summaryrefslogtreecommitdiff
path: root/drivers/infiniband
diff options
context:
space:
mode:
authorJubin John <jubin.john@intel.com>2016-04-12 20:47:00 +0300
committerDoug Ledford <dledford@redhat.com>2016-04-28 19:00:39 +0300
commite6d2e0176e1f3c1e1534851b66c0b972f03ff069 (patch)
treee06b598385a9d2d070a4d1f2d6f02537be834909 /drivers/infiniband
parent849e3e9398608c26a7c54bf9fbf3288f7ced6bfb (diff)
downloadlinux-e6d2e0176e1f3c1e1534851b66c0b972f03ff069.tar.xz
IB/rdmavt: Fix send scheduling
call_send is used to determine whether to send immediately or schedule a send for later. The current logic in rdmavt is inverted and has a negative impact on the latency of the hfi1 and qib drivers. Fix this regression by correctly calling send immediately when call_send is set. Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Jubin John <jubin.john@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Diffstat (limited to 'drivers/infiniband')
-rw-r--r--drivers/infiniband/sw/rdmavt/qp.c4
1 files changed, 2 insertions, 2 deletions
diff --git a/drivers/infiniband/sw/rdmavt/qp.c b/drivers/infiniband/sw/rdmavt/qp.c
index bd82a6948dc8..a9e3bcc522c4 100644
--- a/drivers/infiniband/sw/rdmavt/qp.c
+++ b/drivers/infiniband/sw/rdmavt/qp.c
@@ -1637,9 +1637,9 @@ bail:
spin_unlock_irqrestore(&qp->s_hlock, flags);
if (nreq) {
if (call_send)
- rdi->driver_f.schedule_send_no_lock(qp);
- else
rdi->driver_f.do_send(qp);
+ else
+ rdi->driver_f.schedule_send_no_lock(qp);
}
return err;
}