FIX: Trigger OTA from a TTN Downlink.

.github/workflows/ci-cd.yml

@@ -73,13 +73,13 @@ jobs:
 
           echo "Flash used: $FLASH_USED bytes"
 
-          # Check if flash usage exceeds 1,300,000 bytes
-          if [ "$FLASH_USED" -gt 1300000 ]; then
-            echo "❌ ERROR: Flash usage ($FLASH_USED bytes) exceeds limit of 1,300,000 bytes"
-            echo "Flash usage: $FLASH_USED / 1,300,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1300000" | bc)%)"
+          # Check if flash usage exceeds 1,800,000 bytes (90% of 1.875MB app partition)
+          if [ "$FLASH_USED" -gt 1800000 ]; then
+            echo "❌ ERROR: Flash usage ($FLASH_USED bytes) exceeds limit of 1,800,000 bytes"
+            echo "Flash usage: $FLASH_USED / 1,800,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1800000" | bc)%)"
             exit 1
           else
-            echo "✅ Flash usage OK: $FLASH_USED / 1,300,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1300000" | bc)%)"
+            echo "✅ Flash usage OK: $FLASH_USED / 1,800,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1800000" | bc)%)"
           fi
 
       - name: Store map files as artifacts
@@ -88,9 +88,7 @@ jobs:
           name: build-artifacts-branch-${{ github.sha }}
           path: |
             .pio/build/ttgo-lora32-v1/firmware.map
-            .pio/build/ttgo-lora32-v1/firmware.elf
             .pio/build/minimal/firmware.map
-            .pio/build/minimal/firmware.elf
           retention-days: 30
         if: always()
 
@@ -163,13 +161,13 @@ jobs:
 
           echo "Flash used: $FLASH_USED bytes"
 
-          # Check if flash usage exceeds 1,300,000 bytes
-          if [ "$FLASH_USED" -gt 1300000 ]; then
-            echo "❌ ERROR: Flash usage ($FLASH_USED bytes) exceeds limit of 1,300,000 bytes"
-            echo "Flash usage: $FLASH_USED / 1,300,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1300000" | bc)%)"
+          # Check if flash usage exceeds 1,800,000 bytes (90% of 1.875MB app partition)
+          if [ "$FLASH_USED" -gt 1800000 ]; then
+            echo "❌ ERROR: Flash usage ($FLASH_USED bytes) exceeds limit of 1,800,000 bytes"
+            echo "Flash usage: $FLASH_USED / 1,800,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1800000" | bc)%)"
             exit 1
           else
-            echo "✅ Flash usage OK: $FLASH_USED / 1,300,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1300000" | bc)%)"
+            echo "✅ Flash usage OK: $FLASH_USED / 1,800,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1800000" | bc)%)"
           fi
 
       - name: Store map files as artifacts
@@ -178,9 +176,7 @@ jobs:
           name: build-artifacts-main-${{ github.sha }}
           path: |
             .pio/build/ttgo-lora32-v1/firmware.map
-            .pio/build/ttgo-lora32-v1/firmware.elf
             .pio/build/minimal/firmware.map
-            .pio/build/minimal/firmware.elf
           retention-days: 30
         if: always()
 
@@ -286,20 +282,19 @@ jobs:
 
           echo "Flash used: $FLASH_USED bytes"
 
-          # Check if flash usage exceeds 1,300,000 bytes
-          if [ "$FLASH_USED" -gt 1300000 ]; then
-            echo "❌ ERROR: Flash usage ($FLASH_USED bytes) exceeds limit of 1,300,000 bytes"
-            echo "Flash usage: $FLASH_USED / 1,300,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1300000" | bc)%)"
+          # Check if flash usage exceeds 1,800,000 bytes (90% of 1.875MB app partition)
+          if [ "$FLASH_USED" -gt 1800000 ]; then
+            echo "❌ ERROR: Flash usage ($FLASH_USED bytes) exceeds limit of 1,800,000 bytes"
+            echo "Flash usage: $FLASH_USED / 1,800,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1800000" | bc)%)"
             exit 1
           else
-            echo "✅ Flash usage OK: $FLASH_USED / 1,300,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1300000" | bc)%)"
+            echo "✅ Flash usage OK: $FLASH_USED / 1,800,000 bytes ($(echo "scale=2; $FLASH_USED * 100 / 1800000" | bc)%)"
           fi
 
