[Qemu-devel] [PATCH v10 0/7] PTimer fixes/features and ARM MPTimer conversion

[Qemu-devel] [PATCH v10 0/7] PTimer fixes/features and ARM MPTimer conversion

[Dmitry Osipenko]

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.

    V8: No change.

    V9: No change.

    V10: Correctly handle cases when counter = load = 0 and prescaler != 0,
         i.e. triggering interrupt in that case. Call ptimer_* only when
         certain MPTimer state was changed, like prescaler change. Factor out
         timerblock_set_count from timerblock_run and inline both.
         Tests updated.

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


Patches for ptimer are introduced since V5 of "ARM MPTimer conversion".

Changelog for the ptimer patches:

    V5: Only fixed ptimer_set_limit() for the disabled timer.

    V6: As was pointed by Peter Maydell, there are other issues beyond
        ptimer_set_limit(), so V6 supposed to cover all those issues.

    V7: Added accidentally removed !use_icount check.
        Added missed "else" statement.

    V8: Adjust period instead of the limit and do it for periodic timer only
        (.limit adjusting bug). Added patch/fix for freq/period change and
        ptimer_get_count() improvement.

    V9: Don't do wrap around if counter == 0, otherwise polled periodic
        timer won't ever return counter = 0.

    V10: Addressed V8/9 review comments.
         Adjust timer period based on delta instead of limit.
         Don't wrap around when in icount mode.
         New patches: "on the fly" mode switch, silence error msg when
                      delta = load = 0, introduce ptimer_get_limit.

Dmitry Osipenko (7):
  hw/ptimer: Fix issues caused by the adjusted timer limit value
  hw/ptimer: Perform tick and counter wrap around if timer already
    expired
  hw/ptimer: Update .delta on period/freq change
  hw/ptimer: Support "on the fly" timer mode switch
  hw/ptimer: Legalize running with delta = load = 0
  hw/ptimer: Introduce ptimer_get_limit
  arm_mptimer: Convert to use ptimer

 hw/core/ptimer.c               | 109 +++++++++++++++++++++++------------
 hw/timer/arm_mptimer.c         | 127 ++++++++++++++++++++---------------------
 include/hw/ptimer.h            |   1 +
 include/hw/timer/arm_mptimer.h |   5 +-
 4 files changed, 137 insertions(+), 105 deletions(-)

-- 
2.6.4

[Qemu-devel] [PATCH v10 1/7] hw/ptimer: Fix issues caused by the adjusted timer limit value

[Dmitry Osipenko]

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

1) ptimer_get_count() returns incorrect counter value for the disabled
timer after loading the counter with a small value, because adjusted 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 adjusted 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 adjusting 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 adjusting timer period instead of the limit.
Perform the adjustment for periodic timer only. Use the delta value instead
of the limit to make decision whether adjustment is required, as limit could
be altered while timer is running, resulting in incorrect value returned by
ptimer_get_count.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
 hw/core/ptimer.c | 51 +++++++++++++++++++++++++++++++--------------------
 1 file changed, 31 insertions(+), 20 deletions(-)

diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
index edf077c..c3d31cb 100644
--- a/hw/core/ptimer.c
+++ b/hw/core/ptimer.c
@@ -34,6 +34,9 @@ static void ptimer_trigger(ptimer_state *s)
 
 static void ptimer_reload(ptimer_state *s)
 {
+    uint32_t period_frac = s->period_frac;
+    uint64_t period = s->period;
+
     if (s->delta == 0) {
         ptimer_trigger(s);
         s->delta = s->limit;
@@ -44,10 +47,24 @@ static void ptimer_reload(ptimer_state *s)
         return;
     }
 
+    /*
+     * 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 ((s->enabled == 1) && !use_icount && (s->delta * period < 10000)) {
+        period = 10000 / s->delta;
+        period_frac = 0;
+    }
+
     s->last_event = s->next_event;
-    s->next_event = s->last_event + s->delta * s->period;
-    if (s->period_frac) {
-        s->next_event += ((int64_t)s->period_frac * s->delta) >> 32;
+    s->next_event = s->last_event + s->delta * period;
+    if (period_frac) {
+        s->next_event += ((int64_t)period_frac * s->delta) >> 32;
     }
     timer_mod(s->timer, s->next_event);
 }
@@ -82,6 +99,13 @@ uint64_t ptimer_get_count(ptimer_state *s)
             uint64_t div;
             int clz1, clz2;
             int shift;
+            uint32_t period_frac = s->period_frac;
+            uint64_t period = s->period;
+
+            if ((s->enabled == 1) && !use_icount && (s->delta * period < 10000)) {
+                period = 10000 / s->delta;
+                period_frac = 0;
+            }
 
             /* We need to divide time by period, where time is stored in
                rem (64-bit integer) and period is stored in period/period_frac
@@ -94,7 +118,7 @@ uint64_t ptimer_get_count(ptimer_state *s)
             */
 
             rem = s->next_event - now;
-            div = s->period;
+            div = period;
 
             clz1 = clz64(rem);
             clz2 = clz64(div);
@@ -103,13 +127,13 @@ uint64_t ptimer_get_count(ptimer_state *s)
             rem <<= shift;
             div <<= shift;
             if (shift >= 32) {
-                div |= ((uint64_t)s->period_frac << (shift - 32));
+                div |= ((uint64_t)period_frac << (shift - 32));
             } else {
                 if (shift != 0)
-                    div |= (s->period_frac >> (32 - shift));
+                    div |= (period_frac >> (32 - shift));
                 /* Look at remaining bits of period_frac and round div up if 
                    necessary.  */
-                if ((uint32_t)(s->period_frac << shift))
+                if ((uint32_t)(period_frac << shift))
                     div += 1;
             }
             counter = rem / div;
@@ -181,19 +205,6 @@ 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;
-- 
2.6.4

[Qemu-devel] [PATCH v10 2/7] hw/ptimer: Perform tick and counter wrap around if timer already expired

[Dmitry Osipenko]

ptimer_get_count() might be called while QEMU timer already been expired.
In that case ptimer would return counter = 0, which might be undesirable
in case of polled timer. Do counter wrap around for periodic timer to keep
it distributed. In order to achieve more accurate emulation behaviour of
certain hardware, don't perform wrap around when in icount mode and return
counter = 0 in that case (that doesn't affect polled counter distribution).

