> From: Sarah Sharp [mailto:sarah.a.sharp@xxxxxxxxxxxxxxx] > Sent: Friday, February 15, 2013 2:05 PM > > On Tue, Jan 29, 2013 at 07:40:38PM +0000, Paul Zimmerman wrote: > > Resending with updated xhci.h and signed-off-by added. > > > > Hi Sarah, > > > > We (Synopsys) would like to donate the attached code for driving the debug > > capability of the xHCI controller. > > > > We used this code to test the debug capability of our USB3 IP core. As it > > stands, this is test code rather than fully-fledged Linux driver code, but it > > can be used as a starting point for a complete driver. So although it is > > provided as a patch, it is not intended to be applied in its current form. > > > > This code does not implement a gdb interface; rather it has a kernel thread > > which can do two things - either source a continuous stream of packets on > > the IN endpoint and sink whatever is sent to it on the OUT endpoint, or > > receive whatever is sent to it on the OUT endpoint and echo it back on the > > IN endpoint. The mode is selected by the DBC_SRC_SINK #define at the > > top of the file. We used this, along with the usbtest module and testusb > > program on the host side, to send/receive a constant stream of packets > > over the two bulk endpoints. > > It sounds like Synopsys doesn't want to implement the full xHCI USB debug > capability right now, is that correct? Yes, that's right. > If so, I'm not really sure how helpful this new code is. As you said, > it's useful for testing new hosts against usbtest, but it's not going to > be of much use to normal Linux users. Full debug capability would be > more useful IMO. Sure. But we don't have a resource available to work on that. I thought I remembered you mentioning in an email that you planned to implement debug capability support. So I thought this code could give you a head start on that. Or, if someone else wanted to work on it that would be fine too. > Basically, if I end up merging this code, it isn't going to have very > many users, and it will probably just end up bit-rotting. But if we > could instead get full gdb debug capability like we have under EHCI, I > think it would be worth it. I explicitly mentioned that this code is _not_ for merging as it stands. > So what is Synopsys' commitment level to getting this code cleaned up to > the point that someone could use it to add full debug capability > support? I'm not sure what you mean by that. I think it's usable right now, except maybe for the missing error handling. And I don't know if bulk STALL support is needed for the debug capability, if so that would need to be added. But both of those should be pretty straightforward, I think. dbc_poll_events() and dbc_complete() are the two routines which implement the source/sink functionality. Those are the routines that need to be modified/replaced to implement the gdb support. The rest of the code should not need much modification. -- 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