       - name: Create release artifacts
         run: |
           mkdir -p artifacts
           cp .pio/build/ttgo-lora32-v1/firmware.bin artifacts/firmware-${{ steps.validate-tag.outputs.tag_name }}.bin
-          cp .pio/build/ttgo-lora32-v1/firmware.elf artifacts/firmware-${{ steps.validate-tag.outputs.tag_name }}.elf
           cp .pio/build/ttgo-lora32-v1/firmware.map artifacts/firmware-${{ steps.validate-tag.outputs.tag_name }}.map
           cp .pio/build/ttgo-lora32-v1/partitions.bin artifacts/partitions-${{ steps.validate-tag.outputs.tag_name }}.bin
 
@@ -312,7 +307,7 @@ jobs:
             "branch": "${{ github.ref_name }}",
             "files": {
               "firmware": "firmware-${{ steps.validate-tag.outputs.tag_name }}.bin",
-              "elf": "firmware-${{ steps.validate-tag.outputs.tag_name }}.elf",
+              "map": "firmware-${{ steps.validate-tag.outputs.tag_name }}.map",
               "partitions": "partitions-${{ steps.validate-tag.outputs.tag_name }}.bin"
             }
           }
@@ -339,9 +334,8 @@ jobs:
             - **Commit**: ${{ github.sha }}
 
             ### Artifacts
-            - `firmware-${{ steps.validate-tag.outputs.tag_name }}.bin` - Binary firmware file
-            - `firmware-${{ steps.validate-tag.outputs.tag_name }}.elf` - ELF debug file
-            - `firmware-${{ steps.validate-tag.outputs.tag_name }}.map` - Memory map file (for debugging memory regressions)
+            - `firmware-${{ steps.validate-tag.outputs.tag_name }}.bin` - Binary firmware file for flashing
+            - `firmware-${{ steps.validate-tag.outputs.tag_name }}.map` - Memory map file (for debugging memory usage)
             - `partitions-${{ steps.validate-tag.outputs.tag_name }}.bin` - Partition table
             - `build-info-${{ steps.validate-tag.outputs.tag_name }}.json` - Build metadata
 

.github/workflows/size-baseline.yml

@@ -48,12 +48,8 @@ jobs:
           # Copy build log first (always available)
           cp ci-build-release.log ci-baseline-artifacts/
 
-          # Copy generated files if they exist
-          if [ -f ".pio/build/ttgo-lora32-v1/firmware.elf" ]; then
-            cp .pio/build/ttgo-lora32-v1/firmware.elf ci-baseline-artifacts/release-firmware.elf
-          else
-            echo "WARNING: release firmware.elf not found" >> ci-baseline-artifacts/build-warnings.txt
-          fi
+          # Copy generated files if they exist (excluding ELF files to reduce size)
+          # ELF files are too large for artifacts and not needed for size baseline
 
           if [ -f ".pio/build/ttgo-lora32-v1/firmware.map" ]; then
             cp .pio/build/ttgo-lora32-v1/firmware.map ci-baseline-artifacts/release-firmware.map
@@ -98,12 +94,8 @@ jobs:
           # Copy debug build log
           cp ci-build-debug.log ci-baseline-artifacts/
 
-          # Copy debug generated files if they exist
-          if [ -f ".pio/build/minimal/firmware.elf" ]; then
-            cp .pio/build/minimal/firmware.elf ci-baseline-artifacts/debug-firmware.elf
-          else
-            echo "WARNING: debug firmware.elf not found" >> ci-baseline-artifacts/build-warnings.txt
-          fi
+          # Copy debug generated files if they exist (excluding ELF files to reduce size)
+          # ELF files are too large for artifacts and not needed for size baseline
 
           if [ -f ".pio/build/minimal/firmware.map" ]; then
             cp .pio/build/minimal/firmware.map ci-baseline-artifacts/debug-firmware.map

Makefile

@@ -24,7 +24,6 @@ minimal:
 # Test target that calls minimal build for CI readability
 test: minimal  ## Build minimal env (alias for CI)
 
-
 .PHONY: build-cli
 build-cli:
 	cd cli && go build -o lora-sensor-cli .

