I experimented with your patch and could not find any noticeable
change in BW or interrupt rate so I guess we can use your ideas.
Still, note that NAPI_howto.txt does not read *budget to limit the
number of polls and the code bellow from kernel 2.6.17.7 takes into
account that budget can become negative.
static void net_rx_action(struct softirq_action *h)
{
struct softnet_data *queue = &__get_cpu_var(softnet_data);
unsigned long start_time = jiffies;
int budget = netdev_budget;
void *have;
local_irq_disable();
while (!list_empty(&queue->poll_list)) {
struct net_device *dev;
if (budget <= 0 || jiffies - start_time > 1)
goto softnet_break;
...
}
>I took a stab at implementing this myself, and it
>
> You might want to respin your patch against my for-2.6.19 branch
Do you think I should work on this or you plan to push your code?
>> The biggest problem I have with this is that I don't know what to
call
>> the feature bit. Any suggestions?
>Maybe set bit for the lack of the feature? REQUIRES_POLL_AFTER_ARM?
We can take Michael's suggestion or use NOT_REQUIRES_POLL_AFTER_ARM so
we can implement this for mthca without touching ipath or ehca at the
first step.
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