In addition, there is no reason to keep expired timer tick deferred, so
just perform the tick from ptimer_get_count().

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
 hw/core/ptimer.c | 26 +++++++++++++++++++-------
 1 file changed, 19 insertions(+), 7 deletions(-)

diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
index c3d31cb..cf329eb 100644
--- a/hw/core/ptimer.c
+++ b/hw/core/ptimer.c
@@ -83,14 +83,17 @@ static void ptimer_tick(void *opaque)
 
 uint64_t ptimer_get_count(ptimer_state *s)
 {
-    int64_t now;
+    int enabled = s->enabled;
     uint64_t counter;
 
-    if (s->enabled) {
-        now = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
+    if (enabled) {
+        int64_t now = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
+        int64_t next = s->next_event;
+        int expired = (now - next >= 0);
+        int oneshot = (enabled == 2);
+
         /* Figure out the current counter value.  */
-        if (now - s->next_event > 0
-            || s->period == 0) {
+        if (s->period == 0 || (expired && (use_icount || oneshot))) {
             /* Prevent timer underflowing if it should already have
                triggered.  */
             counter = 0;
@@ -102,7 +105,7 @@ uint64_t ptimer_get_count(ptimer_state *s)
             uint32_t period_frac = s->period_frac;
             uint64_t period = s->period;
 
-            if ((s->enabled == 1) && !use_icount && (s->delta * period < 10000)) {
+            if (!oneshot && !use_icount && (s->delta * period < 10000)) {
                 period = 10000 / s->delta;
                 period_frac = 0;
             }
@@ -117,7 +120,7 @@ uint64_t ptimer_get_count(ptimer_state *s)
                backwards.
             */
 
-            rem = s->next_event - now;
+            rem = expired ? now - next : next - now;
             div = period;
 
             clz1 = clz64(rem);
@@ -137,6 +140,15 @@ uint64_t ptimer_get_count(ptimer_state *s)
                     div += 1;
             }
             counter = rem / div;
+
+            if (expired && (counter != 0)) {
+                /* Wrap around periodic counter.  */
+                counter = s->limit - counter % s->limit;
+            }
+        }
+
+        if (expired) {
+            ptimer_tick(s);
         }
     } else {
         counter = s->delta;
-- 
2.6.4

[Qemu-devel] [PATCH v10 3/7] hw/ptimer: Update .delta on period/freq change

[Dmitry Osipenko]

Delta value must be updated on period/freq change, otherwise running timer
would be restarted (counter reloaded with old delta). Only m68k/mcf520x
and arm/arm_timer devices are currently doing freq change correctly, i.e.
stopping the timer. Perform delta update to fix affected devices and
eliminate potential further mistakes.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Reviewed-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
---
 hw/core/ptimer.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
index cf329eb..0a54212 100644
--- a/hw/core/ptimer.c
+++ b/hw/core/ptimer.c
@@ -194,6 +194,7 @@ void ptimer_stop(ptimer_state *s)
 /* Set counter increment interval in nanoseconds.  */
 void ptimer_set_period(ptimer_state *s, int64_t period)
 {
+    s->delta = ptimer_get_count(s);
     s->period = period;
     s->period_frac = 0;
     if (s->enabled) {
@@ -205,6 +206,7 @@ void ptimer_set_period(ptimer_state *s, int64_t period)
 /* Set counter frequency in Hz.  */
 void ptimer_set_freq(ptimer_state *s, uint32_t freq)
 {
+    s->delta = ptimer_get_count(s);
     s->period = 1000000000ll / freq;
     s->period_frac = (1000000000ll << 32) / freq;
     if (s->enabled) {
-- 
2.6.4

[Qemu-devel] [PATCH v10 4/7] hw/ptimer: Support "on the fly" timer mode switch

[Dmitry Osipenko]

Allow to switch between periodic <-> oneshot modes while timer is running.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
 hw/core/ptimer.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
index 0a54212..6960738 100644
--- a/hw/core/ptimer.c
+++ b/hw/core/ptimer.c
@@ -167,16 +167,16 @@ void ptimer_set_count(ptimer_state *s, uint64_t count)
 
 void ptimer_run(ptimer_state *s, int oneshot)
 {
-    if (s->enabled) {
-        return;
-    }
-    if (s->period == 0) {
+    int was_disabled = !s->enabled;
+    if (was_disabled && s->period == 0) {
         fprintf(stderr, "Timer with period zero, disabling\n");
         return;
     }
     s->enabled = oneshot ? 2 : 1;
-    s->next_event = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
-    ptimer_reload(s);
+    if (was_disabled) {
+        s->next_event = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
+        ptimer_reload(s);
+    }
 }
 
 /* Pause a timer.  Note that this may cause it to "lose" time, even if it
-- 
2.6.4

[Qemu-devel] [PATCH v10 5/7] hw/ptimer: Legalize running with delta = load = 0

[Dmitry Osipenko]

Currently ptimer would print error message and clear enable flag for an
arming timer that has delta = load = 0. That actually could be a valid case
for some hardware, like instant IRQ trigger for oneshot timer or continuous
in periodic mode. Support those cases by printing error message only when
period = 0.

In addition, don't load one-shot timer when delta = 0 and actually stop the
timer by timer_del().

