On Mon, Jun 20, 2022 at 10:40 AM Chee, Tien Fong tien.fong.chee@intel.com wrote:
Hi,
-----Original Message----- From: Paweł Anikiel pan@semihalf.com Sent: Friday, 17 June, 2022 6:47 PM To: Vasut, Marek marex@denx.de; simon.k.r.goldschmidt@gmail.com; Chee, Tien Fong tien.fong.chee@intel.com; michal.simek@xilinx.com Cc: u-boot@lists.denx.de; sjg@chromium.org; festevam@denx.de; jagan@amarulasolutions.com; andre.przywara@arm.com; Armstrong, Neil narmstrong@baylibre.com; pbrobinson@gmail.com; tharvey@gateworks.com; paul.liu@linaro.org; christianshewitt@gmail.com; adrian.fiergolski@fastree3d.com; marek.behun@nic.cz; Denk, Wolfgang wd@denx.de; Lim, Elly Siew Chin elly.siew.chin.lim@intel.com; upstream@semihalf.com; amstan@chromium.org; Paweł Anikiel pan@semihalf.com Subject: [PATCH v3 08/11] socfpga: arria10: Replace delays with busy waiting in cm_full_cfg
Using udelay while the clocks aren't fully configured causes the timer system to save the wrong clock rate. Use sdelay and wait_on_value instead (the values used in these functions were found experimentally).
Signed-off-by: Paweł Anikiel pan@semihalf.com
arch/arm/mach-socfpga/clock_manager_arria10.c | 31 +++++++++++++-----
1 file changed, 22 insertions(+), 9 deletions(-)
diff --git a/arch/arm/mach-socfpga/clock_manager_arria10.c b/arch/arm/mach-socfpga/clock_manager_arria10.c index 58d5d3fd8a..b48a2b47bc 100644 --- a/arch/arm/mach-socfpga/clock_manager_arria10.c +++ b/arch/arm/mach-socfpga/clock_manager_arria10.c
Did you try to call timer_init() after cm_basic_init() in board_init_f? If that's working, then no change is required to fix this clock issue.
Seems like timer_init() isn't implemented on Arria 10 (it defaults to the return 0 stub). I also tried dm_timer_init(), no luck.
I did some code digging, the clock rate is read by clk_get_rate(), and the timer rate is set by dw_apb_timer_probe() (drivers/timer/dw-apb-timer.c:77), and there doesn't seem to be a good way of updating that value later.
The only other function I could find that sets the timer rate is timer_pre_probe() from drivers/timer/timer-uclass.c, which very much looks like what we need, but it's static and the name suggests it shouldn't be called manually anyway.
Regards, Paweł