valgrind：instrumentation_tool，记录parent何时发生fork
我很好奇是否有人（有一点 Valgrind 经验）知道如何告诉 Valgrind 的核心工具客户端发生了分叉。
考虑一下简单的分析工具的走狗。当一个人用走狗跟踪指令并且客户端分叉时，你将有两个“走狗”，一个跟踪父级，另一个跟踪子级。然而;我想知道是否有人知道如何在叉子处或叉子后不久“阻止”父走狗。
希望这是有道理的，想法吗？
I'm curious if anyone (with a bit of Valgrind experience) has an idea how to tell Valgrind's coretool that a fork in the client has occurred.
Consider the simple profiling tool lackey. When one traces instructions with lackey and the client forks, you will have two "lackeys", one tracing the parent and the other tracing the child. However; I'm wondering if anyone has an idea how to "stop" the parent lackey at the point of the fork or shortly after.
Hope this makes sense, thoughts?
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（2）
我没有想到直接的解决方案。我脑子里唯一的想法取决于任务。如果您是为了进行一些分析而使用二进制仪器，那么您可能希望将“分析”部分分离到另一个进程中，因此，只有仪器部分分叉，而您可以决定将它们连接到“分析过程”（并在那里发送日志信息或您想要的信息）。我希望会出现更好的解决方案:)。
no direct solution come to my mind. The only thoughts that come to my mind are dependent on task. If you are binary instrumentation in order to do some analysis, than you might like to separate "analysis" part into another process, thanks to that, only instrumented part forks, while you can decide with of them are connected to "analysis process" (and send log info there or what you want). I hope better solution will appear :).
好吧，我很快就明白了，但直到现在才忽略这个答案。
Valgrind 具有跟踪系统调用的能力，因此可以在系统调用调用之前和之后采取行动。 pub_tool 接口让它们愉快地等待使用和检测。
Well I figured it out shortly after but neglected this answer until now.
Valgrind has the capability of tracking system calls thus actions can be taken pre and post system call invokation. The pub_tool interface has them happily waiting to be used and instrumented.