xfs: widen ondisk inode timestamps to deal with y2038+

Redesign the ondisk inode timestamps to be a simple unsigned 64-bit
counter of nanoseconds since 14 Dec 1901 (i.e. the minimum time in the
32-bit unix time epoch).  This enables us to handle dates up to 2486,
which solves the y2038 problem.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Gao Xiang <hsiangkao@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>

8 files changed, 134 insertions, 10 deletions

diff --git a/fs/xfs/libxfs/xfs_format.h b/fs/xfs/libxfs/xfs_format.h
index d9b629f12c00..6aabe15786b8 100644
--- a/fs/xfs/libxfs/xfs_format.h
+++ b/fs/xfs/libxfs/xfs_format.h
@@ -467,6 +467,7 @@ xfs_sb_has_ro_compat_feature(
 #define XFS_SB_FEAT_INCOMPAT_FTYPE	(1 << 0)	/* filetype in dirent */
 #define XFS_SB_FEAT_INCOMPAT_SPINODES	(1 << 1)	/* sparse inode chunks */
 #define XFS_SB_FEAT_INCOMPAT_META_UUID	(1 << 2)	/* metadata UUID */
+#define XFS_SB_FEAT_INCOMPAT_BIGTIME	(1 << 3)	/* large timestamps */
 #define XFS_SB_FEAT_INCOMPAT_ALL \
 		(XFS_SB_FEAT_INCOMPAT_FTYPE|	\
 		 XFS_SB_FEAT_INCOMPAT_SPINODES|	\
@@ -565,6 +566,12 @@ static inline bool xfs_sb_version_hasreflink(struct xfs_sb *sbp)
 		(sbp->sb_features_ro_compat & XFS_SB_FEAT_RO_COMPAT_REFLINK);
 }
 
+static inline bool xfs_sb_version_hasbigtime(struct xfs_sb *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 &&
+		(sbp->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_BIGTIME);
+}
+
 /*
  * Inode btree block counter.  We record the number of inobt and finobt blocks
  * in the AGI header so that we can skip the finobt walk at mount time when
@@ -858,6 +865,13 @@ struct xfs_agfl {
  * Therefore, the ondisk min and max defined here can be used directly to
  * constrain the incore timestamps on a Unix system.  Note that we actually
  * encode a __be64 value on disk.
+ *
+ * When the bigtime feature is enabled, ondisk inode timestamps become an
+ * unsigned 64-bit nanoseconds counter.  This means that the bigtime inode
+ * timestamp epoch is the start of the classic timestamp range, which is
+ * Dec 31 20:45:52 UTC 1901.  Because the epochs are not the same, callers
+ * /must/ use the bigtime conversion functions when encoding and decoding raw
+ * timestamps.
  */
 typedef __be64 xfs_timestamp_t;
 
@@ -880,6 +894,50 @@ struct xfs_legacy_timestamp {
 #define XFS_LEGACY_TIME_MAX	((int64_t)S32_MAX)
 
 /*
+ * Smallest possible ondisk seconds value with bigtime timestamps.  This
+ * corresponds (after conversion to a Unix timestamp) with the traditional
+ * minimum timestamp of Dec 13 20:45:52 UTC 1901.
+ */
+#define XFS_BIGTIME_TIME_MIN	((int64_t)0)
+
+/*
+ * Largest supported ondisk seconds value with bigtime timestamps.  This
+ * corresponds (after conversion to a Unix timestamp) with an incore timestamp
+ * of Jul  2 20:20:24 UTC 2486.
+ *
+ * We round down the ondisk limit so that the bigtime quota and inode max
+ * timestamps will be the same.
+ */
+#define XFS_BIGTIME_TIME_MAX	((int64_t)((-1ULL / NSEC_PER_SEC) & ~0x3ULL))
+
+/*
+ * Bigtime epoch is set exactly to the minimum time value that a traditional
+ * 32-bit timestamp can represent when using the Unix epoch as a reference.
+ * Hence the Unix epoch is at a fixed offset into the supported bigtime
+ * timestamp range.
+ *
+ * The bigtime epoch also matches the minimum value an on-disk 32-bit XFS
+ * timestamp can represent so we will not lose any fidelity in converting
+ * to/from unix and bigtime timestamps.
+ *
+ * The following conversion factor converts a seconds counter from the Unix
+ * epoch to the bigtime epoch.
+ */
+#define XFS_BIGTIME_EPOCH_OFFSET	(-(int64_t)S32_MIN)
+
+/* Convert a timestamp from the Unix epoch to the bigtime epoch. */
+static inline uint64_t xfs_unix_to_bigtime(time64_t unix_seconds)
+{
+	return (uint64_t)unix_seconds + XFS_BIGTIME_EPOCH_OFFSET;
+}
+
+/* Convert a timestamp from the bigtime epoch to the Unix epoch. */
+static inline time64_t xfs_bigtime_to_unix(uint64_t ondisk_seconds)
+{
+	return (time64_t)ondisk_seconds - XFS_BIGTIME_EPOCH_OFFSET;
+}
+
+/*
  * On-disk inode structure.
  *
  * This is just the header or "dinode core", the inode is expanded to fill a
@@ -1104,12 +1162,22 @@ static inline void xfs_dinode_put_rdev(struct xfs_dinode *dip, xfs_dev_t rdev)
 #define XFS_DIFLAG2_DAX_BIT	0	/* use DAX for this inode */
 #define XFS_DIFLAG2_REFLINK_BIT	1	/* file's blocks may be shared */
 #define XFS_DIFLAG2_COWEXTSIZE_BIT   2  /* copy on write extent size hint */
+#define XFS_DIFLAG2_BIGTIME_BIT	3	/* big timestamps */
+
 #define XFS_DIFLAG2_DAX		(1 << XFS_DIFLAG2_DAX_BIT)
 #define XFS_DIFLAG2_REFLINK     (1 << XFS_DIFLAG2_REFLINK_BIT)
 #define XFS_DIFLAG2_COWEXTSIZE  (1 << XFS_DIFLAG2_COWEXTSIZE_BIT)
+#define XFS_DIFLAG2_BIGTIME	(1 << XFS_DIFLAG2_BIGTIME_BIT)
 
