From: Peng Fan <[email protected]>
For ARM64 platform such as i.MX8ULP which has ARMv8 generic timer as sched
clock, which is much faster compared with tpm sched clock. Reading the
tpm count register in i.MX8ULP requires about 290ns, this is slow and
introduce scheduler latency. So exclude tpm sched clock for ARM64
platform.
Signed-off-by: Peng Fan <[email protected]>
---
drivers/clocksource/timer-imx-tpm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/clocksource/timer-imx-tpm.c b/drivers/clocksource/timer-imx-tpm.c
index 2cdc077a39f5..2bd530b9adc4 100644
--- a/drivers/clocksource/timer-imx-tpm.c
+++ b/drivers/clocksource/timer-imx-tpm.c
@@ -150,10 +150,10 @@ static int __init tpm_clocksource_init(void)
tpm_delay_timer.read_current_timer = &tpm_read_current_timer;
tpm_delay_timer.freq = timer_of_rate(&to_tpm) >> 3;
register_current_timer_delay(&tpm_delay_timer);
-#endif
sched_clock_register(tpm_read_sched_clock, counter_width,
timer_of_rate(&to_tpm) >> 3);
+#endif
return clocksource_mmio_init(timer_base + TPM_CNT,
"imx-tpm",
--
2.25.1
> Subject: [PATCH] clocksource/drivers/imx-tpm: exclude sched clock for
> ARM64
Gentle ping..
Thanks,
Peng.
>
> From: Peng Fan <[email protected]>
>
> For ARM64 platform such as i.MX8ULP which has ARMv8 generic timer as
> sched clock, which is much faster compared with tpm sched clock. Reading
> the tpm count register in i.MX8ULP requires about 290ns, this is slow and
> introduce scheduler latency. So exclude tpm sched clock for ARM64 platform.
>
> Signed-off-by: Peng Fan <[email protected]>
> ---
> drivers/clocksource/timer-imx-tpm.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/clocksource/timer-imx-tpm.c
> b/drivers/clocksource/timer-imx-tpm.c
> index 2cdc077a39f5..2bd530b9adc4 100644
> --- a/drivers/clocksource/timer-imx-tpm.c
> +++ b/drivers/clocksource/timer-imx-tpm.c
> @@ -150,10 +150,10 @@ static int __init tpm_clocksource_init(void)
> tpm_delay_timer.read_current_timer = &tpm_read_current_timer;
> tpm_delay_timer.freq = timer_of_rate(&to_tpm) >> 3;
> register_current_timer_delay(&tpm_delay_timer);
> -#endif
>
> sched_clock_register(tpm_read_sched_clock, counter_width,
> timer_of_rate(&to_tpm) >> 3);
> +#endif
>
> return clocksource_mmio_init(timer_base + TPM_CNT,
> "imx-tpm",
> --
> 2.25.1
The following commit has been merged into the timers/core branch of tip:
Commit-ID: 10720e120e2b1d66172dccb06eb4f346a665cca6
Gitweb: https://git.kernel.org/tip/10720e120e2b1d66172dccb06eb4f346a665cca6
Author: Peng Fan <[email protected]>
AuthorDate: Wed, 05 Jan 2022 20:43:04 +08:00
Committer: Daniel Lezcano <[email protected]>
CommitterDate: Wed, 16 Feb 2022 14:43:30 +01:00
clocksource/drivers/imx-tpm: Exclude sched clock for ARM64
For ARM64 platform such as i.MX8ULP which has ARMv8 generic timer as sched
clock, which is much faster compared with tpm sched clock. Reading the
tpm count register in i.MX8ULP requires about 290ns, this is slow and
introduce scheduler latency. So exclude tpm sched clock for ARM64
platform.
Signed-off-by: Peng Fan <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Daniel Lezcano <[email protected]>
---
drivers/clocksource/timer-imx-tpm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/clocksource/timer-imx-tpm.c b/drivers/clocksource/timer-imx-tpm.c
index df80641..60cefc2 100644
--- a/drivers/clocksource/timer-imx-tpm.c
+++ b/drivers/clocksource/timer-imx-tpm.c
@@ -150,10 +150,10 @@ static int __init tpm_clocksource_init(void)
tpm_delay_timer.read_current_timer = &tpm_read_current_timer;
tpm_delay_timer.freq = timer_of_rate(&to_tpm) >> 3;
register_current_timer_delay(&tpm_delay_timer);
-#endif
sched_clock_register(tpm_read_sched_clock, counter_width,
timer_of_rate(&to_tpm) >> 3);
+#endif
return clocksource_mmio_init(timer_base + TPM_CNT,
"imx-tpm",