On Oct 23, 2014, at 3:33 PM, Thomas Graf <[email protected]> wrote:
> On 10/23/14 at 02:15pm, Jarno Rajahalme wrote:
>> @@ -472,8 +470,8 @@ pvector_verify(const struct pvector *pvec)
>> PVECTOR_FOR_EACH (ptr, pvec) {
>> priority = cursor__.vector[cursor__.entry_idx].priority;
>> if (priority > prev_priority) {
>> - VLOG_ABORT("Priority vector is out of order (%u > %u)",
>> - priority, prev_priority);
>> + ovs_abort(0, "Priority vector is out of order (%u > %u)",
>> + priority, prev_priority);
>> }
>> prev_priority = priority;
>> }
>> @@ -534,14 +532,14 @@ check_tables(const struct classifier *cls, int
>> n_tables, int n_rules,
>> PVECTOR_FOR_EACH (iter, &cls->subtables) {
>> if (iter == table) {
>> if (found) {
>> - VLOG_ABORT("Subtable %p duplicated in 'subtables'.",
>> - table);
>> + ovs_abort(0, "Subtable %p duplicated in 'subtables'.",
>> + table);
>> }
>> found = true;
>> }
>> }
>> if (!found) {
>> - VLOG_ABORT("Subtable %p not found from 'subtables'.", table);
>> + ovs_abort(0, "Subtable %p not found from 'subtables'.", table);
>> }
>>
>> assert(!cmap_is_empty(&table->rules));
>
> I found the VLOG_DECLARE_THIS_MODULE() approach in the original
> patch cleaner than this.
The above changes are actually not required for this approach, as I could have
defined a test_classifier module. However, most of the test-classifier.c did
not use vlog at all, so I decided to convert these to ovs_abort()s instead. It
seems that I had introduced these VLOG_ABORT()s here without thinking about it
too much. I did not intend to depend on the classifier blog module definition
for them to work.
Jarno
_______________________________________________
dev mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/dev
