Repository navigation
WLED-AP still active when connected over ethernet - expected to deactivate Wifi completely #5762
Description
Activity
- changed the title
[-]Deactivate Wifi completely[/-][+]Wled-ap still active when connected over ethernet - expected to deactivate Wifi completely[/+]on Aug 2, 2026 - changed the title
[-]Wled-ap still active when connected over ethernet - expected to deactivate Wifi completely[/-][+]WLED-AP still active when connected over ethernet - expected to deactivate Wifi completely[/+]on Aug 2, 2026 similar to
however PR 5394 intended to save flash by excluding all WiFi functionality. PR 5394 was considered too dangerous for normal use, because it will render esp-now remotes unusable, plus it makes Ethernet config impossible, because AP mode is removed completely.
Reacted by chrisp1001You can also disable the AP mode in the config
You can also disable the AP mode in the config
I couldn't make it work as described above. Let me know if you have a different solution.
Btw, this issue has been addresses here in past: tartinesKiller@47943c4
@chrisp1001 - no just use
@chrisp1001 - no just use
It doesn't work like this unfortunately. You may have read this in the link I mentioned in my inital post.
@coderabbitai please analyse the source code in our
16_xbranch. What happens if WLED is connected via ethernet, and apMode is set to 'Never'? Is the behaviour different from previous releases likev0.14.4(tag)?🧩 Analysis chain
🏁 Script executed:
#!/bin/bash set -euo pipefail printf '%s\n' '== Relevant refs ==' git branch -a --no-color | grep -E '16_x|v0\.14\.4' || true git tag -l 'v0.14.4' || true printf '%s\n' '== Locate configuration and network symbols ==' rg -n -i -C 3 '\bapMode\b|AP\s*mode|WIFI_AP|ETH\.|Ethernet|ETHERNET' \ -g '*.{cpp,h,ino,json}' . | head -n 500 printf '%s\n' '== Candidate source files ==' fd -t f -e cpp -e h -e ino . | grep -Ei '(wifi|network|ethernet|wled)' | head -n 200
Length of output: 34057
🏁 Script executed:
#!/bin/bash set -euo pipefail repo='wled/WLED' refs=('16_x' 'v0.14.4') files=( 'wled00/wled.cpp' 'wled00/network.cpp' 'wled00/cfg.cpp' 'wled00/const.h' 'wled00/wled.h' 'wled00/src/dependencies/network/Network.cpp' 'wled00/src/dependencies/network/Network.h' ) tmp=$(mktemp -d) trap 'rm -rf "$tmp"' EXIT for ref in "${refs[@]}"; do safe_ref=${ref//[^A-Za-z0-9]/_} mkdir -p "$tmp/$safe_ref" printf '\n== Resolve %s ==\n' "$ref" gh api "repos/$repo/commits/$ref" --jq '.sha' for file in "${files[@]}"; do out="$tmp/$safe_ref/${file//\//__}" if gh api -H 'Accept: application/vnd.github.raw+json' \ "repos/$repo/contents/$file?ref=$ref" >"$out" 2>/dev/null; then printf 'Fetched %s\n' "$file" else printf 'Not present at %s: %s\n' "$ref" "$file" fi done done printf '\n== Definitions and decision paths: 16_x ==\n' rg -n -i -C 8 'AP_BEHAVIOR|apBehavior|initAP\s*\(|initInterfaces\s*\(|findWiFi\s*\(|isConnected\s*\(|isEthernet\s*\(|ETH_GOT_IP|ETH_DISCONNECTED' "$tmp/16_x" || true printf '\n== Definitions