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
 hw/core/ptimer.c | 21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
index 6960738..42e44f9 100644
--- a/hw/core/ptimer.c
+++ b/hw/core/ptimer.c
@@ -36,13 +36,20 @@ static void ptimer_reload(ptimer_state *s)
 {
     uint32_t period_frac = s->period_frac;
     uint64_t period = s->period;
+    int periodic = (s->enabled == 1);
 
-    if (s->delta == 0) {
+    if (s->delta == 0 && period != 0) {
         ptimer_trigger(s);
-        s->delta = s->limit;
+        if (periodic) {
+            s->delta = s->limit;
+        }
     }
-    if (s->delta == 0 || s->period == 0) {
-        fprintf(stderr, "Timer with period zero, disabling\n");
+    if (s->delta == 0 || period == 0) {
+        if (period == 0) {
+            fprintf(stderr, "Timer with period zero, disabling\n");
+            s->delta = 0;
+        }
+        timer_del(s->timer);
         s->enabled = 0;
         return;
     }
@@ -56,7 +63,7 @@ static void ptimer_reload(ptimer_state *s)
      * on the current generation of host machines.
      */
 
-    if ((s->enabled == 1) && !use_icount && (s->delta * period < 10000)) {
+    if (periodic && !use_icount && (s->delta * period < 10000)) {
         period = 10000 / s->delta;
         period_frac = 0;
     }
@@ -86,14 +93,14 @@ uint64_t ptimer_get_count(ptimer_state *s)
     int enabled = s->enabled;
     uint64_t counter;
 
-    if (enabled) {
+    if (enabled && s->delta != 0) {
         int64_t now = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
         int64_t next = s->next_event;
         int expired = (now - next >= 0);
         int oneshot = (enabled == 2);
 
         /* Figure out the current counter value.  */
-        if (s->period == 0 || (expired && (use_icount || oneshot))) {
+        if (expired && (use_icount || oneshot)) {
             /* Prevent timer underflowing if it should already have
                triggered.  */
             counter = 0;
-- 
2.6.4

[Qemu-devel] [PATCH v10 6/7] hw/ptimer: Introduce ptimer_get_limit

[Dmitry Osipenko]

Currently ptimer users are used to store copy of the limit value, because
ptimer doesn't provide facility to retrieve the limit. Let's provide it.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
 hw/core/ptimer.c    | 5 +++++
 include/hw/ptimer.h | 1 +
 2 files changed, 6 insertions(+)

diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
index 42e44f9..0201d1b 100644
--- a/hw/core/ptimer.c
+++ b/hw/core/ptimer.c
@@ -235,6 +235,11 @@ void ptimer_set_limit(ptimer_state *s, uint64_t limit, int reload)
     }
 }
 
+uint64_t ptimer_get_limit(ptimer_state *s)
+{
+    return s->limit;
+}
+
 const VMStateDescription vmstate_ptimer = {
     .name = "ptimer",
     .version_id = 1,
diff --git a/include/hw/ptimer.h b/include/hw/ptimer.h
index 8ebacbb..e397db5 100644
--- a/include/hw/ptimer.h
+++ b/include/hw/ptimer.h
@@ -19,6 +19,7 @@ typedef void (*ptimer_cb)(void *opaque);
 ptimer_state *ptimer_init(QEMUBH *bh);
 void ptimer_set_period(ptimer_state *s, int64_t period);
 void ptimer_set_freq(ptimer_state *s, uint32_t freq);
+uint64_t ptimer_get_limit(ptimer_state *s);
 void ptimer_set_limit(ptimer_state *s, uint64_t limit, int reload);
 uint64_t ptimer_get_count(ptimer_state *s);
 void ptimer_set_count(ptimer_state *s, uint64_t count);
-- 
2.6.4

[Qemu-devel] [PATCH v10 7/7] arm_mptimer: Convert to use ptimer

[Dmitry Osipenko]

Current ARM MPTimer implementation uses QEMUTimer for the actual timer,
this implementation isn't complete and mostly tries to duplicate of what
generic ptimer is already doing fine.

Conversion to ptimer brings the following benefits and fixes:
	- Simple timer pausing implementation
	- Fixes counter value preservation after stopping the timer
	- Correctly handles prescaler != 0 cases
	- Code simplification and reduction

Bump VMSD to version 3, since VMState is changed and is not compatible
with the previous implementation.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
 hw/timer/arm_mptimer.c         | 127 ++++++++++++++++++++---------------------
 include/hw/timer/arm_mptimer.h |   5 +-
 2 files changed, 63 insertions(+), 69 deletions(-)

diff --git a/hw/timer/arm_mptimer.c b/hw/timer/arm_mptimer.c
index 3e59c2a..5568300 100644
--- a/hw/timer/arm_mptimer.c
+++ b/hw/timer/arm_mptimer.c
@@ -19,8 +19,9 @@
  * with this program; if not, see <http://www.gnu.org/licenses/>.
  */
 
+#include "hw/ptimer.h"
 #include "hw/timer/arm_mptimer.h"
-#include "qemu/timer.h"
+#include "qemu/main-loop.h"
 #include "qom/cpu.h"
 
 /* This device implements the per-cpu private timer and watchdog block
@@ -42,33 +43,33 @@ static inline void timerblock_update_irq(TimerBlock *tb)
 }
 
 /* Return conversion factor from mpcore timer ticks to qemu timer ticks.  */
-static inline uint32_t timerblock_scale(TimerBlock *tb)
+static inline uint32_t timerblock_scale(uint32_t control)
 {
-    return (((tb->control >> 8) & 0xff) + 1) * 10;
+    return (((control >> 8) & 0xff) + 1) * 10;
 }
 
-static void timerblock_reload(TimerBlock *tb, int restart)
+static inline void timerblock_set_count(TimerBlock *tb, uint32_t control,
+                                        uint64_t *count)
 {
-    if (tb->count == 0) {
-        return;
+    /* PTimer would immediately trigger interrupt for periodic timer
+     * when counter set to 0, MPtimer under certain condition only.  */
+    if ((control & 3) == 3 && (*count == 0) && (control & 0xff00) == 0) {
+        *count = ptimer_get_limit(tb->timer);
     }
-    if (restart) {
-        tb->tick = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
+    ptimer_set_count(tb->timer, *count);
+}
+
+static inline void timerblock_run(TimerBlock *tb, uint32_t control, int cond)
+{
+    if (cond) {
+        ptimer_run(tb->timer, !(control & 2));
     }
-    tb->tick += (int64_t)tb->count * timerblock_scale(tb);
-    timer_mod(tb->timer, tb->tick);
 }
 
 static void timerblock_tick(void *opaque)
 {
     TimerBlock *tb = (TimerBlock *)opaque;
     tb->status = 1;
-    if (tb->control & 2) {
-        tb->count = tb->load;
-        timerblock_reload(tb, 0);
-    } else {
-        tb->count = 0;
-    }
     timerblock_update_irq(tb);
 }
 
