用procstack进行问题排查-优快云博客

本文记录了一次程序异常卡顿的排查过程，通过使用procstack工具定位到问题根源在于两个库中存在相同类名导致的析构混乱。最终通过注释掉其中一个库解决了问题。

上次换包后系统出现一个奇怪的问题，函数A返回后按理应该继续往下走，结果却一直僵在那卡住，注释掉函数A调用则返回正常。排除数据库锁的问题，一直找不到问题所在。

向小强咨询，提示可以用procstack和truss查看一下进程状态，怎么把这两个家伙给忘了。

运行procstack如下：

procstack 1851632
0x3294b738 UcsSubsComponent::~UcsSubsComponent()(0x7a21d988, 0x2) + 0x10
0x329571ac void std::_Destroy<UcsSubsComponent>(UcsSubsComponent*)(0x7a21d988) + 0x18
0x329559cc std::allocator<UcsSubsComponent>::destroy(UcsSubsComponent*)(0x2ff1f1d4, 0x7a21d988) + 0x18
0x329548d8 std::vector<UcsSubsComponent,std::allocator<UcsSubsComponent> >::_Destroy(UcsSubsComponent*,UcsSubsComponent*)(0x2ff1f1d4, 0x7a21d988, 0x3fe64f58) + 0x24
0x329673a0 std::vector<UcsSubsComponent,std::allocator<UcsSubsComponent> >::_Clear()(0x2ff1f1d4) + 0x24
0x32960004 std::vector<UcsSubsComponent,std::allocator<UcsSubsComponent> >::~vector()(0x2ff1f1d4, 0x2) + 0x18
0x323fe3b4 cu_rnd_entity::MuOp::~MuOp()(0x2ff1f1d0, 0x2, 0x0) + 0x98
0x323ef498 dealMuCtl(const BAF_Transaction&,const BMS_USER&)(0x310d24a0, 0x2ff1f340) + 0x1e8
0x323b5078 User_Normalize(const BAF_Transaction&,BAF_FuncResult&)(0x310d24a0, 0x2ff20238) + 0xd28
0x31f0b404 cb_FixNewComplete(bmsOrder&,BAF_Transaction&)(0x2ff219b8, 0x310d24a0) + 0x19d8
0x311fe048 cb_OrderDecomposeXML(bmsOrder&,BAF_Transaction&)(0x2ff219b8, 0x310d24a0) + 0xe38
0x3020f25c OrderTransection::run()(0x2ff219b8) + 0x2c4
0x3012b250 Bms_Action_Batch(void*,Tcl_Interp*,int,const char**)(0x0, 0x3000d578, 0x3, 0x2ff21c00) + 0x164
0xd2634a94 TclInvokeStringCommand() + 0x98
0xd2633efc TclEvalObjvInternal() + 0x1d0
0xd2635f0c Tcl_EvalEx() + 0x1a8
0xd265b784 Tcl_FSEvalFile() + 0x1e8
0xd2680edc Tcl_Main() + 0x2f0
0x10000350 main() + 0x10
0x100001c0 __start() + 0x98