RPI4 - Problems with USB3 HDD - Crash after a few hours

bismosa · April 17, 2020 at 2:26 PM

Hello,

I am still setting up my Raspberry 4 with LibreELEC 9.2.1

Every few hours my system is crashing. This only happens with a connected USB3 HDD. With a connected USB2 HDD everything works fine.

I have tested with my RPi 4 (2GB) and my RPi 4 (4GB). Both are the same result.

dmesg generates the following output:

Code

[17465.830002] xhci_hcd 0000:01:00.0: WARN Cannot submit Set TR Deq Ptr
[17465.830008] xhci_hcd 0000:01:00.0: A Set TR Deq Ptr command is pending.
[17465.953911] usb 2-2: reset SuperSpeed Gen 1 USB device number 2 using xhci_hcd
[17481.190151] xhci_hcd 0000:01:00.0: WARN Cannot submit Set TR Deq Ptr
[17481.190157] xhci_hcd 0000:01:00.0: A Set TR Deq Ptr command is pending.
[17481.314050] usb 2-2: reset SuperSpeed Gen 1 USB device number 2 using xhci_hcd
[17497.190307] xhci_hcd 0000:01:00.0: WARN Cannot submit Set TR Deq Ptr
[17497.190312] xhci_hcd 0000:01:00.0: A Set TR Deq Ptr command is pending.
[17497.314195] usb 2-2: reset SuperSpeed Gen 1 USB device number 2 using xhci_hcd
[17528.977259] xhci_hcd 0000:01:00.0: WARN Cannot submit Set TR Deq Ptr
[17528.977264] xhci_hcd 0000:01:00.0: A Set TR Deq Ptr command is pending.
[17529.101161] usb 2-2: reset SuperSpeed Gen 1 USB device number 2 using xhci_hcd
[17544.337394] xhci_hcd 0000:01:00.0: WARN Cannot submit Set TR Deq Ptr
[17544.337399] xhci_hcd 0000:01:00.0: A Set TR Deq Ptr command is pending.
[17544.461300] usb 2-2: reset SuperSpeed Gen 1 USB device number 2 using xhci_hcd

Display More

If i than try to read Data from the Drive i get

Code

[18066.706030] usb 2-2: reset SuperSpeed Gen 1 USB device number 2 using xhci_hcd
[18066.724117] sd 0:0:0:0: [sda] tag#0 UNKNOWN(0x2003) Result: hostbyte=0x03 driverbyte=0x00
[18066.724128] sd 0:0:0:0: [sda] tag#0 CDB: opcode=0x28 28 00 3a 38 57 80 00 00 08 00
[18066.724135] print_req_error: I/O error, dev sda, sector 976770944

Sometimes (but not always) i can reboot with the command:

Code

reboot -f

I have found following tips that did not solve this problem:

- Use another power adaptor. This problem is also when i use a powered USB2.0 hub.

- From ( STICKY: If you have a Raspberry Pi 4 and are getting bad speeds transferring data to/from USB3.0 SSDs, read this - Raspberry Pi Forums)

Add to cmdline.txt:

Code

usb-storage.quirks=XXXX:XXXX:u

Add to config.txt

Code

force_turbo=1
over_voltage=2

But this problem still exists.

What can i do to solve this?

The HDD works on my PI3 without problems. Or is the HDD about to die?

Regards

Bismosa

**Klojum** · April 17, 2020 at 3:54 PM Official Post

I'd leave out the overclocking and start without it to get some sort of 'baseline' which "should" work.

And maybe your USB3.0 controller is a bit too quirky.

**HiassofT** · April 17, 2020 at 4:16 PM

It's hard to tell anything without full dmesg. Could be undervoltage (that'd be logged in dmesg), could be the USB HDD adapter (model/vendor would be in dmesg) or something else.

so long,

Hias

bismosa · April 17, 2020 at 6:44 PM

Hello,

i have a full dmesg log in Attachment. Sorry, I could have done it right away.

I have now tried an other HDD with USB3.0. Same problem here.

Many thanks!

Regards

Bismsoa

**HiassofT** · April 18, 2020 at 9:51 AM

Thanks for dmesg, couldn't spot anything unusual except for the warnings you posted.

Can you check if the VIA USB firmware is up to date (in LibreELEC Settings->Update)?

It could be that your drive is acting up or dying, hard to tell though - haven't seen your specific issues yet and 2 WD drives (3.5" MyBook and 2.5" MyPassport) are working fine here on RPi4/USB3.

so long,

Hias

bismosa · April 18, 2020 at 1:23 PM

Hello,

many thanks!
There are Updates available i will test them:

Bootloader EEPROM from 2019-07-15 to 2019-09-10

VIA USB3 from 00013701 to 000137ad

I think i have found my problem. I have a script running to monitor the system. There is this query included:

Code

hdparm -C /dev/sda

if i disable this, everything works for hours without problems.

It seems there is a bug with this query.

Many thanks for the help!

Regards

Bismosa