Re: [Qemu-devel] [kvm-unit-tests PATCHv7 2/3] arm: pmu: Check cycle count increases

[cov]

Hi Wei,

Thanks for your work on this.

On 2016-11-02 16:22, Wei Huang wrote:
> Ensure that reads of the PMCCNTR_EL0 are monotonically increasing,
> even for the smallest delta of two subsequent reads.
>
> Signed-off-by: Christopher Covington <address@hidden>
> Signed-off-by: Wei Huang <address@hidden>
> ---
>  arm/pmu.c | 100 
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 100 insertions(+)
>
> diff --git a/arm/pmu.c b/arm/pmu.c
> index 42d0ee1..65b7df1 100644
> --- a/arm/pmu.c
> +++ b/arm/pmu.c
> @@ -14,6 +14,9 @@
>   */
>  #include "libcflat.h"
>
> +#define NR_SAMPLES 10
> +#define ARMV8_PMU_CYCLE_IDX 31
> +
>  #if defined(__arm__)
>  static inline uint32_t get_pmcr(void)
>  {
> @@ -22,6 +25,43 @@ static inline uint32_t get_pmcr(void)
>         asm volatile("mrc p15, 0, %0, c9, c12, 0" : "=r" (ret));
>         return ret;
>  }
> +
> +static inline void set_pmcr(uint32_t pmcr)
> +{
> +       asm volatile("mcr p15, 0, %0, c9, c12, 0" : : "r" (pmcr));
> +}
> +
> +static inline void set_pmccfiltr(uint32_t filter)
> +{
> +       uint32_t cycle_idx = ARMV8_PMU_CYCLE_IDX;
> +
> +       asm volatile("mcr p15, 0, %0, c9, c12, 5" : : "r" (cycle_idx));
> +       asm volatile("mcr p15, 0, %0, c9, c13, 1" : : "r" (filter));
> +}

Down the road I'd like to add tests for the regular events. What if you 
added
separate PMSELR and PMXEVTYPER accessors now and used them (with 
PMSELR.SEL = 31)
for setting PMCCFILTR? Then we wouldn't need a specialized set_pmccfiltr
function for the cycle counter versus PMSELR and PMXEVTYPER for the 
regular events.

> +/*
> + * While PMCCNTR can be accessed as a 64 bit coprocessor register, 
> returning 64
> + * bits doesn't seem worth the trouble when differential usage of the 
> result is
> + * expected (with differences that can easily fit in 32 bits). So just 
> return
> + * the lower 32 bits of the cycle count in AArch32.
> + */
> +static inline unsigned long get_pmccntr(void)
> +{
> +       unsigned long cycles;
> +
> +       asm volatile("mrc p15, 0, %0, c9, c13, 0" : "=r" (cycles));
> +       return cycles;
> +}
> +
> +static inline void enable_counter(uint32_t idx)
> +{
> +       asm volatile("mcr p15, 0, %0, c9, c12, 1" : : "r" (1 << idx));
> +}

My personal preference, that I think would make this function look and
act like the other system register accessor functions, would be to
name the function "set_pmcntenset" and do a plain write of the input
parameter without a shift, letting the shift be done in the C code.
(As we scale up, the system register accessor functions should probably
be generated by macros from a concise table.)

> +static inline void disable_counter(uint32_t idx)
> +{
> +       asm volatile("mrc p15, 0, %0, c9, c12, 1" : : "r" (1 << idx));
> +}

This function doesn't seem to be used yet. Consider whether it might
make sense to defer introducing it until there is a user.

>  #elif defined(__aarch64__)
>  static inline uint32_t get_pmcr(void)
>  {
> @@ -30,6 +70,34 @@ static inline uint32_t get_pmcr(void)
>         asm volatile("mrs %0, pmcr_el0" : "=r" (ret));
>         return ret;
>  }
> +
> +static inline void set_pmcr(uint32_t pmcr)
> +{
> +       asm volatile("msr pmcr_el0, %0" : : "r" (pmcr));
> +}
>
> +static inline void set_pmccfiltr(uint32_t filter)
> +{
> +       asm volatile("msr pmccfiltr_el0, %0" : : "r" (filter));
> +}

As above, consider whether using PMSELR and PMXEVTYPER might be a more
reusable pair of accessors.

> +static inline unsigned long get_pmccntr(void)
> +{
> +       unsigned long cycles;
> +
> +       asm volatile("mrs %0, pmccntr_el0" : "=r" (cycles));
> +       return cycles;
> +}
> +
> +static inline void enable_counter(uint32_t idx)
> +{
> +       asm volatile("msr pmcntenset_el0, %0" : : "r" (1 << idx));
> +}

Same thought as above about uniformity and generatability.

> +static inline void disable_counter(uint32_t idx)
> +{
> +       asm volatile("msr pmcntensclr_el0, %0" : : "r" (1 << idx));
> +}

As above, this function doesn't seem to be used yet.

Thanks,
Cov