github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-05 23:53:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[PR #7139] [MERGED] llama: Decouple patching script from submodule #43572

New Issue
Closed
opened 2026-04-24 23:10:52 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-24 23:10:52 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/7139
Author: @dhiltgen
Created: 10/8/2024
Status: ✅ Merged
Merged: 10/17/2024
Merged by: @dhiltgen

Base: main ← Head: go_server_patching

📝 Commits (2)

  • 86b43f6 Refine llama.cpp vendoring workflow tools
  • 6b74155 Run new make sync and patch flow

📊 Changes

16 files changed (+374 additions, -181 deletions)

View changed files

📝 .github/workflows/test.yaml (+1 -1)
📝 .gitignore (+2 -1)
📝 llama/Makefile (+3 -0)
📝 llama/README.md (+77 -3)
📝 llama/build-info.cpp (+1 -27)
➕ llama/make/Makefile.sync (+190 -0)
📝 llama/patches/0001-cuda.patch (+11 -0)
📝 llama/patches/0002-pretokenizer.patch (+9 -0)
📝 llama/patches/0003-metal.patch (+9 -0)
📝 llama/patches/0004-ggml-metal.patch (+9 -0)
📝 llama/patches/0005-embeddings.patch (+16 -8)
📝 llama/patches/0006-clip-unicode.patch (+9 -0)
📝 llama/patches/0007-solar-pro.patch (+17 -1)
📝 llama/patches/0008-conditional-fattn.patch (+11 -2)
📝 llama/patches/0009-blas.patch (+9 -0)
➖ llama/sync.sh (+0 -138)

📄 Description

The current patching script relies on the existing llama.cpp submodule, which we plan to remove in favor of the vendored code soon, so this updates the patching flow to no longer rely on that. It still uses the git commit from the submodule to ensure we don't drift while the submodule is still present. In the commit that removes the submodule in the future, this would be switched to a manifest file recording the git hash we're sync'd to.

This adjusts our patches to be mailbox format patches which can be round-tripped through the system deterministically. In doing so, this should improve developer experience when updating the llama.cpp commit, or working on fixes/features that impact the native code.

Fixes #2534

🔄 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/7139 **Author:** [@dhiltgen](https://github.com/dhiltgen) **Created:** 10/8/2024 **Status:** ✅ Merged **Merged:** 10/17/2024 **Merged by:** [@dhiltgen](https://github.com/dhiltgen) **Base:** `main` ← **Head:** `go_server_patching` --- ### 📝 Commits (2) - [`86b43f6`](https://github.com/ollama/ollama/commit/86b43f6f95c326f56359578aecce00ff81cb898a) Refine llama.cpp vendoring workflow tools - [`6b74155`](https://github.com/ollama/ollama/commit/6b74155896a25b734a34bd9b2035a154eeef1f7b) Run new make sync and patch flow ### 📊 Changes **16 files changed** (+374 additions, -181 deletions) <details> <summary>View changed files</summary> 📝 `.github/workflows/test.yaml` (+1 -1) 📝 `.gitignore` (+2 -1) 📝 `llama/Makefile` (+3 -0) 📝 `llama/README.md` (+77 -3) 📝 `llama/build-info.cpp` (+1 -27) ➕ `llama/make/Makefile.sync` (+190 -0) 📝 `llama/patches/0001-cuda.patch` (+11 -0) 📝 `llama/patches/0002-pretokenizer.patch` (+9 -0) 📝 `llama/patches/0003-metal.patch` (+9 -0) 📝 `llama/patches/0004-ggml-metal.patch` (+9 -0) 📝 `llama/patches/0005-embeddings.patch` (+16 -8) 📝 `llama/patches/0006-clip-unicode.patch` (+9 -0) 📝 `llama/patches/0007-solar-pro.patch` (+17 -1) 📝 `llama/patches/0008-conditional-fattn.patch` (+11 -2) 📝 `llama/patches/0009-blas.patch` (+9 -0) ➖ `llama/sync.sh` (+0 -138) </details> ### 📄 Description The current patching script relies on the existing llama.cpp submodule, which we plan to remove in favor of the vendored code soon, so this updates the patching flow to no longer rely on that. It still uses the git commit from the submodule to ensure we don't drift while the submodule is still present. In the commit that removes the submodule in the future, this would be switched to a manifest file recording the git hash we're sync'd to. This adjusts our patches to be mailbox format patches which can be round-tripped through the system deterministically. In doing so, this should improve developer experience when updating the llama.cpp commit, or working on fixes/features that impact the native code. Fixes #2534 --- <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 2026-04-24 23:10:52 -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#43572
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?