On Thu, Apr 25, 2002 at 08:50:45PM -0600, Ronald G Minnich wrote: > > consider this case. We drop from linux back into linuxbios. If linuxbios > knows it is a reboot, it can just add to the log buffer instead of > resetting it. Otherwise, linuxbios might have to assume that the log > buffer is invalid every time it is started.
There's probably clever ways to detect a valid log buffer, and figure out where the end is, using magic numbers. That might avoid requiring knowledge of reboot vs. powercycle. For instance, record a magic number at the beginning to mark the buffer as valid. Then (assuming it's a circular buffer), force placement of a tag every X bytes that identifies (a) you're still reading valid entries, (b) this was log message yyy (yyy increments). If you're parsing the buffer and yyy goes backwards, you know where the buffer wrapped.
