Help needed testing STinG drivers

All about the serious stuff.

Moderators: Mug UK, Zorro 2, Moderator Team

Post Reply
ThorstenOtto
Atari God
Atari God
Posts: 1521
Joined: Sun Aug 03, 2014 5:54 pm

Help needed testing STinG drivers

Post by ThorstenOtto »

While working on the STinG driver for the (hopefully upcoming) network support for Hatari, i still encounter strange problems. The symptom is that tools like Ping and Traceroute work for a while, but then suddenly don't receive any packets any more.

Since i had some general problems with network setup in aranym too, this is a bit hard to debug, and can be caused by any of the components involed: could be a bug in the driver, or in STInG, or in the interface between STING & Hatari/ARanYM, or in my setup, or just in the tools that generate the packages. I've already checked that the packages go out of the host (by adding debug output in Hatari), but both tools just don't get any answer after a while.

So i would ask: can somebody run those tools with other drivers (eg. for lightning) and tell me if they work as intended? I would be interested in a result of both the tools from the original package, and the ones attached below.
You do not have the required permissions to view the files attached to this post.
User avatar
christos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2506
Joined: Tue Apr 13, 2004 8:24 pm
Location: Greece

Re: Help needed testing STinG drivers

Post by christos »

Do you need these tests from real hardware? I can check on my MSTE with Ethernec
ThorstenOtto
Atari God
Atari God
Posts: 1521
Joined: Sun Aug 03, 2014 5:54 pm

Re: Help needed testing STinG drivers

Post by ThorstenOtto »

On any hardware with an already working STinG stack. But that would exclude most emulators, anyway ;)
User avatar
christos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2506
Joined: Tue Apr 13, 2004 8:24 pm
Location: Greece

Re: Help needed testing STinG drivers

Post by christos »

Cool, I'll check it out later today.
User avatar
christos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2506
Joined: Tue Apr 13, 2004 8:24 pm
Location: Greece

Re: Help needed testing STinG drivers

Post by christos »

Here are the results
tracert.jpg
ping.jpg

The tools in my installation work as well. Do you need anything else?
You do not have the required permissions to view the files attached to this post.
slingshot
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2254
Joined: Mon Aug 06, 2018 3:05 pm

Re: Help needed testing STinG drivers

Post by slingshot »

Tried on MiST with STinG and Ethernec, it works (waited for 200 pings).
artik-wroc
Captain Atari
Captain Atari
Posts: 180
Joined: Fri Jan 06, 2006 1:30 pm
Location: Wroclaw - POLAND
Contact:

Re: Help needed testing STinG drivers

Post by artik-wroc »

MegaST, PAK, Panther/2 it works.
ThorstenOtto
Atari God
Atari God
Posts: 1521
Joined: Sun Aug 03, 2014 5:54 pm

Re: Help needed testing STinG drivers

Post by ThorstenOtto »

Many thanks ! So it looks like there's just still something wrong with my network setup.
User avatar
TheNameOfTheGame
Atari God
Atari God
Posts: 1736
Joined: Mon Jul 23, 2012 8:57 pm
Location: Almost Heaven, West Virginia

Re: Help needed testing STinG drivers

Post by TheNameOfTheGame »

Glad to hear it got tested with the Panther/2. My system is disassembled at the moment since I am working on the re-casing of it into the XT case otherwise I would have gladly tested it.
User avatar
Ektus
Captain Atari
Captain Atari
Posts: 276
Joined: Mon May 24, 2010 2:58 am
Location: Germany
Contact:

Re: Help needed testing STinG drivers

Post by Ektus »

Trying to test on my Milan 040 running MagiC-Net with GlueSTiK: Exception #4 from ping, warning not to use with STiK from tracerout.
Schneider CPC464 (long retired), Atari Mega ST4 (retired), Falcon+CT2A, Falcon+CT63+CTPCI+Radeon 9250, Milan040+SCSI+Rage
ThorstenOtto
Atari God
Atari God
Posts: 1521
Joined: Sun Aug 03, 2014 5:54 pm

Re: Help needed testing STinG drivers

Post by ThorstenOtto »

