Trap E Help - OS2

This is a discussion on Trap E Help - OS2 ; V2.0 OS/2 486-50 MHz PS2 IBM Model 8595. I've written a few new REXX programs for this system and now I get a trap E every few days. I am using two functions from REXXLIB (DosRename and DosCopy) that I ...

+ Reply to Thread
Results 1 to 6 of 6

Thread: Trap E Help

  1. Trap E Help

    V2.0 OS/2 486-50 MHz PS2 IBM Model 8595.

    I've written a few new REXX programs for this system and now I get a trap E
    every few days. I am using two functions from REXXLIB (DosRename and
    DosCopy) that I don't think was used before.

    I have no idea which program is causing the Trap, how can I determine which
    program to look at? I am completely in the dark as to how to continue. I
    have the trap values, but since I do not know where these programs were
    executing I don't know where to look.

    Is there a writeup somewhere that would help me step through the analysis of
    the values in the Trap?

    Thanks, George Barrowcliff




  2. Re: Trap E Help

    On Tue, 23 Nov 2004 15:14:43 GMT, George W. Barrowcliff
    wrote:
    > I have no idea which program is causing the Trap, how can I determine which
    > program to look at? I am completely in the dark as to how to continue. I
    > have the trap values, but since I do not know where these programs were
    > executing I don't know where to look.


    Your best bet is probably to post the trap screen contents in
    comp.os.os2.bugs and ask for assistance there.

    If you're not deeply familiar with OS/2's internals, trap screen analysis
    can still be done to a limited extent, but it takes experience.

    I'm not too familiar with v2.0 so I can't say much more with certainty.
    More recent kernels in the 4.5x line include the name of the trapping
    module in the trap screen, for earlier versions you have to do some
    digging.

    Generally, the first thing people tell you to look at is the CSLIM value.
    If it's FFFFFFFF then you've trapped in 32-bit code. This isn't too
    helpful, except that AFAIK there isn't really much 32-bit code that runs
    in Ring 0, so if you have any 'usual suspects' you might be able to make
    an educated guess. If not, you'd need more technical expertise than I can
    offer.

    If CSLIM is not FFFFFFFF, OTOH, then it's a 16-bit module that
    trapped, and you can actually use the value to identify the module in
    question. I believe you need a utility called EXEINFO (available from
    Hobbes) for that purpose, and also know how to use it. Others can help
    you there, I'm sure.

    --
    Alex Taylor
    http://www.cs-club.org/~alex

  3. Re: Trap E Help


    > I've written a few new REXX programs for this system and now I
    > get a trap E every few days.


    I don't know the details of it, but could it be related to the
    timer with an 'overflow' after about 4 days?



    ---

  4. Re: Trap E Help

    In <41a35be4$0$33284$45beb828@newscene.com>, on 11/23/2004
    at 09:51 AM, Alex Taylor said:

    >Your best bet is probably to post the trap screen contents in
    >comp.os.os2.bugs and ask for assistance there.


    Don't forget to include the kernel version.

    Also, it's not entirely clear from the OP's description if this is a
    kernel trap or an application trap.

    >I'm not too familiar with v2.0 so I can't say much more with certainty.


    Anything that old is not going to do a module decode.

    >Generally, the first thing people tell you to look at is the CSLIM value.
    >If it's FFFFFFFF then you've trapped in 32-bit code. This isn't too
    >helpful, except that AFAIK there isn't really much 32-bit code that runs
    >in Ring 0,


    Other that the kernel that is. :-)

    >If CSLIM is not FFFFFFFF, OTOH, then it's a 16-bit module that trapped,


    Right. Either a 16-bit driver or 16-bit code in the kernel.

    >I
    >believe you need a utility called EXEINFO (available from Hobbes) for
    >that purpose, and also know how to use it.


    That will work in most cases.

    Steven


    --
    --------------------------------------------------------------------------------------------
    Steven Levine MR2/ICE 2.60b #10183
    Warp4/FP15/14.093c_W4 www.scoug.com irc.fyrelizard.com #scoug (Wed 7pm PST)
    --------------------------------------------------------------------------------------------


  5. Re: Trap E Help

    On 23 Nov 2004 09:51:18 -0600, Alex Taylor wrote:

    > Generally, the first thing people tell you to look at is the CSLIM value.
    > If it's FFFFFFFF then you've trapped in 32-bit code. This isn't too
    > helpful, except that AFAIK there isn't really much 32-bit code that runs
    > in Ring 0


    Really? What about all that nice kernel code that Scott keeps tweaking?
    That's quite a lot of ring 0 code in my book.

    > If CSLIM is not FFFFFFFF, OTOH, then it's a 16-bit module that
    > trapped, and you can actually use the value to identify the module in
    > question.


    You "might be" able to identify the module in question.

  6. Re: Trap E Help

    On Tue, 23 Nov 2004 23:02:57 GMT, Steven Levine
    wrote:
    >>Your best bet is probably to post the trap screen contents in
    >>comp.os.os2.bugs and ask for assistance there.

    >
    > Don't forget to include the kernel version.


    Well, I figure that should be included in the trap screen...


    >>Generally, the first thing people tell you to look at is the CSLIM value.
    >>If it's FFFFFFFF then you've trapped in 32-bit code. This isn't too
    >>helpful, except that AFAIK there isn't really much 32-bit code that runs
    >>in Ring 0,

    >
    > Other that the kernel that is. :-)


    Doh! I should've said '32-bit _drivers_'.

    --
    Alex Taylor
    http://www.cs-club.org/~alex

+ Reply to Thread