 #define XFS_DIFLAG2_ANY \
-	(XFS_DIFLAG2_DAX | XFS_DIFLAG2_REFLINK | XFS_DIFLAG2_COWEXTSIZE)
+	(XFS_DIFLAG2_DAX | XFS_DIFLAG2_REFLINK | XFS_DIFLAG2_COWEXTSIZE | \
+	 XFS_DIFLAG2_BIGTIME)
+
+static inline bool xfs_dinode_has_bigtime(const struct xfs_dinode *dip)
+{
+	return dip->di_version >= 3 &&
+	       (dip->di_flags2 & cpu_to_be64(XFS_DIFLAG2_BIGTIME));
+}
 
 /*
  * Inode number format:
diff --git a/fs/xfs/libxfs/xfs_fs.h b/fs/xfs/libxfs/xfs_fs.h
index 84bcffa87753..2a2e3cfd94f0 100644
--- a/fs/xfs/libxfs/xfs_fs.h
+++ b/fs/xfs/libxfs/xfs_fs.h
@@ -249,6 +249,7 @@ typedef struct xfs_fsop_resblks {
 #define XFS_FSOP_GEOM_FLAGS_SPINODES	(1 << 18) /* sparse inode chunks   */
 #define XFS_FSOP_GEOM_FLAGS_RMAPBT	(1 << 19) /* reverse mapping btree */
 #define XFS_FSOP_GEOM_FLAGS_REFLINK	(1 << 20) /* files can share blocks */
+#define XFS_FSOP_GEOM_FLAGS_BIGTIME	(1 << 21) /* 64-bit nsec timestamps */
 
 /*
  * Minimum and maximum sizes need for growth checks.
diff --git a/fs/xfs/libxfs/xfs_ialloc.c b/fs/xfs/libxfs/xfs_ialloc.c
index 7c1ece5a5f05..974e71bc4a3a 100644
--- a/fs/xfs/libxfs/xfs_ialloc.c
+++ b/fs/xfs/libxfs/xfs_ialloc.c
@@ -2807,6 +2807,10 @@ xfs_ialloc_setup_geometry(
 	uint64_t		icount;
 	uint			inodes;
 
+	igeo->new_diflags2 = 0;
+	if (xfs_sb_version_hasbigtime(&mp->m_sb))
+		igeo->new_diflags2 |= XFS_DIFLAG2_BIGTIME;
+
 	/* Compute inode btree geometry. */
 	igeo->agino_log = sbp->sb_inopblog + sbp->sb_agblklog;
 	igeo->inobt_mxr[0] = xfs_inobt_maxrecs(mp, sbp->sb_blocksize, 1);
diff --git a/fs/xfs/libxfs/xfs_inode_buf.c b/fs/xfs/libxfs/xfs_inode_buf.c
index 4dc43be9ad96..c667c63f2cb0 100644
--- a/fs/xfs/libxfs/xfs_inode_buf.c
+++ b/fs/xfs/libxfs/xfs_inode_buf.c
@@ -157,14 +157,29 @@ xfs_imap_to_bp(
 	return 0;
 }
 
