Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Sun, 12 May 2019 21:20:12 -0400
From:   "Alex Xu (Hello71)" <alex_y_xu@yahoo.ca>
Subject: [REGRESSION] ptrace broken from "cgroup: cgroup v2 freezer" (76f969e)
To:     linux-kernel@vger.kernel.org, tj@kernel.org, guro@fb.com
Cc:     oleg@redhat.com, kernel-team@fb.com
MIME-Version: 1.0
User-Agent: astroid/0.14.0 (https://github.com/astroidmail/astroid)
Message-Id: <1557709124.798rxdb4l3.astroid@alex-desktop.none>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Hi,

I was trying to use strace recently and found that it exhibited some=20
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=20
strace output:

[ pre-main omitted ]
write(1, "a", 1)                        =3D ? ERESTARTSYS (To be restarted =
if SA_RESTART is set)
write(1, "a", 1)                        =3D ? ERESTARTSYS (To be restarted =
if SA_RESTART is set)
write(1, "a", 1)                        =3D ? ERESTARTSYS (To be restarted =
if SA_RESTART is set)
write(1, "a", 1)                        =3D ? ERESTARTSYS (To be restarted =
if SA_RESTART is set)
write(1, "a", 1)                        =3D ? ERESTARTSYS (To be restarted =
if SA_RESTART is set)
write(1, "a", 1)                        =3D ? ERESTARTSYS (To be restarted =
if SA_RESTART is set)
[ repeats forever ]

The correct result is of course:

[ pre-main omitted ]
write(1, "a", 1)                        =3D 1
exit_group(0)                           =3D ?
+++ exited with 0 +++

Strangely, this only occurs when outputting to a tty-like output.=20
Running "strace ./a.out" from a native Linux x86 console or a terminal=20
emulator causes the abnormal behavior. However, the following commands=20
work correctly:

- strace ./a.out >/dev/null
- strace ./a.out >/tmp/a # /tmp is a standard tmpfs
- strace ./a.out >&- # causes -1 EBADF (Bad file descriptor)

"strace -o /tmp/a ./a.out" hangs and produces the above (infinite)=20
output to /tmp/a.

I bisected this to 76f969e, "cgroup: cgroup v2 freezer". I reverted the=20
entire patchset (reverting only that one caused a conflict), which=20
resolved the issue. I skimmed the patch and came up with this=20
workaround, which also resolves the issue. I am not at all clear on the=20
technical workings of the patchset, but it seems to me like a process's=20
frozen status is supposed to be "suspended" when a frozen process is=20
ptraced, and "unsuspended" when ptracing ends. Therefore, it seems=20
suspicious to always "enter frozen" whether or not the cgroup is=20
actually frozen. It seems like the code should instead check if the=20
cgroup is actually frozen, and if so, restore the frozen status.

I am using systemd but not any other cgroup features. I tried in an=20
initramfs environment (no systemd, /init -> shell script) and reproduced=20
the failing test case.

Please CC me on replies.

Thanks,
Alex.

---
 kernel/signal.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/signal.c b/kernel/signal.c
index 62f9aea4a15a..47145d9d89ca 100644
--- a/kernel/signal.c
+++ b/kernel/signal.c
@@ -2110,7 +2110,7 @@ static void ptrace_stop(int exit_code, int why, int c=
lear_code, kernel_siginfo_t
                preempt_disable();
                read_unlock(&tasklist_lock);
                preempt_enable_no_resched();
-               cgroup_enter_frozen();
+               //cgroup_enter_frozen();
                freezable_schedule();
        } else {
                /*
@@ -2289,7 +2289,7 @@ static bool do_signal_stop(int signr)
                }
=20
                /* Now we don't run again until woken by SIGCONT or SIGKILL=
 */
-               cgroup_enter_frozen();
+               //cgroup_enter_frozen();
                freezable_schedule();
                return true;
        } else {
--=20
2.21.0