We are designing a PCI board and when the prototype came back from assembly,
we discovered a bug. It asks for megabytes of I/O space on BAR 1, when it
doesn’t need any I/O space and this is causing problems with resource
allocation. It only needs some memory space, via BAR 0.
This will be fixed before handing it off to the client, but as a temporary
workaround, I figured I could handle IRP_MN_FILTER_RESOURCE_REQUIREMENTS in
the driver. Starting with a code sample from Walter’s book, I have been
unable to make this work. My first thought was to just change the I/O
requirement to a more reasonable 32 bytes, didn’t work. Then I tried
removing the I/O requirement from the resource list, didn’t work. The driver
would not start in either case.
Should this work or do I just have a bug in my code I have been unable to
find? Any insights?
Resource filtering does work for me.
You are filtering the Irp on its way back up, are you?
Robin
-----Original Message-----
From: James A [mailto:xxxxx@hotmail.com]
Sent: Wednesday, February 05, 2003 17:50
To: NT Developers Interest List
Subject: [ntdev] Question on: IRP_MN_FILTER_RESOURCE_REQUIREMENTS
We are designing a PCI board and when the prototype came back from assembly,
we discovered a bug. It asks for megabytes of I/O space on BAR 1, when it
doesn’t need any I/O space and this is causing problems with resource
allocation. It only needs some memory space, via BAR 0.
This will be fixed before handing it off to the client, but as a temporary
workaround, I figured I could handle IRP_MN_FILTER_RESOURCE_REQUIREMENTS in
the driver. Starting with a code sample from Walter’s book, I have been
unable to make this work. My first thought was to just change the I/O
requirement to a more reasonable 32 bytes, didn’t work. Then I tried
removing the I/O requirement from the resource list, didn’t work. The driver
would not start in either case.
Should this work or do I just have a bug in my code I have been unable to
find? Any insights?