Jump to content

Recommended Posts

Posted (edited)

I want to share some information about an issue that I see on my OPi5+. Since I got it I am messing around with it and everything works great so far with the edge 6.7 kernel.

One thing I noticed though was, that right after installation with nothing connected and when idle, there was one kernel process that was constantly consuming around 10% CPU time and the load stays at 1

 

The process was irq/136-hym8563. The number varies of course after each boot. Kernel driver is rtc_hym8563. Since it is build into the kernel I was not able to unload it as a first test. Side node, the issue is not seen on the Rock5b, so it seems OPi5+ hardware specific. Even if both are relative similar.

 

I have not found anything useful so far on the internet. Don't know if it is just my device or if others are seeing a similar thing on their OPi5+ with the 6.7 edge kernel.

 

IRQs going up like crazy for a few interrupts, compared to when driver rtc-hym8563 is not there.

arch_timer
rk_timer
fec80000.i2c
hym8563

 

Here are some debugging information.

 

top output.

root@orangepi5-plus ~# top -b -n 4 -p (pgrep hym)

top - 16:30:43 up 3 min,  1 user,  load average: 1.08, 0.63, 0.27
Tasks:   1 total,   0 running,   1 sleeping,   0 stopped,   0 zombie
%Cpu(s):  0.0 us,  1.2 sy,  0.0 ni, 97.6 id,  0.0 wa,  1.2 hi,  0.0 si,  0.0 st
MiB Mem :  15718.1 total,  15299.5 free,    422.6 used,    212.1 buff/cache
MiB Swap:   7859.0 total,   7859.0 free,      0.0 used.  15295.5 avail Mem

    PID USER      PR  NI    VIRT    RES    SHR S  %CPU  %MEM     TIME+ COMMAND
    438 root     -51   0       0      0      0 D   0.0   0.0   0:16.90 irq/136-hym8563

top - 16:30:46 up 3 min,  1 user,  load average: 1.08, 0.63, 0.27
Tasks:   1 total,   1 running,   0 sleeping,   0 stopped,   0 zombie
%Cpu(s):  0.0 us,  0.7 sy,  0.0 ni, 97.9 id,  0.0 wa,  1.3 hi,  0.0 si,  0.0 st
MiB Mem :  15718.1 total,  15299.5 free,    422.6 used,    212.1 buff/cache
MiB Swap:   7859.0 total,   7859.0 free,      0.0 used.  15295.5 avail Mem

    PID USER      PR  NI    VIRT    RES    SHR S  %CPU  %MEM     TIME+ COMMAND
    438 root     -51   0       0      0      0 R   7.7   0.0   0:17.13 irq/136-hym8563

top - 16:30:49 up 3 min,  1 user,  load average: 1.07, 0.63, 0.27
Tasks:   1 total,   0 running,   1 sleeping,   0 stopped,   0 zombie
%Cpu(s):  0.0 us,  0.7 sy,  0.0 ni, 97.9 id,  0.0 wa,  1.3 hi,  0.0 si,  0.0 st
MiB Mem :  15718.1 total,  15299.5 free,    422.6 used,    212.1 buff/cache
MiB Swap:   7859.0 total,   7859.0 free,      0.0 used.  15295.5 avail Mem

    PID USER      PR  NI    VIRT    RES    SHR S  %CPU  %MEM     TIME+ COMMAND
    438 root     -51   0       0      0      0 D   8.0   0.0   0:17.37 irq/136-hym8563

top - 16:30:52 up 4 min,  1 user,  load average: 0.99, 0.62, 0.27
Tasks:   1 total,   0 running,   1 sleeping,   0 stopped,   0 zombie
%Cpu(s):  0.0 us,  0.8 sy,  0.0 ni, 97.9 id,  0.0 wa,  1.3 hi,  0.0 si,  0.0 st
MiB Mem :  15718.1 total,  15299.5 free,    422.6 used,    212.1 buff/cache
MiB Swap:   7859.0 total,   7859.0 free,      0.0 used.  15295.5 avail Mem

    PID USER      PR  NI    VIRT    RES    SHR S  %CPU  %MEM     TIME+ COMMAND
    438 root     -51   0       0      0      0 D   8.0   0.0   0:17.61 irq/136-hym8563

 

Interrupt increase over 1 minute. 100k for hym8563 and 1million for fec8000.i2c

root@orangepi5-plus ~# uptime && date && cat /proc/interrupts | grep -iE "(hym|fec80000|_timer)" && sleep 60 && date && cat /proc/interrupts | grep -iE "(hym|fec80000|_timer)"

 17:28:15 up 35 min,  1 user,  load average: 1.01, 1.07, 1.00
Mon Jan 22 05:28:15 PM CET 2024
 19:      26038     125123      73155     438814      46260      13649       3574     299665     GICv3  26 Level     arch_timer
 25:      22900      83767      54823     279426      36468      15155      10963       9513     GICv3 321 Level     rk_timer
 48:          0          0          0          0          0          0          0   30250774     GICv3 355 Level     fec80000.i2c
136:          0          0          0          0          0          0          0    5041747  rockchip_gpio_irq   8 Level     hym8563
Mon Jan 22 05:29:15 PM CET 2024
 19:      26366     125128      91142     438815      46849      13652       3574     307458     GICv3  26 Level     arch_timer
 25:      23366      84133      67225     279722      37149      15171      10976       9752     GICv3 321 Level     rk_timer
 48:          0          0          0          0          0          0          0   31073437     GICv3 355 Level     fec80000.i2c
136:          0          0          0          0          0          0          0    5178856  rockchip_gpio_irq   8 Level     hym8563

