Re: 2.4 kernel, problems shutting down eth0 interface
Hi. I just want to let you know that I have run into exactly the same behavior as you did. Originally I compiled my kernel with the PCMCIA driver as a module. I recompiled to make the PCMCIA driver integral (i.e., I selected "y" instead of "m" for PCMCIA in "make xconfig") and since then I have not had this problem. However I haven't had to connect my computer to Ethernet since this time either, and I think that every time I got the error message before I happened to be connected to the Ethernet, so I fear it may be this, and not the integralness of the PCMCIA driver, that has made the difference. Please let me know if compiling the PCMCIA driver integrally fixes the problem for you. Please reply to jdthood_AT_yahoo.co.uk, not to the return address of this e-message. Thomas jdthood_AT_yahoo.co.uk --- Original message follows --- Yesterday I compiled a 2.4 kernel using make-kpkg. All seems to have gone well. I used the pcmcia features in the kernel, and chose not to separately compile pcmcia modules. All seems to work well, I boot fine, start up pcmcia fine, get a dhcp address fine. However, I do have one problem, I can't shut down or reboot without cutting the power to the laptop. When I try, I get the following error: "Shutting down PCMCIA services: cardmgrunregister_netdevice waiting for eth0 to become free. Usage count = 0" No matter how long I wait, this message keeps popping up every 3-4 seconds. It doesn't stop, and the rest of the shutdown/reboot process seems to be waiting for this to finish. Does anybody have any ideas as to what is going on? Also, one other question: I have been using dhcpcd. In the bootup process, I now see something that says "pump -r -i eth0 /sbin/ifconfig eth0 down cardmgr". Don't know what this means, but I do know that I don't have pump installed on my computer. Any ideas? Thanks! Bryan Walton ___ Send a cool gift with your E-Card http://www.bluemountain.com/giftcenter/
Re: Turning off esd
Please note that this problem is addressed in the scripts for the ALSA sound drivers furnished in the alsa-base package (... get the latest version). The package includes a file /usr/lib/alsa-base/snd-dev-utils which contains bash functions which you may find useful. Thomas Please reply not to the return address of this message but to jdthood_AT_yahoo.co.uk > I would like to automatically turn off the Enlightenment Sound Deamon > (esd) and remove the sound modules before suspending my laptop and turn > everything back on when resuming. Otherwise sound does not work anymore > after resuming and I have to reload the modules manually. I tried > putting "esdctl off; modprobe -r i810_audio, etc." into a script in > /etc/apm/event.d. However this does not seem to work since esdctl is > executed from the root account and esd was started from a normal user > account by Gnome. > > Did anyone already solve this problem? I appreciate all suggestions! ___ Send a cool gift with your E-Card http://www.bluemountain.com/giftcenter/
Re: thinkpad 560E serial port
> anyone else have trouble getting the serial port on a thinkpad > 560e to work? the standard PC io port and irq values don't > seem to work. Check the "Serial port" section my web page for info that may help you: Thomas Hood's Debian GNU/Linux on a ThinkPad 600 site: http://panopticon.csustan.edu/thood/tp600lnx.htm Thomas jdthood_AT_mail.com
Re: thinkpad 560E serial port
Tom Vier, Excuse me, but did I give you permission to post my e-mail address on debian-laptop? I deliberately disguise my e-mail address so that I don't get on spam lists, but it seems you're too stupid^H^H^H^H^H^Hmentally challenged to figure that out. You undisguised the address and posted away. Thomas Hood Tom Vier wrote: > > On Sun, Feb 04, 2001 at 06:01:01PM -0500, Thomas Hood wrote: > > > anyone else have trouble getting the serial port on a thinkpad > > > 560e to work? the standard PC io port and irq values don't > > > seem to work. > > > > Check the "Serial port" section my web page for info that may > > yeah, i finally figured out that ps2.exe has two serial options. now my > problem is that when i insert my linksys ethernet pcmcia card, serial stops > working. i get this: > > rs_close: bad serial port count; tty->count is 1, state->count is 2 > > anyone else have problems like this with 2.2.19pre8+pcmcia? > > -- > Tom Vier <[EMAIL PROTECTED]> > DSA Key id 0x27371A2C
Re: Turning off esd
Please note that this problem is addressed in the scripts for the ALSA sound drivers furnished in the alsa-base package (... get the latest version). The package includes a file /usr/lib/alsa-base/snd-dev-utils which contains bash functions which you may find useful. Thomas Please reply not to the return address of this message but to jdthood_AT_yahoo.co.uk > I would like to automatically turn off the Enlightenment Sound Deamon > (esd) and remove the sound modules before suspending my laptop and turn > everything back on when resuming. Otherwise sound does not work anymore > after resuming and I have to reload the modules manually. I tried > putting "esdctl off; modprobe -r i810_audio, etc." into a script in > /etc/apm/event.d. However this does not seem to work since esdctl is > executed from the root account and esd was started from a normal user > account by Gnome. > > Did anyone already solve this problem? I appreciate all suggestions! ___ Send a cool gift with your E-Card http://www.bluemountain.com/giftcenter/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4 kernel, problems shutting down eth0 interface
Hi. I just want to let you know that I have run into exactly the same behavior as you did. Originally I compiled my kernel with the PCMCIA driver as a module. I recompiled to make the PCMCIA driver integral (i.e., I selected "y" instead of "m" for PCMCIA in "make xconfig") and since then I have not had this problem. However I haven't had to connect my computer to Ethernet since this time either, and I think that every time I got the error message before I happened to be connected to the Ethernet, so I fear it may be this, and not the integralness of the PCMCIA driver, that has made the difference. Please let me know if compiling the PCMCIA driver integrally fixes the problem for you. Please reply to jdthood_AT_yahoo.co.uk, not to the return address of this e-message. Thomas jdthood_AT_yahoo.co.uk --- Original message follows --- Yesterday I compiled a 2.4 kernel using make-kpkg. All seems to have gone well. I used the pcmcia features in the kernel, and chose not to separately compile pcmcia modules. All seems to work well, I boot fine, start up pcmcia fine, get a dhcp address fine. However, I do have one problem, I can't shut down or reboot without cutting the power to the laptop. When I try, I get the following error: "Shutting down PCMCIA services: cardmgrunregister_netdevice waiting for eth0 to become free. Usage count = 0" No matter how long I wait, this message keeps popping up every 3-4 seconds. It doesn't stop, and the rest of the shutdown/reboot process seems to be waiting for this to finish. Does anybody have any ideas as to what is going on? Also, one other question: I have been using dhcpcd. In the bootup process, I now see something that says "pump -r -i eth0 /sbin/ifconfig eth0 down cardmgr". Don't know what this means, but I do know that I don't have pump installed on my computer. Any ideas? Thanks! Bryan Walton ___ Send a cool gift with your E-Card http://www.bluemountain.com/giftcenter/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: thinkpad 560E serial port
> anyone else have trouble getting the serial port on a thinkpad > 560e to work? the standard PC io port and irq values don't > seem to work. Check the "Serial port" section my web page for info that may help you: Thomas Hood's Debian GNU/Linux on a ThinkPad 600 site: http://panopticon.csustan.edu/thood/tp600lnx.htm Thomas jdthood_AT_mail.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: thinkpad 560E serial port
Tom Vier, Excuse me, but did I give you permission to post my e-mail address on debian-laptop? I deliberately disguise my e-mail address so that I don't get on spam lists, but it seems you're too stupid^H^H^H^H^H^Hmentally challenged to figure that out. You undisguised the address and posted away. Thomas Hood Tom Vier wrote: > > On Sun, Feb 04, 2001 at 06:01:01PM -0500, Thomas Hood wrote: > > > anyone else have trouble getting the serial port on a thinkpad > > > 560e to work? the standard PC io port and irq values don't > > > seem to work. > > > > Check the "Serial port" section my web page for info that may > > yeah, i finally figured out that ps2.exe has two serial options. now my > problem is that when i insert my linksys ethernet pcmcia card, serial stops > working. i get this: > > rs_close: bad serial port count; tty->count is 1, state->count is 2 > > anyone else have problems like this with 2.2.19pre8+pcmcia? > > -- > Tom Vier <[EMAIL PROTECTED]> > DSA Key id 0x27371A2C -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: persistent interface
> My problem has to do with the network interfaces. My > /etc/network/interfaces file looks like this: > ># The loopback interface ># automatically added when upgrading >auto lo >iface lo inet loopback > ># For the built-in NIC: >iface eth0 inet dhcp > ># For the wireless card ># >iface eth1 inet dhcp >mapping hotplug >script echo It looks as if you are under the impression that "mapping" is an option for "iface eth1 inet dhcp". Actually it should be a separate stanza. Here is your file reformatted to show the effective structure better: # Physical interfaces to bring up at boot time auto lo # Mappings # Bring up hot plugged interfaces with logical interface name # the same as their physical interface name mapping hotplug script echo # Logical interface lo # The logical interface as which the loopback interface is brought up iface lo inet loopback # Logical interface eth0 # The logical interface as which the built-in NIC eth0 is brought up iface eth0 inet dhcp # Logical interface eth1 # The logical interface as which the Wi-Fi card is brought up iface eth1 inet dhcp > I think that this should bring up only the loopback interface at > boot-time. Strictly speaking, this means that only the loopback interface "lo" is brought up by /etc/rcS.d/S40networking. Other interfaces can be brought up while the system is booting via the hotplug mechanism. Your stanza mapping hotplug script echo enables this mechanism. > Eth1 (pcmcia wireless card) should be brought up when the > hotplug system detects an insertion, and eth0 (associated with the > built-in NIC) should be brought up only when ifup is executed. > And indeed when I do: > >/etc/init.d/networking stop >/etc/init.d/networking start > > after the boot process, this is exactly what happens. Right. > The problem, though, is that at boot-time, the system always tries to > bring up eth0 (which corresponds to the built-in NIC). The kernel detects the interface and calls hotplug to notify user space that the interface is available. [...] > suggests that the attempt to bring up eth0 is triggered by loading of > the driver module (8139too), and indeed lsmod reports that this module > is loaded by the end of the boot process. Correct. > What I *cannot* for the life of me figure out is why this is > happening. I've been all through the directory /etc/modprobe.d/ and > there's no reference there to either 8139too or eth0. I've run > update-modules to make sure that there was no ancient module > configuration haunting the system. There is no reference to either > 8139too or to eth0 in /lib/modules/modprobe.conf. > > Can anyone tell me where else I could look to pin this down and figure > out why it is happening? I don't know who is loading the module, sorry. However, you can prevent eth0 from being ifup'ped via the hotplug mechanism by changing the mapping stanza to this: mapping hotplug script grep map eth1 Then only eth1 will be brought up on hotplug. -- Thomas Hood A man does not show his greatness by being at one extremity, but rather by touching both at once.-- Pascal -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: WLAN-Card on Sarge
> Ok, now I'm confused. In the pcmcia-cs package there's the cardmgr > program that, if I understand correctly, reads /etc/pcmcia/config* files > on startup. Then when a card is inserted cardmgr matches the card to a > device id (from the config file(s)) and load modules. It then loads the > associated "class" script -- which is the network.opts script you > mentioned, and if nothing matches there (on Debian) ifup or ifdown is > called. And as you say, ifup/ifdown use /etc/network/interfaces to > configure things. That's right. > But then I see you have a "mapping hotplug" stanza. So is hotplug > managing your pcmcia card inserts? That is, what's confusing me is I > though[t] pcmcia-cs (i.e. cardmgr) handled this, yet you are showing a > stanza defined for hotplug events. cardmgr manages PCMCIA cards via /etc/pcmcia/network. Hotplug can do the same thing but it is disabled by default. If you enable the hotplug mechanism by adding mapping hotplug script echo to /etc/network/interfaces then you should disable the cardmgr mechanism by putting an exit 0 at the top of /etc/pcmcia/network. > In my network.opts script I modified it to bring down eth0 before > bringing up the wireless interface (eth1 in my case), and when the card > is removed call ifup eth0. What I'd like to figure out is how to not > bring up eth0 unless a cable is plugged in. You can use ifplugd for that. > I'm having so many problems with wireless now. I think my hardware must > be failing as I'm having the same problems with different drivers. > Lately when I remove my wireless card my machine will freeze hard. > Seems like these problems started with upgrading to 2.6.5, but I also > suspect my wireless hardware is failing since I often have to hard reset > my AP. It is most likely buggy drivers causing lockups. Expect further problems. Debian's support for dynamic configuration sucks. To gain an understanding of how badly it sucks, start reading the bugs open against hotplug and ifupdown. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: WLAN-Card on Sarge