README.md

@@ -76,6 +76,59 @@ This project uses a single optimized build configuration:
 
 For detailed build information, see [docs/BUILD.md](docs/BUILD.md).
 
+## Watchdog Management for OTA Updates
+
+The firmware implements comprehensive watchdog timer management and stack overflow prevention to prevent system crashes during OTA (Over-The-Air) firmware updates:
+
+### Watchdog Reset Points
+
+- **WiFi Connection Loop**: Reset every 100ms during WiFi connection attempts (8-second timeout)
+- **HTTP Download Loop**: Reset every 1000ms during firmware download (improved from 2000ms)
+- **WiFi Test Function**: Reset every 400ms during WiFi connectivity testing
+- **OTA Progress Updates**: Watchdog reset occurs with each progress update (every 5 seconds or 50KB chunks)
+
+### Stack Overflow Prevention
+
+To prevent stack canary watchpoint panics that can occur during OTA operations:
+
+- **Heap-Allocated Download Buffer**: Uses `malloc()` to allocate a 2KB download buffer on the heap instead of stack
+- **Reduced Buffer Size**: Download buffer reduced from 4KB to 2KB to minimize memory usage
+- **Proper Error Cleanup**: All error paths properly free allocated memory with appropriate cleanup
+- **Memory Safety**: Buffer is set to `nullptr` after freeing to prevent double-free issues
+
+### Timeout Constants
+
+- `WIFI_CONNECTION_TIMEOUT`: 8000ms (8 seconds) - Maximum time to wait for WiFi connection
+- `HTTP_TOTAL_TIMEOUT`: 30000ms (30 seconds) - Total HTTP request timeout for OTA downloads
+- `HTTP_CONNECT_TIMEOUT`: 15000ms (15 seconds) - HTTP connection establishment timeout
+- `CONNECTIVITY_TEST_TIMEOUT`: 10000ms (10 seconds) - Internet connectivity test timeout
+- `WATCHDOG_RESET_INTERVAL_DOWNLOAD`: 1000ms (1 second) - Watchdog reset frequency during download
+- `WATCHDOG_RESET_INTERVAL_WIFI_TEST`: 400ms - Watchdog reset frequency during WiFi testing
+- `PROGRESS_UPDATE_INTERVAL`: 5000ms (5 seconds) - Progress update and watchdog reset frequency
+- `CHUNK_SIZE_THRESHOLD`: 51200 bytes (50KB) - Alternative trigger for progress updates
+
+### HTTPUpdate Library Integration
+
+The OTA system now uses ESP32's built-in HTTPUpdate library for improved reliability:
+
+- **Automatic HTTPS Support**: Handles both HTTP and HTTPS URLs with redirect following
+- **MD5 Verification**: Uses HTTPUpdate's native MD5 verification with manual header injection
+  - GitHub doesn't provide `x-MD5` headers, so we inject the MD5 from TTN downlink via request callback
+  - HTTPUpdate then performs automatic MD5 verification during download
+- **Progress Monitoring**: Real-time download progress with watchdog management
+- **Error Handling**: Comprehensive error reporting and cleanup
+- **Memory Efficiency**: No large buffer allocation - HTTPUpdate handles streaming internally
+- **Redirect Support**: Automatic handling of HTTP redirects (strict mode enabled)
+
+### Network Diagnostics
+
+The OTA system includes comprehensive network diagnostics:
+
+- **HTTPS Support**: Automatic detection and handling of HTTPS URLs with certificate validation bypass
+- **Timeout Configuration**: Explicit HTTP timeouts prevent indefinite blocking
+- **Redirect Handling**: Automatic following of HTTP redirects using HTTPUpdate's built-in support
+- **Error Reporting**: Detailed error codes and diagnostic messages for troubleshooting
+
 ## Build Commands
 