@@ -76,21 +77,11 @@ static uint64_t timerblock_read(void *opaque, hwaddr addr,
                                 unsigned size)
 {
     TimerBlock *tb = (TimerBlock *)opaque;
-    int64_t val;
     switch (addr) {
     case 0: /* Load */
-        return tb->load;
+        return ptimer_get_limit(tb->timer);
     case 4: /* Counter.  */
-        if (((tb->control & 1) == 0) || (tb->count == 0)) {
-            return 0;
-        }
-        /* Slow and ugly, but hopefully won't happen too often.  */
-        val = tb->tick - qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
-        val /= timerblock_scale(tb);
-        if (val < 0) {
-            val = 0;
-        }
-        return val;
+        return ptimer_get_count(tb->timer);
     case 8: /* Control.  */
         return tb->control;
     case 12: /* Interrupt status.  */
@@ -104,39 +95,48 @@ static void timerblock_write(void *opaque, hwaddr addr,
                              uint64_t value, unsigned size)
 {
     TimerBlock *tb = (TimerBlock *)opaque;
-    int64_t old;
+    uint32_t control = tb->control;
     switch (addr) {
     case 0: /* Load */
-        tb->load = value;
-        /* Fall through.  */
-    case 4: /* Counter.  */
-        if ((tb->control & 1) && tb->count) {
-            /* Cancel the previous timer.  */
-            timer_del(tb->timer);
+        /* Setting load to 0 stops the timer if prescaler == 0.  */
+        if ((control & 1) && (value == 0) && (control & 0xff00) == 0) {
+            ptimer_stop(tb->timer);
+            control &= ~1;
         }
-        tb->count = value;
-        if (tb->control & 1) {
-            timerblock_reload(tb, 1);
+        ptimer_set_limit(tb->timer, value, 1);
+        timerblock_run(tb, control, (control & 1));
+        break;
+    case 4: /* Counter.  */
+        /* Setting counter to 0 stops the one-shot timer if prescaler == 0.  */
+        if ((control & 3) == 1 && (value == 0) && (control & 0xff00) == 0) {
+            ptimer_stop(tb->timer);
+            control &= ~1;
         }
+        timerblock_set_count(tb, control, &value);
+        timerblock_run(tb, control, (value != 0) && (control & 1));
         break;
     case 8: /* Control.  */
-        old = tb->control;
-        tb->control = value;
-        if (value & 1) {
-            if ((old & 1) && (tb->count != 0)) {
-                /* Do nothing if timer is ticking right now.  */
-                break;
+        if ((value & 1) && (control & 3) != (value & 3)) {
+            uint64_t count = (value & 0xff00) ? 1 : ptimer_get_count(tb->timer);
+            if ((count == 0) && (value & 2)) {
+                timerblock_set_count(tb, value, &count);
             }
-            if (tb->control & 2) {
-                tb->count = tb->load;
-            }
-            timerblock_reload(tb, 1);
-        } else if (old & 1) {
-            /* Shutdown the timer.  */
-            timer_del(tb->timer);
+            timerblock_run(tb, value, (count != 0));
+        } else if ((control & 1) && !(value & 1)) {
+            ptimer_stop(tb->timer);
+        }
+        if ((control & 0xff00) != (value & 0xff00)) {
+            ptimer_set_period(tb->timer, timerblock_scale(value));
         }
+        tb->control = value;
         break;
     case 12: /* Interrupt status.  */
+        /* Simulate continuous IRQ gen.  */
+        if ((control & 3) == 3 && (control & 0xff00) != 0) {
+            if (ptimer_get_limit(tb->timer) == 0) {
+                break;
+            }
+        }
         tb->status &= ~value;
         timerblock_update_irq(tb);
         break;
@@ -184,13 +184,12 @@ static const MemoryRegionOps timerblock_ops = {
 
 static void timerblock_reset(TimerBlock *tb)
 {
-    tb->count = 0;
-    tb->load = 0;
     tb->control = 0;
     tb->status = 0;
-    tb->tick = 0;
     if (tb->timer) {
-        timer_del(tb->timer);
+        ptimer_stop(tb->timer);
+        ptimer_set_limit(tb->timer, 0, 1);
+        ptimer_set_period(tb->timer, timerblock_scale(0));
     }
 }
 
@@ -235,7 +234,8 @@ static void arm_mptimer_realize(DeviceState *dev, Error **errp)
      */
     for (i = 0; i < s->num_cpu; i++) {
         TimerBlock *tb = &s->timerblock[i];
-        tb->timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, timerblock_tick, tb);
+        QEMUBH *bh = qemu_bh_new(timerblock_tick, tb);
+        tb->timer = ptimer_init(bh);
         sysbus_init_irq(sbd, &tb->irq);
         memory_region_init_io(&tb->iomem, OBJECT(s), &timerblock_ops, tb,
                               "arm_mptimer_timerblock", 0x20);
@@ -245,26 +245,23 @@ static void arm_mptimer_realize(DeviceState *dev, Error **errp)
 
 static const VMStateDescription vmstate_timerblock = {
     .name = "arm_mptimer_timerblock",
-    .version_id = 2,
-    .minimum_version_id = 2,
+    .version_id = 3,
+    .minimum_version_id = 3,
     .fields = (VMStateField[]) {
-        VMSTATE_UINT32(count, TimerBlock),
-        VMSTATE_UINT32(load, TimerBlock),
         VMSTATE_UINT32(control, TimerBlock),
         VMSTATE_UINT32(status, TimerBlock),
-        VMSTATE_INT64(tick, TimerBlock),
-        VMSTATE_TIMER_PTR(timer, TimerBlock),
+        VMSTATE_PTIMER(timer, TimerBlock),
         VMSTATE_END_OF_LIST()
     }
 };
 
 static const VMStateDescription vmstate_arm_mptimer = {
     .name = "arm_mptimer",
-    .version_id = 2,
-    .minimum_version_id = 2,
+    .version_id = 3,
+    .minimum_version_id = 3,
     .fields = (VMStateField[]) {
         VMSTATE_STRUCT_VARRAY_UINT32(timerblock, ARMMPTimerState, num_cpu,
-                                     2, vmstate_timerblock, TimerBlock),
+                                     3, vmstate_timerblock, TimerBlock),
         VMSTATE_END_OF_LIST()
     }
 };
diff --git a/include/hw/timer/arm_mptimer.h b/include/hw/timer/arm_mptimer.h
index b34cba0..c46d8d2 100644
--- a/include/hw/timer/arm_mptimer.h
+++ b/include/hw/timer/arm_mptimer.h
@@ -27,12 +27,9 @@
 
