Re: [PATCH] cpu-topology: Fix the potential data corruption

[Sudeep Holla]

On Sat, Feb 29, 2020 at 01:41:47AM +0000, Zengtao (B) wrote:
> > -----Original Message-----
> > From: Sudeep Holla [mailto:sudeep.holla@xxxxxxx]
> > Sent: Friday, February 28, 2020 6:41 PM
> > To: Zengtao (B)
> > Cc: Linuxarm; Greg Kroah-Hartman; Rafael J. Wysocki;
> > linux-kernel@xxxxxxxxxxxxxxx; Sudeep Holla
> > Subject: Re: [PATCH] cpu-topology: Fix the potential data corruption
> >
> > On Fri, Feb 28, 2020 at 04:35:45PM +0800, Zeng Tao wrote:
> > > Currently there are only 10 bytes to store the cpu-topology info.
> > > That is:
> > > snprintf(buffer, 10, "cluster%d",i);
> > > snprintf(buffer, 10, "thread%d",i);
> > > snprintf(buffer, 10, "core%d",i);
> > >
> > > In the boundary test, if the cluster number exceeds 100, there will be a
> >
> > I don't understand you mention of 100 in particular above. I can see
> > issue
> > if there are cluster with more than 2-digit id. Though highly unlikely for
> > now, but I don't have objection to the patch.
> >
>
> The same meaning, more than 2-digit id equals to more than 100, right?

Yes. May be it is obvious but I prefer to word the commit message accordingly.
Mention of 100 specifically makes at-least me think something very specific
to 100 and not applicable for any more than 2-digit number.

> Here 100 is for from tester/user perspective.
> And we found this issue when test with QEMU.

OK.

--
Regards,
Sudeep