On 11/5/21 7:22 AM, Dave Chinner wrote:
No. Some filesystems don't track inode metadata dirty status using the VFS inode; instead they track it more efficiently in internal inode and/or journal based structures. Hence the only way to get "inode needs journal flush for data stability" information to generic IO code is to have a specific per-IO mapping flag for it.
Could we add IOMAP_REPORT_DIRTY flag in the flags field of struct iomap_iter to indicate whether the IOMAP_F_DIRTY flag needs to be set or not? Currently the IOMAP_F_DIRTY flag is only checked in iomap_swapfile_activate(), dax_iomap_fault() and iomap_dio_rw() (To be more specific, only the write path in dax_iomap_fault() and iomap_dio_rw()). So it would be unnecessary to set the IOMAP_F_DIRTY flag in dax_iomap_rw() called in the previous tests. Other file systems that set the IOMAP_F_DIRTY flag efficiently could ignore the IOMAP_REPORT_DIRTY flag. Best, Zhongwei.