and decision paths: v0.14.4 ==\n' rg -n -i -C 8 'AP_BEHAVIOR|apBehavior|initAP\s*\(|initInterfaces\s*\(|findWiFi\s*\(|isConnected\s*\(|isEthernet\s*\(|ETH_GOT_IP|ETH_DISCONNECTED' "$tmp/v0_14_4" || true printf '\n== Focused source diff ==\n' diff -u \ "$tmp/v0_14_4/wled00__wled.cpp" \ "$tmp/16_x/wled00__wled.cpp" \ | grep -E -C 12 'AP_BEHAVIOR|apBehavior|initAP|findWiFi|Ethernet|ETH|^[+-]{1}[^+-]' \ | head -n 800 || true diff -u \ "$tmp/v0_14_4/wled00__network.cpp" \ "$tmp/16_x/wled00__network.cpp" \ | grep -E -C 12 'AP_BEHAVIOR|apBehavior|initAP|findWiFi|Ethernet|ETH|^[+-]{1}[^+-]' \ | head -n 800 || true
Length of output: 116250
🏁 Script executed:
#!/bin/bash set -euo pipefail repo='wled/WLED' refs=('16_x' 'v0.14.4') files=( 'wled00/data/settings_wifi.htm' 'wled00/data/settings_wifi.js' 'wled00/data/settings_wifi.htm.gz' 'wled00/wled.cpp' 'wled00/network.cpp' 'wled00/cfg.cpp' 'wled00/src/dependencies/network/Network.cpp' ) tmp=$(mktemp -d) trap 'rm -rf "$tmp"' EXIT for ref in "${refs[@]}"; do d="$tmp/${ref//[^A-Za-z0-9]/_}" mkdir -p "$d" for file in "${files[@]}"; do gh api -H 'Accept: application/vnd.github.raw+json' \ "repos/$repo/contents/$file?ref=$ref" >"$d/${file//\//__}" 2>/dev/null || true done printf '\n== %s: UI mapping and AP options ==\n' "$ref" rg -n -i -C 4 'Never|button.only|AP_BEHAVIOR|behav|access point|AP mode' "$d" || true printf '\n== %s: initAP/initConnection/setup sequence ==\n' "$ref" awk ' /void WLED::initAP\(/,/^}/ || /void WLED::initConnection\(/,/^}/ || /void WLED::setup\(/,/^void WLED::beginStrip\(/ ' "$d/wled00__wled.cpp" | sed -n '1,520p' printf '\n== %s: Ethernet events and connection predicate ==\n' "$ref" awk ' /void WiFiEvent\(/,/^}/ ' "$d/wled00__network.cpp" | sed -n '1,260p' sed -n '70,95p' "$d/wled00__src__dependencies__network__Network.cpp" done
Length of output: 14458
@softhack007, I compared16_x(7230b2c) withv0.14.4(2f6f0d2).“Never” maps to
apBehavior == 3, which isAP_BEHAVIOR_BUTTON_ONLY.When Ethernet has link and an IP address
In both versions:
- WLED treats Ethernet as a network connection.
- The disconnected path does not run.
- The AP does not start.
- If code calls
initAP()for the normal startup or reconnect flow,initAP()returns immediately forAP_BEHAVIOR_BUTTON_ONLY. - The Ethernet-connected event calls
WiFi.disconnect(true)when the AP is not active.
Therefore, with AP opens: Never and a working Ethernet connection, WLED operates through Ethernet and disables the Wi-Fi radio. ESP-NOW needs the Wi-Fi radio. Thus, ESP-NOW remotes are not usable in this intended Ethernet-connected state.
This behavior is not new in
16_x.v0.14.4has the same AP guard and the sameWiFi.disconnect(true)call in the Ethernet-connected event.Difference from
v0.14.4The relevant difference is Ethernet link-loss detection:
16_xreports Ethernet connected only ifETH.localIP()is non-zero andETH.linkUp()is true.v0.14.4reports Ethernet connected whenETH.localIP()is non-zero. It does not checkETH.linkUp().
