On Monday 14 December 2009, Øyvind Harboe wrote: > > I tried restoring the workaround for that (TLR) and added a comment > > about that being IMX-specific. Could you give this patch a try? > > I ran a real quick test. > > It's better I think. > > I can do a "reset run" and the "reset init" error messages are prettier.
Those are good things. :) Compared to current code, is this better for iMX31? > "reset init" *fails* w/debug_level 1 and *works* with debug_level 3.... How about the default, level 2? > This is indicative that some pause is needed somewhere... I removed one hard-wired delay, since that stuff is better suited to config files. (Not all targets want it...) Did you update the config file with something like "jtag nsrst_delay 5" to make up for that? > Using > zy1000, debug_level 3 can have a non-trivial performance impact > compared to a PC... There's some kind of bug with memory access, for sure. I've seen reads misbehave. I'm not sure what's up, but it doesn't seem related to this patch. By the way, from reading that log ... one bug the iMX31 has is resetting the debug logic with SRST. It should not have taken the core out of halting debug mode. I may do an experiment to see if OMAP2 also does that. (Or even stick a warning in!) - Dave _______________________________________________ Openocd-development mailing list [email protected] https://lists.berlios.de/mailman/listinfo/openocd-development
