Coverity spotted the leaks.
Here's a proposed fix.
I didn't bother with a separate diagnostic for the unlikely
event that we read a negative number from the pipe.
Seems far-fetched enough not to bother, but it's easy to
add, if anyone cares.
>From f3439c7eae46681eacf5b469a6f0e22cb8fec1b4 Mon Sep
According to Jim Meyering on 2/25/2010 12:01 PM:
>> You're still keeping with fscanf.
>
> Yes. There are plenty of bigger fish to fry, for now ;-)
> As it is, this is more of a rearrangement than I generally
> prefer "just to fix a leak".
Indeed, I have no problem with checking in your patch as-
Eric Blake wrote:
> According to Jim Meyering on 2/25/2010 11:30 AM:
>
> ACK on plugging the leak. However,...
Thanks for the review.
>> @@ -979,18 +980,12 @@ int openvzGetVEID(const char *name) {
>> return -1;
>> }
>>
>> -if (fscanf(fp, "%d\n", &veid ) != 1) {
>> +ok = fsc
According to Jim Meyering on 2/25/2010 11:30 AM:
ACK on plugging the leak. However,...
> @@ -979,18 +980,12 @@ int openvzGetVEID(const char *name) {
> return -1;
> }
>
> -if (fscanf(fp, "%d\n", &veid ) != 1) {
> +ok = fscanf(fp, "%d\n", &veid ) == 1;
You're still keeping