PTimer fix and ARM MPTimer conversion

Multiple issues here related to the timer with a corrected .limit value:

1) ptimer_get_count() returns incorrect value for the disabled timer after loading the counter with a small value, because corrected limit value is used instead of the original.

For instance:
1) ptimer_stop(t)
2) ptimer_set_period(t, 1)
3) ptimer_set_limit(t, 0, 1)
4) ptimer_get_count(t) <-- would return 10000 instead of 0

2) ptimer_get_count() might return incorrect value for the timer running with a corrected limit value.

For instance:
1) ptimer_stop(t)
2) ptimer_set_period(t, 1)
3) ptimer_set_limit(t, 10, 1)
4) ptimer_run(t)
5) ptimer_get_count(t) <-- might return value > 10

3) Neither ptimer_set_period() nor ptimer_set_freq() are correcting the limit value, so it is still possible to make timer timeout value arbitrary small.

For instance:
1) ptimer_set_period(t, 10000)
2) ptimer_set_limit(t, 1, 0)
3) ptimer_set_period(t, 1) <-- bypass limit correction

Fix all of the above issues by moving timeout value correction to the ptimer_reload(). Instead of changing limit value, set new ptimer struct member "next_event_corrected" when next_event was corrected and make ptimer_get_count() always return 1 for the timer running with corrected next_event value. Bump VMSD version since ptimer VM state is changed, but keep .minimum_version_id as it is backward compatible.

Signed-off-by Dmitry Osipenko <[email protected]>

---
 hw/core/ptimer.c | 37 +++++++++++++++++++++++--------------
 1 file changed, 23 insertions(+), 14 deletions(-)

hw/core/ptimer.c +18 -0

index 8437bd6..2441943 100644
@@ -19,6 +19,7 @@ struct ptimer_state
     int64_t period;
     int64_t last_event;
     int64_t next_event;
+    uint8_t next_event_corrected;
     QEMUBH *bh;
     QEMUTimer *timer;
 };

@@ -48,6 +49,23 @@ static void ptimer_reload(ptimer_state *s)
     if (s->period_frac) {
         s->next_event += ((int64_t)s->period_frac * s->delta) >> 32;
     }
+
+    /*
+     * Artificially limit timeout to something
+     * achievable under QEMU.  Otherwise, QEMU spends all
+     * its time generating timer interrupts, and there
+     * is no forward progress.
+     * About ten microseconds is the fastest that really works
+     * on the current generation of host machines.
+     */
+
+    s->next_event_corrected = 0;
+
+    if (!use_icount && (s->next_event - s->last_event < 10000)) {
+        s->next_event = s->last_event + 10000;
+        s->next_event_corrected = 1;
+    }
+
     timer_mod(s->timer, s->next_event);
 }
 

hw/core/ptimer.c +2 -15

         } else {
             uint64_t rem;
             uint64_t div;

@@ -180,19 +201,6 @@ void ptimer_set_freq(ptimer_state *s, uint32_t freq)
    count = limit.  */
 void ptimer_set_limit(ptimer_state *s, uint64_t limit, int reload)
 {
-    /*
-     * Artificially limit timeout rate to something
-     * achievable under QEMU.  Otherwise, QEMU spends all
-     * its time generating timer interrupts, and there
-     * is no forward progress.
-     * About ten microseconds is the fastest that really works
-     * on the current generation of host machines.
-     */
-
-    if (!use_icount && limit * s->period < 10000 && s->period) {
-        limit = 10000 / s->period;
-    }
-
     s->limit = limit;
     if (reload)
         s->delta = limit;

@@ -204,7 +212,7 @@ void ptimer_set_limit(ptimer_state *s, uint64_t limit, int reload)
 
 const VMStateDescription vmstate_ptimer = {
     .name = "ptimer",
-    .version_id = 1,
+    .version_id = 2,
     .minimum_version_id = 1,
     .fields = (VMStateField[]) {
         VMSTATE_UINT8(enabled, ptimer_state),

@@ -214,6 +222,7 @@ const VMStateDescription vmstate_ptimer = {
         VMSTATE_INT64(period, ptimer_state),
         VMSTATE_INT64(last_event, ptimer_state),
         VMSTATE_INT64(next_event, ptimer_state),
+        VMSTATE_UINT8(next_event_corrected, ptimer_state),
         VMSTATE_TIMER_PTR(timer, ptimer_state),
         VMSTATE_END_OF_LIST()
     }
--

2.6.1

This looks like a give-up without trying to get the correct value. If the calculated value (using the normal-path logic below) is sane, you should just use it. If it comes out bad then you should clamp to 1.

I am wondering whether this clamping policy (as in original code as well) is correct at all though. The value of a free-running short-interval periodic timer (poor mans random number generator) without any actual interrupt generation will be affected by QEMUs asynchronous handling of timer events. So if I set limit to 100, then sample this timer every user keyboard stroke, I should get a uniform distribution on [0,100]. Instead in am going to get lots of 1s. This

is more broken in the original code (as you state), as I will get > 100, but I think we have traded broken for slightly less broken. I think the correct semantic is to completely ignoring rate limitin except for the scheduling on event callbacks. That is, the timer

interval is not rate limited, instead the timer_mod interval (next_event -last_event) just has a 10us clamp.

The means the original code semantic of returning counter = 0 for an already triggered timer is wrong. It should handle in-the-past wrap-arounds as wraparounds by triggering the timer and redoing the math with the new interval values. So instead the logic would be something like:

timer_val = -1;

for(;;) {

  if (!enabled) {
     return delta;
  }

timer_val = (next-event - now) * scaling();

  if (timer_val >=  0) {
      return timer_val;

}
/* Timer has actually expired but we missed it, reload it and try again */ ptimer_tick();
}

ptimer_reload() then needs to be patched to make sure it always timer_mod()s in the future, otherwise this loop could iterate a large number of times.

