On 7/12/09 5:01 PM, "Xiaofan Chen" <[email protected]> wrote:
>>> >>> I do not know anything about svn-bisect. But I just used Google for >>> 5 minutes to locate the svn commit for the above lines. >>> >>> 1) r1507 by Zach to fix J-Link reset >>> https://lists.berlios.de/pipermail/openocd-svn/2009-April/000291.html >>> + jlink_end_state(TAP_RESET); >>> + jlink_state_move(); >>> 2) r1616 by mlu >>> + jtag_sleep(5000); >> >> Hmmmm!!! I wonder if this has any relationship to the first part of my V2 >> patch that I recently addressed. If you recall I found that the j-link >> dongle would ignore the first reset command intermittently immediately after >> the USB session was started, but worked subsequently. I speculated that the >> j-link dongle was doing something critical and dropped the ball on occasion >> (probably a bug inside the dongle itself). The physical outcome I observed >> if the reset didn't occur was that the JTAG port was messed up afterwards. >> >> I guess we will need to ask Zach or Oyvind about this one. Hopefully they >> can shed some light on it. > > Need to go back to this after the git-svn side topic. It seems to me this > needs sometime to get fixed and may be not ready for 0.2 release... > > Still Spen could commit Gary's V3 J-Link patch to fix the problem > for Windows users. > https://lists.berlios.de/pipermail/openocd-development/2009-July/009438.html > > Technically it would be the V4 patch, but that is a minor detail. :) As for the problem above, I agree that it will not be possible to get this finished for the 0.2 release. While we tentatively now know what is causing one problem with "reset run" the other reset conditions involving halt and init are still broken at least on my target (and possibly others). Those problems will require additional troubleshooting to identify and correct. The benefit of forgoing the 0.2 release is that it will also give us more time to better test any proposed fixes. I personally worry when patches are rushed because expediency usually introduces as many problems as it corrects. Gary _______________________________________________ Openocd-development mailing list [email protected] https://lists.berlios.de/mailman/listinfo/openocd-development
