Re: [PATCH] xen/arm: Avoid integer overflow using MIDR_IMPLEMENTOR_MASK

[Julien Grall <julien@xxxxxxx>]

Hi,

On 25/02/2022 10:59, Andrew Cooper wrote:
> On 25/02/2022 10:54, Julien Grall wrote:
> > Hi Michal,
> >
> > On 25/02/2022 08:38, Michal Orzel wrote:
> > > Value of macro MIDR_IMPLEMENTOR_MASK exceeds the range of integer
> > > and can lead to overflow. Currently there is no issue as it is used
> > > in an expression implicitly casted to u32 in MIDR_IS_CPU_MODEL_RANGE.
> > > To avoid possible problems, fix the macro.
> > Correct me if I am wrong, it sounds like this is only for hardening
> > purpose at the moment.
> >
> > As this code is coming from Linux, I would prefer if we first upstream
> > to Linux and then port to Xen once merged.
> Well.  The expression is undefined behaviour in C, because of shifting
> into the sign bit.
>
> In principle, the compiler is free to optimise is_affected_midr_range()
> to "return true" as a consequence, even if compilers tend not to be that
> malicious.
Are you arguing against fixing Linux first and the backport it to Xen?

Cheers,

--
Julien Grall