Memory footprints in website

[Krishna Chaitanya]

On Fri, Oct 13, 2023 at 9:31 PM Jouni Malinen <j at w1.fi> wrote:
>
> On Tue, Oct 10, 2023 at 06:27:47PM +0530, Krishna Chaitanya wrote:
> > Looking at https://w1.fi/wpa_supplicant/ the memory footprint for a
> > WPA2-personal binary
> > is stated as 50KB on the x86 target without any debugging.
> >
> > I have tried to compile a minimal configuration WPA as below on RPI
> > (had some issues with 32bit compilation on my 64bit machine):
>
> > but got the binary size of ~289K, am I missing something? or are the
> > numbers outdated?
>
> I think those numbers are from the last time I looked at the minimal
> binary sizes which was probably around 2005 or so.. A lot more
> functionality has been added over the last 18 years and there has not
> really been much focus to trying to make sure the minimum size would
> remain small..
Thanks for the context.
>
> > But the help text for CONFIG_NO_WPA mentions ~35-50KB reduction, with
> >
> > CONFIG_NO_WPA=y
> > CONFIG_IEEE8021X_EAPOL=n
> >
> > we get ~237K, so, that is 52K reduction, so, can I assume that the
> > "wording" should be a reduction and not absolute binary size?
>
> I'm not sure which wording you are referring to here..  This 35-50 kB
> reduction is from 2005 as well and at the time and this one is reduction
> as it says while the numbers on the web site are for the full binary.
As NO_WPA reduces the footprint by 50K, I thought its "reduces binary size by"
vs. "binary size is", but as you say, its just a coincidence that both are 50K.
>
> > Either way looks like the website text needs an update, as customers
> > just refers to that expect to match reality.
>
> Well, I'm not sure whether there is really much interest on keeping this
> type of number up-to-date and as such, that update might mean simply
> deletion of any numbers or adding a note pointing out that the values
> were from a snapshot in 2005 and do not indicated more or less anything
> about the current snapshot.
IMHO, we could just delete those.