);
     int lis_unlink(char *name);
     int lis_mount(char *dev_name, char *dir_name, char *fstype,
                   unsigned long rwfla g, void *data);
     int lis_umount(char *file, int flags);

4 Conformance

4.1 STREAMS Compatibility

Linux STREAMS (LiS) provides some degree of compatibility with other STREAMS implementation as follows:

— SVR 4.2 ES/MP

Linux STREAMS (LiS) provides some degree of operational compatibility with SVR 4.2 ES/MP to ease portability and common comprehension, see SVR 4.2 Compatibility.

— AIX 5L Version 5.1

Linux STREAMS (LiS) provides some degree of operational compatibility with AIX 5L Version 5.1 to ease portability and common comprehension, see AIX Compatibility.

— HP-UX 11.0i v2

Linux STREAMS (LiS) provides some degree of operational compatibility with HP-UX 11.0i v2 to ease portability and common comprehension, see HP-UX Compatibility.

— OSF/1 1.2/Digital UNIX/True 64

Linux STREAMS (LiS) provides some degree of operational compatibility with OSF/1 1.2/Digital UNIX to ease portability and common comprehension, see OSF/1 Compatibility.

— UnixWare 7.1.3 (OpenUnix 8)

Linux STREAMS (LiS) provides some degree of operational compatibility with UnixWare 7.1.3 (OpenUnix 8) to ease portability and common comprehension, see UnixWare Compatibility.

— Solaris 9/SunOS 5.9

Linux STREAMS (LiS) provides some degree of operational compatibility with Solaris 9/SunOS 5.9 to ease portability and common comprehension, see Solaris Compatibility.

— SUPER-UX

Linux STREAMS (LiS) provides some degree of operational compatibility with SUPER-UX to ease portability and common comprehension, see SUX Compatibility.

— UXP/V

Linux STREAMS (LiS) provides some degree of operational compatibility with UXP/V to ease portability and common comprehension, see UXP Compatibility.

— LiS-2.16.18

Linux STREAMS (LiS) provides some degree of operational compatibility with LiS 2.16 to ease portability and common comprehension, see LiS Compatibility.

For additional details, see About This Manual.

4.2 Porting

— SVR 4.2 ES/MP

Linux STREAMS (LiS) provides compatibility functions for source level compatibility with SVR 4.2 ES/MP and to ease porting of modules and drivers to Linux STREAMS (LiS). Portability considerations are maintained in a separate manual: see Porting from SVR 4.2 MP.

— AIX 5L Version 5.1

Linux STREAMS (LiS) provides compatibility functions for source level compatibility with AIX 5L Version 5.1 and to ease porting of modules and drivers to Linux STREAMS (LiS). Portability considerations are maintained in a separate manual: see Porting from AIX 5L Version 5.1.

— HP-UX 11.0i v2

Linux STREAMS (LiS) provides compatibility functions for source level compatibility with HP-UX 11.0i v2 and to ease porting of modules and drivers to Linux STREAMS (LiS). Portability considerations are maintained in a separate manual: see Porting from HP-UX 11.0i v2.

— OSF/1 1.2/Digital UNIX/True 64

Linux STREAMS (LiS) provides compatibility functions for source level compatibility with OSF/1 1.2/Digital UNIX/True 64 and to ease porting of modules and drivers to Linux STREAMS (LiS). Portability considerations are maintained in a separate manual: see Porting from OSF/1 1.2/Digital UNIX.

— UnixWare 7.1.3 (OpenUnix 8)

Linux STREAMS (LiS) provides compatibility functions for source level compatibility with UnixWare 7.1.3 (OpenUnix 8) and to ease porting of modules and drivers to Linux STREAMS (LiS). Portability considerations are maintained in a separate manual: see Porting from UnixWare 7.1.3 (OpenUnix 8).

— Solaris 9/SunOS 5.9

Linux STREAMS (LiS) provides compatibility functions for source level compatibility with Solaris 9/SunOS 5.9 and to ease porting of modules and drivers to Linux STREAMS (LiS). Portability considerations are maintained in a separate manual: see Porting from Solaris 9/SunOS 5.9.

— SUPER-UX

Linux STREAMS (LiS) provides compatibility functions for source level compatibility with SUPER-UX and to ease porting of modules and drivers to Linux STREAMS (LiS). Portability considerations are maintained in a separate manual: see Porting from SUPER-UX.

— UXP/V

Linux STREAMS (LiS) provides compatibility functions for source level compatibility with UXP/V and to ease porting of modules and drivers to Linux STREAMS (LiS). Portability considerations are maintained in a separate manual: see Porting from UXP/V.

— LiS-2.16.18

Linux STREAMS (LiS) provides compatibility functions for source level compatibility with LiS-2.16.18 and to ease porting of modules and drivers to Linux STREAMS (LiS). Portability considerations are maintained in a separate manual: see Porting from Linux STREAMS (LiS) 2.16.18.

For additional details, see About This Manual.

5 Releases

This is the OpenSS7 Release of the Linux STREAMS (LiS) core, tools, drivers and modules that implement the Linux STREAMS (LiS) SVR 4 STREAMS utility for Linux.

The following sections provide information on Linux STREAMS (LiS) releases as well as compatibility information of OpenSS7 release to the original GCOM releases of these modules and drivers, as well as Linux kernel compatibility.

5.1 Prerequisites

The quickest and easiest way to ensure that all prerequisites are met is to download and install this package from within the OpenSS7 Master Package, openss7-0.9.2.F, instead of separately.

Prerequisites for the Linux STREAMS (LiS) package are as follows:

1. Linux distribution, somewhat Linux Standards Base compliant, with a 2.4 or 2.6 kernel and the appropriate tool chain for compiling out-of-tree kernel modules. Most recent Linux distributions are usable out of the box, but some development packages must be installed. For more information, see Compatibility.

   	− A fairly LSB compliant GNU/Linux distribution.6
   	− Linux 2.4 kernel (2.4.10 - 2.4.27), or
   	− Linux 2.6 kernel (2.6.3 - 2.6.21);
   	− glibc2 or better.
   	− GNU info (for info files).
   	− GNU groff (for man pages).7

If you need to rebuild the package from sources with modifications, you will need a larger GNU toolchain as described in See Downloading from CVS.

5.2 Compatibility

This section discusses compatibility with major prerequisites.

5.2.1 GNU/Linux Distributions

Linux STREAMS (LiS) is compatible with the following Linux distributions:⁸

• CentOS Enterprise Linux 3.4 (centos34)
• CentOS Enterprise Linux 4.0 (centos4)
• CentOS Enterprise Linux 4.92 (centos48)
• CentOS Enterprise Linux 5.0 (centos5)
• Debian 3.0r2 Woody (deb3.0) – TBD
• Debian 3.1r0a Sarge (deb3.1)
• Debian 4.0r1 Etch (deb4.0) (untested)
• Fedora Core 1 (FC1) – TBD
• Fedora Core 2 (FC2) – TBD
• Fedora Core 3 (FC3) – TBD
• Fedora Core 4 (FC4) – TBD
• Fedora Core 5 (FC5)
• Fedora Core 6 (FC6)
• Fedora 7 (FC7)
• Gentoo 2006.1 (untested)
• Lineox 4.026 (LEL4) – TBD
• Lineox 4.053 (LEL4)
• Mandrakelinux 9.2 (MDK92) – TBD
• Mandrakelinux 10.0 (MDK100) – TBD
• Mandrakelinux 10.1 (MDK101) – TBD
• Mandriva Linux LE2005 (MDK102) – TBD
• Mandriva Linux LE2006 (MDK103)
• Mandriva One (untested)
• Performance Technlogies NexusWare24 – TBD
• Performance Technologies NexusWare 8.0
• RedHat Linux 7.2 (RH7)
• RedHat Linux 7.3 (RH7)
• RedHat Linux 8.0 (RH8) – TBD
• RedHat Linux 9 (RH9) – TBD
• RedHat Enterprise Linux 3.0 (EL3)
• RedHat Enterprise Linux 4 (EL4)
• RedHat Enterprise Linux 5 (EL5)
• SuSE 8.0 Professional (SuSE8.0) – TBD
• SuSE 9.1 Personal (SuSE9.1) – TBD
• SuSE 9.2 Professional (SuSE9.2) – TBD
• SuSE OpenSuSE (SuSEOSS)
• SuSE 10.0 (SuSE10.0)
• SuSE 10.1 (SuSE10.1)
• SuSE 10.2 (SuSE10.2)
• SLES 9 (SLES9)
• SLES 9 SP2 (SLES9)
• SLES 9 SP3 (SLES9)
• SLES 10 (SLES10)
• Ubuntu 5.10 (ubu5.10)
• Ubuntu 6.03 LTS (ubu6.03)
• Ubuntu 6.10 (ubu6.10)
• Ubuntu 7.04 (ubu7.04)
• WhiteBox Enterprise Linux 3.0 (WBEL3)
• WhiteBox Enterprise Linux 4 (WBEL4)

When installing from the tarball (see Installing the Tar Ball), this distribution is probably compatible with a much broader array of distributions than those listed above. These are the distributions against which the current maintainer creates and tests builds.

5.2.2 Kernel

The Linux STREAMS (LiS) package compiles as a Linux kernel module. It is not necessary to patch the Linux kernel to build or use the package.⁹ Nor do you have to recompile your kernel to build or use the package. OpenSS7 packages use autoconf scripts to adapt the package source to your existing kernel. The package builds and runs nicely against production kernels from the distributions listed above. Rather than relying on kernel versions, the autoconf scripts interrogate the kernel for specific features and variants to better adapt to distribution production kernels that have had patches applied over the official kernel.org sources.

The Linux STREAMS (LiS) package is compatible with 2.4 kernel series after 2.4.10 and has been tested up to and including 2.4.27. It has been tested from 2.6.3 up to and including 2.6.21 (with Fedora 7 patchsets). Please note that your mileage may vary if you use a kernel more recent than 2.6.21: it is difficult to anticipate changes that kernel developers will make in the future. Many kernels in the 2.6 series now vary widely by release version and if you encounter problems, try a kernel within the supported series.

UP validation testing for kernels is performed on all supported architectures. SMP validation testing is performed on UP machines, as well as on an Intel 3.0GHz Pentium IV 630 with HyperThreading enabled. Because HyperThreading is not as independent as multiple CPUs, SMP validation testing is limited.

5.2.3 Architectures

The Linux STREAMS (LiS) package compiles and installs on a wide range of architectures. Although it is believed that the package will work on all architectures supported by the Linux kernel being used, validation testing has only been performed with the following architectures:

• ix86
• x86_64
• ppc (MPC 860)
• ppc64

32-bit compatibility validation testing is performed on all 64-bit architectures supporting 32-bit compatibility. If you would like to validate an OpenSS7 package on a specific machine architecture, you are welcome to sponsor the project with a test machine.

5.2.3.1 Kernel Version 2.3.x

For LiS version 2.7 and later and for kernel version 2.3.x there are some significant compatibility issues. Version 2.3 of the Linux kernel brings with it some compatibility issues that need to be addressed by the LiS user. The two most important ones concern the file <sys/stropts.h> and the major device numbers used by LiS.

stropts.h Compatibility

There are no more compatibility problems with <sys/stropts.h> with glibc-2.1 and LiS-2.10. The following is more for historical purposes than practical necessity.

Beginning at least with egcs-2.91.66 (egcs-1.1.2 release), which comes with Red Hat 6.0, there is a file in the standard include directory named <sys/stropts.h>. This file has constant definitions that are incompatible with those used in LiS/include/sys/stropts.h. If you compile an application against the glibc version of stropts.h, and compile LiS using its own version then certain ioctls may not work correctly. You should be aware of this problem and be sure to include "-I/usr/src/LiS/include" in the compiler options that you use in compiling your STREAMS based applications.

In this version of LiS, some of the constants in stropts.h have been changed to conform to the values used by UnixWare and Solaris. These are different values than previously used in LiS. When you install LiS the installation procedure will ask you whether you want LiS compiled with the backward-compatible LiS constants, or the UnixWare/Solaris compatible constants. Logically speaking, it does not matter which set you use as long as LiS and your application code are both compiled with the same values.

I highly recommend that you use the UnixWare/Solaris compatible version, however. A future release of egcs, utilizing glibc 2.2, will contain an updated version of its stropts.h which has constants that are compatible with UnixWare, Solaris and LiS. So by selecting the UnixWare/Solaris compatible version at this time you can ensure that your applications will be fully compatible with these values in the future.

With any luck, these constants will never have to change again.

Major Device Number Compatibility

The second major compatibility issue concerns the major device numbers that LiS assigns to STREAMS devices. In the past LiS based these device numbers at 50, since the Linux kernel did not pre-define many major device numbers. As of kernel version 2.3.x there are major device numbers defined up to 220 and beyond! So starting with LiS-2.12, we have used the major number of 240 as the base for STREAMS device files. This range is supposed to be reserved for "experimental drivers" which should make it safe to use.

What this means is that you must be sure to run the strmakenodes program before running any STREAMS applications after installing LiS-2.12. This need not concern you overly, since doing a "make install" in the /usr/src/LiS directory causes strmakenodes to be run anyway. This is more a concern if you are compiling LiS on one machine and then loading it onto another for execution. In such cases you may need to load the new strmakenodes program and run it. I am hoping that the kernel developers will expand the major and minor device number spaces for 2.6. If they do that then LiS should be able to get a block of majors allocated to it.

5.2.3.2 Kernel Version 2.2.x

For LiS version 2.5 and later and for kernel version 2.2.x there are no compatibility issues; there are no kernel patches whatsoever required to install LiS. You will need LiS-2.4 at minimum to run in a 2.2.x kernel.

5.2.3.3 Kernel Version 2.0.36

The latest version of LiS has not been tested on 2.0 kernels. Therefore, do not be surprised if it does not install or execute correctly in these kernels. If you are using an old kernel, you must also use an older version of LiS, perhaps LiS-2.5.

For LiS version 2.5 and later and for kernel version 2.0.36 there are no kernel patches required to run LiS as a "bottom half" process. A one-line patch is required to run LiS as a kernel daemon process. The installation default is to run as a bottom half process in 2.0.36. LiS-1.25 or later should install properly with 2.0.36. The more recent the version of LiS, the less kernel patching is required.

5.2.4 Linux STREAMS

Linux Fast-STREAMS provides a suitable replacement for the (now deprecated) Linux STREAMS (LiS) 2.18.0 package formerly maintained by Dave Goethe of GCOM.

5.2.4.1 LiS-2.18 Kernel and Driver Compatibility

There are several issues that needed to be addressed for compatibility with the 2.6 Linux kernel. You are encouraged to follow the links in the paragraphs below to see more detailed information on each of these topics.

1. The 2.6 kernel redefined the size of the dev_t structure. LiS has extended its internal dev_t structure to be compatible with the 2.6 method for some time.
2. The 2.6 kernel changed the approach to building and installing kernel modules. This affects LiS as a whole and also affects how you install separate loadable STREAMS drivers. LiS provides a mechanism that allows STREAMS drivers and modules to be easily installed.
3. The 2.6 kernel offers an option to compile the kernel using machine registers to pass parameters to functions. LiS takes this into account.
4. The 2.6 kernel needs GCC version 3.3.3 (sic) to be compiled properly. LiS needs to be compiled using the same version of the compiler when running with the 2.6 kernel.
5. You may have to edit the file /etc/rc.d/rc.sysinit to get demand loadable modules to work correctly. This is especially true when hosting a 2.6 kernel on a 2.4 distribution.

