2021-10-11 13:26:14

by Yajun Deng

Subject: [PATCH net-next] netpoll: Fix carrier_timeout for msleep()

It should be sleep carrier_timeout seconds rather than 4 seconds if
carrier_timeout has been modified.
Add start variable, hence atleast and atmost use the same jiffies, and
use msecs_to_jiffies() and MSEC_PER_SEC match with jiffies.
At the same time, msleep() is not for 1ms - 20ms, use usleep_range()
instead, see Documentation/timers/timers-howto.rst.

Fixes: bff38771e106 ("netpoll: Introduce netpoll_carrier_timeout kernel option")
Signed-off-by: Yajun Deng <[email protected]>
---
net/core/netpoll.c | 14 ++++++++------
1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index edfc0f8011f8..88e4ff3b9e95 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -682,7 +682,7 @@ int netpoll_setup(struct netpoll *np)
}

if (!netif_running(ndev)) {
- unsigned long atmost, atleast;
+ unsigned long atmost, atleast, start;

np_info(np, "device %s not up yet, forcing it\n", np->dev_name);

@@ -694,14 +694,15 @@ int netpoll_setup(struct netpoll *np)
}

rtnl_unlock();
- atleast = jiffies + HZ/10;
- atmost = jiffies + carrier_timeout * HZ;
+ start = jiffies;
+ atleast = start + msecs_to_jiffies(MSEC_PER_SEC / 10);
+ atmost = start + msecs_to_jiffies(carrier_timeout * MSEC_PER_SEC);
while (!netif_carrier_ok(ndev)) {
if (time_after(jiffies, atmost)) {
np_notice(np, "timeout waiting for carrier\n");
break;
}
- msleep(1);
+ usleep_range(USEC_PER_MSEC, 2 * USEC_PER_MSEC);
}

/* If carrier appears to come up instantly, we don't
@@ -710,8 +711,9 @@ int netpoll_setup(struct netpoll *np)
*/

if (time_before(jiffies, atleast)) {
- np_notice(np, "carrier detect appears untrustworthy, waiting 4 seconds\n");
- msleep(4000);
+ np_notice(np, "carrier detect appears untrustworthy, waiting %d seconds\n",
+ carrier_timeout);
+ msleep(carrier_timeout * MSEC_PER_SEC);
}
rtnl_lock();
}
--
2.32.0

2021-10-12 00:14:21

by Jakub Kicinski

Subject: Re: [PATCH net-next] netpoll: Fix carrier_timeout for msleep()

On Mon, 11 Oct 2021 16:57:53 +0800 Yajun Deng wrote:
> It should be sleep carrier_timeout seconds rather than 4 seconds if
> carrier_timeout has been modified.

carrier_timeout is for changing the upper bound of the wait,
not for controlling how long to wait if carrier is untrustworthy.

> Add start variable, hence atleast and atmost use the same jiffies, and
> use msecs_to_jiffies() and MSEC_PER_SEC match with jiffies.
> At the same time, msleep() is not for 1ms - 20ms, use usleep_range()
> instead, see Documentation/timers/timers-howto.rst.
>
> Fixes: bff38771e106 ("netpoll: Introduce netpoll_carrier_timeout kernel option")
> Signed-off-by: Yajun Deng <[email protected]>