> From: Pratyush Anand [mailto:pratyush.anand@xxxxxxxxx] > Sent: Wednesday, March 21, 2012 2:36 AM > > Hi Paul, > > I tried to use these patches , but I see that all isoc OUT test does > pass, but it > fails with iosc IN. > I do use device as synopsys dwc3(in HS mode) and host as synopsys ehci > controller. > > May be I will investigate it. But if you see something at a glance > then please let me know. > Regards > Pratyush > > root@(none):/testusb# ./testusb -D /proc/bus/usb/001/002 -t 15 -g 5 > unknown speed /proc/bus/usb/001/usbtest 1-1:3.0: TEST 15: write > 1000 iso, 5 entries of 512 bytes > 002 > usbtest 1-1:3.0: ... iso period 8 microframes, wMaxPacket 1400 > usbtest 1-1:3.0: ... total 5000 msec (5000 packets) > /proc/bus/usb/001/002 test 15, 5.027093 secs > root@(none):/testusb# ./testusb -D /proc/bus/usb/001/002 -t 16 -g 2 > unknown speed /proc/bus/usb/001/002 > usbtest 1-1:3.0: TEST 16: read 1000 iso, 2 entries of 512 bytes > usbtest 1-1:3.0: ... iso period 8 microframes, wMaxPacket 1400 > usbtest 1-1:3.0: ... total 2000 msec (2000 packets) > usbtest 1-1:3.0: iso test, 2000 errors out of 2000 > /proc/bus/usb/001/002 test 16 --> 5 (Input/output error) > root@(none):/testusb# ./testusb -D /proc/bus/usb/001/002 -t 22 -g 2 > unknown speed /proc/bus/usb/001/dwc3 dwc3: incomplete IN transfer ep2in-isoc > dwc3 dwc3: ep2in-isoc's TRB (c3009020) still owned by HW > dwc3 dwc3: incomplete IN transfer ep2in-isoc > dwc3 dwc3: ep2in-isoc's TRB (c3009030) still owned by HW > dwc3 dwc3: incomplete IN transfer ep2in-isoc > dwc3 dwc3: ep2in-isoc's TRB (c3009040) still owned by HW > dwc3 dwc3: incomplete IN transfer ep2in-isoc > 002 > usbtest 1-1:3.0: TEST 22: write 1000 iso odd, 2 entries of 512 bytes > usbtest 1-1:3.0: ... iso period 8 microframes, wMaxPacket 1400 > usbtest 1-1:3.0: ... total 2000 msec (2000 packets) > /proc/bus/usb/001/002 test 22, 2.026283 secs > root@(none):/testusb# ./testusb -D /proc/bus/usb/001/002 -t 23 -g 2 > unknown speed /proc/bus/usb/001/usbtest 1-1:3.0: TEST 23: read 1000 > iso odd, 2 entries of 512 bytes > 002 > usbtest 1-1:3.0: ... iso period 8 microframes, wMaxPacket 1400 > usbtest 1-1:3.0: ... total 2000 msec (2000 packets) > dwc3 dwc3: incomplete IN transfer ep2in-isoc > dwc3 dwc3: incomplete IN transfer ep2in-isoc > dwc3 dwc3: incomplete IN transfer ep2in-isoc > dwc3 dwc3: incomplete IN transfer ep2in-isoc > usbtest 1-1:3.0: iso test, 2000 errors out of 2000 > /proc/bus/usb/001/002 test 23 --> 5 (Input/output error) Hi Pratyush, Please don't top-post. I did not test these patches with the dwc3 driver yet, nor with the EHCI host, only xHCI. With xHCI, you need the ring expansion patchset in order for testusb to work. Also, I think there may be some issue with dwc3 and testusb, maybe Felipe can comment on that. -- Paul > On Wed, Mar 14, 2012 at 1:54 AM, Paul Zimmerman > <Paul.Zimmerman@xxxxxxxxxxxx> wrote: > >> From: Alan Stern [mailto:stern@xxxxxxxxxxxxxxxxxxx] > >> Sent: Tuesday, March 13, 2012 7:05 AM > >> > >> On Mon, 12 Mar 2012, Paul Zimmerman wrote: > >> > >> > > Anyway, it's possible to do this sort of testing already by using > >> > > gadgetfs with the "usb.c" test program, if you compile it with the > >> > > -DAIO option. > >> > > >> > Hmm, I see. Is gadget zero considered obsolete then, if everything > >> > can be done using usb.c and gadgetfs? I wonder why gadget zero is > >> > still maintained? It has support for super speed, for example, while > >> > gadgetfs does not. > >> > >> I think the answer is that gadgetfs is pretty much unmaintained at this > >> point (the original author passed away and nobody else has taken over). > >> Felipe is the closest approximation, since he now maintains the entire > >> gadget subsystem, but he hasn't done much direct work on gadgetfs. > >> > >> > Do you think there could be a problem moving data at high- > >> > bandwidth super speed rates (48K bytes every 125 usec) using > >> > gadgetfs? > >> > >> It would be somewhat less efficient than using an in-kernel driver. > >> As far as I know, nobody has ever tried to measure exactly how much > >> less efficient. And whether or not it would cause a problem depends > >> on the speed of the system. > > > > Thanks Alan. > > > > Felipe, are these patches something you would consider taking? If not, > > I'll just maintain them out of tree. I took a look at inode.c, but I > > don't think my kernel fu is strong enough to update that for super > > speed and fix the inevitable bugs it would introduce. > > > > -- > > Paul > > > > -- > > 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 -- 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
