acache memory leak found

Shane Kerr Shane_Kerr at isc.org
Mon May 15 10:54:52 UTC 2006


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

All,

Okay, it took ages but I did find the leak with acache. Basically, the
invoking function in rbtdb.c didn't handle the error case properly
(overmem), and leaked memory. It had attached the memory to the
database, so it wasn't leaked according to the memory tracking stuff.

Yes, this means that putting the check in to not exceed our maximum
memory introduced a memory leak. The irony *almost* made me smile.

I was able to reproduce the leak, fixed it, and confirmed that the fix
works. The change is committed to the rt16034 branch, and it's on the
bind9-review queue. It would be nice to merge this into v9_4, so we can
try it on senna. :)

I don't know what other v9_4 changes are pending. I guess maybe the core
dump reported (support ticket #908) needs fixing? Although it might not
be the most efficient, I don't mind looking at this as well.

Now, back to the DHCP!

- --
Shane
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFEaF38MsfZxBO4kbQRAjjnAKDAxZkQVraG/zvWKUVopTHRW/6lEQCbByub
ezvKDcbT+invTof4H0z7xrI=
=qVEf
-----END PGP SIGNATURE-----


More information about the bind-workers mailing list