Hello, this patch series makes UDF use the same address_space_operations for both normal and in-ICB files as switching aops on live files is prone to races as spotted by syzbot. When already dealing with this code, switch readpage, writepage, in-ICB expanding functions from using kmap_atomic() to use kmap_local_page(). I plan to queue these patches to my tree. Changes since v1: * Added Reviewed-by tags from Christoph * Added cleanup patches for udf_adinicb_writepage() and udf_expand_file_adinicb() Honza Previous versions: Link: http://lore.kernel.org/r/20230124120221.31585-1-jack@xxxxxxx # v1
