sowmini.varadhan at sun.com wrote:
> On (01/29/08 22:22), Erik Nordmark wrote:
>>> http://cr.opensolaris.org/~sowmini/nce_walker/
>> Would it have been simpler to have the kernel allocate the nces using a
>> separate kmem_cache? (That way some existing walker infrastructure in mdb
>> can be used.)
>> Reason I'm asking is because I have some code being developed which moves
>> the nce allocation to use kmem_cache_alloc.
>
> Do you mean like the ire entries?
Yes.
> That would certainly make the
> walker much easier to write, but the existing code did not do a
> kmem_cache_alloc for nce's so I didn't want to change that
> infrastructure merely to make the mdb macro easier to write.
OK.
(But we waste an mblk_t+dblk_t for every nce without any benefits AFAICT
- I'll add it to my list to fix.)
Erik
