kevin/meshtastic_firmware
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Fix heap blowout on TBeams (#10155)

Browse Source 
* Fix heap blowout on TBeams

* Update src/graphics/draw/MessageRenderer.cpp

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* Set MESSAGE_HISTORY_LIMIT to 10 for original ESP32 to optimize RAM usage

* Optimize message frame allocation to prevent excessive memory usage

* Refine message history limits for resource-constrained builds and cap cached lines to prevent heap overflow

* Update src/graphics/draw/MessageRenderer.cpp

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

---------

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
This commit is contained in:
Ben Meadors
2026-04-14 13:05:58 -05:00
co-authored by Copilot
parent a67eb15ad3
commit d24d8806e1
2 changed files with 27 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/graphics/draw/MessageRenderer.cpp
+20 -2
View File
@@ -422,6 +422,17 @@ void drawTextMessageFrame(OLEDDisplay *display, OLEDDisplayUiState *state, int16
std::vector<bool> isMine; // track alignment
std::vector<bool> isHeader; // track header lines
std::vector<AckStatus> ackForLine;
// Hard limit on total cached lines to prevent unbounded growth from a single long message.
// Reserve to the actual cache cap up front, because a single message can expand to many more
// wrapped display lines than a small per-message estimate would predict. For a display
// rendering only ~5-30 lines at a time, caching more than this limit wastes heap. Stop
// appending once we reach MAX_CACHED_LINES to prevent a single message from blowing out the
// heap.
constexpr size_t MAX_CACHED_LINES = 100U; // ~5-6KB for std::string overhead on 32-bit (if each ~50-60 bytes avg)
allLines.reserve(MAX_CACHED_LINES);
isMine.reserve(MAX_CACHED_LINES);
isHeader.reserve(MAX_CACHED_LINES);
ackForLine.reserve(MAX_CACHED_LINES);
for (auto it = filtered.rbegin(); it != filtered.rend(); ++it) {
const auto &m = *it;
@@ -565,16 +576,23 @@ void drawTextMessageFrame(OLEDDisplay *display, OLEDDisplayUiState *state, int16
int wrapWidth = mine ? rightTextWidth : leftTextWidth;
std::vector<std::string> wrapped = generateLines(display, "", msgText, wrapWidth);
// Per-message wrap-line limit: even if wrapping produces many lines, cap them to prevent
// a single long message from consuming most or all of the cache.
constexpr size_t MAX_WRAPPED_LINES_PER_MSG = 20U;
size_t wrappedCount = 0;
for (auto &ln : wrapped) {
allLines.push_back(ln);
if (allLines.size() >= MAX_CACHED_LINES || wrappedCount >= MAX_WRAPPED_LINES_PER_MSG)
break; // Cache limit or per-message limit reached; stop adding lines from this message
allLines.emplace_back(std::move(ln));
isMine.push_back(mine);
isHeader.push_back(false);
ackForLine.push_back(AckStatus::NONE);
++wrappedCount;
}
}
// Cache lines and heights
cachedLines = allLines;
cachedLines.swap(allLines);
cachedHeights = calculateLineHeights(cachedLines, emotes, isHeader);
std::vector<MessageBlock> blocks = buildMessageBlocks(isHeader, isMine);