5.2.4.2 LiS-2.16 Kernel and Driver Compatibility

LiS-2.16 is a small change from LiS-2.15. The change is that it no longer uses Linux system calls to implement getpmsg and putpmsg. Instead it overloads the read and write file system functions with particular values for the count parameter, values that are otherwise invalid.¹⁰

5.2.4.3 LiS-2.15 Kernel and Driver Compatibility

LiS-2.15 continues to insulate STREAMS drivers from the Linux kernel. It works with 2.2, 2.4, and 2.5 versions of the kernel. Support for 2.0 kernels has been dropped.

Driver writers will need to recompile their drivers against LiS-2.15 include files. You will see the following major changes.

• LiS spin locks and semaphores have been rearranged so that the kernel memory is at the end of the structure instead of the beginning.
• The former change allows for there to be dynamic allocation routines for spin locks and semaphores. (see LiS Spin Locks)
• LiS now provides an abstraction for read/write locks, with dynamic allocation. (see LiS Read/Write Locks)
• Those experimenting with 2.5 kernels will notice that the "sleep while holding spin lock" problems have been fixed.
• Porting to 2.5 has necessitated some changes to the major/minor device structure handling. (see Major/Minor Device Numbering)
• The fattach related functions are functional on kernels version 2.4.7 and later.
• STREAMS pipes and FIFOs are now functional.
• OS interface code has been added for the kernel's DMA mapping functions.

There is one known bug in LiS-2.15 relative to 2.5 kernels. It has to do with a memory leak involving timer structures, and may prove to be a kernel bug rather than an LiS bug. Since the 2.5 kernel is not suitable for general use I am saving the investigation of this bug for later.

5.2.4.4 LiS-2.14 Kernel and Driver Compatibility

LiS-2.13 was a series of beta releases. LiS-2.14 represents the culmination of this series. There should be enough distribution and kernel compatibility that LiS-2.14 will hold up for some time.

The known fattach and FIFO bugs have still not been fixed. The author of those subsystems has not found the time to put in the fixes, nor have I.

5.2.4.5 LiS-2.13 Kernel and Driver Compatibility

This version of LiS has been tested with 2.4 kernels up to 2.4.16. LiS does not yet support the fattach/fdetach functions on kernel versions 2.4.7 and beyond. There are also known bugs in the LiS pipe/FIFO code. All of these problems are scheduled to be fixed in early 2002.

LiS-2.13 adds the ability for drivers to make their own "/dev" nodes via the lis_mknod function (see System Calls from within the Kernel). Also provided is an lis_unlink function that allows drivers to remove their device files.

There is almost no new functionality added by LiS-2.13. The differences between LiS-2.13 and LiS-2.12 are almost entirely kernel compatibility issues and bug fixes.

5.2.4.6 LiS-2.12 Kernel and Driver Compatibility

This version of LiS is compatible with all 2.2.x versions of the kernel and with early versions of the 2.4.x kernel, at least up to 2.4.2 and perhaps later versions as well.

If you have drivers that have worked with LiS-2.10 or LiS-2.11 (or earlier) please recompile them using the header files from LiS-2.12. This may be the last recompile in quite some time that you will need for your driver code.

LiS-2.12 contains a sufficient Driver/Kernel Interface (DKI), (see Development), that it is straightforward to write a STREAMS driver that can be compiled against LiS-2.12 and the resulting object modules used either on a 2.2 or 2.4 kernel, with only LiS needing recompilation on the target machine.

When run on 2.4 kernels, LiS makes full use of multiple CPUs (see LiS SMP Implementation). It forks a queue runner task for each CPU and locks each task onto its CPU. Queue runner tasks are awakened to assist with service procedure processing as the number of scheduled queues increases.

Because of this aggressive use of processors, you may find that your drivers do not function properly when run with LiS-2.12 in a multi-CPU SMP environment. You should expect that drivers that worked in single-CPU environments will continue to work as before.

Making your drivers MP safe involves the use of spin locks. The DKI documentation contains advice on the use of these locks. See LiS Spin Locks.

This version of LiS also contains a rewrite of the flushing code and tests added to strtst for flushing. In particular the details of the rules for flushing queue bands are now adhered to. See Flushing Queue Bands. Be advised, however, that Solaris STREAMS does not adhere strictly to these rules so there may be some subtle differences in behaviour between LiS and Solaris when flushing queue bands.

Speaking of queue bands, the queue band handling code has been debugged a bit more and a test added to strtst to illustrate its correct behaviour.

5.2.4.7 LiS-2.10 Kernel and Driver Compatibility

This version of LiS is compatible with all 2.2.x versions of the Linux kernel. It may work with 2.4.x kernels, but you should probably wait for LiS-2.11 for that.

If you have drivers that worked with LiS-2.8 or earlier, you must recompile your drivers in the context of the LiS-2.10 header files. The queue_t structure has changed in size since LiS-2.8 which means that the old RD and WR macros will not compute the correct addresses.

LiS-2.10 contains features that are intended to greatly reduce the necessity of recompiling STREAMS driver code in future versions of LiS or future versions of the kernel. The goal is to be able to compile STREAMS drivers against LiS-2.10 header files and use the resulting object code on both 2.2.x kernels and 2.4.x kernels.

For more details about the interface between STREAMS drivers and the kernel, see the Driver/Kernel Interface documentation, (see Development).

5.2.5 Linux Fast-STREAMS

The Linux STREAMS (LiS) package is no longer receiving active development or support. The Linux STREAMS (LiS) package is so fraught with bugs that it is unusable as far as The OpenSS7 Project is concerned. Linux Fast-STREAMS is the preferred replacement for Linux STREAMS (LiS).

5.3 Release Notes

The sections that follow provide information on OpenSS7 releases of the
Linux STREAMS (LiS) package.

Major changes for release LiS-2.18.6

This is an internal release of LiS. It is only available for download by subscribers and sponsors of the OpenSS7 Project. There are too many packages that cannot build against LiS due to its implementation deficiencies.

Do not use this release (even if you are a subscriber or sponsor). Port to Linux Fast-STREAMS. Do not report bugs on this release. The OpenSS7 Project no longer actively maintains LiS.

Major features since the last public release are as follows:

• Support build on openSUSE 10.2.
• Support build on Fedora 7 with 2.6.21 kernel.
• Support build on CentOS 5.0 (RHEL5).
• Support build on Ubuntu 7.04.
• Updated to gettext 0.16.1.
• Changes to support build on 2.6.20-1.2307.fc5 and 2.6.20-1.2933.fc6 kernel.
• Supports build on Fedora Core 6.
• Support for recent distributions and tool chains.

Major changes for release LiS-2.18.5

This is an internal release of LiS. There are too many packages now that cannot build against LiS because it is lacking fundamental capabilities in the Stream head.

Do not use this release. Port to Linux Fast-STREAMS. Do not report bugs on this release. (Yup, there are lots of 'em, but that's LiS.)

Major features since the last public release are as follows:

• Added versions to all exported symbols. Made LiS functions GPL export.
• Improvements to the common build environment with better support for standalone package builds on 2.4 kernels.
• Support for autoconf 2.61, automake 2.10 and gettext 0.16.
• Support for Ubuntu 6.10 distribution and bug fixes for i386 kenels.

Major changes for release LiS-2.18.4

This is the final release of LiS. There are too many packages now that cannot build against LiS because it is lacking fundamental capabilities in the Stream head.

This release builds both 32-bit compatibility and 64-bit native libraries and functions on 64-bit architectures. One of the major reasons for doing this for LiS was to demonstrate its sad lack of ability to sustain any form of 32-bit compatibility on most 64-bit architectures. Test suites now run first 64-bit native and then 32-bit compatibility tests to demonstrate LiS' dismal failure in this regard.

Do not use this release. Port to Linux Fast-STREAMS. Do not report bugs on this release. (Yup, there are lots of 'em, but that's LiS.)

Major features since the last public release are as follows:

• Support for most recent 2.6.18 kernels (including Fedora Core 5 with inode diet patchset).
• Added send-pr scripts for automatic problem report generation. (Please do not report bugs on LiS.)
• Added --disable-devel configure option to suppress building and installing development environment. This feature is for embedded or pure runtime targets that do not need the development environment (static libraries, manual pages, documentation).
• Corrected start of assignment of major devices numbers (back to 230).
• Improved compiler flag generation and optimizations for recent gcc compilers and some idiosyncratic behaviour for some distributions (primarily SUSE).
• Optimized compilation is now available also for user level programs in addition to kernel programs. Added new --with-optimize option to configure to accomplish this.
• Better detection of SUSE distributions, release numbers and SLES distributions: support for additional SuSE distributions on ix86 as well as x86_64. Added distribution support includes SLES 9, SLES 9 SP2, SLES 9 SP3, SLES 10, SuSE 10.1.
• Many documentation updates for all OpenSS7 packages. Automated release file generation making for vastly improved and timely text documentation present in the release directory.
• Bug fixes to STREAMS library for proper asynchronous thread cancellation handling. Versions for all library symbols.
• Now builds 32-bit compatibility libraries and tests them against 64-bit kernel modules and drivers. The `make installcheck' target will now automatically test both 64-bit native and 32-bit compatibility versions, one after the other, on 64-bit platforms.
• The package will now build doxygen(1) html documentation with the 'doxy' make target. See 'make help' or README-make in the distribution for more information.

Major changes for release LiS-2.18.4.rc3

Third release candidate.

LiS is really not supported any longer. Use Linux Fast-STREAMS instead. The purpose of this release is to demonstrate the inability of the LiS package to properly support 32-bit compatibility on 64-bit architectures.

• Minor bug fixes to STREAMS library. isastream(3) and fattach(3) must not contain an asynchronous thread cancellation point, but they could. Added asynchronous thread cancellation protection to these functions to remove any thread cancellation points.
• The package will now build doxygen(1) html documentation with the 'doxy' make target. See 'make help' or README-make in the distribution for more information.
• Now builds 32-bit compatibility libraries and tests them against 64-bit kernel modules and drivers. The `make installcheck' target will now automatically test both 64-bit native and 32-bit compatibility versions, one after the other, on 64-bit platforms.
• Added versions for all library symbols.
• Automated release file generation making for vastly improved and timely text documentation present in the release directory.
• Many documentation updates for all OpenSS7 packages.
• This release candidate includes the changes made to the strsctp drivers at the 2006 SCTP Interop at the University of British Columbia. This version was interoperability tested with all implementations present.
• This release candidate provides support for additional SuSE distributions on ix86 as well as x86_64. Added distribution support includes SLES 9, SLES 9 SP2, SLES 9 SP3, SLES 10, SuSE 10.1.
• Better detection of SUSE distributions, release numbers and SLES distributions.
• Optimized compilation is now available also for user level programs in addition to kernel programs. Added new --with-optimize option to configure to accomplish this.
• Improved compiler flag generation and optimizations for recent gcc compilers and some idiosyncratic behaviour for some distributions (primarily SUSE):

  −	Remove -fno-reorder-blocks and -fno-reorder-functions options added by some recent 2.6 Makefiles for `x86_64' architecture: it impedes performance optimizations.
  −	Remove -ffunction-sections option added by some recent 2.6 Makefiles for `x86_64' architecture: this is an insane option and should never have been used.
  −	Add -ffreestanding that some older 2.6 Makefiles (such as that with SLES 9 2.6.5 kernel) neglect to add to the gcc command line.
  −	SLES 10 expands the directory before autoconf.h on the gcc command line for some reason. configure script watches out for this now.

• Updated init scripts for proper addition and removal of modules.
• Start assigning majors at major device number 231 instead of major device number 230. Assign major device number 230 explicitly to the clone device.
• Linux Fast-STREAMS now supports expanded addressable minor device numbers, permitting 2^16 addressable minor devices per major device number on 2.6 kernels: LiS cannot support this change.

This was an internal alpha test release candidate and was not released publicly. This release was only available to subscribers to and sponsors of the OpenSS7 Project.

Major changes for release LiS-2.18.4.rc2

Second release candidate.

• Added `--enable-devel' configure option for embedded targets.
• Added send-pr script for automatic problem report generation.

This was an internal alpha test release candidate and was not released publicly. This release was only available to subscribers to and sponsors of the OpenSS7 Project.

Major changes for release LiS-2.18.4rc1

First release candidate.

This was an internal alpha test release candidate and was not released publicly. This release was only available to subscribers to and sponsors of the OpenSS7 Project.

Major changes for release LiS-2.18.3

Corrections for and testing of 64-bit clean compile and test runs on x86_64 architecture. Some bug corrections resulting from gcc 4.0.2 compiler warnings.

Corrected build flags for Gentoo and 2.6.15 kernels as reported on mailing list. Builds on FC4 2.6.15 kernel and with gcc 4.0.2.

Added in many of Paul's 64-bit corrections.

The Linux STREAMS (LiS) 2.18.3 is still largely unusable on 64-bit or SMP kernels. Linux STREAMS (LiS) package is deprecated. Do not use it. The package contains many unresovled bugs. Use Linux Fast-STREAMS instead.

Major changes for release LiS-2.18.2

Cross-build support for newer NexusWare releases. Build support for (recent FC4) 2.6.14 kernel. Corrected installation support (init scripts) for SuSE 9.2.

Binary compatibility backward compatible to GCOM 2.18.0 included. This includes exported symbols changed to not generate version symbols on 2.4 kernels. Also, exported symbols are always compiled attribute((regparm(0))) on regparm capable architectures, regardless of the kernel version or compile options.¹¹ For actual binary compatibility packaging, see the strcompat package.

A major change for 2.18.2 is the port-back of POSIX/SUSv3 XSR/XSI conformance test suites and performance programs from Linux Fast-STREAMS. The purpose of porting back theses tests suites and supporting modules and drivers is to provide the ability to do comparison tests between LiS and Linux Fast-STREAMS.

Another change is a module unloading safe vstrlog hook register and unregister functions register_strlog() and unregister_strlog().

Some bug corrections and fixes for glaring SMP errors reported by Kutluck.

This might be the last OpenSS7 release of Linux STREAMS (LiS). You should seriously consider using the Linux Fast-STREAMS package (streams-0.7a.4 or later) instead. If you need compatibility to LiS (or other STREAMS implementation), investigate the strcompat package, which provides some binary compatibility to LiS under Linux Fast STREAMS.

Major changes for release LiS-2.18.1

Initial autoconf/RPM packaging of the LiS release.

This is a port forward of most of the build and patches from 2.16.19 forward and applied over 2.18.0. This is our first LiS-2.18 release. All further development on 2.16.19 will now cease. 2.18.1 is maintained on both 2.4 and 2.6 kernels. No active development will be performed on 2.18.1, only maintenance. For an active development release, see the Linux Fast-STREAMS releases.

Major changes from LiS-2.18.0 include all of the autoconf build system, manual pages and texi/pdf manual for LiS that were applied on the 2.16.19 release. This includes a number of 64 bit, HPPA, PARISC, printf, atomic stats, HZ calculations for 64bit machines, DMA patch for mblk buffer alignment, flush handling patch, panic patch, smp patch, parisc syscall patch, appq patch, and multi-threaded test program patches, POSIX threads compliant library functions.

Additional changes made to support later 2.6 kernels and distributions. Switched putpmsg()/getpmsg() to use ioctl for system call emulation instead of read()/write(), primarily because 2.6.11 kernels check for a valid count before calling the driver's read()/write() file operations. Updates to the build system to support a wider range of kernels and distributions. See the installation and reference manual for a complete list of supported kernels and distributions.

