Raspberry Pi 3 B+ lockups

[nina]

my 3B+ was running fine until i decided to put a heatsink on the cpu (with 3M thermal pad). then i experienced random lockups like the ones described in this topic. no further changes, no clues in syslog.

after two weeks i removed the heatsink, cleaned cpu. running fine now for a week, a little hotter, but no more lockups since.

[jamesh]

my 3B+ was running fine until i decided to put a heatsink on the cpu (with 3M thermal pad). then i experienced random lockups like the ones described in this topic. no further changes, no clues in syslog.

after two weeks i removed the heatsink, cleaned cpu. running fine now for a week, a little hotter, but no more lockups since.

Thats weird!

[jahboater]

my 3B+ was running fine until i decided to put a heatsink on the cpu (with 3M thermal pad). then i experienced random lockups like the ones described in this topic. no further changes, no clues in syslog.

after two weeks i removed the heatsink, cleaned cpu. running fine now for a week, a little hotter, but no more lockups since.

Are you sure it was a proper 3M thermal tape?
There have been instances of cheap heatsinks being supplied with thermal tape that is actually insulating!

I use this heat sink because it is bolted on and therefore can use high quality thermal paste instead of sticky pads or tape.
https://www.amazon.co.uk/Kintaro-Alumin ... spberry+pi
Its the best cooling solution I know for the Pi3+, slightly better than the metal FLIRC case.
It also cools the USB/ethernet chip which can get very got.

[e-raser]

Even I‘m pretty sure it has nothing to do with it in my case (as I described: memory / kswapd issue): I use a FLIRC case.

[nina]

Even I‘m pretty sure it has nothing to do with it in my case (as I described: memory / kswapd issue): I use a FLIRC case.

yes, i think jahboater and you might both be right. a while ago i bought a bunch of heatsinks from aliexpress. if i put some of them together i can read the name of the doublesided tape which seems to be 3M 200MP. and according to google that isn't supposed to be heat conductive.
secondly i failed to mention i also put one of the smaller heatsinks on the usb/ethernet chip. and because i run raspbian from an external ssd it might have gotten too hot.
so thank you both for pointing that out, still running stable, mystery resolved. i'll probably install a small fan because now i'm a little worried about the temperature of the usb chip.

[abhilash7]

I recently started using the RPi 3 B+. Initially I was running LibreELEC on it, but it kept on locking up when streaming content. I noticed that this happens only when the RPi was connected to internet via ethernet, never on WiFi. I always had to hard reset in order to use it again. Switching to OSMC didn't help.

In order to make sure that there was something wrong with my RPi 3 B+, I ran the same installations on my RPi 3, with the same setup (ethernet, power supply, microSD card) and it never froze.

I tried the latest Raspbian on the B+ to make sure I was not missing any crucial kernel/firmware updates, but that didn't help as well. On Raspbian I can always recreate this lockup by initiating download of a large file.

Configuration wise I've tried limiting the ethernet connection speed to 100 Mbps, setting

Code: Select all

dtparam=eee=off

(since I'm using a slightly longer ethenet cable) but it never solved the issue.

I have been lurking on this thread for weeks now and I have read it through and through but I didn't find a solution. Can someone please help me?

[Raspi-ana]

Hello,

my 3B+ unit is acting similarly - a bit longer streaming or file transfer with moderate sized (few MBs) WIRED, a crash will occur. Unit needs to be powered off. Can be reproduced. WIRELESS works OK.

I am using a powerline adapters (TP-Link TL-PA4010P) to distribute wired network through the house, I was wondering if it could cause the crash. Or the raspberry unit is faulty.

[abhilash7]

Hello,

my 3B+ unit is acting similarly - a bit longer streaming or file transfer with moderate sized (few MBs) WIRED, a crash will occur. Unit needs to be powered off. Can be reproduced. WIRELESS works OK.

I am using a powerline adapters (TP-Link TL-PA4010P) to distribute wired network through the house, I was wondering if it could cause the crash. Or the raspberry unit is faulty.

