On Thu, Feb 06, 2003 at 10:50:28AM -0800, Julian Elischer wrote: > > The only way it makes sense to talk about suspending > a process in user space from the kernel, would be in a MP system. in a > UP system, if you are in the kernel, then there is no process in user > space. (it trapped into the kernel whenever whatever interrupt that > stasrted you running happenned).
In other words: on UP system it is possible to remove a process from the run queue and then this process can be considered as "suspended", because current syscall can't be preempted, and PCB of suspended process is valid (that is the kernel saved it). Is this correct? > In SMP you may need to set some flag > and send an IP to all teh processes to ensure that the process enters > the kernel to see the flag. In this case, as I understand, I need to modify some parts of the kernel. What is IP ? > Depending on what you wnat to do, it is > possible that ptrace (as DES says) is what you want, or maybe just a > SIGSTOP will do. > I will think about ptrace(2) and about SIGSTOP again (the problem with SIGSTOP I described in previous letter: signal will be delivered to a target process asynchronously with respect to the current process and I need to stay in a loop in a syscall and check when a SIGSTOP signal will be delivered to a process). I would like somebody to comment other questions in my previous latter, probably this will help me. Thanks Julian and DES for your help! To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
