On Tue, Feb 10, 2009 at 8:57 PM, Greg Freemyer <[email protected]> wrote: [snip] >> >> kernel does not handle each of them. There is a user space scheduler >> which takes care of them. >> Kernel is oblivious to usespace threads. >> > > I'm not an expert on linux threading, but I think the above is misleading. > > There are a number of threading models supported in linux. Most are > 5+ years old and are pure userspace implementations. > > Native POSIX threading came to the 2.6.x kernel a while ago. > > http://en.wikipedia.org/wiki/Native_POSIX_Thread_Library > > IF this is the threading model under discussion, then there is kernel > involvement.
Yes Greg, there is kernel involvement while creation of threads. As an example, if you try getpid on all the threads of a process you 'll get same pid. Effectively only one kernel task_struct for the process and constituting threads. ( This was basis of my assumption) :-/ I stand corrected as I spoke of M:N model and I think I am wrong here :). But here is a general doubt. I create a large amount of user space threads(via a rogue application for brevity) . does it means that equivalent task_structs will be created for all of them? Won't i use up the kernel heap this way too fast? Thanks > > Greg > -- > Greg Freemyer > Litigation Triage Solutions Specialist > http://www.linkedin.com/in/gregfreemyer > First 99 Days Litigation White Paper - > http://www.norcrossgroup.com/forms/whitepapers/99%20Days%20whitepaper.pdf > > The Norcross Group > The Intersection of Evidence & Technology > http://www.norcrossgroup.com > -- Pradeep -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to [email protected] Please read the FAQ at http://kernelnewbies.org/FAQ
