Who killed my keys? - Xwindows

This is a discussion on Who killed my keys? - Xwindows ; I received a support call 2 weeks ago where an NC900 at a customer site was reported to have lost some of its keymappings. Certain keys such as Page UP, Page Down, keypad ENTER, and the Sun L-keys (we use ...

+ Reply to Thread
Results 1 to 5 of 5

Thread: Who killed my keys?

  1. Who killed my keys?

    I received a support call 2 weeks ago where an NC900 at a customer site
    was reported to have lost some of its keymappings. Certain keys such as
    Page UP, Page Down, keypad ENTER, and the Sun L-keys (we use a Sun IV kbd)
    were all dead or non-functional. This unit had been in use for a few hours
    before it was reported. The keys are essential to the use of our custom
    apps so an outage lasting all morning would not have gone unreported.
    The assumption is that the keymap, once established during login, was fine
    for hours then took a hit.

    I asked the user to log out and back in. Problem solved. But what caused
    it? I don't know. If anyone reading this has had similar experiences,
    please post a reply.

    Normally X-Terminal keymaps don't suddenly become corrupt.

    In our system, we run XDM on the UNIX host. When the customer logs on,
    XDM kicks off our wsapp script which runs a set of xmodmap lines then
    launches our custom X app suite. Once the keymap is set up, we
    make no edits to it for the life of the login. So for the above problem
    to occur, I have to consider maybe a hardware failure? The problem was
    blown away by asking the customer to log off: XDM resets the NC's X
    display server and thereby reinitialized the keymap.

    There is the possibility of the NC being modified by an xmodmap run
    from another location on the production LAN, but that is highly
    unlikely. That LAN is not connected to the public Internet nor the
    development LAN.

    Any help would be appreciated.

    Jim Olson
    Software Engineer
    PHLX - The Philadelphia Stock Exchange
    www.phlx.com
    Email: e1-0j2z-sn0w-jcbd@emailias.com
    (Temporary address - use this to spam me and I'll expire it,
    rendering your list obsolete!)

  2. Re: Who killed my keys?

    James Olson wrote:
    >I received a support call 2 weeks ago where an NC900 at a customer site
    >was reported to have lost some of its keymappings. Certain keys such as
    >Page UP, Page Down, keypad ENTER, and the Sun L-keys (we use a Sun IV kbd)
    >were all dead or non-functional. This unit had been in use for a few hours
    >before it was reported. The keys are essential to the use of our custom
    >apps so an outage lasting all morning would not have gone unreported.
    >The assumption is that the keymap, once established during login, was fine
    >for hours then took a hit.
    >
    >I asked the user to log out and back in. Problem solved. But what caused
    >it? I don't know. If anyone reading this has had similar experiences,
    >please post a reply.
    >
    >Normally X-Terminal keymaps don't suddenly become corrupt.
    >
    >In our system, we run XDM on the UNIX host. When the customer logs on,
    >XDM kicks off our wsapp script which runs a set of xmodmap lines then
    >launches our custom X app suite. Once the keymap is set up, we
    >make no edits to it for the life of the login. So for the above problem
    >to occur, I have to consider maybe a hardware failure? The problem was
    >blown away by asking the customer to log off: XDM resets the NC's X
    >display server and thereby reinitialized the keymap.
    >
    >There is the possibility of the NC being modified by an xmodmap run
    >from another location on the production LAN, but that is highly
    >unlikely. That LAN is not connected to the public Internet nor the
    >development LAN.
    >
    >Any help would be appreciated.


    I haven't got a clue about what would be changing the keymap.

    I have tracked this sort of problem in the past by running all clients
    in the `untrusted' mode; changing the keymap is one of many things
    that an untrusted client cannot do.

    You can make all your auth keys untrusted with a script similar to
    the following:

    #!/bin/sh

    for i in `xauth list|cut -f1 -d' '`; do
    echo generate $i MIT-MAGIC-COOKIE-1 timeout 0 | xauth
    done


