test(lmbench): 引入 lmbench 测试框架集成和自动测试脚本 - #1433
Open
mistcoversmyeyes wants to merge 39 commits into
Open
Conversation
mistcoversmyeyes
force-pushed
the
feat/lmbench
branch
6 times, most recently
from
December 17, 2025 15:46
4ca963e to
b5138f4
Compare
mistcoversmyeyes
force-pushed
the
feat/lmbench
branch
4 times, most recently
from
December 18, 2025 17:32
2edfdee to
cc88ddc
Compare
mistcoversmyeyes
force-pushed
the
feat/lmbench
branch
from
December 26, 2025 08:23
cc88ddc to
836a2c1
Compare
mistcoversmyeyes
force-pushed
the
feat/lmbench
branch
2 times, most recently
from
January 13, 2026 11:13
5657c94 to
a05fcb8
Compare
mistcoversmyeyes
force-pushed
the
feat/lmbench
branch
2 times, most recently
from
July 30, 2026 04:24
8ef1628 to
a9a963e
Compare
mistcoversmyeyes
marked this pull request as ready for review
July 31, 2026 11:05
mistcoversmyeyes
force-pushed
the
feat/lmbench
branch
from
August 12, 2026 14:07
671b758 to
1f409e9
Compare
- 从 DragonOS 镜像站安装 lmbench 二进制包 - 添加 lmbench 测试用例脚本，涵盖内存、进程、文件系统和网络等方面的性能测试 - 修改使用 ext2 文件系统的测试脚本为使用 ext4 文件系统 - 添加测试环境初始化脚本 init.sh ，清理脚本 clean_up.sh和环境变量配置脚本 env.sh
- 修复 `lmbench/mem_map_lat` 测例的权限问题，确保测试能够正确执行。 - 在 `ramfs_copy_files_bw` 测试中添加 `sudo`，以确保有足够的权限进行文件操作。 - 修复 `vfs_xx_lat` 系列测试缺少临时文件的问题，确保测试环境完整。
…(container-confirmed)
…NER hang lmbench's BENCH_INNER macro (bench.h) loops `while(__result < 0.95 * get_enough(0))`. Under virtualized timing compute_enough() can return SHORT=1000000, making __iterations explode (<<=3 past the 1<<27 break that resets __result=0) into an infinite while — the observed benchmp hangs on WSL2 (lat_fifo/bw_unix/lat_sem/bw_tcp/lat_select tcp/lat_udp/ lat_unix_connect all busy-loop state=R until timeout). Fix: precompute ENOUGH once via lmbench's `enough` tool (the standard config-run flow: `ENOUGH=`../bin/$OS/enough``), then clamp to <=100000 so BENCH_INNER's inner calibration converges, with 50000 (REAL_SHORT) fallback on hang/failure. Enough<=100000 makes the while(__result < 0.95*enough) target small enough to converge under timing jitter. Root cause confirmed via systematic-debugging on host: ENOUGH<=100000 unblocks all hung tools (lat_pipe/lat_unix/lat_sem/bw_unix output normal), ENOUGH=1000000 hangs, TIMING_O=0 LOOP_O=0 unblocks (skips BENCH_INNER). Not a DragonOS kernel issue; guest 8/8 e2e implies guest timing is stable.
…te serial EINVAL - unblock lmbench in app-blocklist so make installs the suite - remove sudo / unsupported -W -N args from copy/mmap/ramfs test cases - ramfs lat_fs: use default iterations + explicit tmp dir (was 360s timeout) - init.sh: shrink fixtures 512M->64M, tolerate dd stderr EINVAL on serial close - env.sh: enable LMBENCH_CREATE_TEST_FILES Guest e2e verified: ramfs/copy metrics now ok (31 ok / 47 metrics).
… fixtures
- init.sh: always recreate 64MB fixtures; stale undersized files broke the
size checks in bw_mmap_rd/bw_file_rd/lat_mmap/lat_pagefault
- mem_mmap_bw: 16m (<=64MB fixture); mem_pagefault_lat: dedicated 8MB file;
vfs_read_pagecache_bw: 512m -> 64m (bw_file_rd validates size <= file)
- tcp_loopback_bw_*.meta: SEARCH_PATTERN now matches real bw_tcp output
("<size> <MB/s> MB/sec") instead of the guessed "socket connection"
- tcp_loopback_http_bw: build a web root, DOCROOT + port 8080, abs file_list
- tcp_virtio_*: proper server/client + -S shutdown lifecycle (was leaking
daemons that held the fixed ports -> EADDRINUSE)
- unix_connect_lat/semaphore_lat: 10s bound; guest kernel can hang on the
missing unix-socket bind / sem_open instead of failing fast
Guest e2e verified: 32 metrics, 23 ok (mem/pagecache/process fixes confirmed).
mistcoversmyeyes
force-pushed
the
feat/lmbench
branch
from
August 12, 2026 14:41
1f409e9 to
364364f
Compare
sparkzky
reviewed
Aug 13, 2026
lmbench server mode (-s) double-forks: $! captures the launcher PID (already exited), not the daemon. When set -e aborts the script on client failure, the explicit -S shutdown in the body never runs, and the cleanup trap kill $SERVER_PID misses the daemon. Leaked daemons persist on fixed ports (31234/31236/31237), corrupting subsequent network tests that share the port. Add the -S shutdown command to cleanup() in all 11 network test scripts so the daemon is stopped on every exit path (set -e failure, timeout SIGTERM, normal exit). Signed-off-by: sparkzky <sparkhhhhhhhhhh@outlook.com>
Member
|
几个 commit 里提到的数字不太一样（8/8、31/47、32/23），想确认下以最新代码为准，现在是多少 ok / 总共多少个？有没有逐个测例的通过清单？ |
mistcoversmyeyes
left a comment
mistcoversmyeyes
left a comment
Contributor
Author
There was a problem hiding this comment.
What's this
Something to be clean
mistcoversmyeyes
left a comment
mistcoversmyeyes
left a comment
Contributor
Author
There was a problem hiding this comment.
几个 commit 里提到的数字不太一样（8/8、31/47、32/23），想确认下以最新代码为准，现在是多少 ok / 总共多少个？
由于 Full Run 的时候，中间部分测试用例会随机卡住，阻塞后续用例运行，故而每次 Full Run OK 的测试用例数目都不一样，但是大概是 31/47 这样。
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
3 participants
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.
No description provided.