This means that when the qemu_timer() actually ticks, a large number or cycles may have occured, but we can justify that in that callback event latency (usually interrupts) is undefined anyway and coalescing of multiples may have happened as part of that. This usually matches expectations of real guests where interrupt latency is ultimately undefined.

Regards,
Peter

24.10.2015 22:45, Peter Crosthwaite пишет:

Quoted 6 lines

This looks like a give-up without trying to get the correct value. If the calculated value (using the normal-path logic below) is sane, you should just use it. If it comes out bad then you should clamp to 1.

Quoted 7 lines

I am wondering whether this clamping policy (as in original code as well) is correct at all though. The value of a free-running short-interval periodic timer (poor mans random number generator) without any actual interrupt generation will be affected by QEMUs asynchronous handling of timer events. So if I set limit to 100, then sample this timer every user keyboard stroke, I should get a uniform distribution on [0,100]. Instead in am going to get lots of 1s. This

Right, that's a good example. What about to scale ptimer period to match adjusted timer_mod interval?

Quoted 4 lines

is more broken in the original code (as you state), as I will get > 100, but I think we have traded broken for slightly less broken. I think the correct semantic is to completely ignoring rate limitin except for the scheduling on event callbacks. That is, the timer

I'm missing you here. What event callbacks?

Quoted 17 lines

interval is not rate limited, instead the timer_mod interval (next_event -last_event) just has a 10us clamp.

The means the original code semantic of returning counter = 0 for an already triggered timer is wrong. It should handle in-the-past wrap-arounds as wraparounds by triggering the timer and redoing the math with the new interval values. So instead the logic would be something like:

timer_val = -1;

for(;;) {

  if (!enabled) {
     return delta;
  }

Quoted 8 lines

timer_val = (next-event - now) * scaling();

  if (timer_val >=  0) {
      return timer_val;

}
/* Timer has actually expired but we missed it, reload it and try again
*/
ptimer_tick();
}

Why do you think that ptimer_get_count() == 0 in case of the running periodic timer that was expired while QEMU was "on the way" to ptimer code is bad and wrong?

From the guest point of view it's okay (no?), do we really need to overengineer that corner case?

Quoted 11 lines

ptimer_reload() then needs to be patched to make sure it always timer_mod()s in the future, otherwise this loop could iterate a large number of times.

This means that when the qemu_timer() actually ticks, a large number or cycles may have occured, but we can justify that in that callback event latency (usually interrupts) is undefined anyway and coalescing of multiples may have happened as part of that. This usually matches expectations of real guests where interrupt latency is ultimately undefined.

ptimer_tick() is re-arm'ing the qemu_timer() in case of periodic mode. Hope I haven't missed your point here.

Quoted 4 lines

Regards,
Peter

--
Dmitry

Quoted 16 lines

24.10.2015 22:45, Peter Crosthwaite пишет:

Quoted 6 lines

This looks like a give-up without trying to get the correct value. If the calculated value (using the normal-path logic below) is sane, you should just use it. If it comes out bad then you should clamp to 1.

Quoted 7 lines

I am wondering whether this clamping policy (as in original code as well) is correct at all though. The value of a free-running short-interval periodic timer (poor mans random number generator) without any actual interrupt generation will be affected by QEMUs asynchronous handling of timer events. So if I set limit to 100, then sample this timer every user keyboard stroke, I should get a uniform distribution on [0,100]. Instead in am going to get lots of 1s. This

Quoted 3 lines

Right, that's a good example. What about to scale ptimer period to match adjusted timer_mod interval?

Thats just as incorrect as changing the limit IMO. The guest could get confused with a timer running at the wrong frequency.

Quoted 8 lines

Quoted 4 lines

is more broken in the original code (as you state), as I will get > 100, but I think we have traded broken for slightly less broken. I think the correct semantic is to completely ignoring rate limitin except for the scheduling on event callbacks. That is, the timer

I'm missing you here. What event callbacks?

when timer_mod() hits, and it turn triggers some device specific event (usually an interrupt).

There are two basic interactions for any QEMU timer. There are synchronous events, i.e. the guest reading (polling) the counter which is what this patch tries to fix. The second is the common case of periodic interrupt generation. My proposal is that rate limiting does not affect synchronous operation, only asynchronous (so my keystroke RNG case works). In the current code, if ptimer_get_count() is called when the event has passed it returns 0 under the assumption that the timer_mod callback is about to happen. With rate-limiting that may be well in the future.

Quoted 27 lines

Quoted 17 lines

interval is not rate limited, instead the timer_mod interval (next_event -last_event) just has a 10us clamp.

The means the original code semantic of returning counter = 0 for an already triggered timer is wrong. It should handle in-the-past wrap-arounds as wraparounds by triggering the timer and redoing the math with the new interval values. So instead the logic would be something like:

timer_val = -1;

for(;;) {

   if (!enabled) {
      return delta;
   }

Quoted 8 lines

   timer_val = (next-event - now) *  scaling();
   if (timer_val >=  0) {
       return timer_val;
   }
   /* Timer has actually expired but we missed it, reload it and try again

*/
ptimer_tick();
}

Quoted 3 lines

Why do you think that ptimer_get_count() == 0 in case of the running periodic timer that was expired while QEMU was "on the way" to ptimer code is bad and wrong?

Because you may have gone past the time when it was actually zero and reloaded and started counting again. It should return the real (reloaded and resumed) value. This is made worse by rate limiting as the timer will spend a long time at the clamp value waiting for the rate-limited tick to fix it.