+static inline struct timespec64 xfs_inode_decode_bigtime(uint64_t ts)
+{
+	struct timespec64	tv;
+	uint32_t		n;
+
+	tv.tv_sec = xfs_bigtime_to_unix(div_u64_rem(ts, NSEC_PER_SEC, &n));
+	tv.tv_nsec = n;
+
+	return tv;
+}
+
 /* Convert an ondisk timestamp to an incore timestamp. */
 struct timespec64
 xfs_inode_from_disk_ts(
+	struct xfs_dinode		*dip,
 	const xfs_timestamp_t		ts)
 {
 	struct timespec64		tv;
 	struct xfs_legacy_timestamp	*lts;
 
+	if (xfs_dinode_has_bigtime(dip))
+		return xfs_inode_decode_bigtime(be64_to_cpu(ts));
+
 	lts = (struct xfs_legacy_timestamp *)&ts;
 	tv.tv_sec = (int)be32_to_cpu(lts->t_sec);
 	tv.tv_nsec = (int)be32_to_cpu(lts->t_nsec);
@@ -226,9 +241,9 @@ xfs_inode_from_disk(
 	 * a time before epoch is converted to a time long after epoch
 	 * on 64 bit systems.
 	 */
-	inode->i_atime = xfs_inode_from_disk_ts(from->di_atime);
-	inode->i_mtime = xfs_inode_from_disk_ts(from->di_mtime);
-	inode->i_ctime = xfs_inode_from_disk_ts(from->di_ctime);
+	inode->i_atime = xfs_inode_from_disk_ts(from, from->di_atime);
+	inode->i_mtime = xfs_inode_from_disk_ts(from, from->di_mtime);
+	inode->i_ctime = xfs_inode_from_disk_ts(from, from->di_ctime);
 
 	to->di_size = be64_to_cpu(from->di_size);
 	to->di_nblocks = be64_to_cpu(from->di_nblocks);
@@ -241,7 +256,7 @@ xfs_inode_from_disk(
 	if (xfs_sb_version_has_v3inode(&ip->i_mount->m_sb)) {
 		inode_set_iversion_queried(inode,
 					   be64_to_cpu(from->di_changecount));
-		to->di_crtime = xfs_inode_from_disk_ts(from->di_crtime);
+		to->di_crtime = xfs_inode_from_disk_ts(from, from->di_crtime);
 		to->di_flags2 = be64_to_cpu(from->di_flags2);
 		to->di_cowextsize = be32_to_cpu(from->di_cowextsize);
 	}
@@ -266,11 +281,15 @@ out_destroy_data_fork:
 /* Convert an incore timestamp to an ondisk timestamp. */
 static inline xfs_timestamp_t
 xfs_inode_to_disk_ts(
+	struct xfs_inode		*ip,
 	const struct timespec64		tv)
 {
 	struct xfs_legacy_timestamp	*lts;
 	xfs_timestamp_t			ts;
 
+	if (xfs_inode_has_bigtime(ip))
+		return cpu_to_be64(xfs_inode_encode_bigtime(tv));
+
 	lts = (struct xfs_legacy_timestamp *)&ts;
 	lts->t_sec = cpu_to_be32(tv.tv_sec);
 	lts->t_nsec = cpu_to_be32(tv.tv_nsec);
@@ -297,9 +316,9 @@ xfs_inode_to_disk(
 	to->di_projid_hi = cpu_to_be16(from->di_projid >> 16);
 
 	memset(to->di_pad, 0, sizeof(to->di_pad));
-	to->di_atime = xfs_inode_to_disk_ts(inode->i_atime);
-	to->di_mtime = xfs_inode_to_disk_ts(inode->i_mtime);
-	to->di_ctime = xfs_inode_to_disk_ts(inode->i_ctime);
+	to->di_atime = xfs_inode_to_disk_ts(ip, inode->i_atime);
+	to->di_mtime = xfs_inode_to_disk_ts(ip, inode->i_mtime);
+	to->di_ctime = xfs_inode_to_disk_ts(ip, inode->i_ctime);
 	to->di_nlink = cpu_to_be32(inode->i_nlink);
 	to->di_gen = cpu_to_be32(inode->i_generation);
 	to->di_mode = cpu_to_be16(inode->i_mode);
@@ -318,7 +337,7 @@ xfs_inode_to_disk(
 	if (xfs_sb_version_has_v3inode(&ip->i_mount->m_sb)) {
 		to->di_version = 3;
 		to->di_changecount = cpu_to_be64(inode_peek_iversion(inode));
-		to->di_crtime = xfs_inode_to_disk_ts(from->di_crtime);
+		to->di_crtime = xfs_inode_to_disk_ts(ip, from->di_crtime);
 		to->di_flags2 = cpu_to_be64(from->di_flags2);
 		to->di_cowextsize = cpu_to_be32(from->di_cowextsize);
 		to->di_ino = cpu_to_be64(ip->i_ino);
@@ -538,6 +557,11 @@ xfs_dinode_verify(
 	if (fa)
 		return fa;
 
