On Mar 24, 2014, at 9:28 PM, Mateusz Guzik <m...@freebsd.org> wrote:
> Author: mjg > Date: Tue Mar 25 03:28:58 2014 > New Revision: 263704 > URL: http://svnweb.freebsd.org/changeset/base/263704 > > Log: > Make /dev/devctl mpsafe. > > MFC after: 1 week > > Modified: > head/sys/kern/subr_bus.c > > Modified: head/sys/kern/subr_bus.c > ============================================================================== > --- head/sys/kern/subr_bus.c Tue Mar 25 03:25:30 2014 (r263703) > +++ head/sys/kern/subr_bus.c Tue Mar 25 03:28:58 2014 (r263704) > @@ -358,15 +358,16 @@ device_sysctl_fini(device_t dev) > /* Deprecated way to adjust queue length */ > static int sysctl_devctl_disable(SYSCTL_HANDLER_ARGS); > /* XXX Need to support old-style tunable hw.bus.devctl_disable" */ > -SYSCTL_PROC(_hw_bus, OID_AUTO, devctl_disable, CTLTYPE_INT | CTLFLAG_RW, > NULL, > - 0, sysctl_devctl_disable, "I", "devctl disable -- deprecated"); > +SYSCTL_PROC(_hw_bus, OID_AUTO, devctl_disable, CTLTYPE_INT | CTLFLAG_RW | > + CTLFLAG_MPSAFE, NULL, 0, sysctl_devctl_disable, "I", > + "devctl disable -- deprecated”); This can likely be deleted now... > #define DEVCTL_DEFAULT_QUEUE_LEN 1000 > static int sysctl_devctl_queue(SYSCTL_HANDLER_ARGS); > static int devctl_queue_length = DEVCTL_DEFAULT_QUEUE_LEN; > TUNABLE_INT("hw.bus.devctl_queue", &devctl_queue_length); > -SYSCTL_PROC(_hw_bus, OID_AUTO, devctl_queue, CTLTYPE_INT | CTLFLAG_RW, NULL, > - 0, sysctl_devctl_queue, "I", "devctl queue length"); > +SYSCTL_PROC(_hw_bus, OID_AUTO, devctl_queue, CTLTYPE_INT | CTLFLAG_RW | > + CTLFLAG_MPSAFE, NULL, 0, sysctl_devctl_queue, "I", "devctl queue > length"); > > static d_open_t devopen; > static d_close_t devclose; > @@ -376,7 +377,6 @@ static d_poll_t devpoll; > > static struct cdevsw dev_cdevsw = { > .d_version = D_VERSION, > - .d_flags = D_NEEDGIANT, > .d_open = devopen, > .d_close = devclose, > .d_read = devread, > @@ -420,23 +420,31 @@ devinit(void) > static int > devopen(struct cdev *dev, int oflags, int devtype, struct thread *td) > { > + > if (devsoftc.inuse) > return (EBUSY); Why not delete these two lines? Since this isn’t a performance critical part of the code, it is clearer and safer to just test inuse inside the locked section. > + mtx_lock(&devsoftc.mtx); > + if (devsoftc.inuse) { > + mtx_unlock(&devsoftc.mtx); > + return (EBUSY); > + } > /* move to init */ > devsoftc.inuse = 1; > devsoftc.nonblock = 0; > devsoftc.async_proc = NULL; > + mtx_unlock(&devsoftc.mtx); > return (0); > } > > static int > devclose(struct cdev *dev, int fflag, int devtype, struct thread *td) > { > - devsoftc.inuse = 0; > + > mtx_lock(&devsoftc.mtx); > + devsoftc.inuse = 0; > + devsoftc.async_proc = NULL; > cv_broadcast(&devsoftc.cv); > mtx_unlock(&devsoftc.mtx); > - devsoftc.async_proc = NULL; > return (0); > } _______________________________________________ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
