PCI Express powering off when screen locks + time passes?

I have an issue…

I was copying data onto my drive in Dom0.
The screen locked and since I was copying over 300 GB of data I was just waiting.
I unlocked again after I got back from making coffee and the writing had an I/O Error… And I find that the drives are no longer in the lists, and can’t be seen by Domain-0 any more.

I can see the PCI device, but I can’t see the drives.
There is no longer any /dev/sda or sdb.

What is going on here please?
Is there some setting I need to change in Fedora or Xen configuration to stop things from vanishing?

Thanks in advance.

Just to be sure that I’ve given you accurate information below, any chance you could post the output of lspci, and the last few lines of dmesg (redacting any information you see fit to redact)?


What I think has happened based on the info available so far

(will edit if more information becomes available)

I’m assuming that this is an internal drive connected to a PCI storage controller (you’re describing it like it is, at least), so you can’t really just unplug it and plug it back in like an external one.

I don’t think this has anything to do with Xen, Qubes, or dom0. This sounds like you’ve got yourself a loose connection, the drive was knocked/bumped while writing, or you’ve got yourself a faulty drive, unfortunately…

…and it also sounds like the firmware running on your PCI storage controller was coded to cut power to a drive that threw an error (most likely to protect anything else connected, in case it was an electrical fault…).

If you want, you can check the output of dmesg in your dom0 terminal, and the last few lines of output should be bright red and full of errors, which will confirm this :frowning:


Maybe this might work…

If you want, you can try to rescan the PCI devices by running this in your dom0 terminal:


<PCI_DEVICE> is the XXXX:XX:XX.X value at the beginning of the line in the output of lspci that corresponds to your PCI storage controller.


# Become root. (sudo will not work with echo commands to sysfs)
sudo su -

# If you're boot drive is connected to the same PCI device,
# DO NOT RUN THIS COMMAND.  Your computer will NOT like it...
# SKIP THIS COMMAND if this is the case.
echo 1 > /sys/bus/pci/devices/<PCI_DEVICE>/remove

# Tell dom0 to check all PCI slots for what's there, and try and
# power them up with applicable firmware from /lib/firmware/
echo 1 > /sys/bus/pci/devices/rescan

# Get out of root
logout # or press Ctrl+D

If that doesn’t get your drive to show up again in lsblk, then I’m afraid you’ll likely have to open up your case and check your connects, and maybe even replace the drive :frowning:

Well, it would seem that that drive is actually failing.
Literally as I set it up as a backup drive and copy on 900 GB of data, it starts to fail…
So that’s why there is the I/O error.

Connections are all fine, there is no issue with them.
The PCI Device is still there.

01:00.0 SATA controller: ASMedia Technology Inc. Device 1064 (rev 02)

So it is there after the issue. I did check that at the time.

But it lost the drives because I believe it was powered off as the drives were powered down, but the card should always be on, meaning the PCIE shouldn’t suspended or put to low power or anything.

I’ll get the DMESG output next time it happens.

1 Like