> ... problems with USB enclosure hangs with Fedora Core 2 ... A friend of mine is working on this problem. It is real, and there seems to be a kernel flaw causing it - something to do with host_reset. Contact me and I will forward his email address to you. I have seen problems with large drives in some USB enclosures and not others. I first saw a permanent hang with *any* size drive with late 2.4 kernels and external drives using the Cypress CY7C68013 USB2 chipset. I did not see the problem with two other USB enclosures that used different - and slower - chipsets. The 2.6 kernels do not have the old 2.4 problem (a permanent hang). The new version of the problem is a big slowdown in transfer rate after an initial burst on small ( <137 GB ) drives. On large drives, the slowdown is extreme - the data still moves, but there are 5 minute (yes, 300 second!) pauses in the data transfer. Again, this only happens with the Cypress chipsets, and the problem does not appear under Windoze. My guess is that it has something to do with the speed of the Cypress chipset, and the size of the data chunks with LBA48 addressing. Most USB2 chipsets are single buffered; there are gaps in the data transfer allowing the kernel to do housekeeping. The Cypress chipset is double buffered, and fills one buffer while spewing another, allowing it to go at full USB2 bandwidth during write. It will help to learn the chipset in your enclosure, which may appear in /proc/bus/usb/devices . You might also open the case and read the chipset markings off the big LSI on the USB2 interface board. I BCCed my friend, so perhaps he can weigh in here. Please help the young fellow achieve fame and fortune (or at least resume points) by providing observations that will help him debug this pesky problem. Keith -- Keith Lofstrom keithl@xxxxxxxx Voice (503)-520-1993 KLIC --- Keith Lofstrom Integrated Circuits --- "Your Ideas in Silicon" Design Contracting in Bipolar and CMOS - Analog, Digital, and Scan ICs
