Bug#440161: wrong /etc/fstab when installing on RAID using Adaptec 2100S - Debian

This is a discussion on Bug#440161: wrong /etc/fstab when installing on RAID using Adaptec 2100S - Debian ; I have looked into the problem of using Adaptec 2100S with Debian Edu/Etch, have have a few observations. The problem seem to originate from the fact that two different kernel modules, dpt_i2o and i2o_block/i2o_core support these cards, and that these ...

+ Reply to Thread
Results 1 to 6 of 6

Thread: Bug#440161: wrong /etc/fstab when installing on RAID using Adaptec 2100S

  1. Bug#440161: wrong /etc/fstab when installing on RAID using Adaptec 2100S


    I have looked into the problem of using Adaptec 2100S with Debian
    Edu/Etch, have have a few observations. The problem seem to originate
    from the fact that two different kernel modules, dpt_i2o and
    i2o_block/i2o_core support these cards, and that these modules produce
    different device nodes /dev/sda (dpt_i2o) and /dev/i2o/hda
    (i2o_block). During boot, it seem to be arbitary which of the modules
    are loaded by udev. If one is loaded during installation, and another
    is loaded during boot, the device path in /etc/fstab is wrong and the
    boot fail. The selected kernel module can change from boot to boot.

    A quick look in /lib/modules/2.6.18-5-686/modules.pcimap tell me that
    the conflicting kernel module issue only affect one PCI id. These are
    the devices supported by the two modules:

    1044:a511 dpt_i2o i2o_core
    1044:a501 dpt_i2o
    8086:1962 i2o_core

    There are at least three problems in Etch related to this issue.

    - The i2o_core and i2o_block modules is missing in the
    scsi-extra-modules udeb.
    - The automatic partitioner do not accept /dev/i2o/hda as an hard
    drive, and fail.
    - Any of the drivers might grab the device during boot, and this is
    not consistent from boot to boot.

    The first one was partly addressed in d-i svn by replacing dpt_i2o
    with i2o_core, but it will break support for PCI id 1044:a501. A fix
    for this is to reinsert dpt_i2o.

    The others are not addressed as far as I know.

    A fix for the last problem is to blacklist one of the drivers in
    /etc/modprobe.d/blacklist (or equivalent, could be another file), to
    make sure the same driver is used during installation and later boot.
    Perhaps something like this in an appropriate place (must be before
    the kernel modules are loaded):

    # Force the use of dpt_i2o for devices where both it and i2o_core can
    # support the card, to avoid device name confusion.
    if lspci -n | grep -q " 1044:a511 " ; then
    cat >>/etc/modprobe.d/blacklist < # Make sure dpt_i2o is used for PCI device 1044:a511 (Adaptec 2100 RAID)
    blacklist i2o_core
    blacklist i2o_block
    EOF
    fi

    It would select the dpt_i2o driver for the card supported by both
    drivers. One could also select the i2o_core driver and blacklist the
    dpt_i2o driver. I have no idea which of these drivers are preferred,
    but must admit I personally do not like the /dev/i2o/ device path.

    Happy hacking,
    --
    Petter Reinholdtsen



    --
    To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST@lists.debian.org
    with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

  2. Bug#440161: wrong /etc/fstab when installing on RAID using Adaptec 2100S

    [Petter Reinholdtsen]
    > A quick look in /lib/modules/2.6.18-5-686/modules.pcimap tell me that
    > the conflicting kernel module issue only affect one PCI id. These are
    > the devices supported by the two modules:
    >
    > 1044:a511 dpt_i2o i2o_core
    > 1044:a501 dpt_i2o
    > 8086:1962 i2o_core


    This proved to be wrong.

    PCI id 1044:a501 is supported by i2o_core too. I'm not sure why as
    the entries in /lib/modules/2.6.18-5-686/modules.pcimap told me
    otherwise, but on my test machine with that device the i2o_core kernel
    module would load and I would end up with the /dev/i2o/hda device.

