[PATCH v2 1/3] cppc_cpufreq: Return desired perf in ->get() if feedback counters are 0
Jie Zhan
zhanjie9 at hisilicon.com
Thu Sep 12 00:22:29 PDT 2024
The CPPC performance feedback counters could return 0 when the target cpu
is in a deep idle state, e.g. powered off, and those counters are not
powered. In this case, cppc_cpufreq_get_rate() returns 0, and hence,
cpufreq_online() gets a false error and doesn't generate a cpufreq policy,
which happens in cpufreq_add_dev() when a new cpu device is added.
Don't take it as an error and return the frequency corresponding to the
desired perf when the feedback counters are 0.
Fixes: 6a4fec4f6d30 ("cpufreq: cppc: cppc_cpufreq_get_rate() returns zero in all error cases.")
Signed-off-by: Jie Zhan <zhanjie9 at hisilicon.com>
Reviewed-by: Zeng Heng <zengheng4 at huawei.com>
---
drivers/cpufreq/cppc_cpufreq.c | 19 +++++++++++++++++--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/drivers/cpufreq/cppc_cpufreq.c b/drivers/cpufreq/cppc_cpufreq.c
index bafa32dd375d..6aa3af56924b 100644
--- a/drivers/cpufreq/cppc_cpufreq.c
+++ b/drivers/cpufreq/cppc_cpufreq.c
@@ -748,18 +748,33 @@ static unsigned int cppc_cpufreq_get_rate(unsigned int cpu)
ret = cppc_get_perf_ctrs(cpu, &fb_ctrs_t0);
if (ret)
- return 0;
+ goto out_err;
udelay(2); /* 2usec delay between sampling */
ret = cppc_get_perf_ctrs(cpu, &fb_ctrs_t1);
if (ret)
- return 0;
+ goto out_err;
delivered_perf = cppc_perf_from_fbctrs(cpu_data, &fb_ctrs_t0,
&fb_ctrs_t1);
return cppc_perf_to_khz(&cpu_data->perf_caps, delivered_perf);
+
+out_err:
+ /*
+ * Feedback counters could be 0 when cores are powered down.
+ * Take desired perf for reflecting frequency in this case.
+ */
+ if (ret == -EFAULT) {
+ ret = cppc_get_desired_perf(cpu, &delivered_perf);
+ if (ret)
+ return 0;
+
+ return cppc_perf_to_khz(&cpu_data->perf_caps, delivered_perf);
+ }
+
+ return 0;
}
static int cppc_cpufreq_set_boost(struct cpufreq_policy *policy, int state)
--
2.33.0
More information about the linux-arm-kernel
mailing list