On 2020-11-26 01:08, Xie He wrote:
Hi Martin,
Since we are going to assume lapb->state would remain in LAPB_STATE_0
when
the carrier is down (as understood by me. Right?), could we add a check
in
lapb_connect_request to reject the upper layer's "connect" instruction
when
the carrier is down? Like this:
No, because this will break the considered "on demand" calling feature.
diff --git a/include/linux/lapb.h b/include/linux/lapb.h
index eb56472f23b2..7923b1c6fc6a 100644
--- a/include/linux/lapb.h
+++ b/include/linux/lapb.h
@@ -14,6 +14,7 @@
#define LAPB_REFUSED 5
#define LAPB_TIMEDOUT 6
#define LAPB_NOMEM 7
+#define LAPB_NOCARRIER 8
#define LAPB_STANDARD 0x00
#define LAPB_EXTENDED 0x01
diff --git a/net/lapb/lapb_iface.c b/net/lapb/lapb_iface.c
index 3c03f6512c5f..c909d8db1bef 100644
--- a/net/lapb/lapb_iface.c
+++ b/net/lapb/lapb_iface.c
@@ -270,6 +270,10 @@ int lapb_connect_request(struct net_device *dev)
if (!lapb)
goto out;
+ rc = LAPB_NOCARRIER;
+ if (!netif_carrier_ok(dev))
+ goto out_put;
+
rc = LAPB_OK;
if (lapb->state == LAPB_STATE_1)
goto out_put;
Also, since we are going to assume the lapb->state would remain in
LAPB_STATE_0 when the carrier is down, are the
"lapb->state == LAPB_STATE_0" checks in carrier-up/device-up event
handling necessary? If they are not necessary, it might be better to
remove them because it may confuse people reading the code.
They are still necessary, because if the link setup is initiated by
upper layers, we've already entered the respective state by
lapb_connect_request().
Every suggestion for improvement is really welcome, but please let this
patch set pass now, if you don't find any more gross errors.
Martin
