[ovs-discuss] ovs 2.6.1 crash due to __nr_readers overflow

[夏宇]

Hello OVS team,


I have a ovs-vswitched crash of ovs version 2.6.1. From the backtrace I could see that the crash is because of __nr_readers overflow. 



(gdb) bt
#0  0x00007f31539f71d7 in __GI_raise (sig=sig at entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
#1  0x00007f31539f88c8 in __GI_abort () at abort.c:90
#2  0x00007f31558287ae in ovs_abort_valist (err_no=<optimized out>, format=<optimized out>, args=args at entry=0x7f30a3ff9f20) at lib/util.c:335
#3  0x00007f3155828844 in ovs_abort (err_no=<optimized out>, format=format at entry=0x7f31558ea6b0 "%s: pthread_%s_%s failed") at lib/util.c:327
#4  0x00007f31557fb162 in ovs_rwlock_rdlock_at (l_=0x7f3156ff5d98, where=where at entry=0x7f31558d2420 "ofproto/ofproto-dpif-xlate.c:2152") at lib/ovs-thread.c:77
#5  0x00007f3155763e37 in update_learning_table (xbridge=0x7f31570583c0, flow=0x7f3094006ec0, wc=0x7f30a3ffa0b0, vlan=2226, in_xbundle=0x7f3156fdf240)
    at ofproto/ofproto-dpif-xlate.c:2152
#6  0x00007f3155765b89 in xlate_cache_normal (ofproto=<optimized out>, flow=0x7f3094006ec0, vlan=2226) at ofproto/ofproto-dpif-xlate.c:5857
#7  0x00007f315576def0 in xlate_push_stats (xcache=<optimized out>, stats=stats at entry=0x7f30a3ffa3f0) at ofproto/ofproto-dpif-xlate.c:5898
#8  0x00007f315575f282 in revalidate_ukey (udpif=udpif at entry=0x7f3156fe4420, ukey=ukey at entry=0x7f30d0007c90, stats=stats at entry=0x7f30a3ffcc58, 
    odp_actions=odp_actions at entry=0x7f30a3ffaee0, reval_seq=reval_seq at entry=8575315723, recircs=recircs at entry=0x7f30a3ffaed0) at ofproto/ofproto-dpif-upcall.c:1844
#9  0x00007f3155761ad4 in revalidate (revalidator=0x7f3156fc8008) at ofproto/ofproto-dpif-upcall.c:2199
#10 0x00007f3155761dcb in udpif_revalidator (arg=0x7f3156fc8008) at ofproto/ofproto-dpif-upcall.c:881
#11 0x00007f31557fb066 in ovsthread_wrapper (aux_=<optimized out>) at lib/ovs-thread.c:342
#12 0x00007f31546ebdc5 in start_thread (arg=0x7f30a3fff700) at pthread_create.c:308
#13 0x00007f3153ab976d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:113
(gdb) frame 4
#4  0x00007f31557fb162 in ovs_rwlock_rdlock_at (l_=0x7f3156ff5d98, where=where at entry=0x7f31558d2420 "ofproto/ofproto-dpif-xlate.c:2152") at lib/ovs-thread.c:77
77    LOCK_FUNCTION(rwlock, rdlock)
(gdb) p * l
$2 = {lock = {__data = {__lock = 0, __nr_readers = 4294967295, __readers_wakeup = 72951523, __writer_wakeup = 128782028, __nr_readers_queued = 0, __nr_writers_queued = 0, 
      __writer = 0, __shared = 0, __pad1 = 0, __pad2 = 0, __flags = 0}, __size = "\000\000\000\000\377\377\377\377\343&Y\004\314\016\255\a", '\000' <repeats 39 times>, 
    __align = -4294967296}, where = 0x7f31558cb9de "<unlocked>"}


ovs-vswitchd --version
ovs-vswitchd (Open vSwitch) 2.6.1


cat /proc/version 
Linux version 3.10.0-514.26.2.el7.x86_64 (builder at kbuilder.dev.centos.org) (gcc version 4.8.5 20150623 (Red Hat 4.8.5-11) (GCC) ) #1 SMP Tue Jul 4 15:04:05 UTC 2017





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openvswitch.org/pipermail/ovs-discuss/attachments/20190428/c743c0f7/attachment.html>