On Mon, Mar 01, 2004 at 02:57:46PM +1000, Peter Grehan wrote: > > Any further thoughts? > > The problem is that make_dev() calls devlock(), which requires > curthread to be valid to take out a mutex. > > The fix is to put the hand-crafting of proc0, thread0 and pcpu > data and mutex_init to be *before* the call to cninit(). This > is why i386 works. > > I had the same problem on PPC, and fortunately had the psim simulator > to work out what was going on. Thanks!!! I think that's what my latest patch does, by moving the final console initialization after the proc0/thread0/pcpu setup. I was most worried that what was truly causing the problems was extra function calls consuming more stack than before and that starting to cause problems. It makes me much happier that there is another cause of these problems and that it's deterministic. :-) -- Ken Smith - From there to here, from here to | kensmith_at_cse.buffalo.edu there, funny things are everywhere. | - Theodore Geisel |Received on Sun Feb 29 2004 - 20:07:55 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:45 UTC