Re: [PATCH] 64 bit jiffies for 2.4.23-pre2

[Marcelo Tosatti]

On Sun, 31 Aug 2003, Tim Schmielau wrote:

> Hi Marcelo,
> 
> this patch makes the jiffies wrap after 497 days transparent to the user.
> While it allows uptimes of more than 497 days to be displayed, it's main
> use is to keep 'ps aux' output correct after the wrap. Without this patch,
> ps output will look very messy then with processes being reported as
> having started in the future.
> 
> For minimal intrusiveness it does not touch jiffies itself but introduces
> a (2.6 compatible) get_jiffies_64() function that keeps a private counter
> of the upper 31 bits, updated by a timer. user, nice, system, and idle
> times are handled similarely if the /proc filesystem is present.
> Process start time is extended to 64 bits.
> The patch should collapse to a no-op on 64 bit systems except for
> extending the per_cpu_user, per_cpu_nice, and per_cpu_system variables in
> struct kernel_stat to 64 bits (needed even there for correct display after
> 497 days).
> 
> The patch has been around for almost two years, discussed extensivly on
> lkml, tested with simulated uptime, and recently got its first real life
> success report (thanks Willy!).
> I think it's the most appropriate solution for 2.4. But you need to decide
> whether you take this, whether someone should backport the (intrusive and
> architecture-dependent) 2.6 fix, or whether you drop it completly and let
> people just upgrade to 2.6.

Sincerely, I prefer people use 2.6. 

Thanks

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/