 /* State of a single timer or watchdog block */
 typedef struct {
-    uint32_t count;
-    uint32_t load;
     uint32_t control;
     uint32_t status;
-    int64_t tick;
-    QEMUTimer *timer;
+    struct ptimer_state *timer;
     qemu_irq irq;
     MemoryRegion iomem;
 } TimerBlock;
-- 
2.6.4

Re: [Qemu-devel] [PATCH v10 1/7] hw/ptimer: Fix issues caused by the adjusted timer limit value

[Peter Crosthwaite]

On Sat, Jan 9, 2016 at 9:39 AM, Dmitry Osipenko <digetx@gmail.com> wrote:
> Multiple issues here related to the timer with a adjusted .limit value:
>
> 1) ptimer_get_count() returns incorrect counter value for the disabled
> timer after loading the counter with a small value, because adjusted 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 adjusted 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 adjusting 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 adjusting timer period instead of the limit.
> Perform the adjustment for periodic timer only. Use the delta value instead
> of the limit to make decision whether adjustment is required, as limit could
> be altered while timer is running, resulting in incorrect value returned by
> ptimer_get_count.
>
> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>

Reviewed-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>

> ---
>  hw/core/ptimer.c | 51 +++++++++++++++++++++++++++++++--------------------
>  1 file changed, 31 insertions(+), 20 deletions(-)
>
> diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
> index edf077c..c3d31cb 100644
> --- a/hw/core/ptimer.c
> +++ b/hw/core/ptimer.c
> @@ -34,6 +34,9 @@ static void ptimer_trigger(ptimer_state *s)
>
>  static void ptimer_reload(ptimer_state *s)
>  {
> +    uint32_t period_frac = s->period_frac;
> +    uint64_t period = s->period;
> +
>      if (s->delta == 0) {
>          ptimer_trigger(s);
>          s->delta = s->limit;
> @@ -44,10 +47,24 @@ static void ptimer_reload(ptimer_state *s)
>          return;
>      }
>
> +    /*
> +     * 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 ((s->enabled == 1) && !use_icount && (s->delta * period < 10000)) {
> +        period = 10000 / s->delta;
> +        period_frac = 0;
> +    }
> +
>      s->last_event = s->next_event;
> -    s->next_event = s->last_event + s->delta * s->period;
> -    if (s->period_frac) {
> -        s->next_event += ((int64_t)s->period_frac * s->delta) >> 32;
> +    s->next_event = s->last_event + s->delta * period;
> +    if (period_frac) {
> +        s->next_event += ((int64_t)period_frac * s->delta) >> 32;
>      }
>      timer_mod(s->timer, s->next_event);
>  }
> @@ -82,6 +99,13 @@ uint64_t ptimer_get_count(ptimer_state *s)
>              uint64_t div;
>              int clz1, clz2;
>              int shift;
> +            uint32_t period_frac = s->period_frac;
> +            uint64_t period = s->period;
> +
> +            if ((s->enabled == 1) && !use_icount && (s->delta * period < 10000)) {
> +                period = 10000 / s->delta;
> +                period_frac = 0;
> +            }
>
>              /* We need to divide time by period, where time is stored in
>                 rem (64-bit integer) and period is stored in period/period_frac
> @@ -94,7 +118,7 @@ uint64_t ptimer_get_count(ptimer_state *s)
>              */
>
>              rem = s->next_event - now;
> -            div = s->period;
> +            div = period;
>
>              clz1 = clz64(rem);
>              clz2 = clz64(div);
> @@ -103,13 +127,13 @@ uint64_t ptimer_get_count(ptimer_state *s)
>              rem <<= shift;
>              div <<= shift;
>              if (shift >= 32) {
> -                div |= ((uint64_t)s->period_frac << (shift - 32));
> +                div |= ((uint64_t)period_frac << (shift - 32));
>              } else {
>                  if (shift != 0)
> -                    div |= (s->period_frac >> (32 - shift));
> +                    div |= (period_frac >> (32 - shift));
>                  /* Look at remaining bits of period_frac and round div up if
>                     necessary.  */
> -                if ((uint32_t)(s->period_frac << shift))
> +                if ((uint32_t)(period_frac << shift))
>                      div += 1;
>              }
>              counter = rem / div;
> @@ -181,19 +205,6 @@ 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;
> --
> 2.6.4
>

Re: [Qemu-devel] [PATCH v10 2/7] hw/ptimer: Perform tick and counter wrap around if timer already expired

[Peter Crosthwaite]

On Sat, Jan 9, 2016 at 9:39 AM, Dmitry Osipenko <digetx@gmail.com> wrote:
> ptimer_get_count() might be called while QEMU timer already been expired.
> In that case ptimer would return counter = 0, which might be undesirable
> in case of polled timer. Do counter wrap around for periodic timer to keep
> it distributed. In order to achieve more accurate emulation behaviour of
> certain hardware, don't perform wrap around when in icount mode and return
> counter = 0 in that case (that doesn't affect polled counter distribution).
>
> In addition, there is no reason to keep expired timer tick deferred, so
> just perform the tick from ptimer_get_count().
>
> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
> ---
>  hw/core/ptimer.c | 26 +++++++++++++++++++-------
>  1 file changed, 19 insertions(+), 7 deletions(-)
>
> diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
> index c3d31cb..cf329eb 100644
> --- a/hw/core/ptimer.c
> +++ b/hw/core/ptimer.c
> @@ -83,14 +83,17 @@ static void ptimer_tick(void *opaque)
>
>  uint64_t ptimer_get_count(ptimer_state *s)
>  {
> -    int64_t now;
> +    int enabled = s->enabled;

You haven't added any additional usages of s->enabled to really
warrant the new local variable, I think it should just stay as
s->enabled to avoid churn.

>      uint64_t counter;
>
> -    if (s->enabled) {
> -        now = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
> +    if (enabled) {
> +        int64_t now = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
> +        int64_t next = s->next_event;
> +        int expired = (now - next >= 0);

bool.

