| Message ID | 0141bb9117a653a75a24128dbf4b9d68fbb54e40.1549488381.git.swise@opengridcomputing.com (mailing list archive) |
|---|---|
| State | Mainlined |
| Commit | 4133b013faf31fbab61d4deb626fcb4390042eba |
| Delegated to: | Jason Gunthorpe |
| Headers | show |
| Series | [rdma-next] lib/irq_poll: support schedules in non-interrupt contexts | expand |
> -----Original Message----- > From: linux-rdma-owner@vger.kernel.org <linux-rdma- > owner@vger.kernel.org> On Behalf Of Steve Wise > Sent: Wednesday, February 6, 2019 3:12 PM > To: dledford@redhat.com; jgg@mellanox.com > Cc: linux-rdma@vger.kernel.org; hch@infradead.org; sagi@grimberg.me > Subject: [PATCH rdma-next] lib/irq_poll: support schedules in non-interrupt > contexts > > Do not assume irq_poll_sched() is called from an interrupt context only. > So use raise_softirq_irqoff() instead of __raise_softirq_irqoff() so it > will kick the ksoftirqd if the schedule is from a non-interrupt context. > > This is required for RDMA drivers, like soft iwarp, that generate cq > completion notifications in a workqueue or kthread context. Without this > change, siw completion notifications to the ULP can take several hundred > usecs, depending on the system load. > > Signed-off-by: Steve Wise <swise@opengridcomputing.com> > --- > lib/irq_poll.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/lib/irq_poll.c b/lib/irq_poll.c > index 86a709954f5a..2f17b488d58e 100644 > --- a/lib/irq_poll.c > +++ b/lib/irq_poll.c > @@ -35,7 +35,7 @@ void irq_poll_sched(struct irq_poll *iop) > > local_irq_save(flags); > list_add_tail(&iop->list, this_cpu_ptr(&blk_cpu_iopoll)); > - __raise_softirq_irqoff(IRQ_POLL_SOFTIRQ); > + raise_softirq_irqoff(IRQ_POLL_SOFTIRQ); > local_irq_restore(flags); > } > EXPORT_SYMBOL(irq_poll_sched); > -- > 1.8.3.1 Are there any comments on this patch? Thanks, Steve.
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Jason, care to pick it up?
On Wed, Feb 06, 2019 at 01:11:49PM -0800, Steve Wise wrote: > Do not assume irq_poll_sched() is called from an interrupt context only. > So use raise_softirq_irqoff() instead of __raise_softirq_irqoff() so it > will kick the ksoftirqd if the schedule is from a non-interrupt context. > > This is required for RDMA drivers, like soft iwarp, that generate cq > completion notifications in a workqueue or kthread context. Without this > change, siw completion notifications to the ULP can take several hundred > usecs, depending on the system load. > > Signed-off-by: Steve Wise <swise@opengridcomputing.com> > Reviewed-by: Sagi Grimberg <sagi@grimberg.me> > --- > lib/irq_poll.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) Applied to for-next, thanks Jason
diff --git a/lib/irq_poll.c b/lib/irq_poll.c index 86a709954f5a..2f17b488d58e 100644 --- a/lib/irq_poll.c +++ b/lib/irq_poll.c @@ -35,7 +35,7 @@ void irq_poll_sched(struct irq_poll *iop) local_irq_save(flags); list_add_tail(&iop->list, this_cpu_ptr(&blk_cpu_iopoll)); - __raise_softirq_irqoff(IRQ_POLL_SOFTIRQ); + raise_softirq_irqoff(IRQ_POLL_SOFTIRQ); local_irq_restore(flags); } EXPORT_SYMBOL(irq_poll_sched);
Do not assume irq_poll_sched() is called from an interrupt context only. So use raise_softirq_irqoff() instead of __raise_softirq_irqoff() so it will kick the ksoftirqd if the schedule is from a non-interrupt context. This is required for RDMA drivers, like soft iwarp, that generate cq completion notifications in a workqueue or kthread context. Without this change, siw completion notifications to the ULP can take several hundred usecs, depending on the system load. Signed-off-by: Steve Wise <swise@opengridcomputing.com> --- lib/irq_poll.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)