On Wed, Mar 30, 2005 at 08:54:24AM -0800, Mark Wong wrote: > I'm still panicking. Is still more helpful (config attached)? Mark, Sorry I haven't had more time to look into this, but I got pulled into some higher priority things. :( I went to OSDL and ran some debug code on Mark's machine a couple of weeks ago. Mark's machine is panicing because the OF device tree strings are being overwritten/corrupted. Hence, the call to of_find_node_by_type() in xics_init_IRQ fails and panics. Upon further examination, I discovered that the device strings were still good at the beginning of paging_init(), but were overwritten by the time we were leaving paging_init(). Note that paging_init() initializes the zone and page structs. I suspect a problem in the memmap_init() (actually memmap_init_zone) routine. However, I've taken a good look here and nothing really stands out. Hopefully, others can use this info and help track down the root cause. Most likely, this is a ppc64 specific problem. But, previously I did find a memmap_init bug (arch independent) that was only found via the ppc64 code. I haven't given up on this, but any help would be appreciated. Thanks, -- Mike
