This commit doesn't add any supported events, but provides the framework
for adding them. We store the pm_event structs in a simple array, and
provide the mapping from the event numbers to array indexes in the
supported_event_map array. Because the value of PMCEID[01] depends upon
which events are supported at runtime, generate it dynamically.

Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org>
---
 target/arm/cpu.c    | 20 +++++++++++++-------
 target/arm/cpu.h    | 10 ++++++++++
 target/arm/cpu64.c  |  4 ----
 target/arm/helper.c | 42 ++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 65 insertions(+), 11 deletions(-)

diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index 9e54c56379..d1c766d180 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -957,9 +957,19 @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp)
     if (!cpu->has_pmu) {
         unset_feature(env, ARM_FEATURE_PMU);
         cpu->id_aa64dfr0 &= ~0xf00;
-    } else if (!kvm_enabled()) {
-        arm_register_pre_el_change_hook(cpu, &pmu_pre_el_change, 0);
-        arm_register_el_change_hook(cpu, &pmu_post_el_change, 0);
+    }
+    if (arm_feature(env, ARM_FEATURE_PMU)) {
+        uint64_t pmceid = get_pmceid(&cpu->env);
+        cpu->pmceid0 = extract64(pmceid, 0, 32);
+        cpu->pmceid1 = extract64(pmceid, 32, 32);
+
+        if (!kvm_enabled()) {
+            arm_register_pre_el_change_hook(cpu, &pmu_pre_el_change, 0);
+            arm_register_el_change_hook(cpu, &pmu_post_el_change, 0);
+        }
+    } else {
+        cpu->pmceid0 = 0x00000000;
+        cpu->pmceid1 = 0x00000000;
     }
 
     if (!arm_feature(env, ARM_FEATURE_EL2)) {
@@ -1601,8 +1611,6 @@ static void cortex_a7_initfn(Object *obj)
     cpu->id_pfr0 = 0x00001131;
     cpu->id_pfr1 = 0x00011011;
     cpu->id_dfr0 = 0x02010555;
-    cpu->pmceid0 = 0x00000000;
-    cpu->pmceid1 = 0x00000000;
     cpu->id_afr0 = 0x00000000;
     cpu->id_mmfr0 = 0x10101105;
     cpu->id_mmfr1 = 0x40000000;
@@ -1647,8 +1655,6 @@ static void cortex_a15_initfn(Object *obj)
     cpu->id_pfr0 = 0x00001131;
     cpu->id_pfr1 = 0x00011011;
     cpu->id_dfr0 = 0x02010555;
-    cpu->pmceid0 = 0x0000000;
-    cpu->pmceid1 = 0x00000000;
     cpu->id_afr0 = 0x00000000;
     cpu->id_mmfr0 = 0x10201105;
     cpu->id_mmfr1 = 0x20000000;
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 92282cd976..f991ff370e 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -991,6 +991,16 @@ void pmu_op_finish(CPUARMState *env);
 void pmu_pre_el_change(ARMCPU *cpu, void *ignored);
 void pmu_post_el_change(ARMCPU *cpu, void *ignored);
 
+/*
+ * get_pmceid
+ * @env: CPUARMState
+ *
+ * Return the PMCEID[01] register values corresponding to the counters which
+ * are supported given the current configuration (0 is low 32, 1 is high 32
+ * bits)
+ */
+uint64_t get_pmceid(CPUARMState *env);
+
 /* SCTLR bit meanings. Several bits have been reused in newer
  * versions of the architecture; in that case we define constants
  * for both old and new bit meanings. Code which tests against those
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
index 873f059bf2..a1aad772fa 100644
--- a/target/arm/cpu64.c
+++ b/target/arm/cpu64.c
@@ -138,8 +138,6 @@ static void aarch64_a57_initfn(Object *obj)
     cpu->isar.id_isar6 = 0;
     cpu->isar.id_aa64pfr0 = 0x00002222;
     cpu->id_aa64dfr0 = 0x10305106;
-    cpu->pmceid0 = 0x00000000;
-    cpu->pmceid1 = 0x00000000;
     cpu->isar.id_aa64isar0 = 0x00011120;
     cpu->id_aa64mmfr0 = 0x00001124;
     cpu->dbgdidr = 0x3516d000;
@@ -246,8 +244,6 @@ static void aarch64_a72_initfn(Object *obj)
     cpu->isar.id_isar5 = 0x00011121;
     cpu->isar.id_aa64pfr0 = 0x00002222;
     cpu->id_aa64dfr0 = 0x10305106;
-    cpu->pmceid0 = 0x00000000;
-    cpu->pmceid1 = 0x00000000;
     cpu->isar.id_aa64isar0 = 0x00011120;
     cpu->id_aa64mmfr0 = 0x00001124;
     cpu->dbgdidr = 0x3516d000;
diff --git a/target/arm/helper.c b/target/arm/helper.c
index 6724d97346..b9d8441497 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -1009,6 +1009,48 @@ static inline uint64_t pmu_counter_mask(CPUARMState *env)
   return (1 << 31) | ((1 << pmu_num_counters(env)) - 1);
 }
 
