FreeBSD entry on Wikipedia

[Andy Gilligan]

On 22 Jun 2005, at 21:11, Vadim Goncharov wrote:
> AG> On 21 Jun 2005, at 01:38, Bob Martin wrote:
>>> IMHO, if they can roll out a patch to a major security flaw in a
>>> day, they should have been able to fix the uptime clock at some
>>> point in the last decade. Odd that MS can do something that Linux
>>> can't.
>>>
> AG> This was fixed about 3 years ago.
>
> Sure? So why Netcraft states about it aven nowadays? Even more, it  
> says
> that some versions of FreeBSD (newer, as I understood) also wrap that
> uptime counter?

Linux used to wrap at 497 days, but there were a few changes in 2.5.x to
change the counters to 64-bit. -- In theory 2.6.x should be fine.

I'm not 100% sure when or if this was backported to a 2.4 kernel  
release,
but the patches have been around since early 2002.  (around 2.4.19)
-- http://lwn.net/2002/0307/a/uptime.php3

As for why Netcraft is reporting it... I guess it's probably safe enough
to say that "most" linux machines will wrap at 497 days, as not everyone
patches or upgrades their kernel.

On a more personal note, I don't believe having a long uptime is always
a good thing... If there are indeed Linux machines with uptimes of over
1000 days, they are no doubt full of security holes, but this applies
equally to any OS.

And yes, nearly all my machines are FreeBSD, before anyone asks ;)

Best regards,
-Andy