Expiring multiple action results causes crash

[Scotty1701]

Bug report

Required Info:

• Operating System:
  • Windows 10
• Installation type:
  • Chocolatey Install
• Version or commit hash:
  • Foxy
• DDS implementation:
  • Fast-RTPS
• Client library (if applicable):
  • rclcpp_action

Steps to reproduce issue

Start the example fibonacci action server:
ros2 run action_tutorials_cpp fibonacci_action_server

Send 2 goals to the server in quick succession:
ros2 action send_goal /fibonacci action_tutorials_interfaces/action/Fibonacci "{}"

Wait 15 minutes (or however long the result_timeout for the action server is set for).

Expected behavior

Action Server continues running

Actual behavior

Action Server crashes with exception code 0xc0000409

Additional information

Sending one goal, waiting for the result to expire, and then sending another does not cause a crash, only when there are multiple cached goals does a crash occur. To avoid a 15 minute wait before a crash, you can set the result_timeout.nanoseconds field of an rcl_action_server_options_t object and pass that to the rclcpp_action::create_server function.

[fujitatomoya]

I see this is the problem in theory and logically, but cannot make it happen.

[KavenYau]

#0  __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:50
#1  0x00007f8bb0ac3859 in __GI_abort () at abort.c:79
#2  0x00007f8bb0b2e3ee in __libc_message (action=action@entry=do_abort, fmt=fmt@entry=0x7f8bb0c5807c "*** %s ***: terminated\n") at ../sysdeps/posix/libc_fatal.c:155
#3  0x00007f8bb0bd0b4a in __GI___fortify_fail (msg=msg@entry=0x7f8bb0c58064 "stack smashing detected") at fortify_fail.c:26
#4  0x00007f8bb0bd0b16 in __stack_chk_fail () at stack_chk_fail.c:24
#5  0x00007f8bb0776571 in rclcpp_action::ServerBase::execute_check_expired_goals() () from /home/kaven/ros2_ws/install/rclcpp_action/lib/librclcpp_action.so
#6  0x7deabfa5e2ea4802 in ?? ()
#7  0x3707acb461434114 in ?? ()
#8  0x162168ace7f673b2 in ?? ()
#9  0x7d9d995e8448ee4d in ?? ()
#10 0x3a0160f661434115 in ?? ()
#11 0xa5dfa2a0f83d899f in ?? ()
#12 0x58790a2639635c28 in ?? ()
#13 0x09444a6861434117 in ?? ()
#14 0x61dc685b9a6a2f3b in ?? ()
#15 0x06d52596f0b1e101 in ?? ()
#16 0x0c416ec761434118 in ?? ()
#17 0x957f8a9bd93a0845 in ?? ()
#18 0xee55f19df09ba4a3 in ?? ()
#19 0x0f214afd61434119 in ?? ()
#20 0x28243330a9c0ca32 in ?? ()
#21 0x760d2152abb21d59 in ?? ()
#22 0x12239e8d6143411a in ?? ()
#23 0x6fd24b3bc403950a in ?? ()
#24 0x64edc8aa0c34c025 in ?? ()
#25 0x151307016143411b in ?? ()
#26 0x00007fff333a5bd0 in ?? ()
#27 0x00007fff333a5b80 in ?? ()
#28 0x00007f8bb17176d6 in rclcpp::executors::SingleThreadedExecutor::spin() () from /home/kaven/ros2_ws/install/rclcpp/lib/librclcpp.so
Backtrace stopped: Cannot access memory at address 0x92933d130abe9902

I think I got this issue on ubuntu 20.04 too.

[clalancette]

@KavenYau Can you confirm that the latest code with #931 applied fixes your problem?

[KavenYau]

@clalancette Yes, with applying #931, it was fixed.