As a result, after an Ethernet cable disconnects,
16_xcan enter the disconnected/reconnect path as soon as the link is down.v0.14.4can still consider Ethernet connected while the interface retains its old IP address.With AP opens: Never, this difference still does not enable the AP. The
initAP()guard prevents it. It can affect Wi-Fi reconnection and the general disconnected state after Ethernet link loss.Relevant
16_xlocations:settings_wifi.htm— “Never” is value3wled.cpp—initAP()rejects the normal call for behavior3network.cpp— Ethernet connection disables Wi-Fi when no AP is activeNetwork.cpp— Ethernet now requires both IP address and link state
You are interacting with an AI system.
@coderabbitai please compare the WiFi and ETH implementations between arduino-esp32 v1.0.6 (old release) and arduino-esp32 v2.0.17 (release 16.0). Could it be that differences in the framework's ETH or Network implementation can lead to "Never" not working as expected when ethernet is connected? Please look into static and dynamic(DHCP) IP adress assignments.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash set -euo pipefail framework_repo='espressif/arduino-esp32' wled_repo='wled/WLED' framework_refs=('1.0.6' '2.0.17') framework_files=( 'libraries/WiFi/src/WiFiSTA.cpp' 'libraries/WiFi/src/WiFiGeneric.cpp' 'libraries/WiFi/src/WiFiSTA.h' 'libraries/ETH/src/ETH.cpp' 'libraries/ETH/src/ETH.h' ) wled_files=( 'wled00/network.cpp' 'wled00/wled.cpp' 'wled00/cfg.cpp' 'wled00/wled.h' 'wled00/src/dependencies/network/Network.cpp' 'wled00/src/dependencies/network/Network.h' ) tmp="$(mktemp -d)" trap 'rm -rf "$tmp"' EXIT fetch_raw() { local repo="$1" ref="$2" file="$3" out="$4" gh api -H 'Accept: application/vnd.github.raw+json' \ "repos/${repo}/contents/${file}?ref=${ref}" >"$out" 2>/dev/null } printf '%s\n' '== Framework tag SHAs ==' for ref in "${framework_refs[@]}"; do printf '%s: ' "$ref" gh api "repos/${framework_repo}/commits/${ref}" --jq '.sha' done for ref in "${framework_refs[@]}"; do dir="$tmp/framework-$ref" mkdir -p "$dir" for file in "${framework_files[@]}"; do out="$dir/${file//\//__}" if fetch_raw "$framework_repo" "$ref" "$file" "$out"; then printf 'Fetched %s @ %s\n' "$file" "$ref" else printf 'Missing %s @ %s\n' "$file" "$ref" fi done done dir="$tmp/wled-16_x" mkdir -p "$dir" for file in "${wled_files[@]}"; do out="$dir/${file//\//__}" fetch_raw "$wled_repo" '16_x' "$file" "$out" || true done printf '%s\n' $'\n== WLED 16_x: configuration and state call sites ==' rg -n -C 8 \ '\b(ETH\.config|WiFi\.config|WiFi\.disconnect|ETH\.localIP|ETH\.linkUp|ETH_CONNECTED|ARDUINO_EVENT_ETH|SYSTEM_EVENT_ETH|apBehavior|AP_BEHAVIOR_BUTTON_ONLY)\b' \ "$dir" || true for ref in "${framework_refs[@]}"; do dir="$tmp/framework-$ref" printf '%s\n' $"\n== arduino-esp32 $ref: public APIs and implementations ==" rg -n -C 10 \ '\b(config|disconnect|begin|localIP|linkUp|setHostname)\s*\(|(ETH_CONNECTED|ETH_GOT_IP|ETH_DISCONNECTED|ARDUINO_EVENT_ETH|SYSTEM_EVENT_ETH)|tcpip_adapter|esp_netif' \ "$dir" || true done printf '%s\n' $'\n== Focused framework diffs ==' for stem in 'libraries__WiFi__src__WiFiSTA.cpp' 'libraries__WiFi__src__WiFiGeneric.cpp' 'libraries__ETH__src__ETH.cpp'; do old="$tmp/framework-1.0.6/$stem" new="$tmp/framework-2.0.17/$stem" if [[ -f "$old" && -f "$new" ]]; then printf '\n-- %s --\n' "$stem" diff -u "$old" "$new" | \ grep -E -C 12 '^[+-].*(config|disconnect|tcpip_adapter|esp_netif|ETH|WiFi)|^@@' | \ head -n 1200 || true fi done
Length of output: 143914
🏁 Script executed:
#!/bin/bash set -euo pipefail framework_repo='espressif/arduino-esp32' tmp="$(mktemp -d)" trap 'rm -rf "$tmp"' EXIT fetch() { local ref="$1" path="$2" gh api -H 'Accept: application/vnd.github.raw+json' \ "repos/${framework_repo}/contents/${path}?ref=${ref}" >"$tmp/${ref//./_}-$(basename "$path")" 2>/dev/null } printf '%s\n' '== Locate Ethernet source paths in each framework tag ==' for ref in 1.0.6 2.0.17; do printf '\n-- %s --\n' "$ref" gh api "repos/${framework_repo}/git/trees/${ref}?recursive=1" --paginate --jq \ '.tree[].path | select(test("(^|/)(ETH|Network)\\.(cpp|h)$|libraries/(ETH|Network)/"))' \ | head -n 100 done for ref in 1.0.6 2.0.17; do for path in \ 'libraries/ETH/src/ETH.cpp' \ 'libraries/ETH/src/ETH.h' \ 'libraries/Network/src/NetworkInterface.cpp' \ 'libraries/Network/src/NetworkInterface.h'; do fetch "$ref" "$path" || true done done printf '%s\n' $'\n== Framework: Ethernet method implementations ==' rg -n -C 14 \ '\b(ETHClass::(begin|config|localIP|linkUp|end)|NetworkInterface::(config|localIP|linkUp)|set_esp_interface_ip|esp_netif_dhcpc_|tcpip_adapter_dhcpc_|tcpip_adapter_set_ip_info)\b' \ "$tmp" || true printf '%s\n' $'\n== WLED 16_x: actual Ethernet configuration and event handling ==' rg -n -C 16 \ '\b(ETH\.begin|ETH\.config|ETH\.localIP|ETH\.linkUp|WiFi\.disconnect|ETH_GOT_IP|ETH_CONNECTED|ETH_DISCONNECTED|ARDUINO_EVENT_ETH|SYSTEM_EVENT_ETH|apBehavior|AP_BEHAVIOR_BUTTON_ONLY)\b' \ wled00 platformio.ini 2>/dev/null || true printf '%s\n' $'\n== WLED 16_x: PlatformIO framework pins ==' rg -n -C 3 \ '(platform_packages|framework-arduinoespressif32|arduino-esp32|espressif32)' \ platformio.ini platforms.ini 2>/dev/null || true
Length of output: 56856
🏁 Script executed:
#!/bin/bash set -euo pipefail repo='espressif/arduino-esp32' tmp="$(mktemp -d)" trap 'rm -rf "$tmp"' EXIT declare -A paths=( ['1.0.6']='libraries/WiFi/src/ETH.cpp libraries/WiFi/src/ETH.h libraries/WiFi/src/WiFiGeneric.cpp' ['2.0.17']='libraries/Ethernet/src/ETH.cpp libraries/Ethernet/src/ETH.h libraries/WiFi/src/WiFiGeneric.cpp' ) for ref in 1.0.6 2.0.17; do mkdir -p "$tmp/$ref" for path in ${paths[$ref]}; do outfile="$tmp/$ref/${path//\//__}" gh api -H 'Accept: application/vnd.github.raw+json' \ "repos/$repo/contents/$path?ref=$ref" >"$outfile" done done printf '%s\n' '== Public Ethernet API signatures ==' for ref in 1.0.6 2.0.17; do printf '\n-- %s --\n' "$ref" rg -n -C 4 '\b(bool|IPAddress|uint8_t)\s+(begin|config|linkUp|localIP|setHostname)\s*\(' \ "$tmp/$ref"/*ETH.h done printf '%s\n' $'\n== Ethernet implementation: begin, config, link, and address access ==' for ref in 1.0.6 2.0.17; do printf '\n-- %s --\n' "$ref" rg -n -C 18 \ '\bETHClass::(begin|config|linkUp|localIP|setHostname|end)\s*\(|\b(tcpip_adapter|esp_netif|esp_eth|SYSTEM_EVENT_ETH|ARDUINO_EVENT_ETH|IP_EVENT_ETH_GOT_IP)\b' \ "$tmp/$ref"/*ETH.cpp done printf '%s\n' $'\n== Framework Ethernet event translation and dispatch ordering ==' for ref in 1.0.6 2.0.17; do printf '\n-- %s --\n' "$ref" rg -n -C 12 \ '(SYSTEM_EVENT_ETH_CONNECTED|SYSTEM_EVENT_ETH_GOT_IP|ETHERNET_EVENT_CONNECTED|IP_EVENT_ETH_GOT_IP|ARDUINO_EVENT_ETH_CONNECTED|ARDUINO_EVENT_ETH_GOT_IP)' \ "$tmp/$ref"/*WiFiGeneric.cpp done
Length of output: 24309
@softhack007, yes. There is a concrete DHCP compatibility problem between these framework versions.Main result
WLED
16_xselects Ethernet DHCP with:ETH.config(INADDR_NONE, INADDR_NONE, INADDR_NONE);
This works in Arduino-ESP32
v1.0.6.It does not select DHCP in Arduino-ESP32
v2.0.17.INADDR_NONEis255.255.255.255. Inv2.0.17,ETHClass::config()tests only whetherlocal_ipis non-zero. It therefore treatsINADDR_NONEas a static address.Framework ETH.config(INADDR_NONE, ...)resultArduino-ESP32 v1.0.6DHCP. The code recognizes INADDR_NONEas the non-static case and starts the DHCP client.Arduino-ESP32 v2.0.17Static IPv4 configuration with address 255.255.255.255. The code does not restart the DHCP client.This difference can break Ethernet DHCP in the
v2.0.17build.Effect on “Never”
This does not change the
AP_BEHAVIOR_BUTTON_ONLYguard. “Never” still prevents normal AP startup.However, WLED disables Wi-Fi at Ethernet link-up, not when Ethernet obtains a valid IPv4 address:
case ARDUINO_EVENT_ETH_CONNECTED: if (!apActive) { WiFi.disconnect(true); }
The framework emits
ETH_CONNECTEDbeforeETH_GOT_IPin both versions.Therefore, this sequence is possible with DHCP on
v2.0.17:- Ethernet link becomes active.
- WLED receives
ARDUINO_EVENT_ETH_CONNECTED. - WLED calls
WiFi.disconnect(true). - The Wi-Fi radio turns off.
- The WLED DHCP call has configured
255.255.255.255instead of enabling DHCP. - Ethernet does not receive a valid DHCP lease.
- No Wi-Fi AP becomes available because “Never” suppresses it.
There is also an incorrect connection-state risk.
