On Wed, May 12, 2010 at 09:24:24PM +0530, Ramya Desai wrote: > On Tue, May 11, 2010 at 7:50 PM, Sarah Sharp > <sarah.a.sharp@xxxxxxxxxxxxxxx> wrote: > > If you want to test John's patch, it's on a branch called > > synopsis-link-trbs. I just pushed this morning, sorry about the delay. > > If you see that branch get deleted, you know that Greg has queued the > > patch, and I've moved the patch to the master branch. > > > > If you find the patch helps you, please reply to John's patch and add a > > line like > > > > Tested-by: Ramya Desai <ramya.desai@xxxxxxxxx> > > > > Greg will add that line to the patch before it goes upstream. > > > > Sarah Sharp > > Dear Sarah, > > Thank you very much for the information. > > I checked out the code from your git and followed by > synopsis-link-trbs branch. I used the following commands for it. > > # git clone git://git.kernel.org/pub/scm/linux/kernel/git/sarah/xhci.git > # cd xhci > # git checkout –b synopsis-link-trbs origin/ synopsis-link-trbs > > Then, I tested my UASP device for queue depth 2. Unfortunately, I saw > the issue, again. The xHCI driver submits the requests to the host > controller, but xHCI driver does not receive any interrupts for that > particular submitted commands. > > Please see the attached kernel log file, in the ZIP format. The > following table shows about a few commands start and end line numbers > in the log file. > > Stream ID Command Started Command Completed > at line number at line number > ---------------------------------------------------------------------------------- > 2 136987 137354 > 1 137176 137484 > 2 137360 137671 > 1 137490 137786 > 2 137677 138018 > 1 137792 138178 > 2 138024 ? > 1 138184 ? The transfer starting at 138024 definitely starts at the top of the stream ring, so John's patch should have fixed it if the problem is link TRB activation. I see on line 137720 that the link TRB wasn't given back to the hardware at the end of the transfer before that (I added some debug on top of John's patch). Line 138033 shows that the debug statement in prepare_ring() got hit, and the link TRB should have been given back then, before the transfer at 138024. So it looks like John's patch is working properly, but I can't be sure unless I see the state of the transfer rings. Can you pull the synopsis-link-trb branch and test again? I've added more debugging. John, can you look over Ramya's log file and make sure your patch is working as intended? I've posted it at http://minilop.net/~sarah/ramya-2010-05-12.zip Ramya, have you tried hooking up a bus analyzer and seeing what is happening there? It would be good to know if the host is sending the transfer at all, or if there is something wrong with the way software setup the transfer. Sarah Sharp -- 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
