Merge remote-tracking branch 'upstream/dev' into fast-millis-esp32

# Conflicts:
#	esphome/core/application.h

esphome/components/api/__init__.py

@@ -291,12 +291,12 @@ CONFIG_SCHEMA = cv.All(
             cv.SplitDefault(
                 CONF_MAX_CONNECTIONS,
                 esp8266=4,  # ~40KB free RAM, each connection uses ~500-1000 bytes
-                esp32=8,  # 520KB RAM available
+                esp32=5,  # 520KB RAM available
                 rp2040=4,  # 264KB RAM but LWIP constraints
-                bk72xx=8,  # Moderate RAM
-                rtl87xx=8,  # Moderate RAM
+                bk72xx=5,  # Moderate RAM
+                rtl87xx=5,  # Moderate RAM
                 host=8,  # Abundant resources
-                ln882x=8,  # Moderate RAM
+                ln882x=5,  # Moderate RAM
             ): cv.int_range(min=1, max=20),
             # Maximum queued send buffers per connection before dropping connection
             # Each buffer uses ~8-12 bytes overhead plus actual message size
@@ -336,8 +336,7 @@ async def to_code(config: ConfigType) -> None:
     cg.add(var.set_batch_delay(config[CONF_BATCH_DELAY]))
     if CONF_LISTEN_BACKLOG in config:
         cg.add(var.set_listen_backlog(config[CONF_LISTEN_BACKLOG]))
-    if CONF_MAX_CONNECTIONS in config:
-        cg.add(var.set_max_connections(config[CONF_MAX_CONNECTIONS]))
+    cg.add_define("MAX_API_CONNECTIONS", config[CONF_MAX_CONNECTIONS])
     cg.add_define("API_MAX_SEND_QUEUE", config[CONF_MAX_SEND_QUEUE])
 
     # Set USE_API_USER_DEFINED_ACTIONS if any services are enabled

esphome/components/api/api_server.cpp

@@ -118,7 +118,7 @@ void APIServer::loop() {
     this->accept_new_connections_();
   }
 
-  if (this->clients_.empty()) {
+  if (this->api_connection_count_ == 0) {
     // Check reboot timeout - done in loop to avoid scheduler heap churn
     // (cancelled scheduler items sit in heap memory until their scheduled time)
     if (this->reboot_timeout_ != 0) {
@@ -135,15 +135,15 @@ void APIServer::loop() {
   // Check network connectivity once for all clients
   if (!network::is_connected()) {
     // Network is down - disconnect all clients
-    for (auto &client : this->clients_) {
+    for (auto &client : this->active_clients()) {
       client->on_fatal_error();
       client->log_client_(ESPHOME_LOG_LEVEL_WARN, LOG_STR("Network down; disconnect"));
     }
     // Continue to process and clean up the clients below
   }
 
-  size_t client_index = 0;
-  while (client_index < this->clients_.size()) {
+  uint8_t client_index = 0;
+  while (client_index < this->api_connection_count_) {
     auto &client = this->clients_[client_index];
 
     // Common case: process active client
@@ -161,7 +161,7 @@ void APIServer::loop() {
   }
 }
 
-void APIServer::remove_client_(size_t client_index) {
+void APIServer::remove_client_(uint8_t client_index) {
   auto &client = this->clients_[client_index];
 
 #ifdef USE_API_USER_DEFINED_ACTION_RESPONSES
@@ -179,14 +179,17 @@ void APIServer::remove_client_(size_t client_index) {
   // Close socket now (was deferred from on_fatal_error to allow getpeername)
   client->helper_->close();
 
-  // Swap with the last element and pop (avoids expensive vector shifts)
-  if (client_index < this->clients_.size() - 1) {
-    std::swap(this->clients_[client_index], this->clients_.back());
+  // Swap-and-reset: move the removed client to the trailing slot and null it out so slots
+  // [api_connection_count_, N) remain nullptr.
+  const uint8_t last_index = this->api_connection_count_ - 1;
+  if (client_index < last_index) {
+    std::swap(this->clients_[client_index], this->clients_[last_index]);
   }
-  this->clients_.pop_back();
+  this->clients_[last_index].reset();
+  this->api_connection_count_--;
 
   // Last client disconnected - set warning and start tracking for reboot timeout
-  if (this->clients_.empty() && this->reboot_timeout_ != 0) {
+  if (this->api_connection_count_ == 0 && this->reboot_timeout_ != 0) {
     this->status_set_warning(LOG_STR("waiting for client connection"));
     this->last_connected_ = App.get_loop_component_start_time();
   }
@@ -210,8 +213,8 @@ void __attribute__((flatten)) APIServer::accept_new_connections_() {
     sock->getpeername_to(peername);
 
     // Check if we're at the connection limit
-    if (this->clients_.size() >= this->max_connections_) {
-      ESP_LOGW(TAG, "Max connections (%d), rejecting %s", this->max_connections_, peername);
+    if (this->api_connection_count_ >= MAX_API_CONNECTIONS) {
+      ESP_LOGW(TAG, "Max connections (%d), rejecting %s", MAX_API_CONNECTIONS, peername);
       // Immediately close - socket destructor will handle cleanup
       sock.reset();
       continue;
@@ -220,11 +223,11 @@ void __attribute__((flatten)) APIServer::accept_new_connections_() {
     ESP_LOGD(TAG, "Accept %s", peername);
 
     auto *conn = new APIConnection(std::move(sock), this);
-    this->clients_.emplace_back(conn);
+    this->clients_[this->api_connection_count_++].reset(conn);
     conn->start();
 
     // First client connected - clear warning and update timestamp
-    if (this->clients_.size() == 1 && this->reboot_timeout_ != 0) {
+    if (this->api_connection_count_ == 1 && this->reboot_timeout_ != 0) {
       this->status_clear_warning();
       this->last_connected_ = App.get_loop_component_start_time();
     }
@@ -237,7 +240,7 @@ void APIServer::dump_config() {
                 "  Address: %s:%u\n"
                 "  Listen backlog: %u\n"
                 "  Max connections: %u",
-                network::get_use_address(), this->port_, this->listen_backlog_, this->max_connections_);
+                network::get_use_address(), this->port_, this->listen_backlog_, MAX_API_CONNECTIONS);
 #ifdef USE_API_NOISE
   ESP_LOGCONFIG(TAG, "  Noise encryption: %s", YESNO(this->noise_ctx_.has_psk()));
   if (!this->noise_ctx_.has_psk()) {
@@ -255,7 +258,7 @@ void APIServer::handle_disconnect(APIConnection *conn) {}
   void APIServer::on_##entity_name##_update(entity_type *obj) { /* NOLINT(bugprone-macro-parentheses) */ \
     if (obj->is_internal()) \
       return; \
-    for (auto &c : this->clients_) { \
+    for (auto &c : this->active_clients()) { \
       if (c->flags_.state_subscription) \
         c->send_##entity_name##_state(obj); \
     } \
@@ -337,7 +340,7 @@ API_DISPATCH_UPDATE(water_heater::WaterHeater, water_heater)
 void APIServer::on_event(event::Event *obj) {
   if (obj->is_internal())
     return;
-  for (auto &c : this->clients_) {
+  for (auto &c : this->active_clients()) {
     if (c->flags_.state_subscription)
       c->send_event(obj);
   }
@@ -349,7 +352,7 @@ void APIServer::on_event(event::Event *obj) {
 void APIServer::on_update(update::UpdateEntity *obj) {
   if (obj->is_internal())
     return;
-  for (auto &c : this->clients_) {
+  for (auto &c : this->active_clients()) {
     if (c->flags_.state_subscription)
       c->send_update_state(obj);
   }
@@ -360,7 +363,7 @@ void APIServer::on_update(update::UpdateEntity *obj) {
 void APIServer::on_zwave_proxy_request(const ZWaveProxyRequest &msg) {
   // We could add code to manage a second subscription type, but, since this message type is
   //  very infrequent and small, we simply send it to all clients
-  for (auto &c : this->clients_)
+  for (auto &c : this->active_clients())
     c->send_message(msg);
 }
 #endif
@@ -375,7 +378,7 @@ void APIServer::send_infrared_rf_receive_event([[maybe_unused]] uint32_t device_
   resp.key = key;
   resp.timings = timings;
 
-  for (auto &c : this->clients_)
+  for (auto &c : this->active_clients())
     c->send_infrared_rf_receive_event(resp);
 }
 #endif
@@ -392,7 +395,7 @@ void APIServer::set_batch_delay(uint16_t batch_delay) { this->batch_delay_ = bat
 
 #ifdef USE_API_HOMEASSISTANT_SERVICES
 void APIServer::send_homeassistant_action(const HomeassistantActionRequest &call) {
-  for (auto &client : this->clients_) {
+  for (auto &client : this->active_clients()) {
     client->send_homeassistant_action(call);
   }
 }
@@ -532,7 +535,7 @@ bool APIServer::update_noise_psk_(const SavedNoisePsk &new_psk, const LogString
         return;
       }
       ESP_LOGW(TAG, "Disconnecting all clients to reset PSK");
-      for (auto &c : this->clients_) {
+      for (auto &c : this->active_clients()) {
         DisconnectRequest req;
         c->send_message(req);
       }
@@ -583,7 +586,7 @@ bool APIServer::clear_noise_psk(bool make_active) {
 
 #ifdef USE_HOMEASSISTANT_TIME
 void APIServer::request_time() {
-  for (auto &client : this->clients_) {
+  for (auto &client : this->active_clients()) {
     if (!client->flags_.remove && client->is_authenticated()) {
       client->send_time_request();
       return;  // Only request from one client to avoid clock conflicts
@@ -593,8 +596,8 @@ void APIServer::request_time() {
 #endif
 
 bool APIServer::is_connected_with_state_subscription() const {
-  for (const auto &client : this->clients_) {
-    if (client->flags_.state_subscription) {
+  for (uint8_t i = 0; i < this->api_connection_count_; i++) {
+    if (this->clients_[i]->flags_.state_subscription) {
       return true;
     }
   }
@@ -609,7 +612,7 @@ void APIServer::on_log(uint8_t level, const char *tag, const char *message, size
     // we would be filling a buffer we are trying to clear
     return;
   }
-  for (auto &c : this->clients_) {
+  for (auto &c : this->active_clients()) {
     if (!c->flags_.remove && c->get_log_subscription_level() >= level)
       c->try_send_log_message(level, tag, message, message_len);
   }
@@ -618,7 +621,7 @@ void APIServer::on_log(uint8_t level, const char *tag, const char *message, size
 
 #ifdef USE_CAMERA
 void APIServer::on_camera_image(const std::shared_ptr<camera::CameraImage> &image) {
-  for (auto &c : this->clients_) {
+  for (auto &c : this->active_clients()) {
     if (!c->flags_.remove)
       c->set_camera_state(image);
   }
@@ -635,7 +638,7 @@ void APIServer::on_shutdown() {
   this->batch_delay_ = 5;
 
