Hi Rudolf, Jordan, > Well we had the problem with i2c-viapro and vi686a. Best solution was > to fail from the probe function - like this: > > /* Always return failure here. This is to allow other drivers to bind > * to this pci device. We don't really want to have control over the > * pci device, we only wanted to read as few register values from it. > */ > return -ENODEV; > > release_region: > release_region(vt596_smba, 8); > return error; > } True, but we were lucky. You can only do that if you don't need to access the PCI configuration space during run-time (after the initialization step) in either driver "sharing" the PCI device. The i2c-viapro driver qualified, but for example we couldn't do that for the i2c-i801 driver (block transactions in I2C mode require PCI configuration register access.) So depending on the exact hardware context, Jordan may or may not be able to use the same trick. The other (cleaner) solution is to have an additional module which registers the PCI device and exports some kind of interface for the other modules to access the parts they need - but this requires extra code and makes things much more complex too. -- Jean Delvare