Hm, the exception is maybe from the gem library, i think it is still using movep somewhere. And for Gluestik, i think that is only supporting the original STiK interface, not STinG (they are slightly different).
ThorstenOtto
Atari God
Atari God
Posts: 1521
Joined: Sun Aug 03, 2014 5:54 pm

Re: Help needed testing STinG drivers

Post by ThorstenOtto »

I've now fixed traceroute so that it should also work with Stik/Gluestik. But the other problem will remain. Can you verify that the exception is caused by a movep instruction? Or is it maybe some other instruction not supported by '040? (the program does not use floating point)
User avatar
TheNameOfTheGame
Atari God
Atari God
Posts: 1736
Joined: Mon Jul 23, 2012 8:57 pm
Location: Almost Heaven, West Virginia

Re: Help needed testing STinG drivers

Post by TheNameOfTheGame »

Interesting. I tested two configurations with GlueSTiK and both crashed when running PING.PRG. TRACROUT.PRG on both machines gave the "Don't use this program with STiK!" message.

GlueSTiK v0.15
STE/MagiC - PING.PRG crashed with "Runtime error 163 at offset $000A3FB"
Hades060/MagiC - PING.PRG crashed with "Priveledge violation"
ThorstenOtto
Atari God
Atari God
Posts: 1521
Joined: Sun Aug 03, 2014 5:54 pm

Re: Help needed testing STinG drivers

Post by ThorstenOtto »

Yes, tracrout does not get the exception because it aborts right away, because the MODULE_LAYER interface is not supported by STik/Gluestik. However i might be wrong with my assumption about movep, because the form_alert() apparently works ;)
Unfortunately i was also wrong with my statement about having tracrout fixed, there is another function needed that i missed in the first place (IP_send). Haven't figured out yet what's the equivalent when just using STiK. Ping should now work theoretically, but it's compiled using the StinG definitions, and i might have missed some subtle differences to STiK.
Its also a bit strange tat you get different exceptions. And "offset $A3FB" is also strange, the program isn't that large...

Edit: did you get any answer from Vassilis yet regarding the source of MagCNet? (https://www.atari-forum.com/viewtopic.php?f=16&t=40009) I found some sources as part of iconnect (https://github.com/ggnkua/Atari_ST_Sour ... onnect_src), but that seems to be a different stack that was onyl used by IConnect/Emailer. I just wonder whether they have anything in common....
User avatar
Ektus
Captain Atari
Captain Atari
Posts: 276
Joined: Mon May 24, 2010 2:58 am
Location: Germany
Contact:

Re: Help needed testing STinG drivers

Post by Ektus »

Here is the crash output form my Milan:

Code: Select all

Exception #4: PC=00000000 (TEXT+FEA16888), USP=015ED4E0, SR=14                       
Dn:00000001 015EB0CC 00000000 FFFFFFFF 00000000 00000000 00000000 00000000           
An:00000000 015ED4E2 015ED9DE 015E9678 015EC51C 015EB5EE 00000000 015977B0           
PD:015E9678 TEXT:015E9778 DATA:015EB044 BSS:015EB1C2
Schneider CPC464 (long retired), Atari Mega ST4 (retired), Falcon+CT2A, Falcon+CT63+CTPCI+Radeon 9250, Milan040+SCSI+Rage
ThorstenOtto
Atari God
Atari God
Posts: 1521
Joined: Sun Aug 03, 2014 5:54 pm

Re: Help needed testing STinG drivers

Post by ThorstenOtto »

Hm. PC=0 looks it tried to jump through some callback that isn't there. Maybe have to check how gluestik for MagiC works.
ThorstenOtto
Atari God
Atari God
Posts: 1521
Joined: Sun Aug 03, 2014 5:54 pm

Re: Help needed testing STinG drivers

Post by ThorstenOtto »

Yes, looks like Gluestik for MagiC was based on an older version that does not implement some of the Functions (the last function implemented is CNgets, and all the following function are not supported). I can work around that so it does not crash, but that will just mean that it cannot work. I hope i can build a new version, but may take a few days.

PS.: i should have known, since i fixed the same problem already in gluestik for mint: https://github.com/freemint/freemint/co ... 2cee46121e
Post Reply

Return to “Applications”