On Thu, Apr 16, 2009 at 02:55:15PM +0530, pradeep singh wrote: > On Thu, Apr 16, 2009 at 9:56 AM, Jeffrey Cao <jcao.linux@xxxxxxxxx> wrote: > > On 2009-04-16, Jeffrey Cao <jcao.linux@xxxxxxxxx> wrote: > >> On 2009-04-15, Peter Teoh <htmldeveloper@xxxxxxxxx> wrote: > >> > >> > >>> "No process context" actually means that the taskstruct's > >>> mm_struct->mm is NULL. This means that the pagetable CR3 are not > >> That's not correct. mm is NULL does not mean no porcess context. > >> Since all kernel processes/threads share the same upper 1G line address space, > >> they do not have their own individual line address like user sapce proesses. > >> All user space processes have their individual page table maps, but kernel > >> processes share one page table maps. So mm is NULL for kernel processes. > > I did some kernel source code investigation to prove my thought. > > kernel_thread() calles do_fork() with the parameter CLONE_VM. that means > > the kernel process will share the memory space of its parent. And this way > > all the kernel threads share one memory space. > > Which is memory space of kernel itself, right? Yes, CLONE_VM colones the kernel address space. All the kernel threads share this address space. > And off course this means a kernel thread can be cloned from an existing kernel > thread only. Sure, you can't fork a kernel thread from user space. Jeffrey -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
