Re: [libvirt] [PATCH v3 10/14] qemu: Add support for external swtpm TPM emulator
[...]
>> virCommandPtr cmd,
>> virQEMUCapsPtr qemuCaps,
>> int *tpmfd,
>> - int *cancelfd)
>> + int *cancelfd,
>> + char **chardev)
>> {
>> - const virDomainTPMDef *tpm = def->tpm;
>> + virDomainTPMDef *tpm = def->tpm;
> Don't lose the "const"
>
>> virBuffer buf = VIR_BUFFER_INITIALIZER;
>> - const char *type = virDomainTPMBackendTypeToString(tpm->type);
>> + const char *type = NULL;
>> char *cancel_path = NULL, *devset = NULL;
>> const char *tpmdev;
>> *tpmfd = -1;
>> *cancelfd = -1;
>> + switch (tpm->type) {
>> + case VIR_DOMAIN_TPM_TYPE_PASSTHROUGH:
>> + case VIR_DOMAIN_TPM_TYPE_EMULATOR:
>> + type = virDomainTPMBackendTypeToString(tpm->type);
>> + break;
>> + case VIR_DOMAIN_TPM_TYPE_LAST:
> default:
> virReportEnumRangeError(virDomainTPMBackendType, tpm->type);
>
> We need some sort of error message otherwise we get failed for some
> reason which is never fun to diagnose.
All other cases I see use the same function without error message. Not
sure what you mean. We seem to follow a pattern with this now.
yeah this is one of those "inconsistent" things we have. It's probably
just reflex action to see a place that returns -1 without an error
message and say - we need one here; however, since the ->type would
already have been validated, not necessary then.
There are some of these switches where the EnumRange is given - so well
it's just habit for me.
John
[...]