Thanks for the PR! The progress bar looks good and the fundamentals check out: step numbering is correct on every screen (including the deferred show-words build), bars don't accumulate across state transitions since lv_obj_clean() runs before each rebuild, and the layout works on all board sizes. A few things to address before merge:

1. The flow can end at step 3 of 4 (pin_page.c:381)

When eFuse provisioning is declined by the user, fails, or pin_efuse_check() returns PIN_EFUSE_ERROR, the PIN is saved directly from the split screen (split_confirm_cb at pin_page.c:700-706, efuse_confirm_result at pin_page.c:796) and STATE_SETUP_SHOW_WORDS never appears. Declining the irreversible eFuse burn is a routine choice, so on those devices the bar promises a 4th step that never comes, on first setup and on every subsequent change-PIN flow. Either make the total dynamic based on whether show-words will run, or drop the show-words step from the bar entirely.

2. Redundant mode gate (pin_page.c:403)

The current_mode != PIN_PAGE_UNLOCK check is dead code: every state reachable in UNLOCK mode returns 0 from setup_step_for_state() (the unlock→setup transition at pin_page.c:261-262 only fires in CHANGE mode), so if (step > 0) alone decides visibility in all three modes. In CHANGE mode the unlock/delay screens pass the mode check anyway and only the step check suppresses the bar, so the comment "never shown during unlock" describes the wrong mechanism. Please drop the mode check and keep just the step check.

3. Step count can silently drift (pin_page.c:381)

PIN_SETUP_STEP_COUNT and the setup_step_for_state() switch have no compile-time link, and lv_bar silently clamps values above the max, so inserting a setup screen and forgetting the define would render a full bar on the last two screens with no build error. Suggest deriving the total from the switch itself, e.g. setup_step_for_state(STATE_SETUP_SHOW_WORDS), so there's one source of truth.

4. Minor: NULL-anchor fallback in theme_create_progress_bar (theme_widgets.c:222)

The anchor == NULL branch is currently unreachable (the only caller always passes the title), and if it were used, theme_default_padding() from the top lands inside the corner-button band that theme_create_page_title deliberately avoids. Fine to keep the tolerant parameter, but the fallback position should account for the button band, or just require a non-NULL anchor.

5. Question: change-PIN flow shows "n of 4" across 5 screens

In PIN_PAGE_CHANGE mode the user sees 5 screens (verify current PIN + the 4 setup screens) but the bar only counts the setup portion. Treating the verify screen as a gate rather than a step is a reasonable design, just confirming it's intentional.

Thank you😊 for the detailed review @odudex.

For 1 & 3 -- I'll compute the total step count dynamically based on eFuse availability instead of hardcoding 4. STATE_SETUP_SHOW_WORDS will either not count as a step, or the total will drop to 3 when that path is skipped.

For 2 -- I'll remove the PIN_PAGE_UNLOCK guard so the bar appears correctly in change-PIN mode too.

For 4 -- I'll require a non-NULL anchor (since the only caller always passes the title) and add an assertion to make that explicit.

For 5-- Yes, intentional , the verify screen in change-PIN mode is a gate, not a setup step. The bar only counts the steps where the user is constructing the new PIN.

✌️🙌🤞🤞✌️

@odudex , Please check the updated PR