> +        int oneshot = (enabled == 2);

bool.

Otherwise:

Reviewed-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>

> +
>          /* Figure out the current counter value.  */
> -        if (now - s->next_event > 0
> -            || s->period == 0) {
> +        if (s->period == 0 || (expired && (use_icount || oneshot))) {
>              /* Prevent timer underflowing if it should already have
>                 triggered.  */
>              counter = 0;
> @@ -102,7 +105,7 @@ uint64_t ptimer_get_count(ptimer_state *s)
>              uint32_t period_frac = s->period_frac;
>              uint64_t period = s->period;
>
> -            if ((s->enabled == 1) && !use_icount && (s->delta * period < 10000)) {
> +            if (!oneshot && !use_icount && (s->delta * period < 10000)) {
>                  period = 10000 / s->delta;
>                  period_frac = 0;
>              }
> @@ -117,7 +120,7 @@ uint64_t ptimer_get_count(ptimer_state *s)
>                 backwards.
>              */
>
> -            rem = s->next_event - now;
> +            rem = expired ? now - next : next - now;
>              div = period;
>
>              clz1 = clz64(rem);
> @@ -137,6 +140,15 @@ uint64_t ptimer_get_count(ptimer_state *s)
>                      div += 1;
>              }
>              counter = rem / div;
> +
> +            if (expired && (counter != 0)) {
> +                /* Wrap around periodic counter.  */
> +                counter = s->limit - counter % s->limit;
> +            }
> +        }
> +
> +        if (expired) {
> +            ptimer_tick(s);
>          }
>      } else {
>          counter = s->delta;
> --
> 2.6.4
>

Re: [Qemu-devel] [PATCH v10 4/7] hw/ptimer: Support "on the fly" timer mode switch

[Peter Crosthwaite]

On Sat, Jan 9, 2016 at 9:39 AM, Dmitry Osipenko <digetx@gmail.com> wrote:
> Allow to switch between periodic <-> oneshot modes while timer is running.
>

"Allow switching"

> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
> ---
>  hw/core/ptimer.c | 12 ++++++------
>  1 file changed, 6 insertions(+), 6 deletions(-)
>
> diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
> index 0a54212..6960738 100644
> --- a/hw/core/ptimer.c
> +++ b/hw/core/ptimer.c
> @@ -167,16 +167,16 @@ void ptimer_set_count(ptimer_state *s, uint64_t count)
>
>  void ptimer_run(ptimer_state *s, int oneshot)
>  {
> -    if (s->enabled) {
> -        return;
> -    }
> -    if (s->period == 0) {
> +    int was_disabled = !s->enabled;

bool

blank line here.

Otherwise:

Reviewed-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>

Regards,
Peter

> +    if (was_disabled && s->period == 0) {
>          fprintf(stderr, "Timer with period zero, disabling\n");
>          return;
>      }
>      s->enabled = oneshot ? 2 : 1;
> -    s->next_event = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
> -    ptimer_reload(s);
> +    if (was_disabled) {
> +        s->next_event = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
> +        ptimer_reload(s);
> +    }
>  }
>
>  /* Pause a timer.  Note that this may cause it to "lose" time, even if it
> --
> 2.6.4
>

Re: [Qemu-devel] [PATCH v10 2/7] hw/ptimer: Perform tick and counter wrap around if timer already expired

[Dmitry Osipenko]

10.01.2016 03:44, Peter Crosthwaite пишет:
> On Sat, Jan 9, 2016 at 9:39 AM, Dmitry Osipenko <digetx@gmail.com> wrote:
>> ptimer_get_count() might be called while QEMU timer already been expired.
>> In that case ptimer would return counter = 0, which might be undesirable
>> in case of polled timer. Do counter wrap around for periodic timer to keep
>> it distributed. In order to achieve more accurate emulation behaviour of
>> certain hardware, don't perform wrap around when in icount mode and return
>> counter = 0 in that case (that doesn't affect polled counter distribution).
>>
>> In addition, there is no reason to keep expired timer tick deferred, so
>> just perform the tick from ptimer_get_count().
>>
>> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
>> ---
>>   hw/core/ptimer.c | 26 +++++++++++++++++++-------
>>   1 file changed, 19 insertions(+), 7 deletions(-)
>>
>> diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
>> index c3d31cb..cf329eb 100644
>> --- a/hw/core/ptimer.c
>> +++ b/hw/core/ptimer.c
>> @@ -83,14 +83,17 @@ static void ptimer_tick(void *opaque)
>>
>>   uint64_t ptimer_get_count(ptimer_state *s)
>>   {
>> -    int64_t now;
>> +    int enabled = s->enabled;
>
> You haven't added any additional usages of s->enabled to really
> warrant the new local variable, I think it should just stay as
> s->enabled to avoid churn.
>

Compiler doesn't know that ptimer struct is supposed to be thread-safe there and 
infers memory load instructions rather than use pure cpu registers. Local 
variable would help to produce somewhat more rational. However, given how 
optimized and complex modern cpu's, I'm not really sure that it would bring any 
benefit and agree that it might not worth churning. Thanks for review!

-- 
Dmitry

Re: [Qemu-devel] [PATCH v10 7/7] arm_mptimer: Convert to use ptimer

[Dmitry Osipenko]

I missed case where periodic timer should stop in the following case:

load = 0, counter != 0 -> run -> set counter = 0 -> should stop

Test added. Will fix it in V11.

-- 
Dmitry

Re: [Qemu-devel] [PATCH v10 7/7] arm_mptimer: Convert to use ptimer

[Dmitry Osipenko]

10.01.2016 21:05, Dmitry Osipenko пишет:
> I missed case where periodic timer should stop in the following case:
>
> load = 0, counter != 0 -> run -> set counter = 0 -> should stop
>
> Test added. Will fix it in V11.
>

Forgot to mention that prescaler must be 0 in that case. Prescaler != 0 would 
result in unclearable IRQ while periodic timer is enabled and counter = load = 0.

