[vdr] Re: [PATCH] time warp

[Rainer Zocholl]

marcel.wiesweg at gmx.de(Marcel Wiesweg)  22.06.05 19:09


>Am Mittwoch 22 Juni 2005 00:10 schrieb Rainer Zocholl:
>> udo_richter at gmx.de(Udo Richter)  19.06.05 16:40
>>>Klaus Schmidinger wrote:
>>>> Isn't all that data protected by CRC32 checksums?
>>>> How could "garbage" get in there?
>>>>
>>>> Maybe somebody should try to find out why that "garbage"
>>>> is produced in the first place...
>>>
>>>I've seen this - though at least one year ago - with bad reception
>>>due to bad weather. For half an hour, the system time was jumping
>>>back and forward like crazy, starting and stopping some timers.
>>>Luckily, all timers were repeating, so none of them was gone after
>>>reception was back.
>>
>> So: Who is doing the CRC checks?
>> Maybe it's turned off or only an error bit is set but nobody
>> cares about in further processing? (Maybe the cause for
>> "unkown picture type" too?)
>>
>> It's very unlikely that garbage can so often pass a CRC32 check.

>The TDT is not specified to contain a CRC32 checksum.

That makes (a bit) sense because that information is "transient" anyway
(what's the use of a time 10 minutes ago when i woudl have to know it
was 10min ago??)

Ok, then we can "forget" to use raw TDT to set system time.

>All larger tables have this, NIT, EIT, SDT, also TOT.
>The "small" ones TDT and RST have no checksum.

So the only way use the time table would be to
implement tests like "ntp" uses.

(taking at least 3 probes, limit the time skip to only few seconds etc.)


Rainer