 

irq cpu time is also visible in mpstat 

root@orangepi5-plus ~# mpstat -P ALL

Linux 6.7.1-edge-rockchip-rk3588 (orangepi5-plus) 	01/22/2024 	_aarch64_	(8 CPU)

07:43:50 PM  CPU    %usr   %nice    %sys %iowait    %irq   %soft  %steal  %guest  %gnice   %idle
07:43:50 PM  all    0.13    0.00    0.68    0.01    1.33    0.03    0.00    0.00    0.00   97.82
07:43:50 PM    0    0.03    0.00    1.30    0.01    0.98    0.08    0.00    0.00    0.00   97.61
07:43:50 PM    1    0.03    0.00    0.26    0.01    0.18    0.02    0.00    0.00    0.00   99.51
07:43:50 PM    2    0.03    0.00    3.06    0.01    2.46    0.07    0.00    0.00    0.00   94.37
07:43:50 PM    3    0.02    0.00    0.77    0.01    0.58    0.02    0.00    0.00    0.00   98.60
07:43:50 PM    4    0.59    0.00    0.11    0.01    0.03    0.02    0.00    0.00    0.00   99.24
07:43:50 PM    5    0.22    0.00    0.05    0.01    0.01    0.01    0.00    0.00    0.00   99.70
07:43:50 PM    6    0.06    0.00    0.01    0.01    0.01    0.00    0.00    0.00    0.00   99.91
07:43:50 PM    7    0.04    0.00    0.02    0.01    6.49    0.03    0.00    0.00    0.00   93.42

 

procinfo output shows as well the high number for irq 136 and 48.

root@orangepi5-plus ~# procinfo

Memory:        Total        Used        Free     Buffers
RAM:        16095284      585248    15510036       22584
Swap:        8047640           0     8047640

Bootup: Mon Jan 22 16:53:14 2024   Load average: 1.29 1.09 1.02 1/231 9335

user  :   00:01:51.76   0.1%  page in :           238872
nice  :   00:00:00.03   0.0%  page out:           118633
system:   00:09:19.33   0.7%  page act:            59088
IOwait:   00:00:09.19   0.0%  page dea:                0
hw irq:   00:18:07.58   1.3%  page flt:          1335085
sw irq:   00:00:24.98   0.0%  swap in :                0
idle  :   22:09:57.44  97.8%  swap out:                0
uptime:   02:51:46.79         context :        142704160

irq  18:          0                      irq  66:          0  427819017 Edge ae
irq  19:    4672570                      irq  77:          0  570425352 Edge PC
irq  20:          0                      irq  78:          0  570425353 Edge ae
irq  23:          0                      irq  89:          0  285212680 Edge PC
irq  25:    2504720                      irq  90:          0  285212681 Edge ae
irq  31:          0                      irq  91:          2
irq  32:         15                      irq  92:          0
irq  33:          1                      irq  93:         32
irq  34:          0                      irq  94:          0
irq  35:          0                      irq  95:      75881  428343296 Edge en
irq  36:          0                      irq  96:          0  570949632 Edge en
irq  37:          0                      irq  97:          0
irq  38:          0                      irq  98:          0
irq  39:          0                      irq  99:          0
irq  40:          0                      irq 100:          0
irq  41:          2                      irq 101:        514
irq  43:     653459                      irq 102:          0
irq  44:       2278                      irq 119:          0
irq  45:       1346                      irq 130:          0  8 Edge PCIe PME
irq  46:        203                      irq 131:          0  9 Edge aerdrv
irq  47:      41964                      irq 132:      99660
irq  48:  141047548                      irq 133:      56265
irq  49:        108                      irq 135:          0
irq  50:          0                      irq 136:   23507838
irq  51:          0                      irq 137:        324
irq  52:          2                      irq 138:          0
irq  53:          0                      irq 139:          0
irq  54:          0                      irq 140:          0
irq  65:          0  427819016 Edge PC

 

Since I do not need the RTC on my system, my idea was to create an own custom image, where rtc_hym8563 is build as a module. Where I can disable the module in the hope the process is gone and nothing is consuming cpu time when idle.

It turns out that worked for me and I am running it like this for now. 

 

Here are my steps.

 

Change kernel config in the build system.

 

# config/kernel/linux-rockchip-rk3588-edge.config
CONFIG_RTC_DRV_HYM8563=y

 

After new image is installed and booted, it is indeed now a module.

