Re: [PATCH -next] tpm: change the return type of calc_tpm2_event_size to size_t

[Jason Gunthorpe]

On Tue, Feb 19, 2019 at 12:35:32PM +0200, Jarkko Sakkinen wrote:
> On Tue, Feb 19, 2019 at 05:15:47PM +0800, YueHaibing wrote:
> > On 2019/2/19 16:59, Jarkko Sakkinen wrote:
> > > On Tue, Feb 19, 2019 at 03:26:18PM +0800, YueHaibing wrote:
> > >> calc_tpm2_event_size return size of the event which type is
> > >> size_t, If it is an invalid event, returns 0. And all the
> > >> caller use a size_t variable to check the return value, so
> > >> no need to convert to the return value type to int.
> > >>
> > >> Signed-off-by: YueHaibing <yuehaibing@xxxxxxxxxx>
> > > 
> > > I don't see this patch make any value.
> > 
> > calc_tpm2_event_size return size_t 'size' which derive from u32 'event_field->event_size',
> > 
> > convert it to int may result in truncation.
> 
> Two remarks:
> 
> - Your real name is formatted incorrectly. It should be like "Yue Haibig"
>   as shown in [1].
> - If there is no actual regression, this change is useless.

Clarity of the code is a laudible goal - not everything is about
fixing bugs or adding features.

It is much easier to write the code properly than to try and prove
that event_size can not exceed INT_MAX.

Jason