> So, what's the advantage of using hotplug for bringing up the interface > over cardmgr running /etc/pcmcai/network(.opts)? Hotplug is the buggy work-in-progress standard mechanism of the future whereas cardmgr is the debugged bloated obsolete mechanism of the past. Take your pick. > I tried that once, and maybe didn't understand the config. I was > actually looking for a way in /etc/network/interfaces to run a script > and have the exit code control how the interface is brought up. Look at the ifupdown package: http://panopticon.csustan.edu/thood/ifupdown-roam.html Even if you don't want to install the package, the README file may be helpful to you. > > Expect further problems. Debian's support for dynamic configuration > > sucks. > > To gain an understanding of how badly it sucks, start reading the bugs > > open against hotplug and ifupdown. > > Is that a Linux issue or Debian? It's a Debian issue. The hotplug and ifupdown maintainers are inactive. However, I don't know of any other distros that have better support for automatic dynamic configuration of laptops. Do you? > I did notice that with 2.4.21 and wlan-ng that cardmgr was better at > removing the card -- I'd pull the card and the drivers would be removed > without delay. Now with 2.6.5 and Orinoco|HostAP there's either a long > delay or a system hang. I don't think that's a cardmgr issue, though -- > as I've tried manually removing the drivers (ifdown eth1; rmmod > ) and have the same problems. So, I'll agree that it's either a > problem with the driver dealing with being removed, or the kernel. I should make one thing clear. If you have PCMCIA cards then you always need cardmgr per se. When you put an "exit 0" at the top of /etc/pcmcia/network you only disable the interface configuration code. However, cardmgr is still used to load drivers, to control the sockets, to beep, etc. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: problems getting static ip to work
> I have a laptop running Debian, whose network interface is a pcmcia card. > I use it as an IMAP mail server. Recently, I tried giving it a static IP > address. > by changing /etc/pcmcia/network.opts and filling in the appropriate > fields (I actually used pcnetconfig, so it generated the file for me) > My /etc/network/interfaces looks like this > auto lo > iface lo inet loopback > eth0 > iface eth0 inet static > address 192.168.2.6 > netmask 255.255.255.0 > network 192.168.2.0 > gateway 192.168.2.1 This is invalid. The 'eth0' on its own line shouldn't be there. The rest is OK: auto lo iface lo inet loopback iface eth0 inet static address 192.168.2.6 netmask 255.255.255.0 network 192.168.2.0 gateway 192.168.2.1 > ifconfig -a > does seem to give me information that eth0 is configured (I can even it > ping it from another machine on the network) > > I am noticing 2 problems. > > On boot up I get a "neighbor table overflow" message. I was told this > has something to do with the loopback not being configured, but I > thought my interfaces file should take care of this. Yes it should. > More importantly, my IMAP daemon does not seem to be retrieving any mail > from my Pop server. > I tried doing a > % telnet 127.0.0.1 imap2 > > and it hangs.. > > I think I'm going in circles here and I'd appreciate it if someone can > explain to me what the problem is and how I might fix it? Try eliminating the 'network' line from the eth0 iface definition. -- Thomas -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: persistent interface
> Is there good documentation somewhere about mapping stanzas in the > interfaces file? I've read the `interfaces' man page, and the stuff in > /usr/share/doc/ifupdown. They're both very helpful, but a bit cryptic. You might also want to read the Networking chapter of the Debian Reference. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: no sound on thinkpad 600 - kernel 2.6.8
On Fri, 17 Sep 2004 06:00:11 +0200, Carlos Gabriel Drach wrote: > Hi, i have no sound (i cant hear nothing), but see that modules > (cs46xx) are loaded. cs46xx is not an ALSA driver; it is an OSS driver. What you want is snd-cs46xx. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: APM: will not sleep with AC plugged in?
User space cannot veto APM suspend requests in current Linux kernels. What is keeping your machine from suspending is the firmware. If you have a PCMCIA card then try ejecting it before attempting to suspend. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Apt-get update killed i810 audio?
On Wed, 22 Dec 2004 10:10:14 +0100, Hanspeter Kunz wrote: > On Wed, 2004-12-22 at 12:52 +1000, Ben wrote: >> Hi all, >> >> Just wondering if anyone else has had this problem recently... >> >> I'm running debian testing. did an apt-get update three days ago and now >> my sound doesn't work! >> dmesg doesn't show anything interesting, but i have noticed on boot when >> pci loads(?) that i810 is blacklisted > > Yes it is. The reason is I guess that it interferes with the ALSA > drivers, which are prefered. If you want to continue using OSS just take > it out of the hotplug blacklist (in /etc/hotplug). The latest ALSA does blacklist all OSS modules. The reason, of course, is that OSS modules interfere with ALSA modules in that they try to control the same hardware. If you have alsa-base installed on your system and you are still using OSS sound modules then you may experience loss of sound on upgrade to the latest alsa-base release. This is not a bug in alsa-base. If you want to use OSS modules then you should not have alsa/libasound packages installed. If you would like to resume using OSS modules then purge alsa* and libasound*. If you would like to begin using ALSA modules then most probably all you need to do is load them. Many people arrange to have this done automatically at boot time either by installing the discover package or by adding the name of the sound card driver module to /etc/modules. (I am not sure whether or not hotplug loads ALSA modules at boot time.) -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Debian vs. Fedora on Laptops
On Sun, 26 Dec 2004 03:30:07 +0100, Ryan D'Baisse wrote: > 3. My primary machine is a laptop (Toshiba 5005-S507). Most of the > info I can find on Google, pertaining to laptops, is for Fedora or > Mandrake. How well does Debian handle laptops? If you have a laptop then there is no distribution of GNU/Linux that doesn't give you headaches once in a while. Debian is no worse than the others, overall, and is possibly somewhat better in some ways. However, you do have to educate yourself to solve the problems that arise. One way in which Debian is different is in its poverty of wizard type programs, compared with other distros. > 4. My second biggest problem on Fedora was/is wireless support. I am > currently using FC2 with Linuxant's DriverLoader software on my > Linksys WPC54G PCMCIA NIC. Be honest; am I going to be crying if I > try to set this up? No. Take reassurance from the fact that Debian is one of the major GNU/Linux distributions and it has a lot of capable people working on it. While some things will not work without effort, it is generally true that with some effort, and with the help of many excellent tools developed for Debian, you can achieve your goal. > 5. And, finally, my biggest problem, and one of the reasons I am > looking to leave Fedora, is ACPI. I have to bypass it with later > versions of the FC2 kernel and with the base install of FC3. And, > unfortunately, there are problems with my NIC if I bypass it. How > tightly integrated is ACPI with Debian? In a sense, it isn't integrated at all. Debian is highly modular. > Any chance I can get away > from these issues by switching from Fedora to Debian. I don't think that you should switch from one distribution to another in order to solve a single problem like this. Debian has its own weaknesses and Fedora is an excellent distribution; many capable people are working on it including many of the Linux hackers. Choose Debian if you want freedom, modularity, APT, and attention to technical detail. Choose Libranet or Ubuntu if you want all of the above plus easier installation and configuration, minus some freedom. Choose RedHat if you want to use what the majority of corporate users of GNU/Linux are using. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Debian vs. Fedora on Laptops
On Tue, 28 Dec 2004 01:30:06 +0100, norton wrote: >> It also seemed to easy to corrupt >> the system by installing the wrong program from one of the downloaded CDs. That is less likely to happen with Debian than with any other distro, if you are using apt. >> (Sarge comes with 14 CDs. People like me have no idea what most of those >> programs are or how they are supposed to be used.) > > yeah... but download only the 2 first... that was all I needed for one of > the laptops... the other I used a net install. But I agree that > documentation is really bad. People are working on this: http://debtags.alioth.debian.org/ Please take into account that Debian is by far the largest distribution. >> I think that Knoppix is probably the best distro, but FC3 comes in right >> afterwards, and RH9 follows that. I might have a more favorable opinion of >> Debian if it installed correctly, if it supported more hardware out of the >> box, if it allowed me to log in as 'Root' (Which I sometimes need), and if >> the software on the CDs worked without corrupting the operating system. Of course you can log in as root. The new installer is much improved and is being widely tested. It will never work perfectly for everyone, however. >> But as you can see, my thoughts are not limited to Debian. I feel that >> entire Linux scene is like a toy where children get things almost done, >> then they loose interest and try to pass the buck on to someone else. Yes, that is often what happens. Furthermore, Debian's social structure does not promote collaborative maintenance as much as it could and should. The result is that some Debian packages are buggy and out of date. > we know it is not like that... those guys are working for peanuts, and are > doing a bloody good job. Even Aple is using linux, and they have many good > guys working for them... so it is not like this. Apple's operating system is not based on Linux. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: kernel 2.6.9 and thinkpad source package
On Sun, 02 Jan 2005 13:30:15 +0100, S. Anthony Sequeira wrote: > I have followed the README instructions to compile the thinkpad modules > for my custom 2.6.9 kernel, built using make-kpkg. > > Unfortunately, the thinkpad modules seem to be installed > in /lib/modules/2.6.9 instead of what seems to be the right place for > custom kernels /lib/modules/2.6.9y. > > Any ideas please? Probably you need to give make-kpkg the "--append-to-version=y" option. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: my kde sound problem
On Tue, 04 Jan 2005 18:20:05 +0100, Jose Carlos Garcia Sogo wrote: > ALSA is most advanced, and OSS is deprecated in new kernels. You should > install alsa-base, alsa-utils, alsa-oss (emulation layer for OSS in > ALSA) and libesd0-alsa (yes, though you are using arts for KDE, it > depends on libesd). That should be libesd-alsa0. BTW, you don't need alsa-oss unless you intend to use the aoss program. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: /dev/thinkpad keeps disappearing
On Sat, 29 Jan 2005 22:40:12 +0100, Pollywog wrote: > The bigger problem is that /dev/thinkpad disappears after reboots. If you have udev then the file should reappear when you load the thinkpad module. You can set up a Linux system so that an application's attempt to open a device file results in a module being automatically loaded. On a udev system, however, device files do not exist until modules are loaded, so auto-loading by the latter mechanism is not possible. (devfs was better than udev in this respect: because it lived in the kernel it could detect attempts to open non-existent device files and could load modules and create the device files just-in-time.) If you do this: $ sudo mknod /dev/thinkpad c 10 170 then running tpctl will cause the thinkpad module to be auto-loaded. Alternatively you can add 'thinkpad' to /etc/modules. You should install the latest version (5.8-3) of thinkpad-base in order to get the /etc/udev/permissions.d/thinkpad.permissions file. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: debian-alsa bug?