To ensure that your powerline adapters are not faulty, run the 3B+ with a headless raspbian installation directly connected to your router and try downloading a file.

[Raspi-ana]

Could and should try that. In any case, the contrast between wired and wireless functionality is vast - with wired network my RB3B+ unit could sometimes crash even when trying to add an add-on file to LibreElec. I did go the exact same route as you, tried OSMC, same thing....Raspian, LibreElec, OSMC, Batocera etc...whenever wired, unit would freeze totally upon a min/moderate network traffic. Wireless - stable as a calm horse in a stable.

Actually, I could not use noobs to install OSMC or Lakka (via wired) , unit froze every time during the installation process.

This problem is not about the CPU or GPU heating too much. Something else. PoE? (one difference between 3b and 3b+)

My power unit is the official Raspberry one.

[nina]

just to confirm, i tried putting another small heatsink on only the usb/ethernet chip this time and it locked up within 30 minutes. i first suspected radio interference from my phone, but it's probably just overheating because of the isolating 'thermal' pad. removed it and running stable again.

i'm using the official 2.5A power supply, raspbian on a usb drive and the wired ethernet is connected to a gigabit hub.

[abhilash7]

I recently started using the RPi 3 B+. Initially I was running LibreELEC on it, but it kept on locking up when streaming content. I noticed that this happens only when the RPi was connected to internet via ethernet, never on WiFi. I always had to hard reset in order to use it again. Switching to OSMC didn't help.

In order to make sure that there was something wrong with my RPi 3 B+, I ran the same installations on my RPi 3, with the same setup (ethernet, power supply, microSD card) and it never froze.

I tried the latest Raspbian on the B+ to make sure I was not missing any crucial kernel/firmware updates, but that didn't help as well. On Raspbian I can always recreate this lockup by initiating download of a large file.

Configuration wise I've tried limiting the ethernet connection speed to 100 Mbps, setting

Code: Select all

dtparam=eee=off

(since I'm using a slightly longer ethenet cable) but it never solved the issue.

I have been lurking on this thread for weeks now and I have read it through and through but I didn't find a solution. Can someone please help me?

@jamesh or any of the other moderators here can help me (and others) with this issue? I have tried the updates, but it hasn't been fixed yet.

[e-raser]

Hi there,

I´m back with my issue which basically is the same. To remember (details here): kswapd0 is going crazy on a regular basis.

Updates:

• I think I found the trigger for that: a cron job (nextcloud) triggers this behaviour, cause it´s pretty RAM intensive
• What I did some time ago was editing the watchdog settings so it does not kick in and reboot the system ("max-load-1 = 25" changed to "max-load-1 = 60" where I´ve seen a load-1 at around 50 once)
• This way the system is able to survive, but is extreeeeeeeeeeeemly slow and freezes all services. It takes years to complete some memory intensive tasks!

Therefore I´m - again - totall out of ideas cause the initial problem is still there - after more than 6 months!
??? Is it the microSD card which is in the silent process of going to death? FSCK finds and fixes some minor things but nothing remarkable.
??? Is it the CPU throttling?

As it´s for sure memory/SD card related I´ll try two things next:
1) test with another power supply
2) switch the SD card to another one, probably by 1 on 1 cloning it.

Any ideas?

[e-raser]

Update: /var/log/kern.log shows this when the system is on heavy RAM load --> does it indicate SD card problems???

