Re: Fork bombing a Linux machine as a non-root user

From: Scot L. Harris (webid_at_cfl.rr.com)
Date: 03/18/05

  • Next message: bruce: "RE: booting an intel 815 fc3 without monitor" 
    To: For users of Fedora Core releases <fedora-list@redhat.com>
Date: Fri, 18 Mar 2005 14:24:56 -0500

    On Fri, 2005-03-18 at 13:23, Matthew Miller wrote:
    > On Fri, Mar 18, 2005 at 01:08:31PM -0500, M.Rudra wrote:
    > > As a non-geek user I am concerned as I thought Linux is very secure.
    > > So I googled "fork bombing" and after reading them, I have some doubts :
    >
    > This isn't really a general security issue. If a user on your system does
    > this, find them and (metaphorically) smack them around.
    >
    > If you want to prevent this from happening accidentally, or if you have a
    > critical server with users you don't trust (a situation to be avoided!) you
    > can _reduce_ the chances of something like this happening by configuring
    > resource limits, but since all computers do have finite resources, it's hard
    > to have a generally-useful shared machine without some possibility of
    > someone going out of their way to use more than their share.

    This is the wrong approach. A user level account should not be allowed
    to consume resources to the point that the whole system crashes.

    Per the article, this is an old trick that was fixed in many different
    versions of Unix. There should be limits on the resources a user
    process can request, or at least throttling so that once a certain
    threshold is reached the user process does not get anymore until it
    releases some.

    Similar limit is placed on a user allocating disk space. The system
    reserves about 5% of the disk which only root can use. This prevents
    someone from eating all of /tmp or some other file system and preventing
    root from being able to execute programs that require some disk
    resources, possibly while trying to fix the problem the user process
    created.

    While this is not an exploit that results in root level access it is a
    denial of service that could be used by anyone that achieves user level
    access to a system. And users should never be trusted. They will find
    some way to mess things up. :)

    Definitely something that should be fixed. 

    -- 
Scot L. Harris
webid@cfl.rr.com
Those who can, do; those who can't, write.
Those who can't write work for the Bell Labs Record. 
-- 
fedora-list mailing list
fedora-list@redhat.com
To unsubscribe: http://www.redhat.com/mailman/listinfo/fedora-list
  • Next message: bruce: "RE: booting an intel 815 fc3 without monitor"

    Relevant Pages

    • RE: Concurrently streaming a file to HttpResponse and file IO
      ... about the same time to read a resource from the disk as it does to select it ... I'm implementing support for disk based caching of binary resources ... (continue to serve requests for resources) ... an approach is to start streaming the resource from ...
      (microsoft.public.dotnet.languages.csharp)
    • Re: Cluster Freezes
      ... When testing the offline times I notice that the file shares go offline in a ... the Physical disk seem to take atleast 30-45 seconds to ... The failover time for all the resources is really good maybe 20 seconds tops ... the real problem is coming back online. ...
      (microsoft.public.windows.server.clustering)
    • Re: Concurrently streaming a file to HttpResponse and file IO
      ... I would like to keep the application responsive (continue to serve requests for resources) while streaming resources to disk. ... In order to serve each request, an approach is to start streaming the resource from DB to the client request - and simultaneously queue a task to the threadpool that streams the resource to disk. ...
      (microsoft.public.dotnet.languages.csharp)
    • Concurrently streaming a file to HttpResponse and file IO
      ... I would like to keep the application responsive (continue to serve requests for resources) while streaming resources to disk. ... In order to serve each request, an approach is to start streaming the resource from DB to the client request - and simultaneously queue a task to the threadpool that streams the resource to disk. ...
      (microsoft.public.dotnet.languages.csharp)
    • Re: MSDTC Disk Problem
      ... As i told you it fails the DTC resources,instead i can move ip and network ... error that the resources can't open the log file that reside on the disk ... >> rejoin the crashed node in the cluster ...
      (microsoft.public.windows.server.clustering)