再續《linux 3.10 一次softlock排查》，看運行態進程數量之多：

crash> machMACHINE TYPE: x86_64MEMORY SIZE: 90 GBCPUS: 24HYPERVISOR: KVMPROCESSOR SPEED: 2394 MhzHZ: 1000PAGE SIZE: 4096KERNEL VIRTUAL BASE: ffff880000000000KERNEL VMALLOC BASE: ffffc90000000000KERNEL VMEMMAP BASE: ffffea0000000000KERNEL START MAP: ffffffff80000000KERNEL MODULES BASE: ffffffffc0000000KERNEL STACK SIZE: 16384IRQ STACK SIZE: 16384IRQ STACKS:CPU 0: ffff880b52a00000CPU 1: ffff880b52a40000CPU 2: ffff880b52a80000CPU 3: ffff880b52ac0000CPU 4: ffff880b52b00000CPU 5: ffff880b52b40000CPU 6: ffff880b52b80000CPU 7: ffff880b52bc0000CPU 8: ffff880b52c00000CPU 9: ffff880b52c40000CPU 10: ffff880b52c80000CPU 11: ffff880b52cc0000CPU 12: ffff881692200000CPU 13: ffff881692240000CPU 14: ffff881692280000CPU 15: ffff8816922c0000CPU 16: ffff881692300000CPU 17: ffff881692340000CPU 18: ffff881692380000CPU 19: ffff8816923c0000CPU 20: ffff881692400000CPU 21: ffff881692440000CPU 22: ffff881692480000CPU 23: ffff8816924c0000 DOUBLEFAULT STACK SIZE: 4096

但是，我們來看，有多少個running進程呢？

ps |grep -i RU > 0 0 0 ffffffff81a09480 RU 0.0 0 0 [swapper/0]0 0 1 ffff88016e798fd0 RU 0.0 0 0 [swapper/1]0 0 2 ffff88016e799fa0 RU 0.0 0 0 [swapper/2]0 0 3 ffff88016e79af70 RU 0.0 0 0 [swapper/3]0 0 4 ffff88016e79bf40 RU 0.0 0 0 [swapper/4]0 0 5 ffff88016e79cf10 RU 0.0 0 0 [swapper/5]0 0 6 ffff88016e79dee0 RU 0.0 0 0 [swapper/6]0 0 7 ffff88016e79eeb0 RU 0.0 0 0 [swapper/7] > 0 0 8 ffff88016e008000 RU 0.0 0 0 [swapper/8]0 0 9 ffff88016e008fd0 RU 0.0 0 0 [swapper/9]0 0 10 ffff88016e009fa0 RU 0.0 0 0 [swapper/10] > 0 0 11 ffff88016e00af70 RU 0.0 0 0 [swapper/11] > 0 0 12 ffff880beeaceeb0 RU 0.0 0 0 [swapper/12]0 0 13 ffff880beeacdee0 RU 0.0 0 0 [swapper/13]0 0 14 ffff880beeaccf10 RU 0.0 0 0 [swapper/14] > 0 0 15 ffff880beeacbf40 RU 0.0 0 0 [swapper/15]0 0 16 ffff880beeacaf70 RU 0.0 0 0 [swapper/16]0 0 17 ffff880beeac9fa0 RU 0.0 0 0 [swapper/17]0 0 18 ffff880beeb10000 RU 0.0 0 0 [swapper/18]0 0 19 ffff880beeb10fd0 RU 0.0 0 0 [swapper/19]0 0 20 ffff880beeb11fa0 RU 0.0 0 0 [swapper/20]0 0 21 ffff880beeb12f70 RU 0.0 0 0 [swapper/21]0 0 22 ffff880beeb13f40 RU 0.0 0 0 [swapper/22]0 0 23 ffff880beeb14f10 RU 0.0 0 0 [swapper/23]1 0 21 ffff880beeac8000 RU 0.0 201332 14472 systemd10 2 4 ffff88016e726eb0 RU 0.0 0 0 [rcu_sched]12 2 1 ffff88016e00eeb0 RU 0.0 0 0 [watchdog/1]14 2 1 ffff88016e00cf10 RU 0.0 0 0 [ksoftirqd/1]17 2 2 ffff88016e040fd0 RU 0.0 0 0 [watchdog/2]22 2 3 ffff88016e045ee0 RU 0.0 0 0 [watchdog/3]24 2 3 ffff88016e098000 RU 0.0 0 0 [ksoftirqd/3]27 2 4 ffff88016e09af70 RU 0.0 0 0 [watchdog/4]42 2 7 ffff88016e111fa0 RU 0.0 0 0 [watchdog/7]52 2 9 ffff88016e163f40 RU 0.0 0 0 [watchdog/9]57 2 10 ffff88016e198fd0 RU 0.0 0 0 [watchdog/10]73 2 13 ffff880beeb3bf40 RU 0.0 0 0 [watchdog/13]83 2 15 ffff880beeb7dee0 RU 0.0 0 0 [watchdog/15]88 2 16 ffff880beebaaf70 RU 0.0 0 0 [watchdog/16]93 2 17 ffff880beebf0000 RU 0.0 0 0 [watchdog/17]98 2 18 ffff880beebf4f10 RU 0.0 0 0 [watchdog/18]103 2 19 ffff880bee021fa0 RU 0.0 0 0 [watchdog/19]108 2 20 ffff880bee026eb0 RU 0.0 0 0 [watchdog/20]110 2 20 ffff880bee050fd0 RU 0.0 0 0 [ksoftirqd/20]113 2 21 ffff880bee053f40 RU 0.0 0 0 [watchdog/21]118 2 22 ffff880bee078fd0 RU 0.0 0 0 [watchdog/22]409 22481 10 ffff880b06e15ee0 RU 0.5 13597708 511816 zte_uep1410 22481 21 ffff88077fe08000 RU 0.5 13597708 511816 zte_uep1413 22481 1 ffff8814af3fbf40 RU 0.5 13597708 511816 zte_uep1425 17770 10 ffff8807967d6eb0 RU 0.5 2267176 439500 java636 1 15 ffff880b4f5b5ee0 RU 0.0 45452 1740 systemd-journal865 22481 23 ffff8809a4aacf10 RU 0.5 13597708 511816 zte_uep1924 22481 17 ffff880036a6cf10 RU 0.5 13597708 511816 zte_uep1932 2 13 ffff880036022f70 IN 0.0 0 0 [xfs_mru_cache]
。。。。。。。。。。。。。。。。。。。。列不完

