Created attachment 70582 [details] dmesg log (from boot) I have a specific USB harddisk that keeps getting resets and never appears as a device on the system. It used to work just fine, but as I did not use it for anything important when it stopped working, I did not do the proper paperwork back then. Unfortunately this means I cannot tell you when it last worked; I guess it has been over a year and perhaps even 2 years. I have used USB pens after the harddisk stopped, so I have a feeling it is specific to the device. It is a "Packard Bell Store & Save" (~750 GB) and the model number is "HD D2 U2". It has an external power and has a little "USB 2.0" next to its USB plug. I have tried in once in a Windows a while back, but without any luck. Though I could not tell if this was because the disk was broken or because Windows had no clue to handling the underlying file system[1]. I am attaching my dmesg log (fresh boot and with the USB being turned on when the boot has been completed). I will follow up with a lsusb -v output. The machine uses a 64bit kernel with 32bit user land. This is a forward of the Debian bug #639462[2]. I have used the Debian kernel for my testing and (to my knowledge) it contains the "EHCI: fix direction handling for interrupt data toggles" commit (see [2] comment #22) ~Niels [1] It is encrypted with cryptsetup luksOpen/luksClose and has (as I recall) an xfs layer beneath that. [2] http://bugs.debian.org/639462
Created attachment 70592 [details] lsusb -v output As far as I can tell, the harddisk is the last entry. Other than it is identified as "Cypress"[1] and the MaxPower is 2mA[2] I do see anything that is immediately odd (but then again, I do not work with this a lot). [1] The disk itself does not mention "Cypress" anywhere; only Packard Bell. Though I guess this could comparable to a "Dell" with "Intel Inside". [2] The power line and the disk seems to say "5V 2A" and "12V 1.5A" (yes, both appears on the disk and the transformer on the powerline).
The oldest file (I can think of) I would have moved to the drive (but didn't) is from is from early Sep 2010 and I can think of some files around late Dec 2009 that I most likely moved to the disk. I created (or updated) an (autofs) symlink to the drive in Sep 2008, which implies it definitely worked back then. ~Niels
All USB bugs should be sent to the linux-usb@vger.kernel.org mailing list, and not entered into bugzilla. Please bring this issue up there, if it is still a problem in the latest kernel release.