Please note that the entire package is released under GPL.

Major changes for release LiS-2.16.19

Not publicly released.

Major changes for release LiS-2.16.18-22

Replaced m4 and automake files with common equivalents. This allows the same set of m4 macros and automake fragments to be used with all of the OpenSS7 release packages. Maintenance is easier as one correction will propagate across all items. Performed similar function with texinfo documentation pieces.

Major changes for release LiS-2.16.18-21

Removed all XTI/TLI and Linux networking code, headers and documentation from LiS distribution and bumped epoch to 2. Linux networking code has been migrated to the strxns, strxnet, strinet and strsctp packages. The purpose for doing this was to allow the Linux networking to build against Linux Fast-STREAMS as well as Linux STREAMS (LiS) and is a preparation for phasing out LiS and phasing in LfS.

Added missing configure.nexusware to distribution. LiS cache options now default to 'no' because of instabilities with timers.

Not publicly released.

Major changes for release LiS-2.16.18-20

Minor corrections: made conflicting manpage xti_sctp.3 dependent on OpenSS7 SCTP kernel.

Not publicly released.

Major changes for release LiS-2.16.18-19

Changes to compile, install and builds rpms for Fedora Core 1 (FC1), Whitebox Enterprise Linux (WBEL) and RedHat Enterprise Linux 3 (EL3). Included explicit epoch in internal dependencies in spec file for RPM versions 4.2.1, 4.2.2 and higher. Added hugemem kernel detection and moved getpmsg and putpmsg manual pages.

Correction to symbolic linking and system map file location during non-rpm autoconf installation.

Correction to zero maxlen behaviour in t_rcvconnect().

Major changes for release LiS-2.16.18-18

Added check for CONFIG_REGPARM, addition of -mregparm=3 CFLAGS, addition of regparm_ prefix for exported ksyms.

Minor corrections to separate build directory install of devices and caching of detected ksyms.

Major changes for release LiS-2.16.18-17

Added option --disable-k-modversions to suppress versions for LiS exported symbols.

A couple of corrections to the build process reported by Gurol. Changed order of build in `make rebuild' target to build tools last so that the rpm debug package is built correctly on RH9.

Change MODULE_PARM to static so that make install-strip does not strip module parameter symbols.

Added lis_check_mem_region(), lis_release_mem_region() and lis_request_mem_region() for memory mapped io instead of just io.

Added printk patches discussed on linux-stream mailing list. Added gcc printf checking and corrected errors in LiS debugging printk statements.

Added HP patches. There are a couple of questionable components in the HP patches that I reversed. They include;

• modification of lis_msgsize to lis_msgdsize. This would change the calculation of queue counts. Queue counts aren't M_DATA counts, they are "data" message counts. LiS probably doesn't have this the right way, but lis_msgdsize is not correct either.
• addition of qi_mstat->ms_pcnt increment on lis_safe_putmsg. Same for ms_scnt, ms_ocnt, ms_ccnt. STREAMS is not supposed to increment counts. It is the module writer's responsibility to increment counts in their own queue procedures.

Added HP ldl patches.

Made modifications to putq(), putbq(), insq() and appq() discussed on linux-streams mailing list. These do not free messages on failure. Modified all occurrences internal to LiS to free the message on error to ensure old behaviour.

Added HP dejagnu patches to strtst and added dejagnu testsuite directory and file. Added the make check target. Use DEJATOOLS on the make command line to invoke the tests, such as `make DEJATOOLS=strtst check' to invoke the tests. Because a patched netperf is not commonly available and netperf will not be distributed with the package, GNU autotest might be a better choice. But that's for a later release.

Major changes for release LiS-2.16.18-16

General updates to the build process, optimization options, build options. Corrected library linkage. Synced TLI modules and INET driver to Linux Fast-STREAMS. Removed deadlock from INET driver and loosened locking. Unfortunately suitable libraries must be installed before distcheck will clear.

Smoother and more reliable stripping of kernel symbols, starts with /proc/ksyms if applicable then System.map then modversions.h to attempt to choose symbols most closely synced with an installed or running kernel.

Improvements to autoconf installation of manpages (autocompressed now) and info and pdf manuals are distributed. install-strip target will actually properly strip kernel modules.

Included an option to build and install only kernel or user parts of package to speed rpm rebuild process for multiple kernel. Added `rebuild' target to rebuild the rpms from srpm for multiple kernel and architectures. Added a `sign' and `resign' target to sign srpm and rebuilt rpms respectively.

Greatly enhanced cross-build and cross-compile support, primarily in support of the NexusWare embedded target. Added NexusWare helper script and documentation. DESTDIR is now a blessed environment variable used by configure to set the cross-build root as well as the install root. Try adding –with-k-optimize='size' to configure to optimize for size for embedded targets. Builds clean against NexusWare24 (810p0674.10-rc4).

Added start of an option to build as linkable object for embedded targets rather than loadable kernel module.

Major changes for release LiS-2.16.18-15

Fixed several symbol errors that made -13 and -14 unusable. Corrected error in calculation of kernel debug flags.

Major changes for release LiS-2.16.18-14

A few more enhancements to the build process to work with autoconf 2.59.

Major changes for release LiS-2.16.18-13

Enhanced build process for autoconf-2.59, automake-1.8.3, gettext-0.14.1, and libtool-1.5.6.

Major changes for release LiS-2.16.18-12

Added defaults for SK_WMEM_MAX and SK_RMEM_MAX for lastest 2.4.25 and 2.4.26 kernel builds.

Enhanced build process.

All kernel symbols exported by LiS have versions on kernels that have versions on symbols. This makes it safer to compile kernel modules against kernel/LiS combinations. This is in preparation for splitting off the strxnet package, and the technique was imported from the Linux Fast-STREAMS build.

Major changes for release LiS-2.16.18-11

Ripped three additional kernel symbols in support of INET driver that were missing in -10 release.

Major changes for release LiS-2.16.18-10

Added support for cooked manpages both for non-rpm systems and for rpm systems. It is still better to leave manpages uncooked for rpm releases because they are much smaller that way. Give the –with-cooked-manpages flag to configure if you want cooked manpages. You still need grefer on the build system.

Updates to all manual pages in man7, and some others (xti) in man3. Removed unused .macros and .refs files.

Moved automake fragments into separate directory. Cleaned up automake fragments.

Rearranged header files in the xti subdirectory to install in LiS package include directory instead. Reworked xti, tihdr and tiuser file groups to include properly from kernel or user space independent of order. tiuser and xti still cannot be included together. Added older TLI interface <tiuser.h> that is still consistent with newer XTI interface. Changed references in man pages to XTI/TLI instead of just XTI.

Added ticlts.h, ticots.h and ticotsord.h header files. Updated dlpi.h and npi.h header files. Removed sys/LiS/tpicommon.h because it is largely replaced by sys/tli.h and sys/tpi.h. Removed the, now redundant, xti header file subdirectory.

A series of bug fixes to xnet.c (libxnet) that resulted from discussions with Gurol Akman on openss7-develop mailing list. Mostly surrounding t_alloc and t_getinfo behaviour and the behaviour when NULL pointers are passed to various XTI/TLI library calls. Updated xti documentation as well.

Many changes to the inet.c INET driver. Wildcard IP addresses can now be bound and wildcard addresses will be assigned with no address is passed to most providers. (/dev/rawip still requires an address or TNOADDR is returned.) Option management has been extensively rewritten to closer conform to XNS documentation. Test programs test-inet_raw, test-inet_udp, test-inet_tcp have been upgraded and converted to multiple child processes. A number of fixes to SMP lock behaviour and M_FLUSH have been added as reported by Dave Grothe. Corrected all level and TBADOPT behaviour on negotiation.

Although this driver is now closer to expected behaviour, it has not been tested for XNS 5.2 compliance, nor will it be until someone has the time to extend the test programs to handle all test cases in a similar manner as was done for the library. Your mileage many vary. Remember, there is no warranty.

Major changes for release LiS-2.16.18-9

Changes primarily in support of builds on HPPA (PARISC) architectures. LiS doesn't build too well on PARISC so some modifications where used from the Linux Fast-STREAMS package to correct deficiencies. Better building on recent 2.4 kernels (2.4.23, 2.4.24, 2.4.25) is also provided.

Major changes for release LiS-2.16.18-8

Changes to permit better builds on recent RedHat kernels, and especially kernel-2.4.20-30.9.

Major changes for release LiS-2.16.18-7

Fixed a module loading bug in LiS. Previously modules would not demand load.

Major changes for release LiS-2.16.18-6

Fixed a possible null pointer dereference in libxnet. Corrected t_bind to return TNOADDR instead of TBADADDR on wildcard bind attempt. Module loading bug patched.

Major changes for release LiS-2.16.18-5

Fixes a t_open and t_bind problem in libxnet. Fixes alignment of data portion of mblks. Adds (untested) ticots_ord, ticots and ticlts devices over UNIX domain sockets.

Major changes for release LiS-2.16.18-4

Adds back in missing strms_up/down/status scripts to distribution and install.

Major changes for release LiS-2.16.18-3

Not publicly released.

Major changes for release LiS-2.16.18-2

Not publicly released.

Major changes for release LiS-2.16.18-1

This OpenSS7 release of LiS-2.16.18 updates the previous LiS-2.16.16 rpm release to the latest LiS-2.16 release level.

Initial release LiS-2.16.16-1

This OpenSS7 release of LiS-2.16.16 includes autoconf for configuration, complete manual pages and documentation, and packaging in source and binary RPMs for ease and repeatability of installation. The package also builds and installs properly versions LiS shared object libraries.

Before the OpenSS7 release of LiS, it was necessary to have a significant working knowledge of the Linux kernel, kernel source, headers and other intricacies. This made it difficult to distribute software based on LiS to users not proficient in those areas. The OpenSS7 release removes the configuration and installation tasks from the user and permits distribution of applications, modules and driver software based on LiS to users without sufficient kernel expertise to install the package.

This OpenSS7 release fixes few of the outstanding bugs and deficiencies of the LiS software. This release is intended to package and distribute LiS in an efficient manner and, for the most part, does not address LiS deficiencies or errors.

This OpenSS7 release is compatible with Linux 2.4 kernels only and will refuse to configure for older or newer kernels.

Following are the new features of the OpenSS7 release of LiS:

• adds configuration using the GNU tools, autoconf, automake and autotest.

  These tools greatly enhance the ability to maintain a repeatable and testable release cycle as well as being compatible with most major package managers such as Redhat's RPM.

• adds long options to all LiS utilities.

  This change was necessitated because we use GNITS (the strictest level) of configuration with autoconf that requires for distribution checking that all utility programs support the `--help' and `--version' long options without side-effects.

• provides a source and binary release mechanism using both autoconf distributions as well as RedHat source and binary RPMs.

  Use of the RPM mechanism for release permits add-on packages to ensure that they have sufficient level of support and versions for the LiS load during their build and installation process. It is now also possible to ensure that add on binaries are compatible with a loaded LiS during installation.

• includes a complete set of kernel programmer manual pages for all LiS exported kernel functions for use by STREAMS module and driver developers.
• includes a complete set of user manual pages for all libLiS functions and separate administrative utilities.
• includes the OpenSS7 strinet driver providing XTI/TLI access to the Linux native NET4 IP stack including TCP, UDP, IP and (and OpenSS7 SCTP if your kernel is so equipped).
• includes functional tirdwr and timod modules for use with the included XTI/TLI library.¹²
• includes complete, thread-safe XNS 5.2 XTI/TLI library support with the libxnet library, complete manual pages and documentation released under the LGPL (see GNU Lesser General Public License).

The next release may include some strss7 software.

5.4 Maturity

The OpenSS7 Project adheres to the following release philosophy:

• pre-alpha release
• alpha release
• beta release
• gamma release
• production release
• unstable release

5.4.1 Pre-Alpha Releases

Pre-alpha releases are releases that have received no testing whatsoever. Code in the release is not even known to configure or compile. The purpose of a pre-alpha release is to make code and documentation available for inspection only, and to solicit comments on the design approach or other characteristics of the software package.

Pre-alpha release packages ship containing warnings recommending that the user not even execute the contained code.

5.4.2 Alpha Releases

Alpha release are releases that have received little to no testing, or that have been tested and contains known bugs or defects that make the package unsuitable even for testing. The purpose for an alpha release are the same as for the pre-alpha release, with the additional purpose that it is an early release of partially functional code that has problems that an external developer might be willing to fix themselves and contribute back to the project.

Alpha release packages ship containing warnings that executing the code can crash machines and might possibly do damage to systems upon which it is executed.

5.4.3 Beta Releases

Beta releases are releases that have received some testing, but the testing to date is not exhaustive. Beta release packages do not ship with known defects. All known defects are resolved before distribution; however, as exhaustive testing has not been performed, unknown defects may exist. The purpose for a beta release is to provide a baseline for other organizations to participate in the rigorous testing of the package.

Beta release packages ship containing warnings that the package has not been exhaustively tested and that the package may cause systems to crash. Suitability of software in this category for production use is not advised by the project; however, as always, is at the discretion of the user of the software.

5.4.4 Gamma Releases

Gamma release are releases that have received exhaustive testing within the project, but external testing has been minimal. Gamma release packages do not ship with known defects. As exhaustive internal testing has been performed, unknown defects should be few. Please remember that there is NO WARRANTY on public release packages.

Gamma release packages typically resolve problems in previous beta releases, and might not have had full regression testing performed. Suitability of software in this category for production use is at the discretion of the user of the software. The OpenSS7 Project recommends that the complete validation test suites provided with the package be performed and pass on target systems before considering production use.

5.4.5 Production Releases

Production releases are releases that have received exhaustive testing within the project and validated on specific distributions and architectures. Production release packages do not ship with known defects. Please remember that there is NO WARRANTY on public release packages.

Production packages ship containing a list of validated distributions and architectures. Full regression testing of any maintenance changes is performed. Suitability of software in this category for production use on the specified target distributions and architectures is at the discretion of the user. It should not be necessary to preform validation tests on the set of supported target systems before considering production use.

5.4.6 Unstable Releases

Unstable releases are releases that have received extensive testing within the project and validated on a a wide range of distributions and architectures; however, is has tested unstable and found to be suffering from critical problems and issues that cannot be resolved. Maintenance of the package has proved impossible. Unstable release packages ship with known defects (and loud warnings). Suitability of software in this category for production use is at the discretion of the user of the software. The OpenSS7 Project recommends that the problems and issues be closely examined before this software is used even in a non-production environment. Each failing test scenario should be completely avoided by the application. OpenSS7 beta software is more stable that software in this category.

5.5 Bugs

5.5.1 Defect Notices

Linux STREAMS (LiS) has many and critical known defects. This is an unstable release. Some defects could be harmful. Validation testing has been performed by the OpenSS7 Project on this software and it has revealed itself to be unstable and irreparable. The software might not even configure or compile. The OpenSS7 Project recommends that you do not use this software. Use at your own risk. Remember that there is NO WARRANTY.¹³

Linux STREAMS (LiS), both releases from OpenSS7 and GCOM, contain many known bugs. These are unstable releases. Although there are no bugs known directly to be harmful, the OpenSS7 Project has tested the release and found defects that cause the kernel to lock or crash. Use at your own risk. Remember that there is NO WARRANTY¹⁴ and that the package is no longer actively maintained.

