On Mon, Jun 23, 2008 at 11:31:30AM -0700, Waskiewicz Jr, Peter P wrote:

quoted

Commit d62733c8e437fdb58325617c4b3331769ba82d70
([SCHED]: Qdisc changes and sch_rr added for multiqueue)
added a NET_SCH_RR option that was unused since the code
went unconditionally into sch_prio.

I don't agree we should remove this option, purely for the fact that
there isn't any way a user would know the qdisc existed without reading
the code.

Most users of the kernel will anyway use a distribution kernel, and 
therefore never see it.

If the problem is a lack of any user documentation then kconfig is wrong 
place to attack that.

I'd rather see a change to sch_prio.c to correctly wrap the
RR code, which I should have done in the first place.  My hurdle is when
the module is selected, NET_SCH_RR isn't defined to be used as an #ifdef
check in the code. Only when it's selected to be built-in is the symbol
defined.  If I can get around that, I'll generate a patch to properly
wrap sch_prio.c code.  I know I'm missing something silly, so any
guidance is appreciated.

In this case NET_SCH_RR is not a module but a feature inside 
NET_SCH_PRIO, and that can be expressed as follows:

config NET_SCH_RR
        bool "Multi Band Round Robin Queuing (RR)"
        depends on NET_SCH_PRIO

But I doubt making these #ifdef's makes much sense.

IMHO the best you can do is to add a note about RR to the NET_SCH_PRIO 
help text.

-PJ Waskiewicz

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed