On Wednesday, November 5, 2003, at 09:01  AM, Paul Kranenburg wrote:

>>> Is anyone working on fixes for the recently intorduced UVM RB tree 
>>> allocator?
>>> Currently, it doesn't try hard enough to honour the address hint 
>>> passed to
>>> mmap calls, which breaks sbrk().
>>
>> could anyone please check if uvm_map.c rev.1.149 changes the 
>> situation?
>> thanks.
>
> It changes... as follows:
>
> root file system type: ffs
> init: copying out path `/sbin/init' 11
> panic: kernel diagnostic assertion "hint <= tmp->end" failed: file 
> "../../../../uvm/uvm_map.c", line 1576
> Stopped in pid 1.1 (init)

Is this on a sun4m or sun4c?  Could the rbtree code be interacting with 
the sun4c "vm hole" in some unfortunate way?

         -- Jason R. Thorpe <thorpej@wasabisystems.com>