Following on from before, we don't want the async stuff to affect sync. As the async callbacks are heavily affected by rate limiting we don't want the polled timer having to rely on the callbacks (async path) at all for correct operation. That's what the current implementation of ptimer_get_count assumes with that 0-clamp.

Quoted 3 lines

From the guest point of view it's okay (no?), do we really
need to overengineer that corner case?

Depends on your use case. Your bug report is correct in that the timer should never be outside the bounds of the limit. But you are fixing that very specifically with a minimal change rather than correcting the larger ptimer_get_count() logic which I think is more broken than you suggest it is.

Quoted 13 lines

Quoted 11 lines

ptimer_reload() then needs to be patched to make sure it always timer_mod()s in the future, otherwise this loop could iterate a large number of times.

This means that when the qemu_timer() actually ticks, a large number or cycles may have occured, but we can justify that in that callback event latency (usually interrupts) is undefined anyway and coalescing of multiples may have happened as part of that. This usually matches expectations of real guests where interrupt latency is ultimately undefined.

Quoted 3 lines

ptimer_tick() is re-arm'ing the qemu_timer() in case of periodic mode. Hope I haven't missed your point here.

Yes. But it is also capable of doing the heavy lifting for our already expired case. Basic idea is, if the timer is in a bad state (should have hit) but hasn't, do the hit to put the timer into a good state (by calling ptimer_tick) then just do the right thing. That's what the loop does. It should also work for an in-the-past one-shot.

25.10.2015 02:55, Peter Crosthwaite пишет:

Quoted 40 lines · Dmitry Osipenko

Quoted 19 lines

24.10.2015 22:45, Peter Crosthwaite пишет:

Quoted 7 lines

This looks like a give-up without trying to get the correct value. If the calculated value (using the normal-path logic below) is sane, you should just use it. If it comes out bad then you should clamp to 1.

Quoted 7 lines

I am wondering whether this clamping policy (as in original code as well) is correct at all though. The value of a free-running short-interval periodic timer (poor mans random number generator) without any actual interrupt generation will be affected by QEMUs asynchronous handling of timer events. So if I set limit to 100, then sample this timer every user keyboard stroke, I should get a uniform distribution on [0,100]. Instead in am going to get lots of 1s. This

Quoted 3 lines

Right, that's a good example. What about to scale ptimer period to match adjusted timer_mod interval?

Thats just as incorrect as changing the limit IMO. The guest could get confused with a timer running at the wrong frequency.

Quoted 9 lines

Quoted 4 lines

is more broken in the original code (as you state), as I will get > 100, but I think we have traded broken for slightly less broken. I think the correct semantic is to completely ignoring rate limitin except for the scheduling on event callbacks. That is, the timer

I'm missing you here. What event callbacks?

when timer_mod() hits, and it turn triggers some device specific event (usually an interrupt).

Quoted 10 lines

There are two basic interactions for any QEMU timer. There are synchronous events, i.e. the guest reading (polling) the counter which is what this patch tries to fix. The second is the common case of periodic interrupt generation. My proposal is that rate limiting does not affect synchronous operation, only asynchronous (so my keystroke RNG case works). In the current code, if ptimer_get_count() is called when the event has passed it returns 0 under the assumption that the timer_mod callback is about to happen. With rate-limiting that may be well in the future.

ptimer_tick() would happen on the next QEMU loop cycle, so it might be more reasonable to return counter = 1 here, wouldn't it?

Quoted 41 lines

Quoted 29 lines

Quoted 22 lines

interval is not rate limited, instead the timer_mod interval (next_event -last_event) just has a 10us clamp.

The means the original code semantic of returning counter = 0 for an already triggered timer is wrong. It should handle in-the-past wrap-arounds as wraparounds by triggering the timer and redoing the math with the new interval values. So instead the logic would be something like:

timer_val = -1;

for(;;) {

   if (!enabled) {
      return delta;
   }

   timer_val = (next-event - now) *  scaling();
   if (timer_val >=  0) {
       return timer_val;
   }
   /* Timer has actually expired but we missed it, reload it and try

Quoted 4 lines

again
*/
ptimer_tick();
}

Quoted 4 lines

Why do you think that ptimer_get_count() == 0 in case of the running periodic timer that was expired while QEMU was "on the way" to ptimer code
is bad and wrong?

Because you may have gone past the time when it was actually zero and reloaded and started counting again. It should return the real (reloaded and resumed) value. This is made worse by rate limiting as the timer will spend a long time at the clamp value waiting for the rate-limited tick to fix it.

Quoted 6 lines

Following on from before, we don't want the async stuff to affect sync. As the async callbacks are heavily affected by rate limiting we don't want the polled timer having to rely on the callbacks (async path) at all for correct operation. That's what the current implementation of ptimer_get_count assumes with that 0-clamp.

Alright, that make sense now. Thanks for clarifying.

Quoted 29 lines

Quoted 3 lines

From the guest point of view it's okay (no?), do we really
need to overengineer that corner case?

Depends on your use case. Your bug report is correct in that the timer should never be outside the bounds of the limit. But you are fixing that very specifically with a minimal change rather than correcting the larger ptimer_get_count() logic which I think is more broken than you suggest it is.

Quoted 14 lines

Quoted 11 lines

ptimer_reload() then needs to be patched to make sure it always timer_mod()s in the future, otherwise this loop could iterate a large number of times.

This means that when the qemu_timer() actually ticks, a large number or cycles may have occured, but we can justify that in that callback event latency (usually interrupts) is undefined anyway and coalescing of multiples may have happened as part of that. This usually matches expectations of real guests where interrupt latency is ultimately undefined.

