[Gluster-devel] crypt xlator bug

[Jeff Darcy]

> > I found one issue that local is not allocated using GF_CALLOC and with a
> > mem-type.
> > This is a patch which *might* fix it.
> 
> It does. The memory corruption disapeared and the test can complete.

Interesting.  I suspect this means that we *are* in the case where the
previous comment came from.  Mem_get can allocate objects two ways:

* As one of many objects in a slab, tracking internally.

* As a singleton, directly via GF_*ALLOC.

In mem_put, we do some pretty nasty pointer arithmetic to figure out
which way an object was allocated.  If we get it wrong, and therefore
use the wrong *de*allocate method (either way I believe) then we'll
corrupt memory.  The symptoms so far suggest that an object was
allocated within a slab, then deallocated as a singleton (causing its
memory to be poisoned).  That sucks.