 - `make release`: Build the main release environment (ttgo-lora32-v1)

cli/cmd/update_device.go

@@ -177,12 +177,10 @@ func updateDevice(ctx context.Context, cmd *cli.Command) error {
 	// Step 10: Publish chunks to TTN MQTT topic
 	topic := fmt.Sprintf("ttn/soil-conditions/devices/%s/down/push", deviceName)
 	fmt.Printf("Publishing to topic: %s\n", topic)
+	fmt.Printf("Publishing %d chunks as single TTN message\n", len(chunks))
 
-	for i, chunk := range chunks {
-		fmt.Printf("Publishing chunk %d/%d\n", i+1, len(chunks))
-		if err := mqttClient.Publish(topic, chunk); err != nil {
-			return fmt.Errorf("failed to publish chunk %d: %w", i+1, err)
-		}
+	if err := mqttClient.PublishChunks(topic, chunks); err != nil {
+		return fmt.Errorf("failed to publish chunks: %w", err)
 	}
 
 	fmt.Println("Firmware update payload published successfully!")

cli/pkg/mqtt/mqtt.go

@@ -22,13 +22,16 @@ type Client struct {
 	client mqtt.Client
 }
 
-// TTNMessage represents a TTN MQTT message structure
+// TTNDownlink represents a single TTN downlink message
+type TTNDownlink struct {
+	FPort      int    `json:"f_port"`
+	FRMPayload string `json:"frm_payload"`
+	Priority   string `json:"priority"`
+}
+
+// TTNMessage represents a TTN MQTT message structure with chunked downlinks
 type TTNMessage struct {
-	DownlinkMessage struct {
-		FPort      int    `json:"f_port"`
-		FRMPayload string `json:"frm_payload"`
-		Priority   string `json:"priority"`
-	} `json:"downlink_message"`
+	Downlinks []TTNDownlink `json:"downlinks"`
 }
 
 // NewClient creates a new MQTT client
@@ -67,13 +70,21 @@ func (c *Client) Disconnect() {
 	c.client.Disconnect(250)
 }
 
-// Publish publishes a chunk as a TTN MQTT message
-func (c *Client) Publish(topic string, chunk []byte) error {
-	// Create TTN message structure
-	message := TTNMessage{}
-	message.DownlinkMessage.FPort = 1 // Using fport 1 for OTA chunks
-	message.DownlinkMessage.FRMPayload = base64.StdEncoding.EncodeToString(chunk)
-	message.DownlinkMessage.Priority = "NORMAL"
+// PublishChunks publishes multiple chunks as a single TTN MQTT message with downlinks array
+func (c *Client) PublishChunks(topic string, chunks [][]byte) error {
+	// Create TTN message structure with multiple downlinks
+	message := TTNMessage{
+		Downlinks: make([]TTNDownlink, len(chunks)),
+	}
+
+	// Create downlink for each chunk with incrementing f_port
+	for i, chunk := range chunks {
+		message.Downlinks[i] = TTNDownlink{
+			FPort:      i + 1, // f_port starts at 1 and increments
+			FRMPayload: base64.StdEncoding.EncodeToString(chunk),
+			Priority:   "NORMAL",
+		}
+	}
 
 	// Marshal to JSON
 	payload, err := json.Marshal(message)
@@ -90,6 +101,11 @@ func (c *Client) Publish(topic string, chunk []byte) error {
 	return nil
 }
 
+// Publish publishes a single chunk as a TTN MQTT message (deprecated, use PublishChunks)
+func (c *Client) Publish(topic string, chunk []byte) error {
+	return c.PublishChunks(topic, [][]byte{chunk})
+}
+
 // IsConnected returns true if the client is connected
 func (c *Client) IsConnected() bool {
 	return c.client.IsConnected()