diff options
author | jkh <jkh@FreeBSD.org> | 1996-03-17 11:44:22 +0000 |
---|---|---|
committer | jkh <jkh@FreeBSD.org> | 1996-03-17 11:44:22 +0000 |
commit | 783a3fe9363d0ad82dbc66c6b59a3383734d1612 (patch) | |
tree | 75bfc0c05f1901af5d18f776841fb4d5d7d465c3 /share/man/man4/man4.i386/fe.4 | |
parent | 6aa1e2ad8bdc3a860a591d0d89c2767673225796 (diff) | |
download | FreeBSD-src-783a3fe9363d0ad82dbc66c6b59a3383734d1612.zip FreeBSD-src-783a3fe9363d0ad82dbc66c6b59a3383734d1612.tar.gz |
Whoops, don't forget the man page that goes with the previous commit.
Submitted-by: M. Sekiguchi <seki@sysrap.cs.fujitsu.co.jp>
Diffstat (limited to 'share/man/man4/man4.i386/fe.4')
-rw-r--r-- | share/man/man4/man4.i386/fe.4 | 252 |
1 files changed, 62 insertions, 190 deletions
diff --git a/share/man/man4/man4.i386/fe.4 b/share/man/man4/man4.i386/fe.4 index dfcc0bf..b610979 100644 --- a/share/man/man4/man4.i386/fe.4 +++ b/share/man/man4/man4.i386/fe.4 @@ -17,27 +17,24 @@ .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS DOCUMENT, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. -.\" Contributed by M.S. <seki@sysrap.cs.fujitsu.co.jp> -.\" for fe driver version 0.8 .\" -.\" $Id: fe.4,v 1.1 1995/04/23 18:54:40 wollman Exp $ -.Dd April 23, 1995 +.\" Contributed by M. Sekiguchi <seki@sysrap.cs.fujitsu.co.jp>. +.\" for fe driver. +.\" +.\" $Id: fe.4,v 1.2 1996/02/11 22:37:34 mpp Exp $ +.Dd March 3, 1996 .Dt FE 4 .Sh NAME .Nm fe .Nd Fujitsu MB86960A/MB86965A based Ethernet adapters .Sh SYNOPSIS -.Cm "device fe0 at isa? port ? net irq ? vector feintr" +.Cm "device fe0 at isa? port 0x300 net irq ? vector feintr" .Sh DESCRIPTION The .Nm fe is a network device driver -for Fujitsu MB86960A/MB86965A based Ethernet adapters. -.Pp -Performance decline to support multicast is kept minimum -by the assistance of multicast address filter hardware -built in the chip. -The driver also supports duplex operation. +for Ethernet adapters based on Fujitsu MB86960A, MB86965A, +or other compatible chips. .Pp The driver provides automatic I/O port address configuration and automatic IRQ configuration, @@ -48,7 +45,8 @@ It gives a fair performance. Shared memory is never used, even if the adapter has one. .Pp It currently works with Fujitsu FMV-180 series for ISA, -and Allied-Telesis AT1700 series and RE2000 series for ISA. +Allied-Telesis AT1700 series and RE2000 series for ISA, +and Fujitsu MBH10302 PC card. .Ss Parameters On the kernel configuration file, two parameters, .Ar port @@ -140,8 +138,7 @@ Setting the .Dv "FE_DEBUG" option to a value which is not listed here may cause undocumented behavior. -The default setting of this option varies from version to version. -For version 0.8, the default is 1. +The default setting of this option is 1. .Bl -bullet .It Setting @@ -174,37 +171,47 @@ or dumping register values are logged under this setting. .El .El -.Sh HARDWARE DEPENDENCY +.Sh FEATURES SPECIFIC TO HARDWARE MODELS The .Nm fe -driver has some features and limitations which depend on adapter hardwares. +driver has some features and limitations +which depend on adapter hardware models. The following is a summary of such dependency. .Ss Fujitsu FMV-180 series adapters Both automatic IRQ detection and automatic I/O port address detection is available with these adapters. .Pp -Automatic I/O port address detection feature works -even if there are two or more FMV-180 cards in a system. -It is recomended to use +Automatic I/O port address detection feature of +.Nm fe +works mostly fine for FMV-180 series. +It works even if there are two or more FMV-180s in a system. +However, some combination of other adapters may confuse the driver. +It is recommended not to use .Nm port ? -always. +when you experience some difficulties with hardware probe. .Pp -The hardware setting of IRQ is always taken from the configuration EEPROM -on the adapter. -The user-specified IRQ is never used if the +Automatic IRQ detection feature of .Nm fe -driver is configured for FMV-180. +works reliably for FMV-180 series. +It is recommended to use +.Nm irq ? +always for FMV-180. +The hardware setting of IRQ is read +from the configuration EEPROM on the adapter, +even when the kernel config file specifies an IRQ value. +The driver will generate a warning message, +if the IRQ setting specified in the