| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
docs: restructure docs, consolidate intro into README, add developer quick start - Replace long README with concise overview: use cases, features, quick start, supported distros, main commands table, installation layout, how-it-works (run namespaces, install stages). - Add docs/index.md as doc index - Add docs/user-guide/developer-quick-start.md: build deps, standard/static build, dev loop, try a channel, testing (unit, solver, lua, e2e). - Add docs/user-guide/: getting-started, environments, package-operations, advanced. Align examples with current epkg output (e.g. DEPTH/SIZE/PACKAGE, env list columns). Add dependency resolution note, best practices. - Add docs/reference/: commands, repositories, paths. Document PATHS from epkg help; clarify cache vs store in paths.md. Signed-off-by: Wu Fengguang <wfg@mail.ustc.edu.cn> | 5 个月前 | |
docs: update developer-quick-start.md - Add supported platforms and host OS section - Add Windows build instructions (cross-windows) - Add Windows debugging env vars (WSLENV, EPKG_DEBUG_LIBKRUN, RUST_LOG) - Add PowerShell debugging example - Add release builds from Debian section Signed-off-by: Wu Fengguang <wfg@mail.ustc.edu.cn> | 3 个月前 | |
docs: add ebin exposure documentation Add comprehensive documentation for ebin exposure mechanism: - zh/user-guide/ebin-exposure.md: Usage concepts and scenarios - zh/architecture/ebin-exposure.md: Design details and implementation - zh/troubleshooting/ebin-exposure.md: Debug commands and common issues | 4 个月前 | |
docs: add ELF Loader documentation (3 perspectives) Add comprehensive documentation for ELF Loader mechanism: - user-guide/elf-loader.md: Usage concepts and scenarios - architecture/elf-loader.md: Design details and implementation - troubleshooting/elf-loader.md: Debug commands and common issues Structure follows the same pattern as ebin-exposure docs. | 4 个月前 | |
epkg run --sandbox: rename param to --isolate Rename SandboxMode to IsolateMode too. Signed-off-by: Wu Fengguang <wfg@mail.ustc.edu.cn> | 4 个月前 | |
docs: sync getting-started.md with README quick start changes Updates: - Install: curl|bash with platform labels (Linux/macOS/WSL2), add Windows PowerShell section - Example env name: myenv -> alpine for consistency with README - Run command: add --isolate=env|fs|vm examples showing three isolation modes Signed-off-by: Wu Fengguang <wfg@mail.ustc.edu.cn> | 3 个月前 | |
docs: add user notes about search vs list output stability - Add note to epkg search -h output about parallel optimization - Update package-operations.md (EN and ZH) with search behavior explanation - Clarify that search output may vary between runs while list is sorted Signed-off-by: Wu Fengguang <wfg@mail.ustc.edu.cn> | 5 个月前 | |
epkg run --sandbox: rename param to --isolate Rename SandboxMode to IsolateMode too. Signed-off-by: Wu Fengguang <wfg@mail.ustc.edu.cn> | 4 个月前 | |
libkrun: fix KVM resource leak by explicitly calling krun_free_ctx before exit Problem: libkrun randomly fails with "Out of memory (os error 12)" when running VMs in rapid succession without delay (e.g., "sleep 3" between runs). Root Cause Analysis: - std::process::exit() does NOT call Drop on local variables - KrunContext's Drop implementation calls krun_free_ctx() to clean up KVM resources - Because Drop was not called, KVM resources (VM fd, memory mappings, vCPU) were not released until the kernel reclaimed them asynchronously - The next epkg process would then encounter resource conflicts, causing ENOMEM The previous "6.12/6.13 kernel bug" theory was incorrect. The actual issue is that the host kernel is 6.12.43, but the guest kernel is 6.19.6. The bug is in epkg's resource management, not the kernel. Solution: - Save ctx_id before moving ctx to the VM thread - Explicitly call krun_free_ctx(ctx_id) before all std::process::exit() calls - This ensures KVM resources are released immediately, not asynchronously Testing: - Without fix: random failures when running without sleep - With fix: 10/10 consecutive runs succeeded without sleep Fixes: run_command_in_krun() fails randomly with: Failure during vcpu run: Out of memory (os error 12) Signed-off-by: Wu Fengguang <wfg@mail.ustc.edu.cn> | 4 个月前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 5 个月前 | ||
| 3 个月前 | ||
| 4 个月前 | ||
| 4 个月前 | ||
| 4 个月前 | ||
| 3 个月前 | ||
| 5 个月前 | ||
| 4 个月前 | ||
| 4 个月前 |