   // Send disconnect requests to all connected clients
-  for (auto &c : this->clients_) {
+  for (auto &c : this->active_clients()) {
     DisconnectRequest req;
     if (!c->send_message(req)) {
       // If we can't send the disconnect request directly (tx_buffer full),
@@ -653,7 +656,7 @@ bool APIServer::teardown() {
   this->loop();
 
   // Return true only when all clients have been torn down
-  return this->clients_.empty();
+  return this->api_connection_count_ == 0;
 }
 
 #ifdef USE_API_USER_DEFINED_ACTION_RESPONSES

esphome/components/api/api_server.h

@@ -21,6 +21,8 @@
 #include "esphome/components/camera/camera.h"
 #endif
 
+#include <array>
+#include <memory>
 #include <vector>
 
 namespace esphome::api {
@@ -63,7 +65,6 @@ class APIServer final : public Component,
   void set_batch_delay(uint16_t batch_delay);
   uint16_t get_batch_delay() const { return batch_delay_; }
   void set_listen_backlog(uint8_t listen_backlog) { this->listen_backlog_ = listen_backlog; }
-  void set_max_connections(uint8_t max_connections) { this->max_connections_ = max_connections; }
 
   // Get reference to shared buffer for API connections
   APIBuffer &get_shared_buffer_ref() { return shared_write_buffer_; }
@@ -186,9 +187,26 @@ class APIServer final : public Component,
   void send_infrared_rf_receive_event(uint32_t device_id, uint32_t key, const std::vector<int32_t> *timings);
 #endif
 
-  bool is_connected() const { return !this->clients_.empty(); }
+  bool is_connected() const { return this->api_connection_count_ != 0; }
   bool is_connected_with_state_subscription() const;
 
+  // Range-for view over the populated slice [0, api_connection_count_). Read-only with respect
+  // to ownership — callers get `const unique_ptr&` so they can invoke non-const methods on the
+  // APIConnection but cannot reset/move the slot and break the count invariant.
+  using APIConnectionPtr = std::unique_ptr<APIConnection>;
+  class ActiveClientsView {
+    const APIConnectionPtr *begin_;
+    const APIConnectionPtr *end_;
+
+   public:
+    ActiveClientsView(const APIConnectionPtr *b, const APIConnectionPtr *e) : begin_(b), end_(e) {}
+    const APIConnectionPtr *begin() const { return this->begin_; }
+    const APIConnectionPtr *end() const { return this->end_; }
+  };
+  ActiveClientsView active_clients() const {
+    return {this->clients_.data(), this->clients_.data() + this->api_connection_count_};
+  }
+
 #ifdef USE_API_HOMEASSISTANT_STATES
   struct HomeAssistantStateSubscription {
     const char *entity_id;  // Pointer to flash (internal) or heap (external)
@@ -234,8 +252,8 @@ class APIServer final : public Component,
  protected:
   // Accept incoming socket connections. Only called when socket has pending connections.
   void __attribute__((noinline)) accept_new_connections_();
-  // Remove a disconnected client by index. Swaps with last element and pops.
-  void __attribute__((noinline)) remove_client_(size_t client_index);
+  // Remove a disconnected client by index. Swaps with the last populated slot and resets it.
+  void __attribute__((noinline)) remove_client_(uint8_t client_index);
 
 #ifdef USE_API_NOISE
   bool update_noise_psk_(const SavedNoisePsk &new_psk, const LogString *save_log_msg, const LogString *fail_log_msg,
@@ -273,8 +291,9 @@ class APIServer final : public Component,
   uint32_t reboot_timeout_{300000};
   uint32_t last_connected_{0};
 
+  // Slots [0, api_connection_count_) are populated; trailing slots are always nullptr.
+  std::array<std::unique_ptr<APIConnection>, MAX_API_CONNECTIONS> clients_{};
   // Vectors and strings (12 bytes each on 32-bit)
-  std::vector<std::unique_ptr<APIConnection>> clients_;
   // Shared proto write buffer for all connections.
   // Not pre-allocated: all send paths call prepare_first_message_buffer() which
   // reserves the exact needed size. Pre-allocating here would cause heap fragmentation
@@ -309,10 +328,10 @@ class APIServer final : public Component,
   uint16_t port_{6053};
   uint16_t batch_delay_{100};
   // Connection limits - these defaults will be overridden by config values
-  // from cv.SplitDefault in __init__.py which sets platform-specific defaults
+  // from cv.SplitDefault in __init__.py which sets platform-specific defaults.
   uint8_t listen_backlog_{4};
-  uint8_t max_connections_{8};
   bool shutting_down_ = false;
+  uint8_t api_connection_count_{0};
   // 7 bytes used, 1 byte padding
 
 #ifdef USE_API_NOISE

esphome/components/debug/debug_component.cpp

@@ -30,7 +30,7 @@ void DebugComponent::dump_config() {
 
   char device_info_buffer[DEVICE_INFO_BUFFER_SIZE];
   ESP_LOGD(TAG, "ESPHome version %s", ESPHOME_VERSION);
-  size_t pos = buf_append_printf(device_info_buffer, DEVICE_INFO_BUFFER_SIZE, 0, "%s", ESPHOME_VERSION);
+  size_t pos = buf_append_str(device_info_buffer, DEVICE_INFO_BUFFER_SIZE, 0, ESPHOME_VERSION);
 
   this->free_heap_ = get_free_heap_();
   ESP_LOGD(TAG, "Free Heap Size: %" PRIu32 " bytes", this->free_heap_);

esphome/components/debug/debug_esp32.cpp

@@ -224,17 +224,21 @@ size_t DebugComponent::get_device_info_(std::span<char, DEVICE_INFO_BUFFER_SIZE>
   const char *model = ESPHOME_VARIANT;
 
   // Build features string
-  pos = buf_append_printf(buf, size, pos, "|Chip: %s Features:", model);
+  pos = buf_append_str(buf, size, pos, "|Chip: ");
+  pos = buf_append_str(buf, size, pos, model);
+  pos = buf_append_str(buf, size, pos, " Features:");
   bool first_feature = true;
   for (const auto &feature : CHIP_FEATURES) {
     if (info.features & feature.bit) {
-      pos = buf_append_printf(buf, size, pos, "%s%s", first_feature ? "" : ", ", feature.name);
+      pos = buf_append_str(buf, size, pos, first_feature ? "" : ", ");
+      pos = buf_append_str(buf, size, pos, feature.name);
       first_feature = false;
       info.features &= ~feature.bit;
     }
   }
   if (info.features != 0) {
-    pos = buf_append_printf(buf, size, pos, "%sOther:0x%" PRIx32, first_feature ? "" : ", ", info.features);
+    pos = buf_append_str(buf, size, pos, first_feature ? "" : ", ");
+    pos = buf_append_printf(buf, size, pos, "Other:0x%" PRIx32, info.features);
   }
   pos = buf_append_printf(buf, size, pos, " Cores:%u Revision:%u", info.cores, info.revision);
 
@@ -267,17 +271,20 @@ size_t DebugComponent::get_device_info_(std::span<char, DEVICE_INFO_BUFFER_SIZE>
   // Framework detection
 #ifdef USE_ARDUINO
   ESP_LOGD(TAG, "  Framework: Arduino");
-  pos = buf_append_printf(buf, size, pos, "|Framework: Arduino");
+  pos = buf_append_str(buf, size, pos, "|Framework: Arduino");
 #else
   ESP_LOGD(TAG, "  Framework: ESP-IDF");
-  pos = buf_append_printf(buf, size, pos, "|Framework: ESP-IDF");
+  pos = buf_append_str(buf, size, pos, "|Framework: ESP-IDF");
 #endif
 
-  pos = buf_append_printf(buf, size, pos, "|ESP-IDF: %s", esp_get_idf_version());
+  pos = buf_append_str(buf, size, pos, "|ESP-IDF: ");
+  pos = buf_append_str(buf, size, pos, esp_get_idf_version());
   pos = buf_append_printf(buf, size, pos, "|EFuse MAC: %02X:%02X:%02X:%02X:%02X:%02X", mac[0], mac[1], mac[2], mac[3],
                           mac[4], mac[5]);
-  pos = buf_append_printf(buf, size, pos, "|Reset: %s", reset_reason);
-  pos = buf_append_printf(buf, size, pos, "|Wakeup: %s", wakeup_cause);
+  pos = buf_append_str(buf, size, pos, "|Reset: ");
+  pos = buf_append_str(buf, size, pos, reset_reason);
+  pos = buf_append_str(buf, size, pos, "|Wakeup: ");
+  pos = buf_append_str(buf, size, pos, wakeup_cause);
 
   return pos;
 }

esphome/components/debug/debug_libretiny.cpp

@@ -38,9 +38,12 @@ size_t DebugComponent::get_device_info_(std::span<char, DEVICE_INFO_BUFFER_SIZE>
            lt_get_version(), lt_cpu_get_model_name(), lt_cpu_get_model(), lt_cpu_get_freq_mhz(), mac_id,
            lt_get_board_code(), flash_kib, ram_kib, reset_reason);
 
-  pos = buf_append_printf(buf, size, pos, "|Version: %s", LT_BANNER_STR + 10);
-  pos = buf_append_printf(buf, size, pos, "|Reset Reason: %s", reset_reason);
-  pos = buf_append_printf(buf, size, pos, "|Chip Name: %s", lt_cpu_get_model_name());
+  pos = buf_append_str(buf, size, pos, "|Version: ");
+  pos = buf_append_str(buf, size, pos, LT_BANNER_STR + 10);
+  pos = buf_append_str(buf, size, pos, "|Reset Reason: ");
+  pos = buf_append_str(buf, size, pos, reset_reason);
+  pos = buf_append_str(buf, size, pos, "|Chip Name: ");
+  pos = buf_append_str(buf, size, pos, lt_cpu_get_model_name());
   pos = buf_append_printf(buf, size, pos, "|Chip ID: 0x%06" PRIX32, mac_id);
   pos = buf_append_printf(buf, size, pos, "|Flash: %" PRIu32 " KiB", flash_kib);
   pos = buf_append_printf(buf, size, pos, "|RAM: %" PRIu32 " KiB", ram_kib);

esphome/components/debug/debug_zephyr.cpp

@@ -162,14 +162,18 @@ size_t DebugComponent::get_device_info_(std::span<char, DEVICE_INFO_BUFFER_SIZE>
   const char *supply_status =
       (nrf_power_mainregstatus_get(NRF_POWER) == NRF_POWER_MAINREGSTATUS_NORMAL) ? "Normal voltage." : "High voltage.";
   ESP_LOGD(TAG, "Main supply status: %s", supply_status);
-  pos = buf_append_printf(buf, size, pos, "|Main supply status: %s", supply_status);
+  pos = buf_append_str(buf, size, pos, "|Main supply status: ");
+  pos = buf_append_str(buf, size, pos, supply_status);
 
