On Wed, Jun 22, 2016 at 6:54 AM, Arnd Bergmann <a...@arndb.de> wrote: > On Sunday, June 19, 2016 5:27:08 PM CEST Deepa Dinamani wrote: >> mutex_lock(&sbi->s_alloc_mutex); >> lvidiu->impIdent.identSuffix[0] = UDF_OS_CLASS_UNIX; >> lvidiu->impIdent.identSuffix[1] = UDF_OS_ID_LINUX; >> + ktime_get_real_ts(&ts); >> udf_time_to_disk_stamp(&lvid->recordingDateAndTime, >> - CURRENT_TIME); >> + timespec_trunc(ts, sb->s_time_gran)); >> lvid->integrityType = cpu_to_le32(LVID_INTEGRITY_TYPE_OPEN); >> >> lvid->descTag.descCRC = cpu_to_le16( >> > > I think we don't need the timespec_trunc here, and introducing the > call might complicate matters in the future.
This was done so that it could be similar to inode timestamps as both have same format and use same formatting functions. > IMHO timespec_trunc() really only makes sense when assigning into > an inode timestamp, whereas udf_time_to_disk_stamp() already truncates > the resulting nanoseconds to microseconds. I reconsidered our discussion. I think you are correct. I missed that the prints using microseconds format would actually convert seconds also to microseconds and not just nanoseconds. I will remove the timespec_trunc here and post a v3. Thanks, -Deepa
