[01:26] 有人没? [01:26] You are banned from this server- You have a host listed in the DroneBL. For more information, visit http://eu1.dronebl.org/lookup_branded.do?ip [01:27] 我连irc的时候显示ip被封了? 已经请求解封过了 怎么用原来的IP还是没法连irc [01:30] https://i.loli.net/2017/10/31/59f7d2093239f.png [01:30] 有人遇到这种情况吗? [01:30] ,, [01:35] SJX: 知道吗? [01:36] 不清楚啊,昨天刚用上irc... [02:09] iMadper: 壕早 [02:12] ooOO_OOoo: 早. [02:12] ooOO_OOoo: 王老板日子这么滋润? [02:12] ooOO_OOoo: 我还想请教几个ltp的问题呢 [02:12] iMadper: 王老板现在niu了呢 [02:13] huntxu: 我 [02:13] huntxu: 修好 [02:13] huntxu: 输入法 [02:13] huntxu: 了 [02:13] ooOO_OOoo: 肯定的啊 [02:13] ooOO_OOoo: 王老板离senior多远? [02:13] ooOO_OOoo: 一年? [02:13] ooOO_OOoo: 他说重装了系统,刚把 irc server 发给他了 [02:13] iMadper: 估计也快了 [02:15] ooOO_OOoo: 昂. [02:21] iMadper: ltp啥问题,王老板还没上线。。 你急的话我问问他 [02:25] iMadper [02:25] wangli: 王老板早 [02:25] 谢老板早 [02:29] iMadper, [02:30] wangli: 早 [02:30] wangli: 问一下, 现在systemd情况下, 跑ltp会有很多失败啊 [02:30] wangli: cgroup相关的 [02:30] 啥失败 [02:30] wangli: 因为systemd一上来就会按照他需要的拓扑mount很多cgroup的实例 [02:31] iMadper, 贴出来看看 [02:31] wangli: 然后ltp的case假设是干净的系统 [02:31] wangli: 好的 [02:33] wangli: https://pastebin.com/B0Znwcu2 [02:34] 对啊 [02:34] 高级了 [02:34] iMadper, checking [02:41] iMadper, hi, which ltp-version did you run? and what's the kernel-version/arch/platform? [02:42] wangli: LTP Version: 20170516 kernel: modified 4.11 aarch64 [02:42] iMadper, it seems not a systemd problem, i guess it is related to cgroup or kernel bug, but still need do some more research [02:43] wangli: So it's not a known bug? [02:43] iMadper sorry for input English, my workstation always hung with input Chinese:( [02:43] MangHuoEr: ^^ [02:44] wangli: I was expecting it's a conflict between ltp and systemd. loool [02:44] iMadper, not sure, and can you run the latest ltp-20170929 version? [02:45] wangli: any modifications against the cgroup cases? [02:45] memcg_stat_test 6 TFAIL: hierarchical_memory_limit is 65536, 131072 expected [02:46] wangli: Yes, because the parent node has a limit of 65536... [02:47] iMadper, so you guess the limit set to 65536 in systemd? [02:47] wangli: *nod* [02:48] iMadper, sounds reasonable, but we need some more debugging [02:48] wangli: so you never met that issue? [02:50] iMadper, i didn't run the cgroup cases before, let me try it now [02:50] cat /sys/fs/cgroup/memory/memory.use_hierarchy ? [02:51] wangli: okay. Shouldn't memcg belong to ktd-team? [02:52] iMadper, it's not my working area, but i can do it for you [02:52] iMadper, just wait.. [02:54] wangli: thanks tons. [02:59] 怎么他突然e文了 [02:59] 是的 [02:59] iMadper, can you perform 'cat /sys/fs/cgroup/memory/memory.use_hierarchy' as MangHuoEr mentioned above? [03:00] wangli: 1 [03:01] iMadper, ok, it's not the hierarchy problem [03:01] wangli: what's the value in your side? [03:01] i just noticed that ltp disable it in the cgroup case [03:01] echo 0 > memory.use_hierarchy [03:02] so it's not related to systemd i think [03:03] iMadper, the page_size on your system is? [03:04] wangli: however if it has a sub-node, we can't change the value. [03:04] wangli: and by default, systemd creates sub-nodes. [03:05] ========== [03:05] testcase_6() [03:05] { [03:05] echo 0 > memory.use_hierarchy [03:05] mkdir subgroup [03:05] echo $PAGESIZE > memory.limit_in_bytes [03:05] echo $((PAGESIZE*2)) > subgroup/memory.limit_in_bytes [03:05] wangli:. .., 别刷屏,超过6行 大段文字 请贴至paste.ubuntu.com +q 23s [03:05] cd subgroup [03:05] check_mem_stat "hierarchical_memory_limit" $((PAGESIZE*2)) [03:05] cd .. [03:05] rmdir subgroup [03:05] } [03:06] memcg_stat_test 6 TPASS: hierarchical_memory_limit is 8192 as expected [03:06] echo 0 > memory.use_hierarchy <--- this line fails. [03:06] iMadper, I get pass on my x86_64 box [03:06] iMadper, really? [03:06] do you have a sub-node before executing the test case? [03:07] try to create one before executing it. [03:07] i didn't, i just run ltp by default [03:07] sure [03:10] * iMadper brb [03:11] # echo 0 > /dev/memcg/wangli/memory.use_hierarchy [03:11] # cat /dev/memcg/wangli/memory.use_hierarchy [03:11] 0 [03:12] iMadper, ^ [03:12] wangli: so you can still change parent's memory.use_hierarchy? [03:13] wangli: the child node could always be modified. However, in the meanwhile, parent's can't [03:15] iMadper, hmm, what kind of situation you hit? I didn't fully understand you mean [03:16] wangli: okay, you have a sub-node "wangli" which located in /dev/memcg/wangli. could you still echo 0/1 > /dev/memcg/memory.use_hierarchy ? [03:17] wangli: What I'm trying to express is: Given a memcg node A. Once A has a child node, The _A_'s use_hierarchy can't be changed. [03:17] iMadper, understand [03:18] iMadper, i can't modify the parent's too [03:20] wangli: Okay. So by default there is no sub-node inside memcg on rhel7? [03:20] iMadper, seems yes! [03:21] wangli: I see. [03:21] iMadper, if so, we probably need to modify the ltp-case to cover your system [03:22] wangli: yeah, for ubuntu... [03:23] wangli: it seems the default behavior of ubuntu systemd leads the issue. [03:24] iMadper, can you show me the cgroup dir in your system? [03:26] wangli: iMadper http://pastebin.ubuntu.com/25855776/ === iMadper is now known as iMadper|eat === iMadper|eat is now known as iMadper [08:29] . [08:29] test [08:29] sjd_zeus:点点点. 16:29 [09:21] test [09:21] MangHuoEr:点点点. 17:21 [09:22] sjd_zeus: 色鸡蛋! [09:22] .......... [11:02] knownbad, 老屁眼 === _d0048 is now known as d0048 [18:02] knownbad, 在么?