eppp_link 1.0.1 CONFIG_EPPP_LINK_USES_PPP forcibly enables PPP server, causing local IP assignment and MQTT connect failures (IDFGH-16352)

[kadir-guzel]

Answers checklist.

• I have read the documentation for esp-protocols components and the issue is not addressed there.
• I have updated my esp-protocols branch (master or release) to the latest version and checked that the issue is present there.
• I have searched the issue tracker for a similar issue and not found a similar issue.

General issue report

Bug description:
I discovered this issue while running the host example of eppp_link. When enabling CONFIG_EPPP_LINK_USES_PPP, the Kconfig automatically enables CONFIG_LWIP_PPP_SERVER_SUPPORT and doesn’t allow the user to disable it via menuconfig.

This causes esp_netif_lwip_ppp.c to assign a local IP address to the device. As a result, the MQTT client attempts to connect using this local IP instead of the expected IP from the PPP peer (e.g., modem), and the connection fails with esp-tls errors.

Steps to reproduce:

1. Open the host example of eppp_link.
2. Enable CONFIG_EPPP_LINK_USES_PPP in menuconfig.
3. Build and run the example.
4. Observe that CONFIG_LWIP_PPP_SERVER_SUPPORT is automatically enabled.
5. Application receives local IP from esp_netif_lwip_ppp.c.
6. MQTT connection fails since it tries to use that local IP.

Observed log snippet:

ip4_output_if: call netif->output()
pppos_netif_output[1]: proto=0x21, len = 44
...
|  192  |  168  |  11  |  2  | (src) // local ip from esp_netif_lwip_ppp.c
...
E (591827) esp-tls: [sock=54] select() timeout
E (591827) esp-tls: Failed to open new connection
E (591827) mqtt_client: Error transport connect

Expected behavior:

• CONFIG_LWIP_PPP_SERVER_SUPPORT should not be automatically forced when CONFIG_EPPP_LINK_USES_PPP is selected.
• User should be able to choose client/server PPP mode explicitly.
• For PPP client use-cases, the IP assignment should come from the PPP peer, not from esp_netif_lwip_ppp.c.

Proposed fix:

• Remove the hard dependency in Kconfig so CONFIG_LWIP_PPP_SERVER_SUPPORT can be disabled.
• Allow EPPP_LINK_USES_PPP to work in PPP client mode without assigning a local IP from esp_netif_lwip_ppp.c.

Additional context:
Tested on ESP32-S3 using the host example. This issue affects any project that enables CONFIG_EPPP_LINK_USES_PPP, not just the example.

[david-cermak]

Thanks for reporting the problem. Would it work if you reset the default addresses:

diff --git a/components/eppp_link/examples/host/main/app_main.c b/components/eppp_link/examples/host/main/app_main.c
index b082137e..c19437a8 100644
--- a/components/eppp_link/examples/host/main/app_main.c
+++ b/components/eppp_link/examples/host/main/app_main.c
@@ -125,6 +125,8 @@ void app_main(void)
     config.transport = EPPP_TRANSPORT_SDIO;
     config.sdio.is_host = true;
 #endif
+    config.ppp.our_ip4_addr.addr = 0;
+    config.ppp.their_ip4_addr.addr = 0;
     esp_netif_t *eppp_netif = eppp_connect(&config);
     if (eppp_netif == NULL) {
         ESP_LOGE(TAG, "Failed to connect");

[kadir-guzel]

Thanks for the quick response, I tested your suggested change and it worked. The host now receives the correct IP from the nRF9160 SLM side during PPP negotiation, and the MQTT connection succeeds without hitting the local 192.168.x.x issue.

This resolves the problem on my side. Thanks again for the fix!

[david-cermak]

Thanks for letting me know.

Please keep this open, I'd like to address it in the codebase in some way. Ideally letting users to choose client/server side runtime (when CONFIG_LWIP_PPP_SERVER_SUPPORT is enabled)