In the current version, udev rules are generated with hardcoded KERNEL udev entries for PCI devices that were found when the sdk_setup.sh script is run.
But, if the number of FPGA change, the udev rule is not updated without re-run the script.
This can be an issue if building an AMI on an instance with a single FPGA (or even no FPGA) and then using the AMI to start an instance type with more FPGA devices.
So, the udev rule was changed to dynamically detect devices independently to the initial configuration.
In the current version, udev rules are generated with hardcoded KERNEL udev entries for PCI devices that were found when the sdk_setup.sh script is run.
But, if the number of FPGA change, the udev rule is not updated without re-run the script. This can be an issue if building an AMI on an instance with a single FPGA (or even no FPGA) and then using the AMI to start an instance type with more FPGA devices.
So, the udev rule was changed to dynamically detect devices independently to the initial configuration.