WLEDNetwork::isEthernet()checks:return (ETH.localIP()[0] != 0) && ETH.linkUp();
255.255.255.255is non-zero. If the Ethernet link stays up, WLED can treat this invalid address as an Ethernet connection. This can suppress normal disconnected/reconnect handling.Static IPv4 configuration
Static Ethernet configuration does not have this
INADDR_NONEproblem.WLED uses static Ethernet only when both
multiWiFi[0].staticIPandmultiWiFi[0].staticGWare non-zero:ETH.config( multiWiFi[0].staticIP, multiWiFi[0].staticGW, multiWiFi[0].staticSN, dnsAddress );
Both framework versions stop DHCP and install the specified IPv4 configuration for this case. The framework migration from
tcpip_adaptertoesp_netifdoes not materially change this WLED static-IP path.Framework changes
Arduino-ESP32
v2.0.17adds anesp_netifEthernet interface duringETH.begin(). It still implements the examinedETH.config()path throughtcpip_adaptercompatibility APIs.The important behavior change is the DHCP sentinel check:
v1.0.6:local_ip != 0 && local_ip != INADDR_NONEv2.0.17:local_ip != 0
Thus, the problem is not general Ethernet event handling. It is the meaning of WLED’s DHCP configuration call in
v2.0.17.Recommended correction
For Arduino-ESP32
v2.0.17, WLED should request DHCP with zero addresses:ETH.config(IPAddress((uint32_t)0), IPAddress((uint32_t)0), IPAddress((uint32_t)0));
This matches WLED’s Wi-Fi DHCP path and makes
v2.0.17restart the Ethernet DHCP client.A second hardening change would validate that
ETH.localIP()is a usable unicast address beforeWLEDNetwork::isEthernet()reports connected. This would prevent255.255.255.255from being accepted as a valid Ethernet connection.Relevant sources:
- WLED
16_x: Ethernet static/DHCP configuration - WLED
16_x: Wi-Fi disabled on Ethernet link-up - WLED
16_x: Ethernet connection predicate - Arduino-ESP32
v1.0.6:ETHClass::config() - Arduino-ESP32
v2.0.17:ETHClass::config() - Arduino-ESP32
v2.0.17: Ethernet link and got-IP event mapping
You are interacting with an AI system.
Reacted by Frank MöhleFor Arduino-ESP32 v2.0.17, WLED should request DHCP with zero addresses:
ETH.config(IPAddress((uint32_t)0), IPAddress((uint32_t)0), IPAddress((uint32_t)0));@coderabbitai please review the arduino-esp32 migration guides for version 2.0.x. is the DHCP suspicion supported by espressif documentation?
🧩 Analysis chain
🤖 web_search executed:
Search Espressif Arduino-ESP32 official documentation and GitHub repository documentation for Arduino-ESP32 2.0 migration guides. Determine whether the guides mention EthernetETH.config(), DHCP configuration,IPAddress(0U),INADDR_NONE, or a behavior change from Arduino-ESP32 1.0.x. Also find the official API documentation or examples forETH.config()in Arduino-ESP32 2.0.x.Length of output: 16994
🤖 web_search executed:
`site:docs.espressif.com/projects/arduino-esp32/en/2.0.17 "Migration from 1.x to 2.0" OR "Migration from 1.x to 2.x" Arduino ESP32
site:docs.espressif.com/projects/arduino-esp32/en/2.0.17 ETH.config DHCP INADDR_NONE IPAddress
site:github.com/espressif/arduino-esp32 2.0.17 "INADDR_NONE" "ETHClass::config"
site:github.com/espressif/arduino-esp32 2.0.17 "Migration" "DHCP" Ethernet`
Length of output: 17639
@softhack007, no. The Arduino-ESP32 2.0 migration documentation does not explicitly document this DHCP behavior change.I did not find a 1.x-to-2.0 migration-guide statement that says any of the following:
ETH.config(INADDR_NONE, ...)is unsupported.INADDR_NONEchanged meaning inETH.config().IPAddress(0U)is the required DHCP request value.- Ethernet DHCP configuration changed between Arduino-ESP32 1.x and 2.0.x.
The official Ethernet guide does support one related point. It treats link-up and IP acquisition as separate states:
ARDUINO_EVENT_ETH_CONNECTEDmeans the Ethernet link is active.ARDUINO_EVENT_ETH_GOT_IPmeans Ethernet has an IP address.
The guide also states that the hostname must be set after Ethernet starts and before DHCP. This supports the conclusion that a link-up event does not prove a usable DHCP address.
