[PATCH] iw: Use ifname if available, print new/del/stop events.

[PATCH] iw: Use ifname if available, print new/del/stop events.

[greearb]

From: Ben Greear <greearb@candelatech.com>

If the kernel sends the IFNAME in the netlink message, then
use it to save the call to map ifindex to name (and
potentially fix races, especially when kernel is
busy deleting an interface.

Signed-off-by: Ben Greear <greearb@candelatech.com>
---
 event.c | 22 +++++++++++++++++++---
 1 file changed, 19 insertions(+), 3 deletions(-)

diff --git a/event.c b/event.c
index 929854b..b87a076 100644
--- a/event.c
+++ b/event.c
@@ -317,15 +317,22 @@ static int print_event(struct nl_msg *msg, void *arg)
 	nla_parse(tb, NL80211_ATTR_MAX, genlmsg_attrdata(gnlh, 0),
 		  genlmsg_attrlen(gnlh, 0), NULL);
 
-	if (tb[NL80211_ATTR_IFINDEX] && tb[NL80211_ATTR_WIPHY]) {
+	ifname[0] = 0;
+	if (tb[NL80211_ATTR_IFNAME]) {
+		memset(ifname, 0, sizeof(ifname));
+		strncpy(ifname, nla_data(tb[NL80211_ATTR_IFNAME]), IFNAMSIZ);
+	}
+	else if (tb[NL80211_ATTR_IFINDEX]) {
 		if_indextoname(nla_get_u32(tb[NL80211_ATTR_IFINDEX]), ifname);
+	}
+
+	if (ifname[0] && tb[NL80211_ATTR_WIPHY]) {
 		printf("%s (phy #%d): ", ifname, nla_get_u32(tb[NL80211_ATTR_WIPHY]));
 	} else if (tb[NL80211_ATTR_WDEV] && tb[NL80211_ATTR_WIPHY]) {
 		printf("wdev 0x%llx (phy #%d): ",
 			(unsigned long long)nla_get_u64(tb[NL80211_ATTR_WDEV]),
 			nla_get_u32(tb[NL80211_ATTR_WIPHY]));
-	} else if (tb[NL80211_ATTR_IFINDEX]) {
-		if_indextoname(nla_get_u32(tb[NL80211_ATTR_IFINDEX]), ifname);
+	} else if (ifname[0]) {
 		printf("%s: ", ifname);
 	} else if (tb[NL80211_ATTR_WDEV]) {
 		printf("wdev 0x%llx: ", (unsigned long long)nla_get_u64(tb[NL80211_ATTR_WDEV]));
@@ -599,6 +606,15 @@ static int print_event(struct nl_msg *msg, void *arg)
 		}
 		printf("\n");
 		break;
+	case NL80211_CMD_STOP_AP:
+		printf("stop AP\n");
+		break;
+	case NL80211_CMD_NEW_INTERFACE:
+		printf("new interface\n");
+		break;
+	case NL80211_CMD_DEL_INTERFACE:
+		printf("delete interface\n");
+		break;
 	default:
 		printf("unknown event %d (%s)\n",
 		       gnlh->cmd, command_name(gnlh->cmd));
-- 
1.9.3

Re: [PATCH] iw: Use ifname if available, print new/del/stop events.

[Johannes Berg]

On Wed, 2015-06-10 at 18:17 -0400, greearb@candelatech.com wrote:
> From: Ben Greear <greearb@candelatech.com>
> 
> If the kernel sends the IFNAME in the netlink message, then
> use it to save the call to map ifindex to name (and
> potentially fix races, especially when kernel is
> busy deleting an interface.

Please split the "print new/del/stop events" part into a separate patch
- I can apply that now.

johannes

Re: [PATCH] iw: Use ifname if available, print new/del/stop events.

[Ben Greear]

On 06/17/2015 02:14 AM, Johannes Berg wrote:
> On Wed, 2015-06-10 at 18:17 -0400, greearb@candelatech.com wrote:
>> From: Ben Greear <greearb@candelatech.com>
>>
>> If the kernel sends the IFNAME in the netlink message, then
>> use it to save the call to map ifindex to name (and
>> potentially fix races, especially when kernel is
>> busy deleting an interface.
>
> Please split the "print new/del/stop events" part into a separate patch
> - I can apply that now.

Some existing events (interface-delete, I think) already have the ifname, by the way..so the iw patch
is useful without any kernel changes.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com

Re: [PATCH] iw: Use ifname if available, print new/del/stop events.

[Johannes Berg]

On Wed, 2015-06-17 at 06:45 -0700, Ben Greear wrote:
> 
> On 06/17/2015 02:14 AM, Johannes Berg wrote:
> > On Wed, 2015-06-10 at 18:17 -0400, greearb@candelatech.com wrote:
> > > From: Ben Greear <greearb@candelatech.com>
> > > 
> > > If the kernel sends the IFNAME in the netlink message, then
> > > use it to save the call to map ifindex to name (and
> > > potentially fix races, especially when kernel is
> > > busy deleting an interface.
> > 
> > Please split the "print new/del/stop events" part into a separate 
> > patch
> > - I can apply that now.
> 
> Some existing events (interface-delete, I think) already have the 
> ifname, by the way..so the iw patch
> is useful without any kernel changes.
> 

Fair enough. Still, splitting the patch would make sense I think.

johannes