On Mon, 31 Jan 2005 19:30:14 +0100, Wendy Leigh Vandoolaeghe wrote: > wlv:/home/wlv21# /etc/init.d/alsa stop Storing ALSA mixer > settings...failed. nymphie:/home/wlv21# /etc/init.d/alsa start Starting > ALSA.../etc/init.d/alsa: Error: alsactl restore failed with message 'al > sactl: load_state:1134: No soundcards found...'. done. You need to load the driver modules for your card. See the "loading modules" section of alsa-base README.Debian file. Also, I suggest you upgrade to alsa-base 1.0.8-4 and alsa-utils 1.0.8-1. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: How to use guessnet with wireless
On Tue, 01 Mar 2005 18:00:26 +0100, Otavio Salvador wrote: > I'm really interested in to use the guessnet to setup my wireless > connection but I didn't found docs about it. Use waproamd to associate automatically. Guessnet can detect the network to which the Wi-Fi card has associated. E.g., here is the logical interface definition I use for the wireless network with name 'lubber'. iface home-wireless-static inet static test wireless essid lubber address 192.168.1.9 netmask 255.255.255.0 gateway 192.168.1.1 dns-nameservers 192.168.1.1 -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Help, ALSA and Nvidia
On Tue, 01 Mar 2005 04:40:08 +0100, Rony wrote: > Sounds like radio with no > station, some buzzing sound, even i didn't play any sound. Should i use > "module-assistant auto-install alsa" with kernel-image-2.6.8-2-686? Or > the kernel supports ALSA out of the box? This is the msg from kern.log: > > ALSA > usr/src/modules/alsa-driver/alsa-kernel/pci/ac97/ac97_codec.c:2003: > MC'97 1 converters and GPIO not ready (0xf200) Did a previous version of ALSA work? Which version(s)? -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: make-kpkg won't compile thinkpad-source
On Wed, 09 Mar 2005 16:00:16 +0100, Karl Ebener wrote: > I have an IBM Thinkpad R50 and I want to use the tpctl tools. I > installed thinkpad-source and thinkpad-base. As far as I understood it, > when compiling a kernel with make-kpkg, the thinkpad-modules should be > compiles from thinkpad-source "automatically". First you have to unpack the source tarball manually. cd /usr/src rm -rf modules/thinkpad tar zxvf thinkpad.tar.gz -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: snd-cs4236 and Thinkpad 600E
On Thu, 10 Mar 2005 18:40:21 +0100, Antonio-Blasco Bonito wrote: > FATAL: Error inserting snd_cs4236 > (/lib/modules/2.6.8/kernel/sound/isa/cs423x/snd-cs4236.ko): No such device > FATAL: Error running install command for snd_cs4236 Did you enable all the sound hardware with the DOS utility PS2.EXE? -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: snd-cs4236 and Thinkpad 600E
On Sun, 13 Mar 2005 01:20:08 +0100, Antonio-Blasco Bonito wrote: > % modprobe --ignore-install snd-cs4236 isapnp=0 index=0 port=0x530 > cport=0x538 fm_port=0x388 irq=5 dma1=1 dma2=0 Do you want to automate this? Do you have alsa-base installed? If so then the way to add options to modules is to find the line for snd-cs4236 in /etc/modprobe.d/alsa-base: install snd-cs4236 modprobe --ignore-install snd-cs4236 && /lib/alsa/modprobe-post-install snd-cs4236 and add the required options there, e.g.: install snd-cs4236 modprobe --ignore-install snd-cs4236 dma1=1 dma2=0 && /lib/alsa/modprobe-post-install snd-cs4236 The procedure is slightly different if you are running a 2.4 kernel. Then you have to edit /etc/modutils/alsa-base and add a line like this: options snd-cs4236 dma1=1 dma2=0 -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Problems with /etc/interfaces
On Wed, 06 Apr 2005 18:10:16 +0200, Tim Murison wrote:
> I want to be able to tell /etc/interfaces to only auto enable the
> wireless card, without knowing what that device is.
You mean /etc/network/interfaces, I presume.
> I know how to retrieve the device's name on boot (/proc/net/wireless),
> but I cannot figure out how to combine that knowledge with the rest of
> the /etc/interfaces stuff.
The name assigned to a network interface by the kernel can differ from
occasion to occasion.
There are various ways to deal with this fact.
The easiest way is to install and set up ifrename so that the name
assigned by the kernel ('wifi0' or whatever) is changed to something else
(e.g., 'wifi_a') that is always the same for that card.
--
Thomas Hood
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: apm sleep doesn't work under 2.4.19 and 2.4.20
On Mon, 2003-01-27 at 06:27, Angus McMorland wrote: > This looks a lot like Debian bug #142940, which was closed several > months ago, but, from what I can tell, without any satisfactory > resolution or solution. The solution there is: "Don't call cardctl from apm event scripts". I don't think that this is the same bug. > Does anyone have any ideas? The error is being returned by the firmware. The apm driver simply reports this and exits with an I/O error. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Experimental package available: ifupdown-roaming
ifup. Also make sure that whereami is not called at boot or on APM events. If you want to make use of the nifty scripts in the whereami package then the safest thing to do is install the package, copy the scripts and then purge the package. I would like to know which scripts from other packages (whereami, etc.) you would like to see included in ifupdown-roaming. This includes both custom detection scripts and action scripts such as set-named-forwarders and set-mail-relay. ifupdown-roaming is little more than a proof of concept at the moment, even though it Works For Me (tm). The ifscout script parses /etc/network/interfaces and isn't won't be very tolerant of non-standard formatting at present. You are welcome to try it by downloading the deb: http://panopticon.csustan.edu/thood/ifupdown-roaming.html provided you forgive me for the abominable awk script in ifscout and resolve to send me all and only constructive criticisms (which could includes the criticism "This was a big waste of time, because yadda yadda yadda."). -- Thomas Hood [EMAIL PROTECTED] __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: Experimental package available: ifupdown-roaming
On Wed, 2003-01-29 at 16:20, Sam Clegg wrote: > Great!.. i've been wanting this for ages. Is this packages in > exaperimental now? Do you have any plans for wireless stuff > like automatcially configuring devices when certain SSIDs are > detected? No, it's not in the experimental distro. You can get the latest package via http://panopticon.csustan.edu/thood/ifupdown-roaming.html It will be easy to support everything that whereami supports because we can simply copy over the scripts from that package (Thanks, guys! :) and do a little tweaking to eliminate any side effects on network configuration. (Side effects are OK in whereami, but not in a utility that wants to play nicely with ifupdown.) Since you have expressed interest in SSID detection, I'll work on that today. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: cardmgr, usbmgr, and hotplug
On Thu, 2003-01-30 at 20:43, Jason Kraftcheck wrote: > What is the relation between cardmgr from the pcmcia-cs package, usbmgr, > and hotplug? pcmcia-cs suggests hotplug, but other than that there is no > relation in the package dependencies. > Does hotplug replace cardmgr and/or usbmgr? Hot plug PCI (Cardbus) drivers are not managed by cardmgr; they are managed by the "hotplug" subsystem. > If I have hotplug installed, can I remove pcmcia-cs and usbmgr? > Should I? If you want to use non-Cardbus PCMCIA cards then you still need cardmgr. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: pcmcia freezes
On Fri, 2003-01-31 at 07:14, cyn wrote: > Whenever I eject a pcmcia card, my machine freezes Sounds like you have interrupt conflicts. What card(s) are you using and what IRQs are they assigned? Send your /etc/pcmcia/config.opts, /proc/interrupts, and all other info that you think might be relevant. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
ifupdown-roaming -- automatic network configuration selection for ifup
I have added wireless SSID detection to ifscout, the main component of ifupdown-roaming. ifupdown-roaming is a package that extends the ifupdown package so that it better supports machines that roam from one network environment to another. It is intended to be an easily configured and ifupdown-compatible alternative to whereami, laptop-net, intuitively, and so on. See http://panopticon.csustan.edu/thood/ifupdown-roaming.html for more information and for the latest package. The latest package version is ifupdown-roaming_0.1-10_all.deb but that may have changed by the time you read this. :) Please let me know if and how it works for you. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: pcmcia freezes
On Fri, 2003-01-31 at 14:59, cyn wrote: > Yes, you're right - this was why I enabled acpi in the first place - my > cardbus socket, onboard ethernet, and onboard modem are all using IRQ 11. > I've never had the misfortune to deal with configuring IRQ's (well, not in > a long time) under linux - how do I go about managing this The PCI subsystem usually handles interrupts intelligently. When problems arise they most often involve ISA interrupts. Try adding "exclude irq 3" to your /etc/pcmcia/config.opts. My guess is that the orinoco driver is conflicting with a built-in serial port. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: Choosing network settings at boot?
