On 1/23/17 6:13 PM, Darrick J. Wong wrote: >> So I guess the question is: If the superblock stripe geometry >> is inconsistent by the time we get to is_multidisk_filesystem(), >> what should we do? >> >> 1) catch it earlier in verify_sb, and copy from backup? >> 2) keep that as it is, but only return true here if all 3 are set? > I'm ok with running in slow mode if the stripe data seems garbage. > TBH I'd prefer to have as few asserts in the repair program as possible, > particularly since this is a performance optimization. yeah. > return xfs_sb_version_hasdalign(sbp) && sbp->sb_unit && sbp->s_width; Backing up, I guess it's really not all that important in this function, as long as we don't ASSERT. I was actually going to suggest returning true if all 3 were set, yes. It just seems that this all should have been consistent by the time we get here, having already gone through superblock verification routines. Might make sense to add it, but I guess that may as well be another patch. -Eric -- To unsubscribe from this list: send the line "unsubscribe linux-xfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
