M. Gerards wrote:
I assume you mean the password server with the evil program. The password
server is the one who dies and needs to be debugged. I don't see any changes to
the password server and its interfaces in CVS... I think another debian package
caused this, I remember a change to /etc/passwd or something like that. (This
is just a guess, but I hope it helps :)).
I tought this too, but I looked at /etc/passwd and all was OK.
I had encripted password directly in /etc/passwd, no /etc/shadow.. not MD5
Also, why "su" worked if the password file was corrupted?
Instead I think it was a combination of incompatible versions of "hurd"
(so the "login" program
and the password server) and "glibc":
upgrading the hurd all gone OK.. so..
1) or the bug has been fixed yet
2) or that is one of the strange things of life: so common on
computers.. in this case it's probably better try some
debugging on the evil program: maybe there's a sleepy bug!
but if i don't know which one I prefer to not lose my time since the
latest version of hurd works.