[Linux-aus] strange NVMe errors when USB is busy

[Russell Coker]

On Sunday, 26 January 2025 16:18:00 AEDT Info wrote:
> If the enclosure is USB 3 instead of 3.1 or later, check the USB ID of the
> USB chip. There are a lot of USB 2.5 chips that implement only parts of USB
> 3 because USB 3 was a really sloppy standard.

Bus 002 Device 005: ID 152d:8561 JMicron Technology Corp. / JMicron USA 
Technology Corp. salcar docking station two disks

This is the device.

> It could also be the cable. I throw out about half of the USB 3 cables after
> intermittent errors. I replace them with USB 3.1 Gen 2.

The USB device and cable could have problems.  I don't think that they do but 
it's definitely possible.  But the NVMe is connected by PCIe so if everything 
else is working well then no combination of USB broken behaviour should be 
able to interfere with NVMe.

Things apparently aren't running as they should for unknown reasons.

The system has ECC RAM with advanced ECC enabled (chipkill using ECC across 
pairs of DIMMs).  I think this allows correcting 2 bit errors and detecting 3 
or more bit errors.  I have not seen any sort of ECC error reported so the 
core of the system is working OK.

> The WD SSDs have controllers from everywhere. They buy some from Toshiba,
> now named Kioxia. If you can find the name of the SSD controller chip in
> the WD, you can look it up for known errors and perhaps a firmware update.

It has the latest firmware according to fwupdmgr.

# fwupdmgr update
WARNING: UEFI capsule updates not available or enabled in firmware setup
See https://github.com/fwupd/fwupd/wiki/PluginFlag:capsules-unsupported for 
more information.
Devices with the latest available firmware version:
 • KXG70PNV2T04 NVMe KIOXIA 2048GB
 • PC SN810 NVMe WDC 2048GB
 • UEFI dbx
Devices with no available firmware updates: [USB hard drives]
 • ST8000AS0002-1NA17Z
 • ST8000AS0002-1NA17Z


-- 
My Main Blog         http://etbe.coker.com.au/
My Documents Blog    http://doc.coker.com.au/