>>> On 7/4/2014 at 01:59 AM, David Henningsson <david.henningsson at canonical.com>
>
> On 2014-07-02 22:12, Scott Reeves wrote:
>>>>> On 7/1/2014 at 12:12 AM, David Henningsson <david.henningsson at canonical.com>
>> wrote:
>>> From: Ricardo Salveti de Araujo <ricardo.salveti at canonical.com>
>>>
>>> Since rtkit v11, the top limit for rttime is 200 ms (previously it
>>> was wrongly limited to 2 seconds).
>>>
>>
>> Hi David,
>>
>> Just want to verify my understanding of this as far as not needing to
> backport this commit. Testing this issue on SUSE I don't see us hitting this
> because of the clamping to rtkit_get_rttime_usec_max() that set_scheduler()
> performs ?
>>
>> Thanks,
>> Scott.
>
> We hit this when upgrading from rtkit 0.10 to rtkit 0.11. What version
> of rtkit are you using?
We are using rtkit 0.11. I fully agree with the patch as far as setting the values correct in the first place. I was just asking about the need to backport this to existing installations. At least on openSUSE I don't believe we need to do that as the incorrect values are caught by the check in set_scheduler()
rttime = rtkit_get_rttime_usec_max(bus);
r = getrlimit(RLIMIT_RTTIME, &rl);
if (r >= 0 && (long long) rl.rlim_max > rttime) {
pa_log_info("Clamping rlimit-rttime to %lld for RealtimeKit\n", rttime);
rl.rlim_cur = rl.rlim_max = rttime;
...
