Andy Green wrote: > and both are themselves canned actions. There is a set of common canned solutions, such as changing the loglevels, adding or removing consoles, increasing the dmesg buffer, or overriding init, but there's no telling what happens in the next iteration of debugging.
E.g., if some people have problems with the SD driver and you add an option to vary the timing, that's probably something you can't easily predict. And if your rootfs is on SD, handling this when user space is up may not be an option either. > Maybe there's a more adaptive > Windows 95 style solution possible where the device-specific code can > mark something like PMU spare register bit in Qi and clear it in > userspace... if we see the bit set next time in Qi we append these > troubleshooting options ourselves in Qi. Hmm, but would this really add anything that can't also be done through the manually activated secondary boot choice ? - Werner
