Repository navigation
Rebuild libvgpu with the timeslice lock fixes - #24
Merged
Merged
Conversation
Both binaries come from HAMi-core 4c9b2d0dee9ee9fec6fcf2ac07782ac62e3ebbc7 (v0.1.3-43-g4c9b2d0), which carries the client half of the timeslice lock repair -- a drop is now answered with `released` even when the lock had already been let go early -- on top of the upstream sync to Project-HAMi f01e9f2. Nothing records which HAMi-core commit a bundled binary was built from; the build embeds no revision string, so this message is the only provenance. Worth fixing separately. Co-authored-by: Cursor <cursoragent@cursor.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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Refreshes both bundled
libvgpu.sobinaries from HAMi-core4c9b2d0(v0.1.3-43-g4c9b2d0, Above-Os/HAMi-core#16).What the client half of that change does: a
drop:from the scheduler is now answered withreleasedeven when the lock had already been let go early. Previously the handler stayed silent wheneverown_lockwas 0, which is also the state left behind by the idle timer. If the scheduler had not recorded that earlier release it still considered us the owner — precisely why it was dropping a lock we no longer thought we held — so the two sides stayed permanently out of sync and the waiter blocked forever.The binaries also carry the sync to Project-HAMi/HAMi-core
f01e9f2, so this is a larger delta than the lock fix alone.Both architectures are rebuilt:
lib/nvidia/libvgpu.solib/nvidia/libvgpu.arm64.soThe arm64 one needed a fix to HAMi-core's
build-arm-in-dockertarget first: its prebuilt image ships no git whileDOCKER_BUILD_CMDstarts withgit config, so that target had been failing and the arm64 binary could only ever lag behind.One gap worth flagging separately: nothing records which HAMi-core commit a bundled binary came from. The build embeds no revision string, so the commit message is the only provenance we have.