[ros-dev] Loading failure of rtl8139.sys driver = OK+ Ndisfunction unimplemented

[Eric Kohl]

"Filip Navara" <xnavara at volny.cz> wrote:

> No. In fact it was registry handle till NDIS4 and since NDIS5 it's a
> structure. I have this part partly rewritten in my tree.

Ah, I see!


> Incidentally I have NdisMQueryAdapterResource implemented in my local
> tree. It gets the resources from the AllocatedResources field in the
> NDIS_MINIPORT_BLOCK structure which is itself filled in the
> IRP_MN_START_DEVICE dispatch code. There are two problems with this:
> 1) The PnP resource assigning in kernel isn't implemented yet (only the
> untraslated boot configuration is reported to device now, but it might
> be enough for this case) We need to assign the resources to each device
> based on the resource requirements (see
> IRP_MN_QUERY_RESOURCE_REQUIREMENTS, IRP_MN_FILTER_RESOURCE_REQUIREMENTS
> and _DEVICE_NODE) and then translate them to bus neutral form. (I've
> written the translation code, but I seriously doubt it's correct.)

Yep! That's where I got stuck because the hardware I have needed more work
on other places. Could you commit your experimental code? Maybe it is a
usable foundation.

Btw, I am removing variables like HardwareId from DEVICE_NODE (everything
below the '/* Not NT's */' comment) and store it in the registry instead.
IoGetDeviceProperty will read them from the registry.


> 2) Currently the RTL8193 card isn't found by our PCI driver (which
> reports the resources) and so we need this too to proceed further.
> (Gerard/Hartmut, if you want to help me with this problem, please run
> http://www.volny.cz/xnavara/devcon.zip on your system under Windows and
> send me the results back).

An RTL8193 is enumerated properly on one of my machines but it freezes on
the other one.


Regards,
Eric