My approach should be functionally equivalent.
old approach when implicit_tag==TRUE
read tag and length, ignore tag but use length to know how long
the field is.
my approach when implicit_tag==TRUE
tag and length have already been stripped off by the parent dissector,
use tvb_length_remaining() to find the length of the field.
i had to do it that way to get cmip to behave.
On Wed, 26 Jan 2005 07:28:38 +0100, Tomas Kukosa
<tomas.kukosa@xxxxxxxxxxx> wrote:
> My idea concerning implicit tags used in the asn2eth compiler and my dissectors is little
> bit different.
>
> If implicit tag ocurs it is checked in the 'parent structure' and it is not cut but
> passesd to the 'field decoder' with implicit_tag=TRUE.
> And the 'field decoder' does not check tag value but can use all information about length
> etc. Event it can pass that tag into the next subdecoder if nested.
>
> Which constructs for CMIP could make problem with that approach?
> Could you send me an exmaple?
>
> Regards,
> Tom
>
>
> ronnie sahlberg wrote:
>
> > the idea was to make some CMIP stuff decode properly.
> >
> > the idea is that IF implitic_tag is set, then the identifier and
> > length is not checked at all
> > instead it is assume the value spans until the end of the tvb, some
> > cases therefore need, to create a new sub tvb that only spans the
> > length.
> >
> > i.e. CHOICE {
> > [1] IMPLICIT integer foo;
> >
> > then the identifier and length is read in the coice dissector, and
> > since it was IMPLICIT
> > it creates a new tvb only containing as much data as length indicated.
> > then it calls the subdissector for integer/foo since implicit_tag
> > is now set instead of reading the identifier/length it assumes the
> > length is tvb_length_remaining()
> >
> >
> > it was the best i could come up with to handle some of the constructs for CMIP.
> >
> > If you have a better approach, please have a go at it.
> > I can provide the cmip captures i had problems with for you to test with.
> >
> >
> >
> >
> > On Tue, 25 Jan 2005 16:00:32 +0100, Tomas Kukosa
> > <tomas.kukosa@xxxxxxxxxxx> wrote:
> >
> >>Hi Ronnie,
> >> I have found with delay that some my dissectors does not work properly.
> >>The reason is different behaviour of dissect_ber_sequence() if implicit_tag is TRUE.
> >>
> >>What was your idea concernig implicit tag?
> >>My original idea was that BER identifier and length is decoded for implicit tag too but
> >>the tag is not checked.
> >>But after your changes BER identifier and length is not decoded anymore for implicit tag.
> >>
> >>Could we clarify it?
> >>
> >>Regards,
> >> Tom
> >>
> >>_______________________________________________
> >>Ethereal-dev mailing list
> >>Ethereal-dev@xxxxxxxxxxxx
> >>http://www.ethereal.com/mailman/listinfo/ethereal-dev
> >>
> >
> >
> > _______________________________________________
> > Ethereal-dev mailing list
> > Ethereal-dev@xxxxxxxxxxxx
> > http://www.ethereal.com/mailman/listinfo/ethereal-dev
> >
>
> --
> _________________________________________________
>
> ANF DATA ANF DATA spol. s r. o.
> a SIEMENS Company MediaHall, Bidláky 20
> 639 00 Brno
> Czech Republic
> Tomáš Kukosa Tel.: +420 - 533 337 227
> PSE ECT PSS 2 Fax : +420 - 533 337 252
> Room: 2.64 mailto:tomas.kukosa@xxxxxxxxxxx
> _________________________________________________
>
>