strace显示线程触发SIGSEGV但主进程以退出码0结束，原因何在？

问题描述 #

对某进程执行strace时发现，部分线程触发了SIGSEGV信号，但主进程仍持续执行并最终以退出状态码0结束，不符合“进程应返回对应错误码”的常规逻辑，需分析原因。

环境信息 #

strace -- version 5.18
kernel -- 5.18.15-1.el8.elrepo.x86_64

相关strace输出 #

--- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=7, si_uid=20000, si_status=0, si_utime=0, si_stime=0} ---
strace: Process 2744190 attached
[pid 2744190] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=NULL} ---
strace: Process 2744191 attached
strace: Process 2744192 attached
strace: Process 2744193 attached
strace: Process 2744194 attached
strace: Process 2744195 attached
strace: Process 2744196 attached
strace: Process 2744197 attached
strace: Process 2744198 attached
[pid 2744196] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x7fa89a131d00} ---
[pid 2744190] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0xc} ---
[pid 2744195] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x7fa89a131b80} ---
strace: Process 2744248 attached
strace: Process 2744249 attached
strace: Process 2744250 attached
[pid 2744190] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x7fa89a131a80} ---
strace: Process 2744277 attached
strace: Process 2744278 attached
strace: Process 2744279 attached
[pid 2744279] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0xc} ---
strace: Process 2744280 attached
[pid 2744280] +++ exited with 0 +++
strace: Process 2744281 attached
[pid 2744196] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x7fa89a131d00} ---
[pid 2744279] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x7fa89a131c00} ---
[pid 2744279] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x7fa89a131c00} ---
[pid 2744196] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x7fa89a131d00} ---
strace: Process 2744445 attached
[pid 2744445] +++ exited with 0 +++
strace: Process 2744446 attached
[pid 2744446] +++ exited with 0 +++
[pid 2744198] +++ exited with 0 +++
[pid 2744191] +++ exited with 0 +++
[pid 2744190] +++ exited with 0 +++
[pid 2744281] +++ exited with 0 +++
[pid 2744277] +++ exited with 0 +++
[pid 2744249] +++ exited with 0 +++
[pid 2744279] +++ exited with 0 +++
[pid 2744278] +++ exited with 0 +++
[pid 2744250] +++ exited with 0 +++
[pid 2744248] +++ exited with 0 +++
[pid 2744197] +++ exited with 0 +++
[pid 2744196] +++ exited with 0 +++
[pid 2744195] +++ exited with 0 +++
[pid 2744194] +++ exited with 0 +++
[pid 2744193] +++ exited with 0 +++
[pid 2744192] +++ exited with 0 +++
+++ exited with 0 +++

核心原因分析 #

• 自定义SIGSEGV信号处理函数：程序通过sigaction()等系统调用注册了自定义的SIGSEGV处理逻辑。当线程触发段错误时，系统会调用该处理函数而非直接终止进程。处理函数可能完成了错误修复（比如动态映射缺失的内存区域、修正非法指针），让线程继续执行；或者触发线程的优雅退出流程，不影响主进程的正常运行。
• 线程级错误隔离：SIGSEGV信号默认作用于触发它的线程而非整个进程。如果信号处理函数没有终止整个进程，仅让出错线程退出（从strace输出可见，所有触发SIGSEGV的线程最终都以状态码0退出），主进程的主线程会继续执行，等待所有子线程结束后正常退出，返回状态码0。
• 特定业务逻辑触发的“预期内段错误”：部分程序（比如内存调试工具、动态加载框架）会故意触发SIGSEGV来实现特定功能，比如检测内存边界、延迟加载代码/数据。这类场景下，段错误是预期内的流程，程序会在信号处理函数中完成相应逻辑后恢复执行或正常退出。
• SIGSEGV类型与处理适配：输出中存在两种SIGSEGV类型：SEGV_MAPERR（访问未映射内存）和SEGV_ACCERR（访问已映射但无权限内存），程序可能针对这两种场景分别做了处理，确保错误被局限在单个线程内，不扩散到整个进程。

内容的提问来源于stack exchange，提问作者Uruk-hai