[Gluster-devel] crypt xlator bug

[Jeff Darcy]

> As I understand it, mem_get0 is a valid (and even more
> efficient) way to allocate such objects.  The frame cleanup
> code should recognize which method to use when deallocating.
> If that's broken, we're going to have more numerous and
> serious problems than this.  I'll look into it further.

I don't see anything obviously wrong, but I did find this
gem of a comment in mem_get:

* I am working around this by performing a regular allocation
* , just the way the caller would've done when not using the
* mem-pool. That also means, we're not padding the size with
* the list_head structure because, this will not be added to
* the list of chunks that belong to the mem-pool allocated
* initially.
*
* This is the best we can do without adding functionality for
* managing multiple slabs. That does not interest us at present
* because it is too much work knowing that a better slab
* allocator is coming RSN.

Now I'm curious to find out what effect your change will have,
but I suspect we'll still be a while figuring this out.