This software is unstable software. As such, it will lock or crash your kernel. Installation of the software will irreparably mangle your header files or Linux distribution in such a way as to make it unusable. Crashes will lock your system and rebooting the system will not repair the problem. You will loose all the data on your system. Because this software has tested unstable in a number of test cases, simply running the validation test cases can cause locks or crashes. Because this software will lock or crash your kernel, the resulting unstable system can destroy computer hardware or You will void the warranty on any system on which you run this software. YOU HAVE BEEN WARNED.

5.5.2 Known Defects

With the exception of packages not originally created by the OpenSS7 Project, the OpenSS7 Project software does not ship with known bugs in any release stage except pre-alpha. Linux STREAMS (LiS) had many known bugs at the time of release.

Linux STREAMS (LiS) has many known bugs. Under some architectures, the test cases in the conformance test suite cause the kernel to lock or crash. Linux STREAMS (LiS) contains many races and defects and is unsuitable for production environments. This section provides a summary of some (but not all) known defects.

1. A substantial group of test cases in the POSIX conformance test suite fail. This is largely due to non-fatal defects in LiS.
2. A number of test cases fail under any architecture and result in a kernel lock. In particular if a significant number of modules are pushed onto a Stream using I_PUSH, the kernel will lock when the Stream is closed. The number of modules pushed to cause a crash depends on the speed of the machine upon which the test case is run. The test case in the test suite pushes 64 modules and has always resulted in a kernel lock regardless of the machine speed upon which it was tested. Pushing 20 modules results in a kernel lock on some of the OpenSS7 Project 2.57GHz test machines.

   As a result, any test case that pushes a number of modules, and the performance tests (that push modules for measurement) will cause the kernel to lock.

3. A large number of test cases fail when running under an SMP kernel, regardless of the number of processors on the test system. These test cases cause the kernel to lock. The kernel locks are apparently due to locking defects in the implementation. None of these implementation defects have been repaired.
4. The original LiS-2.18.0 release from GCOM has a large number of defects that were repaired in the OpenSS7 LiS-2.18.1 release.
5. The original LiS-2.18.0 release has defects in the description of data types and handling under 64-bit architectures. 32-bit compatibility for 64-bit architectures is all but non-existent in the LiS-2.18.0 release. Because of binary compatibility issues, many of these defects persist in the OpenSS7 LiS-2.18.1 and LiS-2.18.2 releases. LiS is largely unusable in a 64-bit and almost completely unusable in a mixed architecture.

The work-around for these defects is to not use LiS at all: use the OpenSS7 Linux Fast-STREAMS release instead. The OpenSS7 Linux Fast-STREAMS, being a completely independent implementation, does not suffer from this extensive set of LiS defects.

5.5.3 Defect History

This section contains historical bugs that were encountered during development and their resolutions. This list serves two purposes:

1. It captures bugs encountered between releases during development that could possibly reoccur (and the Moon is made of blue cheese). It therefore provides a place for users to look if they encounter a problem.
2. It provides a low overhead bug list between releases for developers to use as a TODO list.

Bugs

LiS contains way too many bugs to be useful. This list only represents those bugs that were discovered in the development of Linux Fast-STREAMS that were easy enough to fix in LiS.

Do not use LiS. Use Linux Fast-STREAMS instead.

001. 2006-09-24T20:02:00+0000

Discovered asynchronous thread cancellation inconsistencies in libLiS libpLiS by inspection during documentation. isastream(2), fattach(2) were not performing proper asynchronous thread cancellation suppression so that these function contained a cancellation point when the should not.

*fixed* in LiS-2.18.4.rc3

5.6 Schedule

The OpenSS7 release of Linux STREAMS (LiS) was in maintenance mode for over a year and the latest `2.18' releases are quite stable. However, LiS is fraught with bugs and suffers from a poor design approach making it unsuitable for production use. It fails many of the conformance test cases in the test suite and even locks the kernel on a number of cases on `SMP' kernels regardless of their being run on a `UP' machine.

The OpenSS7 Project wrote a new STREAMS package for Linux called Linux Fast-STREAMS. Linux Fast-STREAMS is now far more stable than LiS, and The OpenSS7 Project will no longer perform development on the LiS package, unless at the bequest of a project sponsor. As such, LiS is deprecated. Use the latest streams-0.9.2.3 package from the OpenSS7 download site instead.

Things to do:

• Last remaining to do item is to release the LiS-2.18.6 release that might very well be the last release.

  *done*

5.7 History

For the latest developments with regard to history of changes, please see the ChangeLog file in the release package.

6 Installation

6.1 Downloading

The Linux STREAMS (LiS) package releases can be downloaded from the downloads page of The OpenSS7 Project. The package is available as a binary RPM (for popular architectures) a source RPM, Debian binary DEB and source DSC, or as a tar ball. If you are using a browsable viewer, you can obtain the OpenSS7 release of LiS from the links in the sections that follow.

By far the easiest (most repeatable and manageable) form for installing and using OpenSS7 packages is to download and install individual packages from binary RPM or DEB. If binary RPMs or DEBs are not available for your distribution, but your distribution supports rpm(1) or dpkg(1), the next best method for installing and using OpenSS7 packages is to download and rebuild the source RPMs or DSCs.

If your architecture does not support rpm(1) or dpkg(1) at all, or you have special needs (such as cross-compiling for embedded targets), the final resort method is to download, configure, build and install from tarball. In this later case, the easiest way to build and install OpenSS7 packages from tarball is to use the tarball for the OpenSS7 Master Package, openss7-0.9.2.F.

6.1.1 Downloading the Binary RPM

To install from binary RPM, you will need several of the RPM for a complete installation. Binary RPM fall into several categories. To download and install a complete package requires the appropriate RPM from each of the several categories below, as applicable. Some release packages do not provide RPMs in each of the several categories.

To install from Binary RPM, you will need all of the following kernel independent packages for your architecture, and one of the kernel-dependent packages from the next section.

Independent RPM

Independent RPM are not dependent on the Linux kernel version. For example, the source package `LiS-source-2.18.6-1.7.2.noarch.rpm', is not dependent on kernel.

All of the following kernel independent RPM are required for your architecture. Binary RPMs listed here are for example only: additional binary RPMs are available from the downloads site. If your architecture is not available, you can build binary RPM from the source RPM (see see Building from the Source RPM).

Architecture Independent

LiS-dev-2.18.6-1.7.2.noarch.rpm

The LiS-dev package contains the device definitions necessary to run applications programs developed for Linux STREAMS (LiS).¹⁵

LiS-doc-2.18.6-1.7.2.noarch.rpm

The LiS-doc package contains this manual in plain text, postscript, pdf and html forms, along with the meta-information from the LiS package. It also contains all of the manual pages necessary for developing Linux STREAMS (LiS) applications and Linux STREAMS (LiS) STREAMS modules or drivers.

LiS-init-2.18.6-1.7.2.noarch.rpm

The LiS-init package contains the init scripts and provides the `postinst' scripts necessary to create kernel module preloads and modules definitions for all kernel module `core' subpackages.

LiS-source-2.18.6-1.7.2.noarch.rpm

The LiS-source package contains the source code necessary for building the Linux STREAMS (LiS) release. It includes the autoconf(1) configuration utilities necessary to create and distribute tarballs, rpm and deb/dsc.

Architecture Dependent

LiS-devel-2.18.6-1.7.2.i686.rpm

The LiS-devel package contains library archives for static compilation, header files to develop Linux STREAMS (LiS) modules and drivers. This also includes the header files and static libraries required to compile Linux STREAMS (LiS) applications programs.

LiS-lib-2.18.6-1.7.2.i686.rpm

The LiS-lib package contains the run-time shared libraries necessary to run application programs and utilities developed for the LiS package.

LiS-util-2.18.6-1.7.2.i686.rpm

The LiS-util package provides administrative and configuration test utilities and commands associated with the Linux STREAMS (LiS) package.

Kernel-Dependent RPM

Kernel-Dependent RPM are dependent on specific Linux Kernel Binary RPM releases. Packages are provided for popular released RedHat kernels. Packages dependent upon RedHat or other kernel RPM will have the `_kversion' kernel package version in the package name.

One of the following Kernel-Dependent packages is required for your architecture and kernel version. If your architecture or kernel version is not on the list, you can build binary RPM from the source RPM (see see Building from the Source RPM).¹⁶

LiS-core-2.4.20-28.7-2.18.6-1.7.2.i686.rpm

The LiS-core package contains the loadable kernel modules that depend only on the kernel. This package is heavily tied to the kernel for which it was compiled. This particular package applies to kernel version `2.4.20-28.7'.¹⁷

LiS-info-2.4.20-28.7-2.18.6-1.7.2.i686.rpm

The LiS-info package¹⁸ contains the module symbol version information for the core subpackage, above. It is possible to load this subpackage and compile modules that use the exported symbols without loading the actual kernel modules (from the core subpackage above). This package is heavily tied to the kernel for which it was compiled. This particular package applies to kernel version `2.4.20-28.7'.¹⁹

Configuration and Installation

To configure, build and install the binary RPM, See Configuring the Binary RPM.

6.1.2 Downloading the Debian DEB

To install from binary DEB, you will need several of the DEB for a complete installation. Binary DEB fall into several categories. To download and install a complete package requires the appropriate DEB from each of the several categories below, as applicable. Some release packages do not provide DEBs in each of the several categories.

To install from Binary DEB, you will need all of the following kernel independent packages for your architecture, and one of the kernel-dependent packages from the next section.

Independent DEB

Independent DEB are not dependent on the Linux kernel version. For example, the source package `LiS-source_2.18.6-0_i386.deb', is not dependent on kernel.

All of the following kernel independent DEB are required for your architecture. Binary DEBs listed here are for example only: additional binary DEBs are available from the downloads site. If your architecture is not available, you can build binary DEB from the Debian DSC (see see Building from the Debian DSC).

Architecture Independent

LiS-dev_2.18.6-0_all.deb

The LiS-dev package contains the device definitions necessary to run applications programs developed for Linux STREAMS (LiS). ²⁰

LiS-doc_2.18.6-0_all.deb

The LiS-doc package contains this manual in plain text, postscript, pdf and html forms, along with the meta-information from the LiS package. It also contains all of the manual pages necessary for developing Linux STREAMS (LiS) applications and Linux STREAMS (LiS) STREAMS modules or drivers.

LiS-init_2.18.6-0_all.deb

The LiS-init package contains the init scripts and provides the postinst scripts necessary to create kernel module preloads and modules definitions for all kernel module `core' subpackages.

LiS-source_2.18.6-0_all.deb

The LiS-source package contains the source code necessary for building the Linux STREAMS (LiS) release. It includes the autoconf(1) configuration utilities necessary to create and distribute tarballs, rpms and deb/dscs. !ignore ²¹ !end ignore

Architecture Dependent

LiS-devel_2.18.6-0_i386.deb

The LiS-devel package contains library archives for static compilation, header files to develop Linux STREAMS (LiS) modules and drivers. This also includes the header files and static libraries required to compile Linux STREAMS (LiS) applications programs.

LiS-lib_2.18.6-0_i386.deb

The LiS-lib package contains the run-time shared libraries necessary to run application programs and utilities developed for the LiS package.

Kernel-Dependent DEB

Kernel-Dependent DEB are dependent on specific Linux Kernel Binary DEB releases. Packages are provided for popular released Debian kernels. Packages dependent upon Debian or other kernel DEB will have the `_kversion' kernel package version in the package name.

One of the following Kernel-Dependent packages is required for your architecture and kernel version. If your architecture or kernel version is not on the list, you can build binary DEB from the source DEB (see see Building from the Debian DSC).²²

LiS-core-2.4.20-28.7_2.18.6-0_i386.deb

The LiS-core package contains the loadable kernel modules that depend only on the kernel. This package is heavily tied to the kernel for which it was compiled. This particular package applies to kernel version `2.4.20-28.7'.²³

LiS-info-2.4.20-28.7_2.18.6-0_i386.deb

The LiS-info package²⁴ contains the module symbol version information for the core subpackage, above. It is possible to load this subpackage and compile modules that use the exported symbols without loading the actual kernel modules (from the core subpackage above). This package is heavily tied to the kernel for which it was compiled. This particular package applies to kernel version `2.4.20-28.7'.²⁵

Configuration and Installation

To configure, build and install the Debian DEB, See Configuring the Debian DEB.

6.1.3 Downloading the Source RPM

If you cannot obtain a binary RPM for your architecture, or would like to roll you own binary RPM, download the following source RPM.

LiS-2.18.6-1.src.rpm

This is the source RPM for the package. From this source RPM it is possible to build binary RPM for any supported architecture and for any 2.4 or 2.6 kernel.

Configuration

To configure the source RPM, See Configuring the Source RPM.

6.1.4 Downloading the Debian DSC

If you cannot obtain a binary DEB for your architecture, or would like to roll your own DEB, download the following Debian DSC.

LiS_2.18.6-0.dsc

LiS_2.18.6-0.tar.gz

This is the Debian DSC for the package. From this Debian DSC it is possible to build binary DEB for any supported architecture and for any 2.4 or 2.6 kernel.

Configuration

To configure the source RPM, See Configuring the Debian DSC.

6.1.5 Downloading the Tar Ball

For non-rpm(1) architectures, such as NexusWare embedded target, download the tarball as follows:

LiS-2.18.6.tar.gz

LiS-2.18.6.tar.bz2

These are the tar(1) balls for the release. These tar(1) balls contain the autoconf(1) distribution which includes all the source necessary for building and installing the package. These tarballs will even build Source RPM and Binary RPM on rpm(1) architectures and Debian DSC and DEB on dpkg(1) architectures.

The tar ball may be downloaded easily with wget(1) as follows:

     % wget http://www.openss7.org/LiS-2.18.6.tar.bz2

or

     % wget http://www.openss7.org/LiS-2.18.6.tar.gz

Note that you will need an OpenSS7 Project user name and password to download release candidates (which are only available to subscribers and sponsors of the OpenSS7 Project).

Unpacking the Archive

After downloading one of the tar balls, unpack the archive using one of the following commands:

     % wget http://www.openss7.org/LiS-2.18.6.tar.gz
     % tar -xzvf LiS-2.18.6.tar.gz

or

     % wget http://www.openss7.org/LiS-2.18.6.tar.bz2
     % tar -xjvf LiS-2.18.6.tar.bz2

Either will create a subdirectory name LiS-2.18.6 containing all of the files and subdirectories for the LiS package.

Configuration

To configure and install the tar ball, See Configuring the Tar Ball.

6.1.6 Downloading from CVS

If you are a subscriber or sponsor of The OpenSS7 Project with CVS archive access privileges then you can download release, mid-release or release candidate versions of the LiS package from the project CVS archive.

The Linux STREAMS (LiS) package is located in the LiS module of /var/cvs. For release tag information, see Releases.

To access the archive from the project CVS pserver, use the following commands to check out a version from the archive:

     % export CVSROOT='-d:pserver:username@cvs.openss7.com:2401/var/cvs'
     % cvs login
     Password: *********
     % cvs co -r LiS_2.18.6 LiS
     % cvs logout

It is, of course, possible to check out by date or by other criteria. For more information, see cvs(1).

Preparing the CVS Working Directory

Although public releases of the LiS package do not require reconfiguration, creating a configurable directory from the CVS archive requires tools not normally distributed with the other releases.