Quoted 4 lines

ptimer_tick() is re-arm'ing the qemu_timer() in case of periodic mode. Hope
I haven't missed your point here.

Quoted 6 lines

Yes. But it is also capable of doing the heavy lifting for our already expired case. Basic idea is, if the timer is in a bad state (should have hit) but hasn't, do the hit to put the timer into a good state (by calling ptimer_tick) then just do the right thing. That's what the loop does. It should also work for an in-the-past one-shot.

Summarizing what we have now:

There are two issues with ptimer:

1) ptimer_get_count() return wrong values with adjusted .limit

Patch V7 doesn't solve that issue, but makes it slightly better by clamping returned value to [0, 1]. That's not what we want, we need to return counter value in it's valid range [0, limit].

You are rejecting variant of scaling ptimer period, saying that it might affect software behavior inside the guest. But by adjusting the timer, we might already causing same misbehavior in case of blazing fast host machine.

I'll scratch my head a bit more on it. If you have any better idea, please share.

2) ptimer_get_count() return fake 0 value in case of the expired qemu_timer() without triggering ptimer_tick()

You're suggesting to solve it by running ptimer_tick(). So if emulated device uses ptimer tick event (scheduled qemu bh) to raise interrupt, it would do it by one QEMU loop cycle earlier.

My question here: is it always legal for the guest software to be able to get counter = 0 on poll while CPU interrupt on timer expire hasn't happened yet (would happen after one QEMU cycle).

I guess it might cause software
misbehavior if it assumes that the real hardware has CPU and timer running in the same clock domain, i.e. such situation might be not possible.

So I'm
suggesting to return counter = 1 and allow ptimer_tick() happen on it's own.

--
Dmitry

Quoted 149 lines

25.10.2015 02:55, Peter Crosthwaite пишет:

Quoted 40 lines · Dmitry Osipenko

Quoted 19 lines

24.10.2015 22:45, Peter Crosthwaite пишет:

Quoted 7 lines

This looks like a give-up without trying to get the correct value. If the calculated value (using the normal-path logic below) is sane, you should just use it. If it comes out bad then you should clamp to 1.

Quoted 7 lines

I am wondering whether this clamping policy (as in original code as well) is correct at all though. The value of a free-running short-interval periodic timer (poor mans random number generator) without any actual interrupt generation will be affected by QEMUs asynchronous handling of timer events. So if I set limit to 100, then sample this timer every user keyboard stroke, I should get a uniform distribution on [0,100]. Instead in am going to get lots of 1s. This

Quoted 3 lines

Right, that's a good example. What about to scale ptimer period to match adjusted timer_mod interval?

Thats just as incorrect as changing the limit IMO. The guest could get confused with a timer running at the wrong frequency.

Quoted 9 lines

Quoted 4 lines

is more broken in the original code (as you state), as I will get > 100, but I think we have traded broken for slightly less broken. I think the correct semantic is to completely ignoring rate limitin except for the scheduling on event callbacks. That is, the timer

I'm missing you here. What event callbacks?

when timer_mod() hits, and it turn triggers some device specific event (usually an interrupt).

Quoted 10 lines

There are two basic interactions for any QEMU timer. There are synchronous events, i.e. the guest reading (polling) the counter which is what this patch tries to fix. The second is the common case of periodic interrupt generation. My proposal is that rate limiting does not affect synchronous operation, only asynchronous (so my keystroke RNG case works). In the current code, if ptimer_get_count() is called when the event has passed it returns 0 under the assumption that the timer_mod callback is about to happen. With rate-limiting that may be well in the future.

ptimer_tick() would happen on the next QEMU loop cycle, so it might be more reasonable to return counter = 1 here, wouldn't it?

Quoted 41 lines

Quoted 29 lines

Quoted 13 lines

interval is not rate limited, instead the timer_mod interval (next_event -last_event) just has a 10us clamp.

The means the original code semantic of returning counter = 0 for an already triggered timer is wrong. It should handle in-the-past wrap-arounds as wraparounds by triggering the timer and redoing the math with the new interval values. So instead the logic would be something like:

timer_val = -1;

for(;;) {

Quoted 13 lines

    if (!enabled) {
       return delta;
    }

    timer_val = (next-event - now) *  scaling();
    if (timer_val >=  0) {
        return timer_val;
    }
    /* Timer has actually expired but we missed it, reload it and try
again
*/
    ptimer_tick();
}

Quoted 4 lines

Why do you think that ptimer_get_count() == 0 in case of the running periodic timer that was expired while QEMU was "on the way" to ptimer code
is bad and wrong?

Because you may have gone past the time when it was actually zero and reloaded and started counting again. It should return the real (reloaded and resumed) value. This is made worse by rate limiting as the timer will spend a long time at the clamp value waiting for the rate-limited tick to fix it.

Quoted 6 lines

Following on from before, we don't want the async stuff to affect sync. As the async callbacks are heavily affected by rate limiting we don't want the polled timer having to rely on the callbacks (async path) at all for correct operation. That's what the current implementation of ptimer_get_count assumes with that 0-clamp.

Alright, that make sense now. Thanks for clarifying.

Quoted 29 lines

Quoted 3 lines

From the guest point of view it's okay (no?), do we really
need to overengineer that corner case?

Depends on your use case. Your bug report is correct in that the timer should never be outside the bounds of the limit. But you are fixing that very specifically with a minimal change rather than correcting the larger ptimer_get_count() logic which I think is more broken than you suggest it is.

Quoted 14 lines

Quoted 11 lines

ptimer_reload() then needs to be patched to make sure it always timer_mod()s in the future, otherwise this loop could iterate a large number of times.

