Sponsored link
Vyatta –
Linux & Open Source
Alternative to Cisco –
Advanced Routing,
Firewall, VPN, QoS..
Free Download ->
| |||||||||||
reiserfsreiserfsPosted Aug 3, 2003 22:43 UTC (Sun) by yem (guest, #1138)Parent article: Reiser4 is coming I keep hearing about a problem with reiserfs that results in data being overwritten (corruption) completely silently. Perhaps something to do with a hash collision. This seems to come up in every reiserfs discussion, but I haven't seen a explanation or response from Hans. Can anyone confirm this problem exists or point to Namesys's official position on the matter? I haven't hit the problem (that I know of - it is a silent failure) but it keeps coming up so some official info would be much appreciated.
(Log in to post comments)
reiserfs Posted Aug 7, 2003 11:37 UTC (Thu) by joib (guest, #8541) [Link] I read somewhere that reiserfs does, in fact, not handle hash collisions at all. Thus the silent overwriting. Supposedly this is because of speed, certainly hash collision detection and handling reduces performance. I guess they figured that their hash algorithm is so good that these collision occur extremely seldom, and avoiding them is not worth the performance loss. And perhaps they were right, I have never heard of a problem that was actually caused by a hash collision, so perhaps it only is a theoretical problem?
reiserfs Posted Sep 11, 2005 20:54 UTC (Sun) by gst (guest, #21487) [Link] theoretical yes... as long as you can't intentionally create hash collisions on their r5 hashing algorithm. since md5 and sha1 don't seem to be so secure anymore i wonder how hard it would be to create collisions on r5.
|
|||||||||||