  3. Re: Who killed my keys?

    In comp.windows.x, Kip Rugger

    wrote
    on 9 Jan 2004 13:00:00 -0600
    :
    > James Olson wrote:
    >>I received a support call 2 weeks ago where an NC900 at a customer site
    >>was reported to have lost some of its keymappings. Certain keys such as
    >>Page UP, Page Down, keypad ENTER, and the Sun L-keys (we use a Sun IV kbd)
    >>were all dead or non-functional. This unit had been in use for a few hours
    >>before it was reported. The keys are essential to the use of our custom
    >>apps so an outage lasting all morning would not have gone unreported.
    >>The assumption is that the keymap, once established during login, was fine
    >>for hours then took a hit.
    >>
    >>I asked the user to log out and back in. Problem solved. But what caused
    >>it? I don't know. If anyone reading this has had similar experiences,
    >>please post a reply.
    >>
    >>Normally X-Terminal keymaps don't suddenly become corrupt.
    >>
    >>In our system, we run XDM on the UNIX host. When the customer logs on,
    >>XDM kicks off our wsapp script which runs a set of xmodmap lines then
    >>launches our custom X app suite. Once the keymap is set up, we
    >>make no edits to it for the life of the login. So for the above problem
    >>to occur, I have to consider maybe a hardware failure? The problem was
    >>blown away by asking the customer to log off: XDM resets the NC's X
    >>display server and thereby reinitialized the keymap.
    >>
    >>There is the possibility of the NC being modified by an xmodmap run
    >>from another location on the production LAN, but that is highly
    >>unlikely. That LAN is not connected to the public Internet nor the
    >>development LAN.
    >>
    >>Any help would be appreciated.

    >
    > I haven't got a clue about what would be changing the keymap.
    >
    > I have tracked this sort of problem in the past by running all clients
    > in the `untrusted' mode; changing the keymap is one of many things
    > that an untrusted client cannot do.
    >
    > You can make all your auth keys untrusted with a script similar to
    > the following:
    >
    > #!/bin/sh
    >
    > for i in `xauth list|cut -f1 -d' '`; do
    > echo generate $i MIT-MAGIC-COOKIE-1 timeout 0 | xauth
    > done
    >


    Pedant point: should that be

    for i in `xauth list|cut -f1 -d' '`; do
    echo generate $i MIT-MAGIC-COOKIE-1 timeout 0
    done | xauth

    instead? :-) It's mostly a matter of running xauth once with a
    bunch of 'generate' commands as opposed to running it multiple
    times with a single 'generate' each.

    I for one don't see how a hardware failure would cause a corrupted
    keymap unless the NC has bad memory, in which case there should be
    other symptoms.

    *scratches head*

    --
    #191, ewill3@earthlink.net
    It's still legal to go .sigless.

  4. Re: Who killed my keys?

    The Ghost In The Machine wrote:
    >In comp.windows.x, Kip Rugger
    >
    > wrote
    >> #!/bin/sh
    >>
    >> for i in `xauth list|cut -f1 -d' '`; do
    >> echo generate $i MIT-MAGIC-COOKIE-1 timeout 0 | xauth
    >> done
    >>

    >
    >Pedant point: should that be
    >
    >for i in `xauth list|cut -f1 -d' '`; do
    > echo generate $i MIT-MAGIC-COOKIE-1 timeout 0
    >done | xauth
    >
    >instead? :-) It's mostly a matter of running xauth once with a
    >bunch of 'generate' commands as opposed to running it multiple
    >times with a single 'generate' each.


    Absolutely -- since you are going to loose the ability to do xauth,
    my broken script runs the risk of not being able to finish the list.


  5. Re: Who killed my keys?

    e1-0j2z-sn0w-jcbd@emailias.com (James Olson) wrote in message news:<4a15e416.0401081155.56cd70e6@posting.google.com>...
    > I received a support call 2 weeks ago where an NC900 at a customer site
    > was reported to have lost some of its keymappings. Certain keys such as
    > Page UP, Page Down, keypad ENTER, and the Sun L-keys (we use a Sun IV kbd)
    > were all dead or non-functional. This unit had been in use for a few hours
    > before it was reported. The keys are essential to the use of our custom
    > apps so an outage lasting all morning would not have gone unreported.
    > The assumption is that the keymap, once established during login, was fine
    > for hours then took a hit.
    >
    > I asked the user to log out and back in. Problem solved. But what caused
    > it? I don't know. If anyone reading this has had similar experiences,
    > please post a reply.
    >
    > Normally X-Terminal keymaps don't suddenly become corrupt.
    >
    >
    >
    > Jim Olson
    > Software Engineer
    > PHLX - The Philadelphia Stock Exchange
    > www.phlx.com
    > Email: e1-0j2z-sn0w-jcbd@emailias.com
    > (Temporary address - use this to spam me and I'll expire it,
    > rendering your list obsolete!)


    Thanks guys for the xauth script. I go back on production coverage
    next week (Jan 20-23, 2004) and I'll be staying alert to that
    customer site. Without a repeat I have to write it off as a single incident.
    While I have seen customers do things behind our backs, given that the
    restrictions on the NC (our apps are 100% custom), customers can't run apps
    if we don't put them on the root menu. This leads me to suspect a bug
    somewhere in our code. If there is, it should happen again. I'll post
    another message if anything happens next week or if we get lucky. I'll
    also ask my coworkers who had coverage during the past 5 weeks (we have
    a 6-man rotation).

    Jim Olson, PHLX

+ Reply to Thread