On Fri, 24 Aug 2012, Sami Farin wrote: > On Sun, Jul 15, 2012 at 11:14:13 -0400, Alan Stern wrote: > ... > > In past years I have found that Genesys's devices tend to be rather > > buggy. This card reader continues the tradition. Basically, although > > the firmware works reasonably well at high speed, the SuperSpeed > > support is too buggy to be usable. > > > > My advice is to exchange the card reader for one of a different brand. > > I got �6 reader from eBay, sanXi usb3 sdcx reader.. > It reset itself to usb2 speed when I tried it in usb3 port > and in usb2 port it is quite slow (14 MB/s). > There are benchmarks done by other people with sanXi where they get good > speed (90 MB/s) but presumably they do not have any ASMedia near their computers. > Also, (after first use) the usb3 port I used sanXi in was DEAD till I rebooted > the machine; also, other devices were not detected in the port, either. > > > Alan Stern > > > > P.S.: Sarah, the lspci information shows this xHCI controller to be an > > ASM1042 controller from ASMedia. Is this one of those known to have > > the short-packet bug? The usbmon trace shows a cancelled transfer, for > > which 0 bytes must have been received, but with actual_length equal to > > the transfer_buffer_length. > > 3.6.0-rc2-g6dab7ede has some funny printk's , but it panics on network use > so I am back to 3.5.0-rc7-g84a1caf: > > <5>[ 5.163846] scsi 10:0:0:0: bad extra_isize (%u != %u)Direct-Access Generic STORAGE DEVICE TS07 PQ: 0 ANSI: 0 > <3>[ 2277.412255] sd 10:0:0:1: 6mpf: %llx > > > > This would seem to indicate another type of bug in the host controller. > > Apparently it reported normal completion for a transfer that was > > cancelled. > > Can these be worked around in kernel, or is firmware update needed > for asmedia? I have latest BIOS from Asus... I don't know the answers to all this. Sarah is the person to ask, but she will have a lot of other work to get through first when she returns from vacation. All I can tell you is that the ASM1042 controller appears to have a bug for which there is a workaround in the kernel. But that workaround may be targeted at other models, so the kernel may need a change in order to apply the workaround to the ASM1042 as well. 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
