@Jurassicc

It's strange that it doesn't work for you, because I have also tested a WPN311 in a Peg-II, albeit a G4 version. Maybe it's down to the OS version. What version are you running?

@Cara

To be clear, by "hang" I meant that it didn't print more output after what's shown in the screen grab you sent me. Does it?

I don't know what program will show you the interrupt number. Maybe Ranger or UBoot, or maybe one of the OS4 experts here can advise

@Cara

Thanks, I got it. Does it hang indefinitely at that point? If so, it could be an interrupt problem. Do you know what interrupt number is assigned to the device?

@Cara

Please post the output from running "C:WirelessManager atheros5000.device VERBOSE" manually.

Please test this new version that uses the new DMA functions.

Anyone who's experiencing crashes when starting the driver, please try this test program and paste what it prints out here (and if it crashes).

@utri007

I'm fairly sure the same problem would occur with any Atheros card for now.

It looks like CachePostDMA() is causing the crash. Do any programmers have any comments or experience with this function on the SAM (Flex)?

@Cara

The crash occurs while calling the Exec function CachePostDMA(). Here's a *nasty* debug version that will either crash for you as before or hang the program or system. Please report which of these it does, as it will help me find the cause of the problem.

@zzd10h:

Not in the short term. The problem isn't that your card is PCI-E, but that it uses a newer 802.11n-capable chipset. The driver is based on the official Atheros HAL, and I'm not sure if the HAL properly supports 802.11n-capable cards. It will take some time to investigate this, and maybe upgrade the driver to use a newer version of the HAL.

@Cara, utri007 and others:

Please test this new debug version. There may have been problems with DMA alignment.

@imagodespira

The new drivers are linked in post #7. You will need to rename the debug driver to the normal name before use.

@imagodespira

Please try both new driver versions in your X1000. If you still get the GR, post/send the log from the debug version.

The problem with your Sam is probably different and unrelated to the driver.

I don't have an OS4 system with PCIe (nor even such a PC).

Here's a new version that works with the DWL-G520 in my A1SE (I hope it's not just luck), and here's a debug version.

@imagodespira

I found your email in my spam folder.

Quote:

Take the battery out? (I take no responsibility...)

@utri007

I also have a DWL-G520, and, as the driver manual indicates, it works in my Pegasus-II but not in my A1SE. I currently don't have any ideas on how to debug this. Maybe someone with low-level PPC knowledge could give some advice. I already had to add eieio and sync instructions around register accesses to get the driver working on the Peg-II. The weird thing about the G520 bug is that the instruction pointer seems to end up at zero! Maybe it's interrupt related.

@imagodespira

Sorry, I don't think I got any communication from you. Did you send an email? It does sound like the problem with your card may be easier to debug. I might post a test driver with debugging symbol information.

@billt

What computer do you have? The Prism-II card does indeed need to be reflashed. I don't think AmigaKit-supplied cards were sold already reflashed in the past.

MiniPCI cards in adapters will probably work.

There is only one DWL-G520 AFAIK, and I have one myself that works. Don't worry about the extra marketing terms

Quote:

"heti"?

Quote:

Sure, the email address in the archive and Aminet readme is also my PayPal address.

Have you tried the D-Link card yet, or are you waiting for it to arrive?

Quote:

Do you mean DWL-G510? If so, it must be a revision B card to work with atheros5000.device.

The Sam fix for prism2.device should be already included in atheros5000.device (and prism2 V2.x).

It seems to me there are several variants of the G3SE. Mine never worked properly with UDMA, but works fine 99.9% of the time with PIO (I occasionally get disk error requesters, but clicking Retry always cures them).

I agree though that it should never have made it to market. It's not a sign of a good computer that you have to fill up all your PCI slots to get basic functionality.

EDIT: The OP might want to check if the 3Com Ethernet is listed as a PCI device. PCI IDs are shown at boot, and I think there are utilities to list them in AmigaOS.

Quote:

I don't think it's memory related. That message is a typical symptom of a locked-up driver or NIC. The driver has taken all of Roadshow's fixed number of buffers and isn't giving them back.