Messages in this thread | | | Date | Wed, 1 Aug 2012 08:50:12 +0200 (CEST) | From | Thomas Gleixner <> | Subject | Re: boot panic regression introduced in 3.5-rc7 |
| |
On Mon, 30 Jul 2012, John Stultz wrote: > On 07/29/2012 08:51 PM, CAI Qian wrote: > Now, one of the reasons Thomas and I changed the logic was that using the > precalculated realtime_offset was slightly more efficient then re-adding xtime > and wall_to_monotonic's components separately. But how valuable this > unmeasured slight efficiency is vs extra robustness for crazy time values is > questionable.
Well, I guess it unearthed a weakness which has been there forever:
Trusting random values which are supplied by cmos or whatever.
So the right fix is:
> 2) Validate that time values we accept are smaller the ktime_t before using > them.
I really don't like the magic workaround by nulling out the effect of crap data with a more expensive calculation.
Thanks,
tglx
| |