This means that when the qemu_timer() actually ticks, a large number or cycles may have occured, but we can justify that in that callback event latency (usually interrupts) is undefined anyway and coalescing of multiples may have happened as part of that. This usually matches expectations of real guests where interrupt latency is ultimately undefined.

Quoted 4 lines

ptimer_tick() is re-arm'ing the qemu_timer() in case of periodic mode. Hope
I haven't missed your point here.

Quoted 6 lines

Yes. But it is also capable of doing the heavy lifting for our already expired case. Basic idea is, if the timer is in a bad state (should have hit) but hasn't, do the hit to put the timer into a good state (by calling ptimer_tick) then just do the right thing. That's what the loop does. It should also work for an in-the-past one-shot.

Summarizing what we have now:

There are two issues with ptimer:

1) ptimer_get_count() return wrong values with adjusted .limit

Quoted 7 lines

Patch V7 doesn't solve that issue, but makes it slightly better by clamping returned value to [0, 1]. That's not what we want, we need to return counter value in it's valid range [0, limit].

You are rejecting variant of scaling ptimer period, saying that it might affect software behavior inside the guest. But by adjusting the timer, we might already causing same misbehavior in case of blazing fast host machine.

It is a different misbehaviour. We are modelling the polled timer perfectly but limiting the frequency of callbacks (interrupts). I think this is the lesser of two evils.

Quoted 10 lines

I'll scratch my head a bit more on it. If you have any better idea, please share.

2) ptimer_get_count() return fake 0 value in case of the expired qemu_timer() without triggering ptimer_tick()

You're suggesting to solve it by running ptimer_tick(). So if emulated device uses ptimer tick event (scheduled qemu bh) to raise interrupt, it would do it by one QEMU loop cycle earlier.

Yes, this is ok, as even in a rate limited scenario there is no reason to absolutely force the rate limit. If a poll happens it should just flush the waiting interrupt.

Quoted 3 lines

My question here: is it always legal for the guest software to be able to get counter = 0 on poll while CPU interrupt on timer expire hasn't happened yet (would happen after one QEMU cycle).

Yes. And I am going a step further by saying it is ok for the guest software to see the timer value wrapped around before the expire too.

Quoted 3 lines

I guess it might cause software
misbehavior if it assumes that the real hardware has CPU and timer running in the same clock domain, i.e. such situation might be not possible.

Assumptions about the CPU clocking only make sense in icount mode, where the rate limiter is disable anyway.

Quoted 3 lines

So I'm
suggesting to return counter = 1 and allow ptimer_tick() happen on it's own.

My alternate suggestion is, if you detect that the tick should have already happened, just make it happen. I don't see the need to rate limit a polled timer.

25.10.2015 20:39, Peter Crosthwaite пишет:

Quoted 191 lines · Dmitry Osipenko

Quoted 166 lines

25.10.2015 02:55, Peter Crosthwaite пишет:

On Sat, Oct 24, 2015 at 3:22 PM, Dmitry Osipenko <[email protected]>

Quoted 45 lines

Quoted 21 lines · wrote:

24.10.2015 22:45, Peter Crosthwaite пишет:

Quoted 7 lines

This looks like a give-up without trying to get the correct value. If the calculated value (using the normal-path logic below) is sane, you should just use it. If it comes out bad then you should clamp to 1.

Quoted 8 lines

I am wondering whether this clamping policy (as in original code as well) is correct at all though. The value of a free-running short-interval periodic timer (poor mans random number generator) without any actual interrupt generation will be affected by QEMUs asynchronous handling of timer events. So if I set limit to 100, then sample this timer every user keyboard stroke, I should get a uniform distribution on [0,100]. Instead in am going to get lots of 1s. This

Quoted 5 lines

Right, that's a good example. What about to scale ptimer period to match
adjusted timer_mod interval?

Thats just as incorrect as changing the limit IMO. The guest could get confused with a timer running at the wrong frequency.

is more broken in the original code (as you state), as I will get >

Quoted 10 lines

Quoted 4 lines

100, but I think we have traded broken for slightly less broken. I think the correct semantic is to completely ignoring rate limitin except for the scheduling on event callbacks. That is, the timer

I'm missing you here. What event callbacks?

when timer_mod() hits, and it turn triggers some device specific event (usually an interrupt).

Quoted 11 lines

There are two basic interactions for any QEMU timer. There are synchronous events, i.e. the guest reading (polling) the counter which is what this patch tries to fix. The second is the common case of periodic interrupt generation. My proposal is that rate limiting does not affect synchronous operation, only asynchronous (so my keystroke RNG case works). In the current code, if ptimer_get_count() is called when the event has passed it returns 0 under the assumption that the timer_mod callback is about to happen. With rate-limiting that may be well in the future.

ptimer_tick() would happen on the next QEMU loop cycle, so it might be more
reasonable to return counter = 1 here, wouldn't it?

interval is not rate limited, instead the timer_mod interval

Quoted 42 lines

Quoted 29 lines

Quoted 12 lines

(next_event -last_event) just has a 10us clamp.

The means the original code semantic of returning counter = 0 for an already triggered timer is wrong. It should handle in-the-past wrap-arounds as wraparounds by triggering the timer and redoing the math with the new interval values. So instead the logic would be something like:

timer_val = -1;

for(;;) {

Quoted 14 lines

    if (!enabled) {
       return delta;
    }

    timer_val = (next-event - now) *  scaling();
    if (timer_val >=  0) {
        return timer_val;
    }
    /* Timer has actually expired but we missed it, reload it and try
again
*/
    ptimer_tick();
}

Quoted 5 lines

Why do you think that ptimer_get_count() == 0 in case of the running periodic timer that was expired while QEMU was "on the way" to ptimer code
is bad and wrong?

