On 2/21/13 4:27 AM, Warner Losh wrote: > Author: imp > Date: Thu Feb 21 00:27:51 2013 > New Revision: 247066 > URL: http://svnweb.freebsd.org/changeset/base/247066 > > Log: > Replace splhigh() with critical_enter()/leave() to ensure we write the > config mode unlock sequence quickly enough. This likely isn't too critical, > since splhigh() has been a noop for a decade... > > Modified: > head/sys/dev/ppc/ppc.c > > Modified: head/sys/dev/ppc/ppc.c > ============================================================================== > --- head/sys/dev/ppc/ppc.c Thu Feb 21 00:26:31 2013 (r247065) > +++ head/sys/dev/ppc/ppc.c Thu Feb 21 00:27:51 2013 (r247066) > @@ -74,6 +74,22 @@ static void ppcintr(void *arg); > > #define DEVTOSOFTC(dev) ((struct ppc_data *)device_get_softc(dev)) > > +/* > + * We use critical enter/leave for the simple config locking needed to > + * detect the devices. We just want to make sure that both of our writes > + * happen without someone else also writing to those config registers. Since > + * we just do this at startup, Giant keeps multiple threads from executing, > + * and critical_enter() then is all that's needed to keep us from being > preempted > + * during the critical sequences with the hardware. > + * > + * Note: this doesn't prevent multiple threads from putting the chips into > + * config mode, but since we only do that to detect the type at startup the > + * extra overhead isn't needed since Giant protects us from multiple entry > + * and no other code changes these registers. > + */ > +#define PPC_CONFIG_LOCK(ppc) critical_enter() > +#define PPC_CONFIG_UNLOCK(ppc) critical_leave() > +
s/critical_leave/critical_exit/? -- Andrey Zonov
signature.asc
Description: OpenPGP digital signature
