What does this PR do?

This PR replaces the usage of non-async-safe functions (fprintf, printf, exit) in the SIGSEGV handler with async-safe alternatives (write, _exit). This aims to ensure more reliable behavior when a segmentation fault occurs.

Motivation

https://docs.oracle.com/cd/E19455-01/806-5257/gen-26/index.html

When a SIGSEGV is triggered, the program’s state may already be corrupt. Using non-async-safe functions inside the signal handler can lead to reentrancy issues or additional faults. By switching to write and _exit, we minimize these risks and maintain a more stable crash-handling sequence.

Additional Notes

• backtrace_symbols_fd is left as is for now, even though it’s not guaranteed to be fully async-safe, in order to keep the existing logic mostly unchanged.
• By TODO comment in the code, it appears that removing fprintf-family functions was already planned, suggesting that this problem was anticipated.
• I am not c++ developer, but the issue has already arisen and was known. This code must be updated in an async-safe manner, whether it’s through my proposal or another approach.

How to test the change?

• Trigger a SIGSEGV intentionally and verify that the updated handler logs the message and terminates without additional faults.
• Compare the behavior with the previous handler to ensure no unexpected regressions.
• Observe whether the logs or any diagnostic outputs align with the expected error information under stress tests.