Because you may have gone past the time when it was actually zero and reloaded and started counting again. It should return the real (reloaded and resumed) value. This is made worse by rate limiting as the timer will spend a long time at the clamp value waiting for the rate-limited tick to fix it.

Quoted 7 lines

Following on from before, we don't want the async stuff to affect sync. As the async callbacks are heavily affected by rate limiting we don't want the polled timer having to rely on the callbacks (async path) at all for correct operation. That's what the current implementation of ptimer_get_count assumes with that 0-clamp.

Alright, that make sense now. Thanks for clarifying.

From the guest point of view it's okay (no?), do we really

Quoted 29 lines

Quoted 3 lines

need to overengineer that corner case?

Depends on your use case. Your bug report is correct in that the timer should never be outside the bounds of the limit. But you are fixing that very specifically with a minimal change rather than correcting the larger ptimer_get_count() logic which I think is more broken than you suggest it is.

Quoted 14 lines

Quoted 11 lines

ptimer_reload() then needs to be patched to make sure it always timer_mod()s in the future, otherwise this loop could iterate a large number of times.

This means that when the qemu_timer() actually ticks, a large number or cycles may have occured, but we can justify that in that callback event latency (usually interrupts) is undefined anyway and coalescing of multiples may have happened as part of that. This usually matches expectations of real guests where interrupt latency is ultimately undefined.

Quoted 5 lines

ptimer_tick() is re-arm'ing the qemu_timer() in case of periodic mode. Hope
I haven't missed your point here.

Quoted 7 lines

Yes. But it is also capable of doing the heavy lifting for our already expired case. Basic idea is, if the timer is in a bad state (should have hit) but hasn't, do the hit to put the timer into a good state (by calling ptimer_tick) then just do the right thing. That's what the loop does. It should also work for an in-the-past one-shot.

Summarizing what we have now:

There are two issues with ptimer:

1) ptimer_get_count() return wrong values with adjusted .limit

Patch V7 doesn't solve that issue, but makes it slightly better by clamping
returned value to [0, 1]. That's not what we want, we need to return counter
value in it's valid range [0, limit].

Quoted 5 lines

You are rejecting variant of scaling ptimer period, saying that it might affect software behavior inside the guest. But by adjusting the timer, we might already causing same misbehavior in case of blazing fast host machine.

It is a different misbehaviour. We are modelling the polled timer perfectly but limiting the frequency of callbacks (interrupts). I think this is the lesser of two evils.

I'll scratch my head a bit more on it. If you have any better idea, please

Quoted 10 lines

share.

2) ptimer_get_count() return fake 0 value in case of the expired qemu_timer() without triggering ptimer_tick()

You're suggesting to solve it by running ptimer_tick(). So if emulated device uses ptimer tick event (scheduled qemu bh) to raise interrupt, it would do it by one QEMU loop cycle earlier.

Yes, this is ok, as even in a rate limited scenario there is no reason to absolutely force the rate limit. If a poll happens it should just flush the waiting interrupt.

Quoted 10 lines

My question here: is it always legal for the guest software to be able to

Quoted 4 lines

get counter = 0 on poll while CPU interrupt on timer expire hasn't happened
yet (would happen after one QEMU cycle).

Yes. And I am going a step further by saying it is ok for the guest software to see the timer value wrapped around before the expire too.

Let's imagine a hardware with a such restriction: timer interrupt has highest priority and CPU immediately switches to the interrupt handler in a such way that it won't ever could see counter = 0 / wraparound (with interrupt enabled) before entering the handler.

Is it unrealistic?

For instance (on QEMU):

CPU | Timer
---------------------------------------------------------------------
start_periodic_timer | timer starts ticking

                                .....
  QEMU starts to execute          |
  translated block                |
                                  |       QEMU timer expires
                                  |
  CPU reads the timer register,   |       ptimer_get_count() return
  ptimer_get_count() called       |       wrapped around value
                                .....
  CPU interrupt handler kicks in  |       timer continue ticking, so
                                  |       any value is valid here
  CPU stops the timer and sets    |
  counter to 0, returns from the  |
  handler                         |
                                .....
  Now, for some reason, software  |
  sees that timer is stopped      |
  and do something using read     |
  value                           |

Program code sketch:

timer_interrupt_handler()

{
        write32(1, TIMER_STOP);
        write32(0, TIMER_COUNTER);
        write32(TIMER_IRQ_CLEAR, TIMER_STATUS);

        return IRQ_HANDLED;
}

program()

{
        .....

        ..... <--- timer expired here
        ..... <--- interrupt handler executed here on real HW

        var1 = read32(TIMER_COUNTER); <--- Emulated got wrapped,
                                           real got 0

        ..... <--- interrupt handler executed here on QEMU

        if (read32(TIMER_STATUS) & TIMER_RUNNING) {
                .....
        } else {
                .....

                write(var1 >> 16, SOME_DEV_REGISTER);
        }

        .....
}

Might emulated program behave differently from the real HW after it? Probably.

I want to mention that not only beefy generic CPU's are the ptimer users.

However, it seems that no one of the current ptimer users has a such restriction since it would already return 0 on expire and ptimer_tick() would happen after it. We can agree on keeping ptimer less universal in favor of
the expire optimization, so somebody may improve it later if it would be needed.

Do we agree?

I guess it might cause software

Quoted 5 lines

Quoted 4 lines

misbehavior if it assumes that the real hardware has CPU and timer running
in the same clock domain, i.e. such situation might be not possible.

Quoted 4 lines

Assumptions about the CPU clocking only make sense in icount mode, where the rate limiter is disable anyway.

Timer limiter has nothing to do with a returned value for the expired timer. Clock cycle accurate execution isn't relevant to upstream QEMU, I meant clocking in general. Emulated behaviour shouldn't diverge from the real HW.

