Kai, On Thursday 15 March 2007 09:03, Kai Ponte wrote: > On Wednesday 14 March 2007 09:29:34 pm Randall R Schulz wrote: > > On Wednesday 14 March 2007 20:56, Kai Ponte wrote: > > > ... > > > > > > ...and don't forget that WinNT (XP/Vista/2003) is not preemptive > > > multitasking either. At least not at the kernel level like Linux > > > 2.6+ is. > > > > You really ought to check your facts. You'll give software managers > > a bad name... > > Huh? > > Kernel 2.6+ has preemptive multitasking at the kernel level - a.k.a. > a preemptable kernel. > > WinNT does not.
I don't know where you got this idea, but it's flatly false. If you have access to the IEEE digital library, you can retrieve and read this article: <http://ieeexplore.ieee.org/iel4/2/15590/00722284.pdf?arnumber=722284>. You will see that process and thread control is located in kernel space and is interrupt-driven. To wit: "· The process and thread manager creates and terminates processes and threads. The underlying support for processes and threads is implemented in the Windows NT kernel; the executive adds additional semantics and functions to these lower-level objects. " ... "Kernel The kernel performs the most fundamental operations in Windows NT, determining how the OS uses the processor or processors and ensuring that they are used prudently. It is the lowest layer in NTOSKRNL.EXE. These are the primary functions the kernel provides · Thread scheduling and dispatching. · Trap handling and exception dispatching. · Interrupt handling and dispatching. · Multiprocessor synchronization." > It does do user-level preemptive multitasking but not kernel level. Perhaps you're drawing some real distinction here, but I'm not sure what it is. > Never had it, supposedly will sometime soon. Of course, since the > linux and windows kernels are fairly similar in design I'm sure > Windows will borrow whatever was learned from Linux's implementation > and then sue Novell for patent violations. > > > (And you thought i was a pointy-haired guy who reboots his laptop by > holding it upside down and shaking.) > > > > I was just trying to kill some processes on a Win2003 system > > > today and had to wait for the kernel to finish some tasks before > > > it would die. > > > > That can happen on Linux, too. Try to kill a process in a 'D' wait > > state. It's not possible. Extended duration of a D wait does imply > > a bug (usually in a disk or file system driver), but it happens. > > Ooh, that's just nasty. And if the distinction you suggest betwen user-level and kernel multitasking is what I think it is, this shows that either both kernels have this form of multitasking or both do not. > -- > kai Randall Schulz -- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
