Hi there guys,
I need some help in regards with thread programming on Linux. At the moment I am
working on a scheduler which uses a number of kernel threads running on a quad
processor system. Now I am using the clone() call to create these kernel threads
since I require that each thread shares the same working space (if there is a
different call please tell me about it !!!). My problem is that I need to discern
between each kernel thread. Is there a way how to know which thread is which, yet
still not calling a kernel function to obtain the pid (I cannot afford the long wait
to enter the kernel since the pid would be needed quite often, also it would entail a
semaphore block to obtain it). I was told that in the SunOS system, kernel threads
which share the same memory (LWPs) have a section for private memory. I am under the
impression that this does not exist on the Linux system. I hope I am wrong since it
would be an easy task for each thread to store an identifier of itself in here.
I need each kernel thread to know itself (says its number) since I need it to store
some information about its current state which would be different from other clone'd
threads. I know that the real true personal resources a thread has is the processor it
is working on, yet for several reasons I would prefer not to use one of the
processor's registers to store an identifier. As a last resort I would consider it
though.
Is there maybe some one out there who can hint me on some programming trick which
would allow me for each thread to have its personal data. I know that if after the
clone each thread makes a malloc() call, the pointer returned would be different, yet
how can I store it if they are using the same working space.
Hope someone can help
Thanks a lot
Joseph Cordina
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/dmentre/smp-howto/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
