Communication issue with serial port

[videobruce]

I have two identical MB's both using XP Pro w/sp2
They are the Foxconn N570SM2AA.

With two different serial devices (a FTA receiver and a RF signal level meter) I have to first disable the serial port, then re-enable it to be able to communicate with either device. If I don't, I get a error message.

My previous MB had a nForce2 chipset and used 2k. I had no issue with two different FTA receivers using the serial port.

With another piece of test equipment, there is a different problem. The software locks up after 5 seconds after hitting the 'start' button to try to connect with the device (after the s/w is opened). I believe this is aseparate issue since the disable/enable doesn't help. (There was no issue with the previous MB and O/S.)

Any ideas about why I have to disable, then re-enable the port everytime?
BTW; no conflicts in DM and it is the stock Windows driver that is loaded.

[Gizmo]

Now that's weird.

Both of them original Windows loads (i.e. they were loaded on the mobo they are running on now)?

[videobruce]

Yes to that question, but it gets better.

Elsewhere, I was told to check the serial port settings in the bios. To set them manually to 3F8 & IRQ4. Also, to set the PIC bus to manual.
I first checked Device Mgr. The resources were; 3F8 & IRQ7.

I went into the bios, the serial port was already set to manual, 3F8 & IRQ4.
I changed the PnP/PCI config. from auto to manual. That was all I did.

This appears to work.

Ok, Was the fact the PnP/PVCI config. was set to auto caused Windows to change it in DM to 7 from 4?
If, so, why would disabling and then re-enabling the port temporarily solve the problem??

[Gizmo]

Setting the PnP/PCI config to manual in the BIOS essentially tells the OS that it can't reassign those resources. With it set to auto, Windows was free to reassign IRQs if it thought there was a need to do so, and that was apparently what caused the issue.

As for why disabling and re-enabling the port would temporarily resolve the issue, I can only surmise that the driver would initialize the port to IRQ 4, and then when it got reset after the app exited, the Windows PnP configuration would re-assign the IRQ to 7. But that's just a guess. I'm not convinced ANYBODY really knows what is going on inside Windows any more (though Mark Minassi probably knows more than most).

[videobruce]

I'm not convinced ANYBODY really knows what is going on inside Windows any more

Does that inclide M$?

With the above in mind, is the "assigning" is done at boot time? If so, then disabling and reenabling the port puts it to the bios setting bypassing the PnP process?

Does that make sense?

[Áedán]

That's a good question... For a system using a HAL that is ACPI aware, the resources are allegedly configured at boot time. MS claim that modern hardware is complex enough that they don't attempt to support rebalancing. They cite things like multiple roots, IOAPICs and other reasons...

[videobruce]

You sort of lost me with most of that.

[Áedán]

Sorry. Windows NT/2K/XP/2K3/Vista/2K8 all have what's called a "Hardware Abstraction Layer" or HAL for short. Diffferent HALs are used in different circumstances. For an x86 based ACPI based platform, Windows installs a particular HAL that talks to the ACPI stuff in the BIOS. Apparently due to the complexities of the hardware, MS don't attempt to do anything but configure resources at boot time. Interrupts have become a lot more complex, especially when there's multiple processors.