George Michaelson wrote:
>>>isn't the same as usertime probe in USB.
>>
>>The basic cause of the problem is well known - it comes down to memory
>>fragmentation, the kernel needs to allocate a contiguous chunk of
>>memory for the scsibus at the time its created.  On a system that has
>>been running a while, especially with smaller memory, sometimes it
>>can't find a large enough chunk to use.
> 
> 
> ok. I have 256M in my laptop. how big is this chunk? can I not pre-reserve
> 'enough' in my kernel to have at least one instance? I mean, if this is really
> only some low k's of memory, can't we consume that as potentially wasted, but
> useful bufferspace?
All it wants is 64k.  In the olden days, when the umass driver first
appeared, this never seemed to be a problem.  I'm not sure what kernel
changes (UVM?, UBC?, something else?) caused kernel memory to become
more fragmented.

	-- Lennart