Community
Participate
Working Groups
Created attachment 258164 [details] Screenshot and dump Hi Eugene, In the attachement, you can find a backtrace that "repeats itself" forever. We are stopped at 0xb6f55830 which is 0xce830 in dump.txt ce830 is a jump to _exit. When I look at _exit, there is no detectable epilogue. But the sp register has been changed, pc is seen unchanged, so we have a new level and so on. This happens when we are stopped in a function that has been spawned by pthread_create (we have an agent running on wrLinux arm). With the design of the arm crawler, I see no way to fix this. Any ideas ? PS : As you can see in the screenshot, the debugger seems not able to display the function names. libc-2.20.so has NO .symtab section, only .dynsym. Can TCF agent can take usage of .dynsym to retrieve the symbol name ?) Thanks ! Xavier.
At the moment, I don't see how to improve detection of end of stack. > Can TCF agent can take usage of .dynsym to retrieve the symbol name ? Maybe. I'll look into it.
Thanks ! This way, we could use the BOTTOM_STACK_HOOK.
> Can TCF agent can take usage of .dynsym to retrieve the symbol name ? I have changed the code to always add dynamic symbols to search results.
Thanks !