CODEWITHSUNDEEP
clinuxstat
Aaron Campbell
Aaron Campbell

Reputation: 612

major(stat.st_rdev) and minor(stat.st_rdev) on non-device files

Simple questions: in Linux, I stat() a file that is not a device. what is the expected value of the st_rdev field? Is it ok for me to run major(stat.st_rdev) and minor(stat.st_rdev)? If so, what are the expected return values?

Code example:

struct stat sstat = stat("myfileordevice");
ProcessFileOrDevice("myfileordevice",
                    sstat.st_mode,
                    major(stat.st_rdev),
                    minor(stat.st_rdev));

Upvotes: 2

Views: 8045

Answers (1)

Nominal Animal
Nominal Animal

Reputation: 39426

What value stat.st_rdev should have if I stat() a non-device filesystem entry?

It depends on the filesystem that the entry is on. The internal structures and functions in Linux (such asfs/stat.c) allow each filesystem to define the values.

In general, the value should be zero. Quick testing (stat -c '%t:%T' some-files-and-directories) indicates it tends to be. I just cannot find any guarantees or claims it should be zero.

Is it safe to use major(stat.st_rdev) and minor(stat.st_rdev)

Of course; they only apply some arithmetic to the specified value, and return the result.

Personally, I'd be inclined to use something like

    unsigned int maj, min;
    struct stat  info;

    /* stat(), fstat(), fstatat() or something
       to fill the info buffer */

    if (S_ISCHR(info.st_mode) || S_ISBLK(info.st_mode)) {
        maj = major(info.st_rdev);
        min = minor(info.sr_rdev);
    } else {
        maj = 0U;
        min = 0U;
    }

    ProcessFileOrDevice("myfileordevice", info.st_mode, maj, min);

just to make sure. Note that the ProcessFileOrDevice() function could just as well do the check itself, and ignore maj and min values unless S_ISCHR(info.st_mode) or S_ISBLK(info.st_mode) values are true.

The man 2 stat man page contains further useful details.

Upvotes: 4

Related Questions