The strongest support for the
INADDR_NONEconcern is source history, not the migration guide. Espressif previously fixed the exact defect in an olderETHClass::config()implementation:“Fix ETH not enabling DHCP when configured with INADDR_NONE”
That fix added an explicit
local_ip != INADDR_NONEtest before selecting static configuration. The associated Espressif issue describes the failure mode as static configuration with255.255.255.255and DHCP disabled.- Espressif fix for
INADDR_NONE - Espressif issue describing the
255.255.255.255result - Official Ethernet event documentation
Therefore, I should qualify the earlier conclusion:
- The claim is not supported by a documented 2.0 migration requirement.
- The claim is supported by the 1.0.6 and 2.0.17 implementation comparison.
- The old Espressif fix proves that
INADDR_NONEhandling inETH.config()has been a known compatibility-sensitive behavior. ETH.config(IPAddress(0U), ...)is a source-compatible mitigation for the examined 2.0.17 implementation. Espressif migration documentation does not prescribe it.
A hardware test is still necessary. Test DHCP on the exact release framework package, with both
INADDR_NONEand zero-address calls. LogARDUINO_EVENT_ETH_GOT_IP,ETH.localIP(), gateway, subnet, and DNS values.You are interacting with an AI system.
Btw, this issue has been addresses here in past: tartinesKiller@47943c4
@coderabbitai please analyse the commit referenced by the OP. Can this change solve the DHCP connection state problem? If not, what does it achieve?
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash set -euo pipefail repo='tartinesKiller/WLED' sha='47943c4107b09ad566f2f71166c17bed890c0fd6' tmp="$(mktemp -d)" trap 'rm -rf "$tmp"' EXIT echo '== Commit metadata and changed files ==' gh api "repos/$repo/commits/$sha" \ --jq '{sha, parents: [.parents[].sha], message: .commit.message, files: [.files[] | {filename, status, additions, deletions, patch}]}' \ >"$tmp/commit.json" jq '{sha, parents, message, files: [.files[] | del(.patch)]}' "$tmp/commit.json" jq -r '.files[] | "\n== \(.filename) ==\n\(.patch // "(No inline patch returned)")"' "$tmp/commit.json" parent="$(jq -r '.parents[0]' "$tmp/commit.json")" echo echo "== Parent: $parent ==" echo '== Relevant source before and after ==' jq -r '.files[].filename' "$tmp/commit.json" | while IFS= read -r path; do echo echo "--- $path @ parent ---" gh api -H 'Accept: application/vnd.github.raw+json' \ "repos/$repo/contents/$path?ref=$parent" 2>/dev/null | \ rg -n -C 20 'ETH\.config|WiFi\.config|INADDR_NONE|staticIP|staticGW|ETH_GOT_IP|ETH_CONNECTED|WiFi\.disconnect|localIP|linkUp' || true echo "--- $path @ commit ---" gh api -H 'Accept: application/vnd.github.raw+json' \ "repos/$repo/contents/$path?ref=$sha" 2>/dev/null | \ rg -n -C 20 'ETH\.config|WiFi\.config|INADDR_NONE|staticIP|staticGW|ETH_GOT_IP|ETH_CONNECTED|WiFi\.disconnect|localIP|linkUp' || true done echo echo '== Corresponding paths in WLED 16_x ==' for path in wled00/network.cpp wled00/src/dependencies/network/Network.cpp wled00/wled.cpp; do echo "--- $path ---" gh api -H 'Accept: application/vnd.github.raw+json' \ "repos/wled/WLED/contents/$path?ref=16_x" 2>/dev/null | \ rg -n -C 20 'ETH\.config|WiFi\.config|INADDR_NONE|staticIP|staticGW|ETH_GOT_IP|ETH_CONNECTED|WiFi\.disconnect|localIP|linkUp' || true done
Length of output: 36995
@softhack007, no. The referenced commit cannot solve the Ethernet DHCP connection-state problem.The commit changes only
wled00/wled.cpp. It does not change:ETH.config()INADDR_NONE- Ethernet DHCP startup
ETH.localIP()ETH.linkUp()NetworkClass::isEthernet()- Ethernet events such as
ARDUINO_EVENT_ETH_CONNECTEDorARDUINO_EVENT_ETH_GOT_IP
Therefore, it cannot correct either of these DHCP-related failures:
ETH.config(INADDR_NONE, ...)configures255.255.255.255instead of enabling DHCP in a framework that does not recognizeINADDR_NONEas the DHCP sentinel.NetworkClass::isEthernet()accepts a nonzero but invalid Ethernet address while the Ethernet link remains up.