   // Regulator stage 0
   if (nrf_power_mainregstatus_get(NRF_POWER) == NRF_POWER_MAINREGSTATUS_HIGH) {
     const char *reg0_type = nrf_power_dcdcen_vddh_get(NRF_POWER) ? "DC/DC" : "LDO";
     const char *reg0_voltage = regout0_to_str((NRF_UICR->REGOUT0 & UICR_REGOUT0_VOUT_Msk) >> UICR_REGOUT0_VOUT_Pos);
     ESP_LOGD(TAG, "Regulator stage 0: %s, %s", reg0_type, reg0_voltage);
-    pos = buf_append_printf(buf, size, pos, "|Regulator stage 0: %s, %s", reg0_type, reg0_voltage);
+    pos = buf_append_str(buf, size, pos, "|Regulator stage 0: ");
+    pos = buf_append_str(buf, size, pos, reg0_type);
+    pos = buf_append_str(buf, size, pos, ", ");
+    pos = buf_append_str(buf, size, pos, reg0_voltage);
 #ifdef USE_NRF52_REG0_VOUT
     if ((NRF_UICR->REGOUT0 & UICR_REGOUT0_VOUT_Msk) >> UICR_REGOUT0_VOUT_Pos != USE_NRF52_REG0_VOUT) {
       ESP_LOGE(TAG, "Regulator stage 0: expected %s", regout0_to_str(USE_NRF52_REG0_VOUT));
@@ -177,13 +181,14 @@ size_t DebugComponent::get_device_info_(std::span<char, DEVICE_INFO_BUFFER_SIZE>
 #endif
   } else {
     ESP_LOGD(TAG, "Regulator stage 0: disabled");
-    pos = buf_append_printf(buf, size, pos, "|Regulator stage 0: disabled");
+    pos = buf_append_str(buf, size, pos, "|Regulator stage 0: disabled");
   }
 
   // Regulator stage 1
   const char *reg1_type = nrf_power_dcdcen_get(NRF_POWER) ? "DC/DC" : "LDO";
   ESP_LOGD(TAG, "Regulator stage 1: %s", reg1_type);
-  pos = buf_append_printf(buf, size, pos, "|Regulator stage 1: %s", reg1_type);
+  pos = buf_append_str(buf, size, pos, "|Regulator stage 1: ");
+  pos = buf_append_str(buf, size, pos, reg1_type);
 
   // USB power state
   const char *usb_state;
@@ -197,7 +202,8 @@ size_t DebugComponent::get_device_info_(std::span<char, DEVICE_INFO_BUFFER_SIZE>
     usb_state = "disconnected";
   }
   ESP_LOGD(TAG, "USB power state: %s", usb_state);
-  pos = buf_append_printf(buf, size, pos, "|USB power state: %s", usb_state);
+  pos = buf_append_str(buf, size, pos, "|USB power state: ");
+  pos = buf_append_str(buf, size, pos, usb_state);
 
   // Power-fail comparator
   bool enabled;
@@ -302,14 +308,18 @@ size_t DebugComponent::get_device_info_(std::span<char, DEVICE_INFO_BUFFER_SIZE>
           break;
       }
       ESP_LOGD(TAG, "Power-fail comparator: %s, VDDH: %s", pof_voltage, vddh_voltage);
-      pos = buf_append_printf(buf, size, pos, "|Power-fail comparator: %s, VDDH: %s", pof_voltage, vddh_voltage);
+      pos = buf_append_str(buf, size, pos, "|Power-fail comparator: ");
+      pos = buf_append_str(buf, size, pos, pof_voltage);
+      pos = buf_append_str(buf, size, pos, ", VDDH: ");
+      pos = buf_append_str(buf, size, pos, vddh_voltage);
     } else {
       ESP_LOGD(TAG, "Power-fail comparator: %s", pof_voltage);
-      pos = buf_append_printf(buf, size, pos, "|Power-fail comparator: %s", pof_voltage);
+      pos = buf_append_str(buf, size, pos, "|Power-fail comparator: ");
+      pos = buf_append_str(buf, size, pos, pof_voltage);
     }
   } else {
     ESP_LOGD(TAG, "Power-fail comparator: disabled");
-    pos = buf_append_printf(buf, size, pos, "|Power-fail comparator: disabled");
+    pos = buf_append_str(buf, size, pos, "|Power-fail comparator: disabled");
   }
 
   auto package = [](uint32_t value) {

esphome/components/ld2412/ld2412.cpp

@@ -766,32 +766,38 @@ void LD2412Component::get_distance_resolution_() { this->send_command_(CMD_QUERY
 void LD2412Component::query_light_control_() { this->send_command_(CMD_QUERY_LIGHT_CONTROL, nullptr, 0); }
 
 void LD2412Component::set_basic_config() {
+  uint8_t min_gate = 1;
+  uint8_t max_gate = TOTAL_GATES;
+  uint16_t timeout = DEFAULT_PRESENCE_TIMEOUT;
+  uint8_t out_pin_level = 0x01;
+
 #ifdef USE_NUMBER
-  if (!this->min_distance_gate_number_->has_state() || !this->max_distance_gate_number_->has_state() ||
-      !this->timeout_number_->has_state()) {
-    return;
+  if (this->min_distance_gate_number_ != nullptr) {
+    if (!this->min_distance_gate_number_->has_state())
+      return;
+    min_gate = static_cast<int>(this->min_distance_gate_number_->state);
+  }
+  if (this->max_distance_gate_number_ != nullptr) {
+    if (!this->max_distance_gate_number_->has_state())
+      return;
+    max_gate = static_cast<int>(this->max_distance_gate_number_->state) + 1;
+  }
+  if (this->timeout_number_ != nullptr) {
+    if (!this->timeout_number_->has_state())
+      return;
+    timeout = static_cast<int>(this->timeout_number_->state);
   }
 #endif
 #ifdef USE_SELECT
-  if (!this->out_pin_level_select_->has_state()) {
-    return;
+  if (this->out_pin_level_select_ != nullptr) {
+    if (!this->out_pin_level_select_->has_state())
+      return;
+    out_pin_level = find_uint8(OUT_PIN_LEVELS_BY_STR, this->out_pin_level_select_->current_option().c_str());
   }
 #endif
 
   uint8_t value[5] = {
-#ifdef USE_NUMBER
-      lowbyte(static_cast<int>(this->min_distance_gate_number_->state)),
-      lowbyte(static_cast<int>(this->max_distance_gate_number_->state) + 1),
-      lowbyte(static_cast<int>(this->timeout_number_->state)),
-      highbyte(static_cast<int>(this->timeout_number_->state)),
-#else
-      1,    TOTAL_GATES, DEFAULT_PRESENCE_TIMEOUT, 0,
-#endif
-#ifdef USE_SELECT
-      find_uint8(OUT_PIN_LEVELS_BY_STR, this->out_pin_level_select_->current_option().c_str()),
-#else
-      0x01,  // Default value if not using select
-#endif
+      lowbyte(min_gate), lowbyte(max_gate), lowbyte(timeout), highbyte(timeout), out_pin_level,
   };
   this->set_config_mode_(true);
   this->send_command_(CMD_BASIC_CONF, value, sizeof(value));

esphome/components/lock/__init__.py

@@ -35,9 +35,11 @@ LockStateForwarder = lock_ns.class_("LockStateForwarder")
 LockState = lock_ns.enum("LockState")
 
 LOCK_STATES = {
+    "OPEN": LockState.LOCK_STATE_OPEN,
     "LOCKED": LockState.LOCK_STATE_LOCKED,
     "UNLOCKED": LockState.LOCK_STATE_UNLOCKED,
     "JAMMED": LockState.LOCK_STATE_JAMMED,
+    "OPENING": LockState.LOCK_STATE_OPENING,
     "LOCKING": LockState.LOCK_STATE_LOCKING,
     "UNLOCKING": LockState.LOCK_STATE_UNLOCKING,
 }

esphome/components/lock/lock.cpp

@@ -8,9 +8,10 @@ namespace esphome::lock {
 
 static const char *const TAG = "lock";
 
-// Lock state strings indexed by LockState enum (0-5): NONE(UNKNOWN), LOCKED, UNLOCKED, JAMMED, LOCKING, UNLOCKING
+// Lock state strings indexed by LockState enum.
 // Index 0 is UNKNOWN (for LOCK_STATE_NONE), also used as fallback for out-of-range
-PROGMEM_STRING_TABLE(LockStateStrings, "UNKNOWN", "LOCKED", "UNLOCKED", "JAMMED", "LOCKING", "UNLOCKING");
+PROGMEM_STRING_TABLE(LockStateStrings, "UNKNOWN", "LOCKED", "UNLOCKED", "JAMMED", "LOCKING", "UNLOCKING", "OPENING",
+                     "OPEN");
 
 const LogString *lock_state_to_string(LockState state) {
   return LockStateStrings::get_log_str(static_cast<uint8_t>(state), 0);
@@ -74,12 +75,16 @@ LockCall &LockCall::set_state(optional<LockState> state) {
   return *this;
 }
 LockCall &LockCall::set_state(const char *state) {
-  if (ESPHOME_strcasecmp_P(state, ESPHOME_PSTR("LOCKED")) == 0) {
+  if (ESPHOME_strcasecmp_P(state, ESPHOME_PSTR("OPEN")) == 0) {
+    this->set_state(LOCK_STATE_OPEN);
+  } else if (ESPHOME_strcasecmp_P(state, ESPHOME_PSTR("LOCKED")) == 0) {
     this->set_state(LOCK_STATE_LOCKED);
   } else if (ESPHOME_strcasecmp_P(state, ESPHOME_PSTR("UNLOCKED")) == 0) {
     this->set_state(LOCK_STATE_UNLOCKED);
   } else if (ESPHOME_strcasecmp_P(state, ESPHOME_PSTR("JAMMED")) == 0) {
     this->set_state(LOCK_STATE_JAMMED);
+  } else if (ESPHOME_strcasecmp_P(state, ESPHOME_PSTR("OPENING")) == 0) {
+    this->set_state(LOCK_STATE_OPENING);
   } else if (ESPHOME_strcasecmp_P(state, ESPHOME_PSTR("LOCKING")) == 0) {
     this->set_state(LOCK_STATE_LOCKING);
   } else if (ESPHOME_strcasecmp_P(state, ESPHOME_PSTR("UNLOCKING")) == 0) {

esphome/components/lock/lock.h

@@ -26,7 +26,9 @@ enum LockState : uint8_t {
   LOCK_STATE_UNLOCKED = 2,
   LOCK_STATE_JAMMED = 3,
   LOCK_STATE_LOCKING = 4,
-  LOCK_STATE_UNLOCKING = 5
+  LOCK_STATE_UNLOCKING = 5,
+  LOCK_STATE_OPENING = 6,
+  LOCK_STATE_OPEN = 7,
 };
 const LogString *lock_state_to_string(LockState state);
 

esphome/components/qmc5883l/qmc5883l.cpp

@@ -24,6 +24,8 @@ static const uint8_t QMC5883L_REGISTER_CONTROL_1 = 0x09;
 static const uint8_t QMC5883L_REGISTER_CONTROL_2 = 0x0A;
 static const uint8_t QMC5883L_REGISTER_PERIOD = 0x0B;
 
