Re: Process Capabilities on 2.2.16, Sendmail problem revisited

From: Andrew Morgan (morgan@transmeta.com)
Date: Tue Jun 13 2000 - 22:49:52 EST


Just to be clear. What is in the 2.2.16 kernel is ok and supports the
features folk had come to make use of with the old code [*] (given the
lack of filesystem capability support in the kernel) with the important
addition that the kernel now defaults to not having the stated sendmail
vulnerability. The next step is being discussed on this list (where more
folk with an active interest are subscribed than here):

 http://www.geocrawler.com/lists/3/SourceForge/4109/0/

For reference, here is some sordid detail of what was wrong and how it
was resolved:

 http://www.geocrawler.com/lists/3/SourceForge/4109/25/3862070/

with typo fixes:

 http://www.geocrawler.com/lists/3/SourceForge/4109/0/3864894/

Cheers

Andrew

[*] to get the old behavior, you lower cap_bset and raise pI in the init
process. This is just not the 2.2.16 default - becuase it is not
provably secure in the face of legacy binaries and there are examples
(like sendmail) for which legacy code becomes exploitable in a pI -> ~0
model.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Thu Jun 15 2000 - 21:00:30 EST