Skip to contentSkip to conversation

afs: stop using time_t for internal times

PATCH RESEND New Arnd Bergmann 1 participant 2 messages Jul 13, 2018 Afs filesystem

lore
Share & tools
lore
#0 lore raw

PATCH RESEND v2 1/2 afs: stop using time_t for internal times

afs uses 32-bit timestamps everywhere, but mixes signed and unsigned usage, which is a bit inconsistent. In particular on 32-bit machines, it currently uses unsigned timestamps (ranging from 1970 to 2106) for locally modified files, but signed timestamps (rand 1902 to 2038) when reading from a remote end. On 64-bit machines, we always interpret timestamps as unsigned here.

This replaces the deprecated time_t with a new explicitly unsigned afs_time32_t to get a consistent interpretation of inode times according the the wire protocol definition.

This avoids the y2038 overflow on 32-bit machines, extending the range to the end of the afs_time32_t in year 2106. On 64-bit machines, using the shorter type saves a few bytes for each afs_file_status and afs_volsync saves a few bytes over time_t or time64_t.

Note that mtime_server and struct afs_volsync are not currently used in any meaningful way and could be removed completely.

---
Originally submitted this on June 20, but got no reply. Resending this in case it got lost somewhere on the way. David, can you pick up both patches?

---
 fs/afs/afs.h | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
fs/afs/afs.h +4 -3
index b4ff1f7ae4ab..a17f4ce06323 100644
@@ -26,6 +26,7 @@
 typedef unsigned			afs_volid_t;
 typedef unsigned			afs_vnodeid_t;
 typedef unsigned long long		afs_dataversion_t;
+typedef unsigned			afs_time32_t;
 
 typedef enum {
 	AFSVL_RWVOL,			/* read/write volume */
@@ -129,8 +130,8 @@ typedef u32 afs_access_t;
 struct afs_file_status {
 	u64			size;		/* file size */
 	afs_dataversion_t	data_version;	/* current data version */
-	time_t			mtime_client;	/* last time client changed data */
-	time_t			mtime_server;	/* last time server changed data */
+	afs_time32_t		mtime_client;	/* last time client changed data */
+	afs_time32_t		mtime_server;	/* last time server changed data */
 	unsigned		abort_code;	/* Abort if bulk-fetching this failed */
 
 	afs_file_type_t		type;		/* file type */
@@ -158,7 +159,7 @@ struct afs_file_status {
  * AFS volume synchronisation information
  */
 struct afs_volsync {
-	time_t			creation;	/* volume creation time */
+	afs_time32_t		creation;	/* volume creation time */
 };
 
 /*
Signature · 1 line
2.9.0
#1 lore raw

PATCH RESEND v2 2/2 afs: avoid deprecated get_seconds()

get_seconds() has a limited range on 32-bit architectures and is deprecated because of that. While AFS uses the same limits for its inode timestamps on the wire protocol, let's just use the simpler current_time() as we do for other file systems.

This will still zero out the 'tv_nsec' field of the timestamps internally.

---
Originally submitted this on June 20, but got no reply. Resending this in case it got lost somewhere on the way. David, can you pick up both patches?

---
 fs/afs/inode.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
fs/afs/inode.c +1 -3
index 479b7fdda124..0507e52e3330 100644
@@ -213,9 +213,7 @@ struct inode *afs_iget_pseudo_dir(struct super_block *sb, bool root)
 	set_nlink(inode, 2);
 	inode->i_uid		= GLOBAL_ROOT_UID;
 	inode->i_gid		= GLOBAL_ROOT_GID;
-	inode->i_ctime.tv_sec	= get_seconds();
-	inode->i_ctime.tv_nsec	= 0;
-	inode->i_atime		= inode->i_mtime = inode->i_ctime;
+	inode->i_ctime = inode->i_atime = inode->i_mtime = current_time(inode);
 	inode->i_blocks		= 0;
 	inode_set_iversion_raw(inode, 0);
 	inode->i_generation	= 0;
Signature · 1 line
2.9.0