having a readahead of 128k with a msize of 128k (with overhead) lead to
reading 124+4k everytime, making two roundtrips needlessly.

tune readahead according to msize when cache is enabled for better
performance

Link: http://lkml.kernel.org/r/20211104120323.2189376-1-asmadeus@codewreck.orgSigned-off-by: Dominique Martinet <asmadeus@codewreck.org>