Jump to content
Compatible Support Forums

martouf

Members
  • Content count

    335
  • Joined

  • Last visited

    Never

Posts posted by martouf


  1. I too have a Dimension 4550 system.

     

    If I had it to do again, I'd be sure to download the BIOS version you have now

    in addition to the new version you intend to try.

     

    I ended up putting the A06 BIOS back because the A08 BIOS messed up the IRQ

    routing where the Atlantis PCI video I/O card wouldn't work with any OS.

     

    As I understand the Dell flash process: the new BIOS is written to a secondary

    location and checksummed. If the checksum passes, the secondary location

    is changed to be the primary. The one time I experienced an error,

    the end result was no change made to the BIOS.

     

    It seems they've implemented a two-phase commit.

     

    I still plug the system into a UPS before I update the BIOS, though. 8)


  2. under Windows, can you use device manager (Start->Run->"devmgmt.msc" ),

    find the SiS network adapter, pull up the properties (right click) and

    report the version number of your driver?

     

    according to this article, you need v1.16 or later

    before the correct MAC address can be revealed by "ipconfig /all".

     

    also, under Linux, when your network interface driver loads correctly

    you will see something that looks like this in /var/log/messages (or via "dmesg" ):

     

    [size:3][tt]sis900.c: v1.07.06 11/07/2000

    eth0: SiS 900 PCI Fast Ethernet at 0xd000, IRQ 10, 00:00:e8:83:7f:a4.

    eth0: SiS 900 Internal MII PHY transceiver found at address 1.

    eth0: Using SiS 900 Internal MII PHY as default[/tt][/color]

     

    Note carefully the first 3 octets of the MAC address.

     

    The MAC address is assigned by SiS (id 0x0000e8) and is unique for the example

    network interface (nbr 0x837fa4). Besides the ethernet broadcast address,

    there is a whole range of addresses reserved for ethernet multicast which

    you should not assign to your network interface.

     

    Instead of trying to reassign your interface's address, I think it would be best

    to fix the driver problems so you will end up using the MAC address already

    allocated for your interface by SiS.

     

    http://www.sis.com/support/index.htm

     

    what is the full identification string returned by "lspci -v" for the interface?

     

    for Windows, the latest driver appears to be v1.18 and for Linux it's v1.08.

     

    Compare your Windows network driver to the contents of the LAN driver archive from ECS.

     


  3. Quote:
    I believe the 'Ethernet HWaddr FF:FF:FF:FF:FF:FF' is the mac address.

    can't be. sorry. all bits on is the ethernet broadcast address.

    Until ifconfig shows the same MAC address as Windows, then you've
    got some kind of configuration problem.


    Quote:
    when i look in windows it doesn't tell me the mac at all! How can I discover that?

    (assuming WinXP)
    Start->Run-> "cmd" -> then "ipconfig /all"

    You'll get full interface information, IP address and MAC address.


    In linux, if you check the output of 'dmesg' or the contents
    of /var/log/messages, does it say anything interesting about 'eth0'?

  4. Yes, the referenced Buffalo USB uses the Broadcom 54g chipset.

     

    As best I recall, the necessary WinXP inf file adds a windriver named 'netu2g54'.

     

    it may help the troubleshooting if you could cut-n-paste the ndiswrapper entries

    from the file /var/log/messages or the command dmesg ..

     

    assuming you've installed the necessary WinXP driver using the ndiswrapper tool

    ("ndiswrapper -i nameofthewindriver.inf"\) and the driver is listed using the tool

    ("ndiswrapper -l"\):

     

    if there's a log entry about disabling an interrupt after you've loaded

    the ndiswrapper module, then you may need to modify your 2.6 kernel boot

    parameters by adding "acpi=noirq".


  5. (frizzo: the 1394 Net Adapter is your ieee1394 aka Firewire interface)

     

    Marbin, famorim, hamade: yeah, you've got it right collectively..

    the Compaq support guy was trying to be helpful but was flat wrong...

