On Sun, Aug 26, 2018 at 10:30 PM Miklos Szeredi <miklos@xxxxxxxxxx> wrote: > > On Sun, Aug 26, 2018 at 6:25 PM, Amir Goldstein <amir73il@xxxxxxxxx> wrote: > > For an overlayfs file/inode, gage io is operating on the real underlying > > file, so the readahead hints set by fadvise64() should also be set on the > > real underlying file to take affect. > > Hmm, how about making this be an f_op? > > Would also fix readahead(2), which can be translated to fadvise(..., > POSIX_FADV_WILLNEED). > Ok. I guess we could match readahead(2) semantics to an advise for fs that implements f_op->advise(). Semantics are not currently the same for the case of !mapping->a_ops->readpages, but they are already the same (i.e. both return 0) for dax_mapping(mapping). Thanks, Amir.