 

root@orangepi5-plus ~# zcat /proc/config.gz | grep -i hym
CONFIG_RTC_DRV_HYM8563=m

 

Disabling the rtc_hym8563 driver

echo "blacklist rtc_hym8563" > /etc/modprobe.d/blacklist-rtc.conf
update-initramfs -u
reboot

 

 

I just wanted to share this information in case someone else see this. 

For now I will live with my workaround and keep an eye on the kernel development to see, if there will be a change.

 

In case there is some hardware expert and has some other tips, please feel free to share.

 

 

Cheers,

Marco

 

 

 

Edited by Marco Schirrmeister
Posted

After some more searching, I found this patch, which looked to me interesting. At least from the little description https://patchwork.kernel.org/project/linux-rockchip/patch/20231202071212.1606800-1-CFSworks@gmail.com/

The RK1 module is also RK3588 based and has a HYM8563 chip as well.

 

No idea what this is or what is different on the various boards and their components. But I thought I give it a try. 

I implemented the hym8563 patch line on the edge-6.7.1 version to see what happens.

 

Result is, it made a change on the OPi5+. No more high cpu usage or heavy interrupt increases with the change from none to up, when the hym8563 driver is loaded.

Since I am not a hardware guy, I have no idea what this is or what it does. If it is good or bad or whatever. 

 

Posted

@Marco Schirrmeister Thank you for your search and pointing towards the fix. I applied the same fix to 6.8.0-rc1-edge kernel and I noticed the same result!

 

 

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: John Doe <john.doe@somewhere.on.planet>
Date: Sun, 11 Feb 2024 02:19:42 +0000
Subject: Patching kernel rockchip-rk3588 files
 arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts to fix hym8563-int
 issue.

Signed-off-by: John Doe <john.doe@somewhere.on.planet>
---
 arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts b/arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts
index 88bfce623..844992180 100644
--- a/arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts
+++ b/arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts
@@ -462,11 +462,11 @@ &pcie3x4 {
 };

 &pinctrl {
        hym8563 {
                hym8563_int: hym8563-int {
-                       rockchip,pins = <0 RK_PB0 RK_FUNC_GPIO &pcfg_pull_none>;
+                       rockchip,pins = <0 RK_PB0 RK_FUNC_GPIO &pcfg_pull_up>;
                };
        };

        leds {
                blue_led_pin: blue-led {
--
Created with Armbian build tools https://github.com/armbian/build
                  

 

Posted (edited)

Thank you for the confirmation @Vijay Gill.

 

Here is the patch I use. I did also add the poweroff support. Splitting it into multiple files might be better, but it is good enough for my test builds.

 

root@dumpster /m/t/t/n/build (main)# cat userpatches/kernel/rockchip-rk3588-edge/1000-arm64-dts-fix-rtc-add-poweroff-support-Orange-Pi-5-Plus.patch

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: John Doe <john.doe@somewhere.on.planet>
Date: Mon, 29 Jan 2024 12:51:13 +0100
Subject: Patching kernel rockchip-rk3588 files
 arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts

Signed-off-by: John Doe <john.doe@somewhere.on.planet>
---
 arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts b/arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts
index 88bfce6237db..70cc6bd5a0af 100644
--- a/arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts
+++ b/arch/arm64/boot/dts/rockchip/rk3588-orangepi-5-plus.dts
@@ -462,11 +462,11 @@ &pcie3x4 {
 };

 &pinctrl {
 	hym8563 {
 		hym8563_int: hym8563-int {
-			rockchip,pins = <0 RK_PB0 RK_FUNC_GPIO &pcfg_pull_none>;
+			rockchip,pins = <0 RK_PB0 RK_FUNC_GPIO &pcfg_pull_up>;
 		};
 	};

 	leds {
 		blue_led_pin: blue-led {
@@ -572,10 +572,12 @@ pmic@0 {
 		pinctrl-names = "default";
 		pinctrl-0 = <&pmic_pins>, <&rk806_dvs1_null>,
 			    <&rk806_dvs2_null>, <&rk806_dvs3_null>;
 		spi-max-frequency = <1000000>;

+		system-power-controller;
+
 		vcc1-supply = <&vcc5v0_sys>;
 		vcc2-supply = <&vcc5v0_sys>;
 		vcc3-supply = <&vcc5v0_sys>;
 		vcc4-supply = <&vcc5v0_sys>;
 		vcc5-supply = <&vcc5v0_sys>;
@@ -592,11 +594,11 @@ pmic@0 {

 		gpio-controller;
 		#gpio-cells = <2>;

 		rk806_dvs1_null: dvs1-null-pins {
-			pins = "gpio_pwrctrl2";
+			pins = "gpio_pwrctrl1";
 			function = "pin_fun0";
 		};

 		rk806_dvs2_null: dvs2-null-pins {
 			pins = "gpio_pwrctrl2";
--
Created with Armbian build tools https://github.com/armbian/build

 

Edited by Marco Schirrmeister
Formatting
Posted

You're doing great :)

Draft PR usually means it is WIP or the author implies it shall not be merged (yet).

So once you think you are done with your adjustments mark it ready for review. It is still possible to do adjustments after that.

 

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...

Important Information

Terms of Use - Privacy Policy - Guidelines