On Fri, 2003-02-07 at 20:44, Brian Kendig wrote: > What's the package that will let me choose my network settings at boot > time? You may be thinking of netenv. > I want to be able to sometimes boot without a network, without having > to wait a few minutes for DHCP to time out during startup. Remove the name of the offending interface from the list of "auto" interfaces in /etc/network/interfaces? -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: How to disable the 'Stand-By' mode on APM? and ACPI problem...
Try standing by by doing apm --standby instead of pressing the "standby" key on your machine. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: setup apm for 'battery low' warning on textconsole
On Tue, 2003-02-11 at 08:35, Andrew McMillan wrote: > Write a cronjob that logs the battery level apmd already does this. (Be sure to get the apmd from unstable.) -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: ls[wireless]
On Thu, 2003-02-13 at 15:49, Jerome BENOIT wrote: > is there a tool to detect wireless hardware ? If your question is "How do I detect the presence of wireless access points?" then you can try iwlist in the wireless-tools package. I would also like to know whether there is a better tool out there because I would like to integrate it into ifupdown-roaming. (Note: ifupdown-roaming version 0.26 is available at http://panopticon.csustan.edu/thood/ifupdown-roaming.html .) -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: ls[wireless]
On Fri, 2003-02-14 at 12:36, Mark Janssen wrote: > Try Wellenreiter... it scans all channels for accesspoints and wireless > traffic telling you what AP's are available... It's mostly useful for > wardriving. My impression is that this program (like other "war driving" utilities) has to put the 802.11b card into rfmon mode in order to work. Is this correct? I would rather avoid putting the card in rfmon mode when doing location detection because some cards (e.g., the Cisco Aironet) have to be cold reset to get them out of that mode. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: ls[wireless]
On Fri, 2003-02-14 at 17:52, Timothy Ball wrote: > Wow... you're extremely wrong about the aironet card (at least wrt to > the lm350 cards). I'm glad to be informed of that. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: ls[wireless]
On Fri, 2003-02-14 at 19:09, Faye Pearson wrote: > If you look at the source code for iwgetid (google iwgetid.c and click > on the version number in the CVS link that comes up) then you'll find > some sample pseudo code for using it to detect the nearest accesspoint. > > I've modified gentoo's scripts to use it and it works well. > > Basically you just initialise the card as normal for Managed use, bring > the interface up without an IP address, wait for it to sync, then call > iwgetid -s $DEVICE > > Then you ifconfig $DEVICE down and do the setup as normal. Thanks for the tip. Have you tried "iwlist scan"? That's what I'm using now. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com
Re: network hopping tricks: an unsolicited plug for ifupdown-roaming
On Thu, 2003-06-05 at 02:38, Tony Godshall wrote: > There's been a lot of discussion on this list about tools to > detect which network a laptop is connected to and configure > various things. [...] I've tried several, and I think I > advocated whereami to someone at some point whereami works but it is harder to configure than ifscout and (more seriousl) is incompatible with ifupdown, the standard network-interface management package. > but it seems Thomas Hood's > ifupdown-roaming is the best so far! His solution has the > configuration reside entirely in /etc/networks/interfaces. > And in many instances needs almost no configuration at all. Yes. Please note that the features of ifscout are in the process of being integrated into guessnet by the intrepid Enrico Zini. The work is not yet complete; the guessnet release in sid (0.21) is broken and lacks some important features. However, once the work is done, guessnet will work like ifscout, only faster. :) -- Thomas Hood <[EMAIL PROTECTED]>
Re: network hopping tricks: an unsolicited plug for ifupdown-roaming
On Thu, 2003-06-05 at 09:58, David Goodenough wrote: > I am not at all sure I understand what you mean that whereami > and ifupdown are incompatible. I use them both on my machine > all the time. ifupdown assumes that it is in charge of configuring interfaces. If someone else adjusts interface settings then ifupdown gets confused. Whereami does so: it adjusts interface settings to do its probing but does not restore the settings afterward. -- Thomas Hood <[EMAIL PROTECTED]>
[ANNOUNCE] resolvconf -- Auto-manage the resolv.conf file
The resolvconf package contains the infrastructure required for sane updating of the /etc/resolv.conf file as interfaces go up and down and DNS caches start and stop. While resolvconf is appropriate for any networked computer, it is especially useful for roaming machines. The package uses hooks provided by pppd, DHCP clients and ifupdown to gather information about what nameservers are available. Based on this information it maintains a correct resolv.conf file for users of the libc resolver and a different resolv.conf file for dnsmasq or a named.conf options fragment for bind and bind9 named. Resolvconf provides hooks so that applications that use the libc resolver can arrange to be notified when the file changes. This solves the problem of various interface configurers contending for control of /etc/resolv.conf. Because the variable resolv.conf file is stored in /var/run/, resolvconf is also compatible with a read-only root filesystem. For the most part it suffices simply to install the package. Support for dhcp3-client, dhcpcd, ifupdown or laptop-net, and dnsmasq is built in. Although seamless support for resolvconf has not yet made it into the pump and bind(9) packages, the README explains how to configure the latter packages by hand. Resolvconf is available as a .deb in the resolvconf download section of the update-resolv project at alioth: http://alioth.debian.org/projects/update-resolv After installing the package, read the README if you want to gain an understanding of how resolvconf works. Please let me know if you run into any difficulties. -- Thomas Hood GNU/Linux on a ThinkPad 600X page: http://panopticon.csustan.edu/thood/tp600lnx.htm
Re: pcmcia-cs trouble
> Russell Coker wrote: > On Tue, 15 Jul 2003 02:00, Derek Broughton wrote: >> iirc, you didn't have a PCMCIA card, it was a Cardbus card. >> In which case, it should skip the pcmcia scripts and use HotPlug... > > Thanks for this advice. It seems that hotplug was running > "ifup eth0=hotplug", I modified it so that it would run "ifup eth0" > and now it works fine. I'm not sure if this is a bug in hotplug or > in my settings for /etc/network/interfaces. It is a shortcoming in your /etc/network/interfaces. Earlier, hotplug's net.agent script ifupped _any_ interface that was hotplugged. Several people objected to this being the default behavior. So, the hotplug maintainer (using an idea from the ifupdown maintainer) decided to run "ifup IFACE=hotplug" instead. Unless configured to do something with this, ifup will do nothing (thus satisfying the objectors). Ifup can be configured to turn "ifup IFACE=hotplug" into "ifup IFACE" (effectively) by adding the following stanza to /etc/network/interfaces. mapping hotplug script echo If you only want _some_ interfaces (wifi0, say) to be ifupped on hot plug and others to be left alone, then add a stanza like the following. mapping hotplug script grep map wifi0 -- Thomas Hood P.S. Sorry this message won't thread properly.
Re: ppp can't get a route?
On Fri, 2003-07-18 at 11:47, Leandro Guimarães Faria Corsetti Dutra wrote: > > It _should_ be automatic. I use ppp once a week, and ethernet the rest of > > the > > time. When I bring up ppp it always overrides the default route. The ppp > > config parameter is something obscure like "defaultroute" :-) > > Looks like it refuses to overwrite the existing default route... pppd won't replace an existing default route, even if it is given the "defaultroute" option. SuSE patched its version of pppd so that it accepts a "replacedefaultroute" option. Debian users see Debian bugreport #93936. -- Thomas
Re: exim on laptops: smarthost on demand
The ifupdown-roaming package provides a script that simplifies the setting of the smarthost for exim. The script is called 'set-mail-relay'. Here are the comments at the top of the file # Usage: # set-mail-relay MTA none # Do not use a mail relay # set-mail-relay MTA RELAYHOST # Make MTA use specified relay host # set-mail-relay MTA queue RELAYHOST # Place on the queue. # # When the queue is flushed, mail # # will be relayed to RELAYHOST You can use this script even if you don't want to use ifscout, the automatic interface mapper that comes with ifupdown-roaming. You can get the latest package at: http://panopticon.csustan.edu/thood/ifupdown-roaming.html -- Thomas Hood On Mon, 2003-07-28 at 16:23, Christian Lyra wrote: > I use to use the ifupdown-roaming package. With this package the machine > automagically detect the right network and may run custom commands. So, to > solve your problem is only a matter of putting a little script on the "up / > down" network interface section. > > You can have two exim configuration files, let´s say, exim.conf-home > and > exim.conf-other. On the "up" interface command you choose the right one, link > it to exim.conf, and restart exim. > > ah... of course... you can hack /etc/init.d/exim too, and choose the > right > config file at start.
Re: How should I manage multiple wireless lan configs?