So I'm

Quoted 9 lines

Quoted 4 lines

suggesting to return counter = 1 and allow ptimer_tick() happen on it's own.

My alternate suggestion is, if you detect that the tick should have already happened, just make it happen. I don't see the need to rate limit a polled timer.

Yes, I got your idea and it is absolutely correct if we agree on the above tradeoff (if that tradeoff exists).

--
Dmitry

On Tue, Oct 27, 2015 at 2:26 PM, Dmitry Osipenko <[email protected] <mailto:[email protected]>> wrote:

25.10.2015 20:39, Peter Crosthwaite пишет:

On Sun, Oct 25, 2015 at 6:23 AM, Dmitry Osipenko <[email protected] <mailto:[email protected]>> wrote:

25.10.2015 02:55, Peter Crosthwaite пишет:

On Sat, Oct 24, 2015 at 3:22 PM, Dmitry Osipenko <[email protected] <mailto:[email protected]>> wrote:

24.10.2015 22:45, Peter Crosthwaite пишет:

This looks like a give-up without trying to get the correct value. If
the calculated value (using the normal-path logic below) is sane, you
should just use it. If it comes out bad then you should clamp to 1.

I am wondering whether this clamping policy (as in original code as
well) is correct at all though. The value of a free-running short-interval periodic timer (poor mans random number generator)
without any actual interrupt generation will be affected by QEMUs
asynchronous handling of timer events. So if I set limit to 100, then
sample this timer every user keyboard stroke, I should get a uniform
distribution on [0,100]. Instead in am going to get lots of 1s. This

Right, that's a good example. What about to scale ptimer period to match
adjusted timer_mod interval?

Thats just as incorrect as changing the limit IMO. The guest could get
confused with a timer running at the wrong frequency.

is more broken in the original code (as you state), as I will get >
100, but I think we have traded broken for slightly less broken. I
think the correct semantic is to completely ignoring rate limitin
except for the scheduling on event callbacks. That is, the timer

I'm missing you here. What event callbacks?

when timer_mod() hits, and it turn triggers some device specific event
(usually an interrupt).

There are two basic interactions for any QEMU timer. There are synchronous events, i.e. the guest reading (polling) the counter which
is what this patch tries to fix. The second is the common case of periodic interrupt generation. My proposal is that rate limiting does
not affect synchronous operation, only asynchronous (so my keystroke RNG case works). In the current code, if ptimer_get_count() is called
when the event has passed it returns 0 under the assumption that the timer_mod callback is about to happen. With rate-limiting that may be
well in the future.

ptimer_tick() would happen on the next QEMU loop cycle, so it might be more
reasonable to return counter = 1 here, wouldn't it?

interval is not rate limited, instead the timer_mod interval (next_event -last_event) just has a 10us clamp.

The means the original code semantic of returning counter = 0 for an
already triggered timer is wrong. It should handle in-the-past
wrap-arounds as wraparounds by triggering the timer and redoing the
math with the new interval values. So instead the logic would be
something like:

timer_val = -1;

for(;;) {

if (!enabled) {
return delta;
}

timer_val = (next-event - now) * scaling(); if (timer_val >= 0) {
return timer_val;
}
/* Timer has actually expired but we missed it,
reload it and try
again
*/
ptimer_tick();
}

Why do you think that ptimer_get_count() == 0 in case of the running
periodic timer that was expired while QEMU was "on the way" to ptimer
code
is bad and wrong?

Because you may have gone past the time when it was actually zero and
reloaded and started counting again. It should return the real (reloaded and resumed) value. This is made worse by rate limiting as the timer will spend a long time at the clamp value waiting for the rate-limited tick to fix it.

Following on from before, we don't want the async stuff to affect sync. As the async callbacks are heavily affected by rate limiting we
don't want the polled timer having to rely on the callbacks (async path) at all for correct operation. That's what the current implementation of ptimer_get_count assumes with that 0-clamp.

Alright, that make sense now. Thanks for clarifying.

From the guest point of view it's okay (no?), do we really
need to overengineer that corner case?

Depends on your use case. Your bug report is correct in that the timer
should never be outside the bounds of the limit. But you are fixing that very specifically with a minimal change rather than correcting the larger ptimer_get_count() logic which I think is more broken than
you suggest it is.

ptimer_reload() then needs to be patched to make sure it always
timer_mod()s in the future, otherwise this loop could iterate a large
number of times.

This means that when the qemu_timer() actually ticks, a large number
or cycles may have occured, but we can justify that in that callback
event latency (usually interrupts) is undefined anyway and coalescing
of multiples may have happened as part of that. This usually matches
expectations of real guests where interrupt latency is ultimately
undefined.

ptimer_tick() is re-arm'ing the qemu_timer() in case of periodic mode.
Hope
I haven't missed your point here.

Yes. But it is also capable of doing the heavy lifting for our already
expired case. Basic idea is, if the timer is in a bad state (should have hit) but hasn't, do the hit to put the timer into a good state (by calling ptimer_tick) then just do the right thing. That's what the
loop does. It should also work for an in-the-past one-shot.

Summarizing what we have now:

There are two issues with ptimer:

1) ptimer_get_count() return wrong values with adjusted .limit

Patch V7 doesn't solve that issue, but makes it slightly better by clamping
returned value to [0, 1]. That's not what we want, we need to return counter
value in it's valid range [0, limit].

You are rejecting variant of scaling ptimer period, saying that it might affect software behavior inside the guest. But by adjusting the timer, we
might already causing same misbehavior in case of blazing fast host machine.

It is a different misbehaviour. We are modelling the polled timer perfectly but limiting the frequency of callbacks (interrupts). I think this is the lesser of two evils.