The build host requires the following GNU tools:

• autoconf 2.61
• automake 1.10
• libtool 1.5.22
• gettext 0.16.1

These tools can be acquired from the FSF website in the free software directory, and also at the following locations:

• autoconf-2.61
• automake-1.10
• libtool-1.5.22
• gettext-0.16.1

It should be stressed that, in particular, the autoconf(1), and automake(1), must be at version releases 2.61 and 1.10. The versions normally distributed in some mainstream GNU/Linux distributions are, in fact, much older than these versions.²⁶ GNU version of these packages configured and installed to default directories will install in /usr/local/ allowing them to coexist with distribution installed versions.

For building documentation, the build host also requires the following documentation tools:

• gs 8.15
• tetex 3.0
• texinfo 4.8
• transfig 3.2.5
• imagemagick 6.2.4
• groff 1.17.2

Most desktop GNU/Linux distributions will have these tools; however, some server-style installations (e.g. Ubuntu-server or SLES 9) will not and they must be installed separately.

For uncooked manual pages, the entire groff(1) package is required on Debian and Ubuntu systems (the base package does not include grefer(1) which is used extensively by uncooked manual pages). The following will get what you need:

     Debian: % apt-get install groff_ext
     Ubuntu: % apt-get install groff

In addition, the build host requires a complete tool chain for compiling for the target host, including kernel tools such as genksyms(8) and others.

If you wish to package rpms on an rpm(1) system, or debs on a dpkg(1) system, you will need the appropriate tool chain. Systems based on rpm(1) typically have the necessary tool chain available, however, dpkg(1) systems do not. The following on a Debian or Ubuntu system will get what you need:

     % apt-get install debhelper
     % apt-get install fakeroot

To generate a configuration script and the necessary scriptlets required by the GNU autoconf(1) system, execute the following commands on the working directory:

     % autoreconf -fiv LiS

where, LiS is the name of the directory to where the working copy was checked out under the previous step. This command generates the configure script and other missing pieces that are normally distributed with the release Tar Balls, SRPMs and DSCs.

Make sure that `autoreconf --version' returns `2.61'. Otherwise, you may need to perform something like the following:

     % PATH="/usr/local/bin:$PATH"
     % autoreconf -fiv LiS

After reconfiguring the directory, the package can then be configured and built using the same instructions as are used for the Tar Ball, see Configuring the Tar Ball, and Building from the Tar Ball.

Do note, however, that make(1) will rebuild the documentation that is normally released with the package. Additional tools may be necessary for building the documentation. To avoid building and installing the documentation, use the --disable-devel option to configure described in Configuring the Tar Ball.

When configuring the package in a working directory and while working a change-compile-test cycle that involves configuration macros or documentation, I find it of great advantage to invoke the GNU configure options --enable-maintainer-mode, --enable-dependency-tracking and --disable-devel. The first of these three options will add maintainer-specific targets to any generated Makefile, the second option will invoke automatic dependency tracking within the Makefile so rebuilds after changes to macro, source or documentation files will be automatically rebuilt; and the last option will suppress rebuilding and reinstalling documentation manual pages and header files. Header files will still be available under the /usr/src directory.

6.2 Configuration

6.2.1 Configuring the Binary RPM

In general the binary RPM do not require any configuration, however, during installation it is possible to relocate some of the installation directories. This allows some degree of customization. Relocations that are available on the binary RPM are as follows:

LiS-core-2.4.20-28.7-2.18.6-1.7.2.i686.rpm

/lib/modules/2.4.20-28.7

This relocatable directory contains the kernel modules that provide the LiS core, drivers and modules.²⁷

LiS-info-2.4.20-28.7-2.18.6-1.7.2.i686.rpm

/usr/include/LiS/2.4.20-28.7

This relocatable directory contains the kernel module exported symbol information that allows other kernel modules to be compiled against the correct version of the LiS package.²⁸

LiS-dev-2.18.6-1.7.2.i686.rpm

(not relocatable)

LiS-devel-2.18.6-1.7.2.i686.rpm

/usr/lib

This relocatable directory contains LiS libraries.

/usr/include/LiS

This relocatable directory contains LiS header files.

LiS-doc-2.18.6-1.7.2.i686.rpm

/usr/share/doc

This relocatable directory contains all package specific documentation (including this manual). The subdirectory in this directory is the LiS-2.18.6 directory.

/usr/share/info

This relocatable directory contains info files (including the info version of this manual).

/usr/share/man

This relocatable directory contains manual pages.

LiS-lib-2.18.6-1.7.2.i686.rpm

/usr/lib

This relocatable directory contains the run-time shared libraries necessary to run applications programs and utilities developed for Linux STREAMS (LiS).

/usr/share/locale

This relocatable directory contains the locale information for shared library files.

LiS-source-2.18.6-1.7.2.i686.rpm

/usr/src

This relocatable directory contains the source code.

LiS-util-2.18.6-1.7.2.i686.rpm

/usr/bin

This relocatable directory contains binary programs and utilities.

/usr/sbin

This relocatable directory contains system binary programs and utilities.

/usr/libexec

This relocatable directory contains test programs.

/etc

This relocatable directory contains init scripts and configuration information.

Installation

To install the binary RPM, See Installing the Binary RPM.

6.2.2 Configuring the Debian DEB

In general the binary DEB do not require any configuration.

Installation

To install the Debian DEB, See Installing the Debian DEB.

6.2.3 Configuring the Source RPM

When building from the source RPM (see Building from the Source RPM), the rebuild process uses a number of macros from the user's .rpmmacros file as described in rpm(8).

Following is an example of the ~/.rpmmacros file that I use for rebuilding RPMS:

     #
     # RPM macros for building rpms
     #
     
     %_topdir /usr/src/openss7.rpms
     
     %vendor OpenSS7 Corporation
     %distribution OpenSS7
     %disturl http://www.openss7.org/
     %packager Brian Bidulock <bidulock@openss7.org>
     %url http://www.openss7.org/
     
     %_signature gpg
     %_gpg_path /home/brian/.gnupg
     %_gpg_name openss7@openss7.org
     %_gpgbin /usr/bin/gpg
     
     %_source_payload w9.bzdio
     %_binary_payload w9.bzdio
     
     %_unpackaged_files_terminate_build 1
     %_missing_doc_files_terminate_build 1
     %_enable_debug_packages 1
     
     #
     # Template for debug information sub-package.
     # with our little addition of release
     #
     %debug_package \
     %ifnarch noarch\
     %global __debug_package 1\
     %package debug\
     Summary: Debug information for package %{name}\
     Group: Development/Debug\
     AutoReqProv: 0\
     %{?fullrelease:Release: %{fullrelease}}\
     %description debug\
     This package provides debug information for package %{name}.\
     Debug information is useful when developing applications that use this\
     package or when debugging this package.\
     %files debug -f debugfiles.list\
     %defattr(-,root,root)\
     %endif\
     %{nil}
     

When building from the source RPM (see Building from the Source RPM), it is possible to pass a number of additional configuration options to the rpmbuild(1) process.

The additional configuration options are described below.

Note that distributions that use older versions of rpm do not have the --with or --without options defined. To achieve the same effect as:

     --with someparm=somearg

do:

     --define "_with_someparm --with-someparm=somearg"

--define "_kversion $PACKAGE_KVERSION"

Specifies the kernel version other than the running kernel for which to build. If _kversion is not defined when rebuilding, the environment variable PACKAGE_KVERSION is used. If the environment variable PACKAGE_KVERSION is not defined, then the version of the running kernel (i.e. discovered with `uname -r') is used as the target version for kernel-dependent packages. This option can also be defined in an .rpmspec file using the macro name `_kversion'.

--with checks

--without checks

Enable or disable preinstall checks. Each packages supports a number of preinstall checks that can be performed by invoking the `check' target with automake(1). These currently consist of checking each kernel module for unresolved kernel symbols, checking for documentation for exported kernel module symbols, checking for documentation for exported library symbols, checking for standard options for build and installable programs, checking for documentation for built and installable programs. Normally these checks are only run in maintainer mode, but can be enabled and disabled with this option.

--with k-optimize=HOW

--without k-optimize

Specify `HOW' optimization, normal, size, speed or quick. size compiles kernel modules -Os, speed compiles kernel modules -O3, and quick compiles kernel modules -O0. The default is normal. Use with care.

--with cooked-manpages

--without cooked-manpages

Some systems do not like grefer(1) references in manual pages.²⁹ This option will cook soelim(1), refer(1), tbl(1) and pic(1) commands from the manual pages and also strip groff(1) comments. The default is to leave manual pages uncooked: they are actually smaller that way.

--with public

--without public

Release public packages or private packages. This option has no effect on the LiS package. The default is to release public packages.

--with k-debug

--without k-debug

Specifies whether kernel debugging is to be performed on the build kernel modules. Mutually exclusive with test and safe below. This has the effect of removing static and inline attributes from functions and invoking all debugging macros in the code. The default is to not perform kernel debugging.

--with k-test

--without k-test

Specifies whether kernel testing is to be performed. Mutually exclusive with debug above and safe below. This has the effect of removing static and inline attributes from functions and invoking most debugging macros in the code. The default is to not perform kernel testing.

--with k-safe

--without k-safe

Specifies whether kernel saftey is to be performed. Mutually exclusive with debug and test above. This has the effect of invoking some more pedantic assertion macros in the code. The default is not to apply kernel safety.

--with k-inline

--without k-inline

Specifies whether kernel inline functions are to be placed inline. This has the effect of adding the -finline-functions flag to CFLAGS for compiling kernel modules. Linux 2.4 kernels are normally compiled -O2 which does not respect the inline directive. This compiles kernel modules with -finline-functions to get closer to -O3 optimization. For better optimization controls, See Configuring the Tar Ball.

--with k-modversions

--without k-modversions

Specifies whether kernel symbol versions are to be applied to symbols exported by package kernel modules. The default is to version exported module symbols. This package does not export symbols so this option has no effect.

--with devfs

--without devfs

Specifies whether the build is for a device file system daemon enabled system with autoloading, or not. The default is to build for devfsd(1) autoloading when CONFIG_DEVFS_FS is defined in the target kernel. The `rebuild' target uses this option to signal to the RPM spec file that the `dev' subpackage need not be built. This option does not appear when the package has no devices.

--with devel

--without devel

Specifies whether to build development environment packages such as those that include header files, static libraries, manual pages and texinfo(1) documentation. The default is to build development environment packages. This option can be useful when building for an embedded target where only the runtime components are desired.

--with tools

--without tools

Specifies whether user space packages are to be built. The default is to build user space packages. This option can be useful when rebuilding for multiple architectures and target kernels. The `rebuild' automake(1) target uses this feature when rebuilding for all available architectures and kernels, to rebuild user packages once per architecture instead of once per kernel.

--with modules

--without modules

Specifies whether kernel modules packages are to be built. The default is to build kernel module packages. This option can be useful when rebuilding for multiple architectures and target kernels. The `rebuild' automake(1) target uses this feature to rebuild for all available architectures and kernels.

In addition, the following rpm options, specific to the Linux STREAMS (LiS) package are available:

--with broken-cpu-flags

Specify int instead of long interrupt flags. Linux uses long interrupt flags exclusively. The default is to use long interrupt flags. This option defaults to `disabled'.

--without lis-development

This option is new to LiS-2.18.2. Disable reporting of source code path traces in debug logs. Linux STREAMS (LiS) passes file name and line number arguments to many of the exported functions. This is exhaustive of stack resources and requires the passing of more than three arguments on the stack in many circumstances. Note that disabling code path traces breaks binary compatibility. The default is perform source code path traces, unless optimized for size or speed, in which case, the default is to not perform source code path traces. This option defaults to `enabled'.

--with k-timers

Invoke Linux kernel caching on timer structures. This increases speed and efficiency but is unsafe on SMP architectures and for buggy drivers. The default is to perform kernel caching on timer structures. This option defaults to `disabled'.

--with k-cache

Invoke Linux kernel caching on primary structures. This increases speed and efficiency. The default is to perform kernel caching on primary structures. This option defaults to `disabled'.

--with atomic-stats

Specify atomic_t element types for elements of the module_stat structure instead of int element types. The default is to use int elements types for the module_stat structure. This option defaults to `disabled'.

--with lis-regparms=NUMBER

--without lis-regparms

This option is new to LiS-2.18.2. Specifies the number of parameters to pass in registers to Linux STREAMS (LiS) exported functions. Specifying other than `0' here may break binary compatibility. Specifying `--without lis-regparms' will default to the number used for kernel exported functions. Specifying greater than `3' will cause the build to fail. The default is to pass no arguments in registers. This option defaults to `0'.

--without solaris-consts

Attempt to be compatible with Solaris constants. The default is to attempt to be compatible with Solaris constants. This option defaults to `enabled'.

--without solaris-cmn_err

Specifies whether Solaris-style cmn_err is used. The Solaris style prints a newline at the end of each statement. The SVR 4.2 style prints a newline at the beginning of each statement unless CE_CONT is specified. Linux kernel logs use the former, so this defaults to Solaris-style. This option defaults to `enabled'.

In general, the default values of these options are sufficient for most purposes and no options need be provided when rebuilding the Source RPMs.

Build

To build from the source RPM, See Building from the Source RPM.

6.2.4 Configuring the Debian DSC

The Debian DSC can be configured by passing options in the environment variable BUILD_DEBOPTIONS. The options placed in this variable take the same form as those passed to the configure script, See Configuring the Tar Ball. For an example, See Building from the Debian DSC.

Build

To build from the Debian DSC, See Building from the Debian DSC.

6.2.5 Configuring the Tar Ball

All of the normal GNU autoconf(1) configuration options and environment variables apply. Additional options and environment variables are provided to tailor or customize the build and are described below.

6.2.5.1 Configure Options

Following are the additional configure options, their meaning and use:

--enable-checks

--disable-checks

Enable or disable preinstall checks. Each release package supports a number of preinstall checks that can be performed by invoking the `check' target with make(1). These currently consist of checking each kernel module for unresolved kernel symbols, checking for documentation for exported kernel module symbols, checking for documentation for exported library symbols, checking for standard options for build and installable programs, checking for documentation for built and installable programs. Normally these checks are only run in maintainer mode, but can be enabled and disabled with this option.

--disable-compress-manpages

Compress manual pages with `gzip -9' or `bzip2 -9' or leave them uncompressed. The default is to compress manual pages with `gzip -9' or `bzip2 -9' if a single compressed manual page exists in the target installation directory (--mandir). This disables automatic compression.

--disable-public

Disable public release. This option is not usable on public releases and only has a usable effect on Linux STREAMS (LiS) when the package is acquired from CVS. In particular, the STREAMS SS7/VoIP/ISDN/SIGTRAN Stacks (strss7-0.9a.7) release package has a large number of non-public components. Specifying this option will cause the package to build and install all private release components in addition to the public release components. This option affects all release packages. Most release packages do not have private release components.

--disable-initscripts

Disables the installation of init scripts. The default is to configure and install init scripts and their associated configuration files.

Although the default is to install init scripts, installation attempts to detect a System V init script configuration, and if one is not found, the init scripts are installed into the appropriate directories, but the symbolic links to the run level script directories are not generated and the script is not invoked. Therefore, it is safe to leave this option unchanged, even on distributions that do not support System V init script layout (such as NexusWare).

--disable-devel

Disables the installation of development environment components such as header files, static libraries, manual pages and texinfo(1) documentation. The default is to install development environment components. This option can be useful when configuring for an embedded target where only the runtime components are desired, or when performing a edit-compile-test cycle.

