You are not logged in.
I have intermittently gotten USB error -110 on my external backup drive, for months. I learned from this thread (https://bbs.archlinux.org/viewtopic.php?id=149384) to power down the system, disconnect power from the USB drive, and hook things back up to work around the problem. This was not a big deal when I only had to do it once or twice a week. But in the past few days, it has started happening at just about every reboot. In fact, I can no longer even do a clean shutdown, because the error prevents the filestem from being unmounted.
Once the problem has been worked around, the external drive works perfectly until I need to reboot or poweroff.
I found other online articles indicating that to actually fix the problem, instead of just working around it, I should set BIOS option IOMMU Controller to Enabled. This was specifically for Gigabyte motherboards, and mine is a Gigabyte Z68-D3-B3. However, I cannot find such a setting in my BIOS configuration screens (Award BIOS). The article said that if that doesn't work, to set linux command option "iommu=soft". I have done that, to no effect.
This is a 2 TB Seagate USB 2 drive. It is several years old. I might replace it with a new drive, but it kind of irritates me that the drive is working for everything except shutdown and restarting the system.
Tim
Last edited by ratcheer (2015-07-17 22:34:38)
Offline
Can you try using another usb cable, another power supply for the usb drive, or unplug other usb devices ?
110 is not enough power to the usb device if i'm not mistaken...
Another option could be to try with 'noacpi' added to your boot options...
Offline
Thanks for your response.
It is a funny USB cable with a tiny connector on the disk drive end. I will look around, but I doubt that I have another cable like that. I have tons of wall warts, I will look around and see if I can find one with satisfactory output specs.
Many of the articles I have found indicate that the drive will probably work perfectly on a Windows PC. I think I will just buy a new drive and move that one to my wife's PC. I will have to resume my backups from ground zero.
Finally, this old drive still works with my openSUSE stable (13.2) instance, which runs an old 3.16 kernel. I wonder whether the latest Linux kernels are just less tolerant of some little glitch in the USB connection? My Arch instance is on 4.1.2-2.
Tim
PS - I just re-tested my openSUSE instance. It boots and reboots with the external drive. I do see the -110 errors during startup, but the system continues and completes startup, successfully. Arch with older kernels started successfully, most of the time, but since kernel 4.1 or so, the error prevents startup from completing and eventually throws the system into rescue mode.
Last edited by ratcheer (2015-07-17 15:01:11)
Offline
You could always install the linux-lts kernel (current is 3.14 and i find it usefull to always have an older kernel available for troubleshooting)
sudo pacman -S linux-ltsedit: just be sure you select the proper kernel on your grub menu...
Last edited by Malkymder (2015-07-17 14:59:33)
Offline
Thanks again , Malkymder. But, one of the reasons I prefer Arch is to stay at the leading edge of Linux. I only use openSUSE when I have to run some crap website with Flash. Well, also for printing with a stupid Brother 32-bit proprietary driver.
Tim
Offline
Installing the lts kernel does not prevent you from staying bleeding edge but gives you a troubleshooting option when you do get bleeding problems with latest kernels allowing you to switch to another kernel for troubleshooting... i have both the latest kernel and lts, when something goes wrong with latest i can still use the lts for troubleshooting instead of using a livecd/usb to try and debug stuff ![]()
Offline
IIRC. you are an EE. I bet you are seeing IR drops on your cable and that that the drive is starving for power. It is supposed to work down to 4.25V (per the USB 2.0 spec), but not all devices thrive there. I have also seen significant inductance in long cables and in some USB ports. I have seen more than one USB port use a common mode choke in order to achieve FCC Part 15 and CE electromagnetic emissions and susceptibility compliance -- resulting in horrible transient load response. Of course, if the voltage drops below 4.25 at the drive, it is game over. Have you a scope? I would look for excursions below the limit while the drive is seeking. Can the drive be powered externally? Have you shorter cables? Without ferrites? How about adding a 100uF low ESR capacitor at the drive end of the cable?
Last edited by ewaller (2015-07-17 23:58:57)
Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
The shortest way to ruin a country is to give power to demagogues.— Dionysius of Halicarnassus
---
How to Ask Questions the Smart Way
Offline
ewaller, thanks, but I am not an EE. I spent 34 years as an IT systems administrator, mostly as a database administrator. Hardware is not my strong suit.
Anyway, I have resolved the problem by buying a new USB3 3TB external drive. I already have my first set of backups on it.
I will move the old one to my wife's Windows PC. If it works, great. If not, I got five or six good years out of it.
Tim
Offline
Just in case, you may want to check your PSU.
One day, my (internal) HDD started to reset spontaneously and I traced it to 20% voltage drops caused by CPU load due to mass failure of PSU capacitors.
Offline
Just in case, you may want to check your PSU.
One day, my (internal) HDD started to reset spontaneously and I traced it to 20% voltage drops caused by CPU load due to mass failure of PSU capacitors.
Interesting. How would one do that?
Tim
Offline
If you are (like me) into collecting tools for everything, get a digital multimeter and just measure voltage in some unused molex plug.
If not, connect this HDD, load all CPUs (e.g. while :; do :; done times the number of cores) and see if the HDD is bothered by that.
You may also get voltage readouts in BIOS setup or using lm_sensors, although the latter may not work without acpi_enforce_resources=lax and may be useless if the voltage sensors are connected to power rails through divisors of unknown ratio.
Offline