diff mbox

[ovs-dev,10/21] ofproto-macros.at: Add another strerror(0) value

Message ID 20151110041902.GC8509@ovn.org
State Not Applicable
Headers show

Commit Message

Ben Pfaff Nov. 10, 2015, 4:19 a.m. UTC 
On Mon, Oct 19, 2015 at 01:29:01PM +0900, YAMAMOTO Takashi wrote:
> On NetBSD, strerror(0) is "Undefined error: 0".
> 
> Signed-off-by: YAMAMOTO Takashi <yamamoto@midokura.com>
> ---
>  tests/ofproto-macros.at | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/tests/ofproto-macros.at b/tests/ofproto-macros.at
> index 94d7c86..368dade 100644
> --- a/tests/ofproto-macros.at
> +++ b/tests/ofproto-macros.at
> @@ -28,7 +28,8 @@ prt==1 { sub(/[ \t]*$/, ""); print $0 }
>  vconn_sub() {
>      sed '
>  s/tcp:127.0.0.1:[0-9][0-9]*:/unix:/
> -s/No error/Success/
> +s/(No error)/(Success)/
> +s/(Undefined error: 0)/(Success)/
>  '
>  }

Since we now have two different OSes that need special treatment here,
it might be better to do something more uniform to avoid the special
treatment.  One idea would be to modify ovs_strerror() to always provide
the same string for error 0.  Another would be to modify vconn.c to
avoid indicating any error when there is none, e.g.:

Comments

Takashi Yamamoto Nov. 19, 2015, 4:13 a.m. UTC | #1 
hi,

On Tue, Nov 10, 2015 at 1:19 PM, Ben Pfaff <blp@ovn.org> wrote:
> On Mon, Oct 19, 2015 at 01:29:01PM +0900, YAMAMOTO Takashi wrote:
>> On NetBSD, strerror(0) is "Undefined error: 0".
>>
>> Signed-off-by: YAMAMOTO Takashi <yamamoto@midokura.com>
>> ---
>>  tests/ofproto-macros.at | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/tests/ofproto-macros.at b/tests/ofproto-macros.at
>> index 94d7c86..368dade 100644
>> --- a/tests/ofproto-macros.at
>> +++ b/tests/ofproto-macros.at
>> @@ -28,7 +28,8 @@ prt==1 { sub(/[ \t]*$/, ""); print $0 }
>>  vconn_sub() {
>>      sed '
>>  s/tcp:127.0.0.1:[0-9][0-9]*:/unix:/
>> -s/No error/Success/
>> +s/(No error)/(Success)/
>> +s/(Undefined error: 0)/(Success)/
>>  '
>>  }
>
> Since we now have two different OSes that need special treatment here,
> it might be better to do something more uniform to avoid the special
> treatment.  One idea would be to modify ovs_strerror() to always provide
> the same string for error 0.  Another would be to modify vconn.c to
> avoid indicating any error when there is none, e.g.:

it's a good idea.

>
> diff --git a/lib/vconn.c b/lib/vconn.c
> index d835943..f50f3e8 100644
> --- a/lib/vconn.c
> +++ b/lib/vconn.c
> @@ -683,7 +683,9 @@ do_send(struct vconn *vconn, struct ofpbuf *msg)
>      } else {
>          char *s = ofp_to_string(msg->data, msg->size, 1);
>          retval = (vconn->vclass->send)(vconn, msg);
> -        if (retval != EAGAIN) {
> +        if (!retval) {
> +            VLOG_DBG_RL(&ofmsg_rl, "%s: sent: %s", vconn->name, s);
> +        } else if (retval != EAGAIN) {
>              VLOG_DBG_RL(&ofmsg_rl, "%s: sent (%s): %s",
>                          vconn->name, ovs_strerror(retval), s);
>          }
>
diff mbox

Patch

diff --git a/lib/vconn.c b/lib/vconn.c
index d835943..f50f3e8 100644
--- a/lib/vconn.c
+++ b/lib/vconn.c
@@ -683,7 +683,9 @@  do_send(struct vconn *vconn, struct ofpbuf *msg)
     } else {
         char *s = ofp_to_string(msg->data, msg->size, 1);
         retval = (vconn->vclass->send)(vconn, msg);
-        if (retval != EAGAIN) {
+        if (!retval) {
+            VLOG_DBG_RL(&ofmsg_rl, "%s: sent: %s", vconn->name, s);
+        } else if (retval != EAGAIN) {
             VLOG_DBG_RL(&ofmsg_rl, "%s: sent (%s): %s",
                         vconn->name, ovs_strerror(retval), s);
         }