On Thu, Apr 08, 2021 at 01:51:11PM +0100, David Howells wrote: > Hi Willy, Jeff, > > I think we need the attached change to readahead_expand() to fix the oops seen > when it tries to dereference ractl->ra when called indirectly from > netfs_write_begin(). > > netfs_write_begin() should also be using DEFINE_READAHEAD() rather than > manually initialising the ractl variable so that Willy can find it;-). ACK. Please fold into the appropriate patches.