What the commit achieves
The commit changes the Wi-Fi fallback policy.
1. It turns Wi-Fi off when no client SSID exists
In
initConnection(), ifWLED_WIFI_CONFIGUREDis false andclientSSIDis empty, it now runs:WiFi.mode(WIFI_OFF); return;
Before this change, WLED started the AP immediately:
if (!apActive) initAP(); return;
This prevents WLED from automatically entering AP mode when the configuration has no Wi-Fi client SSID.
2. It stops repeated Wi-Fi reconnect attempts when no SSID exists
The commit adds this condition:
&& *clientSSID != '\0'WLED then does not call
initConnection()every reconnect interval when it has no SSID to connect to.3. It prevents the 12-second fallback from starting an AP in button-only mode
The commit adds:
apBehavior != AP_BEHAVIOR_BUTTON_ONLYThe disconnected-state fallback no longer calls
initAP()after 12 seconds when AP mode is button-only.This is the part that addresses the historical “AP opens: Never” behavior. In that version,
AP_BEHAVIOR_BUTTON_ONLYcould still reach the periodic fallback path and start an AP.Effect when Ethernet DHCP fails
If Ethernet link is up but DHCP does not obtain a usable address:
- The commit does not start Ethernet DHCP.
- The commit does not validate the Ethernet address.
- The commit does not change
Network.isConnected(). - The commit can prevent Wi-Fi and AP fallback.
Thus, the commit can make the device remain unreachable after an Ethernet DHCP failure. It enforces “do not use Wi-Fi/AP” behavior. It does not repair Ethernet connectivity.
For
16_x, the DHCP investigation must remain focused onwled00/network.cppandwled00/src/dependencies/network/Network.cpp. The referenced commit is relevant to AP fallback behavior, but not to the Ethernet DHCP configuration or Ethernet connection-state defect.You are interacting with an AI system.
the commit can make the device remain unreachable after an Ethernet DHCP failure. It enforces “do not use Wi-Fi/AP” behavior.
Sounds dangerous 🤔 it basically makes the board unreachable if ethernet startup failed.
the commit can make the device remain unreachable after an Ethernet DHCP failure. It enforces “do not use Wi-Fi/AP” behavior.
Sounds dangerous 🤔 it basically makes the board unreachable if ethernet startup failed.
Well, if there is a DHCP failure there must be an error with the DHCP server or any other network issue. This can be easily fixed by plugging it to any other network with working DHCP. So if a user chooses the option AP opens: 'Never' I think this is an expected behaviour.
What happened?
I am using a WLED controller with ethernet port. Therefore I would like to deactivate the WLED AP completely as soon as the controller is running over LAN.
I tried this approach however the Wifi AP is still active: https://www.reddit.com/r/WLED/comments/vmxchv/comment/ie3qxg3/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
To Reproduce Bug
Not setting a network name and setting AP opens to "Never" in Wifi settings
Expected Behavior
Wifi AP switched off permanently
Install Method
Binary from WLED.me
What version of WLED?
16.0
Which microcontroller/board are you seeing the problem on?
ESP32 with ethernet
Relevant log/trace output
Anything else?
No response
Code of Conduct