On 17.12.2013, at 15:26, Thomas Huth <th...@linux.vnet.ibm.com> wrote:
> > Hi Alex, > > On Tue, 17 Dec 2013 14:56:33 +0100 > Alexander Graf <ag...@suse.de> wrote: > >> >> On 17.12.2013, at 14:22, Jens Freimann <jf...@linux.vnet.ibm.com> wrote: >> >>> From: Thomas Huth <th...@linux.vnet.ibm.com> >>> >>> This patch adds the missing START order to the SIGP instruction handler. >> >> Does the spec define what happens on START when the CPU is already running? > > As far as I can see, the spec does not explicitely defines the behavior > in that case, it says only: "The order is effective only when the > addressed CPU is in the stopped state". But according to my experiments > (I wrote a test program that I ran without additional hypervisor), the > START order is simply ignored when the CPU is already running and CC0 is > returned. > > This is also what happens with the code below, since > s390_add_running_cpu() only does something if the "halted" flag is set, > and kicking a CPU that is already running does not hurt either, as far > as I can see. Yup :) > >> Does START modify any register state? > > No, the CPU simply continues running with the state where it has been > stopped before. Ok, great. Then it really does the same thing :) Alex