     

    there are no Broadcom chipset drivers available for Linux. Just Windows.

     

    your choices are DriverLoader (by Linuxant) or ndiswrapper (probably already included

    in the Mepis distro) and use the 'bcmwl5a.inf' file.

     

    I use ndiswrapper with both a miniPCI and a PC Card (cardbus) wlan card having

    the Broadcom 54g chipset.

     

    Marbin: you may find 'acpi=noirq' is enough to turn off only the problematic part

    of ACPI but leave the useful parts enabled.


  6. as I read the posts in the Red Hat forums, to build madwifi from sources means

    you must first either install the 'kernel-debuginfo' package or you must build a new

    kernel yourself. Building madwifi depends on having the full set of kernel build 'leftovers'.

     

    This is not a thing for the noob.

     

    Instead, visit http://atrpms.net/dist/fc3/madwifi/ and pick one of the pre-compiled

    packages for your particular flavor of FC3 from the top of the list.

     

    (Note the source RPM you've already grabbed is at the bottom of the list)

     


  7. 1a. try 'rpm -qa | grep madwi' to see if you already have the madwifi package

    installed.

    In some distros, the madwifi drivers are just rolled into the kernel package

    so for those you must check the kernel modules for 'ath_hal' and 'ath_pci'.

    1b. "find /lib/modules -name \*ath\*" (you'll need the backslashes)

     

    2. you need to check "/var/log/messages" or 'dmesg' for an indication the

    madwifi drivers appropriate for your wireless card are loading.

     

    -> "grep ath /var/log/messages" or "dmesg | grep ath"

    -> "grep wlan /var/log/messages" or "dmesg | grep wlan"

     

    3. if the drivers are loaded, then you should find an "ath0" interface

    listed with 'iwconfig'. if they aren't loaded, you may need to 'modprobe ath_pci'

    and go back to Step 2.


  8. flippantly: because you're running a multi-user operating system. laugh

     

    seriously though, many applications are designed to store configuration data

    and user files in a user's part of the filesystem while the application binaries

    and libraries are stored in the system's part of the filesystem for reasons of sharing

    and protecting the application itself from the actions of users.

     

    this separation of user config+data from app bin+libs is also reflected in the filesystem

    layout and usage accounting in that the /usr/bin directory often lives on a partition

    or disk separate from /home.

     

    Similarly, the system backup strategy is positively affected since the infrequently changing

    and possibly quite large collection app bin+libs files need backing up less often

    than the frequently changing and often relatively small collection of user config+data files.


  9. Applicable extracts from the GRUB Manual:

     

    Quote:
    password

     

    password [--md5] passwd [new-config-file]

    If used in the first section of a menu file, disable all interactive editing control (menu entry editor and command-line) and entries protected by the command lock. If the password passwd is entered, it loads the new-config-file as a new config file and restarts the GRUB Stage 2, if new-config-file is specified. Otherwise, GRUB will just unlock the privileged instructions. You can also use this command in the script section, in which case it will ask for the password, before continuing.

    The option --md5 tells GRUB that passwd is encrypted with md5crypt (see md5crypt).

     

    Quote:
    lock

    Prevent normal users from executing arbitrary menu entries. You must use the command password if you really want this command to be useful (see password).

     

    This command is used in a menu, as shown in this example:

     

    [tt] title This entry is too dangerous to be executed by normal users

    lock

    root (hd0,a)

    kernel /no-security-os[/tt]

     

    In other words:

    If you set a password, then all the boot entries are fixed "as is" and aren't

    editable unless you supply the password.

    In addition to the above, any entry marked with "lock" can't be booted unless

    you supply the password.

     

    crystal clear? got it?

     

    cheers!

     

     


  10. In Windows, do File->Run->"cmd" then "ipconfig /all"

    and then write down the DNS/nameserver IP addresses for

    the LAN interface.

