无法在使用golang时通过corefile从cgo例程中获取堆栈跟踪。

huangapple
go评论106阅读模式
英文:

Unable to get the stack trace with a corefile from a cgo routine when using golang

问题

我正在使用Golang和cgo。当我的C代码引发assert()时，使用cgo时无法看到C代码的堆栈跟踪。

相反，我看到捕获assert的golang运行时的堆栈跟踪。

这是我的C代码示例：

  1. #include <stdio.h>
  2. #include <pthread.h>
  3. #include <assert.h>
  4. #include <string.h>
  6. void fn2(char *arg)
  7. {
  8.     int stackvar2 = 256;
  10.     printf("Argument %s\n", arg);
  12.     assert(1 == 2);
  13. }
  15. void fn1(int arg)
  16. {
  17.     int stackvar3 = 512;
  18.     char var[256];
  20.     strcpy(var, "deadbeef");
  22.     fn2(var);
  23. }
  26. void *thread(void *arg)
  27. {
  28.     printf("Hello from the thread... going in for an assert\n");
  30.     fn1(1092);
  32.     return NULL;
  33. }
  35. void hello_world(char *str)
  36. {
  37.     pthread_t tid;
  39.     printf("Hello World from C and here the str is from Go: %s\n", str);
  41.     pthread_create(&tid, NULL, thread, NULL);
  43.     sleep(100000);
  44. }

这是我的Go代码：

  1. package main
  3. /*
  4. extern void hello_world(char *str);
  5. #cgo LDFLAGS: -L. -lhello
  6. #cgo CFLAGS: -g3
  7. */
  8. import "C"
  10. import (
  11.     _ "fmt"
  12. )
  14. func main() {
  15.     str := "From Golang"
  17.     cStr := C.CString(str)
  18.     C.hello_world(cStr)
  20.     select {}
  21. }

这是我的Makefile：

  1. all:
  2.     gcc -g3 -O0 -c hello.c
  3.     ar cru libhello.a hello.o
  4.     go build hello.go
  6. clean:
  7.     rm -f *.o hello
英文:

I am using Golang and cgo. When my C code raises an assert(), I am unable to see the stack trace of the C code when using cgo.

Instead, I see the stack trace of the golang runtime that caught the assert.

Here is an example of my C code

  1. #include &lt;stdio.h&gt;
  2. #include &lt;pthread.h&gt;
  3. #include &lt;assert.h&gt;
  4. #include &lt;string.h&gt;
  6. void fn2(char *arg)
  7. {
  8.     int stackvar2 = 256;
  10.     printf(&quot;Argument %s\n&quot;, arg);
  12.     assert(1 == 2);
  13. }
  15. void fn1(int arg)
  16. {
  17.     int stackvar3 = 512;
  18.     char var[256];
  20.     strcpy(var, &quot;deadbeef&quot;);
  22.     fn2(var);
  23. }
  26. void *thread(void *arg)
  27. {
  28.     printf(&quot;Hello from the thread... going in for an assert\n&quot;);
  30.     fn1(1092);
  32.     return NULL;
  33. }
  35. void hello_world(char *str)
  36. {
  37.     pthread_t tid;
  39.     printf(&quot;Hello World from C and here the str is from Go: %s\n&quot;, str);
  41.     pthread_create(&amp;tid, NULL, thread, NULL);
  43.     sleep(100000);
  44. }
  47. Here is my Go code
  50.   
  51. package main
  53. /*
  54. extern void hello_world(char *str);
  55. #cgo LDFLAGS: -L. -lhello
  56. #cgo CFLAGS: -g3
  57. */
  58. import &quot;C&quot;
  60. import (
  61. 	_ &quot;fmt&quot;
  62. )
  64. func main() {
  65. 	str := &quot;From Golang&quot;
  67. 	cStr := C.CString(str)
  68. 	C.hello_world(cStr)
  70. 	select {}
  71. }

And here is my Makefile

  1. all:
  2. 	gcc -g3 -O0 -c hello.c
  3. 	ar cru libhello.a hello.o
  4. 	go build hello.go
  6. clean:
  7. 	rm -f *.o hello

答案1

得分: 2

除了显而易见的ulimit -c检查之外，使用GOTRACEBACK=crash运行Go程序。这将打印出更多信息，并允许程序使用SIGABRT退出以触发核心转储。

英文:

Besides the obvious check of ulimit -c, run the go program with GOTRACEBACK=crash. This will print out more information, and allow the program to exit with SIGABRT to trigger a core dump.

答案2

得分: 1

实际上，我需要将这段代码添加到我的Go代码中：signal.Ignore(syscall.SIGABRT)。这样可以让我查看崩溃的C代码的堆栈跟踪。

英文:

Actually I need to add this to my go code: signal.Ignore(syscall.SIGABRT). This allows me to see the stack trace of the C code that crashed.

huangapple
  • 本文由 发表于 2016年2月6日 05:29:52
  • 转载请务必保留本文链接：https://go.coder-hub.com/35233871.html
  • cgo
  • gccgo
  • go
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定