Check on UDP length maybe discards correct responses

FStefanni commented 2 years ago

Describe the bug

Hi,

I have this setup:

a Linux laptop
a Raspberry
a Esp32

On all of tem MDNS is configured (on Esp32, I am using the example sketch WiFiResolvingHostNames)

Linux and Raspberry are able to see each others (ping, ssh and avahi-resolve work fine).
Linux and Raspberry are able to resolve the Esp32 ip (ping and avahi-resolve).
Esp32 is unable to resolve Linux or Raspberry ips.

After some debug, I have found this check in the lib code (near line 963 of file MDNS_Generic_Impl.h):

//KH, to report name Resolve only UDP packet has corect size of 48
if (48 == udp_len)

So I put a debug print, and actually what I got is:

udp_len equals to 40
The data pointed by buf actually contains the correct ip.

So the question/bug is:

Why consider only length of 48? Is truly incorrect checking also for different lengths (as in my case for 40)?
If the length of 48 is the only correct, then why the example sketch is unable to resolve the ip? What am I doing wrongly?

Please note that on Rasberry and Linux no acutal services discoverable via MDNS are running: what I need is to just resolve the ip.

Regards

Steps to Reproduce

I am using the WiFiResolvingHostNames sketch.

Expected behavior

I was expecting to be able to resolve the ip.

Actual behavior

The sketch prints Resolving XXX error

Information

Arduino IDE version: 1.8.19
ESP32 Arduino library version: 2.0.3

khoih-prog commented 2 years ago

Hi @FStefanni

Thanks for your usage of the library and the issue report.

As you're using ESP32, I suggest that you use the native ESP32-core ESPmDNS library

If you still have issue, please post on the ESP32 core issue, with the Minimal, Reproducible Example so that other people can duplicate the issue and help.

I'm sorry I don't have time to spend here as the library is mainly designed for boards besides ESP32/ESP8266, and I'm trying to offload any issue to other cores / libraries if possible.

I certainly reopen, spend some time to investigate and fix the library if you can prove clearly this is a bug of the library.

Thanks again and Good Luck,

khoih-prog commented 2 years ago

Hereafter is the debug terminal showing the correct UDP packet size 48 received and processed

[MDNS] Receiving len =  14 , hostname raspberrypi-02
Resolving 'raspberrypi-02' via Multicast DNS (Bonjour)...
[MDNS] ::resolveName: name= raspberrypi-02.local
[MDNS] ::_sendMDNSMessage: xid= 0
[MDNS] ::_sendMDNSMessage: queryCount= 256
[MDNS] ::_sendMDNSMessage: answerCount= 0
[MDNS] ::_sendMDNSMessage: authorityCount= 0
[MDNS] ::_sendMDNSMessage: additionalCount= 0
[MDNS] ::_sendMDNSMessage: statusCode = 1
[MDNS] ::_processMDNSQuery: UDP parsePacket len= 48
[MDNS] ::_processMDNSQuery: UDP parsePacket len= 48
[MDNS] buffer======
0, 0, 84, 0, 0, 0, 0, 1, 0, 0, 0, 0, e, 72, 61, 73, 70, 62, 65, 72, 
72, 79, 70, 69, 2d, 30, 32, 5, 6c, 6f, 63, 61, 6c, 0, 0, 1, 80, 1, 0, 0, 
0, 78, 0, 4, c0, a8, 2, 70, 
[MDNS] =====
[MDNS] ::_processMDNSQuery: to report IP, buf = 192
[MDNS] ::_finishedResolvingName: name = raspberrypi-02 , IP = 192.168.2.112
The IP address for 'raspberrypi-02' is 192.168.2.112
[MDNS] IP[ 0 ]= 192
[MDNS] IP[ 1 ]= 168
[MDNS] IP[ 2 ]= 2
[MDNS] IP[ 3 ]= 112
[MDNS] ::_processMDNSQuery: UDP parsePacket len= 48
[MDNS] ::_processMDNSQuery: UDP parsePacket len= 48
[MDNS] buffer======
0, 0, 84, 0, 0, 0, 0, 1, 0, 0, 0, 0, e, 72, 61, 73, 70, 62, 65, 72, 
72, 79, 70, 69, 2d, 30, 32, 5, 6c, 6f, 63, 61, 6c, 0, 0, 1, 80, 1, 0, 0, 
0, 78, 0, 4, c0, a8, 2, 7a,

FStefanni commented 2 years ago

Hi,

thank you for the reply. So my doubt is: is it really correct to check only for a length of 48? Why am I receiving a length of 40?

Nevertheless, I am going to check the ESPmDNS library library.

Regards

[EDIT] FYI, ESPmDNS library works fine. Thank you for the suggestion.

khoih-prog / MDNS_Generic