I'll scratch my head a bit more on it. If you have any better idea, please
share.

2) ptimer_get_count() return fake 0 value in case of the expired qemu_timer() without triggering ptimer_tick()

You're suggesting to solve it by running ptimer_tick(). So if emulated device uses ptimer tick event (scheduled qemu bh) to raise interrupt, it would do it by one QEMU loop cycle earlier.

Yes, this is ok, as even in a rate limited scenario there is no reason to absolutely force the rate limit. If a poll happens it should just flush the waiting interrupt.

My question here: is it always legal for the guest software to be able to
get counter = 0 on poll while CPU interrupt on timer expire hasn't happened
yet (would happen after one QEMU cycle).

Yes. And I am going a step further by saying it is ok for the guest software to see the timer value wrapped around before the expire too.

Let's imagine a hardware with a such restriction: timer interrupt has highest priority and CPU immediately switches to the interrupt handler in a such way that it won't ever could see counter = 0 / wraparound (with interrupt enabled) before entering the handler.

Is it unrealistic?

Yes. And if it is possible in real HW, I don't think this is valid for QEMU outside of icount mode.

For instance (on QEMU):

CPU | Timer
---------------------------------------------------------------------
start_periodic_timer | timer starts ticking
.....
QEMU starts to execute |
translated block |
| QEMU timer expires
|
CPU reads the timer register, | ptimer_get_count() return
ptimer_get_count() called | wrapped around value
.....
CPU interrupt handler kicks in | timer continue ticking, so
| any value is valid here
CPU stops the timer and sets |
counter to 0, returns from the |
handler |
.....
Now, for some reason, software |
sees that timer is stopped |
and do something using read |
value |

Program code sketch:

timer_interrupt_handler()
{
write32(1, TIMER_STOP); write32(0, TIMER_COUNTER); write32(TIMER_IRQ_CLEAR, TIMER_STATUS);

return IRQ_HANDLED;
}

program()
{
.....

..... <--- timer expired here
..... <--- interrupt handler executed here on real HW

var1 = read32(TIMER_COUNTER); <--- Emulated got wrapped,
real got 0

..... <--- interrupt handler executed here on QEMU

if (read32(TIMER_STATUS) & TIMER_RUNNING) {
.....
} else {
.....

write(var1 >> 16, SOME_DEV_REGISTER);
}

.....
}

Might emulated program behave differently from the real HW after it? Probably.

I want to mention that not only beefy generic CPU's are the ptimer users.

However, it seems that no one of the current ptimer users has a such restriction since it would already return 0 on expire and ptimer_tick() would happen after it. We can agree on keeping ptimer less universal in favor of the expire optimization, so somebody may improve it later if it would be needed.

I think removing the rate limiter's and clamping-affect on the read value makes it more universal if anything.

Do we agree?

I'm not sure, what are you referring to as the "expire optimizsation"?

I guess it might cause software
misbehavior if it assumes that the real hardware has CPU and timer running
in the same clock domain, i.e. such situation might be not possible.

Assumptions about the CPU clocking only make sense in icount mode, where the rate limiter is disable anyway.

Timer limiter has nothing to do with a returned value for the expired timer. Clock cycle accurate execution isn't relevant to upstream QEMU, I meant clocking in general. Emulated behaviour shouldn't diverge from the real HW.

But when the rate limiter is on for a short interval, it massively distorts this.

Regards,
Peter

So I'm
suggesting to return counter = 1 and allow ptimer_tick() happen on it's own.

My alternate suggestion is, if you detect that the tick should have already happened, just make it happen. I don't see the need to rate limit a polled timer.

Yes, I got your idea and it is absolutely correct if we agree on the above tradeoff (if that tradeoff exists).

- Disable limiting for that host machine?

- Tune timer limit?

Changelog for ARM MPTimer QEMUTimer to ptimer conversion:

V2: Fixed changing periodic timer counter value "on the fly". I added a
test to the gist to cover that issue.

V3: Fixed starting the timer with load = 0 and counter != 0, added tests
to the gist for this issue. Changed vmstate version for all VMSD's, since loadvm doesn't check version of nested VMSD.

V4: Fixed spurious IT bit set for the timer starting in the periodic mode
with counter = 0. Test added.

V5: Code cleanup, now depends on ptimer_set_limit() fix.

V6: No code change, added test to check ptimer_get_count() with corrected
.limit value.

V7: No change.

ARM MPTimer tests: https://gist.github.com/digetx/dbd46109503b1a91941a

Quoted 19 lines

Changelog for ARM MPTimer QEMUTimer to ptimer conversion:

V2: Fixed changing periodic timer counter value "on the fly". I added a
test to the gist to cover that issue.

V3: Fixed starting the timer with load = 0 and counter != 0, added tests
to the gist for this issue. Changed vmstate version for all VMSD's, since loadvm doesn't check version of nested VMSD.

V4: Fixed spurious IT bit set for the timer starting in the periodic mode
with counter = 0. Test added.

V5: Code cleanup, now depends on ptimer_set_limit() fix.

V6: No code change, added test to check ptimer_get_count() with corrected
.limit value.

V7: No change.

Quoted 1 line

ARM MPTimer tests: https://gist.github.com/digetx/dbd46109503b1a91941a

Hi. Given that we're now in the second half of softfreeze, and the discussions about the design of the changes to ptimer don't seem to have reached a conclusion yet, I would prefer to defer this to 2.6. (It feels a bit late in the release cycle to make changes to a core component like ptimer that affects multiple boards.)

My understanding is that although this patchset fixes a bunch of edge cases and odd features in the mptimer, there aren't any bugs that it fixes that cause problems for common guests like Linux, right?