> What I would like, of course, is to be able to configure > my Debian laptop so that, when I boot the laptop at any > of these locations, the laptop sees the SSID and > automatically uses the right settings for that location. Normally you can store several keys in the card's nonvolatile RAM. My Aironet 352 holds five keys, for example. Whether or not this is the case with your card, you can use ifscout in the ifupdown-roam package to find the right wireless and network settings automatically. Read especially the README file at: http://panopticon.csustan.edu/thood/ifupdown-roam.html -- Thomas Hood
Re: ACPI and anacron
> So my workaround is: I changed the acpi path from > /proc/acpi/0/ to /proc/acpi/ac_adapter/ in /usr/bin/on_ac_power. Upgrading your powermgmt-base package would have done that too. -- Thomas Hood
Re: Need help on configuring laptop-net
Xavier Maillard wrote: > Anybpdy can help me to configure it please? Get the laptop-net-doc package. For an alternative autoconfigurer, look at ifupdown-roam. http://panopticon.csustan.edu/thood/ifupdown-roam.html -- Thomas Hood
Re: Need help on configuring laptop-net
> What I expect is a tool able to set the WHOLE laptop > configuration according to the current environment. > > For example, at home I use my wired network with a Xinerama'd > Xsession in 16bits. At work, I am using a wireless connection > and no Xinerama. My printers are different, etc... Ifupdown is more capable than a lot of people realize, although of course it should only be used to change the system configuration as a function of the network environment. You can put arbitrary configuration commands on "up" and "down" lines in /etc/network/interfaces. You can arbitrarily shuffle configuration files around as you wish using the mv command. If you want to organize such shuffling you can make use of switchconf. In most cases, though, I find that there is a better way to implement dynamic configuration than by shuffling configuration files. To set the mail relay use the set-mail-relay script that comes in the ifupdown-roam package[1]. To update /etc/resolv.conf use the resolvconf package[2]. To update the printer list use a script like that included in bug report #204847. To mount and umount drives use a script like that included in #204750 . Etc. [1] http://panopticon.csustan.edu/thood/ifupdown-roam.html [2] http://alioth.debian.org/project/showfiles.php?group_id=1227 -- Thomas
[ANNOUNCE] resolvconf -- Manage the resolv.conf file
The resolvconf package contains the infrastructure required for sane updating of the /etc/resolv.conf file as interfaces go up and down and DNS caches start and stop. While resolvconf is appropriate for any networked computer, it is especially useful on roaming machines. The package uses hooks provided by pppd, DHCP clients and ifupdown to gather information about what nameservers are available. Based on this information it maintains a correct resolv.conf file for users of the libc resolver and a different resolv.conf file for dnsmasq or a named.conf options fragment for bind|bind9 named. Resolvconf provides hooks so that applications that use the libc resolver can arrange to be notified when the file changes. (The maintainer of squid has told me that he will use this feature.) Resolvconf solves the problem of various interface configurers contending for control of /etc/resolv.conf. Also, because the variable resolv.conf file is stored by default in /var/run/, resolvconf is compatible with a read-only root filesystem. For the most part it suffices simply to install the package. Support for pump, dhcp3-client, dhcpcd, ifupdown, laptop-net, and dnsmasq is built in. Although support for resolvconf has not yet been built into the bind and bind9 packages, the README explains how to configure the latter by hand. (It's easy.) Resolvconf is now available in experimental. http://packages.debian.org/resolvconf After installing the package, read the README if you want to gain an understanding of how resolvconf works. Please file bugs through the bug tracking system if you run into any difficulties. -- Thomas Hood Resolvconf package: http://packages.debian.org/resolvconf GNU/Linux on a ThinkPad 600X page: http://panopticon.csustan.edu/thood/tp600lnx.htm ifupdown-roam -- roaming extensions for ifupdown -- home page http://panopticon.csustan.edu/thood/ifupdown-roam.html read-only root information page http://panopticon.csustan.edu/thood/readonly-root.html
Re: Two fixed network cards
Derek Broughton wrote: > And Thomas Hood's ifupdown-roaming will handle this automatically, > with a single interfaces file but, if your only way to tell which > net you're on is whether or not you get an IP served by dhcp, > there'll be a 30 second (iirc) delay while it waits for a response > from the server (obviously you can change this value in your dhcp > client). Most of the automatic network detections will look for a > known MAC address, or similar test, to eliminate this delay. The package name has changed to 'ifupdown-roam' and is available at: http://panopticon.csustan.edu/thood/ifupdown-roam.html Another way of reducing the delay is to put the DHCP logical interface last in the list of logical interfaces that ifscout will search. Whether or not you choose to use ifupdown-roam, David Z Maze is correct: you should only ever need _one_ /etc/network/interfaces file. Victor Cain wrote: > I tried ifplugd on a ThinkPad A20P that has a fixed, wired, > network card and a PCMCIA wireless. It did not configure the > wireless card, but ifstatus correctly tells me when the wire is > unplugged. Anyway, if you get it working, maybe you could let > us know what you did. You should try the very latest release of ifplugd and if that doesn't work then continue trying the releases as they come out. The author of ifplugd is still busy improving the wireless network link detection code. -- Thomas Hood
Re: ignore system APM suspend on AC power
Jaume: If you want system suspend requests to be ignored while on ac power, what you would need to do is set SUSPEND_ON_AC=false in /etc/apm/apmd_proxy.conf. The default is "true". Even so, that won't do it. Current kernels do _not_ support the rejection of suspend requests. It is not a question of kernel configuration choices; the APM driver just doesn't support this yet. Chris: It would be a good idea to make the note exit 1 # Reject (NOTE kernel support must be enabled) clearer. You write: > This note just means that you must have APM support compiled into > the kernel and turned on. From what you write that appears to be > true. but of course it is true whenever the apmd_proxy script is running. What you might want to say is: exit 1 # Reject suspend request # Note that this will only work on kernels that allow # user space to reject suspend requests. -- Thomas signature.asc Description: This is a digitally signed message part
Re: APM Suspend on I8K
--- I've got my Dell I8K running Sid with a 2.4.17 kernel and everything works just fine. I only have one problem with the system. If I use Fn + Esc to suspend, or if I close the cover, the BIOS seems to handle the suspend. That is, none of the APM scripts get called. This causes something of a hassle if I come out of suspend when I'm no longer connected to the network, etc. --- Did you answer Yes to "suspend on AC?" when you installed apmd? You should have. Answering No just causes the suspend scripts to be skipped before the suspend. This is a bug. (apmd 3.0.2-1.19 has corrected this problem.) It's possible that the BIOS isn't notifying the OS of Fn-ESC suspend events. --- Using apm -s or "Suspend" from the GDM login menu puts my system into suspend "properly". (i.e. All of the suspend scripts are run.) --- That is probably because a suspend request generated in userspace is forwarded directly to APM event listeners by the apm driver -- this doesn't depend on the BIOS. --- However, when I come OUT of suspend, GDM is dead and needs to be manually restarted. --- It looks as if no resume event is being processed. What kernel are you using? Please cc: me directly as I do not check debian-laptop very often. -- Thomas signature.asc Description: This is a digitally signed message part
nsc-ircc "wrong chip version"
You aren't the only person who has encountered this problem.
You need to remove the "chip version" test (lines 700-706
in the 2.4.19-pre6 driver) from nsc-ircc.c. Then the module
will load and work fine.
The driver isn't wrong according to the specs available for
the 87338/97338. However, some chips just don't contain the
specified module i.d. You report getting 0x16; someone else
reported getting 0xff; I just tried and got 0x00.
I tried modifying the driver to read the version number many
times to see if it was a timing problem, but that didn't help.
IrDA folks: I think the test should be removed.
If it isn't removed, the code should at least be changed so
that instead of "Wrong chip version 16" it says:
Wrong module I.D. 0x1
where 0x1 is the content of the high nybble.
// Thomas Hood
Petr Stanek wrote to debian-laptop:
When Im trying to load the module (Potato kernel 2.4.18-586tsc),
I get the following massage: "Wrong Chip version: 16".
modprobe ircomm-tty nsc-ircc says:
nsc-ircc, Found chip at base=0x02e
nsc-ircc, Wrong chip version 16
Findchip -v says:
Found NSC PC87338 Controller at 0x2e, DevID=0x0b, Rev. 3
SIR Base 0x2f8, FIR Base 0x2f8
IRQ = 3, DMA = 3
Enabled: Yes, Suspended: no
UART compatible: yes
Half duplex delay = 0 us
signature.asc
Description: This is a digitally signed message part
Re: tpctl on a thinkpad
You (who aren't root) need read permission on /dev/smapi in order to examine settings. You also need write permission in order to change settings. Try sudo tpctl --H -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Using a PS/2 mouse, USB wheel mouse, gpm, and X simultaneously?
gpm can only read one mouse at a time SFAIK; however X can handle multiple mice. See the "multiple mice and X" section of my web page for more info: http://panopticon.csustan.edu/thood/tp600lnx.htm#secmouse signature.asc Description: This is a digitally signed message part
Re: ignore system APM suspend on AC power
Jaume: If you want system suspend requests to be ignored while on ac power, what you would need to do is set SUSPEND_ON_AC=false in /etc/apm/apmd_proxy.conf. The default is "true". Even so, that won't do it. Current kernels do _not_ support the rejection of suspend requests. It is not a question of kernel configuration choices; the APM driver just doesn't support this yet. Chris: It would be a good idea to make the note exit 1 # Reject (NOTE kernel support must be enabled) clearer. You write: > This note just means that you must have APM support compiled into > the kernel and turned on. From what you write that appears to be > true. but of course it is true whenever the apmd_proxy script is running. What you might want to say is: exit 1 # Reject suspend request # Note that this will only work on kernels that allow # user space to reject suspend requests. -- Thomas signature.asc Description: This is a digitally signed message part
Re: APM Suspend on I8K
--- I've got my Dell I8K running Sid with a 2.4.17 kernel and everything works just fine. I only have one problem with the system. If I use Fn + Esc to suspend, or if I close the cover, the BIOS seems to handle the suspend. That is, none of the APM scripts get called. This causes something of a hassle if I come out of suspend when I'm no longer connected to the network, etc. --- Did you answer Yes to "suspend on AC?" when you installed apmd? You should have. Answering No just causes the suspend scripts to be skipped before the suspend. This is a bug. (apmd 3.0.2-1.19 has corrected this problem.) It's possible that the BIOS isn't notifying the OS of Fn-ESC suspend events. --- Using apm -s or "Suspend" from the GDM login menu puts my system into suspend "properly". (i.e. All of the suspend scripts are run.) --- That is probably because a suspend request generated in userspace is forwarded directly to APM event listeners by the apm driver -- this doesn't depend on the BIOS. --- However, when I come OUT of suspend, GDM is dead and needs to be manually restarted. --- It looks as if no resume event is being processed. What kernel are you using? Please cc: me directly as I do not check debian-laptop very often. -- Thomas signature.asc Description: This is a digitally signed message part
nsc-ircc "wrong chip version"
You aren't the only person who has encountered this problem.
