Skip to content

Troubleshooting bambu lab slicer CLI for H2D two-material prints (PLA & TPU) #64

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
Copilot wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

Troubleshooting bambu lab slicer CLI for H2D two-material prints (PLA & TPU) #64

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9190e96
Initial plan
Copilot May 22, 2026
c9d905d
Add BambuStudio CLI PLA/TPU per-part extruder bug diagnosis
Copilot May 22, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
174 changes: 174 additions & 0 deletions manufacturing-notes/bambustudio-cli-multimaterial-bug.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,174 @@
# BambuStudio CLI multi-material PLA/TPU per-part extruder assignment bug

**Status:** investigation only — fix must be filed upstream against
[`bambulab/BambuStudio`](https://github.com/bambulab/BambuStudio); we cannot patch the
slicer from this repo.

**Why this lives here:** Phase-1 of the tensegrity-optimization prototype prints
co-extruded PLA struts + TPU 85A tendons on a multi-extruder Bambu printer
(`reviews/...`, `simulations/printable_design.py`). Any time we re-run the
slicer CLI with refreshed profiles (`--load-settings` + `--load-filaments`) the
job fails out with the bogus error reproduced below, which blocks the build
pipeline.

## Reported symptom

> After re-loading profiles with `--load-settings` / `--load-filaments`: filament
> TPU can not be printed on extruder 21842, under manual mode for multi extruder
> printer → CLI assigns a bogus extruder id to the per-part.

`21842` is not a real extruder index — only extruders `1..new_extruder_count`
exist. The value is an uninitialised / out-of-bounds read.

## Root cause (BambuStudio @ `e150b502`, `src/BambuStudio.cpp`)

The relevant block is the manual-mode printability validator at lines 6764–6796:

```cpp
// L6764
for (int index = 0; index < filament_maps.size(); index++) {
int filament_extruder = filament_maps[index];
if (unprintable_filament_ids[filament_extruder - 1].find(index + 1)
!= unprintable_filament_ids[filament_extruder - 1].end()) { ... }
}

// L6775
for (int f_index = 0; f_index < plate_filaments.size(); f_index++) {
for (int f_index = 0; f_index < plate_filaments.size(); f_index++) { // (1)
if (plate_filaments[f_index] <= filament_count) { // (2)
int filament_extruder = filament_maps[plate_filaments[f_index] - 1]; // (3)
std::string filament_type;
m_print_config.get_filament_type(filament_type, plate_filaments[f_index] - 1);
auto *filament_printable_status =
dynamic_cast<const ConfigOptionInts *>(m_print_config.option("filament_printable"));
if (filament_printable_status &&
(filament_printable_status->values.size() >= plate_filaments[f_index])) {
int status = filament_printable_status->values.at(plate_filaments[f_index] - 1);
if (!(status >> (filament_extruder - 1) & 1)) { // (4)
BOOST_LOG_TRIVIAL(error)
<< boost::format("plate %1% : filament %2% can not be printed on extruder %3%, ...")
% (index + 1) % filament_type % filament_extruder;
...
}
}
}
}
}
```

Three defects compound to produce the symptom:

1. **Shadowed inner loop (line 6775–6776).** Both loops declare `int f_index`;
the inner shadow makes the outer loop iterate exactly once (the inner runs
`plate_filaments.size()` times, then increments the outer `f_index` past the
end). Almost certainly a copy-paste regression. Cosmetic on its own, but it
masks the defect below by appearing to "cover" all filaments.

2. **Missing bounds check on `filament_maps`.** The guard at (2) only checks
`plate_filaments[f_index] <= filament_count`, not against
`filament_maps.size()`. After `--load-filaments` shrinks the active filament
set (e.g. original 3MF had 4 filaments, CLI re-loads 2: PLA + TPU), the
`filament_map` ConfigOptionInts in `m_extra_config` is only resized to the
new `filament_count` in some paths (line 6624–6664 only sanitises `-1`
sentinels; it does not enforce
`filament_maps.size() == filament_count`). Meanwhile `plate_filaments` is
built from per-volume `ModelVolume::config.option("extruder")` on the loaded
3MF — those still reference the *original* slot numbers (3, 4, …).
Indexing `filament_maps[plate_filaments[f_index] - 1]` is then an
out-of-bounds read on a `std::vector<int>`; `operator[]` returns
uninitialised memory → `filament_extruder == 21842`.

