fix(zaparoo): shut down frontend before FPGA reconfig to stop atari launch freeze#13
Merged
Merged
Conversation
…freeze Launching a core from the launcher frontend could hard-freeze the device (silent black screen, reboot required) — reliably reproduced on atari5200 /atari800. The frontend process stays alive holding mmaps into /dev/fb0, the HPS framebuffer the FPGA scans out over the f2sdram bridge. fpga_load_rbf calls do_bridge(0) to reset that bridge while the frontend is still mapped to it, deadlocking the AXI bus mid-reconfiguration. Upstream has no frontend, so it never hits this. Move the frontend teardown ahead of reconfiguration: call alt_launcher_shutdown() at the top of fpga_load_rbf so the frontend (and its framebuffer) is gone before do_bridge(0)/socfpga_load. The call is idempotent, so app_restart's existing call is left untouched. Also harden the frontend VT takeover: replace the unbounded video_chvt() (VT_WAITACTIVE blocks forever if the frontend stalls bringing up video) with a bounded VT_ACTIVATE + VT_GETSTATE poll, and defer the takeover until the FPGA is ready. This fixes a separate poll-cothread wedge during menu+frontend bring-up after a 15kHz core.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
📝 WalkthroughWalkthroughLauncher VT switching now waits for FPGA readiness and uses a bounded activation loop. FPGA loading now shuts down the launcher before OSD disable and bridge reset. ChangesLauncher and FPGA reconfiguration timing
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@support/zaparoo/alt_launcher.cpp`:
- Around line 485-490: The activation wait loop in switch_to_launcher_vt is
blocking the poll cothread by using usleep inside a fixed retry loop. Replace
the busy-sleep with a cooperative yield to the libco scheduler (for example via
scheduler_yield() in the alt_launcher_poll/finalize_spawn path) and keep the
bounded retry behavior using a wall-clock deadline or equivalent non-blocking
timeout logic.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: f03b4a06-1ec3-4cc1-a9c4-3dee46076d64
📒 Files selected for processing (2)
fpga_io.cppsupport/zaparoo/alt_launcher.cpp
switch_to_launcher_vt() ran in the poll cothread. usleep() in libco's cooperative single-thread model sleeps the whole process, so the bounded VT wait stalled both cothreads (no UI/OSD, no poll) for up to 500ms. Replace the fixed usleep retry loop with a wall-clock deadline (GetTimer/CheckTimer) plus scheduler_yield(), keeping the same bound while staying cooperative.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
Launching a core from the launcher frontend could hard-freeze the device — silent black screen, serial output dead, reboot required. Reliably reproduced on atari5200/atari800; the identical launch works on stock MiSTer.
Root cause (traced)
The freeze is silent, right after
Bitstream size: N bytes, with no further serial output:Every loop in the path after that print (
fpgamgr_program_write,fpgamgr_program_poll_*) is bounded and prints on error — there is no software infinite loop and no silent failure path. A totally silent, reboot-requiring freeze is therefore a hardware AXI bus deadlock during FPGA reconfiguration.The only fork-unique factor: the frontend process is still alive, holding mmaps into
/dev/fb0— the HPS framebuffer the FPGA scans out over the same f2sdram bridge thatfpga_load_rbfresets viado_bridge(0). The fork only killed the frontend afterward, inapp_restart. So the live frontend + active framebuffer overlap the bridge teardown → bus deadlock. Upstream has no frontend, so it never hits this. The atari bitstream just reproduces the timing window reliably.Fix
fpga_io.cpp: callalt_launcher_shutdown()at the top offpga_load_rbf, so the frontend (and its framebuffer) is torn down beforedo_bridge(0)/socfpga_load. The call is idempotent, soapp_restart's existing call is left untouched.support/zaparoo/alt_launcher.cpp: harden the VT takeover — replace the unboundedvideo_chvt()(itsVT_WAITACTIVEblocks forever if the frontend stalls bringing up video) with a boundedVT_ACTIVATE+VT_GETSTATEpoll, and defer the takeover until the FPGA is ready. Fixes a separate poll-cothread wedge during menu+frontend bring-up after a 15kHz core.Both changes are surgical and fork-friendly (one added line in the upstream file plus fork-owned code) — minimal merge surface, no upstream reformatting, no Makefile change.
Testing
Built via
./docker-build.shand deployed to hardware:Summary by CodeRabbit