-- 
Dmitry

Re: [Qemu-devel] [PATCH v10 5/7] hw/ptimer: Legalize running with delta = load = 0

[Peter Crosthwaite]

On Sat, Jan 09, 2016 at 08:39:53PM +0300, Dmitry Osipenko wrote:
> Currently ptimer would print error message and clear enable flag for an
> arming timer that has delta = load = 0. That actually could be a valid case
> for some hardware, like instant IRQ trigger for oneshot timer or continuous
> in periodic mode. Support those cases by printing error message only when
> period = 0.
> 

Isn't the continuous-periodic the same as period = 0, so if we were to really
support this, there should be no error message. This would simplify as we
can remove the conditionals of 0 period completely and rely only on the
too-fast clamps you add in previous patches.

Regards,
Peter

> In addition, don't load one-shot timer when delta = 0 and actually stop the
> timer by timer_del().
> 
> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
> ---
>  hw/core/ptimer.c | 21 ++++++++++++++-------
>  1 file changed, 14 insertions(+), 7 deletions(-)
> 
> diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
> index 6960738..42e44f9 100644
> --- a/hw/core/ptimer.c
> +++ b/hw/core/ptimer.c
> @@ -36,13 +36,20 @@ static void ptimer_reload(ptimer_state *s)
>  {
>      uint32_t period_frac = s->period_frac;
>      uint64_t period = s->period;
> +    int periodic = (s->enabled == 1);
>  
> -    if (s->delta == 0) {
> +    if (s->delta == 0 && period != 0) {
>          ptimer_trigger(s);
> -        s->delta = s->limit;
> +        if (periodic) {
> +            s->delta = s->limit;
> +        }
>      }
> -    if (s->delta == 0 || s->period == 0) {
> -        fprintf(stderr, "Timer with period zero, disabling\n");
> +    if (s->delta == 0 || period == 0) {
> +        if (period == 0) {
> +            fprintf(stderr, "Timer with period zero, disabling\n");
> +            s->delta = 0;
> +        }
> +        timer_del(s->timer);
>          s->enabled = 0;
>          return;
>      }
> @@ -56,7 +63,7 @@ static void ptimer_reload(ptimer_state *s)
>       * on the current generation of host machines.
>       */
>  
> -    if ((s->enabled == 1) && !use_icount && (s->delta * period < 10000)) {
> +    if (periodic && !use_icount && (s->delta * period < 10000)) {
>          period = 10000 / s->delta;
>          period_frac = 0;
>      }
> @@ -86,14 +93,14 @@ uint64_t ptimer_get_count(ptimer_state *s)
>      int enabled = s->enabled;
>      uint64_t counter;
>  
> -    if (enabled) {
> +    if (enabled && s->delta != 0) {
>          int64_t now = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
>          int64_t next = s->next_event;
>          int expired = (now - next >= 0);
>          int oneshot = (enabled == 2);
>  
>          /* Figure out the current counter value.  */
> -        if (s->period == 0 || (expired && (use_icount || oneshot))) {
> +        if (expired && (use_icount || oneshot)) {
>              /* Prevent timer underflowing if it should already have
>                 triggered.  */
>              counter = 0;
> -- 
> 2.6.4
> 

Re: [Qemu-devel] [PATCH v10 6/7] hw/ptimer: Introduce ptimer_get_limit

[Peter Crosthwaite]

On Sat, Jan 09, 2016 at 08:39:54PM +0300, Dmitry Osipenko wrote:
> Currently ptimer users are used to store copy of the limit value, because
> ptimer doesn't provide facility to retrieve the limit. Let's provide it.
> 
> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>

Fair call. One less piece of duped state for the VMSDs.

Reviewed-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>

Regards,
Peter

> ---
>  hw/core/ptimer.c    | 5 +++++
>  include/hw/ptimer.h | 1 +
>  2 files changed, 6 insertions(+)
> 
> diff --git a/hw/core/ptimer.c b/hw/core/ptimer.c
> index 42e44f9..0201d1b 100644
> --- a/hw/core/ptimer.c
> +++ b/hw/core/ptimer.c
> @@ -235,6 +235,11 @@ void ptimer_set_limit(ptimer_state *s, uint64_t limit, int reload)
>      }
>  }
>  
> +uint64_t ptimer_get_limit(ptimer_state *s)
> +{
> +    return s->limit;
> +}
> +
>  const VMStateDescription vmstate_ptimer = {
>      .name = "ptimer",
>      .version_id = 1,
> diff --git a/include/hw/ptimer.h b/include/hw/ptimer.h
> index 8ebacbb..e397db5 100644
> --- a/include/hw/ptimer.h
> +++ b/include/hw/ptimer.h
> @@ -19,6 +19,7 @@ typedef void (*ptimer_cb)(void *opaque);
>  ptimer_state *ptimer_init(QEMUBH *bh);
>  void ptimer_set_period(ptimer_state *s, int64_t period);
>  void ptimer_set_freq(ptimer_state *s, uint32_t freq);
> +uint64_t ptimer_get_limit(ptimer_state *s);
>  void ptimer_set_limit(ptimer_state *s, uint64_t limit, int reload);
>  uint64_t ptimer_get_count(ptimer_state *s);
>  void ptimer_set_count(ptimer_state *s, uint64_t count);
> -- 
> 2.6.4
> 

Re: [Qemu-devel] [PATCH v10 5/7] hw/ptimer: Legalize running with delta = load = 0

[Dmitry Osipenko]

12.01.2016 06:58, Peter Crosthwaite пишет:
> On Sat, Jan 09, 2016 at 08:39:53PM +0300, Dmitry Osipenko wrote:
>> Currently ptimer would print error message and clear enable flag for an
>> arming timer that has delta = load = 0. That actually could be a valid case
>> for some hardware, like instant IRQ trigger for oneshot timer or continuous
>> in periodic mode. Support those cases by printing error message only when
>> period = 0.
>>
>
> Isn't the continuous-periodic the same as period = 0, so if we were to really
> support this, there should be no error message. This would simplify as we
> can remove the conditionals of 0 period completely and rely only on the
> too-fast clamps you add in previous patches.
>

I don't think that clamping is needed. Instead doing the ptimer_tick might be 
necessary, so ptimer user could handle that case on it own.

BTW, that printf isn't quite reliable, hw_error or other way of execution abort 
should been used instead.

Thanks for the comment.