3. **No range check before the bitshift (4).** Even when `filament_maps` is
correctly sized, a stale per-part extruder id larger than `new_extruder_count`
feeds `status >> (filament_extruder - 1)` with a shift count ≥ width of `int`
→ implementation-defined behaviour, in practice always producing
"unprintable" and the same misleading error.

## Reproduction

```
bambu-studio-cli \
--load-settings <process.json> <machine_with_2_extruders.json> \
--load-filaments <pla.json> <tpu.json> \
--filament-map-mode "manual" \
--filament-map 1 2 \
--export-3mf out.3mf \
input.3mf # input was sliced with 4 filaments, TPU was slot 3
```

`input.3mf` carries `ModelVolume` configs with `extruder = 3` / `4`. The CLI
keeps those volume-level assignments but truncates `filament_map` to 2 slots,
triggering the OOB read.

## Proposed upstream patch

Three small, independent fixes in `src/BambuStudio.cpp` around line 6775:

```diff
- for (int f_index = 0; f_index < plate_filaments.size(); f_index++) {
- for (int f_index = 0; f_index < plate_filaments.size(); f_index++) {
- if (plate_filaments[f_index] <= filament_count) {
- int filament_extruder = filament_maps[plate_filaments[f_index] - 1];
+ for (int f_index = 0; f_index < (int)plate_filaments.size(); f_index++) {
+ const int fid = plate_filaments[f_index];
+ if (fid >= 1 && fid <= filament_count
+ && (size_t)fid <= filament_maps.size()) {
+ const int filament_extruder = filament_maps[fid - 1];
+ if (filament_extruder < 1 || filament_extruder > new_extruder_count) {
+ BOOST_LOG_TRIVIAL(error) << boost::format(
+ "plate %1% : filament %2% has invalid extruder id %3% "
+ "(expected 1..%4%); per-part extruder map was not rebuilt "
+ "after --load-filaments")
+ % (index + 1) % fid % filament_extruder % new_extruder_count;
+ record_exit_reson(outfile_dir, CLI_INVALID_PARAMS, index + 1,
+ cli_errors[CLI_INVALID_PARAMS], sliced_info);
+ flush_and_exit(CLI_INVALID_PARAMS);
+ }
std::string filament_type;
- m_print_config.get_filament_type(filament_type, plate_filaments[f_index] - 1);
+ m_print_config.get_filament_type(filament_type, fid - 1);
auto *filament_printable_status = dynamic_cast<const ConfigOptionInts *>(m_print_config.option("filament_printable"));
- if (filament_printable_status && (filament_printable_status->values.size() >= plate_filaments[f_index])) {
- int status = filament_printable_status->values.at(plate_filaments[f_index] - 1);
+ if (filament_printable_status
+ && (filament_printable_status->values.size() >= (size_t)fid)) {
+ const int status = filament_printable_status->values.at(fid - 1);
if (!(status >> (filament_extruder - 1) & 1)) {
...
}
}
}
- }
}
```

Independently, the `filament_map` re-sizing branch at L6624–6664 should grow
**and shrink** `filament_maps` to exactly `filament_count` entries after
`--load-filaments`, defaulting any new slots to `1` and dropping the tail.

## Workaround for our pipeline

Until upstream merges a fix, our print scripts should either:

1. Pass `--filament-map` explicitly with exactly `filament_count` entries that
match the post-`--load-filaments` slot order, and confirm via log that the
sanitiser at L6644 prints `filament map default_value 1` (or the intended
extruder) for every slot; **or**
2. Re-author the input 3MF so every `ModelVolume` `extruder` is `1` or `2`
before CLI invocation (e.g. with a small `lib3mf` pre-pass that clamps every
per-volume `extruder` config to `min(extruder, filament_count)`); **or**
3. Avoid `--load-filaments`/`--load-settings` and let the 3MF embedded profiles
drive the slice.

Option (2) is the most reliable for the BO loop because option (1) still trips
the OOB read whenever Ax proposes a design whose pre-rendered 3MF references
more filament slots than the active CLI filament set.

## Upstream issue draft

Title: `CLI: out-of-bounds read on filament_maps after --load-filaments yields
"can not be printed on extruder <garbage>" under manual multi-extruder mode`

Body: paste sections "Reported symptom", "Reproduction", "Root cause", and
"Proposed upstream patch" above. Reference `src/BambuStudio.cpp` line ranges
relative to commit `e150b502b3d2afc98b83dcc9e5720e998f9eb79a`.