Re: Child Process Monitor for intercepting File IO

From: Andrew Bonello (andrew_bonello_at_hotmail.com)
Date: 06/01/03

  • Next message: Leonidas Petrauskas: "message queue init"
    Date: 31 May 2003 22:53:31 -0700
    
    

    "Kurtis D. Rader" <krader@skepticism.us> wrote in message news:<pan.2003.05.31.22.47.20.363349@skepticism.us>...

    > You'll need to tell us the OS you're working with as there is no widely
    > implemented mechanism for achieving your goal. On many implementations
    > what you're trying to do is going to be impossible.
    >
    > Are the programs you're trying to control are dynamically linked and
    > does your OS support the LD_PRELOAD environment variable? If so you
    > might be able to use that to override functions in the standard shared
    > libraries (e.g., replace the standard write(2) function with your own
    > version). See if your system has a man page named ld.so or something
    > similar that documents the dynamic linker/loader for your system.
    >
    > If that doesn't pan out you could try using the debugging facilities of
    > your system. On many implementations this is the ptrace(2) API. However,
    > that is probably going to have undesirable performance implications.
    > Futhermore, it will be extremely difficult to implement such that all
    > boundary conditions are handled correctly.
    >
    > If your OS supports the concept of nested file systems or file system
    > translation layers arguably the best solution is to write a kernel file
    > system to implement the desired behavior.

    The utilities which the wrapper needs to provide the translation layer
    for are running principally under Linux (Redhat 7.2+). I also have
    some utilities running under Irix4.?, but these are less of a
    priority; its the Linux tools I am mainly interested in providing for.

    Do you happen to know if any of your suggested courses of action above
    apply in the case of Linux?

    Andy


  • Next message: Leonidas Petrauskas: "message queue init"

    Relevant Pages

    • Re: How Common Lisp sucks
      ... Except that Linux has a process for managing change, ... SBCL, CLISP, Allegro, etc. are implementations of Common Lisp. ... The POSIX standard is analogous to the hyperspec. ... the Posix community has processes for developing and evolving ...
      (comp.lang.lisp)
    • Re: Starting a New OS called iOS
      ... > codes will be shamelessly copied from a GNU GPL licenced OS like linux ... The architecture of file systems is due for an upgrade. ... The mounting/unmounting of removable devices in *nix is a major hurdle ... shells available on *nix, which should be good enough for you. ...
      (comp.os.linux)
    • Re: Linux - why?
      ... as "Linux Starter Kit", with a DVD of SUSE Linux 10.1 from a City ... microsoft, you generally have an update for your GNU/Linux with in two days. ... the general file systems for Linux are a lot better than those provided ... Windows is increasing. ...
      (alt.linux)
    • Re: How Common Lisp sucks
      ... Except that Linux has a process for managing change, ... SBCL, CLISP, Allegro, etc. are implementations of Common Lisp. ... is what's important for understanding my analogy. ... The POSIX standard is analogous to the hyperspec. ...
      (comp.lang.lisp)
    • Re: Linux Market share?
      ... :>are quaking in their boots over linux, ... why not be forthcoming about access parameters on NTFS ... With over 150 other file systems out there that ARE disclosing their ... any fear that Bill Gates might have either. ...
      (alt.os.linux.suse)