Hi, David, I just tried at version of 0.4.0-dev-snapshot (2009-11-06-12:53), fails again, bug exists as before.
Yes, i think i'd confirmed there is a bug, for FA526 core. I used to check the about code, and i think the bug have biggest possibility lay in here: arm7_9->enable_single_step = arm9tdmi_enable_single_step;(at fa626.c:346) that is, i think fa526 may have different jtag strcture compare with common arm9tdmi, that's why the wrong instrction excuted. Thanks & Regards! ------------------ zzzh.2007 2009-11-06 ------------------------------------------------------------- Sender：David Brownell Date：2009-11-06 20:04:28 Receiver：openocd-development; zzzh.2007 CC： Topic：Re: [Openocd-development] bugs for FA526 core On Thursday 05 November 2009, zzzh.2007 wrote: > Hi, David, > > I'd tried using the 0.3.0-rc1-dev-snapshot , bugs exists as before. I'm not clear what snapshot you used. To be completely clear ... can you verify this happens with 0.3.0 or with current git (0.4.0-dev)? If it fails, you'll have confirmed there's an open bug. Someone with FA526 will need to fix it. I know that on a real arm920, it behaved. > > > For FA526 core, following bugs exist: > > 1. after making a step(in openocd-0.3.0-rc0) at reset vector > > (at addr of 0, which is LDR pc 0x5c), the pc did not skip to > > 0x5c, but to 0xfc(the instruction at addr of 4 is LDR pc 0x100) > Is that issue still open in *current* git? A patch was merged > recently that could affect breakpoint handling on arm920t in > such situations... > > > 2. can't step or set break point via gdb commands(next, break). > > > > > > > > > > > zzzh.2007 > 2009-11-06 > _______________________________________________ Openocd-development mailing list [email protected] https://lists.berlios.de/mailman/listinfo/openocd-development
