Date: Tue, 14 Oct 2003 04:17:34 +1000 (EST) From: Bruce Evans <bde@zeta.org.au> To: Harti Brandt <brandt@fokus.fraunhofer.de> Cc: sparc64@freebsd.org Subject: Re: time_t on sparc64 Message-ID: <20031014035805.F32262@gamplex.bde.org> In-Reply-To: <20031013153219.H45269@beagle.fokus.fraunhofer.de> References: <20031013153219.H45269@beagle.fokus.fraunhofer.de>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 13 Oct 2003, Harti Brandt wrote:
> I just discovered that time_t is 32-bit on sparc64. One of the problems
> is that struct timeval is defined by Posix as
>
> struct timeval {
> time_t tv_secs;
> suseconds_t tv_usecs;
> };
This is a bug in POSIX. In BSD, tv_secs has type long which may be,
and is different from time_t.
> but _timeval.h has
>
> struct timeval {
> long tv_secs;
> suseconds_t tv_usecs;
> }
>
> This means, that our timeval is not Posix compatible. What is the reason
> for time_t not beeing a long on sparc64?
time_t was used in some data structures whose layout shouldn't be changed
even for new arches. Mainly in ufs in Lite2:
%%%
ffs/fs.h: time_t fs_time; /* last time written */
ffs/fs.h: time_t cg_time; /* time last written */
ffs/fs.h: time_t cg_time; /* time last written */
lfs/lfs.h: time_t bi_segcreate; /* origin segment create time */
ufs/quota.h: time_t dqb_btime; /* time limit for excessive disk use */
ufs/quota.h: time_t dqb_itime; /* time limit for excessive files */
%%%
These are now:
%%%
ffs/fs.h: int32_t fs_old_time; /* last time written */
ffs/fs.h: ufs_time_t fs_time; /* last time written */
ffs/fs.h: int32_t cg_old_time; /* time last written */
ffs/fs.h: ufs_time_t cg_time; /* time last written */
/dev/null: time_t bi_segcreate; /* origin segment create time */
ufs/quota.h: int32_t dqb_btime; /* time limit for excessive disk use */
ufs/quota.h: int32_t dqb_itime; /* time limit for excessive files */
%%%
I.e., int32_t is now not mispelled time_t in f^Hufs1 and Y2.038K bugs are
fixed in ffs2 except for quotas.
ffs2 also parametrizes timestamps in inodes better:
%%%
ufs/dinode.h:typedef int64_t ufs_time_t;
ufs/dinode.h: ufs_time_t di_atime; /* 32: Last access time. */
ufs/dinode.h: ufs_time_t di_mtime; /* 40: Last modified time. */
ufs/dinode.h: ufs_time_t di_ctime; /* 48: Last inode change time. */
ufs/dinode.h: ufs_time_t di_birthtime; /* 56: Inode creation time. */
ufs/dinode.h: int32_t di_mtimensec; /* 64: Last modified time. */
ufs/dinode.h: int32_t di_atimensec; /* 68: Last access time. */
ufs/dinode.h: int32_t di_ctimensec; /* 72: Last inode change time. */
ufs/dinode.h: int32_t di_birthnsec; /* 76: Inode creation time. */
[Note that these aren't in a timespec struct, POSIX or otherwise, since the
struct would give MD packing which happens to be inefficient in most cases.]
ufs/dinode.h: int32_t di_atime; /* 16: Last access time. */
ufs/dinode.h: int32_t di_atimensec; /* 20: Last access time. */
ufs/dinode.h: int32_t di_mtime; /* 24: Last modified time. */
ufs/dinode.h: int32_t di_mtimensec; /* 28: Last modified time. */
ufs/dinode.h: int32_t di_ctime; /* 32: Last inode change time. */
ufs/dinode.h: int32_t di_ctimensec; /* 36: Last inode change time. */
[Y2.038K bugs are still in ffs1.]
%%%
To change time_t to 64 bits, all in-use non-transient data structures
need to be changed similarly.
Bruce
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20031014035805.F32262>