+	/* bigtime iflag can only happen on bigtime filesystems */
+	if (xfs_dinode_has_bigtime(dip) &&
+	    !xfs_sb_version_hasbigtime(&mp->m_sb))
+		return __this_address;
+
 	return NULL;
 }
 
diff --git a/fs/xfs/libxfs/xfs_inode_buf.h b/fs/xfs/libxfs/xfs_inode_buf.h
index 3060ecd24a2e..536666143fe7 100644
--- a/fs/xfs/libxfs/xfs_inode_buf.h
+++ b/fs/xfs/libxfs/xfs_inode_buf.h
@@ -32,6 +32,11 @@ struct xfs_icdinode {
 	struct timespec64 di_crtime;	/* time created */
 };
 
+static inline bool xfs_icdinode_has_bigtime(const struct xfs_icdinode *icd)
+{
+	return icd->di_flags2 & XFS_DIFLAG2_BIGTIME;
+}
+
 /*
  * Inode location information.  Stored in the inode and passed to
  * xfs_imap_to_bp() to get a buffer and dinode for a given inode.
@@ -58,6 +63,12 @@ xfs_failaddr_t xfs_inode_validate_cowextsize(struct xfs_mount *mp,
 		uint32_t cowextsize, uint16_t mode, uint16_t flags,
 		uint64_t flags2);
 
-struct timespec64 xfs_inode_from_disk_ts(const xfs_timestamp_t ts);
+static inline uint64_t xfs_inode_encode_bigtime(struct timespec64 tv)
+{
+	return xfs_unix_to_bigtime(tv.tv_sec) * NSEC_PER_SEC + tv.tv_nsec;
+}
+
+struct timespec64 xfs_inode_from_disk_ts(struct xfs_dinode *dip,
+		const xfs_timestamp_t ts);
 
 #endif	/* __XFS_INODE_BUF_H__ */
diff --git a/fs/xfs/libxfs/xfs_sb.c b/fs/xfs/libxfs/xfs_sb.c
index 15d03d967537..5aeafa59ed27 100644
--- a/fs/xfs/libxfs/xfs_sb.c
+++ b/fs/xfs/libxfs/xfs_sb.c
@@ -1166,6 +1166,8 @@ xfs_fs_geometry(
 		geo->flags |= XFS_FSOP_GEOM_FLAGS_RMAPBT;
 	if (xfs_sb_version_hasreflink(sbp))
 		geo->flags |= XFS_FSOP_GEOM_FLAGS_REFLINK;
+	if (xfs_sb_version_hasbigtime(sbp))
+		geo->flags |= XFS_FSOP_GEOM_FLAGS_BIGTIME;
 	if (xfs_sb_version_hassector(sbp))
 		geo->logsectsize = sbp->sb_logsectsize;
 	else
diff --git a/fs/xfs/libxfs/xfs_shared.h b/fs/xfs/libxfs/xfs_shared.h
index 708feb8eac76..c795ae47b3c9 100644
--- a/fs/xfs/libxfs/xfs_shared.h
+++ b/fs/xfs/libxfs/xfs_shared.h
@@ -176,6 +176,9 @@ struct xfs_ino_geometry {
 	unsigned int	ialloc_align;
 
 	unsigned int	agino_log;	/* #bits for agino in inum */
+
+	/* precomputed value for di_flags2 */
+	uint64_t	new_diflags2;
 };
 
 #endif /* __XFS_SHARED_H__ */
diff --git a/fs/xfs/libxfs/xfs_trans_inode.c b/fs/xfs/libxfs/xfs_trans_inode.c
index 18c391d35a3b..90f1d5645052 100644
--- a/fs/xfs/libxfs/xfs_trans_inode.c
+++ b/fs/xfs/libxfs/xfs_trans_inode.c
@@ -132,6 +132,17 @@ xfs_trans_log_inode(
 	}
 
 	/*
+	 * If we're updating the inode core or the timestamps and it's possible
+	 * to upgrade this inode to bigtime format, do so now.
+	 */
+	if ((flags & (XFS_ILOG_CORE | XFS_ILOG_TIMESTAMP)) &&
+	    xfs_sb_version_hasbigtime(&ip->i_mount->m_sb) &&
+	    !xfs_inode_has_bigtime(ip)) {
+		ip->i_d.di_flags2 |= XFS_DIFLAG2_BIGTIME;
+		flags |= XFS_ILOG_CORE;
+	}
+
+	/*
 	 * Record the specific change for fdatasync optimisation. This allows
 	 * fdatasync to skip log forces for inodes that are only timestamp
 	 * dirty.