[libvirt] [PATCH v3 09/14] qemu: Implement a layer for external devices like tpm-emulator

[Stefan Berger]

On 05/08/2018 04:50 PM, John Ferlan wrote:
>
> On 05/04/2018 04:21 PM, Stefan Berger wrote:
> +
>> + cleanup:
>> +    VIR_FREE(shortName);
>> +    VIR_FREE(errbuf);
>> +    virCommandFree(cmd);
>> +
>> +    virObjectUnref(cfg);
>> +
>> +    return ret;
>> +
>> + error:
>> +    VIR_FREE(tpm->data.emulator.source.data.nix.path);
> Still not clear why VIR_FREE here since virDomainTPMDefFree does it.

Removed.

>
>> +
>> +    goto cleanup;
>> +}
>> +
> Right about here you went back to single blank line between functions.

Until you now told me that this is the new rule, I was sometimes looking 
around in the file what others had used there before. ok, in this file I 
was inconsistent.@@ -6194,6 +6203,8 @@ qemuProcessLaunch(virConnectPtr 
conn,
>>       ret = 0;
>>   
>>    cleanup:
>> +    if (ret)
>> +        qemuExtDevicesStop(driver, vm->def);
>>       qemuDomainSecretDestroy(vm);
>>       virCommandFree(cmd);
>>       virObjectUnref(logCtxt);
>> @@ -6614,6 +6625,8 @@ void qemuProcessStop(virQEMUDriverPtr driver,
>>   
>>       qemuDomainCleanupRun(driver, vm);
>>   
>> +    qemuExtDevicesStop(driver, vm->def);
>> +
>>       /* Stop autodestroy in case guest is restarted */
>>       qemuProcessAutoDestroyRemove(driver, vm);
>>   
>>
> Anything need to be done during qemuProcessReconnect?

In all the scenarios I have tried so far I haven't come across having to 
add something to this function. If this is triggered by restart of 
libvirtd, I haven't seen a failure. I could have missed something, though.

   Stefan
>
> John
>