Re: Glibc sunrpc failure?

[Jeff Bailey <jbailey at nisa dot net>]

On Sun, Jan 19, 2003 at 12:49:43PM -0800, Jeff Bailey wrote:

> > (gdb) print *stat_info
> > $5 = {st_fstype = 23, st_fsid = 63, st_ino = 130590, st_gen = 331987, 
> >   st_rdev = 142541374619648, st_mode = 1, st_nlink = 1000, st_uid = 1000, 
> >   st_gid = 2823, st_size = 4479340028585050112, st_atime = 990000, 
> >   st_atime_usec = 1042890342, st_mtime = 0, st_mtime_usec = 1042890342, 
> >   st_ctime = 268787712, st_ctime_usec = 8192, st_blksize = 8, st_blocks = 0, 
> >   st_author = 0, st_flags = 0, st_spare = {0, 0, 0, 0, 0, 0, 0, 18343548}}

> D'oh!  st_rdev looks like it ate st_mode, since st_nlink 1000 doesn't
> make sense, but if you push st_mode and friends to the right one it's
> happier.

Found it!  st_rdev changed sizes with Roland's Oct 23rd commit of
sysdeps/mach/hurd/bits/typesizes.h.  In the olden days:

bits/types.h:61:
typedef int __dev_t;            /* Type of device numbers.  */

Now:

sysdeps/generic/bits/types.h:
typedef __DEV_T_TYPE __dev_t;   /* Type of device numbers.  */

sysdeps/mach/hurd/bits/typesizes.h:
#define __DEV_T_TYPE            __UQUAD_TYPE

sysdeps/generic/bits/types.h:110:
#define        __UQUAD_TYPE            unsigned long long int

(There are possibly other problems in the struct, but they'll be easy to
track now that I know what to look for)

Are these changed intentionally, or do you want a patch to fix the sizes
to what they were before?

Tks,
Jeff Bailey