You need to remove the "chip version" test (lines 700-706
in the 2.4.19-pre6 driver) from nsc-ircc.c. Then the module
will load and work fine.
The driver isn't wrong according to the specs available for
the 87338/97338. However, some chips just don't contain the
specified module i.d. You report getting 0x16; someone else
reported getting 0xff; I just tried and got 0x00.
I tried modifying the driver to read the version number many
times to see if it was a timing problem, but that didn't help.
IrDA folks: I think the test should be removed.
If it isn't removed, the code should at least be changed so
that instead of "Wrong chip version 16" it says:
Wrong module I.D. 0x1
where 0x1 is the content of the high nybble.
// Thomas Hood
Petr Stanek wrote to debian-laptop:
When Im trying to load the module (Potato kernel 2.4.18-586tsc),
I get the following massage: "Wrong Chip version: 16".
modprobe ircomm-tty nsc-ircc says:
nsc-ircc, Found chip at base=0x02e
nsc-ircc, Wrong chip version 16
Findchip -v says:
Found NSC PC87338 Controller at 0x2e, DevID=0x0b, Rev. 3
SIR Base 0x2f8, FIR Base 0x2f8
IRQ = 3, DMA = 3
Enabled: Yes, Suspended: no
UART compatible: yes
Half duplex delay = 0 us
signature.asc
Description: This is a digitally signed message part
Re: tpctl on a thinkpad
You (who aren't root) need read permission on /dev/smapi in order to examine settings. You also need write permission in order to change settings. Try sudo tpctl --H -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Using a PS/2 mouse, USB wheel mouse, gpm, and X simultaneously?
gpm can only read one mouse at a time SFAIK; however X can handle multiple mice. See the "multiple mice and X" section of my web page for more info: http://panopticon.csustan.edu/thood/tp600lnx.htm#secmouse signature.asc Description: This is a digitally signed message part
Re: apm sleep doesn't work under 2.4.19 and 2.4.20
On Mon, 2003-01-27 at 06:27, Angus McMorland wrote: > This looks a lot like Debian bug #142940, which was closed several > months ago, but, from what I can tell, without any satisfactory > resolution or solution. The solution there is: "Don't call cardctl from apm event scripts". I don't think that this is the same bug. > Does anyone have any ideas? The error is being returned by the firmware. The apm driver simply reports this and exits with an I/O error. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Experimental package available: ifupdown-roaming
ifup. Also make sure that whereami is not called at boot or on APM events. If you want to make use of the nifty scripts in the whereami package then the safest thing to do is install the package, copy the scripts and then purge the package. I would like to know which scripts from other packages (whereami, etc.) you would like to see included in ifupdown-roaming. This includes both custom detection scripts and action scripts such as set-named-forwarders and set-mail-relay. ifupdown-roaming is little more than a proof of concept at the moment, even though it Works For Me (tm). The ifscout script parses /etc/network/interfaces and isn't won't be very tolerant of non-standard formatting at present. You are welcome to try it by downloading the deb: http://panopticon.csustan.edu/thood/ifupdown-roaming.html provided you forgive me for the abominable awk script in ifscout and resolve to send me all and only constructive criticisms (which could includes the criticism "This was a big waste of time, because yadda yadda yadda."). -- Thomas Hood [EMAIL PROTECTED] __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Experimental package available: ifupdown-roaming
On Wed, 2003-01-29 at 16:20, Sam Clegg wrote: > Great!.. i've been wanting this for ages. Is this packages in > exaperimental now? Do you have any plans for wireless stuff > like automatcially configuring devices when certain SSIDs are > detected? No, it's not in the experimental distro. You can get the latest package via http://panopticon.csustan.edu/thood/ifupdown-roaming.html It will be easy to support everything that whereami supports because we can simply copy over the scripts from that package (Thanks, guys! :) and do a little tweaking to eliminate any side effects on network configuration. (Side effects are OK in whereami, but not in a utility that wants to play nicely with ifupdown.) Since you have expressed interest in SSID detection, I'll work on that today. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: cardmgr, usbmgr, and hotplug
On Thu, 2003-01-30 at 20:43, Jason Kraftcheck wrote: > What is the relation between cardmgr from the pcmcia-cs package, usbmgr, > and hotplug? pcmcia-cs suggests hotplug, but other than that there is no > relation in the package dependencies. > Does hotplug replace cardmgr and/or usbmgr? Hot plug PCI (Cardbus) drivers are not managed by cardmgr; they are managed by the "hotplug" subsystem. > If I have hotplug installed, can I remove pcmcia-cs and usbmgr? > Should I? If you want to use non-Cardbus PCMCIA cards then you still need cardmgr. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: pcmcia freezes
On Fri, 2003-01-31 at 07:14, cyn wrote: > Whenever I eject a pcmcia card, my machine freezes Sounds like you have interrupt conflicts. What card(s) are you using and what IRQs are they assigned? Send your /etc/pcmcia/config.opts, /proc/interrupts, and all other info that you think might be relevant. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
ifupdown-roaming -- automatic network configuration selection forifup
I have added wireless SSID detection to ifscout, the main component of ifupdown-roaming. ifupdown-roaming is a package that extends the ifupdown package so that it better supports machines that roam from one network environment to another. It is intended to be an easily configured and ifupdown-compatible alternative to whereami, laptop-net, intuitively, and so on. See http://panopticon.csustan.edu/thood/ifupdown-roaming.html for more information and for the latest package. The latest package version is ifupdown-roaming_0.1-10_all.deb but that may have changed by the time you read this. :) Please let me know if and how it works for you. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: pcmcia freezes
On Fri, 2003-01-31 at 14:59, cyn wrote: > Yes, you're right - this was why I enabled acpi in the first place - my > cardbus socket, onboard ethernet, and onboard modem are all using IRQ 11. > I've never had the misfortune to deal with configuring IRQ's (well, not in > a long time) under linux - how do I go about managing this The PCI subsystem usually handles interrupts intelligently. When problems arise they most often involve ISA interrupts. Try adding "exclude irq 3" to your /etc/pcmcia/config.opts. My guess is that the orinoco driver is conflicting with a built-in serial port. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Choosing network settings at boot?
On Fri, 2003-02-07 at 20:44, Brian Kendig wrote: > What's the package that will let me choose my network settings at boot > time? You may be thinking of netenv. > I want to be able to sometimes boot without a network, without having > to wait a few minutes for DHCP to time out during startup. Remove the name of the offending interface from the list of "auto" interfaces in /etc/network/interfaces? -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: How to disable the 'Stand-By' mode on APM? and ACPI problem...