+typedef struct pm_event {
+    uint16_t number; /* PMEVTYPER.evtCount is 16 bits wide */
+    /* If the event is supported on this CPU (used to generate PMCEID[01]) */
+    bool (*supported)(CPUARMState *);
+    /*
+     * Retrieve the current count of the underlying event. The programmed
+     * counters hold a difference from the return value from this function
+     */
+    uint64_t (*get_count)(CPUARMState *);
+} pm_event;
+
+static const pm_event pm_events[] = {
+};
+#define MAX_EVENT_ID 0x0
+#define UNSUPPORTED_EVENT UINT16_MAX
+static uint16_t supported_event_map[MAX_EVENT_ID + 1];
+
+/*
+ * Called upon initialization to build PMCEID0 (low 32 bits) and PMCEID1 (high
+ * 32). We also use it to build a map of ARM event numbers to indices in
+ * our pm_events array.
+ */
+uint64_t get_pmceid(CPUARMState *env)
+{
+    uint64_t pmceid = 0;
+    unsigned int i;
+
+    for (i = 0; i < ARRAY_SIZE(supported_event_map); i++) {
+        supported_event_map[i] = UNSUPPORTED_EVENT;
+    }
+
+    for (i = 0; i < ARRAY_SIZE(pm_events); i++) {
+        const pm_event *cnt = &pm_events[i];
+        assert(cnt->number <= MAX_EVENT_ID);
+        if (cnt->supported(env)) {
+            pmceid |= (1 << cnt->number);
+            supported_event_map[cnt->number] = i;
+        }
+    }
+    return pmceid;
+}
+
 static CPAccessResult pmreg_access(CPUARMState *env, const ARMCPRegInfo *ri,
                                    bool isread)
 {
-- 
2.19.1

On 5 November 2018 at 18:51, Aaron Lindsay <aaron@os.amperecomputing.com> wrote:
> This commit doesn't add any supported events, but provides the framework
> for adding them. We store the pm_event structs in a simple array, and
> provide the mapping from the event numbers to array indexes in the
> supported_event_map array. Because the value of PMCEID[01] depends upon
> which events are supported at runtime, generate it dynamically.
>
> Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org>
> ---
>  target/arm/cpu.c    | 20 +++++++++++++-------
>  target/arm/cpu.h    | 10 ++++++++++
>  target/arm/cpu64.c  |  4 ----
>  target/arm/helper.c | 42 ++++++++++++++++++++++++++++++++++++++++++
>  4 files changed, 65 insertions(+), 11 deletions(-)
>
> diff --git a/target/arm/cpu.c b/target/arm/cpu.c
> index 9e54c56379..d1c766d180 100644
> --- a/target/arm/cpu.c
> +++ b/target/arm/cpu.c
> @@ -957,9 +957,19 @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp)
>      if (!cpu->has_pmu) {
>          unset_feature(env, ARM_FEATURE_PMU);
>          cpu->id_aa64dfr0 &= ~0xf00;
> -    } else if (!kvm_enabled()) {
> -        arm_register_pre_el_change_hook(cpu, &pmu_pre_el_change, 0);
> -        arm_register_el_change_hook(cpu, &pmu_post_el_change, 0);
> +    }
> +    if (arm_feature(env, ARM_FEATURE_PMU)) {
> +        uint64_t pmceid = get_pmceid(&cpu->env);
> +        cpu->pmceid0 = extract64(pmceid, 0, 32);
> +        cpu->pmceid1 = extract64(pmceid, 32, 32);
> +
> +        if (!kvm_enabled()) {
> +            arm_register_pre_el_change_hook(cpu, &pmu_pre_el_change, 0);
> +            arm_register_el_change_hook(cpu, &pmu_post_el_change, 0);
> +        }
> +    } else {
> +        cpu->pmceid0 = 0x00000000;
> +        cpu->pmceid1 = 0x00000000;
>      }

This now sets one of the ID registers for "we have no
PMU" in the first "if (!cpu->has_pmu)" statement (id_aa64dfr0),
and some of them (pmceid0, pcmeid1) in this else clause at the
end. We should put them all in the same place.

