Repository navigation
benchcase: the shipped image's display driver runs in the benchmark boot - #1149
Merged
Merged
Conversation
The benchmark boot measured the T14 with the display as the GOP left it, while images/system.toml runs tgl-display. That driver sets the display engine's watermarks, loads the DMC and allows DC6 (stages 1-3). So the energy rows and the scoreboard read a display engine the shipped system never leaves. The benchmark image now declares the driver with the shipped row's `drives` match, and the comparison with Linux is of the real system. The row receives no `display-output`. Its server is the compositor, which this image does not run, and an image that receives a port no `[boot] start` row serves is refused (build/build.rs). Without the port the driver programs the engine's power as it does in the shipped image, says its presents go unanswered, and keeps the scan-out. On every other bench the match finds no function and nothing starts. Every osbench row boots this image, so the T14's figures that depend on the display engine (the idle package, its residency and the wake rows) change by design. The next scoreboard reads that baseline. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Japabu
added a commit
that referenced
this pull request
Nov 1, 2026
The PC2 issue's APST figures came from the benchmark image. That image started no display driver until #1149, so they read the GOP's display. In PR #1131's r6, Linux's own back-to-back write_4k_fsync on the T14 ran 4.73-5.02 ms in all three runs, against 1.0-1.08 ms in r3. The slow durable-write mode is not ToyOS's alone. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Japabu
marked this pull request as ready for review
November 2, 2026 00:07
Japabu
enabled auto-merge
November 2, 2026 00:07
Japabu
added a commit
that referenced
this pull request
Nov 2, 2026
…se's display row is granted its window, and the PSR restart point is the durable brief Rebased onto main c81fc83. #1062 took syscall 160 and #1116 took 161-163, so SYS_DEVICE_FIRMWARE_WINDOW is 164, the first commit's message notwithstanding. The profile has 166 bins, and the guard names 164. #1149 runs tgl-display in benchcase "as images/system.toml runs it", so its row carries the same `firmware = ["igd-opregion"]` grant. The display track's PSR restart line now links the clean-room brief at its durable evidence commit. Only facts.md there is an implementer's input. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Japabu
added a commit
that referenced
this pull request
Nov 2, 2026
…se's display row is granted its window, and the PSR restart point is the durable brief Rebased onto main c81fc83. #1062 took syscall 160 and #1116 took 161-163, so SYS_DEVICE_FIRMWARE_WINDOW is 164. The profile has 166 bins, and the guard names 164. #1149 runs tgl-display in benchcase "as images/system.toml runs it", so its row carries the same `firmware = ["igd-opregion"]` grant. The display track's PSR restart line now links the clean-room brief at its durable evidence commit. Only facts.md there is an implementer's input. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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
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
tests/images/benchcase.toml, the image every osbench metal row boots, did not run the display driver the shipped image runs. On the T14 the energy rows and the scoreboard therefore measured the display as the GOP left it, whileimages/system.tomlrunstgl-display. That driver sets watermarks and SAGV, loads the DMC and allows DC6. So ToyOS-against-Linux idle compared Linux's real display driver with a display engine ToyOS's shipped system never leaves.Change
benchcase.tomldeclares[programs.tgl-display]with the shipped row'sdrives = ["pci:03:00:00/8086:9a49"].receives = ["display-output"]. That port's server is the compositor, which the benchmark image deliberately leaves out.build/build.rsrefuses an image that receives a port no[boot] startrow serves.system/tgl-display/src/main.rs,serve).Effect, by design
Every osbench row boots this image. The T14's figures that depend on the display engine change:
energy.idle.*, its package C-state residency, and the wake rows. The next scoreboard reads that new baseline. No A/B is part of this PR.Coordination
#1137 adds
firmware = ["igd-opregion"]to the shipped andmetalcaserows. Whichever of the two lands second also givesbenchcasethat line, so the three rows stay alike.Checks
cargo run -- --ci hostat7e97df75c: exit 0, 108 steps.cargo test --test toyos-build -- osbench_capability_groups_exit_0(QEMU, which boots this image): exit 0. The build accepts the manifest, and the image boots with the row, which starts nothing on QEMU's machine.cargo test --test toyos-build -- --metal --metal-readback <dir> osbench_energystaged the T14 energy image: exit 2, the stager's "machine not touched". The image carries thetgl-displayrow and its binary.Risks
🤖 Generated with Claude Code