Code: Select all

Mar 20 20:01:16 raspberrypi kernel: [244162.600111] INFO: task kworker/2:2:8183 blocked for more than 120 seconds.
Mar 20 20:01:19 raspberrypi kernel: [244162.600127]       Tainted: G         C        4.19.25-v7+ #1205
Mar 20 20:01:19 raspberrypi kernel: [244162.600133] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Mar 20 20:01:19 raspberrypi kernel: [244162.600141] kworker/2:2     D    0  8183      2 0x00000000
Mar 20 20:01:19 raspberrypi kernel: [244162.600177] Workqueue: events_freezable mmc_rescan
Mar 20 20:01:19 raspberrypi kernel: [244162.600212] [<8082414c>] (__schedule) from [<808247bc>] (schedule+0x50/0xa8)
Mar 20 20:01:19 raspberrypi kernel: [244162.600230] [<808247bc>] (schedule) from [<8067b510>] (__mmc_claim_host+0x120/0x228)
Mar 20 20:01:19 raspberrypi kernel: [244162.600249] [<8067b510>] (__mmc_claim_host) from [<8067b650>] (mmc_get_card+0x38/0x3c)
Mar 20 20:01:19 raspberrypi kernel: [244162.600267] [<8067b650>] (mmc_get_card) from [<806846b8>] (mmc_sd_detect+0x24/0x7c)
Mar 20 20:01:19 raspberrypi kernel: [244162.600287] [<806846b8>] (mmc_sd_detect) from [<8067dc6c>] (mmc_rescan+0x1cc/0x39c)
Mar 20 20:01:19 raspberrypi kernel: [244162.600319] [<8067dc6c>] (mmc_rescan) from [<8013bcf4>] (process_one_work+0x170/0x458)
Mar 20 20:01:19 raspberrypi kernel: [244162.600340] [<8013bcf4>] (process_one_work) from [<8013c038>] (worker_thread+0x5c/0x5a4)
Mar 20 20:01:19 raspberrypi kernel: [244162.600359] [<8013c038>] (worker_thread) from [<80142324>] (kthread+0x138/0x168)
Mar 20 20:01:19 raspberrypi kernel: [244162.600378] [<80142324>] (kthread) from [<801010ac>] (ret_from_fork+0x14/0x28)
Mar 20 20:01:20 raspberrypi kernel: [244162.600386] Exception stack(0xb2ab9fb0 to 0xb2ab9ff8)
Mar 20 20:01:21 raspberrypi kernel: [244162.600397] 9fa0:                                     00000000 00000000 00000000 00000000
Mar 20 20:01:23 raspberrypi kernel: [244162.600410] 9fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
Mar 20 20:01:24 raspberrypi kernel: [244162.600420] 9fe0: 00000000 00000000 00000000 00000000 00000013 00000000
Mar 20 20:03:18 raspberrypi kernel: [244285.480969] INFO: task kworker/2:2:8183 blocked for more than 120 seconds.
Mar 20 20:03:19 raspberrypi kernel: [244285.480977]       Tainted: G         C        4.19.25-v7+ #1205
Mar 20 20:03:19 raspberrypi kernel: [244285.480979] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Mar 20 20:03:19 raspberrypi kernel: [244285.480983] kworker/2:2     D    0  8183      2 0x00000000
Mar 20 20:03:19 raspberrypi kernel: [244285.481003] Workqueue: events_freezable mmc_rescan
Mar 20 20:03:19 raspberrypi kernel: [244285.481022] [<8082414c>] (__schedule) from [<808247bc>] (schedule+0x50/0xa8)
Mar 20 20:03:19 raspberrypi kernel: [244285.481031] [<808247bc>] (schedule) from [<8067b510>] (__mmc_claim_host+0x120/0x228)
Mar 20 20:03:19 raspberrypi kernel: [244285.481040] [<8067b510>] (__mmc_claim_host) from [<8067b650>] (mmc_get_card+0x38/0x3c)
Mar 20 20:03:19 raspberrypi kernel: [244285.481048] [<8067b650>] (mmc_get_card) from [<806846b8>] (mmc_sd_detect+0x24/0x7c)
Mar 20 20:03:19 raspberrypi kernel: [244285.481057] [<806846b8>] (mmc_sd_detect) from [<8067dc6c>] (mmc_rescan+0x1cc/0x39c)