     

    Then, in linux, edit /etc/resolv.conf to add the lines

    "nameserver 1.2.3.4

    nameserver 1.2.3.5" (only use the two IP addrs collected from Windows)

     

    after editing the file, you don't need to reboot.

    Try again with Mozilla.

     

    cheers!

     


  11. put simply:

    dangerous, perhaps, but only to -your- user account. it's not shared with

    any of the other user accounts you may have on the system.

     

    stuff installed as root usually means software installed for all user accounts

    on the system. shared applications.

     

    if your user account does not have any special privileges, then software

    installed by and running as your user id is not very dangerous.

     


  12. ok, you've turned K56flex off (s109=2) and set max incoming rate (s38=1),

    but you've set max outgoing rate to a fixed 21600 bps (s37=14).

     

    why set the outgoing rate to anything but auto (s37=0) ?

     

    I know there's a "suggested" init string in the manual, but my experience tells me

    not to trust it very much. I would take it as an example of syntax.

     

    I also know the initial handshaking at both ends can go weird if it doesn't

    progress through all of the "expected" states (including trying to negotiate

    an unsupported protocol (like K56flex)).

     

    Have you already tried s37=0 and s109=1 ?

    (I encountered a source here which gives a Lucent chipset recommendation

    of trying s38=0 )

    [Edited by martouf on 2005-02-02 13:20:27]

     


  13. here's an outline of one way to approach your goal:

     

    1. create a boot floppy for the RH9 system

     

    2. install PartitionMagic on another system, and create the

    'emergency' recovery disks (or whatever PartitionMagic calls it).

     

    3. use the PM disks on the RH9 system to resize the existing partitions

    to make room for WinXP.

     

    4. install WinXP in the free space.

     

    5. since WinXP has probably overwritten the boot sector, use the RH9 boot

    floppy to reinstall lilo/grub.

     

    6. enjoy your dual boot system.

     

    if I've left out an important detail, I'm sure someone else here will

    fill in the missing information...

     


  14. hmm.. drmagoo: yes, the acx100 project folks document the WG311v2 Rev A1 as using

    the ACX111 chipset.

     

    Wow, someone please email bruce@linux-wlan.com with the updated Netgear WG311 info!

    The great resource http://www.linux-wlan.org/docs/wlan_adapters.html.gz

    lists only "WG311" (no "v1" and no "v2" ).

     

    nicevision2020: are you going to need WEP support?

     

    If so, then the ACX100 driver does not yet support WEP on the ACX111 chipset.

    If you need WEP or WPA, then you'll have to use ndiswrapper for now.


  15. I'm finding the phrase "ITU V.92 and companion V.44 standard".

     

    Best I can tell, the modem should auto-negotiate this compression protocol

    with the modem at the other end.

     

    Default values (see Actiontec manual) appear to have the modem attempt the

    highest data rates supported by line conditions.

     

    genuse22: are you trying to squeeze a few more Kbps out of your link?

     


  16. check the nmbd log file carefully. if necessary, increase its verbosity

    so you can check the results of the elections.

     

    As I recall, the default 'os level' is set to be the same as most Windows server products.

    I remember the documentation saying it should be sufficient to win the elections,

    but in practice I found the only way was to _ensure_ it would win by setting it one

    higher than the highest I found in the log trace.

     

    You're right 'domain master browser' is a special kind of master browser

    reserved only for the PDC of a domain.

     

    The local master browser becomes the 'keeper of the names' on the network.

    The winner of the election becomes the master for the entire workgroup,

    even if that's your slowest and least reliable PC on the network!

     

    Bleccchh! frown

     

    I say force the issue by choosing one and only one host to be the master browser.

    Samba gives you the means to make it be one all of the Windows hosts choose.

     

    A quick scan of the Samba docs says you'll want:

    preferred master = auto

    local master = yes

    domain master = no

     

    and probably:

    os level = 129

     

     

×