kernel config file +does not match one stored in EEPROM. +Then, it will use the value specified in the config file. +(This behavior has been changed from the previous releases.) .Ss Allied-Telesis AT1700 series and RE2000 series adapters Automatic I/O port address detection -is available with these adapters. -Automatic IRQ detection is also available with a limitation. -.Pp -Automatic I/O port address detection feature works -even if there are two or more AT1700 and/or RE2000 cards in a system. -It is recomended to use -.Nm port ? -always. +is available with Allied-Telesis AT1700 series and RE2000 series, +while it is less reliable than FMV-180 series. +Using the feature with Allied-Telesis adapters +is not recommended. .Pp +Automatic IRQ detection is also available with some limitation. The .Nm fe driver @@ -219,195 +226,60 @@ while the other from 10/11/12/15. Identification of the models are not well known. Hence, automatic IRQ detection with Allied-Telesis adapters may not be reliable. -Specify exact IRQ number if any troubles are encoutered. +Specify an exact IRQ number if any troubles are encountered. .Pp -All of AT1700/RE2000 type adapters are reported as just -.Sq AT1700 -by the driver. Differences between AT1700 series and RE2000 series or minor models in those series are not recognized. -.Pp -Support of this hardware is still under development. -It may or may not work. -.Sh DIAGNOSTICS -.Ss Level 0 messages -The following messages will be reported when -.Dv "FE_DEBUG" -option is set to 0. -.Bl -diag -.It "address %s, type %s" -The message shows an Ethernet station address and an adapter type. -This is a normal message. -.El -.Ss Level 1 messages -The following messages, as well as level 0 messages, -will be reported when -.Dv "FE_DEBUG" -option is set to 1. -.Bl -diag -.It "transmission timeout (%d+%d)" -The driver had passed one or more out-going packets to the MB86960A/MB86965A -to transmit, -but the chip did not respond in an appropriate time limit. -This may be an indication of an inappropriate hardware installation, -e.g., wrong IRQ setting. -.It "transmission timeout (%d+%d) when down" -The kernel informed -.Nm fe -a transmittion timeout condition, -when the device was not working. -This may be an indication of a serious bug. -.It "init() without any address" -The driver has been started, -but no valid network address has been set. -.It "invalid station address (%s)" -The driver found its own Ehternet station address is invalid, -i.e., it was either broadcast address, multicast address, or -of null-vendor. -This may be a result of an attempt to set such address through -some -.Fn ioctl -calls. -.It "cannot empty receive buffer" -When starting up, -.Nm fe -tries to clean out everything in the packet receive buffer memory. -This message indicates a failure to do so. -This can only be caused by a driver bug and/or a broken hardware. -.It "inconsistent txb variables (%d, %d)" -The driver found some internal buffer management variables in -contradicting states. -This is an indication of a driver bug. -.It "transmitter idle with %d buffered packets" -The driver found that the transmitter part of the MB86960A/MB86965A -was not working, -while one or more outgoing packets were queued. -This message suggests a bug in the -.Nm fe -driver, or a bug in the other part of the kernel, -such as another device driver, -breaking work space of the -.Nm fe -driver. -It may be better to reboot the system as soon as possible, -if the message is reported. -.It "got a partial packet (%u bytes) to send" -An upper layer program passed a very short -(shorter than an Ethernet header) -packet to the device. -It must be a bug in kernel network program. -.It "got a big packet (%u bytes) to send" -An upper layer program passed a packet -longer than what allowed in Etherenet specification to send. -It must be a bug in kernel network program. -.El -.Ss Level 2 messages -The following messages, as well as level 0 and 1 messages, -will be reported when -.Li "FE_DEBUG" -option is set to 2. -.Bl -diag -.It "if_fe.c ver. 0.8 / if_fereg.h ver. 0.8" -It shows versions of the driver source files. -This message is normal. Just ignore it. -.It "strange TXBSIZ config; fixing" -The TXBSIZ (transmission buffer size) of the DLCR6 setting was not -appropriate