BUG: cpufreq_proc_read give prematures eof for /proc/cpufreq (in 2.4.x-ac and 2.5)

If a process reads from /proc/cpufreq with a small buffer, i.e. count < len, the function return always eof because len is always zero if the offset is > 0:

static int cpufreq_proc_read (
char *page,
...

{
    char            *p = page;

...

    if (off != 0)
        goto end;
    ...

end:

    len = (p - page);

where p = page.

Is this bug intentional to avoid overload or it's really a mistake?

PS: there is an obvious fix, I didn't attached to avoid you flaming me. :-)

  ricardo galli       GPG id C8114D34