[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20180514223223.25433-15-sthemmin@microsoft.com>
Date: Mon, 14 May 2018 15:32:13 -0700
From: Stephen Hemminger <stephen@...workplumber.org>
To: davem@...emloft.net
Cc: netdev@...r.kernel.org,
Stephen Hemminger <stephen@...workplumber.org>,
Stephen Hemminger <sthemmin@...rosoft.com>
Subject: [PATCH net-stable 14/24] hv_netvsc: fix race in napi poll when rescheduling
From: Stephen Hemminger <stephen@...workplumber.org>
commit d64e38ae690e3337db0d38d9b149a193a1646c4b upstream
There is a race between napi_reschedule and re-enabling interrupts
which could lead to missed host interrrupts. This occurs when
interrupts are re-enabled (hv_end_read) and vmbus irq callback
(netvsc_channel_cb) has already scheduled NAPI.
Signed-off-by: Stephen Hemminger <sthemmin@...rosoft.com>
Signed-off-by: David S. Miller <davem@...emloft.net>
---
drivers/net/hyperv/netvsc.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c
index b7720b65f5d6..7e0eb99571af 100644
--- a/drivers/net/hyperv/netvsc.c
+++ b/drivers/net/hyperv/netvsc.c
@@ -1205,9 +1205,10 @@ int netvsc_poll(struct napi_struct *napi, int budget)
if (send_recv_completions(ndev, net_device, nvchan) == 0 &&
work_done < budget &&
napi_complete_done(napi, work_done) &&
- hv_end_read(&channel->inbound)) {
+ hv_end_read(&channel->inbound) &&
+ napi_schedule_prep(napi)) {
hv_begin_read(&channel->inbound);
- napi_reschedule(napi);
+ __napi_schedule(napi);
}
/* Driver may overshoot since multiple packets per descriptor */
--
2.17.0
Powered by blists - more mailing lists