On Mon, Apr 22, 2002 at 03:27:09PM -0400, Shaya Potter wrote: > People can install many things now to be LD_PRELOAD'd and not include > that info in the bug reports. A better solution would be to make > encourage people to include LD_PRELOAD info in the bug report. Libsafe as a tool changes the behavior of the software it polices. Some programs will quietly continue to work even with buffer overflows and out of bounds references. For example, commercial applications that we do not have control of the source code may exit "unexplainably" during a "mission critical" function. libsafe as a tool requires that the user is thoroughly educated in the possible consequences of running the tool. To force this upon a user as a default may seem like we're helping to protect them. In practice, we would be forcing draconic security policies upon them. It's perceived that Debian gets a bum-rap about being "slow to release" and "hard to integrate commercial applications". Just imagine what would happen if some desktop pen-jockey tries to run his favorite desktop application, only to have it silently fail (Yeah, errors log in ~/.gnome-session or ~/.xsession, but our desktop pen-jockey doesn't know that). "Well, it runs just fine on Red Hat!" would be his quip in a well meaning, but misrepresented article. libsafe is a wonderful security tool, yes. Should it be installed by default? No. Should it's use be encouraged? Yes, but education needs to go along with this. A basic understanding about the dynamic linker and program error output is essential to its use. You and I can piece these things together, but not everyone can. The user should consiously choose to "break" things. -- Chad Walstrom <chewie@wookimus.net> | a.k.a. ^chewie http://www.wookimus.net/ | s.k.a. gunnarr
Attachment:
pgpEw6e6iTCP3.pgp
Description: PGP signature