I'm currently dabbling with a couple of T222's, and have been able to run Occam code compiled in DOS using D7305. However, since I have a couple of Sun workstations, D4305A would be more useful, if only because I could just ssh into them to run the compiler.

Contrary to D7305, D4305 uses licensing. It comes with all de required binaries (including the FlexLM manager), but no code. It just refers to the "delivery manual" which I haven't been able to find. The license manager accepts a code-less entry, but then oc just complains about invalid encryption in the license.

Was D4305A ever released license-free (like D7305) or was a generic code ever issued?

Louis van Dompselaar <vandompselaar@gmail.com> wrote:
> I'm currently dabbling with a couple of T222's, and have been able to run
> Occam code compiled in DOS using D7305. However, since I have a couple of
> Sun workstations, D4305A would be more useful, if only because I could
> just ssh into them to run the compiler.
>
> Contrary to D7305, D4305 uses licensing. It comes with all de required
> binaries (including the FlexLM manager), but no code. It just refers to
> the "delivery manual" which I haven't been able to find. The license
> manager accepts a code-less entry, but then oc just complains about
> invalid encryption in the license.

I don't know anything about Transputer tools, but there's some generic
(mostly Windows) FlexLM hacking information here:
http://web.archive.org/web/20220218000631/http://www.woodmann.com/crackz/Flexlm.htm

including tools to generate keys. I haven't tried any of them.

FlexLM nowadays is network-based, but it sounds like your version is old
enough that it uses a static code, and if you know the version one of the
generator tools might help.

Theo

Op maandag 29 mei 2023 om 12:09:52 UTC+2 schreef Theo:
> > Contrary to D7305, D4305 uses licensing. It comes with all de required
> > binaries (including the FlexLM manager), but no code. It just refers to
> > the "delivery manual" which I haven't been able to find. The license
> > manager accepts a code-less entry, but then oc just complains about
> > invalid encryption in the license.
> I don't know anything about Transputer tools, but there's some generic
> (mostly Windows) FlexLM hacking information here:
> http://web.archive.org/web/20220218000631/http://www.woodmann.com/crackz/Flexlm.htm
>
> including tools to generate keys. I haven't tried any of them.
>
> FlexLM nowadays is network-based, but it sounds like your version is old
> enough that it uses a static code, and if you know the version one of the
> generator tools might help.

Thanks for the pointers. I'll have a look around. At first glance their trick seems to be using security holes in DLL's, which won't be an option here..

Yes, D4305A includes a local FlexLM lmgrd (v2.21), the Inmos license daemon (inmos_ld), and a D4305 license generator tool (that requires the unknown activation code). I'm not sure how these child daemons work. FlexLM launches them, and as the Occam compiler is complaining about key encryption, it appears the license itself is encrypted in the FlexLM license.dat (by the generator tool, probably node-locked as it requires the hostid too) and then decrypted by the Inmos daemon. That would put up another barrier. It feels over-engineered since there is no protection on the DOS tools.

Louis van Dompselaar <vandompselaar@gmail.com> wrote:
>
> Thanks for the pointers. I'll have a look around. At first glance their
> trick seems to be using security holes in DLL's, which won't be an option
> here.
>
> Yes, D4305A includes a local FlexLM lmgrd (v2.21), the Inmos license
> daemon (inmos_ld), and a D4305 license generator tool (that requires the
> unknown activation code). I'm not sure how these child daemons work.

On network installs of recent FlexLM, there's the FlexLM daemon and the
vendor daemon, which listen on different ports. The first is generic, the
second is specific to a particular vendor. I think the client is talking to
the vendor daemon primarily to check out licences (they are both exposed on
different TCP sockets) - maybe the FlexLM daemon is just being a signpost
to find out the port of the vendor daemon?

(in theory you can run multiple vendor daemons on the same machine with a
single FlexLM daemon, in practice everyone wants a different FlexLM version
so it's easier to use VMs or containers)

> FlexLM launches them, and as the Occam compiler is complaining about key
> encryption, it appears the license itself is encrypted in the FlexLM
> license.dat (by the generator tool, probably node-locked as it requires
> the hostid too) and then decrypted by the Inmos daemon.

license.dat is the licence file, that contains the actual licence
information as well as the host information. Typically there will be a list
of 'feature' tags and a licence for each one?

Do you have a valid licence file? For example, this is an old licence file
for Bluespec's compiler (which is now open source and no longer uses FlexLM):

# licence server and MAC address:
SERVER licsrv.example.com 0123456789ab
# path to vendor daemon, whose binary is called 'BLUESPEC'
VENDOR BLUESPEC $BLUESPECDIR/../util/flexlm/<architecture>/BLUESPEC
# #
########################## FEATURE LINES ############################
# # Version: 3.8
# Expiration: 30-May-2020
# INCREMENT BCOMP BLUESPEC 3.8 30-may-2020 250 \
VENDOR_STRING=Q..alphanum_string.. SUPERSEDE ISSUED=02-May-2019 \
START=02-May-2019 SIGN="0123 4567 89ab cdef 0123 4567 89ab cdef ..."
# # BCOMP = the feature name
# other blocks for other features follow

If it's node locked it'll be based on host ID, which is the MAC address on
x86 hardware. If you run it in a VM you are able to (in fact, you have to)
set an arbitrary MAC address...

Theo

Louis van Dompselaar <vandompselaar@gmail.com> writes:

> Was D4305A ever released license-free (like D7305) or was a generic
> code ever issued?

Looking at the released source code for oc, the FlexLM license check is
done in a single place -- check_licence_manager (note -ence not
-ense!). It's #ifdef-ed out in the released code, but assuming the code
is the same in D4305A, it should be possible to patch it to just
return...

--
Adam Sampson <ats@offog.org> <http://offog.org/>

Op dinsdag 30 mei 2023 om 18:30:05 UTC+2 schreef Adam Sampson:
> Looking at the released source code for oc, the FlexLM license check is
> done in a single place -- check_licence_manager (note -ence not
> -ense!). It's #ifdef-ed out in the released code, but assuming the code
> is the same in D4305A, it should be possible to patch it to just
> return...

Thanks. I'll have a look at that.

It's good to see a British company using British spelling ;)