Files
HaloKeymind/src/helpers/ESP32WsTransportFix.cpp
T
agessaman 88c824c175 docs(mqtt): correct what the WS buffer padding actually fixes
The comment claimed the (WS_BUFFER_SIZE + 1) padding makes an oversized
upgrade response "fail cleanly (Upgrade header not found)". Reading
release/v4.4 transport_ws.c against release/v5.3 shows it does not.

v4.4's response loop is

    } while (NULL == strstr(ws->buffer, "\r\n\r\n") && header_len < WS_BUFFER_SIZE);

so it also exits when the buffer fills without the terminator, and the code
then looks for "Sec-WebSocket-Accept:" and returns 0 if it is present. That
header appears early in a response, so an oversized header block yields a
BOGUS SUCCESS rather than a clean failure: the unread remainder stays queued
on the socket and is delivered as the first post-upgrade read, where the
deframer parses HTTP bytes as a WebSocket frame header.

Observed on hardware 2026-08-12 on a Heltec V4: a Cloudflare Page Shield CSP
report-uri header pushed the 101 response past the buffer, and the tail of
that header ("csp-reporting.cloudflare.com/cdn-cgi/script_monitor/report?")
reached the MQTT layer as payload, surfacing as
"Invalid MSG_TYPE response: 3" (0x35 = '5', high nibble 3).

The padding's real and only value is preventing the one-byte overflow of the
heap canary, which is still worth having. Narrow the comment to that claim and
record where the parser fix has to come from: IDF 5.2+ requires the "\r\n\r\n"
delimiter, memmoves the bytes following it, and fails cleanly when the buffer
fills. It cannot be patched here, since transport_ws.c ships precompiled in
libtcp_transport.a on Arduino 2.x.

No functional change.

(cherry picked from commit 9894e65e8ad961704d430a12a065baffda353f50)
2026-08-14 09:47:40 -07:00

