Hi,
Here are first comments about technical details, discussion about
licensing will go in another thread...
>> From: Farahmand, Farinam
>> Sent: Monday, November 22, 2004 7:08 PM
>> To: [EMAIL PROTECTED]; [EMAIL PROTECTED]
>> Subject: RE: eagle-usb 2.0.0 status report + evolutions for Sagem/ADI
>>
>> 1- BNM Files are backward compatible. They should work for all the
>> modems. It will be a good idea to always include the latest known
>> working BNM files in the package with the related CMV text file/
>> config Files.
ok, as soon as CMVs are implemented completely (with installation/config)
>>To report any bug related to BNM files, we need to have
>> the list of cmvs used in the package or the cmv.txt file plus the
>> Layer 1 (DSP) firmware version number.
for the moment, available with eagle-usb driver's version number <=1.9.8 -
1.9.9.1/2.0.0 that correspond to a different set of BNM files, latest will
be with 2.0.1 if all goes well.
>> Firmware version number in
>> Eagle products can be obtained by sending "INFO 10" cmv read command.
>> It is a good idea to add L1 firmware version to "eaglestat" command.
From my understanding, you implemented it in your driver. I have several
questions (don't be afraid, I can understand quickly, but it simply takes
time to explain me ;-) ) :
- Could you precise if in "firmware version number", firmware refers to
what appears in firmware.h or to DSPcode as well (I think only the latter,
DSPcode being "compiled" BNM files for me).
- Is firmware version number obtained at the time the firmware is sent ?
Can it be obtained dynamically ? (I mean afterwards, by sending the
command to the modem independantly of sending the DSPcode) :
Once the modem is synchronized ? Could it be obtained before : once
firmware loaded ? or only after DSPcode loaded ?
- In some case, when the DSPcode is not correct (for example sending
pre-2.0.31-Sagem-driver's DSPcode to a fast800 E2T keeps it in the state
INITIALIZING forever...), can the firmware version number be obtained in
this case ? (I don't think so).
Could it be implemented in :
- eaglectrl, that could store it in the driver at firmware load or later ?
- or eaglestat ? (seems complicated...)
- statically : this would not be reliable as the user can change DSPcode
files...
>> 2 - For each release of DSP firmware(BNM files), there will be a
>> default cmv text file. This default file will be good for almost all
>> the DSLAMs out there. I attached the default Cmv text file for eagle
ok attached to this mail, I'll put them on cvs ASAP if needed
>> III. There are some exceptions to this:
>>
>> - Additional / Different Special settings for Telephonica Spain
>> (POTS): OPTN 4 0x08600002
>> OPTN 19 0x80000000
>> OPTN 70 0x218280aa
>> OPTN 73 0x0001000a
>> DIAG 0 0x26878030
>> - Some addition/different CMVs in case default doesn't work vs. GSPN
>> DSLAM (POTS).
>> MASK 8 = 0x1fffffc0
ok this is related to DSLAM config... having the detailed impacts of each
parameter would be a plus if possible (like OPTN1 for download rate
divided by 10). We are playing a guess-at-chance game currently.
>> - currently ADI is working on ISDN support for EAGLE III.
Fine.
>> - CMV text files are not based on different modem models;
thanksfully...
>> Telco's / ISPs might suggest different CMV text files.
hum, a process may be needed so that we get appropriate values, otherwise
it will be a nightmare to configure... It's already not that easy to
obtain VPI/VCI/encapsulation values as you can see at
http://www.eagle-usb.org/article.php3?id_article=23
The "easiest" may be for each ISP to have at least one technical web page
with all parameters (bottom-up approach) : but it's already difficult to
obtain it for VPI/VCI... this could be part of recommandation or
"know-how" best practices ;-)
Or it could be determined beforehand (top-down approach), I would prefer
this, as it can be changed afterwards with user feedback if needed.
>> 3 - If the proper CMV text file is used for both eagle II and eagle
>> III, there should be no major difference in attenuation reporting.
ok, my test was between EagleI (fast 800 AB) and EagleIII (fast 800 E3) :
respectively ~25dB and 16dB (and no more "97th threshold reboot"...)
my point is that 16dB is *really* better than theorical attenuation
(around 24 as I'm about 2500 meters from the DSLAM).
>> 4 - We should see an slightly better SNR in EAGLE III modems than
>> EAGLE II
The Signal Noise Ratio does not appear directly in eaglestat's result. It
could play on attenuation.
>> 5 - Moving to CMV text file implementation, we have to remove all of
>> the cmv related configuration from the config files.
ok, identified
>> 6 - Looking at the original implementation of cmv msg structure,
>> eu_cmv_msg_t.idx is assigned to a dummy value, so you can pick any
>> number to initialize the structure.
fine, we had guessed correctly
>> 7 - Additional cmv settings for Telefonica is mentioned above. For
>> Wanadoo/Eresmas, the default cmv settings should work ( Sagem can
>> comment on that).
>> 8 - VPI/VCI/Encapsulation will be provided by ISPs. We can implement
>> Auto-Configuration for USB driver to automatically detect the values (
>> refer to DSL forum for more detail )
that would be fine, any URL is welcome
I've found documents like
http://www.dslforum.org/aboutdsl/Technical_Reports/TR-046.pdf
that were made in 2000.
>> 9 - default value for OPTN 1 is 0x994. If this reduces the
>> performance or causes other problems, we should discuss with ISPs or
>> Telco's to set the correct value.
Our case is "Free dégroupé" (French ISP), maybe Sagem knows which DSLAM or
which configuration they have. If that's true for all "Free dégroupé"
users, that should be fine, if that's for some "Free dégroupé" users and
not for others (for example depending on the DSLAM they are connected...),
we currently miss a criteria to choose automatically the configuration,
so we'll provide both configurations, letting the user choose.
That's not critical as it "works", more slowly, but it gets connected.
>> Please let me know if there is any other questions regarding the cmv
>> text file implementation.
Some documentation maybe ? About CMV if available. And Sl33p3r would
*really* enjoy having a description of the state machine in the driver, if
possible.
Regards
Ben'. aka baud123
;POTS for EAGLE
CW cntl 0 1
cw optn 1 0x00000994
cw optn 2 0x63600000
cw optn 3 0x00000028
cw optn 4 0x00600000
cw optn 5 0x00000500
cw optn 70 0x218280aa
cw optn 72 0x006f06eb
cw optn 73 0x00010060
cw mask 8 0x3fffffc0
cw diag 0 0x26870030
cw flag 0 0x00000820
cw optn 0 0x80008066
cw optn 18 0x820200ff
cw optn 19 0x80000000
CW cntl 0 2
;POTS for EAGLE
CW cntl 0 1
cw optn 1 0x00000994
cw optn 2 0x63600000
cw optn 3 0x00000028
cw optn 4 0x00600000
cw optn 5 0x00000500
cw optn 70 0x218280aa
cw optn 72 0x006f06eb
cw optn 73 0x00010060
cw mask 8 0x3fffffc0
cw diag 0 0x26870030
cw flag 0 0x00000820
cw optn 0 0x80020066
cw optn 18 0x00000000
cw optn 19 0x00000000
CW cntl 0 2