Try standing by by doing apm --standby instead of pressing the "standby" key on your machine. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: setup apm for 'battery low' warning on textconsole
On Tue, 2003-02-11 at 08:35, Andrew McMillan wrote: > Write a cronjob that logs the battery level apmd already does this. (Be sure to get the apmd from unstable.) -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: ls[wireless]
On Thu, 2003-02-13 at 15:49, Jerome BENOIT wrote: > is there a tool to detect wireless hardware ? If your question is "How do I detect the presence of wireless access points?" then you can try iwlist in the wireless-tools package. I would also like to know whether there is a better tool out there because I would like to integrate it into ifupdown-roaming. (Note: ifupdown-roaming version 0.26 is available at http://panopticon.csustan.edu/thood/ifupdown-roaming.html .) -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: ls[wireless]
On Fri, 2003-02-14 at 12:36, Mark Janssen wrote: > Try Wellenreiter... it scans all channels for accesspoints and wireless > traffic telling you what AP's are available... It's mostly useful for > wardriving. My impression is that this program (like other "war driving" utilities) has to put the 802.11b card into rfmon mode in order to work. Is this correct? I would rather avoid putting the card in rfmon mode when doing location detection because some cards (e.g., the Cisco Aironet) have to be cold reset to get them out of that mode. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: ls[wireless]
On Fri, 2003-02-14 at 17:52, Timothy Ball wrote: > Wow... you're extremely wrong about the aironet card (at least wrt to > the lm350 cards). I'm glad to be informed of that. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: ls[wireless]
On Fri, 2003-02-14 at 19:09, Faye Pearson wrote: > If you look at the source code for iwgetid (google iwgetid.c and click > on the version number in the CVS link that comes up) then you'll find > some sample pseudo code for using it to detect the nearest accesspoint. > > I've modified gentoo's scripts to use it and it works well. > > Basically you just initialise the card as normal for Managed use, bring > the interface up without an IP address, wait for it to sync, then call > iwgetid -s $DEVICE > > Then you ifconfig $DEVICE down and do the setup as normal. Thanks for the tip. Have you tried "iwlist scan"? That's what I'm using now. -- Thomas Hood <[EMAIL PROTECTED]> __ Do You Yahoo!? Everything you'll ever need on one web page from News and Sport to Email and Music Charts http://uk.my.yahoo.com -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: network hopping tricks: an unsolicited plug forifupdown-roaming
On Thu, 2003-06-05 at 09:58, David Goodenough wrote: > I am not at all sure I understand what you mean that whereami > and ifupdown are incompatible. I use them both on my machine > all the time. ifupdown assumes that it is in charge of configuring interfaces. If someone else adjusts interface settings then ifupdown gets confused. Whereami does so: it adjusts interface settings to do its probing but does not restore the settings afterward. -- Thomas Hood <[EMAIL PROTECTED]> -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: network hopping tricks: an unsolicited plug forifupdown-roaming
On Thu, 2003-06-05 at 02:38, Tony Godshall wrote: > There's been a lot of discussion on this list about tools to > detect which network a laptop is connected to and configure > various things. [...] I've tried several, and I think I > advocated whereami to someone at some point whereami works but it is harder to configure than ifscout and (more seriousl) is incompatible with ifupdown, the standard network-interface management package. > but it seems Thomas Hood's > ifupdown-roaming is the best so far! His solution has the > configuration reside entirely in /etc/networks/interfaces. > And in many instances needs almost no configuration at all. Yes. Please note that the features of ifscout are in the process of being integrated into guessnet by the intrepid Enrico Zini. The work is not yet complete; the guessnet release in sid (0.21) is broken and lacks some important features. However, once the work is done, guessnet will work like ifscout, only faster. :) -- Thomas Hood <[EMAIL PROTECTED]> -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
[ANNOUNCE] resolvconf -- Auto-manage the resolv.conf file
The resolvconf package contains the infrastructure required for sane updating of the /etc/resolv.conf file as interfaces go up and down and DNS caches start and stop. While resolvconf is appropriate for any networked computer, it is especially useful for roaming machines. The package uses hooks provided by pppd, DHCP clients and ifupdown to gather information about what nameservers are available. Based on this information it maintains a correct resolv.conf file for users of the libc resolver and a different resolv.conf file for dnsmasq or a named.conf options fragment for bind and bind9 named. Resolvconf provides hooks so that applications that use the libc resolver can arrange to be notified when the file changes. This solves the problem of various interface configurers contending for control of /etc/resolv.conf. Because the variable resolv.conf file is stored in /var/run/, resolvconf is also compatible with a read-only root filesystem. For the most part it suffices simply to install the package. Support for dhcp3-client, dhcpcd, ifupdown or laptop-net, and dnsmasq is built in. Although seamless support for resolvconf has not yet made it into the pump and bind(9) packages, the README explains how to configure the latter packages by hand. Resolvconf is available as a .deb in the resolvconf download section of the update-resolv project at alioth: http://alioth.debian.org/projects/update-resolv After installing the package, read the README if you want to gain an understanding of how resolvconf works. Please let me know if you run into any difficulties. -- Thomas Hood GNU/Linux on a ThinkPad 600X page: http://panopticon.csustan.edu/thood/tp600lnx.htm -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: pcmcia-cs trouble
> Russell Coker wrote: > On Tue, 15 Jul 2003 02:00, Derek Broughton wrote: >> iirc, you didn't have a PCMCIA card, it was a Cardbus card. >> In which case, it should skip the pcmcia scripts and use HotPlug... > > Thanks for this advice. It seems that hotplug was running > "ifup eth0=hotplug", I modified it so that it would run "ifup eth0" > and now it works fine. I'm not sure if this is a bug in hotplug or > in my settings for /etc/network/interfaces. It is a shortcoming in your /etc/network/interfaces. Earlier, hotplug's net.agent script ifupped _any_ interface that was hotplugged. Several people objected to this being the default behavior. So, the hotplug maintainer (using an idea from the ifupdown maintainer) decided to run "ifup IFACE=hotplug" instead. Unless configured to do something with this, ifup will do nothing (thus satisfying the objectors). Ifup can be configured to turn "ifup IFACE=hotplug" into "ifup IFACE" (effectively) by adding the following stanza to /etc/network/interfaces. mapping hotplug script echo If you only want _some_ interfaces (wifi0, say) to be ifupped on hot plug and others to be left alone, then add a stanza like the following. mapping hotplug script grep map wifi0 -- Thomas Hood P.S. Sorry this message won't thread properly. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: ppp can't get a route?
On Fri, 2003-07-18 at 11:47, Leandro Guimarães Faria Corsetti Dutra wrote: > > It _should_ be automatic. I use ppp once a week, and ethernet the rest of the > > time. When I bring up ppp it always overrides the default route. The ppp > > config parameter is something obscure like "defaultroute" :-) > > Looks like it refuses to overwrite the existing default route... pppd won't replace an existing default route, even if it is given the "defaultroute" option. SuSE patched its version of pppd so that it accepts a "replacedefaultroute" option. Debian users see Debian bugreport #93936. -- Thomas -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: exim on laptops: smarthost on demand
The ifupdown-roaming package provides a script that simplifies the setting of the smarthost for exim. The script is called 'set-mail-relay'. Here are the comments at the top of the file # Usage: # set-mail-relay MTA none # Do not use a mail relay # set-mail-relay MTA RELAYHOST # Make MTA use specified relay host # set-mail-relay MTA queue RELAYHOST # Place on the queue. # # When the queue is flushed, mail # # will be relayed to RELAYHOST You can use this script even if you don't want to use ifscout, the automatic interface mapper that comes with ifupdown-roaming. You can get the latest package at: http://panopticon.csustan.edu/thood/ifupdown-roaming.html -- Thomas Hood On Mon, 2003-07-28 at 16:23, Christian Lyra wrote: > I use to use the ifupdown-roaming package. With this package the machine > automagically detect the right network and may run custom commands. So, to > solve your problem is only a matter of putting a little script on the "up / > down" network interface section. > > You can have two exim configuration files, let´s say, exim.conf-home and > exim.conf-other. On the "up" interface command you choose the right one, link > it to exim.conf, and restart exim. > > ah... of course... you can hack /etc/init.d/exim too, and choose the right > config file at start. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: ACPI and anacron
> So my workaround is: I changed the acpi path from > /proc/acpi/0/ to /proc/acpi/ac_adapter/ in /usr/bin/on_ac_power. Upgrading your powermgmt-base package would have done that too. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: How should I manage multiple wireless lan configs?
> What I would like, of course, is to be able to configure > my Debian laptop so that, when I boot the laptop at any > of these locations, the laptop sees the SSID and > automatically uses the right settings for that location. Normally you can store several keys in the card's nonvolatile RAM. My Aironet 352 holds five keys, for example. Whether or not this is the case with your card, you can use ifscout in the ifupdown-roam package to find the right wireless and network settings automatically. Read especially the README file at: http://panopticon.csustan.edu/thood/ifupdown-roam.html -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Need help on configuring laptop-net
Xavier Maillard wrote: > Anybpdy can help me to configure it please? Get the laptop-net-doc package. For an alternative autoconfigurer, look at ifupdown-roam. http://panopticon.csustan.edu/thood/ifupdown-roam.html -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Need help on configuring laptop-net
> What I expect is a tool able to set the WHOLE laptop > configuration according to the current environment. > > For example, at home I use my wired network with a Xinerama'd > Xsession in 16bits. At work, I am using a wireless connection > and no Xinerama. My printers are different, etc... Ifupdown is more capable than a lot of people realize, although of course it should only be used to change the system configuration as a function of the network environment. You can put arbitrary configuration commands on "up" and "down" lines in /etc/network/interfaces. You can arbitrarily shuffle configuration files around as you wish using the mv command. If you want to organize such shuffling you can make use of switchconf. In most cases, though, I find that there is a better way to implement dynamic configuration than by shuffling configuration files. To set the mail relay use the set-mail-relay script that comes in the ifupdown-roam package[1]. To update /etc/resolv.conf use the resolvconf package[2]. To update the printer list use a script like that included in bug report #204847. To mount and umount drives use a script like that included in #204750 . Etc. [1] http://panopticon.csustan.edu/thood/ifupdown-roam.html [2] http://alioth.debian.org/project/showfiles.php?group_id=1227 -- Thomas -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
[ANNOUNCE] resolvconf -- Manage the resolv.conf file
The resolvconf package contains the infrastructure required for sane updating of the /etc/resolv.conf file as interfaces go up and down and DNS caches start and stop. While resolvconf is appropriate for any networked computer, it is especially useful on roaming machines. The package uses hooks provided by pppd, DHCP clients and ifupdown to gather information about what nameservers are available. Based on this information it maintains a correct resolv.conf file for users of the libc resolver and a different resolv.conf file for dnsmasq or a named.conf options fragment for bind|bind9 named. Resolvconf provides hooks so that applications that use the libc resolver can arrange to be notified when the file changes. (The maintainer of squid has told me that he will use this feature.) Resolvconf solves the problem of various interface configurers contending for control of /etc/resolv.conf. Also, because the variable resolv.conf file is stored by default in /var/run/, resolvconf is compatible with a read-only root filesystem. For the most part it suffices simply to install the package. Support for pump, dhcp3-client, dhcpcd, ifupdown, laptop-net, and dnsmasq is built in. Although support for resolvconf has not yet been built into the bind and bind9 packages, the README explains how to configure the latter by hand. (It's easy.) Resolvconf is now available in experimental. http://packages.debian.org/resolvconf After installing the package, read the README if you want to gain an understanding of how resolvconf works. Please file bugs through the bug tracking system if you run into any difficulties. -- Thomas Hood Resolvconf package: http://packages.debian.org/resolvconf GNU/Linux on a ThinkPad 600X page: http://panopticon.csustan.edu/thood/tp600lnx.htm ifupdown-roam -- roaming extensions for ifupdown -- home page http://panopticon.csustan.edu/thood/ifupdown-roam.html read-only root information page http://panopticon.csustan.edu/thood/readonly-root.html -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Two fixed network cards
Derek Broughton wrote: > And Thomas Hood's ifupdown-roaming will handle this automatically, > with a single interfaces file but, if your only way to tell which > net you're on is whether or not you get an IP served by dhcp, > there'll be a 30 second (iirc) delay while it waits for a response > from the server (obviously you can change this value in your dhcp > client). Most of the automatic network detections will look for a > known MAC address, or similar test, to eliminate this delay. The package name has changed to 'ifupdown-roam' and is available at: http://panopticon.csustan.edu/thood/ifupdown-roam.html Another way of reducing the delay is to put the DHCP logical interface last in the list of logical interfaces that ifscout will search. Whether or not you choose to use ifupdown-roam, David Z Maze is correct: you should only ever need _one_ /etc/network/interfaces file. Victor Cain wrote: > I tried ifplugd on a ThinkPad A20P that has a fixed, wired, > network card and a PCMCIA wireless. It did not configure the > wireless card, but ifstatus correctly tells me when the wire is > unplugged. Anyway, if you get it working, maybe you could let > us know what you did. You should try the very latest release of ifplugd and if that doesn't work then continue trying the releases as they come out. The author of ifplugd is still busy improving the wireless network link detection code. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Apmd Oops -> ALSA Fatal Error
> apmd oopses [...] > * (not important, but funky) Screen remains blank until > I attempt to switch to a different VC. Then all of a sudden > XWindows appears, even though keyboard input goes into > whatever tty I supposedly switched to. Hitting return > causing the display to shift left an inch or so, wrapping > around to the right side. This is fixed by "switching back" > to the X's VC. One of the PCI drivers appears to be oopsing in its PM resume code. Try upgrading the kernel. -- Thomas Hood <[EMAIL PROTECTED]> -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: APM: will not sleep with AC plugged in?