142 lines
5.8 KiB
C++

#ifdef ESP_PLATFORM
// Link-time workaround for an off-by-one heap overflow in ESP-IDF v4.4's
// WebSocket transport (components/tcp_transport/transport_ws.c), which ships
// PRECOMPILED in the Arduino-ESP32 2.x SDK (libtcp_transport.a) and cannot be
// patched at source level.
//
// The bug (transport_ws.c, ws_connect() response-read loop):
//
// header_len += len;
// ws->buffer[header_len] = '\0'; // header_len can reach WS_BUFFER_SIZE
// } while (... && header_len < WS_BUFFER_SIZE);
//
// ws->buffer is malloc(WS_BUFFER_SIZE) (1024). When a wss:// endpoint answers
// the WebSocket upgrade with >= 1024 bytes of HTTP response before the blank
// line terminator (typical for a down/misconfigured broker behind a proxy or
// CDN that serves a large HTML error page), the final iteration writes one
// '\0' one byte past the block. With heap poisoning enabled that zeroes the
// LSB of the tail canary (0xbaad5678 -> 0xbaad5600); the corruption then sits
// silent until the block is freed — which happens in ws_destroy() during
// esp_mqtt_client_destroy(), i.e. MQTTBridge::end() — and the free asserts:
//
// CORRUPT HEAP: Bad tail at 0x.... Expected 0xbaad5678 got 0xbaad5600
// assert failed: multi_heap_free multi_heap_poisoning.c:259
//
// On observer builds that teardown runs at the start of the deferred
// `ota update`, so a single down wss broker made every online OTA panic and
// reboot before the download began (backtrace decoded from a Heltec V3 on
// v1.16.0.11: free <- ws_destroy <- esp_transport_list_destroy <-
// esp_mqtt_client_destroy <- ~PsychicMqttClient <- destroySlotClients <-
// MQTTBridge::end <- MyMesh::setBridgeState <- MyMesh::loop).
//
// Fix: [esp32_base] adds `-Wl,--wrap=esp_transport_ws_init`, so every
// creation of a WS transport (esp-mqtt does one per wss slot) is routed
// through __wrap_esp_transport_ws_init below, which replaces the freshly
// allocated 1024-byte buffer with a (WS_BUFFER_SIZE + 1)-byte one, so the
// out-of-bounds index WS_BUFFER_SIZE lands on our extra byte instead of the
// heap canary. Upstream fixed this in ESP-IDF 5.x, so this file compiles to a
// pass-through there and can be deleted (together with the --wrap flag) when
// the fork moves to Arduino core 3.x.
//
// Scope: this stops the heap corruption and NOTHING else. It does not make an
// oversized response fail the handshake — v4.4's read loop also exits on
// `header_len < WS_BUFFER_SIZE` going false, and then still accepts the
// connection if "Sec-WebSocket-Accept:" was inside those first bytes (it comes
// early, so it usually is). ws_connect() therefore returns success on a partial
// header block and the unread remainder arrives as the first "payload" read,
// where the deframer parses HTTP bytes as a frame header. Observed on hardware
// 2026-08-12: a large Cloudflare CSP header produced exactly that, surfacing as
// `Invalid MSG_TYPE response: 3`. Only IDF 5.2+ fixes it (it requires the
// "\r\n\r\n" delimiter, preserves the bytes after it, and fails cleanly when the
// buffer fills); it cannot be patched here because transport_ws.c is precompiled.
//
// transport_ws_t below is copied verbatim from ESP-IDF release/v4.4
// transport_ws.c (the struct is file-private, so it is not in any shipped
// header). Source fidelity was verified against the shipped binary: addr2line
// on the crash backtrace resolves to the exact line numbers of that file
// (e.g. free(ws->buffer) at transport_ws.c:546). Only the first two members
// (path, buffer) are dereferenced here.
#include "esp_idf_version.h"
#if ESP_IDF_VERSION_MAJOR == 4
#include <stdlib.h>
#include "sdkconfig.h"
#include "esp_transport.h"
#include "esp_transport_ws.h"
#ifndef CONFIG_WS_BUFFER_SIZE
#define CONFIG_WS_BUFFER_SIZE 1024
#endif
// --- copied from ESP-IDF release/v4.4 components/tcp_transport/transport_ws.c ---
typedef struct {
uint8_t opcode;
char mask_key[4];
int payload_len;
int bytes_remaining;
bool header_received;
} ws_transport_frame_state_t;
typedef struct {
char *path;
char *buffer;
char *sub_protocol;
char *user_agent;
char *headers;
bool propagate_control_frames;
ws_transport_frame_state_t frame_state;
esp_transport_handle_t parent;
} transport_ws_t;
// --------------------------------------------------------------------------------
extern "C" {
esp_transport_handle_t __real_esp_transport_ws_init(esp_transport_handle_t parent_handle);
esp_transport_handle_t __wrap_esp_transport_ws_init(esp_transport_handle_t parent_handle) {
esp_transport_handle_t t = __real_esp_transport_ws_init(parent_handle);
if (t != nullptr) {
transport_ws_t* ws = (transport_ws_t*)esp_transport_get_context_data(t);
if (ws != nullptr && ws->buffer != nullptr) {
// The buffer is untouched at this point (allocated moments ago inside
// __real_esp_transport_ws_init), so a swap is safe.
char* padded = (char*)malloc(CONFIG_WS_BUFFER_SIZE + 1);
if (padded != nullptr) {
free(ws->buffer);
ws->buffer = padded;
}
// On alloc failure keep the original buffer: same behavior as before
// this fix, which is still strictly better than failing init here.
}
}
return t;
}
} // extern "C"
#else // ESP_IDF_VERSION_MAJOR != 4
// IDF 5.x fixed the overflow upstream; keep a pass-through so the --wrap flag
// (set for all ESP32 envs in [esp32_base]) still links if anything references
// the symbol.
#include "esp_transport.h"
#include "esp_transport_ws.h"
extern "C" {
esp_transport_handle_t __real_esp_transport_ws_init(esp_transport_handle_t parent_handle);
esp_transport_handle_t __wrap_esp_transport_ws_init(esp_transport_handle_t parent_handle) {
return __real_esp_transport_ws_init(parent_handle);
}
} // extern "C"
#endif // ESP_IDF_VERSION_MAJOR
#endif // ESP_PLATFORM