+void IRAM_ATTR QMC5883LComponent::gpio_intr(QMC5883LComponent *arg) { arg->enable_loop_soon_any_context(); }
+
 void QMC5883LComponent::setup() {
   // Soft Reset
   if (!this->write_byte(QMC5883L_REGISTER_CONTROL_2, 1 << 7)) {
@@ -35,6 +37,12 @@ void QMC5883LComponent::setup() {
 
   if (this->drdy_pin_) {
     this->drdy_pin_->setup();
+    if (this->drdy_pin_->is_internal()) {
+      static_cast<InternalGPIOPin *>(this->drdy_pin_)
+          ->attach_interrupt(&QMC5883LComponent::gpio_intr, this, gpio::INTERRUPT_RISING_EDGE);
+      this->drdy_use_isr_ = true;
+      this->stop_poller();
+    }
   }
 
   uint8_t control_1 = 0;
@@ -65,8 +73,8 @@ void QMC5883LComponent::setup() {
     return;
   }
 
-  if (this->get_update_interval() < App.get_loop_interval()) {
-    high_freq_.start();
+  if (!this->drdy_use_isr_ && this->get_update_interval() < App.get_loop_interval()) {
+    this->high_freq_.start();
   }
 }
 
@@ -84,16 +92,32 @@ void QMC5883LComponent::dump_config() {
   LOG_SENSOR("  ", "Heading", this->heading_sensor_);
   LOG_SENSOR("  ", "Temperature", this->temperature_sensor_);
   LOG_PIN("  DRDY Pin: ", this->drdy_pin_);
+  if (this->drdy_pin_ != nullptr) {
+    ESP_LOGCONFIG(TAG, "  DRDY mode: %s",
+                  this->drdy_use_isr_ ? LOG_STR_LITERAL("interrupt") : LOG_STR_LITERAL("polling"));
+  }
 }
 
 void QMC5883LComponent::update() {
-  i2c::ErrorCode err;
-  uint8_t status = false;
-
-  // If DRDY pin is configured and the data is not ready return.
+  // If DRDY is on an external expander we keep the polling path and early-return
+  // if data is not ready yet. Internal DRDY pins take the ISR path via loop().
   if (this->drdy_pin_ && !this->drdy_pin_->digital_read()) {
     return;
   }
+  this->read_sensor_();
+}
+
+void QMC5883LComponent::loop() {
+  this->disable_loop();
+  if (!this->drdy_use_isr_ || !this->drdy_pin_->digital_read()) {
+    return;
+  }
+  this->read_sensor_();
+}
+
+void QMC5883LComponent::read_sensor_() {
+  i2c::ErrorCode err;
+  uint8_t status = false;
 
   // Status byte gets cleared when data is read, so we have to read this first.
   // If status and two axes are desired, it's possible to save one byte of traffic by enabling

esphome/components/qmc5883l/qmc5883l.h

@@ -32,6 +32,7 @@ class QMC5883LComponent : public PollingComponent, public i2c::I2CDevice {
   void setup() override;
   void dump_config() override;
   void update() override;
+  void loop() override;
 
   void set_drdy_pin(GPIOPin *pin) { drdy_pin_ = pin; }
   void set_datarate(QMC5883LDatarate datarate) { datarate_ = datarate; }
@@ -44,6 +45,9 @@ class QMC5883LComponent : public PollingComponent, public i2c::I2CDevice {
   void set_temperature_sensor(sensor::Sensor *temperature_sensor) { temperature_sensor_ = temperature_sensor; }
 
  protected:
+  static void IRAM_ATTR gpio_intr(QMC5883LComponent *arg);
+  void read_sensor_();
+
   QMC5883LDatarate datarate_{QMC5883L_DATARATE_10_HZ};
   QMC5883LRange range_{QMC5883L_RANGE_200_UT};
   QMC5883LOversampling oversampling_{QMC5883L_SAMPLING_512};
@@ -53,6 +57,7 @@ class QMC5883LComponent : public PollingComponent, public i2c::I2CDevice {
   sensor::Sensor *heading_sensor_{nullptr};
   sensor::Sensor *temperature_sensor_{nullptr};
   GPIOPin *drdy_pin_{nullptr};
+  bool drdy_use_isr_{false};
   enum ErrorCode {
     NONE = 0,
     COMMUNICATION_FAILED,

esphome/components/runtime_stats/runtime_stats.cpp

@@ -137,11 +137,12 @@ bool RuntimeStatsCollector::compare_total_time(Component *a, Component *b) {
   return a->runtime_stats_.total_time_us > b->runtime_stats_.total_time_us;
 }
 
-void RuntimeStatsCollector::process_pending_stats(uint32_t current_time) {
-  if ((int32_t) (current_time - this->next_log_time_) >= 0) {
-    this->log_stats_();
-    this->next_log_time_ = current_time + this->log_interval_;
-  }
+// Slow path for process_pending_stats — gate already checked by the inline
+// wrapper in runtime_stats.h. Out-of-line keeps the log_stats_ machinery out
+// of Application::loop().
+void RuntimeStatsCollector::process_pending_stats_slow_(uint32_t current_time) {
+  this->log_stats_();
+  this->next_log_time_ = current_time + this->log_interval_;
 }
 
 }  // namespace runtime_stats

esphome/components/runtime_stats/runtime_stats.h

@@ -6,6 +6,7 @@
 
 #include <cstdint>
 #include "esphome/core/hal.h"
+#include "esphome/core/helpers.h"
 #include "esphome/core/log.h"
 
 namespace esphome {
@@ -26,14 +27,24 @@ class RuntimeStatsCollector {
   }
   uint32_t get_log_interval() const { return this->log_interval_; }
 
-  // Process any pending stats printing (should be called after component loop)
-  void process_pending_stats(uint32_t current_time);
+  // Process any pending stats printing. Called on every Application::loop()
+  // tick, so the common "not yet time to log" path must be cheap — inline
+  // the gate check and keep the actual logging work out-of-line.
+  void ESPHOME_ALWAYS_INLINE process_pending_stats(uint32_t current_time) {
+    if ((int32_t) (current_time - this->next_log_time_) >= 0) [[unlikely]] {
+      this->process_pending_stats_slow_(current_time);
+    }
+  }
 
   // Record the wall time of one main loop iteration excluding the yield/sleep.
   // Called once per loop from Application::loop().
   //   active_us = total time between loop start and just before yield.
-  //   before_us = time spent in before_loop_tasks_ (scheduler + ISR enable_loop).
-  //   tail_us   = time spent in after_loop_tasks_ + the trailing record/stats prefix.
+  //   before_us = time spent in Phase A (scheduler tick) excluding time
+  //               already attributed to per-component stats.
+  //   tail_us   = time spent in after_component_phase_ + the trailing record/stats
+  //               prefix. Only meaningful on component-phase ticks; reported
+  //               as 0 on Phase A-only ticks (no component phase ran, so any
+  //               overhead between Phase A and stats belongs to "residual").
   // Residual overhead at log time = active − Σ(component) − before − tail,
   // which captures per-iteration inter-component bookkeeping (set_current_component,
   // WarnIfComponentBlockingGuard construction/destruction, feed_wdt_with_time calls,
@@ -55,6 +66,7 @@ class RuntimeStatsCollector {
   }
 
  protected:
+  void process_pending_stats_slow_(uint32_t current_time);
   void log_stats_();
   // Static comparators — member functions have friend access, lambdas do not
   static bool compare_period_time(Component *a, Component *b);

esphome/components/template/text/__init__.py

@@ -3,6 +3,7 @@ import esphome.codegen as cg
 from esphome.components import text
 import esphome.config_validation as cv
 from esphome.const import (
+    CONF_ID,
     CONF_INITIAL_VALUE,
     CONF_LAMBDA,
     CONF_MAX_LENGTH,
@@ -12,6 +13,7 @@ from esphome.const import (
     CONF_RESTORE_VALUE,
     CONF_SET_ACTION,
 )
+from esphome.core import ID
 
 from .. import template_ns
 
@@ -84,8 +86,15 @@ async def to_code(config):
         if initial_value_config := config.get(CONF_INITIAL_VALUE):
             cg.add(var.set_initial_value(initial_value_config))
         if config[CONF_RESTORE_VALUE]:
-            args = cg.TemplateArguments(config[CONF_MAX_LENGTH])
-            saver = TextSaverTemplate.template(args).new()
+            saver_id = ID(
+                f"{config[CONF_ID].id}_value_saver",
+                is_declaration=True,
+                type=TextSaverBase,
+            )
+            saver_type = TextSaverTemplate.template(
+                cg.TemplateArguments(config[CONF_MAX_LENGTH])
+            )
+            saver = cg.Pvariable(saver_id, saver_type.new())
             cg.add(var.set_value_saver(saver))
 
     if CONF_SET_ACTION in config:

esphome/components/zwave_proxy/zwave_proxy.cpp

@@ -101,8 +101,10 @@ void ZWaveProxy::loop() {
   this->status_clear_warning();
 }
 
-void ZWaveProxy::process_uart_() {
-  while (this->available()) {
+void ZWaveProxy::process_uart_slow_() {
+  // Caller (inline process_uart_) has already confirmed available() > 0, so use do/while to
+  // drain bytes — available() is still checked at the tail, but not redundantly on entry.
+  do {
     uint8_t byte;
     if (!this->read_byte(&byte)) {
       this->status_set_warning(LOG_STR("UART read failed"));
@@ -137,7 +139,7 @@ void ZWaveProxy::process_uart_() {
         this->api_connection_->send_message(this->outgoing_proto_msg_);
       }
     }
-  }
+  } while (this->available());
 }
 
 void ZWaveProxy::dump_config() {
@@ -414,7 +416,7 @@ void ZWaveProxy::parse_start_(uint8_t byte) {
   }
 }
 
-bool ZWaveProxy::response_handler_() {
+bool ZWaveProxy::response_handler_slow_() {
   switch (this->parsing_state_) {
     case ZWAVE_PARSING_STATE_SEND_ACK:
       this->last_response_ = ZWAVE_FRAME_TYPE_ACK;

esphome/components/zwave_proxy/zwave_proxy.h

@@ -38,6 +38,13 @@ enum ZWaveParsingState : uint8_t {
   ZWAVE_PARSING_STATE_READ_BL_MENU,
 };
 