看看個數：

crash> ps |grep -w RU |wc -l 1654

經常敲top的人肯定知道，running的進程數，一般是小于等于cpu的核數的，那為啥這個機器上這么多的running呢？為了排除是crash工具的問題，在從runqueue的角度來看下：

crash> p runqueues PER-CPU DATA TYPE:struct rq runqueues; PER-CPU ADDRESSES:[0]: ffff880b52a17c00[1]: ffff880b52a57c00 [2]: ffff880b52a97c00 [3]: ffff880b52ad7c00 [4]: ffff880b52b17c00 [5]: ffff880b52b57c00 [6]: ffff880b52b97c00 [7]: ffff880b52bd7c00 [8]: ffff880b52c17c00 [9]: ffff880b52c57c00 [10]: ffff880b52c97c00 [11]: ffff880b52cd7c00 [12]: ffff881692217c00 [13]: ffff881692257c00 [14]: ffff881692297c00 [15]: ffff8816922d7c00 [16]: ffff881692317c00 [17]: ffff881692357c00 [18]: ffff881692397c00 [19]: ffff8816923d7c00 [20]: ffff881692417c00 [21]: ffff881692457c00 [22]: ffff881692497c00 [23]: ffff8816924d7c00 crash> struct rq.nr_running ffff880b52a17c00 nr_running = 7 crash> struct rq.nr_running ffff880b52a57c00 nr_running = 152

挨個查看一下nr_running個數，然后腳本統計如下：

grep 'nr_running =' 1.txt |awk '{print $3}BEGIN{sum=0;num=0}{sum+=$3;num+=1}END{printf "%d\n",sum}' 7 152 113 138 65 4 4 38 4 148 43 1 1 141 4 43 107 102 91 110 108 105 98 3 1630-------------總數

為什么這個總數和我們ps看到的RU狀態的總數不一致呢？一開始還以為是percpu統計的瞬時值問題，但是這兩個總數的相差：1654-1630=24，剛好和cpu的核數一致。

ps |grep -w 'RU'|grep 'swapper' > 0 0 0 ffffffff81a09480 RU 0.0 0 0 [swapper/0]0 0 1 ffff88016e798fd0 RU 0.0 0 0 [swapper/1]0 0 2 ffff88016e799fa0 RU 0.0 0 0 [swapper/2]0 0 3 ffff88016e79af70 RU 0.0 0 0 [swapper/3]0 0 4 ffff88016e79bf40 RU 0.0 0 0 [swapper/4]0 0 5 ffff88016e79cf10 RU 0.0 0 0 [swapper/5]0 0 6 ffff88016e79dee0 RU 0.0 0 0 [swapper/6]0 0 7 ffff88016e79eeb0 RU 0.0 0 0 [swapper/7] > 0 0 8 ffff88016e008000 RU 0.0 0 0 [swapper/8]0 0 9 ffff88016e008fd0 RU 0.0 0 0 [swapper/9]0 0 10 ffff88016e009fa0 RU 0.0 0 0 [swapper/10] > 0 0 11 ffff88016e00af70 RU 0.0 0 0 [swapper/11] > 0 0 12 ffff880beeaceeb0 RU 0.0 0 0 [swapper/12]0 0 13 ffff880beeacdee0 RU 0.0 0 0 [swapper/13]0 0 14 ffff880beeaccf10 RU 0.0 0 0 [swapper/14] > 0 0 15 ffff880beeacbf40 RU 0.0 0 0 [swapper/15]0 0 16 ffff880beeacaf70 RU 0.0 0 0 [swapper/16]0 0 17 ffff880beeac9fa0 RU 0.0 0 0 [swapper/17]0 0 18 ffff880beeb10000 RU 0.0 0 0 [swapper/18]0 0 19 ffff880beeb10fd0 RU 0.0 0 0 [swapper/19]0 0 20 ffff880beeb11fa0 RU 0.0 0 0 [swapper/20]0 0 21 ffff880beeb12f70 RU 0.0 0 0 [swapper/21]0 0 22 ffff880beeb13f40 RU 0.0 0 0 [swapper/22]0 0 23 ffff880beeb14f10 RU 0.0 0 0 [swapper/23]

?

