Chris Wilson writes:
> Quoting Francisco Jerez (2020-03-23 22:30:13)
>> Chris Wilson writes:
>>
>> > Quoting Francisco Jerez (2020-03-20 22:14:51)
>> >> Francisco Jerez writes:
>> >>
>> >> > Chris Wilson writes:
>> >> >
>> >> >> We dropped calling process_csb prior to handling direct submiss
Quoting Francisco Jerez (2020-03-23 22:30:13)
> Chris Wilson writes:
>
> > Quoting Francisco Jerez (2020-03-20 22:14:51)
> >> Francisco Jerez writes:
> >>
> >> > Chris Wilson writes:
> >> >
> >> >> We dropped calling process_csb prior to handling direct submission in
> >> >> order to avoid the
Chris Wilson writes:
> Quoting Francisco Jerez (2020-03-20 22:14:51)
>> Francisco Jerez writes:
>>
>> > Chris Wilson writes:
>> >
>> >> We dropped calling process_csb prior to handling direct submission in
>> >> order to avoid the nesting of spinlocks and lift process_csb() and the
>> >> major
Quoting Francisco Jerez (2020-03-20 22:14:51)
> Francisco Jerez writes:
>
> > Chris Wilson writes:
> >
> >> We dropped calling process_csb prior to handling direct submission in
> >> order to avoid the nesting of spinlocks and lift process_csb() and the
> >> majority of the tasklet out of irq-of
Quoting Chris Wilson (2020-03-20 17:47:45)
> We dropped calling process_csb prior to handling direct submission in
> order to avoid the nesting of spinlocks and lift process_csb() and the
> majority of the tasklet out of irq-off. However, we do want to avoid
> ksoftirqd latency in the fast path, so
Francisco Jerez writes:
> Chris Wilson writes:
>
>> We dropped calling process_csb prior to handling direct submission in
>> order to avoid the nesting of spinlocks and lift process_csb() and the
>> majority of the tasklet out of irq-off. However, we do want to avoid
>> ksoftirqd latency in the
Chris Wilson writes:
> We dropped calling process_csb prior to handling direct submission in
> order to avoid the nesting of spinlocks and lift process_csb() and the
> majority of the tasklet out of irq-off. However, we do want to avoid
> ksoftirqd latency in the fast path, so try and pull the in
We dropped calling process_csb prior to handling direct submission in
order to avoid the nesting of spinlocks and lift process_csb() and the
majority of the tasklet out of irq-off. However, we do want to avoid
ksoftirqd latency in the fast path, so try and pull the interrupt-bh
local to direct subm
We dropped calling process_csb prior to handling direct submission in
order to avoid the nesting of spinlocks and lift process_csb() and the
majority of the tasklet out of irq-off. However, we do want to avoid
ksoftirqd latency in the fast path, so try and pull the interrupt-bh
local to direct subm