On Tue, Aug 16, 2005 at 02:04:11PM +0900, Tetsuo Handa wrote: > There was a thread about > how to get the number of CPUs available > (started on Sun, 26 Jun 2005 09:14:11 -0500). > And this time, I got a silly question. > > Can the SMP system continue working > when one of the CPUs crashes? > I wonder how the kernel can continue working > if a CPU crashes while holding a lock. > What will happen if a CPU crashes after spin_lock(&dcache_lock) > and before spin_unlock(&dcache_lock)? > I'm afraid that other CPUs attempt to spin_lock(&dcache_lock) and > fall into infinite wait loop. > > I don't need detailed explanations. > I just want to know whether SMP Linux system is fault tolerant > (tolerant against CPU crashes). > Just "yes" or "no" or some URLs is enough. No, but you can make it fault tolerant with a failover system, see the linux high availability project. Another think to look at is CPU hotplugging, which is supported on some architectures (PowerPC, IIRC). Erik -- Erik Mouw J.A.K.Mouw@xxxxxxxxxxxxxx mouw@xxxxxxxxxxxx
Attachment:
signature.asc
Description: Digital signature
