Re: [PATCH 1/3] trace-cmd: Extend host-guest time sync with fraction bits

[Steven Rostedt <rostedt@goodmis.org>]

On Wed, 13 Oct 2021 12:48:55 +0300
Tzvetomir Stoyanov <tz.stoyanov@gmail.com> wrote:

> On Wed, Oct 13, 2021 at 6:04 AM Steven Rostedt <rostedt@goodmis.org> wrote:
> >
> > On Thu, 23 Sep 2021 12:45:24 +0300
> > "Tzvetomir Stoyanov (VMware)" <tz.stoyanov@gmail.com> wrote:  
> > > @@ -175,6 +184,12 @@ static int kvm_open_vcpu_dir(struct kvm_clock_sync *kvm, int cpu, char *dir_str)
> > >                                        dir_str, entry->d_name);
> > >                               kvm->vcpu_scalings[cpu] = strdup(path);
> > >                       }
> > > +                     if (!strncmp(entry->d_name, KVM_DEBUG_FRACTION_FILE,
> > > +                                  strlen(KVM_DEBUG_FRACTION_FILE))) {  
> >
> > I'm curious, why not just use strcmp(), or does "d_name" have more
> > characters than the fraction file?  
> 
> Just to be on the safe side. Currently the fraction file is with the
> longest name, but that may change in the future.\\


I'm confused, what do you mean on the safe side? What are you protecting
against? In other words, what can possibly go wrong? And I notice you do
this for all of them. I don't get it.

The only thing that this is protecting against, is if the file name becomes
"tsc-scaling-ratio-frac-bits-some-garbage-here" then it it will still
match. But other than that, you are not protecting against anything.

-- Steve


> 
> >  
> > > +                             snprintf(path, sizeof(path), "%s/%s",
> > > +                                      dir_str, entry->d_name);
> > > +                             kvm->vcpu_frac[cpu] = strdup(path);
> > > +                     }
> > >               }
> > >       }