    Happy hacking,
    --
    Petter Reinholdtsen



    --
    To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST@lists.debian.org
    with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

  3. Bug#440161: wrong /etc/fstab when installing on RAID using Adaptec 2100S

    Petter Reinholdtsen writes:

    > [Petter Reinholdtsen]
    >> A quick look in /lib/modules/2.6.18-5-686/modules.pcimap tell me that
    >> the conflicting kernel module issue only affect one PCI id. These are
    >> the devices supported by the two modules:
    >>
    >> 1044:a511 dpt_i2o i2o_core
    >> 1044:a501 dpt_i2o
    >> 8086:1962 i2o_core

    >
    > This proved to be wrong.
    >
    > PCI id 1044:a501 is supported by i2o_core too. I'm not sure why as
    > the entries in /lib/modules/2.6.18-5-686/modules.pcimap told me
    > otherwise, but on my test machine with that device the i2o_core kernel
    > module would load and I would end up with the /dev/i2o/hda device.


    Isn't it 1011:a511?

    Look:

    otavio@neumann:~/hacking/d-i$ grep '0x00001044 0x0000a511' /lib/modules/2.6.22-2-686/modules.pcimap
    dpt_i2o 0x00001044 0x0000a511 0xffffffff 0xffffffff 0x00000000 0x00000000 0x0
    i2o_core 0x00001044 0x0000a511 0xffffffff 0xffffffff 0x00000000 0x00000000 0x0

    and

    otavio@neumann:~/hacking/d-i$ grep '0x00001044 0x0000a501' /lib/modules/2.6.22-2-686/modules.pcimap
    dpt_i2o 0x00001044 0x0000a501 0xffffffff 0xffffffff 0x00000000 0x00000000 0x0

    --
    O T A V I O S A L V A D O R
    ---------------------------------------------
    E-mail: otavio@debian.org UIN: 5906116
    GNU/Linux User: 239058 GPG ID: 49A5F855
    Home Page: http://otavio.ossystems.com.br
    ---------------------------------------------
    "Microsoft sells you Windows ... Linux gives
    you the whole house."



    --
    To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST@lists.debian.org
    with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

  4. Bug#440161: wrong /etc/fstab when installing on RAID using Adaptec 2100S

    [Otavio Salvador]
    > Isn't it 1011:a511?


    Nope, it isn't. That surprised me too, as I looked in the
    modules.pcimap file. The machine I test on do not have the 1011:a511,
    but the 1011:a501 device.

    Happy hacking,
    --
    Petter Reinholdtsen



    --
    To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST@lists.debian.org
    with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

  5. Bug#440161: wrong /etc/fstab when installing on RAID using Adaptec 2100S

    Petter Reinholdtsen writes:

    > [Otavio Salvador]
    >> Isn't it 1011:a511?

    >
    > Nope, it isn't. That surprised me too, as I looked in the
    > modules.pcimap file. The machine I test on do not have the 1011:a511,
    > but the 1011:a501 device.


    hum ... no idea _why_ it's being loaded then.

    It was tested using 2.6.22 right?

    --
    O T A V I O S A L V A D O R
    ---------------------------------------------
    E-mail: otavio@debian.org UIN: 5906116
    GNU/Linux User: 239058 GPG ID: 49A5F855
    Home Page: http://otavio.ossystems.com.br
    ---------------------------------------------
    "Microsoft sells you Windows ... Linux gives
    you the whole house."



    --
    To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST@lists.debian.org
    with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

  6. Bug#440161: wrong /etc/fstab when installing on RAID using Adaptec 2100S

    [Otavio Salvador]
    > It was tested using 2.6.22 right?


    I tested using the 2.6.18 kernel in etch.

    Happy hacking,
    --
    Petter Reinholdtsen



    --
    To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST@lists.debian.org
    with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

+ Reply to Thread