CTPCI - strange behaviour with USB or Realtek card

[Kroll]

HI,
I decided to start a new thread about this problem.
I have CTPCI, which doesn't work well with a USB or Realek card. At the beginning I will say that it was tested on different Falcon motherboards, different CT6x, ribbon cables, additional working USB cards and realtek on another CTPCI unit.
The problem is manifested in the fact that the connected mouse to the USB card works for a moment, a message appears on the screen, the computer still works, but already using a normal mouse.

The network card behaves identically, I have an internet connection and after a moment I can see this connection will be broken.

Unfortunately, I can't send photos of this message at the moment, what appears when the mouse stops working. All equipment is at a friend who replace the main XIlinx and PCI chip.
Radeon works without any trouble of course. Uploaded the latest flash from Ektus (2.02 Beta 10).
Unfortunately, we have no idea where the cause lies. Has anyone encountered something like this? Does this mean that I have a copy of CTPCI that will not work with these cards.

Regards
Kroll

[wongck]

I don't really use my USB card on my CTPCI so I cannot say much about the USB part. But I think there is a pop-up due to asking you to use a proper TOS version or something like that.

As for CTPCI not behaving well, I would flash the CTPCI firmware again in case the firmware got some bit rot.

[Latz]

Hi,

did you test the cards one-by-one, means Radeon+Realtek, Radeon+USB? Same behaviour?
Are the voltages on CTPCI slot board stable (with 2 or more cards mounted)?
The pop-up wongck mentioned was caused by FreeMinT, this is fixed since a few months.

Latz

[Kroll]

I don't really use my USB card on my CTPCI so I cannot say much about the USB part. But I think there is a pop-up due to asking you to use a proper TOS version or something like that.

OK I see, but do you use Realtek card? I have a similar problem on the network card, it breaks the connection after a short time

As for CTPCI not behaving well, I would flash the CTPCI firmware again in case the firmware got some bit rot.

I know from my friend @RJ1307. He has my CT60e and CTPC. he did it several times, cleaned, uploaded a new flash

[wongck]

A short time is how long... 15 minutes ?
I have used my CT63, Radeon and Realtex for like 1+ hours transferring some files (FTP) for backup between my PC and Falcon.
I did not really use browser for it. I did not really count the time but it was 1+ hour total.

[Kroll]

A short time is how long... 15 minutes ?

It is much faster a few seconds, maybe a little longer, it's hard to say, it can be seen especially in the case of mice because the message appears, it can take different 20-30 seconds.

I have used my CT63, Radeon and Realtex for like 1+ hours transferring some files (FTP) for backup between my PC and Falcon.
I did not really use browser for it. I did not really count the time but it was 1+ hour total.

This problem is very visible. On my other Falcon in the same configuration I have no trouble with either the mouse or the network card.
Completely, my colleague or me, we have no idea what the reason is, such a problem of this selected CTPCI

[Kroll]

Hi,

did you test the cards one-by-one, means Radeon+Realtek, Radeon+USB? Same behaviour?

Yes exactly

Are the voltages on CTPCI slot board stable (with 2 or more cards mounted)?
The pop-up wongck mentioned was caused by FreeMinT, this is fixed since a few months.

Latz

Yes, completely independent of the number of cards connected. This is regardless of the system whether TOS or FreeMiNT.
In addition, I have a modified PCI board to work also with 5V, I can but do not need to connect an additional 3.3V.

[Kroll]

I have some interesting news from RJ1307
He tested and the same bahaviour with or without Radeon card.
As a curiosity I received from a friend. You turn on the computer and won't move it, nothing happens even for two hours. Just moving the mouse for a few seconds will cause it to hang.
Very strange and not typical behavior.
Tested on several types of mice.

[wongck]

Is that USB mouse? or the USB card ?
If I use USB mouse, the cursor is not moving correctly.
if I use USB keyboard, some keys are not correct.
Only USB thumb drive was working for me, and I do not really need USB drive to do file transfer.
So I don't use USB card.

IDK, but one thing to try is to disable the DMA on your CT60CONF CPX.

[Kroll]

Is that USB mouse? or the USB card ?
If I use USB mouse, the cursor is not moving correctly.

I test with USB mouse and my friend with other usb mouse too, the cursor is moving correctly. I have in my second Falcon with CTPCI and USB mouse is working fine. USB mass storage too.

if I use USB keyboard, some keys are not correct.

Yes, I know that and that's why I don't use a USB keyboard

IDK, but one thing to try is to disable the DMA on your CT60CONF CPX.

In this moment I do not remeber that I tested with or without DMA disable. Additionally Rodolphe told me about it in private mail too.
My friend has already packed it for me and he will send me on Monday.

Simple question: What is default value of DMA: enable or disable after a new flash CT6x and CTPCI ?

[Ektus]

I don't know the default state. Simply test with both? Can't find that setting in my CPX, though. My CPX says "V2.02 Sept 2012".
Settings: FPU "on", speed 95.000MHz, "/5" (what's that mean?), SCSI arbitration "on" with ID 7, Blitter "fast", video.log "on", TOS "Cache 5", boot.log "without", IDE "CTPCI".
It has settings for RTL8139 and XHDI/SCSI in the "Stop/diverse" panel that are both set to "off". Internal network driver in TOS?

