svn commit: r216955 - head/usr.sbin/rtprio
swell.k at gmail.com
Thu Jan 6 23:58:29 UTC 2011
Garrett Cooper <yanegomi at gmail.com> writes:
> On Jan 6, 2011, at 2:41 PM, Alexander Best wrote:
>> On Fri Jan 7 11, Anonymous wrote:
>>> Alexander Best <arundel at freebsd.org> writes:
>>>> On Thu Jan 6 11, John Baldwin wrote:
>>>>> Note that that usage is rather pointless since it means you apply rtprio to
>>>>> the 'rtprio' process that is about to exit. :)
>>>> yeah but at least it makes the usage of -X consistent. ;) also consider the
>>>> following: the current shell has idle priority and you want to run rtprio in
>>>> normal priority. then rtprio -t -0 would be a neat way of doing
>>>> rtprio -t rtprio. ;) wel...not quite, because the priotity gets set to "NORMAL"
>>>> when rtprio is almost finished running. ;)
>>> I think it'd be useful if the syntax allowed smth like
>>> $ rtprio 1 -0 -111 -222 -333 -444 -555 ...
>> defenately, but that would require quite some code. also please bear in mind:
>> in its current form rtprio *DOES* process -0. my code doesn't change that. the
>> only thing that it changes is that before hand -0 was processed *AND* then also
>> executed. now the execution doesn't take place.
> Same thing, no code change:
> sh -c 'for i in 1 -0 -111 -222 -333 -444 -555; do rtprio $i; done'
> Yes, there's more of a processing cost to doing it this way
> with exec/fork jazz and shell logic
I was thinking about rtprio(1) raising its own priority using syntax like
$ rtprio num -0 ...
$ rtprio -t -0 ...
so that subsequent calls to rtprio(2) are under new priority.
Not sure if it makes difference on heavily loaded system.
> -- but how often do you execute
> rtprio, and is the required code change really necessary? Simple is
> better in my book.
More information about the svn-src-head