github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 00:51:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[PR #14919] [MERGED] Fix mlxrunner subprocess deadlocks #20189

New Issue
Closed
opened 2026-04-16 07:29:49 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-16 07:29:49 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/14919
Author: @jessegross
Created: 3/18/2026
Status: ✅ Merged
Merged: 3/20/2026
Merged by: @jessegross

Base: main ← Head: jessegross/mlx-log

📝 Commits (2)

  • 0fdfdcf mlx: fix subprocess log deadlock
  • 011aa26 llm, mlxrunner: fix done channel value consumed by first receiver

📊 Changes

2 files changed (+79 additions, -34 deletions)

View changed files

📝 llm/server.go (+8 -6)
📝 x/mlxrunner/client.go (+71 -28)

📄 Description

  • Deadlock on long stderr lines: The stderr reader used bufio.Scanner which stops at 64KB lines. If the subprocess exceeded this, the OS pipe buffer filled and the subprocess deadlocked. Replaced with an io.Copy-based statusWriter that streams stderr without buffering constraints.
  • Done channel consumed by first receiver: The done chan error only delivers its value once, so whichever of WaitUntilRunning, HasExited, or Close reads it first consumes the signal. This caused HasExited to return false after already returning true, and Close to block for 5s on an already-dead process. Replaced with a closed chan struct{} (receivable any number of times) plus a separate doneErr field. Applied the same fix to llm/server.go.

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/ollama/ollama/pull/14919 **Author:** [@jessegross](https://github.com/jessegross) **Created:** 3/18/2026 **Status:** ✅ Merged **Merged:** 3/20/2026 **Merged by:** [@jessegross](https://github.com/jessegross) **Base:** `main` ← **Head:** `jessegross/mlx-log` --- ### 📝 Commits (2) - [`0fdfdcf`](https://github.com/ollama/ollama/commit/0fdfdcf35e5e53e20300cb73cb3eb168eed807ca) mlx: fix subprocess log deadlock - [`011aa26`](https://github.com/ollama/ollama/commit/011aa263fa62e93cf28a7e03c6ae4b1258d0440a) llm, mlxrunner: fix done channel value consumed by first receiver ### 📊 Changes **2 files changed** (+79 additions, -34 deletions) <details> <summary>View changed files</summary> 📝 `llm/server.go` (+8 -6) 📝 `x/mlxrunner/client.go` (+71 -28) </details> ### 📄 Description - Deadlock on long stderr lines: The stderr reader used bufio.Scanner which stops at 64KB lines. If the subprocess exceeded this, the OS pipe buffer filled and the subprocess deadlocked. Replaced with an io.Copy-based statusWriter that streams stderr without buffering constraints. - Done channel consumed by first receiver: The done chan error only delivers its value once, so whichever of WaitUntilRunning, HasExited, or Close reads it first consumes the signal. This caused HasExited to return false after already returning true, and Close to block for 5s on an already-dead process. Replaced with a closed chan struct{} (receivable any number of times) plus a separate doneErr field. Applied the same fix to llm/server.go. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
GiteaMirror added the pull-request label 2026-04-16 07:29:49 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#20189
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?