On 07/16/2014 11:09 PM, Christian Groessler wrote:
> On 07/16/14 15:42, Pádraig Brady wrote:
>> Note another reason to _not_ apply the patch is that
>> requests to print the statistics can come async through SIGUSR1,
>> and thus increase the chances of inconsistent output.
>
>
> Sorry, I cannot follow. Which inconsistent output are you referring to?
>
> regards,
> chris
It's a bit of an edge case, but if working with 1024 base quantities,
rarely one might get 1000 based statistics as the selector is essentially:
if ((n_written % 1000) && ! (n_written % 1024))
human_opts |= human_base_1024;
So if SIGUSR1 was sent after 1000 blocks were written for example,
then SI stats would be printed rather than IEC.
Yes it's quite the edge case, and not especially problematic I think,
but worth mentioning.
thanks,
Pádraig.
