> "Thomas G. Lockhart" <lockhart@alumni.caltech.edu> writes:
> > Sadly, that is still true also. As I mentioned earlier, the actual
> > segfault happens when memory is free'd toward the end of a query. But of
> > course there is some problem earlier when the memory was allocated...
>
> Has anyone tried building the backend with a debugging malloc library?
>
> dbmalloc, or Electric Fence, or Purify would probably help smoke out
> any remaining coding errors of this kind (using already-freed space,
> writing past the end of a malloc'd block, etc).
>
> Purify is a commercial product, but the other two are freeware and
> shouldn't be hard to find on the net.
Good suggestion.
We did have someone use Purify six months ago, and he came up with some
nice stuff. Electic Fence is good for over-runs and using already
free'ed memory, and was thinking of trying that here, but the idea that
this is was some type of alignment problem had me moving in another
direction.
--
Bruce Momjian | 830 Blythe Avenue
maillist@candle.pha.pa.us | Drexel Hill, Pennsylvania 19026
+ If your life is a hard drive, | (610) 353-9879(w)
+ Christ can be your backup. | (610) 853-3000(h)