--enable-tools

Specifies whether user space programs and libraries are to be built and installed. The default is to build and install user space programs and libraries. This option can be useful when rebuilding for multiple architectures and target kernels, particularly under rpm(1) or dpkg(1). The `rebuild' automake(1) target uses this feature when rebuilding RPMs for all available architectures and kernels, to rebuild user packages once per architecture instead of once per kernel.

--enable-modules

Specifies whether kernel modules are to be built and installed. The default is to build and install kernel modules. This option can be useful when rebuilding for multiple architectures and target kernels, particularly under rpm(1) or dpkg(1). The `rebuild' automake(1) target uses this feature to rebuild for all available architectures and kernels.

--enable-arch

Specifies whether architectural dependent package components are to be built and installed. This option can be useful when rebuilding for multiple architectures and target kernels, particularly under dpkg(1). The default is to configure, build and install architecture dependent package components.

--enable-indep

Specifies whether architecture independent package components are to be built and installed. This option can be useful when rebuilding for multiple architectures and target kernels, particularly under dpkg(1). The default is to configure, build and install architecture independent package components.

--enable-k-inline

Enable kernel inline functions. Most Linux kernels build without -finline-functions. This option adds the -finline-functions and -Winline flags to the compilation of kernel modules. Use with care.

--enable-k-safe

Enable kernel module run-time safety checks. Specifies whether kernel safety is to be performed. This option is mutually exclusive with --enable-k-test and --enable-k-debug below. This has the effect of invoking some more pedantic assertion macros in the code. The default is not to apply kernel safety.

--enable-k-test

Enable kernel module run-time testing. Specifies whether kernel testing is to be performed. This option is mutually exclusive with --enable-k-safe above and --enable-k-debug below. This has the effect of remove static and inline attributes from functions and invoking most non-performance affecting debugging macros in the code. The default is not to perform kernel testing.

--enable-k-debug

Enable kernel module run-time debugging. Specifies whether kernel debugging is to be performed. This option is mutually exclusive with --enable-k-safe and --enable-k-test above. This has the effect of removing static and inline attributes from functions and invoking all debugging macros in the code (including performance-affecting debug macros). The default is to not perform kernel debugging.

--enable-devfs

--disable-devfs

Specifies whether the build is for a device file system daemon enabled system with autoloading, or not. The default is to build for devfsd(8) autoloading when CONFIG_DEVFS_FS is defined in the target kernel. The `reuild' automake(1) target uses this option to signal to the RPM spec file that the `dev' subpackage need not be built. This option has no effect for release packages that do not provide devices.

--with-gpg-user=GNUPGUSER

Specify the gpg(1) `GNUPGUSER' for signing RPMs and tarballs. The default is the content of the environment variable GNUPGUSER. If unspecified, the gpg(1) program will normally use the user name of the account invoking the gpg(1) program. For building source RPMs, the RPM macro `_gpg_name' will override this setting.

--with-gpg-home=GNUPGHOME

Specify the `GNUPGHOME' directory for signing RPMs and tarballs. The default is the user's ~/.gpg directory. For building source RPMs, the RPM macro `_gpg_path' will override this setting.

--with-pkg-epoch=EPOCH

Specifies the epoch for the package. This is neither used for rpm(1) nor dpkg(1) packages, it applies to the tarball release as a whole. The default is the contents of the .pkgepoch file in the release package source directory or, if that file does not exist, zero (0).

--with-pkg-release=RELEASE

Specifies the release for the package. This is neither used for rpm(1) nor dpkg(1) packages, it applies to the tarball release as a whole. The default is the contents of the .pkgrelease file in the release package source directory or, if that file does not exist, one (1). This is the number after the last point in the package version number.

--with-pkg-distdir=DIR

Specifies the distribution directory for the package. This is used by the maintainer for building distributions of tarballs. This is the directory into which archives are copied for distribution. The default is the top build directory.

--with-cooked-manpages

Convert manual pages to remove macro dependencies and grefer(1) references. Some systems do not like grefer(1) references in manual pages.³⁰ This option will cook soelim(1), refer(1), tbl(1) and pic(1) commands from the manual pages and also strip groff(1) comments. The default is to leave manual pages uncooked (they are actually smaller that way).

--with-rpm-epoch=PACKAGE_EPOCH

Specify the `PACKAGE_EPOCH' for the RPM spec file. The default is to use the RPM epoch contained in the release package file .rpmepoch.

--with-rpm-release=PACKAGE_RPMRELEASE

Specify the `PACKAGE_RPMRELEASE' for the RPM spec file. The default is to use the RPM release contained in the release package file .rpmrelease.

--with-rpm-extra=PACKAGE_RPMEXTRA

Specify the `PACKAGE_RPMEXTRA' extra release information for the RPM spec file. The default is to use the RPM extra release information contained in the release package file .rpmextra. Otherwise, this value will be determined from automatic detection of the RPM distribution.

--with-rpm-topdir=PACKAGE_RPMTOPDIR

Specify the `PACKAGE_RPMTOPDIR' top directory for RPMs. If specified with a null `PACKAGE_RPMTOPDIR', the default directory for the RPM distribution will be used. If this option is not provided on the command line, the top build directory will be used as the RPM top directory as well.

--with-deb-epoch=EPOCH

Specify the `PACKAGE_DEBEPOCH' for the DEB control file. The default is to use the DEB epoch contained in the release package file .debepoch.

--with-deb-release=RELEASE

Specify the `PACKAGE_DEBRELEASE' for the DEB control file. The default is to use the DEB release contained in the release package file .debrelease.

--with-deb-topdir=DIR

Specify the `PACKAGE_DEBTOPDIR' top directory for DEBs. If specified with a null `PACKAGE_DEBTOPDIR', the default directory for the DEB distribution will be used. If this option is not provided on the command line, the top build directory will be used as the DEB top directory as well.

--with-k-release=PACKAGE_KRELEASE

Specify the `PACKAGE_KRELEASE' release of the Linux kernel for which the build is targeted. When not cross compiling, if this option is not set, the build will be targeted at the kernel running in the build environment (e.g., `uname -r'). When cross-compiling this option must be specified or the configure script will generate an error and terminate.

--with-k-linkage=PACKAGE_KLINKAGE

Specify the `PACKAGE_KLINKAGE' for kernel module linkage. This can be one of the following:

• `loadable' – loadable kernel modules
• `linkable' – linkable kernel objects

The default is to build loadable kernel modules.

--with-k-modules=K-MODULES-DIR

Specify the `K-MODULES-DIR' directory to which kernel modules will be installed. The default is based on the option --with-k-release, --with-k-prefix and --with-k-rootdir. The default is DESTDIR/K-MODULES-DIR which is typically DESTDIR/lib/modules/PACKAGE_KRELEASE/. This directory is normally located by the configure script and need only be provided for special cross-build environments or when requested by a configure script error message.

--with-k-build=K-BUILD-DIR

Specify the `K-BUILD-DIR' base kernel build directory in which configured kernel source resides. The default is DESTDIR/K-MODULES-DIR/build. This directory is normally located by the configure script and need only be provided for special cross-build environments or when requested by a configure script error message.

--with-k-source=K-SOURCE-DIR

Specify the `K-SOURCE-DIR' base kernel build directory in which configured kernel source resides. The default is DESTDIR/K-MODULES-DIR/source. This directory is normally located by the configure script and need only be provided for special cross-build environments or when requested by a configure script error message.

--with-k-modver=K-MODVER-FILE

Specify the `K-MODVER-FILE' kernel module versions file. The default is K-BUILD-DIR/Module.symvers. This file is normally located by the configure script and need only be provided for special cross-build environments or when requested by a configure script error message.

--with-k-sysmap=K-SYSMAP-FILE

Specify the `K-SYSMAP-FILE' kernel system map file. The default is K-BUILD-DIR/System.map. This file is normally located by the configure script and need only be provided for special cross-build environments or when requested by a configure script error message.

--with-k-archdir=K-ARCHDIR

Specify the `K-ARCHDIR' kernel source architecture specific directory. The default is DESTDIR/K-SOURCE-DIR/arch. This directory is normally located by the configure script and need only be provided for special cross-build environments or when requested by a configure script error message.

--with-k-machdir=K-MACHDIR

Specify the `K-MACHDIR' kernel source machine specific directory. The default is DESTDIR/K-SOURCE-DIR/target_cpu. This directory is normally located by the configure script and need only be provided for special cross-build environments or when requested by a configure script error message.

--with-k-config=K-CONFIG

Specify the `K-CONFIG' kernel configuration file. The default is BOOT/config-K-RELEASE. This configuration file is normally located by the configure script and need only be provided for special cross-build environments or when requested by a configure script error message.

--with-k-optimize=HOW

--without-k-optimize

Specify `HOW' optimization, normal, size, speed or quick. size compiles kernel modules -Os, speed compiles kernel modules -O3, and quick compiles kernel modules -O0. The default is normal. Use with care. The most common use of this option is to specify --with-k-optimize=speed --disable-k-safe to compile for maximum performance. Nevertheless, even these setting are ricing and the resulting kernel modules will only be about 5% faster.

--with-strconf-master=STRCONF_CONFIG

Specify the `STRCONF_CONFIG' file name to which the configuration master file is written. The default is Config.master.

--with-base-major=STRCONF_MAJBASE

Start numbering for major devices at `STRCONF_MAJBASE'. The default is `230'.

In addition, the following configure options, specific to the Linux STREAMS (LiS) package are available:

--enable-user-mode

Enable user mode build. The option defaults to `disabled'. Don't use this option, it hasn't been tested.

--enable-broken-cpu-flags

Specify int instead of long interrupt flags. Linux uses long interrupt flags exclusively. The default is to use long interrupt flags. This option defaults to `disabled'.

--disable-lis-development

This option is new to LiS-2.18.2. Disable reporting of source code path traces in debug logs. Linux STREAMS (LiS) passes file name and line number arguments to many of the exported functions. This is exhaustive of stack resources and requires the passing of more than three arguments on the stack in many circumstances. Note that disabling code path traces breaks binary compatibility. The default is perform source code path traces, unless optimized for size or speed, in which case, the default is to not perform source code path traces. This option defaults to `enabled'.

--enable-k-timers

Invoke Linux kernel cahcing on timer structures. This increases speed and efficiency but is unsafe on SMP architectures and for buggy or high-speed drivers. The default is to not perform kernel caching on timer structures. This option defaults to `disabled'.

--enable-k-cache

Invoke Linux kernel caching on primary structures. This increases speed and efficiency. The default is to not perform kernel caching on primary structures. This option defaults to `disabled'.

--enable-atomic-stats

Enable atomic statistics elements for STREAMS statistics. The changes STREAMS statistics elements from the standard int to atomic_t to permit atomic increment of statistics structures. This is not recommended and the default is to not use atomic_t for statistics structures. This option defaults to `disabled'.

--with-lis-regparms=NUMBER

--without-lis-regparms

This option is new to LiS-2.18.2. Specifies the number of parameters to pass in registers to Linux STREAMS (LiS) exported functions. Specifying other than `0' here may break binary compatibility. Specifying `--without-lis-regparms' will default to the number used for kernel exported functions. Specifying greater than `3' will cause the build to fail. The default is to pass no arguments in registers. This option defaults to `0'.

--without-solaris-consts

Attempt to be compatible with Solaris constants. The default is to attempt to be compatible with Solaris constants. This option defaults to `enabled'.

--without-solaris-cmn_err

Specifies whether Solaris-style cmn_err is used. The Solaris style prints a newline at the end of each statement. The SVR 4.2 style prints a newline at the beginning of each statement unless CE_CONT is specified. Linux kernel logs use the former, so this defaults to Solaris-style. This option defaults to `enabled'.

6.2.5.2 Environment Variables

Following are additional environment variables to configure, their meaning and use:

GPG

GPG signature command. This is used for signing distributions by the maintainer. By default, configure will search for this tool.

GNUPGUSER

GPG user name. This is used for signing distributions by the maintainer.

GNUPGHOME

GPG home directory. This is used for signing distributions by the maintainer.

GPGPASSWD

GPG password for signing. This is used for signing distributions by the maintainer. This environment variable is not maintained by the configure script and should only be used on an isolated system.

SOELIM

Roff source elimination command, soelim(1). This is only necessary when the option --with-cooked-manpages has been specified and configure cannot find the proper soelim(1) command. By default, configure will search for this tool.

REFER

Roff references command, refer(1). This is only necessary when the option --with-cooked-manpages has been specified and configure cannot find the proper refer(1) command. By default, configure will search for this tool.

TBL

Roff table command, tbl(1). This is only necessary when the option --with-cooked-manpages has been specified and configure cannot find the proper tbl(1) command. By default, configure will search for this tool.

PIC

Roff picture command, pic(1). This is only necessary when the option --with-cooked-manpages has been specified and configure cannot find the proper pic(1) command. By default, configure will search for this tool.

GZIP

Default compression options provided to GZIP_CMD.

GZIP_CMD

Manpages (and kernel modules) compression commands, gzip(1). This is only necessary when the option --without-compressed-manpages has not been specified and configure cannot find the proper gzip(1) command. By default, configure will search for this tool.

BZIP2

Default compression options provided to BZIP2_CMD

BZIP2_CMD

Manpages compression commands, bzip2(1). This is only necessary when the option --without-compressed-manpages has not been specified and configure cannot find the proper bzip2(1) command. By default, configure will search for this tool.

MAKEWHATIS

Manpages apropros database rebuild command, makewhatis(8). By default, configure will search for this tool. By default, configure will search for this tool.

CHKCONFIG

Chkconfig command, chkconfig(8). This was used for installation of init scripts. All packages now come with init_install(8) and init_remove(8) scripts used to install and remove init scripts on both RPM and Debian systems.

RPM

Rpm command, rpm(1). This is only necessary for RPM builds. By default, configure will search for this tool.

RPMBUILD

Build RPM command, rpmbuild(1). This is only necessary for RPM builds. By default, configure will search for this tool. rpm(1) will be used instead of rpmbuild(1) only if rpmbuild(1) cannot be found.

DPKG

Dpkg comand, dpkg(1). This command is used for building Debian packages. By default, configure will search for this tool.

DPKG_SOURCE

Dpkg-source command, dpkg-source(1). This command is used for building Debian dsc packages. By default, configure will search for this tool.

DPKG_BUILDPACKAGE

Dpkg-buildpackage command, dpkg-buildpackage(1). This command is used for building Debian deb packages. By default, configure will search for this tool.

DEB_BUILD_ARCH

Debian build architecture. This variable is used for building Debian packages. The default is the autoconf build architecture.

DEB_BUILD_GNU_CPU

Debian build cpu. This variable is used for building Debian packages. The default is the autoconf build cpu.

DEB_BUILD_GNU_SYSTEM

Debian build os. This variable is used for building Debian packages. The default is the autoconf build os.

DEB_BUILD_GNU_TYPE

Debian build alias. This variable is used for building Debian packages. The default is the autoconf build alias.

DEB_HOST_ARCH

Debian host architecture. This variable is used for building Debian packages. The default is the autoconf host architecture.

DEB_HOST_GNU_CPU

Debian host cpu. This variable is used for building Debian packages. The default is the autoconf host cpu.

DEB_HOST_GNU_SYSTEM

Debian host os. This variable is used for building Debian packages. The default is the autoconf host os.

