Date: Mon, 25 Nov 2019 18:33:22 +0000 (UTC) From: Mariusz Zaborski <oshogbo@FreeBSD.org> To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: svn commit: r355097 - in head: sys/kern tests/sys/kern Message-ID: <201911251833.xAPIXM9h040583@repo.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: oshogbo Date: Mon Nov 25 18:33:21 2019 New Revision: 355097 URL: https://svnweb.freebsd.org/changeset/base/355097 Log: procdesc: allow to collect status through wait(1) if process is traced The debugger like truss(1) depends on the wait(2) syscall. This syscall waits for ALL children. When it is waiting for ALL child's the children created by process descriptors are not returned. This behavior was introduced because we want to implement libraries which may pdfork(1). The behavior of process descriptor brakes truss(1) because it will not be able to collect the status of processes with process descriptors. To address this problem the status is returned to parent when the child is traced. While the process is traced the debugger is the new parent. In case the original parent and debugger are the same process it means the debugger explicitly used pdfork() to create the child. In that case the debugger should be using kqueue()/pdwait() instead of wait(). Add test case to verify that. The test case was implemented by markj@. Reviewed by: kib, markj Discussed with: jhb MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D20362 Modified: head/sys/kern/kern_exit.c head/sys/kern/sys_procdesc.c head/tests/sys/kern/ptrace_test.c Modified: head/sys/kern/kern_exit.c ============================================================================== --- head/sys/kern/kern_exit.c Mon Nov 25 18:27:02 2019 (r355096) +++ head/sys/kern/kern_exit.c Mon Nov 25 18:33:21 2019 (r355097) @@ -993,11 +993,14 @@ proc_to_reap(struct thread *td, struct proc *p, idtype switch (idtype) { case P_ALL: - if (p->p_procdesc != NULL) { - PROC_UNLOCK(p); - return (0); + if (p->p_procdesc == NULL || + (p->p_pptr == td->td_proc && + (p->p_flag & P_TRACED) != 0)) { + break; } - break; + + PROC_UNLOCK(p); + return (0); case P_PID: if (p->p_pid != (pid_t)id) { PROC_UNLOCK(p); Modified: head/sys/kern/sys_procdesc.c ============================================================================== --- head/sys/kern/sys_procdesc.c Mon Nov 25 18:27:02 2019 (r355096) +++ head/sys/kern/sys_procdesc.c Mon Nov 25 18:33:21 2019 (r355097) @@ -60,7 +60,6 @@ * * Open questions: * - * - How to handle ptrace(2)? * - Will we want to add a pidtoprocdesc(2) system call to allow process * descriptors to be created for processes without pdfork(2)? */ Modified: head/tests/sys/kern/ptrace_test.c ============================================================================== --- head/tests/sys/kern/ptrace_test.c Mon Nov 25 18:27:02 2019 (r355096) +++ head/tests/sys/kern/ptrace_test.c Mon Nov 25 18:33:21 2019 (r355097) @@ -4135,6 +4135,105 @@ ATF_TC_BODY(ptrace__proc_reparent, tc) ATF_REQUIRE(errno == ECHILD); } +/* + * Ensure that traced processes created with pdfork(2) are visible to + * waitid(P_ALL). + */ +ATF_TC_WITHOUT_HEAD(ptrace__procdesc_wait_child); +ATF_TC_BODY(ptrace__procdesc_wait_child, tc) +{ + pid_t child, wpid; + int pd, status; + + child = pdfork(&pd, 0); + ATF_REQUIRE(child >= 0); + + if (child == 0) { + trace_me(); + (void)raise(SIGSTOP); + exit(0); + } + + wpid = waitpid(child, &status, 0); + ATF_REQUIRE(wpid == child); + ATF_REQUIRE(WIFSTOPPED(status)); + ATF_REQUIRE(WSTOPSIG(status) == SIGSTOP); + + ATF_REQUIRE(ptrace(PT_CONTINUE, child, (caddr_t)1, 0) != -1); + + wpid = wait(&status); + ATF_REQUIRE(wpid == child); + ATF_REQUIRE(WIFSTOPPED(status)); + ATF_REQUIRE(WSTOPSIG(status) == SIGSTOP); + + ATF_REQUIRE(ptrace(PT_CONTINUE, child, (caddr_t)1, 0) != -1); + + /* + * If process was created by pdfork, the return code have to + * be collected through process descriptor. + */ + wpid = wait(&status); + ATF_REQUIRE(wpid == -1); + ATF_REQUIRE(errno == ECHILD); + + ATF_REQUIRE(close(pd) != -1); +} + +/* + * Ensure that traced processes created with pdfork(2) are not visible + * after returning to parent - waitid(P_ALL). + */ +ATF_TC_WITHOUT_HEAD(ptrace__procdesc_reparent_wait_child); +ATF_TC_BODY(ptrace__procdesc_reparent_wait_child, tc) +{ + pid_t traced, debuger, wpid; + int pd, status; + + traced = pdfork(&pd, 0); + ATF_REQUIRE(traced >= 0); + if (traced == 0) { + raise(SIGSTOP); + exit(0); + } + ATF_REQUIRE(pd >= 0); + + debuger = fork(); + ATF_REQUIRE(debuger >= 0); + if (debuger == 0) { + /* The traced process is reparented to debuger. */ + ATF_REQUIRE(ptrace(PT_ATTACH, traced, 0, 0) == 0); + wpid = waitpid(traced, &status, 0); + ATF_REQUIRE(wpid == traced); + ATF_REQUIRE(WIFSTOPPED(status)); + ATF_REQUIRE(WSTOPSIG(status) == SIGSTOP); + + /* Allow process to die. */ + ATF_REQUIRE(ptrace(PT_CONTINUE, traced, (caddr_t)1, 0) == 0); + wpid = waitpid(traced, &status, 0); + ATF_REQUIRE(wpid == traced); + ATF_REQUIRE(WIFEXITED(status)); + ATF_REQUIRE(WEXITSTATUS(status) == 0); + + /* Reparent back to the orginal process. */ + ATF_REQUIRE(close(pd) == 0); + exit(0); + } + + wpid = waitpid(debuger, &status, 0); + ATF_REQUIRE(wpid == debuger); + ATF_REQUIRE(WEXITSTATUS(status) == 0); + + /* + * We have a child but it has a process descriptori + * so we should not be able to collect it process. + */ + wpid = wait(&status); + ATF_REQUIRE(wpid == -1); + ATF_REQUIRE(errno == ECHILD); + + ATF_REQUIRE(close(pd) == 0); +} + ATF_TP_ADD_TCS(tp) { @@ -4198,6 +4297,8 @@ ATF_TP_ADD_TCS(tp) ATF_TP_ADD_TC(tp, ptrace__PT_LWPINFO_stale_siginfo); ATF_TP_ADD_TC(tp, ptrace__syscall_args); ATF_TP_ADD_TC(tp, ptrace__proc_reparent); + ATF_TP_ADD_TC(tp, ptrace__procdesc_wait_child); + ATF_TP_ADD_TC(tp, ptrace__procdesc_reparent_wait_child); return (atf_no_error()); }
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201911251833.xAPIXM9h040583>