Filter received mdns records to chip only. #7527
Conversation
pullapprove
bot
requested review from
chrisdecenzo,
jelderton,
jmartinez-silabs,
LuDuda,
mspang,
saurabhst and
woody-apple
| @@ -238,11 +223,26 @@ void PacketDataReporter::OnResource(ResourceType type, const ResourceData & data | |||
| } | |||
| else if (mDiscoveryType == DiscoveryType::kOperational) | |||
| { | |||
| OnOperationalSrvRecord(data.GetName(), srv); | |||
| // Ensure this is our record. | |||
There was a problem hiding this comment.
nit: reword comment? I am unsure if this helps provide more context or information beyond reading the code.
| @@ -238,11 +223,26 @@ void PacketDataReporter::OnResource(ResourceType type, const ResourceData & data | |||
| } | |||
| else if (mDiscoveryType == DiscoveryType::kOperational) | |||
| { | |||
| OnOperationalSrvRecord(data.GetName(), srv); | |||
| // Ensure this is our record. | |||
| if (HasQNamePart(data.GetName(), kOperationalServiceName)) | |||
There was a problem hiding this comment.
previous code was checking that the suffix was fully _chip._tcp.local.
updated code seems to check only that "_chip" occurs somewhere in the qname. Is this ok and sufficient? Maybe a comment of "note: this does not fully validate path" may be useful if we at some point decide we really need to fully validate the path.
|
Merged: comments were nitpicks |
| } | ||
| else | ||
| { | ||
| mValid = false; |
There was a problem hiding this comment.
This is marking the whole response as invalid even though the record we're looking for might be later in the packet, causing #9099. In fact this is almost always happening in some test scenarios with thread devices.
| @@ -238,11 +223,26 @@ void PacketDataReporter::OnResource(ResourceType type, const ResourceData & data | |||
| } | |||
| else if (mDiscoveryType == DiscoveryType::kOperational) | |||
| { | |||
| OnOperationalSrvRecord(data.GetName(), srv); | |||
| // Ensure this is our record. | |||
| if (HasQNamePart(data.GetName(), kOperationalServiceName)) | |||
Problem
Discovery was returning all sorts of non-chip devices and this was causing out of memory errors and spam.
Fixes #7327
Change overview
Filters to _chip or _chipc devices as appropriate.
Testing
Tested on M5 with chip-device-ctrl discover -all (also tested by mturon@, who was previously seeing other devices)