Re: [REGRESSION] ptrace broken from "cgroup: cgroup v2 freezer" (76f969e)

[Oleg Nesterov]

On 05/12, Alex Xu (Hello71) wrote:
>
> Hi,
>
> I was trying to use strace recently and found that it exhibited some
> strange behavior. I produced this minimal test case:
>
> #include <unistd.h>
>
> int main() {
>     write(1, "a", 1);
>     return 0;
> }
>
> which, when run using "gcc test.c && strace ./a.out" produces this
> strace output:
>
> [ pre-main omitted ]
> write(1, "a", 1)                        = ? ERESTARTSYS (To be restarted if 
> SA_RESTART is set)
> write(1, "a", 1)                        = ? ERESTARTSYS (To be restarted if 
> SA_RESTART is set)
> write(1, "a", 1)                        = ? ERESTARTSYS (To be restarted if 
> SA_RESTART is set)
> write(1, "a", 1)                        = ? ERESTARTSYS (To be restarted if 
> SA_RESTART is set)
> write(1, "a", 1)                        = ? ERESTARTSYS (To be restarted if 
> SA_RESTART is set)
> write(1, "a", 1)                        = ? ERESTARTSYS (To be restarted if 
> SA_RESTART is set)
> [ repeats forever ]

Yes, cgroup_enter_frozen() alone is wrong, we have already discussed this
a bit... see https://lore.kernel.org/lkml/20190508152536.ga17...@redhat.com/

Probably we add leave_frozen(true) after freezable_schedule() for now, then
think try to make something better...

But I am not sure I 100% understand whats going on in this case, could you
try the patch below? (Just in case, of course it is wrong).

Oleg.

--- x/kernel/signal.c
+++ x/kernel/signal.c
@@ -149,8 +149,7 @@
 {
        if ((t->jobctl & (JOBCTL_PENDING_MASK | JOBCTL_TRAP_FREEZE)) ||
            PENDING(&t->pending, &t->blocked) ||
-           PENDING(&t->signal->shared_pending, &t->blocked) ||
-           cgroup_task_frozen(t)) {
+           PENDING(&t->signal->shared_pending, &t->blocked) {
                set_tsk_thread_flag(t, TIF_SIGPENDING);
                return true;
        }