On Fri, Apr 3, 2015, at 07:49 PM, Adam Wolk wrote: > Hi bugs@ > > I'm trying to install OpenBSD -current snapshot from 01-Apr-2015. > The snapshot was obtained from > http://piotrkosoft.net/pub/OpenBSD/snapshots/ > as install56.fs with the following sha512 checksum: > > $ sha512 install57.fs > SHA512 (install57.fs) = > ca4a79f02cc69680987a0a55385e5419f6028666a15df4f45832b4e0c7d3e4468cd7a7db145708f65b6284d2c4d8c487b140630b756fa62400fee08f7fe814a9 > > This is a lenovo 50-70 laptop which I already reported as not working > for wifi, ethernet and some startx failure both in dmesg & a thread on > misc@ (http://marc.info/?l=openbsd-misc&m=142807374616596&w=2). > > Trying to diagnose the not working re0 I hit a reproducible kernel > panic. Details follow (manually transcribed :(). This machine has no > network but I should be able to get sources on to it via sneaker net usb > & I'm willing to test patches (including debugging printf patches)/help > out with debugging this. > > The bug occurs when sh /etc/netstart is executed as root. Please note > that during the installation or boot the system tries to obtain a dhcp > lease from the router but fails without an offer - this does not result > in a panic. The only time the system panics is when netstart is executed > manually. > > dmesg for this snapshot is attached after the ddb session transcript. > > # cat /etc/hostname.re0 > dhcp
One new data point. If the hostname.re0 file is present, after a significant amount of tme after boot (logging in and leaving the laptop on for 10-15m) the watchdog triggers for re0 leading to the exact same kernel panic (trace is exactly the same). So the bug is probably not limite to /etc/netstart just faster to trigger that way. There are no panics if hostname.re0 doesn't exist - even if /etc/netstart is executed. Regards, Adam
