On 10:12 Thu 20 Nov , ruby wrote: > >> >> No, the "kernel" part of the address space is not accessable in user space. >> It is just the physical memory mapped into the address space as a big chunk. >> The purpose is that user-space/kernel space switches (e.g. due to system >> calls >> do not require a rather expensive (TLB flush) address space switch. >> -Michi >> > so does it mean in theory at least, you _don't have to_ map the kernel > code into the 4GB process address space? It is just that when you do > want kernel to run something on your behalf (through syscall), it is > more expensive to do the mapping? or is it just plain not possible?
There was a patch which did that: http://lwn.net/Articles/39283/ If I remember correctly, it has not been merged, because it was thought that it is unmaintainable (and because 64 bit is available) -Michi -- programing a layer 3+4 network protocol for mesh networks see http://michaelblizek.twilightparadox.com -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to [EMAIL PROTECTED] Please read the FAQ at http://kernelnewbies.org/FAQ
