Upon strolling through my ccache log files i noticed numerous entries like: Placed /var/tmp/portage/glibc-2.3.2-r9/work/glibc-2.3.2/buildhere/stdlib/dbl2mpn.os into cache ccache is using this (full) filename to compute the hash which determines a hit/miss, so even small revisionbumps will never generate hits from ccache. Reproducible: Always Steps to Reproduce: emerge a new glibc Actual Results: compiler results are freshly generated Expected Results: use the results from previous compile runs.
Indeed, this is neither a gcc bug nor a ccache issue. A solution is being provided by cachecc1.