for the -.Nm fe -driver. -The driver will automatically adjust the setting and keep working -even after the message is reported. -Double buffering is essential for +.Ss Fujitsu MBH10302 PC card +The .Nm fe -driver to work properly, -so it must be either -.Li 01 , 10 , No or Li 11 . -This message is reported only when -.Sq Nm flags Li 0x0080 -is set. -.It "excessive collision (%d/%d)" -An excessive collision error was detected on the Ethernet. -It is normal that the message is reported when the network is overloaded. -If the message appears frequently, -it may be an indication of the physical problem on the network cable. -(The two numbers appended to the end of the message -are only for debugging and should be ignored.) -.It "received a big packet? (%u bytes)" -The MB86960A/MB86965A reported a receiving of a big packet. -(A packet longer than the Ethernet limit.) -This may be a bug in the driver, -because these packets are usually thrown out by the chip. -.It "received a partial packet? (%u bytes)" -The MB86960A/MB86965A reported a receiving of a very short packet. -(A packet shorter than the Ethernet header.) -This may be a bug in the driver, -because these packets are usually ignored by the chip. -.It "received a short packet? (%u bytes)" -The MB86960A/MB86965A reported a receiving of a short packet. -.Po -A packet shorter than the Ethernet minimum, also known as a -.Dq runt . -.Pc -This may be a bug in the driver, -because these packets are usually ignored by the chip. -.It "no enough mbuf; a packet (%u bytes) dropped" -A kernel memory space called -.Sq mbuf -was exhausted, -and a received packet could not be processed. -It is normal that the message is reported when the system is overloaded. -If this message appears frequently, -it may be an indication of a bug in the driver, -or an inappropriate configuration of the kernel. -.It "packet length mismatch? (%d/%d)" -The driver got a packet to send, enclosed in an mbuf, -whose length information in the mbuf header -and actual length of the packet data disagreed. -This may be a bug in a kernel network program. -.El -.Ss Level 3 messages -More messages, as well as level 0, 1 and 2 messages, -will be reported when -.Dv "FE_DEBUG" -option is set to 3. -They are not documented, since they are for debugging purposes. -See the driver source program for level 3 messages. +driver supports Fujitsu MBH10302 and compatible PC cards. +It requires the PC card (PCMCIA) support package. .Sh SEE ALSO .Xr netstat 1 , .Xr netintro 4 , .Xr ed 4 , -.Xr ifconfig 8 +.Xr crd 4 , +.Xr ifconfig 8 , +.Xr pccardd 8 .Sh BUGS -Following are major known bugs. +Following are major known bugs: .Pp Statistics on the number of collisions maintained by the .Nm fe -driver may no be accurate: +driver is not accurate; the .Fi i option of .Xr netstat 1 -sometimes shows slightly less value than true number of collisions. +shows slightly less value than true number of collisions. .Pp -More mbuf clusters are consumed than expected. +More mbuf clusters are used than expected. The packet receive routine has an intended violation against the mbuf cluster allocation policy. The unnecessarily allocated clusters are freed within short lifetime, and it will not affect long term kernel memory usage. +.Pp +Although XNS and IPX supports are included in the driver, +they are never be tested and must have a lot of bugs. .Sh AUTHOR, COPYRIGHT AND DISCLAIMER .Pp The .Nm fe -driver is written and contributed by M.S. +driver was originally written and contributed by M. Sekiguchi .Li "<seki@sysrap.cs.fujitsu.co.jp>" , following the .Nm ed driver written by David Greenman. -This manual page is written by M.S. +PC card support in +.Nm fe +is written by Hidetoshi Kimura +.Li "<h-kimura@tokyo.se.fujitsu.co.jp>" . +This manual page was written by M. Sekiguchi. .Pp .Em "All Rights Reserved, Copyright (C) Fujitsu Limited 1995" .Pp -This document and the assosicated software may be used, modified, +This document and the associated software may be used, modified, copied, distributed, and sold, in both source and binary form provided that the above copyright, these terms and the following disclaimer are retained. The name of the author and/or the contributor may not be @@ -433,4 +305,4 @@ The .Nm driver appeared in .Tn FreeBSD -2.1. +2.0.5. |