>
>      if (!arm_feature(env, ARM_FEATURE_EL2)) {
> @@ -1601,8 +1611,6 @@ static void cortex_a7_initfn(Object *obj)
>      cpu->id_pfr0 = 0x00001131;
>      cpu->id_pfr1 = 0x00011011;
>      cpu->id_dfr0 = 0x02010555;
> -    cpu->pmceid0 = 0x00000000;
> -    cpu->pmceid1 = 0x00000000;
>      cpu->id_afr0 = 0x00000000;
>      cpu->id_mmfr0 = 0x10101105;
>      cpu->id_mmfr1 = 0x40000000;
> @@ -1647,8 +1655,6 @@ static void cortex_a15_initfn(Object *obj)
>      cpu->id_pfr0 = 0x00001131;
>      cpu->id_pfr1 = 0x00011011;
>      cpu->id_dfr0 = 0x02010555;
> -    cpu->pmceid0 = 0x0000000;
> -    cpu->pmceid1 = 0x00000000;
>      cpu->id_afr0 = 0x00000000;
>      cpu->id_mmfr0 = 0x10201105;
>      cpu->id_mmfr1 = 0x20000000;
> diff --git a/target/arm/cpu.h b/target/arm/cpu.h
> index 92282cd976..f991ff370e 100644
> --- a/target/arm/cpu.h
> +++ b/target/arm/cpu.h
> @@ -991,6 +991,16 @@ void pmu_op_finish(CPUARMState *env);
>  void pmu_pre_el_change(ARMCPU *cpu, void *ignored);
>  void pmu_post_el_change(ARMCPU *cpu, void *ignored);
>
> +/*
> + * get_pmceid
> + * @env: CPUARMState
> + *
> + * Return the PMCEID[01] register values corresponding to the counters which
> + * are supported given the current configuration (0 is low 32, 1 is high 32
> + * bits)
> + */
> +uint64_t get_pmceid(CPUARMState *env);

This doesn't look like the right API, because in AArch64
PMCEID0_EL0 and PMCEID1_EL0 are both 64-bit registers,
so you can't fit them both into a single uint64_t.

thanks
-- PMM

On Nov 16 15:06, Peter Maydell wrote:
> On 5 November 2018 at 18:51, Aaron Lindsay <aaron@os.amperecomputing.com> wrote:
> > --- a/target/arm/cpu.c
> > +++ b/target/arm/cpu.c
> > @@ -957,9 +957,19 @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp)
> >      if (!cpu->has_pmu) {
> >          unset_feature(env, ARM_FEATURE_PMU);
> >          cpu->id_aa64dfr0 &= ~0xf00;
> > -    } else if (!kvm_enabled()) {
> > -        arm_register_pre_el_change_hook(cpu, &pmu_pre_el_change, 0);
> > -        arm_register_el_change_hook(cpu, &pmu_post_el_change, 0);
> > +    }
> > +    if (arm_feature(env, ARM_FEATURE_PMU)) {
> > +        uint64_t pmceid = get_pmceid(&cpu->env);
> > +        cpu->pmceid0 = extract64(pmceid, 0, 32);
> > +        cpu->pmceid1 = extract64(pmceid, 32, 32);
> > +
> > +        if (!kvm_enabled()) {
> > +            arm_register_pre_el_change_hook(cpu, &pmu_pre_el_change, 0);
> > +            arm_register_el_change_hook(cpu, &pmu_post_el_change, 0);
> > +        }
> > +    } else {
> > +        cpu->pmceid0 = 0x00000000;
> > +        cpu->pmceid1 = 0x00000000;
> >      }
> 
> This now sets one of the ID registers for "we have no
> PMU" in the first "if (!cpu->has_pmu)" statement (id_aa64dfr0),
> and some of them (pmceid0, pcmeid1) in this else clause at the
> end. We should put them all in the same place.

I agree that would be cleaner - I'll move the id_aa64dfr0 update to
later else clause.

> > +/*
> > + * get_pmceid
> > + * @env: CPUARMState
> > + *
> > + * Return the PMCEID[01] register values corresponding to the counters which
> > + * are supported given the current configuration (0 is low 32, 1 is high 32
> > + * bits)
> > + */
> > +uint64_t get_pmceid(CPUARMState *env);
> 
> This doesn't look like the right API, because in AArch64
> PMCEID0_EL0 and PMCEID1_EL0 are both 64-bit registers,
> so you can't fit them both into a single uint64_t.

Heh, I think I started working on this long enough ago that I was using
a copy of the ARMv8.0 ARM - before the statistical profiling extensions
were announced. I believe those are the only currently-defined common
events >= 0x4000, so they're the only bits defined in the upper 32 bits
of PMCEID[01].

Now that I look at it, pmceid[01] are currently only defined as
uint32_t, and we don't have PMCEID[23] for the high bits for AArch32.
Perhaps that should be a separate patch.

At any rate, I'll plan to update this the following signature for the
next version, where `which` is [01] for which PMCEID is being requested.
For now I'll probably just put a comment about not supporting the 0x40xx
events, since I don't know that coming up with a sparse array is worth
it, but the signature will be appropriate if we add support later:

uint64_t get_pmceid(CPUARMState *env, unsigned which);

-Aaron