On Fri, 3 Dec 2010, Linux User wrote: > 02.12.2010 07:11, Greg KH пиÑ?еÑ?: > > On Thu, Dec 02, 2010 at 03:40:40AM +0300, Linux User wrote: > > > >> >From time to time I experience problems with external hard drives > >> connected via USB bus to my laptop. An USB device may suddenly go > >> off-line after receiving some amount of data. > >> > > Are you sure this isn't an electrical issue? The log shows that the > > device is electrically disconnecting itself, which is not something that > > the kernel can do. > > > Actually, external drives' behavior is more complex and there are about > three ways a failure may happen and self-disconnecting of a device is > only one of manifestations of the problem. > > 1) The transfer rate is unexpectedly appears to be as low as 3..6Mb/s > instead of normal 20-30Mb/s with continuous resets of the device every > minute or less (reported in system log), same for reading or writing, > but nothing happens when the disk is idle. After n'th reset the device > may disconnect itself. This also may be a power problem. The resets are caused by communication failures, which can be caused by lack of power. > The SATA hard disks themself is healthy and > totally OK, there are no problems when either of them is connected to > desktop PC via (e)SATA-SATA cable. But that doesn't mean the drive's USB interface chip works okay. > 2) The transfer rate is normal, no problems during idle operation or > reading but few resets may happen less frequently, after n'th (n is > 3..5, anyway < 10) reset the device goes off-line and kernel reports > failure to commit to filesystem journal. After that the device may go > on-line again but filesystem may be already corrupted. Again, the resets are symptoms of communication errors probably caused by lack of power. Or they may be caused by noise in the USB cable, bad cable terminations, etc. > 3) After (1) or (2) happens, disconnecting the USB cable and putting it > back may result in error messages telling that the device cannot be > attached to the system (see pieces of log in the attachment). > > If you use a different cable does this go away? How about providing > > more power to the drive with an external power supply > I think that power supply is OK. Those sudden disconnects only look like > problems with power supply, but just the same device works without any > problems with just the same power supply when attached via (e)SATA-SATA > interface. That doesn't mean anything, since using an eSATA or SATA interface doesn't test the drive's USB interface chip. To get more information about the problem, you should record a usbmon trace. See the instructions in Documentation/usb/usbmon.txt. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
