Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Hi Guys, Thanks very much for all your efforts in debugging & fixing this. You certainly seem to have sorted it out now. The show command is now working properly, and things seem to be syncing up correctly now......I can now get down to actually implementing my solution, and hopefully the questions will now be more 'user' rather than 'dev'. ;-) Best Regards Nick Philipp Reisner wrote: > On Wednesday 17 June 2009 20:12:32 Nick Liebmann wrote: > >> Hello again Philipp, >> >> I must be going mad........ >> >> > > Nick, > > I found something the open moko freerunner is good for ;) > > The kernel in the freerunner's default config seems to have a > trap handler for unaligned accesses and fixes them. > > But that can be disabled: > echo 0 > /proc/cpu/alignment > > With that we where able to reproduce the issue. The bug was > not the code, the bug is that gcc produces incorrect code > when using memcpy. -- As soon as we switched that to memmove > it worked. -- But only at certain optimisation levels. > > Without any "-O" it woked even with memcpy. > > I have pushed the patch to the public git repo, please confirm > that it works now for you as well. > > -Phil >