Richard L. Hamilton writes: > _FIOSATIME - why doesn't zfs support this (assuming I didn't just miss it)? > Might be handy for backups.
Roch Bourbonnais writes: > Are these syscall sufficent ? > int utimes(const char *path, const struct timeval times[2]); > int futimesat(int fildes, const char *path, const struct timeval times[2]); No, the difference is that utimes() and futimesat() also update ctime, while the _FIOSATIME ioctl only changes the atime (requires privilege). This allows backup software or file scanning software to save and restore the original atime after processing, where changing the ctime is undesirable. Unfortunately, according to the Solaris source browser, the _FIOSATIME ioctl is only implemented for ufs, not for zfs. For example, we use a locally-written file scanner to detect changes to file checksums, and would like to preserve atime and ctime. I was very happy to discover the _FIOSATIME ioctl, even though it is not an officially supported interface. Also, "star" backup software uses this capability. Assuming _FIOSATIME continues to work on ufs, it would make sense to implement on zfs and other filesystems for consistency. I see there is a separate thread on implementing O_NOATIME for open() or fcntl(), which would be compatible with Linux (and would avoid updating the atime at all, rather than updating it twice). http://www.opensolaris.org/jive/thread.jspa?messageID=195813 This presumably would also require support from the filesystem. I assume that an existing _FIOSATIME ioctl could be implemented for zfs and available sooner than a new O_NOATIME flag for open/fcntl, although uniform support for O_NOATIME makes more sense in the long run, especially for cross-platform compatibility. This message posted from opensolaris.org _______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
