fix(firmware): make STACKCHAN.RON writeback work via short-name overwrite

[andymai]

Summary

Completes the SD config feature: it could mount (#575) and read (#577) but not save. PUT /settings writeback was broken the same way the read path was — the staging file STACKCHAN.NEW (9-char base) couldn't be created by embedded-sdmmc, and the final copy targeted the un-creatable STACKCHAN.RON.

embedded-sdmmc cannot create long-named files at all, but it can overwrite an existing file opened by its short name. So:

• Stage to a valid 8.3 name (STKCFG.NEW) the firmware can create.
• Resolve the existing STACKCHAN.RON short name via the shared lfn_short_name helper (factored out of the read path).
• Truncate-write the rendered config onto it — which leaves the long-name directory entry intact, so the file stays readable as STACKCHAN.RON next boot.

Writeback updates an existing config (the realistic lifecycle — the file must already be present for Wi-Fi/auth to come up). If STACKCHAN.RON is absent, write_config returns FileNotFound with a clear log rather than silently writing a differently-named file.

Test plan

• just clippy-firmware (-D warnings) — clean.
• On-device (CoreS3, real STACKCHAN.RON on the card): a temporary idempotent self-test (rewrite the just-read config, then re-read) — removed before commit — logged:
  • dbg writeback: write_config OK
  • dbg writeback: re-read OK ssid=Interweb hostname=stackchan — round-trip intact, long name preserved.

Surfaced by config now loading (separate, not addressed here)

• sidecar session: persist failed (Write) — a different persisted file; needs its own look.
• wifi: start_async failed (InternalError(NoMem)) — Wi-Fi now actually starts (there's an SSID); a pre-existing init path that was never exercised while config never loaded.

Greptile: please review.

[greptile-apps]

Greptile Summary

This PR completes the SD config writeback feature by fixing the root cause that broke PUT /settings: embedded-sdmmc can't create files with long names, so the staging file STACKCHAN.NEW (9-char base) was unwritable. The fix stages to STKCFG.NEW (a valid 8.3 name), resolves STACKCHAN.RON's mangled short name via the factored-out lfn_short_name helper, and truncate-writes the new content in-place — preserving the LFN directory entry so the file stays accessible as STACKCHAN.RON on the next boot.

• lfn_short_name is extracted from the read path and reused by both read_config and the new write_config logic, reducing duplication.
• write_config now guards against the absent-file case with a clear defmt::warn log and returns FileNotFound rather than silently writing under a different name.
• Two doc comments still reference the old staging name STACKCHAN.NEW instead of the renamed STKCFG.NEW.

Confidence Score: 4/5

Safe to merge — the core logic is well-reasoned and on-device tested; only two stale doc-comment references to the old staging filename remain.

The writeback mechanism is sound: staging to an 8.3-creatable name, resolving the LFN short name, and truncate-writing in place correctly preserves the long-name directory entry. The lfn_short_name refactor is clean. The only issues found are two doc comments that still name STACKCHAN.NEW.

No files require special attention beyond the two stale doc comments in storage.rs.

Important Files Changed

Filename	Overview
crates/stackchan-firmware/src/storage.rs	Fixes STACKCHAN.RON writeback by staging to a valid 8.3 name (STKCFG.NEW) and overwriting the existing file via its resolved FAT short name; also extracts lfn_short_name as a shared helper. Two stale doc comments refer to the old STACKCHAN.NEW name.

Comments Outside Diff (2)

1. crates/stackchan-firmware/src/storage.rs, line 427-429 (link)

   The write_config doc comment still names STACKCHAN.NEW and describes the old copy-then-delete dance — both are out of date after this change. Readers following the doc will be confused about how writeback actually works now.

   Prompt To Fix With AI

   This is a comment left during a code review.
   Path: crates/stackchan-firmware/src/storage.rs
   Line: 427-429
   
   Comment:
   The `write_config` doc comment still names `STACKCHAN.NEW` and describes the old copy-then-delete dance — both are out of date after this change. Readers following the doc will be confused about how writeback actually works now.
   
   
   
   How can I resolve this? If you propose a fix, please make it concise.

   Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

   

2. crates/stackchan-firmware/src/storage.rs, line 126-127 (link)

   This comment still references STACKCHAN.NEW, the old staging name that was renamed to STKCFG.NEW in this PR. The reference is stale.

   Prompt To Fix With AI

   This is a comment left during a code review.
   Path: crates/stackchan-firmware/src/storage.rs
   Line: 126-127
   
   Comment:
   This comment still references `STACKCHAN.NEW`, the old staging name that was renamed to `STKCFG.NEW` in this PR. The reference is stale.
   
   
   
   How can I resolve this? If you propose a fix, please make it concise.

   Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

   

Prompt To Fix All With AI

Fix the following 2 code review issues. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 2
crates/stackchan-firmware/src/storage.rs:427-429
The `write_config` doc comment still names `STACKCHAN.NEW` and describes the old copy-then-delete dance — both are out of date after this change. Readers following the doc will be confused about how writeback actually works now.

```suggestion
    /// Atomically replace `/sd/STACKCHAN.RON` with the rendered RON
    /// of `config`. Stages to `STKCFG.NEW` first (an 8.3 name the
    /// firmware can create), then truncate-writes the rendered bytes
    /// onto the resolved short name of `STACKCHAN.RON`, and cleans
    /// up the staging file.
```

### Issue 2 of 2
crates/stackchan-firmware/src/storage.rs:126-127
This comment still references `STACKCHAN.NEW`, the old staging name that was renamed to `STKCFG.NEW` in this PR. The reference is stale.

```suggestion
/// Atomic-write staging name for the runtime store. Same dance as
/// `STKCFG.NEW`.
```

_{Reviews (1): Last reviewed commit: "fix(firmware): make STACKCHAN.RON writeb..." | Re-trigger Greptile}