+// response_handler_()'s inline fast-path relies on SEND_ACK/CAN/NAK being contiguous in this
+// enum so a single range check (state - SEND_ACK < 3) is equivalent to three equality checks.
+static_assert(ZWAVE_PARSING_STATE_SEND_CAN == ZWAVE_PARSING_STATE_SEND_ACK + 1,
+              "SEND_CAN must immediately follow SEND_ACK for response_handler_ fast-path");
+static_assert(ZWAVE_PARSING_STATE_SEND_NAK == ZWAVE_PARSING_STATE_SEND_ACK + 2,
+              "SEND_NAK must immediately follow SEND_CAN for response_handler_ fast-path");
+
 enum ZWaveProxyFeature : uint32_t {
   FEATURE_ZWAVE_PROXY_ENABLED = 1 << 0,
 };
@@ -72,8 +79,31 @@ class ZWaveProxy : public uart::UARTDevice, public Component {
   void send_simple_command_(uint8_t command_id);
   bool parse_byte_(uint8_t byte);  // Returns true if frame parsing was completed (a frame is ready in the buffer)
   void parse_start_(uint8_t byte);
-  bool response_handler_();
-  void process_uart_();  // Process all available UART data
+  // Inline fast-path: most calls happen with parsing_state_ outside the SEND_* range, so skip the
+  // out-of-line call entirely in the hot path (e.g. every loop() tick) and only pay for the real
+  // work when a response is actually pending. ESPHOME_ALWAYS_INLINE is required because with -Os
+  // gcc otherwise clones the wrapper into a shared $isra$ outline and keeps the call8.
+  ESPHOME_ALWAYS_INLINE bool response_handler_() {
+    if (this->parsing_state_ < ZWAVE_PARSING_STATE_SEND_ACK || this->parsing_state_ > ZWAVE_PARSING_STATE_SEND_NAK) {
+      return false;
+    }
+    return this->response_handler_slow_();
+  }
+  bool response_handler_slow_();
+  // Inline fast-path: UART::available() is cheap (ring-buffer head/tail compare on most backends).
+  // On an idle loop tick we want to skip the call to process_uart_ entirely. When bytes are
+  // pending we fall into the slow path, which drains the UART with a do/while so available() is
+  // only checked once per byte — no redundant re-check on entry.
+  ESPHOME_ALWAYS_INLINE void process_uart_() {
+    if (!this->available()) {
+      return;
+    }
+    this->process_uart_slow_();
+  }
+  // Precondition: caller must guarantee available() > 0 before invoking (see inline
+  // process_uart_ above). The slow path uses do/while and would otherwise set a spurious UART
+  // warning on entry if called with no bytes pending.
+  void process_uart_slow_();
 
   // Pre-allocated message - always ready to send
   api::ZWaveProxyFrame outgoing_proto_msg_;

esphome/core/application.cpp

@@ -93,8 +93,11 @@ void Application::setup() {
     do {
       uint32_t now = MillisInternal::get();
 
-      // Process pending loop enables to handle GPIO interrupts during setup
-      this->before_loop_tasks_(now);
+      // Service scheduler and process pending loop enables to handle GPIO
+      // interrupts during setup. During setup we always run the component
+      // phase (no loop_interval_ gate), so call both helpers unconditionally.
+      this->scheduler_tick_(now);
+      this->before_component_phase_();
 
       for (uint32_t j = 0; j <= i; j++) {
         // Update loop_component_start_time_ right before calling each component
@@ -103,7 +106,7 @@ void Application::setup() {
         this->feed_wdt();
       }
 
-      this->after_loop_tasks_();
+      this->after_component_phase_();
       yield();
     } while (!component->can_proceed() && !component->is_failed());
   }

esphome/core/application.h

@@ -426,8 +426,9 @@ class Application {
   void enable_component_loop_(Component *component);
   void enable_pending_loops_();
   void activate_looping_component_(uint16_t index);
-  inline uint32_t ESPHOME_ALWAYS_INLINE before_loop_tasks_(uint32_t loop_start_time);
-  inline void ESPHOME_ALWAYS_INLINE after_loop_tasks_() { this->in_loop_ = false; }
+  inline uint32_t ESPHOME_ALWAYS_INLINE scheduler_tick_(uint32_t now);
+  inline void ESPHOME_ALWAYS_INLINE before_component_phase_();
+  inline void ESPHOME_ALWAYS_INLINE after_component_phase_() { this->in_loop_ = false; }
 
   /// Process dump_config output one component per loop iteration.
   /// Extracted from loop() to keep cold startup/reconnect logging out of the hot path.
@@ -582,16 +583,25 @@ inline void Application::drain_wake_notifications_() {
 }
 #endif  // USE_HOST
 
-inline uint32_t ESPHOME_ALWAYS_INLINE Application::before_loop_tasks_(uint32_t loop_start_time) {
+// Phase A: drain wake notifications and run the scheduler. Invoked on every
+// Application::loop() tick regardless of whether a component phase runs, so
+// scheduler items fire at their requested cadence even when the caller has
+// raised loop_interval_ for power savings (see Application::loop()).
+// Returns the timestamp of the last scheduler item that ran (or `now`
+// unchanged if none ran), so the caller's WDT feed stays monotonic with the
+// per-item feeds inside scheduler.call() without an extra millis().
+inline uint32_t ESPHOME_ALWAYS_INLINE Application::scheduler_tick_(uint32_t now) {
 #ifdef USE_HOST
   // Drain wake notifications first to clear socket for next wake
   this->drain_wake_notifications_();
 #endif
+  return this->scheduler.call(now);
+}
 
-  // Scheduler::call feeds the WDT per item and returns the timestamp of the
-  // last fired item, or the input unchanged when nothing ran.
-  uint32_t last_op_end_time = this->scheduler.call(loop_start_time);
-
+// Phase B entry: only invoked when a component loop phase is about to run.
+// Processes pending enable_loop requests from ISRs and marks in_loop_ so
+// reentrant modifications during component.loop() are safe.
+inline void ESPHOME_ALWAYS_INLINE Application::before_component_phase_() {
   // Process any pending enable_loop requests from ISRs
   // This must be done before marking in_loop_ = true to avoid race conditions
   if (this->has_pending_enable_loop_requests_) {
@@ -608,7 +618,6 @@ inline uint32_t ESPHOME_ALWAYS_INLINE Application::before_loop_tasks_(uint32_t l
 
   // Mark that we're in the loop for safe reentrant modifications
   this->in_loop_ = true;
-  return last_op_end_time;
 }
 
 inline void ESPHOME_ALWAYS_INLINE Application::loop() {
@@ -623,46 +632,77 @@ inline void ESPHOME_ALWAYS_INLINE Application::loop() {
   // so charging it again to "before" would double-count.
   uint64_t loop_recorded_snap = ComponentRuntimeStats::global_recorded_us;
 #endif
-  // Get the initial loop time at the start
-  uint32_t last_op_end_time = MillisInternal::get();
+  // Phase A: always service the scheduler. Decouples scheduler cadence from
+  // loop_interval_ so raised intervals (for power savings) don't drag scheduled
+  // items forward. A tick that only runs the scheduler is cheap.
+  // scheduler_tick_ returns the timestamp of the last scheduler item that ran
+  // (advanced by its per-item feeds) or `now` unchanged. We adopt it as `now`
+  // so the gate check and WDT feed both reflect actual elapsed time after
+  // scheduler dispatch, without an extra millis() call.
+  uint32_t now = this->scheduler_tick_(MillisInternal::get());
+  // Guarantee one WDT feed per tick even when the scheduler had nothing to
+  // dispatch and the component phase is gated out — covers configs with no
+  // looping components and no scheduler work (setup() has its own
+  // per-component feed_wdt calls, so only do this here, not in scheduler_tick_).
+  this->feed_wdt_with_time(now);
 
-  // Returned timestamp keeps us monotonic with last_wdt_feed_ (advanced by
-  // the scheduler's per-item feeds) without an extra millis() call.
-  last_op_end_time = this->before_loop_tasks_(last_op_end_time);
-  // Guarantee a WDT touch every tick — covers configs with no looping
-  // components and no scheduler work, where the per-item / per-component
-  // feeds never fire. Rate-limited inline fast path, ~free when unneeded.
-  this->feed_wdt_with_time(last_op_end_time);
 #ifdef USE_RUNTIME_STATS
   uint32_t loop_before_end_us = micros();
   uint64_t loop_before_scheduled_us = ComponentRuntimeStats::global_recorded_us - loop_recorded_snap;
+  // Only meaningful when do_component_phase is true; initialized to 0 so the
+  // tail bucket receives 0 on Phase A-only ticks (no component tail happened,
+  // the gate-check / stats-prefix overhead belongs to "residual", not "tail").
+  uint32_t loop_tail_start_us = 0;
 #endif
 
-  for (this->current_loop_index_ = 0; this->current_loop_index_ < this->looping_components_active_end_;
-       this->current_loop_index_++) {
-    Component *component = this->looping_components_[this->current_loop_index_];
+  // Gate the component phase on loop_interval_, an active high-frequency
+  // request, or an explicit wake from a background producer. A scheduler-only
+  // wake (e.g. set_interval firing under a raised loop_interval_) leaves the
+  // component phase gated; an external producer that called wake_loop_*
+  // (MQTT RX, USB RX, BLE event, etc.) needs the component phase to actually
+  // run so its component's loop() can drain the queued work — that is the
+  // long-standing semantic of wake_loop_threadsafe(), and the wake_request
+  // flag preserves it. wake_request_take() exchange-clears the flag; wakes
+  // that arrive during Phase B re-set it and run Phase B again on the next
+  // iteration.
+  const bool high_frequency = HighFrequencyLoopRequester::is_high_frequency();
+  const uint32_t elapsed = now - this->last_loop_;
+  const bool woke = esphome::wake_request_take();
+  const bool do_component_phase = high_frequency || woke || (elapsed >= this->loop_interval_);
 
-    // Update the cached time before each component runs
-    this->loop_component_start_time_ = last_op_end_time;
+  if (do_component_phase) {
+    this->before_component_phase_();
 
-    {
-      this->set_current_component(component);
-      WarnIfComponentBlockingGuard guard{component, last_op_end_time};
-      component->loop();
-      // Use the finish method to get the current time as the end time
-      last_op_end_time = guard.finish();
+    uint32_t last_op_end_time = now;
+    for (this->current_loop_index_ = 0; this->current_loop_index_ < this->looping_components_active_end_;
+         this->current_loop_index_++) {
+      Component *component = this->looping_components_[this->current_loop_index_];
+
+      // Update the cached time before each component runs
+      this->loop_component_start_time_ = last_op_end_time;
+
+      {
+        this->set_current_component(component);
+        WarnIfComponentBlockingGuard guard{component, last_op_end_time};
+        component->loop();
+        // Use the finish method to get the current time as the end time
+        last_op_end_time = guard.finish();
+      }
+      this->feed_wdt_with_time(last_op_end_time);
     }
-    this->feed_wdt_with_time(last_op_end_time);
+
+#ifdef USE_RUNTIME_STATS
+    loop_tail_start_us = micros();
+#endif
+    this->last_loop_ = last_op_end_time;
+    now = last_op_end_time;
+    this->after_component_phase_();
   }
 
