[BusyBox] Serious bug in udhcpc

Wolfgang Denk wd at denx.de
Tue Apr 22 07:19:45 UTC 2003


In message <1050985092.1476.13.camel at russ.local> you wrote:
>
> wait, so the kernel actually boots, looks at the uninitialized hardware,
> thinks for a moment, sticks its arm into the /dev/urandom bag, pulls out
> a winner, and sticks it into seconds since epoch? On all the embedded
> systems I've seen without a battery backed RTC, the clock is set to
> epoch on boot. That would be really broken behavior to never initialize
> the RTC, or to initialize it to a random value.

You haven't seen many embedded systems, then.

Uninitialized time is pretty usual on such systems. In quite a lot of
applications the actual date / time does  not  play  a  role  and  so
nobody bothers about it.

There are even embedded systems _with_ a RTC which come  up  with  an
uninitialized  value.  For  example,  the  MPC8xx  family  of PowerPC
processors has an internal RTC, which can be supplied from a  battery
nbackup,  but  so  only a _very_ small percentage of systems actually
provides a battery. On such systems, after poweron there is a  random
value in the RTC register.

This is normal, nothing wrong here.


If there is something broken  here  than  it  is  the  software  that
depends  on  being  started  after  a certain date. Remember that all
values of a time_t are legal  values;  times  <  1970  are  perfectly
legal, too.

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
I have been over into the future, and it works.
                - Lincoln Steffens in _Letters_ (1938) vol. 1, p. 463


More information about the busybox mailing list