su, 2005-03-27 kello 02:06 +0100, Kenneth Aafl?y kirjoitti: > Call on when you know how firmware.agent is behaving, Crapola. I got a big honking hunch on the problem after eyeing the script, and just had to boot to check (even though it interrupted something, but that can be redone). And lo and behold: The culprit all along was that I have /sys bind-mounted into my x86 chroot. And the kernel is too smart for its own good and marks this as a sysfs mount in /proc/mounts. And here's how firmware.agent decides where to look for sysfs: SYSFS=$(sed -n 's/^.* \([^ ]*\) sysfs .*$/\1/p' /proc/mounts) Now, on this system that returned two lines. What fun. After I unmounted the chroot sysfs version, just plugging the device in worked flawlessly. (This should probably be considered a hotplug bug; perhaps it should just add a head -1 to the above or something. I'll report.) So, you were right all along. Thanks to everyone involved. -- Mikko Rauhala - mjr@xxxxxx - <URL:http://www.iki.fi/mjr/> Transhumanist - WTA member - <URL:http://www.transhumanism.org/> Singularitarian - SIAI supporter - <URL:http://www.singinst.org/>