 #ifdef USE_RUNTIME_STATS
-  uint32_t loop_tail_start_us = micros();
-#endif
-  this->after_loop_tasks_();
-
-#ifdef USE_RUNTIME_STATS
-  // Process any pending runtime stats printing after all components have run
-  // This ensures stats printing doesn't affect component timing measurements
+  // Record per-tick timing on every loop, not just component-phase ticks.
+  // record_loop_active is a small accumulator; process_pending_stats is an
+  // inline gate check that early-outs unless now >= next_log_time_.
   if (global_runtime_stats != nullptr) {
     uint32_t loop_now_us = micros();
     // Subtract scheduled-component time from the "before" bucket so it is
@@ -671,25 +711,40 @@ inline void ESPHOME_ALWAYS_INLINE Application::loop() {
     uint32_t loop_before_overhead_us = loop_before_wall_us > loop_before_scheduled_us
                                            ? loop_before_wall_us - static_cast<uint32_t>(loop_before_scheduled_us)
                                            : 0;
-    global_runtime_stats->record_loop_active(loop_now_us - loop_active_start_us, loop_before_overhead_us,
-                                             loop_now_us - loop_tail_start_us);
-    global_runtime_stats->process_pending_stats(last_op_end_time);
+    // tail_us is only defined when Phase B ran; 0 on Phase A-only ticks so the
+    // stats bucket keeps its "component-phase trailing overhead" meaning.
+    uint32_t loop_tail_us = do_component_phase ? (loop_now_us - loop_tail_start_us) : 0;
+    global_runtime_stats->record_loop_active(loop_now_us - loop_active_start_us, loop_before_overhead_us, loop_tail_us);
+    global_runtime_stats->process_pending_stats(now);
   }
 #endif
 
-  // Use the last component's end time instead of calling millis() again
+  // Compute sleep: bounded by time-until-next-component-phase and the
+  // scheduler's next deadline. When a scheduler timer fires it re-enters
+  // loop(), Phase A services it, and the component phase stays gated by
+  // loop_interval_. When a background producer calls wake_loop_threadsafe()
+  // it sets the wake_request flag and wakes select() / the task notification;
+  // the gate above sees the flag and runs Phase B too so the producer's
+  // component can drain its queued work without waiting up to loop_interval_.
+  //
+  // Re-read HighFrequencyLoopRequester::is_high_frequency() here instead of
+  // reusing the cached `high_frequency` captured above: a component calling
+  // HighFrequencyLoopRequester::start() from within its loop() would
+  // otherwise sit under the stale value and sleep for up to loop_interval_
+  // before the request took effect. That was fine pre-decoupling (the old
+  // main loop also called the function fresh at the sleep point) but now
+  // matters much more — loop_interval_ is a power-saving knob documented
+  // to accept multi-second values, so the stale path could add seconds of
+  // latency on an HF request. The call is a trivial atomic read.
   uint32_t delay_time = 0;
-  auto elapsed = last_op_end_time - this->last_loop_;
-  if (elapsed < this->loop_interval_ && !HighFrequencyLoopRequester::is_high_frequency()) {
-    delay_time = this->loop_interval_ - elapsed;
-    uint32_t next_schedule = this->scheduler.next_schedule_in(last_op_end_time).value_or(delay_time);
-    // next_schedule is max 0.5*delay_time
-    // otherwise interval=0 schedules result in constant looping with almost no sleep
-    next_schedule = std::max(next_schedule, delay_time / 2);
-    delay_time = std::min(next_schedule, delay_time);
+  if (!HighFrequencyLoopRequester::is_high_frequency()) {
+    const uint32_t elapsed_since_phase = now - this->last_loop_;
+    const uint32_t until_phase =
+        (elapsed_since_phase >= this->loop_interval_) ? 0 : (this->loop_interval_ - elapsed_since_phase);
+    const uint32_t until_sched = this->scheduler.next_schedule_in(now).value_or(until_phase);
+    delay_time = std::min(until_phase, until_sched);
   }
   this->yield_with_select_(delay_time);
-  this->last_loop_ = last_op_end_time;
 
   if (this->dump_config_at_ < this->components_.size()) {
     this->process_dump_config_();

esphome/core/defines.h

@@ -177,6 +177,7 @@
 #define USE_API_USER_DEFINED_ACTION_RESPONSES
 #define USE_API_USER_DEFINED_ACTION_RESPONSES_JSON
 #define API_MAX_SEND_QUEUE 8
+#define MAX_API_CONNECTIONS 6
 #define USE_MD5
 #define USE_SHA256
 #define USE_MQTT

esphome/core/helpers.h

@@ -1117,7 +1117,10 @@ inline size_t buf_append_str(char *buf, size_t size, size_t pos, const char *str
     return size;
   }
   size_t remaining = size - pos - 1;  // reserve space for null terminator
-  size_t len = strnlen(str, remaining);
+  size_t len = 0;
+  while (len < remaining && str[len] != '\0') {
+    len++;
+  }
   memcpy(buf + pos, str, len);
   pos += len;
   buf[pos] = '\0';

esphome/core/util.cpp

@@ -1,28 +1,14 @@
 #include "esphome/core/util.h"
-#include "esphome/core/defines.h"
 #include "esphome/core/application.h"
 #include "esphome/core/version.h"
 #include "esphome/core/log.h"
 
-#ifdef USE_API
-#include "esphome/components/api/api_server.h"
-#endif
-
 #ifdef USE_MQTT
 #include "esphome/components/mqtt/mqtt_client.h"
 #endif
 
 namespace esphome {
 
-bool api_is_connected() {
-#ifdef USE_API
-  if (api::global_api_server != nullptr) {
-    return api::global_api_server->is_connected();
-  }
-#endif
-  return false;
-}
-
 bool mqtt_is_connected() {
 #ifdef USE_MQTT
   if (mqtt::global_mqtt_client != nullptr) {

esphome/core/util.h

@@ -1,10 +1,28 @@
 #pragma once
 
 #include <string>
+
+#include "esphome/core/defines.h"
+#include "esphome/core/helpers.h"
+
+#ifdef USE_API
+#include "esphome/components/api/api_server.h"
+#endif
+
 namespace esphome {
 
-/// Return whether the node has at least one client connected to the native API
-bool api_is_connected();
+/// Return whether the node has at least one client connected to the native API.
+///
+/// Inline so that hot-path callers (e.g. component loop() ticks that check connectivity every
+/// iteration) can skip the call8/return pair. With USE_API disabled this trivially returns false
+/// and collapses at compile time.
+#ifdef USE_API
+ESPHOME_ALWAYS_INLINE inline bool api_is_connected() {
+  return api::global_api_server != nullptr && api::global_api_server->is_connected();
+}
+#else
+ESPHOME_ALWAYS_INLINE inline bool api_is_connected() { return false; }
+#endif
 
 /// Return whether the node has an active connection to an MQTT broker
 bool mqtt_is_connected();

esphome/core/wake.cpp

@@ -12,9 +12,22 @@
 
 namespace esphome {
 
+// === Wake-requested flag storage ===
+#ifdef ESPHOME_THREAD_MULTI_ATOMICS
+// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
+std::atomic<uint8_t> g_wake_requested{0};
+#else
+// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
+volatile uint8_t g_wake_requested = 0;
+#endif
+
 // === ESP32 / LibreTiny — IRAM_ATTR entry points ===
 #if defined(USE_ESP32) || defined(USE_LIBRETINY)
 void IRAM_ATTR wake_loop_isrsafe(BaseType_t *px_higher_priority_task_woken) {
+  // ISR-safe: set flag before notify so the wake is visible on the next gate
+  // check. wake_request_set() is just an aligned 8-bit store / atomic store
+  // and is safe from IRAM.
+  wake_request_set();
   esphome_main_task_notify_from_isr(px_higher_priority_task_woken);
 }
 void IRAM_ATTR wake_loop_any_context() { wake_main_task_any_context(); }
@@ -72,6 +85,9 @@ void wakeable_delay(uint32_t ms) {
 // === Host (UDP loopback socket) ===
 #ifdef USE_HOST
 void wake_loop_threadsafe() {
+  // Set flag before sending so the consumer's gate check on the next loop()
+  // entry observes the wake regardless of select() scheduling.
+  wake_request_set();
   if (App.wake_socket_fd_ >= 0) {
     const char dummy = 1;
     ::send(App.wake_socket_fd_, &dummy, 1, 0);

esphome/core/wake.h

@@ -7,6 +7,10 @@
 #include "esphome/core/defines.h"
 #include "esphome/core/hal.h"
 
+#ifdef ESPHOME_THREAD_MULTI_ATOMICS
+#include <atomic>
+#endif
+
 #if defined(USE_ESP32) || defined(USE_LIBRETINY)
 #include "esphome/core/main_task.h"
 #endif
@@ -25,12 +29,48 @@ namespace esphome {
 extern volatile bool g_main_loop_woke;
 #endif
 
+// === wake_request flag — signals Application::loop() that a producer queued
+// work for some component's loop() to drain (MQTT RX, USB RX, BLE event, etc.)
+// and the component phase should run this tick instead of being held off by
+// the loop_interval_ gate. Set by every wake_loop_* entry point; consumed
+// (via exchange-and-clear) at the gate in Application::loop(). ===
+//
+// std::atomic<uint8_t> rather than std::atomic<bool> because GCC on Xtensa
+// generates an indirect function call for atomic<bool> ops instead of inlining
+// them — same workaround applied in scheduler.h for the SchedulerItem::remove
+// flag. On non-atomic platforms a volatile uint8_t suffices: 8-bit aligned
+// loads/stores are atomic on every supported MCU, and the platform signal
+// that follows wake_request_set() (FreeRTOS task-notify, esp_schedule, socket
+// send) provides the cross-thread/cross-core memory barrier.
+#ifdef ESPHOME_THREAD_MULTI_ATOMICS
+// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
+extern std::atomic<uint8_t> g_wake_requested;
+
+__attribute__((always_inline)) inline void wake_request_set() { g_wake_requested.store(1, std::memory_order_release); }
+__attribute__((always_inline)) inline bool wake_request_take() {
+  return g_wake_requested.exchange(0, std::memory_order_acquire) != 0;
+}
+#else
+// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
+extern volatile uint8_t g_wake_requested;
+
+__attribute__((always_inline)) inline void wake_request_set() { g_wake_requested = 1; }
+__attribute__((always_inline)) inline bool wake_request_take() {
+  uint8_t v = g_wake_requested;
+  g_wake_requested = 0;
+  return v != 0;
+}
+#endif
+
 // === ESP32 / LibreTiny (FreeRTOS) ===
 #if defined(USE_ESP32) || defined(USE_LIBRETINY)
 
