On Mon, Nov 27, 2023 at 08:12:19AM +0100, Christoph Hellwig wrote: > On Mon, Nov 27, 2023 at 12:32:38PM +0530, Ritesh Harjani wrote: > > > > > > i_size_read(inode) returns loff_t type. Can we make end_pos also as > > > loff_t type. We anyway use loff_t for > > > folio_pos(folio) + folio_size(folio), at many places in fs/iomap. It > > > would be more consistent with the data type then. > > > > > > Thoughts? > > > > aah, that might also require to change the types in > > iomap_writepage_map(). So I guess the data type consistency change > > should be a follow up change as this patch does only the refactoring. Separate patch for the cleanup, please. Then we can more easily target it for bisection in case there are signed comparison bugs. I hate C. > Yes, I'm trying to stay consistent in the writeback code. IIRC some > of the u64 use was to better deal with overflows, but I'll have to look > up the history. For this patch, Reviewed-by: Darrick J. Wong <djwong@xxxxxxxxxx> --D > >