I don't use my CT63 much, but during my tests didn't have trouble with the network card. The machine has very annoying problems with the Eiffel, though (spurious keyboard event when moving the mouse), and mostly sits there powered down and waiting for me to find a solution to that problem.

[Kroll]

I don't know the default state. Simply test with both? Can't find that setting in my CPX, though. My CPX says "V2.02 Sept 2012".
Settings: FPU "on", speed 95.000MHz, "/5" (what's that mean?), SCSI arbitration "on" with ID 7, Blitter "fast", video.log "on", TOS "Cache 5", boot.log "without", IDE "CTPCI".

First: In CPX I have version 2.02 May 2012 ? The same information during the boot computer ? Have you gat a newest version ?

I have very similar but "Without delay cache and /2 (I dont know what is mean too ?)

The option "USE DMA" I have in Video section.

It has settings for RTL8139 and XHDI/SCSI in the "Stop/diverse" panel that are both set to "off". Internal network driver in TOS?

Yes if you set RTL drivers Yes I think that is network drivers in TOS, but I does not working. I have always use STiNG to connection under TOS.

I don't use my CT63 much, but during my tests didn't have trouble with the network card. The machine has very annoying problems with the Eiffel, though (spurious keyboard event when moving the mouse), and mostly sits there powered down and waiting for me to find a solution to that problem.

With Eiffel I have a similar problem, but since two or three years I use original keyboard.

[Ektus]

I've got several betas, but the number ten from may 6th was the only one to work with the network card. Obviously, I've got a somewhat later CPX installed.

I don't use network with TOS (or TOS at all) on that machine, and the NetUSBee for network when running MagiC with MagiC-Net.

[Kroll]

@Ektus, can you send me later version of CPX.
I see that you use MagiC system. I have all system in my Falcon

[wongck]

DMA is under the last item in pull down - video section.

/number is the divider to the speed of the PCI - you have to get a hardware hack for it - idea was to lower PCI speed to get stability

cache - i always used 5 sec delay - as it works better with programs but it is a bit slower speed but... hey! it is already faster than 16MHz

Internal RTL8139 settings - it works for TOS and stops for Mint. It lack the resolver so it works with ip address only, so only local network. I think mostly used for Didier to transfer binaries from his dev system to the test falcon.
I was thinking you just need to replace the resolver part of the STING cookie offset pointer to a working resolver, then may be it will work for Internet.

For me, both Radeon and Network is working very strong stable. USB only thumb drive is working.

[wongck]

I've got several betas, but the number ten from may 6th was the only one to work with the network card. Obviously, I've got a somewhat later CPX installed.

I have beta 11 but that one causes my HDD to be corrupted. I think Didier was testing with XHDI stuff.

USB Keyboard - according to David, the issue was fixed in beta 12 but I do not have it.

Pity that Didier stopped work.
I think it is the tools that are not working well - like the gcc compiler that gives different results with different settings.

[wongck]

Simple question: What is default value of DMA: enable or disable after a new flash CT6x and CTPCI ?

yes simple question .... answer = no clue - I have not flashed recently.

Actually a lot of things is walking in the dark and discover.
That's why it is also "fun" for me, as I can still discover something new every few "years"

[Kroll]

I have beta 11 but that one causes my HDD to be corrupted. I think Didier was testing with XHDI stuff.

USB Keyboard - according to David, the issue was fixed in beta 12 but I do not have it.

Pity that Didier stopped work.
I think it is the tools that are not working well - like the gcc compiler that gives different results with different settings.

Were all Beta versions 10 and later developed only by Didier?
Exactly what a pity it stopped developing, CTPCI still has many users and is coming, even my friend from Poland . For several weeks he has been a very active user

[tksm]

Hi.
I am new user of CTPCI thanks to my Friend Kroll who is extremally helpful for Atari Community.
I Have CT63E from Willy and I can confirm that with new flash(on CT60E) CTPCI works just fine with radeon, usb card(mouse and pendrive works fine) and network card also.
Pity that IDE on CTPCI is not working.
I would like to thank all of You who make drivers and other stuff for CTPCI

[Galvez]

Were all Beta versions 10 and later developed only by Didier?

Yes, it's all his own work.

[Kroll]

Today my CTPCI came back to me, of course I did tests with DMA or without it. After running TOS, the mouse works for sometimes 20 seconds, sometimes more, but then a message appears on the screen. The system still works, but I can use a mouse connected to the standard keyboard ports.

Mouse.jpg

The network card is disconnected, only working with Radeon.
Does anyone have an idea? What does this error mean?

[wongck]

Sorry
Do not know what the error means.
Also I do not remember seeing this error before.

[mikro]

It comes from ohci-hcd.c from Didier's USB driver within CTPCI TOS. More specifically, from an old version of U-Boot: https://github.com/BPI-SINOVOIP/BPI-M3- ... ohci.h#L81

Maybe David could tell you more?

[Galvez]

Maybe David could tell you more?

Sorry I'm afraid I can't help here. :-/

[wongck]

My guess is that the USB card have 5 ports, and you have another port on the card ?