Re: [Qemu-devel] [PATCH qom-next v2] qom-test: Test QOM properties

[Andreas Färber]

Am 12.03.2014 10:09, schrieb Markus Armbruster:
> Andreas Färber <address@hidden> writes:
> 
>> Recursively walk all properties under /machine and try to retrieve their
>> value. This is a regression test for link<> properties and the
>> DeviceState::hotpluggable property.
>>
>> Cf. be2f78b6b062eec5170e2612299fb8953046993f and
>>     1a37eca107cece3ed454bae29eef0bd1fac4a244
> 
> I'm afraid the last sentence is too terse for me to understand.

Commit hashes of a) link<> property bug fix and b) hotpluggable property
introduction. Both were about segfaults.

> 
>> Signed-off-by: Andreas Färber <address@hidden>
>> ---
>>  v2: Added g_test_message()s.
>>
>>  Applying to qom-next.
>>
>>  tests/qom-test.c | 43 +++++++++++++++++++++++++++++++++++++++++++
>>  1 file changed, 43 insertions(+)
>>
>> diff --git a/tests/qom-test.c b/tests/qom-test.c
>> index b6671fb..4909258 100644
>> --- a/tests/qom-test.c
>> +++ b/tests/qom-test.c
>> @@ -10,6 +10,7 @@
>>  #include <glib.h>
>>  #include <string.h>
>>  
>> +#include "qemu-common.h"
>>  #include "libqtest.h"
>>  #include "qemu/osdep.h"
>>  #include "qapi/qmp/types.h"
>> @@ -43,6 +44,44 @@ static bool is_blacklisted(const char *arch, const char 
>> *mach)
>>      return false;
>>  }
>>  
>> +static void test_properties(const char *path)
>> +{
>> +    char *cmd, *child_path;
>> +    QDict *response, *tuple;
>> +    QList *list;
>> +    QListEntry *entry;
>> +
>> +    g_test_message("Obtaining properties of %s", path);
>> +    cmd = g_strdup_printf("{ 'execute': 'qom-list',"
>> +                          "  'arguments': { 'path': '%s' } }", path);
>> +    response = qmp(cmd);
>> +    g_free(cmd);
>> +    g_assert(response);
>> +
>> +    g_assert(qdict_haskey(response, "return"));
>> +    list = qobject_to_qlist(qdict_get(response, "return"));
>> +    QLIST_FOREACH_ENTRY(list, entry) {
>> +        tuple = qobject_to_qdict(qlist_entry_obj(entry));
>> +        if (strstart(qdict_get_str(tuple, "type"), "child<", NULL)) {
>> +            child_path = g_strdup_printf("%s/%s",
>> +                                         path, qdict_get_str(tuple, 
>> "name"));
>> +            test_properties(child_path);
>> +            g_free(child_path);
>> +        } else {
>> +            const char *prop = qdict_get_str(tuple, "name");
>> +            g_test_message("Testing property %s.%s", path, prop);
>> +            cmd = g_strdup_printf("{ 'execute': 'qom-get',"
>> +                                  "  'arguments': { 'path': '%s',"
>> +                                  "                 'property': '%s' } }",
>> +                                  path, prop);
>> +            response = qmp(cmd);
>> +            g_free(cmd);
>> +            /* This may fail but should not, e.g., segfault. */
>> +            g_assert(response);
> 
> I'm not sure I get your comment.

Executing qom-get may return a QMP error when accessing a property, but
it may not crash. Therefore not much we can assert in the response. If a
segfault occurs, the process is gone at this point.

> 
>> +        }
>> +    }
>> +}
>> +
>>  static void test_machine(gconstpointer data)
>>  {
>>      const char *machine = data;
>> @@ -51,8 +90,12 @@ static void test_machine(gconstpointer data)
>>  
>>      args = g_strdup_printf("-machine %s", machine);
>>      qtest_start(args);
>> +
>> +    test_properties("/machine");
>> +
>>      response = qmp("{ 'execute': 'quit' }");
>>      g_assert(qdict_haskey(response, "return"));
>> +
>>      qtest_end();
>>      g_free(args);
>>  }
> 
> Looks like an excellent addition to the test suite.  Thanks!

Unfortunately adds to the main-loop warnings.

Cheers,
Andreas

-- 
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg