When __WCHILDSIGINFO is specified, copy the child's siginfo into the waitid infop instead of constructing the parent's SIGCHLD perspective. Cc: Christian Brauner <brauner@xxxxxxxxxx> Cc: "Eric W. Biederman" <ebiederm@xxxxxxxxxxxx> Cc: Andy Lutomirski <luto@xxxxxxxxxxxxxx> Cc: "Robert Święcki" <robert@xxxxxxxxxxx> Cc: Jann Horn <jannh@xxxxxxxxxx> Cc: Oleg Nesterov <oleg@xxxxxxxxxx> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx> Cc: linux-api@xxxxxxxxxxxxxxx Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx> --- include/linux/sched.h | 1 + include/uapi/linux/wait.h | 1 + kernel/exit.c | 23 +++++++++++++++++++---- kernel/signal.c | 4 ++++ 4 files changed, 25 insertions(+), 4 deletions(-) diff --git a/include/linux/sched.h b/include/linux/sched.h index f5b2be39a78c..e40789e801ef 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1178,6 +1178,7 @@ struct task_struct { #endif /* Ptrace state: */ unsigned long ptrace_message; + kernel_siginfo_t death_siginfo; kernel_siginfo_t *last_siginfo; struct task_io_accounting ioac; diff --git a/include/uapi/linux/wait.h b/include/uapi/linux/wait.h index 85b809fc9f11..7258cd4510ba 100644 --- a/include/uapi/linux/wait.h +++ b/include/uapi/linux/wait.h @@ -9,6 +9,7 @@ #define WCONTINUED 0x00000008 #define WNOWAIT 0x01000000 /* Don't reap, just poll status. */ +#define __WCHILDSIGINFO 0x10000000 /* Report child's siginfo. */ #define __WNOTHREAD 0x20000000 /* Don't wait on children of other threads in this group */ #define __WALL 0x40000000 /* Wait on all children, regardless of type */ #define __WCLONE 0x80000000 /* Wait only on non-SIGCHLD children */ diff --git a/kernel/exit.c b/kernel/exit.c index b00a25bb4ab9..de6e024976c6 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -953,6 +953,7 @@ struct waitid_info { uid_t uid; int status; int cause; + kernel_siginfo_t siginfo; }; struct wait_opts { @@ -964,7 +965,7 @@ struct wait_opts { int wo_stat; struct rusage *wo_rusage; - wait_queue_entry_t child_wait; + wait_queue_entry_t child_wait; int notask_error; }; @@ -1012,11 +1013,16 @@ static int wait_task_zombie(struct wait_opts *wo, struct task_struct *p) int state, status; pid_t pid = task_pid_vnr(p); uid_t uid = from_kuid_munged(current_user_ns(), task_uid(p)); - struct waitid_info *infop; + struct waitid_info *infop = wo->wo_info; if (!likely(wo->wo_flags & WEXITED)) return 0; + /* Before WNOWAIT so a copy can be extracted without reaping. */ + if (unlikely(wo->wo_flags & __WCHILDSIGINFO)) { + if (infop && p->last_siginfo) + copy_siginfo(&infop->siginfo, p->last_siginfo); + } if (unlikely(wo->wo_flags & WNOWAIT)) { status = (p->signal->flags & SIGNAL_GROUP_EXIT) ? p->signal->group_exit_code : p->exit_code; @@ -1121,7 +1127,6 @@ static int wait_task_zombie(struct wait_opts *wo, struct task_struct *p) release_task(p); out_info: - infop = wo->wo_info; if (infop) { if ((status & 0x7f) == 0) { infop->cause = CLD_EXITED; @@ -1564,7 +1569,7 @@ static long kernel_waitid(int which, pid_t upid, struct waitid_info *infop, unsigned int f_flags = 0; if (options & ~(WNOHANG|WNOWAIT|WEXITED|WSTOPPED|WCONTINUED| - __WNOTHREAD|__WCLONE|__WALL)) + __WNOTHREAD|__WCLONE|__WALL|__WCHILDSIGINFO)) return -EINVAL; if (!(options & (WEXITED|WSTOPPED|WCONTINUED))) return -EINVAL; @@ -1637,6 +1642,10 @@ SYSCALL_DEFINE5(waitid, int, which, pid_t, upid, struct siginfo __user *, if (!infop) return err; + /* __WCHILDSIGINFO */ + if (info.siginfo.si_signo) + return copy_siginfo_to_user(infop, &info.siginfo); + if (!user_write_access_begin(infop, sizeof(*infop))) return -EFAULT; @@ -1780,6 +1789,12 @@ COMPAT_SYSCALL_DEFINE5(waitid, if (!infop) return err; + /* __WCHILDSIGINFO */ + if (info.siginfo.si_signo) + return copy_siginfo_to_user32( + (struct compat_siginfo __user *)infop, + &info.siginfo); + if (!user_write_access_begin(infop, sizeof(*infop))) return -EFAULT; diff --git a/kernel/signal.c b/kernel/signal.c index 9b04631acde8..41f6ba6b7aa7 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -2825,6 +2825,10 @@ bool get_signal(struct ksignal *ksig) } fatal: + /* Allow siginfo to be queried until reaped. */ + copy_siginfo(¤t->death_siginfo, &ksig->info); + current->last_siginfo = ¤t->death_siginfo; + spin_unlock_irq(&sighand->siglock); if (unlikely(cgroup_task_frozen(current))) cgroup_leave_frozen(true); -- 2.30.2