[Adding linux-usb to the CC list] On Wed, 11 Nov 2015, Douglas Gilbert wrote:
> Hi, > I have been running into problems trying to use a USB device port > on an Acme System's Arietta G25 (Atmel at91sam9g25 hence at91sam9x5 > family in the kernel) as a USB mass storage gadget. Since the problem > may be anywhere in the stack, I have sent this post to a USB expert > as well as at91sam9x5 experts. > > In all cases the mass storage device appears on the host system > after I do 'modprobe g_mass_storage file=/tmp/backing_file' on the > Arietta G25: > > # lsscsi > [0:0:0:0] disk ATA INTEL SSDSC2BW18 DC32 /dev/sda > [6:0:0:0] disk Linux File-Stor Gadget 0400 /dev/sdb > > However if the host port is USB 3.0 (e.g. on my Lenovo > X240 laptop) and the kernel is in the 4.0 series then all attempts You mean if the _gadget's_ kernel is >= 4.0, right? Not the host's kernel. > by the host to read data fail on that mass storage gadget. > For example, I see lots of this in my syslog: > > sd 6:0:0:0: [sdb] > Result: hostbyte=DID_ERROR driverbyte=DRIVER_OK > sd 6:0:0:0: [sdb] CDB: > Read(10): 28 00 00 00 00 00 00 00 08 00 > end_request: I/O error, dev sdb, sector 0 > Buffer I/O error on device sdb, logical block 0 > usb 2-2: reset high-speed USB device number 9 using xhci_hcd > xhci_hcd 0000:00:14.0: xHCI xhci_drop_endpoint called with disabled ep > ffff8800ae58f180 > xhci_hcd 0000:00:14.0: xHCI xhci_drop_endpoint called with disabled ep > ffff8800ae58f1c0 > ... > > I have tried around 6 different kernels in the lk 3.19 to lk 4.3.0 > on the mass storage gadget (Arietta) with 3.19 works while lk 4.0.0-rc4 > is the lowest (in my sample) that fails. My host system is Ubuntu 14.04 > with either its stock kernel (3.13.0-68-generic) or my own hand built > lk 4.3.0; both react the same way to this problem. > > Attached are the outputs of 'lsusb -v' for that mass storage gadget as > seen on the host system. The "319" works while the other two fail. One > difference is that wMaxPacketSize is 512 bytes for "319" and 64 bytes > for the other two. Is that significant? [The block size on most > storage devices is typically 512 bytes.] That most certainly _is_ significant. Almost nothing will work right if the maxpacket size is wrong. > Also, my problem seems to be sensitive to the host port being USB > 3.0 since I have an older Lenovo X201 with USB 2.0 ports that does > _not_ exhibit this problem. IOW when I connect my Arietta G25 mass > storage gadget to the X201, it works properly for all G25 kernels > that I have tried. > > > Also tested a W7 host box with both USB 2.0 and USB 3.0 host ports > and my mass storage gadget worked properly with lk 4.0.0-rc4 on > both 2.0 and 3.0 ports. The Device Manager in W7 shows that it has > xHCI driver present. I don't know what's going on with Windows, but this seems like a bug in either the at91sam9g25 UDC driver or the composite gadget core. Either way, your best bet for locating the bug is to continue bisecting between 3.19 and 4.0-rc4. Unless someone on the mailing list has a better suggestion... Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
