| Message ID | 1459870344-16773-10-git-send-email-alex.bennee@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 05/04/16 18:32, Alex Bennée wrote: > +static void kick_tcg_thread(void *opaque) > +{ > + QEMUTimer *self = *(QEMUTimer **) opaque; > + timer_mod(self, > + qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + > + NANOSECONDS_PER_SECOND / 10); > + qemu_cpu_kick_no_halt(); > +} > It would be nice to have some definition (e.g. macro) of TCG thread kick period. (snip) > @@ -1179,6 +1198,14 @@ static void *qemu_tcg_cpu_thread_fn(void *arg) > } > } > > + /* Set to kick if we have to do more than one vCPU */ > + if (CPU_NEXT(first_cpu)) { > + kick_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, kick_tcg_thread, &kick_timer); > + timer_mod(kick_timer, > + qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + > + NANOSECONDS_PER_SECOND / 10); > + } > + I think cpu_ticks_init() could be more natural place to put this initialization in. Kind regards, Sergey
Sergey Fedorov <serge.fdrv@gmail.com> writes: > On 05/04/16 18:32, Alex Bennée wrote: >> +static void kick_tcg_thread(void *opaque) >> +{ >> + QEMUTimer *self = *(QEMUTimer **) opaque; >> + timer_mod(self, >> + qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + >> + NANOSECONDS_PER_SECOND / 10); >> + qemu_cpu_kick_no_halt(); >> +} >> > > It would be nice to have some definition (e.g. macro) of TCG thread kick > period. Will do. > > (snip) > >> @@ -1179,6 +1198,14 @@ static void *qemu_tcg_cpu_thread_fn(void *arg) >> } >> } >> >> + /* Set to kick if we have to do more than one vCPU */ >> + if (CPU_NEXT(first_cpu)) { >> + kick_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, kick_tcg_thread, &kick_timer); >> + timer_mod(kick_timer, >> + qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + >> + NANOSECONDS_PER_SECOND / 10); >> + } >> + > > I think cpu_ticks_init() could be more natural place to put this > initialization in. It would be but I need somewhere to keep kick_timer and doing it inside the thread function keeps it nice and local. > > Kind regards, > Sergey -- Alex Bennée
On 02/06/16 19:00, Alex Bennée wrote: > Sergey Fedorov <serge.fdrv@gmail.com> writes: >> On 05/04/16 18:32, Alex Bennée wrote: >>> @@ -1179,6 +1198,14 @@ static void *qemu_tcg_cpu_thread_fn(void *arg) >>> } >>> } >>> >>> + /* Set to kick if we have to do more than one vCPU */ >>> + if (CPU_NEXT(first_cpu)) { >>> + kick_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, kick_tcg_thread, &kick_timer); >>> + timer_mod(kick_timer, >>> + qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + >>> + NANOSECONDS_PER_SECOND / 10); >>> + } >>> + >> I think cpu_ticks_init() could be more natural place to put this >> initialization in. > It would be but I need somewhere to keep kick_timer and doing it inside > the thread function keeps it nice and local. Fair enough. By the way, this kick timer is only required for round-robin single-threaded CPU loop, right? Kind regards, Sergey
diff --git a/cpus.c b/cpus.c index 8d27fb0..e22bb77 100644 --- a/cpus.c +++ b/cpus.c @@ -1151,11 +1151,30 @@ static void *qemu_dummy_cpu_thread_fn(void *arg) #endif } +/* Single-threaded TCG + * + * In the single-threaded case each vCPU is simulated in turn. If + * there is more than a single vCPU we create a simple timer to kick + * the vCPU and ensure we don't get stuck in a tight loop in one vCPU. + * This is done explicitly rather than relying on side-effects + * elsewhere. + */ static int tcg_cpu_exec(CPUState *cpu); +static void qemu_cpu_kick_no_halt(void); + +static void kick_tcg_thread(void *opaque) +{ + QEMUTimer *self = *(QEMUTimer **) opaque; + timer_mod(self, + qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + + NANOSECONDS_PER_SECOND / 10); + qemu_cpu_kick_no_halt(); +} static void *qemu_tcg_cpu_thread_fn(void *arg) { CPUState *cpu = arg; + QEMUTimer *kick_timer; rcu_register_thread(); @@ -1179,6 +1198,14 @@ static void *qemu_tcg_cpu_thread_fn(void *arg) } } + /* Set to kick if we have to do more than one vCPU */ + if (CPU_NEXT(first_cpu)) { + kick_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, kick_tcg_thread, &kick_timer); + timer_mod(kick_timer, + qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + + NANOSECONDS_PER_SECOND / 10); + } + /* process any pending work */ atomic_mb_set(&exit_request, 1);
Currently we rely on the side effect of the main loop grabbing the iothread_mutex to give any long running basic block chains a kick to ensure the next vCPU is scheduled. As this code is being re-factored and rationalised we now do it explicitly here. Signed-off-by: Alex Bennée <alex.bennee@linaro.org> --- v2 - re-base fixes - get_ticks_per_sec() -> NANOSECONDS_PER_SEC --- cpus.c | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+)