Unix 6th Edition source oddity

The 6th Edition Unix kernel source code has a very weird struct:

/*
 * structure to access an
 * integer in bytes
 */
struct
{
    char    lobyte;
    char    hibyte;
};

That’s from /usr/sys/param.h, but it appears a couple of other times in the source tree.

This doesn’t make sense in modern day C, but in the C of 1975, all struct members were in a single namespace.

You can see remnants today. From man 3type stat on a typical Linux system:

struct stat {
   dev_t      st_dev;      /* ID of device containing file */
   ino_t      st_ino;      /* Inode number */
   mode_t     st_mode;     /* File type and mode */
   nlink_t    st_nlink;    /* Number of hard links */
   uid_t      st_uid;      /* User ID of owner */
   gid_t      st_gid;      /* Group ID of owner */
   dev_t      st_rdev;     /* Device ID (if special file) */
   off_t      st_size;     /* Total size, in bytes */
 ...
}

Elements of the struct have names st_dev, st_ino, rather than dev or device, inode. The st_ prefix ensured that all the various struct stat elements had the correct offset from a struct pointer. Back in 1975, you could use ->st_nlink to get an offset from any old pointer.

The weird, anonymous struct let Thompson and Ritchie access high and low bytes off any pointer address, p->hibyte, p->lobyte, which was important on a 16-bit CPU like the PDP-11s that 6th Edition Unix ran on.