All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] ufs: use ktime_get_real_seconds for sb and cg timestamps
@ 2018-06-19 15:47 Arnd Bergmann
  2018-06-20  8:33 ` Thomas Gleixner
  0 siblings, 1 reply; 2+ messages in thread
From: Arnd Bergmann @ 2018-06-19 15:47 UTC (permalink / raw)
  To: Evgeniy Dushistov
  Cc: y2038, Arnd Bergmann, Thomas Gleixner, Jeff Layton,
	David Howells, linux-kernel

get_seconds() is deprecated because of the 32-bit overflow and will
be removed. All callers in ufs also truncate to a 32-bit number, so
nothing changes during the conversion, but this should be harmless as the
superblock and cylinder group timestamps are not visible to user space,
except for checking the fs-dirty state, wich works fine across the
overflow.

This moves the call to get_seconds() into a new inline function, with
a comment explaining the constraints, while converting it to
ktime_get_real_seconds().

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 fs/ufs/balloc.c |  4 ++--
 fs/ufs/ialloc.c |  2 +-
 fs/ufs/super.c  |  4 ++--
 fs/ufs/util.h   | 14 ++++++++++++++
 4 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/fs/ufs/balloc.c b/fs/ufs/balloc.c
index e727ee07dbe4..075d3d9114c8 100644
--- a/fs/ufs/balloc.c
+++ b/fs/ufs/balloc.c
@@ -547,7 +547,7 @@ static u64 ufs_add_fragments(struct inode *inode, u64 fragment,
 	/*
 	 * Block can be extended
 	 */
-	ucg->cg_time = cpu_to_fs32(sb, get_seconds());
+	ucg->cg_time = ufs_get_seconds(sb);
 	for (i = newcount; i < (uspi->s_fpb - fragoff); i++)
 		if (ubh_isclr (UCPI_UBH(ucpi), ucpi->c_freeoff, fragno + i))
 			break;
@@ -639,7 +639,7 @@ static u64 ufs_alloc_fragments(struct inode *inode, unsigned cgno,
 	if (!ufs_cg_chkmagic(sb, ucg)) 
 		ufs_panic (sb, "ufs_alloc_fragments",
 			"internal error, bad magic number on cg %u", cgno);
-	ucg->cg_time = cpu_to_fs32(sb, get_seconds());
+	ucg->cg_time = ufs_get_seconds(sb);
 
 	if (count == uspi->s_fpb) {
 		result = ufs_alloccg_block (inode, ucpi, goal, err);
diff --git a/fs/ufs/ialloc.c b/fs/ufs/ialloc.c
index e1ef0f0a1353..c678fff2a04d 100644
--- a/fs/ufs/ialloc.c
+++ b/fs/ufs/ialloc.c
@@ -89,7 +89,7 @@ void ufs_free_inode (struct inode * inode)
 	if (!ufs_cg_chkmagic(sb, ucg))
 		ufs_panic (sb, "ufs_free_fragments", "internal error, bad cg magic number");
 
-	ucg->cg_time = cpu_to_fs32(sb, get_seconds());
+	ucg->cg_time = ufs_get_seconds(sb);
 
 	is_directory = S_ISDIR(inode->i_mode);
 
diff --git a/fs/ufs/super.c b/fs/ufs/super.c
index 488088141451..a4e07e910f1b 100644
--- a/fs/ufs/super.c
+++ b/fs/ufs/super.c
@@ -698,7 +698,7 @@ static int ufs_sync_fs(struct super_block *sb, int wait)
 	usb1 = ubh_get_usb_first(uspi);
 	usb3 = ubh_get_usb_third(uspi);
 
-	usb1->fs_time = cpu_to_fs32(sb, get_seconds());
+	usb1->fs_time = ufs_get_seconds(sb);
 	if ((flags & UFS_ST_MASK) == UFS_ST_SUN  ||
 	    (flags & UFS_ST_MASK) == UFS_ST_SUNOS ||
 	    (flags & UFS_ST_MASK) == UFS_ST_SUNx86)
@@ -1342,7 +1342,7 @@ static int ufs_remount (struct super_block *sb, int *mount_flags, char *data)
 	 */
 	if (*mount_flags & SB_RDONLY) {
 		ufs_put_super_internal(sb);
-		usb1->fs_time = cpu_to_fs32(sb, get_seconds());
+		usb1->fs_time = ufs_get_seconds(sb);
 		if ((flags & UFS_ST_MASK) == UFS_ST_SUN
 		  || (flags & UFS_ST_MASK) == UFS_ST_SUNOS
 		  || (flags & UFS_ST_MASK) == UFS_ST_SUNx86) 
diff --git a/fs/ufs/util.h b/fs/ufs/util.h
index 1907be6d5808..1fd3011ea623 100644
--- a/fs/ufs/util.h
+++ b/fs/ufs/util.h
@@ -590,3 +590,17 @@ static inline int ufs_is_data_ptr_zero(struct ufs_sb_private_info *uspi,
 	else
 		return *(__fs32 *)p == 0;
 }
+
+static inline __fs32 ufs_get_seconds(struct super_block *sbp)
+{
+	time64_t now = ktime_get_real_seconds();
+
+	/* Signed 32-bit interpretation wraps around in 2038, which
+	 * happens in ufs1 inode stamps but not ufs2 using 64-bits
+	 * stamps. For superblock and blockgroup, let's assume
+	 * unsigned 32-bit stamps, which are good until y2106.
+	 * Wrap around rather than clamp here to make the dirty
+	 * file system detection work in the superblock stamp.
+	 */
+	return cpu_to_fs32(sbp, lower_32_bits(now));
+}
-- 
2.9.0


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH] ufs: use ktime_get_real_seconds for sb and cg timestamps
  2018-06-19 15:47 [PATCH] ufs: use ktime_get_real_seconds for sb and cg timestamps Arnd Bergmann
@ 2018-06-20  8:33 ` Thomas Gleixner
  0 siblings, 0 replies; 2+ messages in thread
From: Thomas Gleixner @ 2018-06-20  8:33 UTC (permalink / raw)
  To: Arnd Bergmann
  Cc: Evgeniy Dushistov, y2038, Jeff Layton, David Howells, linux-kernel

On Tue, 19 Jun 2018, Arnd Bergmann wrote:

> get_seconds() is deprecated because of the 32-bit overflow and will
> be removed. All callers in ufs also truncate to a 32-bit number, so
> nothing changes during the conversion, but this should be harmless as the
> superblock and cylinder group timestamps are not visible to user space,
> except for checking the fs-dirty state, wich works fine across the
> overflow.
> 
> This moves the call to get_seconds() into a new inline function, with
> a comment explaining the constraints, while converting it to
> ktime_get_real_seconds().

Acked-by: Thomas Gleixner <tglx@linutronix.de>

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2018-06-20  8:34 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-06-19 15:47 [PATCH] ufs: use ktime_get_real_seconds for sb and cg timestamps Arnd Bergmann
2018-06-20  8:33 ` Thomas Gleixner

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.