Re: [MM 0.6] Plumb up the SPN display-rule bits

[Marcel Holtmann <marcel holtmann org>]

Hi Dan,

> > >>         A possible fix to handle the case where we don't know how much
> > >>         we can
> > >>
> > >>         read would be to try to read the first bytes of the record (3 or
> > >>         4 or 5
> > >>         just in case) to get the full record length of the record,
> > >>         assuming 1-3
> > >>         bytes max for the size field, so we read "A3" + size encoded in
> > >>         1 or 2
> > >>         or 3 bytes + "80"); and once we know the whole record size, read the
> > >>         exact list size.
> > >>
> > >>
> > >>     Yeah, I thought about this, but had hoped to avoid it (If the record
> > >>     is present at all, there will be a minimum of 7 bytes -
> > >>     A3058003FFFFFF -  so I think that's the right amount for an initial
> > >>     read). 
> > >>
> > >>
> > >> Here's a version that does the 7-then-more retry. It works on my SIM
> > >> (retrying from 7 to 34 bytes, even though the entries are all FFFFFF);
> > >> see what you think.
> > >>
> > > 
> > > That worked very well for me.
> > > 
> > > Now that we have the 3 values we needed from the SIM card (2 flags +
> > > SPDI list), we could now also implement the "RequiredNameDisplay"
> > > property which contains the whole logic for us, as Dan suggested
> > > earlier. Given that this value depends on the current 3GPP operator ID,
> > > I would really include that new property in the 3GPP interface. Will try
> > > to implement this later on today.
> > > 
> > 
> > I pushed now a '06-api-display-rules' branch in upstream git; with the
> > changes discussed here and a new "RequiredDisplayName" property.
> > 
> > Dan, Nathan, are you guys happy with the new branch? I'm not very
> > convinced yet about the flag property names...
> 
> Need to check it out; but this is what Marcel is always talking about
> when he says that "you really need to implement the SIM filesystem"
> since I believe you can actually read the lengths of various EF records
> out of the SIM using using some of the various commands, and thus you
> know what length to request and you don't have to retry with other
> lengths.  That's not foolproof though, since some firmware
> implementations are simply dumb.  I ran into the length problem a few
> times when trying to read the IMSI and the MNC length; same thing there
> where some modems want one length and others want another, even though
> it's exactly the same damn SIM just swapped between devices.

many modems that have never been used in a real phone have broken SIM
filesystem access. They are getting better, but some of them are still
pretty bad. Try an Ericsson MBM if you want some of the better ones.

You do need to read SIM filesystem information structures first. That is
how you suppose to do the access. And not wildly guess. Otherwise you do
not even know what type you have and how to read it. But for some of
them you have fixed information struct and types of binary, record or
cyclic. Also the access bits and what PIN needs to be entered first to
access this data is nasty. If you really wanna know this complex crazy
from 3GPP, then have a look at oFono and what we needed to do to make
this work properly.

Regards

Marcel