 /// Wake the main loop from any context (ISR or task).
 /// always_inline so callers placed in IRAM keep the whole wake path in IRAM.
 __attribute__((always_inline)) inline void wake_main_task_any_context() {
+  // Set the wake-requested flag BEFORE the task notification so the consumer
+  // (Application::loop() gate) is guaranteed to see it on its next gate check.
+  wake_request_set();
   if (in_isr_context()) {
     BaseType_t px_higher_priority_task_woken = pdFALSE;
     esphome_main_task_notify_from_isr(&px_higher_priority_task_woken);
@@ -50,7 +90,10 @@ __attribute__((always_inline)) inline void wake_main_task_any_context() {
 void wake_loop_isrsafe(BaseType_t *px_higher_priority_task_woken);
 void wake_loop_any_context();
 
-inline void wake_loop_threadsafe() { esphome_main_task_notify(); }
+inline void wake_loop_threadsafe() {
+  wake_request_set();
+  esphome_main_task_notify();
+}
 
 namespace internal {
 inline void wakeable_delay(uint32_t ms) {
@@ -67,6 +110,9 @@ inline void wakeable_delay(uint32_t ms) {
 
 /// Inline implementation — IRAM callers inline this directly.
 inline void ESPHOME_ALWAYS_INLINE wake_loop_impl() {
+  // Set the wake-requested flag BEFORE esp_schedule so the consumer is
+  // guaranteed to see it on its next gate check.
+  wake_request_set();
   g_main_loop_woke = true;
   esp_schedule();
 }
@@ -98,6 +144,9 @@ inline void wakeable_delay(uint32_t ms) {
 #elif defined(USE_RP2040)
 
 inline void wake_loop_any_context() {
+  // Set the wake-requested flag BEFORE the SEV so the consumer is guaranteed
+  // to see it on its next gate check.
+  wake_request_set();
   g_main_loop_woke = true;
   __sev();
 }

tests/component_tests/template/config/template_text_restore.yaml

@@ -0,0 +1,14 @@
+esphome:
+  name: test
+
+host:
+
+text:
+  - platform: template
+    name: "Test Text Restore"
+    id: test_text_restore
+    optimistic: true
+    max_length: 10
+    mode: text
+    initial_value: "hello"
+    restore_value: true

tests/component_tests/template/test_template_text.py

@@ -0,0 +1,44 @@
+"""Tests for the template text component."""
+
+from __future__ import annotations
+
+from collections.abc import Callable
+from pathlib import Path
+
+
+def test_template_text_saver_uses_placement_new_with_templated_subclass(
+    generate_main: Callable[[str | Path], str],
+    component_config_path: Callable[[str], Path],
+) -> None:
+    """Regression test for template text restore saver using placement new.
+
+    When ``restore_value: true``, the saver is its own Pvariable with
+    placement new: storage is sized for ``TextSaver<MAX_LENGTH>``, the
+    declared pointer stays at ``TemplateTextSaverBase *`` for polymorphism,
+    and the templated subclass constructor runs. A regression would either
+    reintroduce the heap ``new TextSaver<...>()`` expression or size the
+    storage for the base class and silently skip the subclass ctor.
+    """
+    main_cpp = generate_main(component_config_path("template_text_restore.yaml"))
+
+    # Storage is sized and aligned for the templated subclass.
+    assert "sizeof(template_::TextSaver<10>)" in main_cpp
+    assert "alignas(template_::TextSaver<10>)" in main_cpp
+    # Pointer declared as base type for polymorphism.
+    assert (
+        "static template_::TemplateTextSaverBase *const test_text_restore_value_saver"
+        in main_cpp
+    )
+    # Placement new runs the templated subclass constructor.
+    assert "new(test_text_restore_value_saver) template_::TextSaver<10>()" in main_cpp
+    # Base-class default ctor must NOT be used.
+    assert (
+        "new(test_text_restore_value_saver) template_::TemplateTextSaverBase()"
+        not in main_cpp
+    )
+    # No heap `new TextSaver<...>()` left over — the pre-fix pattern.
+    assert "new template_::TextSaver<" not in main_cpp
+    # Saver is wired into the text component.
+    assert (
+        "test_text_restore->set_value_saver(test_text_restore_value_saver)" in main_cpp
+    )

tests/integration/fixtures/external_components/wake_test_component/__init__.py

@@ -0,0 +1,19 @@
+import esphome.codegen as cg
+import esphome.config_validation as cv
+from esphome.const import CONF_ID
+
+CODEOWNERS = ["@esphome/tests"]
+
+wake_test_component_ns = cg.esphome_ns.namespace("wake_test_component")
+WakeTestComponent = wake_test_component_ns.class_("WakeTestComponent", cg.Component)
+
+CONFIG_SCHEMA = cv.Schema(
+    {
+        cv.GenerateID(): cv.declare_id(WakeTestComponent),
+    }
+).extend(cv.COMPONENT_SCHEMA)
+
+
+async def to_code(config):
+    var = cg.new_Pvariable(config[CONF_ID])
+    await cg.register_component(var, config)

tests/integration/fixtures/external_components/wake_test_component/wake_test_component.cpp

@@ -0,0 +1,19 @@
+#include "wake_test_component.h"
+#include "esphome/core/application.h"
+#include "esphome/core/log.h"
+#include <chrono>
+#include <thread>
+
+namespace esphome::wake_test_component {
+
+static const char *const TAG = "wake_test_component";
+
+void WakeTestComponent::start_async_wake() {
+  ESP_LOGI(TAG, "Spawning async wake thread (50ms delay)");
+  std::thread([] {
+    std::this_thread::sleep_for(std::chrono::milliseconds(50));
+    App.wake_loop_threadsafe();
+  }).detach();
+}
+
+}  // namespace esphome::wake_test_component

tests/integration/fixtures/external_components/wake_test_component/wake_test_component.h

@@ -0,0 +1,27 @@
+#pragma once
+
+#include "esphome/core/component.h"
+#include <atomic>
+
+namespace esphome::wake_test_component {
+
+class WakeTestComponent : public Component {
+ public:
+  void setup() override {}
+  void loop() override { this->loop_count_.fetch_add(1, std::memory_order_relaxed); }
+
+  int get_loop_count() const { return this->loop_count_.load(std::memory_order_relaxed); }
+
+  // Spawn a detached thread that sleeps briefly then calls
+  // App.wake_loop_threadsafe(). Used by the integration test to verify a
+  // cross-thread wake forces a component-phase iteration even when
+  // loop_interval_ has been raised high enough to gate it off otherwise.
+  void start_async_wake();
+
+  float get_setup_priority() const override { return setup_priority::DATA; }
+
+ protected:
+  std::atomic<int> loop_count_{0};
+};
+
+}  // namespace esphome::wake_test_component

tests/integration/fixtures/loop_interval_decoupling.yaml

@@ -0,0 +1,60 @@
+esphome:
+  name: loop-interval-decouple
+  on_boot:
+    priority: -100
+    then:
+      - lambda: |-
+          // Raise loop_interval_ to 500ms. With the decoupling fix the
+          // component phase should run ~twice per second while the 50ms
+          // scheduler interval below still fires at its requested cadence.
+          App.set_loop_interval(500);
+      # Start measurement after 1s so boot transients settle.
+      - delay: 1000ms
+      - lambda: |-
+          id(loop_at_start) = id(loop_counter)->get_loop_count();
+          id(sched_at_start) = id(sched_count);
+          ESP_LOGI("test", "MEASUREMENT_STARTED loop=%d sched=%d",
+                   id(loop_at_start), id(sched_at_start));
+      # Observe for 2s.
+      - delay: 2000ms
+      - lambda: |-
+          int loop_delta = id(loop_counter)->get_loop_count() - id(loop_at_start);
+          int sched_delta = id(sched_count) - id(sched_at_start);
+          ESP_LOGI("test", "MEASUREMENT_DONE loop_delta=%d sched_delta=%d",
+                   loop_delta, sched_delta);
+
+host:
+api:
+logger:
+  level: INFO
+  logs:
+    loop_test_component: WARN   # Silence per-loop log spam
+
+external_components:
+  - source:
+      type: local
+      path: EXTERNAL_COMPONENT_PATH
+
+globals:
+  - id: sched_count
+    type: int
+    initial_value: "0"
+  - id: loop_at_start
+    type: int
+    initial_value: "0"
+  - id: sched_at_start
+    type: int
+    initial_value: "0"
+
+loop_test_component:
+  components:
+    - id: loop_counter
+      name: loop_counter
+
+interval:
+  # Fast scheduler interval — with the decoupling fix this should fire at
+  # its requested 50ms cadence regardless of loop_interval_.
+  - interval: 50ms
+    then:
+      - lambda: |-
+          id(sched_count) += 1;

tests/integration/fixtures/loop_interval_default_not_pulled_forward.yaml

@@ -0,0 +1,51 @@
+esphome:
+  name: loop-default-not-pulled
+  on_boot:
+    priority: -100
+    then:
+      # Leave loop_interval_ at its default (16 ms → ~62 Hz). Do NOT call
+      # set_loop_interval here. The fast scheduler interval below used to
+      # pull the component phase forward to ~128 Hz via the old
+      # std::max(next_schedule, delay_time / 2) floor.
+      # Start measurement after 1s so boot transients settle.
+      - delay: 1000ms
+      - lambda: |-
+          id(loop_at_start) = id(loop_counter)->get_loop_count();
+          ESP_LOGI("test", "MEASUREMENT_STARTED loop=%d", id(loop_at_start));
+      # Observe for 2s.
+      - delay: 2000ms
+      - lambda: |-
+          int loop_delta = id(loop_counter)->get_loop_count() - id(loop_at_start);
+          ESP_LOGI("test", "MEASUREMENT_DONE loop_delta=%d", loop_delta);
+
+host:
+api:
+logger:
+  level: INFO
+  logs:
+    loop_test_component: WARN   # Silence per-loop log spam
+
+external_components:
+  - source:
+      type: local
+      path: EXTERNAL_COMPONENT_PATH
+
+globals:
+  - id: loop_at_start
+    type: int
+    initial_value: "0"
+
+loop_test_component:
+  components:
+    - id: loop_counter
+      name: loop_counter
+
+interval:
+  # Fast scheduler interval (well under loop_interval_/2 = 8ms). In the
+  # pre-decoupling code this would have pulled the component phase forward
+  # to ~128 Hz. After the decoupling fix the component phase stays at
+  # ~62 Hz regardless.
+  - interval: 5ms
+    then:
+      - lambda: |-
+          // No-op; the presence of a due scheduler item is what matters.

