composefs: Type 1 kernel path on the ESP is too long for HP G10 firmware, boot hangs
維護者通常 2 天內回覆
還沒有人認領這個 Issue。
評估
- 難度
- 4/5
- 預估耗時
- 1-2 天
- 新手友好度
- 49/100
- Issue 類型
- 缺陷
- 描述清晰度
- 基本清楚
- 活躍度
- 活躍
- 技術堆疊
- linux, rust
研究方向
Start by tracing how the composefs backend creates the Type 1 kernel directory and loader entry; the issue does not name source files or tests. The evidence points to the ESP path length as the cause: shortening the directory lets the same kernel boot on affected HP G10 firmware. Done means choosing and implementing a path that avoids the reported hang, then validating it on the affected hardware.
由索引模型根據 Issue 內容生成。
描述
What happened?
On several HP G10 business laptops, firmware hangs (black screen, hard lock) when systemd-boot loads the kernel from the Type 1 directory that the composefs backend writes to the ESP. Every stock install produces a path like this:
/EFI/Linux/bootc_composefs-<128 hex verity>/vmlinuz
That is 163 characters. Reporters in projectbluefin/dakota#1648 bisected the limit by truncating the directory name by hand: a 160 character kernel path boots, 163 hangs. The kernel, initrd and kernel arguments were unchanged, only the directory was renamed.
Affected so far:
- HP EliteBook 845 G10 (AMD, BIOS V82)
- HP EliteBook 840 G10 (Intel graphics)
- HP ZBook Firefly 14 G10 A
We ship Dakota with --composefs-backend and systemd-boot by default, which is probably why this has not shown up elsewhere.
Loader entries
Original entry generated by bootc (hangs):
title Bluefin
version 20260921
linux /EFI/Linux/bootc_composefs-023cf0e317ee0f30a2f8229ed4734a136a406ed38e46abc7406948c732f23071a7b13ca652775b32d540c8b7eaba8f540fb0f479f39527fd2d2ef05094d2e050/vmlinuz
initrd /EFI/Linux/bootc_composefs-023cf0e317ee0f30a2f8229ed4734a136a406ed38e46abc7406948c732f23071a7b13ca652775b32d540c8b7eaba8f540fb0f479f39527fd2d2ef05094d2e050/initrd
options root=LABEL=root rw composefs=023cf0e317ee0f30a2f8229ed4734a136a406ed38e46abc7406948c732f23071a7b13ca652775b32d540c8b7eaba8f540fb0f479f39527fd2d2ef05094d2e050 <image kargs> rd.luks.name=<uuid>=root
sort-key bootc-bluefin-dakota-0
Longest entry that boots, with the last three hex characters of the directory removed (160 characters):
linux /EFI/Linux/bootc_composefs-023cf0e317ee0f30a2f8229ed4734a136a406ed38e46abc7406948c732f23071a7b13ca652775b32d540c8b7eaba8f540fb0f479f39527fd2d2ef05094d2e/vmlinuz
initrd /EFI/Linux/bootc_composefs-023cf0e317ee0f30a2f8229ed4734a136a406ed38e46abc7406948c732f23071a7b13ca652775b32d540c8b7eaba8f540fb0f479f39527fd2d2ef05094d2e/initrd
A fully short directory (/EFI/Linux/bootc_composefs-dakota/) also boots.
I don't have one of these machines myself, but I can build a Dakota image with a candidate fix and ask the reporters in #1648 to verify it on the affected hardware.
Versions
- bootc v1.17.1
- systemd-boot, Type 1 BLS entries, composefs backend
- Image:
ghcr.io/projectbluefin/dakota:stable
- 主要語言
- Rust
- 星號
- 2.3k
- 分支
- 230
- 平均合併
- 2 天 11 小時
- 30 天內合併 PR
- 42
環境準備
從這裡開始
- 先讀完整個 Issue,再讀專案的貢獻指南。
- 在 Issue 下留言說明你要接手 —— 這能避免兩個人做同樣的事。
- Fork 儲存庫,在一個分支上完成修改。
- 送出 Pull Request,並在描述裡引用這個 Issue 編號。
bootc-dev/bootc 的其他 Issue
-
難度 2/5 1-3 小時 新手友好度 76/100
維護者通常 2 天內回覆
-
難度 2/5 1-3 小時 新手友好度 68/100
維護者通常 2 天內回覆
-
install: improve error message when systemd-boot is selected without composefs backend可能重新可做 關聯的 PR 已關閉且未合併。 未關閉
難度 2/5 1-3 小時 新手友好度 76/100
維護者通常 2 天內回覆
-
難度 4/5 3-5 天 新手友好度 55/100
維護者通常 2 天內回覆
-
難度 5/5 一週以上 新手友好度 30/100
維護者通常 2 天內回覆
相似的 Issue
-
[Bug]: layer.move to the place a layer already has fails and stops a script; with several layers it leaves an empty undo step可能已有人在做 @jgregoriorincon 今天認領。 未關閉
難度 2/5 1-3 小時 新手友好度 78/100
opengeos/RustGIS#2621 · 1 則留言 ·
維護者通常 1 天內回覆
-
good first issue web
難度 1/5 1 小時以內 新手友好度 92/100
維護者通常 1 天內回覆
-
難度 2/5 1-3 小時 新手友好度 70/100
-
Wait for the IDS correction dialog to settle in the stale-audit test可能已有人在做 @louistrue 今天認領。 未關閉ready
難度 2/5 1-3 小時 新手友好度 68/100
維護者通常 1 天內回覆
-
難度 2/5 1-3 小時 新手友好度 70/100
維護者通常 1 天內回覆