[15613] in Athena Bugs
SS5 processor bug -- userlevel processes can hang the machine?
daemon@ATHENA.MIT.EDU (John Hawkinson)
Sun Oct 26 12:51:23 1997
Date: Sun, 26 Oct 1997 12:51:20 -0500
To: bugs@MIT.EDU
From: John Hawkinson <jhawk@MIT.EDU>
Jered suggests I forward this here -- it was discussed around here
(i.e. on -c sipb) for some time before being mentioned on bugtraq; it
would be nice if we could contact Sun about it.
It does appear to be a processor bug -- an annulled branch (,a -- means
"don't execute the delay slot instruction") that branches to itself
causes the machine to hang hard. The instruction in the delay slot
doesn't matter.
Only happens on SS5/170s, not on SS5/110s or SS5/70s. I couldn't
find any '85s to test it on.
--jhawk
[5435] daemon@ATHENA.MIT.EDU (Charles M. Hannum) bugtraq 10/24/97 19:29 (17 lines)
Subject: Cute SPARC CPU bug
Date: Fri, 24 Oct 1997 13:42:32 -0400
Reply-To: "Charles M. Hannum" <mycroft@MIT.EDU>
From: "Charles M. Hannum" <mycroft@MIT.EDU>
To: BUGTRAQ@NETSPACE.ORG
Compile the following:
main(){while(1);}
with `gcc -O' on a SPARC. You get, roughly:
L1:
b,a L1
retl
Run it on a SS5/170. Notice the machine is completely wedged.
Power-cycle it.
--[5435]-- (pref = [5434], nref = [5445])