DEB_HOST_GNU_TYPE

Debian host alias. This variable is used for building Debian packages. The default is the autoconf host alias.

LDCONFIG

Configure loader command, ldconfig(8). Command used to configure the loader when libraries are installed. By default, configure will search for this tool.

DESTDIR

Cross build root directory. Specifies the root directory for build and installation. For example, for NexusWare cross-builds, this is set to environment variable NEXUSWARE_PREFIX on configuration to point to the root of the cross-build tree for both configuration and installation.

DEPMOD

Build kernel module dependencies command, depmod(8). This is used during installation of kernel modules to a running kernel to rebuild the modules dependency database. By default, configure will search for this tool.

MODPROBE

Probe kernel module dependencies command, modprobe(8). This is used during installation of kernel modules to a running kernel to remove old modules. By default, configure will search for this tool.

LSMOD

List kernel modules command, lsmod(8). This is used during installation of kernel modules to a running kernel to detect old modules for removal. By default, configure will search for this tool.

LSOF

List open files command, lsof(1). This is used during installation of kernel modules to a running kernel to detect old modules for removal. Processes owning the old kernel modules will be killed and the module removed. If the process restarts, the new module will be demand loaded. By default, configure will search for this tool.

GENKSYMS

Generate kernel symbols command, genksyms(8). This is used for generating module symbol versions during build. By default, configure will search for this tool.

KGENKSYMS

Linux 2.6 generate kernel symbols command, genksyms(8). This is used for generating module symbol version during build. By default, configure will search for this tool.

OBJDUMP

Object dumping command, objdump(1). This is used for listing information about object files. By default, configure will search for this tool.

NM

Object symbol listing command, nm(1). This is used for listing information about object files. By default, configure will search for this tool.

MODPOST_CACHE

Cache file for modpost(1). The version of the modpost.sh script that ships with each package can cache information to a cache file to speed multiple builds. This environment variable is used to specify a cache file.

AUTOM4TE

Autom4te command, autom4te(1). This is the executable used by autotest for pre- and post-installation checks. By default, configure will search for this tool.

AUTOTEST

Autotest macro build command, autom4te(1). This is the executable used by autotest for pre- and post-installation checks. By default, configure will search for this tool.

6.2.5.3 Build

To build from the tar ball, See Building from the Tar Ball.

6.3 Building

6.3.1 Building from the Source RPM

If you have downloaded the necessary source RPM (see Downloading the Source RPM), then the following instructions will rebuild the binary RPMs on your system. Once the binary RPMs are rebuilt, you may install them as described above (see Installing the Binary RPM).

The source RPM is rebuilt to binary RPMs as follows:

     % wget http://www.openss7.org/rpms/SRPMS/LiS-2.18.6-1.src.rpm
     % rpmbuild --rebuild -vv LiS-2.18.6-1.src.rpm

The rebuild process can also recognize a number of options that can be used to tweak the resulting binaries, See Configuring the Source RPM. These options are provided on the rpm(1) command line. For example:

     % rpmbuild --rebuild -vv --target athlon-redhat-linux \
       --define "_kversion 2.4.20-28.7" \
       -- LiS-2.18.6-1.src.rpm

will rebuild binary RPM for the `2.4.20-28.7' kernel for the `athlon' architecture against the LiS STREAMS package. ³¹

Installation

To install the resulting binary RPM, See Installing the Binary RPM.

6.3.2 Building from the Debian DSC

If you have downloaded the necessary Debian DSC (see Downloading the Debian DSC), then the following instructions will rebuild the binary DEBs on your system. Once the binary DEBs are rebuilt, you may install them as described above (see Installing the Debian DEB).

The Debian DSC is rebuilt to binary DEBs as follows:

     % wget http://www.openss7.org/debian/LiS_2.18.6-0.dsc
     % wget http://www.openss7.org/debian/LiS_2.18.6-0.tar.gz
     % dpkg-buildpackage -v LiS_2.18.6-0.dsc

The rebuild process can also recognize a number of options that can be used to tweak the resulting binaries, See Configuring the Debian DSC. These options are provided in the environment variable BUILD_DPKGOPTIONS and have the same form as the options to configure, See Configuring the Tar Ball. For example:

     % BUILD_DEBOPTIONS='
             --with-k-release=2.4.20-28.7
             --host=athlon-debian-linux-gnu'
       dpkg-buildpackage -v \
       LiS_2.18.6-0.dsc

will rebuild binary DEB for the `2.4.20-28.7' kernel for the `athlon' architecture against the LiS STREAMS package. ³²

Installation

To install the resulting binary DEB, See Installing the Debian DEB.

6.3.3 Building from the Tar Ball

If you have downloaded the tar ball (see Downloading the Tar Ball), then the following instructions will rebuild the package on your system. (Note that the build process does not required root privilege.)

6.3.3.1 Native Build

Following is an example of a native build against the running kernel:

     % wget http://www.openss7.org/LiS-2.18.6.tar.bz2
     % tar -xjvf LiS-2.18.6.tar.bz2
     % pushd LiS-2.18.6
     % ./configure
     % make
     % popd

6.3.3.2 Cross-Build

Following is an example for a cross-build. The kernel release version must always be specified for a cross-build.³³ If you are cross-building, specify the root for the build with environment variable DESTDIR. The cross-compile host must also be specified if different from the build host. Either the compiler and other tools must be in the usual places where GNU autoconf(1) can find them, or they must be specified with declarations such as `CC=/u5/NexusWare24/ppc-linux/gcc' on the configure command line. Look in the file configure.nexusware in the release package for an example.

     % wget http://www.openss7.org/LiS-2.18.6.tar.bz2
     % tar -xjvf LiS-2.18.6.tar.bz2
     % pushd LiS-2.18.6
     % ./configure DESTDIR="/some/other/root" \
     	--with-k-release=2.4.18 --host sparc-linux
     % make
     % popd

6.3.3.3 NexusWare Build

Additional support is provided for cross-building for the Performance Technologies Inc. NexusWare embedded target for the CPC-384, CPC-388 and CPC-396 cards. A configuration script wrapper (configure.nexusware) is provided to simplify the cross-build operation for these targets. The following steps describe the process:

1. Follow the normal NexusWare instructions for rebuilding a generic kernel and flash image as follows: (Note that I keep my NexusWare build in /u5/NexusWare24.)

   % pushd /u5/NexusWare24 % source SETUP.sh % make % popd

   For more recent NexusWare releases, the method for rebuilding a kernel is a little different as follows:

   % pushd /u5/NexusWare80 % ./nexus 2.4 % ./nexus 8260 % ./nexus quick % . SETUP.sh % popd

2. Next download, unpack (see Downloading the Tar Ball) and configure (see Configuring the Tar Ball) using the provided configure.nexusware wrapper for configure. This wrapper simply tells the configure script where to find the NexusWare sources and which NexusWare cross-building tools to use for a cross-compile.³⁴

   Any of the normal configure script options (see Configuring the Tar Ball) can be used on the same line as `./configure.nexusware'. One of particular interest to embedded targets is --with-k-optimize=size to attempt to reduce the size of the kernel modules.

   You must specify the kernel version of the kernel for which you are configuring. Add the --with-k-release=2.4.18 option for older NexusWare releases, --with-k-release=2.4.25 or --with-k-release=2.6.12 for more current NexusWare releases.

3. Install as normal (see Installing the Tar Ball), however, for embedded targets the `install-strip' automake(1) target should be used instead of the `install' automake(1) target. The `install-strip' target will strip unnecessary symbols from kernel modules and further reduce the size in the root file system flash image.

Following is what I use for configuration and installation: (My NexusWare tree is rooted at /u5/NexusWare.)

     % pushd /u5/NexusWare80
     % ./nexus 2.4
     % ./nexus 8260
     % ./nexus quick
     % . SETUP.sh
     % popd
     % wget http://www.openss7.org/LiS-2.18.6.tar.bz2
     % tar -xjvf LiS-2.18.6.tar.bz2
     % pushd LiS-2.18.6
     % ./configure.nexusware --with-k-release=2.4.25 --with-k-optimize=size
     % make
     % make DESTDIR="$NEXUSWARE_PREFIX" install-strip
     % popd

Once built and installed in the NexusWare directory, you will have to (currently) hand edit a .spec file to include the components you want in the NexusWare root file system. If you are cross-building for NexusWare you should already know what that means. Objects that you might be interested in copying to the root file system are kernel modules that were installed in $NEXUSWARE_PREFIX/lib/modules/2.4.18/lis, libraries installed in $NEXUSWARE_PREFIX/usr/lib and utility functions installed in $NEXUSWARE_PREFIX/usr/bin and $NEXUSWARE_PREFIX/usr/sbin and test programs in $NEXUSWARE_PREFIX/usr/libexec. If you would prefer that these programs be installed in $NEXUSWARE_PREFIX/lib, $NEXUSWARE_PREFIX/bin, $NEXUSWARE_PREFIX/sbin and $NEXUSWARE_PREFIX/libexec, (say because you want to remote mount the /usr directory after boot), then specify the --exec-prefix=/ option to `./configure.nexusware'.

Because NexusWare does not include an /etc/modules.conf file by default, it will be necessary to add one or edit your rc.4 file to insmod(8) the necessary LiS modules at boot time.

NexusWare does not configure its kernels for CONFIG_KMOD, so any kernel modules must be loaded by the rc.4 init script at boot. On more recent NexusWare releases, the init scripts will be installed in $NEXUSWARE_PREFIX/etc/rc.d/init.d/ but you must manually edit your rc.4 script to invoke these scripts.

Once you have completed the necessary .spec and rc.4 file entries, you need to rebuild the `generic' kernel flash image once more for these objects to be included in the flash file system. It is important that this second build of the kernel image be the same as the first.

When modifying and rebuilding a NexusWare kernel, it will be necessary to rebuild and install LiS. Simply perform the last `make install-strip' stage or start again with `./configure.nexusware'. You can place the unpacked tarball in $NEXUSWARE_PREFIX/usr/src/LiS, and add the following to the top-level NexusWare Makefile to make the build process a single step process instead of dual pass:

     all:
     ...
             (cd kernels/generic; $(MAKE) depend)
             (cd usr/src/pcmcia-cs-3.2.1; $(MAKE) config)
             (cd kernels/generic; $(MAKE))
             (cd usr/src/pcmcia-cs-3.2.1; $(MAKE) pti)
             (cd usr/src/pti; $(MAKE))
             (cd drivers; $(MAKE))
             (cd utility; $(MAKE))
     #       uncomment for LiS build
     #       (cd usr/src/LiS; ./configure.nexusware; $(MAKE) install-strip)
             (cd build/generic; $(MAKE))
     ...

Another, perhaps simpler approach, is to make the necessary edits to the NexusWare top-level Makefile and .spec and rc.4 files, download and unpack the tar ball into the NexusWare directory, and build the NexusWare flash image as normal:

     % wget http://www.openss7.org/LiS-2.18.6.tar.bz2
     % pushd /u5/NexusWare24
     % source SETUP.sh
     % pushd usr/src
     % tar -xjvf ${DIRSTACK[2]}/LiS-2.18.6.tar.bz2
     % ln -sf LiS-2.18.6 LiS
     % popd
     % make
     % popd

The situation is a little more complex for recent NexusWare releases.

6.4 Installing

6.4.1 Installing the Binary RPM

If you have downloaded the necessary binary RPMs (see Downloading the Binary RPM), or have rebuilt binary RPMs using the source RPM (see Building from the Source RPM), then the following instructions will install the RPMs on your system. For additional information on rpm(1), see rpm(8).

     % pushd RPMS/i686
     % rpm -ihv LiS-*-2.18.6-1.7.2.i686.rpm

You must have the correct binary RPMs downloaded or built for this to be successful.

Some of the packages are relocatable and can have final installation directories altered with the --relocate option to rpm(1), see rpm(8). For example, the following will relocate the documentation and info directories:

     % pushd RPMS/i686
     % rpm -ihv \
             --relocate '/usr/share/doc=/usr/local/share/doc' \
             --relocate '/usr/share/info=/usr/local/share/info' \
             -- LiS-doc-2.18.6-1.7.2.i686.rpm

The previous example will install the LiS-doc package by will relocate the documentation an info directory contents to the /usr/local version.

6.4.2 Installing the Debian DEB

If you have downloaded the necessary Debian DEBs (see Downloading the Debian DEB), or have rebuild binary DEBs using the Debian DSC (see Building from the Debian DSC), then the following instructions will install the DEBs on your system. For additional information see dpkg(8).

     % pushd debian
     % dpkg -iv LiS-*_2.18.6-0_*.deb

You must have the correct .deb files downloaded or build for this to be successful.

6.4.3 Installing the Tar Ball

After the build process (see Building from the Tar Ball), installation only requires execution of one of two automake(1) targets:

`make install'

The `install' automake(1) target will install all the components of the package. Root privilege is required to successfully invoke this target.

`make install-strip'

The `install-strip' automake(1) target will install all the components of the package, but will strip unnecessary information out of the objects and compress manual pages. Root privilege is required to successfully invoke this target.

6.5 Removing

6.5.1 Removing the Binary RPM

To remove an installed version of the binary RPMs (whether obtained from the OpenSS7 binary RPM releases, or whether created by the source RPM), execute the following command:

     % rpm -evv `rpm -qa | grep '^LiS-'`

For more information see rpm(1).

6.5.2 Removing the Debian DEB

To remove and installed version of the Debian DEB (whether obtained from the OpenSS7 binary DEB releases, or whether created by the Debian DSC), execute the following command:

     % dpkg -ev `dpkg -l | grep '^LiS-'`

For more information see dpkg(8).

6.5.3 Removing the Source RPM

To remove all the installed binary RPM build from the source RPM, see Removing the Binary RPM. Then simply remove the binary RPM package files and source RPM file. A command such as:

     % find / -name 'LiS-*.rpm' -type f -print0 | xargs --null rm -f

should remove all LiS RPMs from your system.

6.5.4 Removing the Debian DSC

To remove all the installed binary DEB build from the Debian DSC, see Removing the Debian DEB. Then simply remove the binary DEB package files and Debian DSC file. A command such as:

     % find / \( -name 'LiS-*.deb' \
              -o -name 'LiS-*.dsc' \
              -o -name 'LiS-*.tar.* \
              \) -type f -print0 | xargs --null rm -f

should remove all LiS DEBs, DSCs and TARs from your system.

6.5.5 Removing the Tar Ball

To remove a version installed from tar ball, change to the build directory where the package was built and use the `uninstall' automake(1) target as follows:

     % cd /usr/src/LiS
     % make uninstall
     % cd ..
     % rm -fr LiS-2.18.6
     % rm -f LiS-2.18.6.tar.gz
     % rm -f LiS-2.18.6.tar.bz2

If you have inadvertently removed the build directory and, therefore, no longer have a configured directory from which to execute `make uninstall', then perform all of the steps for configuration and installation (see Installing the Tar Ball) except the final installation and then perform the steps above.

6.6 Loading

6.6.1 Normal Module Loading

When Linux STREAMS (LiS) installs, modules and drivers belonging to release packages are normally configured for demand loading. The `install' and `install-strip' automake(1) targets will make the necessary changes to the /etc/modules.conf file and place the modules in an appropriate place in /lib/modules/2.4.20-28.7/lis. The `make install' process should have copied the kernel module files streams-*.o to the directory /lib/modules/2.4.20-28.7/lis. This means that to load any of these modules, you can simply execute, for example, `modprobe stream-somedriver'.³⁵

6.6.1.1 Linux Fast-STREAMS Module Loading

The LiS demand load system supports both the old kerneld and the new kmod mechanisms for demand loading kernel modules.

