Eric Saxe wrote: > Hi John, > > John Harres wrote: > >> One further question... >> >> In regards to using %stick instead of a global lbolt variable, is there >> any guarantee that the %stick registers will be adequately synchronized >> across the system, such that reads of this register on two different >> CPUs (perhaps after a migration) would be similar enough not to cause >> any problems? >> >> > Yes. It is my understanding that %stick is driven by a clock signal > external to any processor, and it should > be synchronized across the system. > Well, depends on the system. Generally it requires software support to *ensure* that the %stick registers stay in sync (e.g. after a core has its clocks stopped it must be resynced when the clocks start up again).
Rather than a guarantee, there is a requirement, so software makes it so (else bad things happen when time appears to move backwards, etc). -sarito > Thanks, > -Eric > _______________________________________________ > tesla-dev mailing list > tesla-dev at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/tesla-dev >
