> I've been picking a little more at the problem I noticed earlier today; I > just installed a patched init and ran 'while true; do dpkg-buildpackage; done' > to trigger the bug. It actually took quite a while and (notably) didn't > include any complaints about default-pager. In fact, all that I got was > several instances of the message "make: Computer bought the farm: Stop." > followed by a complaint by Mach that the essential task proc had died and an > attempt to reboot the computer (which of course didn't happen since I had > applied the patch to spin forever)
Note that the "essential task died" message comes from init, not from the kernel (that's why patching init to not reboot made a difference). This indeed sounds like a new problem, and looks like a proc bug. If you can find a simpler example that produces this result, please do. Either way, please send a report to the bug-tracking system about this, and include the complete and exact details of how to reproduce what you see. Thanks, Roland
