Re: [PATCH v3 2/4] x86/traps: Print non-canonical address on #GP

[Jann Horn]

On Wed, Nov 20, 2019 at 2:16 PM Ingo Molnar <mingo@xxxxxxxxxx> wrote:
> * Jann Horn <jannh@xxxxxxxxxx> wrote:
>
> > On Wed, Nov 20, 2019 at 12:24 PM Borislav Petkov <bp@xxxxxxxxx> wrote:
> > > On Wed, Nov 20, 2019 at 12:18:59PM +0100, Ingo Molnar wrote:
> > > > How was this maximum string length of '90' derived? In what way will
> > > > that have to change if someone changes the message?
> > >
> > > That was me counting the string length in a dirty patch in a previous
> > > thread. We probably should say why we decided for a certain length and
> > > maybe have a define for it.
> >
> > Do you think something like this would be better?
> >
> > char desc[sizeof(GPFSTR) + 50 + 2*sizeof(unsigned long) + 1] = GPFSTR;
>
> I'd much prefer this for, because it's a big honking warning for people
> to not just assume things but double check the limits.

Sorry, I can't parse the start of this sentence. I _think_ you're
saying you want me to make the change to "char desc[sizeof(GPFSTR) +
50 + 2*sizeof(unsigned long) + 1]"?

> I.e. this mild obfuscation of the array size *helps* code quality in the
> long run :-)