Re: [Qemu-devel] [PATCH] Fix compile with --disable-tpm

[BALATON Zoltan]

On Mon, 23 Oct 2017, Philippe Mathieu-Daudé wrote:
> On 10/23/2017 08:09 AM, Philippe Mathieu-Daudé wrote:
> > On 10/23/2017 07:24 AM, BALATON Zoltan wrote:
> [...]
> > >          info = g_new0(TPMInfoList, 1);
> > > +#ifdef CONFIG_TPM
> > >          info->value = tpm_backend_query_tpm(drv);
> > > -
> > > +#endif
> > >          if (!cur_item) {
> > >              head = cur_item = info;
> > >          } else {
> > > diff --git a/vl.c b/vl.c
> > > index 0723835..dbfd06d 100644
> > > --- a/vl.c
> > > +++ b/vl.c
> > > @@ -4905,7 +4905,9 @@ int main(int argc, char **argv, char **envp)
> > >      res_free();
> > >
> > >      /* vhost-user must be cleaned up before chardevs.  */
> > > +#ifdef CONFIG_TPM
> > >      tpm_cleanup();
> > > +#endif
> > >      net_cleanup();
> > >      audio_cleanup();
> > >      monitor_cleanup();
> >
> > Can you instead add tpm_cleanup() in stubs/tpm.c?
>
> and tpm_backend_query_tpm() ;)

The v2 I've sent avoids #ifdefs outside tpm.c so isn't that acceptable? Do 
we still need stubs for these? I don't understand how stubs are compiled 
so if it's needed can you point me to an example what needs to be done to 
add stubs for these? (I'm not familiar with this, just tried to fix a 
compile problem I've met but I don't really know what's the correct way to 
fix this. I'd expect the author of the patch that broke it to fix it 
instead or at least review this to avoid breaking it more.)

Regards,
BALATON Zoltan