Luis Henriques <lhenriques@xxxxxxx> wrote: > [ I wonder why the timestamps don't match between the traces and the > kernel log... ] I've seen that. I wonder if the timestamping of printk lines is delayed by the serial driver outputting things. > So, can we infer from this trace that an evict could actually be on-going > but the old cookie wasn't relinquished yet and hence the collision? It might be illuminating if you can make it print a traceline at the beginning of v9fs_evict_inode() and in v9fs_drop_inode(). Print the cookie pointer in both. David