The convention for LiS kernel loadable object files is:

• Their name start with "streams-".
• They are placed in /lib/modules/2.4.20-28.7/streams/, where `2.4.20-28.7' is an example kernel version.

If your kernel has been built using the kerneld daemon, then LiS kernel modules will automatically load as soon as the STREAMS module is pushed or the driver is opened. The `make install' process makes the necessary changes to the /etc/modules.conf file. After the install, you will see lines like the following added to your /etc/modules.conf file:

     prune modules.lis
     if -f /lib/modules/`uname -r`/modules.lis
     include /lib/modules/`uname -r`/modules.lis
     endif

which will provide for demand loading of the modules if they have been built and installed for the running kernel. The /lib/modules/`uname -r`/modules.lis file looks like this:

     alias char-major-245  streams-some_driver
     alias char-major-246  streams-other_driver

Note that STREAMS modules are not listed in this file, but will be loaded by name using kerneld if available.

Linux Fast-STREAMS has a wider range of kernel module loading mechanisms than is provided by the deprecated LiS. For mechanisms used for kernel module loading under Linux Fast-STREAMS, See About This Manual.

6.6.2 NexusWare Module Loading

Under exceptional circumstances, such as a NexusWare build, it is necessary to hand-edit a .spec and rc.4 file to load the modules at boot time.³⁶

6.6.2.1 Linux STREAMS Module Loading

LiS is deprecated and this section has been deleted.

6.7 Maintenance

6.7.1 Makefile Targets

automake(1) has many targets, not all of which are obvious to the casual user. In addition, OpenSS7 automake(1) files have additional rules added to make maintaining and releasing a package somewhat easier. This list of targets provides some help with what targets can be invoked, what they do, and what they hope to achieve. The available targets are as follows:

6.7.1.1 User Targets

The following are normal targets intended to be invoked by installers of the package. They are concerned with compiling, checking the compile, installing, checking the installation, and removing the package.

`[all]'

This is also the default target. It compiles the package and all release packages selected by configure. This is performed after configuring the source with `configure'. A Makefile stub is provided so that if the package has not had autoreconf(1) run (such as when checked out from CVS, the package will attempt to run `autoreconf -fiv'.

All OpenSS7 Project packages are configured without maintainer mode and without dependency tracking by default. This speeds compilation of the package for one-time builds. This also means that if you are developing using the source package (edit-compile-test cycle), changes made to source files will not cause the automatic rebuilding due to dependencies. There are two ways to enable dependency tracking: specify --enable-maintainer-mode to configure; or, specify --enable-dependency-tracking to configure. I use the former during my edit-compile-test cycle.

This is a standard GNU automake(1) makefile target. This target does not require root privilege.

`check'

All OpenSS7 Project release packages provide check scripts for the check target. This step is performed after compiling the package and will run all of the `check' programs against the compiled binaries. Which checks are performed depends on whether --enable-maintainer-mode was specified to configure. If in maintainer mode, checks that assist with the release of the package will be run (such as checking that all manual pages load properly and that they have required sections.) We recommend running the check stage before installing, because it catches problems that might keep the installed package from functioning properly.

Another way to enable the greater set of checks, without invoking maintainer mode, is to specify --enable-checks to configure. For more information, see Pre-installation Checks.

This is a standard GNU automake(1) makefile target, although the functions performed are customized for the OpenSS7 Project. This target does not require root privilege.

`install'

`install-strip'

The `install' target installs the package by installing each release package. This target also performs some actions similar to the pre- and post-install scripts used by packaging tools such as rpm(1) or dpkg(1). The `install-strip' target strips unnecessary symbols from executables and kernel modules before installing.

This is a standard GNU automake(1) makefile target. This target requires root privilege.

`installcheck'

All OpenSS7 Project packages provide test scripts for the `installcheck' target. Test scripts are created and run using autotest (part of the autoconf(1) package). Which test suites are run and how extensive they are depends on whether --enable-maintainer-mode was specified to configure. When in maintainer mode, all test suites will be run. When not in maintainer mode, only a few post-install checks will be performed, but the test suites themselves will be installed in /usr/libexec/lis³⁷ for later use.

This is a standard GNU automake(1) makefile target. This target might require root privilege. Tests requiring root privilege will be skipped when run as a regular user. Tests requiring regular account privileges will be skipped when run as root.

`retest'

To complement the `installcheck' target above, all OpenSS7 Project packages provide the `retest' target as a means to rerun failed conformance test suite test cases. The `retest' target is provided because some test cases in the test suites have delicate timing considerations that allow them to fail sporadically. Invoking this target will retest the failed cases until no cases that are not expected failures remain.

This is an OpenSS7 Project specific makefile target. As with `installcheck', this target might require root privilege. Tests requiring root privilege will be skipped when run as a regular user. Tests requiring regular account privileges will be skipped when run as root.

`uninstall'

This target will reverse the steps taken to install the package. This target also performs pre- and post- erase scripts used by packaging tools such as rpm or dpkg. You need to have a configured build directory from which to execute this target, however, you do not need to have compiled any of the files in that build directory.³⁸

The `uninstall' target unfortunately removes add-on packages in the same order in which they were installed. This is not good for the OpenSS7 Master Package, where the `remove' target should be used instead.

This is a standard GNU automake(1) makefile target. This target requires root privilege.

`remove'

This target is like `uninstall' with the exception that it removes add-on packages in the reverse order that installation was performed.³⁹

This is an OpenSS7 Project specific makefile target. This target requires root privilege.

6.7.1.2 Maintainer Targets

The following targets are targets intended for use by maintainers of the package, or those responsible for release and packaging of a derivative work of the package. Some of these targets are only effective when maintainer mode has been invoked (--enable-maintainer-mode specified to configure.)

`dist'

Creates a distribution package (tarball) in the top level build directory. OpenSS7 Project packages distribute two archives: a `gzip tar' archive and a `bzip tar' archive. These archives will have the name LiS-2.18.6.tar.gz and LiS-2.18.6.tar.bz2.

This is a standard GNU automake(1) makefile target. This target does not require root privilege.

`distcheck'

This target is intended for use when releasing the package. It creates the tar(1) archives above and then unpacks the tarball in a source directory, configures in a separate build directory, compiles the package, installs the package in a separate install directory, tests the install package to ensure that some components work, and, finally, uses the unpacked source tree to build another tarball. If you have added or removed files from the package, this is a good way to ensure that everything is still stable for release.

This is a standard GNU automake(1) makefile target. This target does not require root privilege.

6.7.1.3 Clean Targets

`mostlyclean'

Cleans out most of the files from the compile stage. This target is helpful if you have not enabled dependency tracking and need to recompile with changes.

This is a standard GNU automake(1) makefile target. This target does not require root privilege.

`clean'

Cleans all the files from the build directory generated during the `make [all]' phase. It does not, however, remove files from the directory left there from the configure run. Use the `distclean' target to remove those too.

This is a standard GNU automake(1) makefile target. This target might require root privilege if the `installcheck' target or the testsuite was invoked with root privilege (leaving files belonging to root).

`distclean'

This target cleans out the directories left behind by `distcheck' and removes all the configure and generated files from the build directory. This will effectively remove all the files in the build directory, with the except of files that belong to you or some other process.

This is a standard GNU automake(1) makefile target. This target might require root privilege if the `installcheck' target or the testsuite was invoked with root privilege (leaving files belonging to root).

`maintainer-clean'

This target not only removes files from the build directory, it removes generated files from the source directory as well. Care should be taken when invoking this target, because it removes files generated by the maintainer and distributed with the archive that might require special tools to regenerate. These special tools might only be available to the maintainer.⁴⁰ It also means that you probably need a full blown Linux system to rebuild the package. For more information, see Downloading from CVS.

This is a standard GNU automake(1) makefile target. This target might require root privilege if the `installcheck' target or the testsuite was invoked with root privilege (leaving files belonging to root).

`check-clean'

This target removes log files left behind by the `check' target. By default, the check scripts append to log files in the top level build directory. This target can be used to clean out those log files before the next run.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

6.7.1.4 Release Targets

The following are targets used to generate complete releases into the package distribution directory. These are good for unattended and NFS builds, which is what I use them for. Also, when building from atop multiple packages, these targets also recurse down through each package.

`release'

Build all of the things necessary to generate a release. On an rpm(1) system this is the distribution archives, the source rpm, and the architecture dependent and architecture independent binary rpms. All items are placed in the package distribution directory that can be specified with the --with-pkg-distdir=DIR option to configure.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`forced-release'

The `release' target will not regenerate any files that already exist in the package distribution directory. This forced target will.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`release-sign'

You will be prompted for a password, unless to specify it to make with the GNUPGPASS variable. For unattended or non-interactive builds with signing, you can do that as: `make GNUPGPASS=mypasswd release-sign'

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`forced-release-sign'

The `release-sign' target will not regenerate any files that already exist in the package distribution directory. This forced target will.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`release-clean'

This target will remove all distribution files for the current package from the package distribution directory.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

6.7.1.5 Logging Targets

For convenience, to log the output of a number of targets to a file, log targets are defined. The log file itself is used as the target to make, but make invokes the target minus a .log suffix. So, for example, to log the results of target `foo', invoke the target `foo.log'. The only target that this does not apply to is `compile.log'. When you invoke the target `compile.log' a simple automake(1) is invoked and logged to the file compile.log. The `foo.log' rule applies to all other targets. This does not work for all targets, just a selected few.⁴¹ Following are the logging targets:

Common Logging Targets

Common logging targets correspond to normal user automake(1) makefile targets as follows:

`compile.log'

This is an OpenSS7 Project specific makefile target, but it invokes the standard GNU automake(1) makefile target `[all]'.

`check.log'

This is an OpenSS7 Project specific makefile target, but it invokes the standard GNU automake(1) makefile target `check'.

`install.log'

This is an OpenSS7 Project specific makefile target, but it invokes the standard GNU automake(1) makefile target `install'.

`installcheck.log'

This is an OpenSS7 Project specific makefile target, but it invokes the standard GNU automake(1) makefile target `installcheck'.

`uninstall.log'

This is an OpenSS7 Project specific makefile target, but it invokes the standard GNU automake(1) makefile target `uninstall'.

`remove.log'

This is an OpenSS7 Project specific makefile target, that invokes the OpenSS7 Project `remove' target.

Maintainer Logging Targets

Maintainer logging targets correspond to maintainer mode automake(1) makefile targets as follows:

`dist.log'

This is an OpenSS7 Project specific makefile target, but it invokes the standard GNU automake(1) makefile target `dist'.

`distcheck.log'

This is an OpenSS7 Project specific makefile target, but it invokes the standard GNU automake(1) makefile target `distcheck'.

`srpm.log'

This is an OpenSS7 Project specific makefile target, that invokes the OpenSS7 Project `srpm' target.

`rebuild.log'

This is an OpenSS7 Project specific makefile target, that invokes the OpenSS7 Project `rebuild' target.

`resign.log'

This is an OpenSS7 Project specific makefile target, that invokes the OpenSS7 Project `resign' target.

`release.log'

This is an OpenSS7 Project specific makefile target, that invokes the OpenSS7 Project `release' target.

`release-sign.log'

This is an OpenSS7 Project specific makefile target, that invokes the OpenSS7 Project `release-sign' target.

If you want to add one, simply add it to LOGGING_TARGETS in Makefile.am.

6.7.1.6 Problem Report Targets

To ease problem report generation, all logging targets will automatically generate a problem report suitable for mailing in the file target.pr for target `target.log'. This problem report file is in the form of an email and can be sent using the included send-pr script or by invoking the `send-pr' makefile target.

There are two additional problem report targets:

`pr'

The `pr' target is for independently generating a problem report outside of the build or installation process. The target will automatically generate a problem report skeleton suitable for editing and mailing in the file problem.pr. This problem report file is in the form of an email and can be edited and sent directly, or sent using the included send-pr script or by invoking the `send-pr' target.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`send-pr'

The `send-pr' target is for finalizing and mailing a problem report generated either inside or outside the build and installation process. The target will automatically finalize and mail the problem.pr problem report if it has changed since the last time that `send-pr' was invoked.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege (unless the problem report file was generated as root).

6.7.1.7 Release Archive Targets

The following targets are used to generate and clean distribution archive and signature files. Whereas the `dist' target affects archives in the top build directory, the `release-archive' targets affects archives in the package distribution directory (either the top build directory or that specified with --with-pkg-distdir=DIR to configure).

You can change the directory to which packages are distributed by using the --with-pkg-distdir=DIR option to configure. The default directory is the top build directory.

`release-archives'

This target creates the distribution archive files if they have not already been created. This not only runs the `dist' target, but also copies the files to the distribution directory, which, by default is the top build directory.

The files generated are named:

LiS-2.18.6.tar.gz and LiS-2.18.6.tar.bz2

You can change this distribution directory with the --with-pkg-distdir option to configure. See `./configure --help' for more details on options.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`release-sign-archives'

This target is like `release-archives', except that it also signs the archives using a GPG detached signature. You will be prompted for a password unless you pass the GNUPGPASS variable to make. For automated or unattended builds, pass the GNUPGPASS variable like so:

`make GNUPGPASS=mypasswd release-sign-archives'

Signature files will be named:

LiS-2.18.6.tar.gz.asc and LiS-2.18.6.tar.bz2.asc

These files will be moved to the package distribution directory with the plain text archives.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`release-clean-archives'

This target will clean the release archives and signature files from the package distribution directory.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

6.7.1.8 RPM Build Targets

On rpm(1) systems, or systems sporting rpm packaging tools, the following targets are used to generate rpm(1) release packages. The epoch and release number can be controlled by the contents of the .rpmepoch and .rpmrelease files, or with the --with-rpm-epoch=EPOCH and --with-rpm-release=RELEASE options to configure. See `configure --help' for more information on options. We always use release number `1'. You can use release numbers above `1'.

`srpm'

This target generates the source rpm for the package (without signing the source rpm). The source rpm will be named: LiS-2.18.6-1.srpm.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`rpms'

This target is responsible for generating all of the package binary rpms for the architecture. The binary rpms will be named:

LiS-*-2.18.6-1.*.rpm

where the stars indicate the subpackage and the architecture. Both the architecture specific subpackages (binary objects) and the architecture independent (.noarch) subpackages will be built unless the the former was disabled with the option --disable-arch, or the later with the option --disable-indep, passed to configure.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`sign'

`srpm-sign'

These two targets are the same. When invoked, they will add a signature to the source rpm file, provided that the file does not already have a signature. You will be prompted for a password if a signature is required. Automated or unattended builds can be achieved by using the emake expect script, included in ${srcdir}/scripts/emake.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`rebuild'

This target accepts searches out a list of kernel names from the ${DESTDIR}/lib/modules directory and builds rpms for those kernels and for each of a set of architectures given in the AM_RPMTARGETS variable to make. This is convenience target for building a group of rpms on a given build machine.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.

`resign'

This target will search out and sign, with a GPG signature, the source rpm, and all of the binary rpms for this package that can be found in the package distribution directory. This target will prompt for a GPG password. Automated or unattended builds can be achieved with the emake expect script located here: ${srcdir}/scripts/emake.

This is an OpenSS7 Project specific makefile target. This target does not require root privilege.