USB dies when using mass-storage
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
linux-meta (Ubuntu) |
Invalid
|
Undecided
|
Unassigned |
Bug Description
On a Dell Inspiron 1501, there are problems with the USB chipset and the kernel. This probably needs to be filed upstream with the kernel developers, because I was able to reproduce the issue with several versions of the Linux kernel, including the latest one I could get my hands on.
During the course of normal operations, as long as I do not use any mass storage class device, all is well.
However, when I plug in an external hard drive, or any other mass storage device, and perform large amounts of data transfers, the entire USB subsystem of the kernel renders itself inactive. Any process using USB devices winds up left in a "D" state (as reported by the output of "ps ax", and cannot be killed via any means. After this point, no USB hardware is able to be used, including mice and printers—it's almost as if something is happening to make the entire USB stack hang. However, *no* information is reported in dmesg as to the unavailability of the USB stack, there are no kernel OOPS reports, and no indication that there is a problem other than the fact that hardware on any USB port is unavailable.
The output of lspci -vv will be attached to this bug momentarily, as well as the output of lshal.
This is critical because it means that I can not perform backups of my laptop to my USB storage devices. This problem does not exhibit itself on computers that have different USB hardware. This is an issue with Feisty and Gutsy, both 32- and 64-bit, on this computer, as well as with other systems that use the Linux kernel (in either 32- or 64-bit builds).
I have no diagnostic output to offer, because there is none being generated.
Here is what happens when USB dies. The CPU usage on I/O wait goes way up (to 100% on one core, and sometimes both), the USB processes and the process that was using USB is stuck in a state of uninterruptable sleep, and the only way to fix the issue is to re-boot. This happens on any activity ranging from the creation of a new 40 GB ext2/3 fs, to checking an ext2 filesystem. to lots of data transfer after mkfs. IOW, it seems to be somewhat random, but I can only actually trigger the issue when using some form of USB mass storage.
I am about to try this again after rebooting with a small flash drive to see if I can trigger it there. Larger devices certainly trigger it all the time.
Sometimes, some error messages show up in dmesg *before* the drive fails, about resetting low speed devices. I don't know why, because everything in this system should be USB 2.0, and I know that the drive is a USB 2.0 device.
I don't think that the APIC errors are related, though I suppose anything is possible. I get them all the time, but I do not know why.