Open nielsonm236 opened 3 months ago
After a lot of testing we're finding that applying a Broadcast filter is causing more trouble than it is worth. The main problem with the filter is that Gratuitious ARP is not operating as expected., causing significant delays in Browser communication with the HW-584, making it look like the device has become unresponsive.
Further testing also suggests that ARP Broadcast traffic is not the source of the problem in issue #221, so the Broadcast filter idea will be abandoned while other solutions are pursued.
I'm removing the "Gratuitous ARP" code as the experiment failed.
Issue #221 reported symptoms that looked like the settings corruption previously seen with faulty power supplies. However, extensive test in the environment did a good job of ruling that out for the case of #221. Subsequently it was found that the exact same device and firmware performed without issue in a home network environment. This pointed to network traffic in the original network as the source provoking the issue, and further investigation determined Broadcast traffic could potentially overwhelm the current firmware design, AND that it is possible to filter out all Broadcast traffic within the hardware of the ENC28J60. Performing this hardware filtering would reduce the processing burden in the firmware, and would likely increase the availability of the limited receive buffer in the ENC28J60 (although visibility to that resource is limited, so actual effect is undetermined).
Sill, performing hardware filtering of Broadcast traffic has got to be beneficial and will be released as soon as possible.