User space cannot veto APM suspend requests in current Linux kernels. What is keeping your machine from suspending is the firmware. If you have a PCMCIA card then try ejecting it before attempting to suspend. -- Thomas Hood
Re: Apmd Oops -> ALSA Fatal Error
> apmd oopses [...] > * (not important, but funky) Screen remains blank until > I attempt to switch to a different VC. Then all of a sudden > XWindows appears, even though keyboard input goes into > whatever tty I supposedly switched to. Hitting return > causing the display to shift left an inch or so, wrapping > around to the right side. This is fixed by "switching back" > to the X's VC. One of the PCI drivers appears to be oopsing in its PM resume code. Try upgrading the kernel. -- Thomas Hood <[EMAIL PROTECTED]>
Re: persistent interface
> My problem has to do with the network interfaces. My > /etc/network/interfaces file looks like this: > ># The loopback interface ># automatically added when upgrading >auto lo >iface lo inet loopback > ># For the built-in NIC: >iface eth0 inet dhcp > ># For the wireless card ># >iface eth1 inet dhcp >mapping hotplug >script echo It looks as if you are under the impression that "mapping" is an option for "iface eth1 inet dhcp". Actually it should be a separate stanza. Here is your file reformatted to show the effective structure better: # Physical interfaces to bring up at boot time auto lo # Mappings # Bring up hot plugged interfaces with logical interface name # the same as their physical interface name mapping hotplug script echo # Logical interface lo # The logical interface as which the loopback interface is brought up iface lo inet loopback # Logical interface eth0 # The logical interface as which the built-in NIC eth0 is brought up iface eth0 inet dhcp # Logical interface eth1 # The logical interface as which the Wi-Fi card is brought up iface eth1 inet dhcp > I think that this should bring up only the loopback interface at > boot-time. Strictly speaking, this means that only the loopback interface "lo" is brought up by /etc/rcS.d/S40networking. Other interfaces can be brought up while the system is booting via the hotplug mechanism. Your stanza mapping hotplug script echo enables this mechanism. > Eth1 (pcmcia wireless card) should be brought up when the > hotplug system detects an insertion, and eth0 (associated with the > built-in NIC) should be brought up only when ifup is executed. > And indeed when I do: > >/etc/init.d/networking stop >/etc/init.d/networking start > > after the boot process, this is exactly what happens. Right. > The problem, though, is that at boot-time, the system always tries to > bring up eth0 (which corresponds to the built-in NIC). The kernel detects the interface and calls hotplug to notify user space that the interface is available. [...] > suggests that the attempt to bring up eth0 is triggered by loading of > the driver module (8139too), and indeed lsmod reports that this module > is loaded by the end of the boot process. Correct. > What I *cannot* for the life of me figure out is why this is > happening. I've been all through the directory /etc/modprobe.d/ and > there's no reference there to either 8139too or eth0. I've run > update-modules to make sure that there was no ancient module > configuration haunting the system. There is no reference to either > 8139too or to eth0 in /lib/modules/modprobe.conf. > > Can anyone tell me where else I could look to pin this down and figure > out why it is happening? I don't know who is loading the module, sorry. However, you can prevent eth0 from being ifup'ped via the hotplug mechanism by changing the mapping stanza to this: mapping hotplug script grep map eth1 Then only eth1 will be brought up on hotplug. -- Thomas Hood A man does not show his greatness by being at one extremity, but rather by touching both at once.-- Pascal
Re: WLAN-Card on Sarge
> Ok, now I'm confused. In the pcmcia-cs package there's the cardmgr > program that, if I understand correctly, reads /etc/pcmcia/config* files > on startup. Then when a card is inserted cardmgr matches the card to a > device id (from the config file(s)) and load modules. It then loads the > associated "class" script -- which is the network.opts script you > mentioned, and if nothing matches there (on Debian) ifup or ifdown is > called. And as you say, ifup/ifdown use /etc/network/interfaces to > configure things. That's right. > But then I see you have a "mapping hotplug" stanza. So is hotplug > managing your pcmcia card inserts? That is, what's confusing me is I > though[t] pcmcia-cs (i.e. cardmgr) handled this, yet you are showing a > stanza defined for hotplug events. cardmgr manages PCMCIA cards via /etc/pcmcia/network. Hotplug can do the same thing but it is disabled by default. If you enable the hotplug mechanism by adding mapping hotplug script echo to /etc/network/interfaces then you should disable the cardmgr mechanism by putting an exit 0 at the top of /etc/pcmcia/network. > In my network.opts script I modified it to bring down eth0 before > bringing up the wireless interface (eth1 in my case), and when the card > is removed call ifup eth0. What I'd like to figure out is how to not > bring up eth0 unless a cable is plugged in. You can use ifplugd for that. > I'm having so many problems with wireless now. I think my hardware must > be failing as I'm having the same problems with different drivers. > Lately when I remove my wireless card my machine will freeze hard. > Seems like these problems started with upgrading to 2.6.5, but I also > suspect my wireless hardware is failing since I often have to hard reset > my AP. It is most likely buggy drivers causing lockups. Expect further problems. Debian's support for dynamic configuration sucks. To gain an understanding of how badly it sucks, start reading the bugs open against hotplug and ifupdown. -- Thomas Hood
Re: WLAN-Card on Sarge
> So, what's the advantage of using hotplug for bringing up the interface > over cardmgr running /etc/pcmcai/network(.opts)? Hotplug is the buggy work-in-progress standard mechanism of the future whereas cardmgr is the debugged bloated obsolete mechanism of the past. Take your pick. > I tried that once, and maybe didn't understand the config. I was > actually looking for a way in /etc/network/interfaces to run a script > and have the exit code control how the interface is brought up. Look at the ifupdown package: http://panopticon.csustan.edu/thood/ifupdown-roam.html Even if you don't want to install the package, the README file may be helpful to you. > > Expect further problems. Debian's support for dynamic configuration > > sucks. > > To gain an understanding of how badly it sucks, start reading the bugs > > open against hotplug and ifupdown. > > Is that a Linux issue or Debian? It's a Debian issue. The hotplug and ifupdown maintainers are inactive. However, I don't know of any other distros that have better support for automatic dynamic configuration of laptops. Do you? > I did notice that with 2.4.21 and wlan-ng that cardmgr was better at > removing the card -- I'd pull the card and the drivers would be removed > without delay. Now with 2.6.5 and Orinoco|HostAP there's either a long > delay or a system hang. I don't think that's a cardmgr issue, though -- > as I've tried manually removing the drivers (ifdown eth1; rmmod > ) and have the same problems. So, I'll agree that it's either a > problem with the driver dealing with being removed, or the kernel. I should make one thing clear. If you have PCMCIA cards then you always need cardmgr per se. When you put an "exit 0" at the top of /etc/pcmcia/network you only disable the interface configuration code. However, cardmgr is still used to load drivers, to control the sockets, to beep, etc. -- Thomas Hood
Re: problems getting static ip to work
> I have a laptop running Debian, whose network interface is a pcmcia card. > I use it as an IMAP mail server. Recently, I tried giving it a static IP > address. > by changing /etc/pcmcia/network.opts and filling in the appropriate > fields (I actually used pcnetconfig, so it generated the file for me) > My /etc/network/interfaces looks like this > auto lo > iface lo inet loopback > eth0 > iface eth0 inet static > address 192.168.2.6 > netmask 255.255.255.0 > network 192.168.2.0 > gateway 192.168.2.1 This is invalid. The 'eth0' on its own line shouldn't be there. The rest is OK: auto lo iface lo inet loopback iface eth0 inet static address 192.168.2.6 netmask 255.255.255.0 network 192.168.2.0 gateway 192.168.2.1 > ifconfig -a > does seem to give me information that eth0 is configured (I can even it > ping it from another machine on the network) > > I am noticing 2 problems. > > On boot up I get a "neighbor table overflow" message. I was told this > has something to do with the loopback not being configured, but I > thought my interfaces file should take care of this. Yes it should. > More importantly, my IMAP daemon does not seem to be retrieving any mail > from my Pop server. > I tried doing a > % telnet 127.0.0.1 imap2 > > and it hangs.. > > I think I'm going in circles here and I'd appreciate it if someone can > explain to me what the problem is and how I might fix it? Try eliminating the 'network' line from the eth0 iface definition. -- Thomas
Re: persistent interface
> Is there good documentation somewhere about mapping stanzas in the > interfaces file? I've read the `interfaces' man page, and the stuff in > /usr/share/doc/ifupdown. They're both very helpful, but a bit cryptic. You might also want to read the Networking chapter of the Debian Reference. -- Thomas Hood
Re: no sound on thinkpad 600 - kernel 2.6.8
On Fri, 17 Sep 2004 06:00:11 +0200, Carlos Gabriel Drach wrote: > Hi, i have no sound (i cant hear nothing), but see that modules > (cs46xx) are loaded. cs46xx is not an ALSA driver; it is an OSS driver. What you want is snd-cs46xx. -- Thomas Hood
Re: failed ifup leaves device up
On Fri, 15 Jul 2005 22:09:12 +0200, Matej Cepl wrote: > whenever turning up network device fails (either because wireless signal is > too weak, or because a server, which is defined as a gateway is down) it > seems to me that ifup doesn't down network device, so that when I re-try > ifup later it fails because the device is already up. Is it a bug in > ifplugd, ifupdown, or my misconfiguration? The behavior is documented in interfaces(5). See also bug #286148. -- Thomas Hood -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