tests/integration/fixtures/wake_loop_forces_phase_b.yaml

@@ -0,0 +1,52 @@
+esphome:
+  name: wake-loop-phase-b
+  on_boot:
+    priority: -100
+    then:
+      - lambda: |-
+          // Raise loop_interval_ to 2000ms. Without the wake-request flag,
+          // a wake_loop_threadsafe() call would only run Phase A (scheduler)
+          // and leave the component phase gated for ~2s.
+          App.set_loop_interval(2000);
+      # Let boot transients settle.
+      - delay: 1000ms
+      - lambda: |-
+          // Snapshot the loop counter, then ask the component to spawn a
+          // background thread that calls App.wake_loop_threadsafe() after
+          // ~50ms. With the fix, that wake forces Phase B on the next tick
+          // and the counter increments well within the 500ms observation
+          // window below.
+          id(count_at_start) = id(wake_counter)->get_loop_count();
+          id(start_time) = millis();
+          id(wake_counter)->start_async_wake();
+          ESP_LOGI("test", "WAKE_STARTED count=%d", id(count_at_start));
+      # Observation window must be much shorter than loop_interval_ (2000ms)
+      # so a "false pass" isn't possible by simply waiting out the gate.
+      - delay: 500ms
+      - lambda: |-
+          int count_now = id(wake_counter)->get_loop_count();
+          int delta = count_now - id(count_at_start);
+          uint32_t elapsed = millis() - id(start_time);
+          ESP_LOGI("test", "WAKE_RESULT delta=%d elapsed=%u", delta, elapsed);
+
+host:
+api:
+logger:
+  level: INFO
+
+external_components:
+  - source:
+      type: local
+      path: EXTERNAL_COMPONENT_PATH
+    components: [wake_test_component]
+
+globals:
+  - id: count_at_start
+    type: int
+    initial_value: "0"
+  - id: start_time
+    type: uint32_t
+    initial_value: "0"
+
+wake_test_component:
+  id: wake_counter

tests/integration/test_loop_interval_decoupling.py

@@ -0,0 +1,75 @@
+"""Test that loop_interval_ no longer clamps scheduler cadence.
+
+Regression test for the decoupling of Application::loop() component-phase
+cadence from scheduler wake timing.
+
+Setup:
+- App.set_loop_interval(500) — raised for power-savings style cadence
+- Scheduler interval at 50ms — should fire at 50ms regardless of loop_interval_
+- Component loop (LoopTestComponent) — should run at 500ms cadence
+
+Before the decoupling fix the old `std::max(next_schedule, delay_time / 2)`
+floor clamped the sleep to ~250ms, so the 50ms scheduler only fired ~8 times
+per 2s (vs the ~40 expected). After the fix the scheduler fires close to its
+requested cadence while the component phase stays gated at loop_interval_.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import re
+
+import pytest
+
+from .types import APIClientConnectedFactory, RunCompiledFunction
+
+
+@pytest.mark.asyncio
+async def test_loop_interval_decoupling(
+    yaml_config: str,
+    run_compiled: RunCompiledFunction,
+    api_client_connected: APIClientConnectedFactory,
+) -> None:
+    """Raised loop_interval_ must not clamp scheduler item cadence."""
+    loop = asyncio.get_running_loop()
+    measurement_done: asyncio.Future[tuple[int, int]] = loop.create_future()
+
+    def on_log_line(line: str) -> None:
+        match = re.search(r"MEASUREMENT_DONE loop_delta=(\d+) sched_delta=(\d+)", line)
+        if match and not measurement_done.done():
+            measurement_done.set_result((int(match.group(1)), int(match.group(2))))
+
+    async with (
+        run_compiled(yaml_config, line_callback=on_log_line),
+        api_client_connected() as client,
+    ):
+        device_info = await client.device_info()
+        assert device_info is not None
+        assert device_info.name == "loop-interval-decouple"
+
+        try:
+            loop_delta, sched_delta = await asyncio.wait_for(
+                measurement_done, timeout=10.0
+            )
+        except TimeoutError:
+            pytest.fail("MEASUREMENT_DONE marker never appeared")
+
+        # Observation window = 2s, loop_interval_ = 500ms.
+        # Component phase should fire ~4 times in 2s. The upper bound must be
+        # less than 8: the pre-decoupling behavior clamped to ~250ms cadence
+        # giving ~8 loops/2s, so allowing 8 would let the old behavior pass.
+        # Lower bound 3 (not 2) keeps the test honest: a >30% slowdown from
+        # the ~4 nominal is not normal CI jitter and should fail.
+        assert 3 <= loop_delta <= 6, (
+            f"Component loop should fire ~4 times in 2s at loop_interval=500ms, "
+            f"got {loop_delta}"
+        )
+
+        # Scheduler interval = 50ms → ~40 fires in 2s. Before the decoupling
+        # fix this clamped to ~8 fires. Assert >= 20 to catch the old clamped
+        # behavior with comfortable jitter headroom for slow CI hosts.
+        assert sched_delta >= 20, (
+            f"50ms scheduler interval should fire ~40 times in 2s but only "
+            f"fired {sched_delta}. This indicates loop_interval_ is still "
+            f"clamping scheduler cadence."
+        )

tests/integration/test_loop_interval_default_not_pulled_forward.py

@@ -0,0 +1,67 @@
+"""Test that a fast scheduler item does not pull the component phase forward.
+
+Regression test for the original ~128 Hz → ~62 Hz bug fixed by decoupling
+Application::loop() component-phase cadence from scheduler wake timing.
+
+Setup:
+- loop_interval_ left at its default (16 ms → ~62 Hz component phase).
+- Scheduler interval at 5 ms (well under the old loop_interval_/2 = 8 ms floor).
+
+Before the decoupling fix the ``std::max(next_schedule, delay_time / 2)`` floor
+clamped the sleep to ~8 ms whenever any scheduler item was due sooner than
+loop_interval_/2. That pulled the component phase forward to ~128 Hz — twice
+what the documented ~62 Hz default promised. After the fix the component
+phase stays at ~62 Hz regardless of scheduler activity.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import re
+
+import pytest
+
+from .types import APIClientConnectedFactory, RunCompiledFunction
+
+
+@pytest.mark.asyncio
+async def test_loop_interval_default_not_pulled_forward(
+    yaml_config: str,
+    run_compiled: RunCompiledFunction,
+    api_client_connected: APIClientConnectedFactory,
+) -> None:
+    """Fast scheduler item must not pull component phase past default ~62 Hz."""
+    loop = asyncio.get_running_loop()
+    measurement_done: asyncio.Future[int] = loop.create_future()
+
+    def on_log_line(line: str) -> None:
+        match = re.search(r"MEASUREMENT_DONE loop_delta=(\d+)", line)
+        if match and not measurement_done.done():
+            measurement_done.set_result(int(match.group(1)))
+
+    async with (
+        run_compiled(yaml_config, line_callback=on_log_line),
+        api_client_connected() as client,
+    ):
+        device_info = await client.device_info()
+        assert device_info is not None
+        assert device_info.name == "loop-default-not-pulled"
+
+        try:
+            loop_delta = await asyncio.wait_for(measurement_done, timeout=10.0)
+        except TimeoutError:
+            pytest.fail("MEASUREMENT_DONE marker never appeared")
+
+        # Observation window = 2s, loop_interval_ default = 16ms → ~62 Hz →
+        # ~125 component-phase iterations expected.
+        # Pre-fix behavior: the 5 ms scheduler interval tripped the old
+        # delay_time/2 = 8 ms floor, pulling the phase to ~128 Hz → ~256.
+        # Upper bound 180 is comfortably below the ~256 pre-fix rate but
+        # above the ~125 nominal with CI jitter.
+        # Lower bound 80 covers very slow CI hosts without permitting a
+        # complete regression.
+        assert 80 <= loop_delta <= 180, (
+            f"Component loop at default loop_interval_ should fire ~125 times "
+            f"in 2s (≈62 Hz × 2s); got {loop_delta}. Values >200 indicate the "
+            f"scheduler is again pulling the component phase forward."
+        )

tests/integration/test_wake_loop_forces_phase_b.py

@@ -0,0 +1,76 @@
+"""Test that wake_loop_threadsafe() forces a component-phase iteration.
+
+Regression test for the wake-request flag added to Application::loop()'s
+Phase A / Phase B gate. Background producers (MQTT RX, USB RX, BLE event,
+etc.) call App.wake_loop_threadsafe() expecting their component's loop()
+to drain queued work; if the component phase stays gated by loop_interval_,
+the work waits up to loop_interval_ ms instead of running on the next tick.
+
+Setup:
+- App.set_loop_interval(2000) — a wide gate that would clearly mask the bug.
+- A test component spawns a detached std::thread that sleeps 50 ms and then
+  calls App.wake_loop_threadsafe() from a non-main thread.
+- The on_boot block snapshots the component's loop counter before/after a
+  500 ms observation window.
+
+Without the fix, delta=0 (the gate holds Phase B for ~2 s).
+With the fix, delta>=1 (the wake forces Phase B within one tick of the wake).
+"""
+
+from __future__ import annotations
+
+import asyncio
+from pathlib import Path
+import re
+
+import pytest
+
+from .types import APIClientConnectedFactory, RunCompiledFunction
+
+
+@pytest.mark.asyncio
+async def test_wake_loop_forces_phase_b(
+    yaml_config: str,
+    run_compiled: RunCompiledFunction,
+    api_client_connected: APIClientConnectedFactory,
+) -> None:
+    """A wake_loop_threadsafe() call from a background thread must trigger the
+    component phase within the next tick, even when loop_interval_ is raised
+    well above the observation window."""
+    external_components_path = str(
+        Path(__file__).parent / "fixtures" / "external_components"
+    )
+    yaml_config = yaml_config.replace(
+        "EXTERNAL_COMPONENT_PATH", external_components_path
+    )
+
+    loop = asyncio.get_running_loop()
+    result: asyncio.Future[tuple[int, int]] = loop.create_future()
+
+    def on_log_line(line: str) -> None:
+        match = re.search(r"WAKE_RESULT delta=(\d+) elapsed=(\d+)", line)
+        if match and not result.done():
+            result.set_result((int(match.group(1)), int(match.group(2))))
+
+    async with (
+        run_compiled(yaml_config, line_callback=on_log_line),
+        api_client_connected() as client,
+    ):
+        device_info = await client.device_info()
+        assert device_info is not None
+        assert device_info.name == "wake-loop-phase-b"
+
+        try:
+            delta, elapsed = await asyncio.wait_for(result, timeout=15.0)
+        except TimeoutError:
+            pytest.fail("WAKE_RESULT marker never appeared")
+
+        # Without the fix, delta would be 0 — loop_interval_=2000ms held
+        # Phase B off for the full 500ms observation window. With the fix
+        # the wake from the background thread (~50ms after start) forces
+        # Phase B on the next tick, so the counter increments at least once.
+        assert delta >= 1, (
+            f"wake_loop_threadsafe() from a background thread should force "
+            f"Phase B within the next tick; observed delta={delta} after "
+            f"{elapsed}ms with loop_interval_=2000ms"
+        )