Mar 20 20:03:19 raspberrypi kernel: [244285.481068] [<8067dc6c>] (mmc_rescan) from [<8013bcf4>] (process_one_work+0x170/0x458)
Mar 20 20:03:19 raspberrypi kernel: [244285.481077] [<8013bcf4>] (process_one_work) from [<8013c038>] (worker_thread+0x5c/0x5a4)
Mar 20 20:03:19 raspberrypi kernel: [244285.481085] [<8013c038>] (worker_thread) from [<80142324>] (kthread+0x138/0x168)
Mar 20 20:03:19 raspberrypi kernel: [244285.481094] [<80142324>] (kthread) from [<801010ac>] (ret_from_fork+0x14/0x28)
Mar 20 20:03:19 raspberrypi kernel: [244285.481099] Exception stack(0xb2ab9fb0 to 0xb2ab9ff8)
Mar 20 20:03:19 raspberrypi kernel: [244285.481103] 9fa0:                                     00000000 00000000 00000000 00000000
Mar 20 20:03:19 raspberrypi kernel: [244285.481109] 9fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
Mar 20 20:03:19 raspberrypi kernel: [244285.481113] 9fe0: 00000000 00000000 00000000 00000000 00000013 00000000
Mar 20 20:05:22 raspberrypi kernel: [244408.362184] INFO: task kworker/2:2:8183 blocked for more than 120 seconds.
Mar 20 20:05:27 raspberrypi kernel: [244408.362200]       Tainted: G         C        4.19.25-v7+ #1205
Mar 20 20:05:27 raspberrypi kernel: [244408.362206] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Mar 20 20:05:27 raspberrypi kernel: [244408.362214] kworker/2:2     D    0  8183      2 0x00000000
Mar 20 20:05:27 raspberrypi kernel: [244408.362248] Workqueue: events_freezable mmc_rescan
Mar 20 20:05:27 raspberrypi kernel: [244408.362282] [<8082414c>] (__schedule) from [<808247bc>] (schedule+0x50/0xa8)
Mar 20 20:05:27 raspberrypi kernel: [244408.362300] [<808247bc>] (schedule) from [<8067b510>] (__mmc_claim_host+0x120/0x228)
Mar 20 20:05:27 raspberrypi kernel: [244408.362318] [<8067b510>] (__mmc_claim_host) from [<8067b650>] (mmc_get_card+0x38/0x3c)
Mar 20 20:05:27 raspberrypi kernel: [244408.362335] [<8067b650>] (mmc_get_card) from [<806846b8>] (mmc_sd_detect+0x24/0x7c)
Mar 20 20:05:27 raspberrypi kernel: [244408.362355] [<806846b8>] (mmc_sd_detect) from [<8067dc6c>] (mmc_rescan+0x1cc/0x39c)
Mar 20 20:05:27 raspberrypi kernel: [244408.362377] [<8067dc6c>] (mmc_rescan) from [<8013bcf4>] (process_one_work+0x170/0x458)
Mar 20 20:05:27 raspberrypi kernel: [244408.362395] [<8013bcf4>] (process_one_work) from [<8013c038>] (worker_thread+0x5c/0x5a4)
Mar 20 20:05:27 raspberrypi kernel: [244408.362412] [<8013c038>] (worker_thread) from [<80142324>] (kthread+0x138/0x168)
Mar 20 20:05:27 raspberrypi kernel: [244408.362431] [<80142324>] (kthread) from [<801010ac>] (ret_from_fork+0x14/0x28)
Mar 20 20:05:27 raspberrypi kernel: [244408.362439] Exception stack(0xb2ab9fb0 to 0xb2ab9ff8)
Mar 20 20:05:27 raspberrypi kernel: [244408.362450] 9fa0:                                     00000000 00000000 00000000 00000000
Mar 20 20:05:27 raspberrypi kernel: [244408.362463] 9fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
Mar 20 20:05:27 raspberrypi kernel: [244408.362473] 9fe0: 00000000 00000000 00000000 00000000 00000013 00000000

[e-raser]

Kernel isue. viewtopic.php?p=1438617#p1438617

[jamesh]

Kernel isue. viewtopic.php?p=1438617#p1438617

Only if you are on 4.19.

We do have some mitigation patches in rpi-updtae, which turn off the features causing the memory leak.