原來是因為idle進程默認在running態，但是并不算到nr_running的計數中：

crash> ps |grep -w 'RU' |grep -w '12' > 0 0 12 ffff880beeaceeb0 RU 0.0 0 0 [swapper/12]12 2 1 ffff88016e00eeb0 RU 0.0 0 0 [watchdog/1]-----------這個不是12cpu的進程22237 1 12 ffff881254b5eeb0 RU 1.0 14321692 912420 1_scheduler crash> struct rq.nr_running ffff881692217c00nr_running = 1

?為啥running這么多呢？查看處于running狀態的task，堆棧都是?__schedule ，也就是就緒隊列很長，

?來挑一些看看堆棧：

crash> bt 30507 PID: 30507 TASK: ffff8809d42daf70 CPU: 3 COMMAND: "java"#0 [ffff88099eb93d58] __schedule at ffffffff816b6165#1 [ffff88099eb93dc0] schedule at ffffffff816b6ae9#2 [ffff88099eb93dd0] schedule_hrtimeout_range_clock at ffffffff816b5852#3 [ffff88099eb93e68] schedule_hrtimeout_range at ffffffff816b5903#4 [ffff88099eb93e78] ep_poll at ffffffff812526de#5 [ffff88099eb93f30] sys_epoll_wait at ffffffff81253b9d#6 [ffff88099eb93f80] system_call_fastpath at ffffffff816c2789RIP: 00007ffa43a652c3 RSP: 00007ffa2595b818 RFLAGS: 00010202RAX: 00000000000000e8 RBX: ffffffff816c2789 RCX: 00000000e14ef930RDX: 0000000000002000 RSI: 00000000047a5000 RDI: 0000000000000067RBP: 00007ffa2595b630 R8: 0000000000000000 R9: 0000000000000003R10: 0000000000007530 R11: 0000000000000293 R12: 0000016547341bcaR13: 0000000000000067 R14: 0000000000007530 R15: 00000000047a5000ORIG_RAX: 00000000000000e8 CS: 0033 SS: 002b crash> bt 31606 PID: 31606 TASK: ffff8809cbe03f40 CPU: 9 COMMAND: "java"#0 [ffff880a4eedbd58] __schedule at ffffffff816b6165#1 [ffff880a4eedbdc0] schedule at ffffffff816b6ae9#2 [ffff880a4eedbdd0] schedule_hrtimeout_range_clock at ffffffff816b5852#3 [ffff880a4eedbe68] schedule_hrtimeout_range at ffffffff816b5903#4 [ffff880a4eedbe78] ep_poll at ffffffff812526de#5 [ffff880a4eedbf30] sys_epoll_wait at ffffffff81253b9d#6 [ffff880a4eedbf80] system_call_fastpath at ffffffff816c2789RIP: 00007f7bb9bb72c3 RSP: 00007f7b8f060510 RFLAGS: 00000246RAX: 00000000000000e8 RBX: ffffffff816c2789 RCX: 00000000ffffffffRDX: 0000000000002000 RSI: 0000000005be3000 RDI: 0000000000000151RBP: 00007f7b8f060310 R8: 0000000000000000 R9: 0000000000000003R10: 00000000000003e8 R11: 0000000000000293 R12: 0000016547349167R13: 0000000000000151 R14: 00000000000003e8 R15: 0000000005be3000ORIG_RAX: 00000000000000e8 CS: 0033 SS: 002b crash> task_struct.state ffff8809cbe03f40state = 0

看看狀態值的宏：

#define TASK_RUNNING 0 #define TASK_INTERRUPTIBLE 1 #define TASK_UNINTERRUPTIBLE 2 #define __TASK_STOPPED 4 #define __TASK_TRACED 8

再看看占有cpu的進程的堆棧，發現基本都在等鎖，原本大家很快都能執行完，然后放到等待隊列里面去，但是由于前面某個進程執行占著cpu不放，那么因為等待資源而得到喚醒的，處于就緒隊列里面的進程自然得不到調度。當然這個是非搶占式內核的特點，高優先級的進程，也得等低優先級的進程主動放棄cpu，哪怕你已經在就緒隊列，哪怕你的狀態已經被改成了running，但就得等著。

?

轉載于:https://www.cnblogs.com/10087622blog/p/9582606.html

總結

以上是生活随笔為你收集整理的linux 再多的running也挡不住锁的全部內容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。