-- 
Dmitry

Re: [Qemu-devel] [PATCH v10 5/7] hw/ptimer: Legalize running with delta = load = 0

[Dmitry Osipenko]

12.01.2016 21:12, Dmitry Osipenko пишет:
> 12.01.2016 06:58, Peter Crosthwaite пишет:
>> On Sat, Jan 09, 2016 at 08:39:53PM +0300, Dmitry Osipenko wrote:
>>> Currently ptimer would print error message and clear enable flag for an
>>> arming timer that has delta = load = 0. That actually could be a valid case
>>> for some hardware, like instant IRQ trigger for oneshot timer or continuous
>>> in periodic mode. Support those cases by printing error message only when
>>> period = 0.
>>>
>>
>> Isn't the continuous-periodic the same as period = 0, so if we were to really
>> support this, there should be no error message. This would simplify as we
>> can remove the conditionals of 0 period completely and rely only on the
>> too-fast clamps you add in previous patches.
>>
>
> I don't think that clamping is needed. Instead doing the ptimer_tick might be
> necessary, so ptimer user could handle that case on it own.
>
> BTW, that printf isn't quite reliable, hw_error or other way of execution abort
> should been used instead.
>
> Thanks for the comment.
>

Looking more at it, I think we should keep period = 0 forbidden. So it's treated 
as undefined behaviour and ptimer user should take care of it. If we'd want to 
handle period = 0 within ptimer, then we should also handle freq = 0, which 
implies potential ptimer VMSD version bump. Also it is uncertain what default 
behaviour should be chosen for period = 0, my guess is that pausing 
(_not_disabling_) the timer (like in freq = 0 case) should be more common among 
various hardware. If you have any thoughts on it, please let me know.

-- 
Dmitry

Re: [Qemu-devel] [PATCH v10 5/7] hw/ptimer: Legalize running with delta = load = 0

[Dmitry Osipenko]

12.01.2016 21:12, Dmitry Osipenko пишет:
> 12.01.2016 06:58, Peter Crosthwaite пишет:
>> On Sat, Jan 09, 2016 at 08:39:53PM +0300, Dmitry Osipenko wrote:
>>> Currently ptimer would print error message and clear enable flag for an
>>> arming timer that has delta = load = 0. That actually could be a valid case
>>> for some hardware, like instant IRQ trigger for oneshot timer or continuous
>>> in periodic mode. Support those cases by printing error message only when
>>> period = 0.
>>>
>>
>> Isn't the continuous-periodic the same as period = 0, so if we were to really
>> support this, there should be no error message. This would simplify as we
>> can remove the conditionals of 0 period completely and rely only on the
>> too-fast clamps you add in previous patches.
>>
>
> I don't think that clamping is needed. Instead doing the ptimer_tick might be
> necessary, so ptimer user could handle that case on it own.
>
> BTW, that printf isn't quite reliable, hw_error or other way of execution abort
> should been used instead.
>
> Thanks for the comment.
>

Just tried with the clamping. It works, but what clamping should be chosen for 
icount mode? delta * period = 1ns is damn slow.

However, I'm still not sure what benefit has clamping over unclearable IRQ... 
only disadvantages in form of slowdown in icount mode, +2 lines of ptimer code 
and possible misbehaviour of blazing fast host machine in non-icount mode.

-- 
Dmitry

Re: [Qemu-devel] [PATCH v10 2/7] hw/ptimer: Perform tick and counter wrap around if timer already expired

[Dmitry Osipenko]

Hi Peter,

10.01.2016 03:44, Peter Crosthwaite пишет:
> On Sat, Jan 9, 2016 at 9:39 AM, Dmitry Osipenko <digetx@gmail.com> wrote:
[cut]
>> In addition, there is no reason to keep expired timer tick deferred, so
>> just perform the tick from ptimer_get_count().
>>
[cut]

I noticed an issue here... The problem is that device reset invokes 
ptimer_stop() that invokes ptimer_get_count() that might cause the tick after 
reset, i.e. bogus bh would be invoked after QEMU reset. I have reproduced that 
issue.

The solution might be to introduce ptimer_reset() that would stop QEMU timer and 
reset delta/load/period without invoking ptimer_get_count. And of course all 
devices should be updated to use new ptimer_reset() prior to "Perform tick and 
counter wrap around if timer already expired" patch, but that's not an issue I 
suppose. Please let me know if you have any objections, I'm leaning to do it in V11.

void ptimer_reset(ptimer_state *s)
{
     timer_del(s->timer);
     s->enabled = 0;
     s->period_frac = 0;
     s->period = 0;
     s->delta = 0;
     s->limit = 0;
}

-- 
Dmitry

Re: [Qemu-devel] [PATCH v10 2/7] hw/ptimer: Perform tick and counter wrap around if timer already expired

[Dmitry Osipenko]

20.01.2016 20:03, Dmitry Osipenko пишет:
> Hi Peter,
>
> 10.01.2016 03:44, Peter Crosthwaite пишет:
>> On Sat, Jan 9, 2016 at 9:39 AM, Dmitry Osipenko <digetx@gmail.com> wrote:
> [cut]
>>> In addition, there is no reason to keep expired timer tick deferred, so
>>> just perform the tick from ptimer_get_count().
>>>
> [cut]
>
> I noticed an issue here... The problem is that device reset invokes
> ptimer_stop() that invokes ptimer_get_count() that might cause the tick after
> reset, i.e. bogus bh would be invoked after QEMU reset. I have reproduced that
> issue.
>
> The solution might be to introduce ptimer_reset() that would stop QEMU timer and
> reset delta/load/period without invoking ptimer_get_count. And of course all
> devices should be updated to use new ptimer_reset() prior to "Perform tick and
> counter wrap around if timer already expired" patch, but that's not an issue I
> suppose. Please let me know if you have any objections, I'm leaning to do it in
> V11.
>
> void ptimer_reset(ptimer_state *s)
> {
>      timer_del(s->timer);
>      s->enabled = 0;
>      s->period_frac = 0;
>      s->period = 0;
>      s->delta = 0;
>      s->limit = 0;
> }
>

However, ptimer_stop() still would be broken. So maybe it's better to keep tick 
deferred.

-- 
Dmitry