There are a few system calls (pselect, ppoll, etc) which replace a task
sigmask while they are running in a kernel-space
When a task calls one of these syscalls, the kernel saves a current
sigmask in task->saved_sigmask and sets a syscall sigmask.
On syscall-exit-stop, ptrace traps a task before restoring the
saved_sigmask, so PTRACE_GETSIGMASK returns the syscall sigmask and
PTRACE_SETSIGMASK does nothing, because its sigmask is replaced by
saved_sigmask, when the task returns to user-space.
This patch fixes this problem. PTRACE_GET_SIGMASK returns saved_sigmask
is it's set. PTRACE_SETSIGMASK drops the TIF_RESTORE_SIGMASK flag.
Link: http://lkml.kernel.org/r/[email protected]
Fixes: 29000caecbe8 ("ptrace: add ability to get/set signal-blocked mask")
Signed-off-by: Andrei Vagin <[email protected]>
Cc: Oleg Nesterov <[email protected]>
Cc: "Eric W. Biederman" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Stephen Rothwell <[email protected]>
---
Hello Andrew,
This patch is in a queue for a long time and it looks like you are
waiting ack from Oleg. He said the patch is fine:
https://www.spinics.net/lists/kernel/msg2972154.html
And promised to send a formal ack in reply to this email.
include/linux/sched/signal.h | 18 ++++++++++++++++++
kernel/ptrace.c | 15 +++++++++++++--
2 files changed, 31 insertions(+), 2 deletions(-)
diff --git a/include/linux/sched/signal.h b/include/linux/sched/signal.h
index ae5655197698..e412c092c1e8 100644
--- a/include/linux/sched/signal.h
+++ b/include/linux/sched/signal.h
@@ -418,10 +418,20 @@ static inline void set_restore_sigmask(void)
set_thread_flag(TIF_RESTORE_SIGMASK);
WARN_ON(!test_thread_flag(TIF_SIGPENDING));
}
+
+static inline void clear_tsk_restore_sigmask(struct task_struct *tsk)
+{
+ clear_tsk_thread_flag(tsk, TIF_RESTORE_SIGMASK);
+}
+
static inline void clear_restore_sigmask(void)
{
clear_thread_flag(TIF_RESTORE_SIGMASK);
}
+static inline bool test_tsk_restore_sigmask(struct task_struct *tsk)
+{
+ return test_tsk_thread_flag(tsk, TIF_RESTORE_SIGMASK);
+}
static inline bool test_restore_sigmask(void)
{
return test_thread_flag(TIF_RESTORE_SIGMASK);
@@ -439,6 +449,10 @@ static inline void set_restore_sigmask(void)
current->restore_sigmask = true;
WARN_ON(!test_thread_flag(TIF_SIGPENDING));
}
+static inline void clear_tsk_restore_sigmask(struct task_struct *tsk)
+{
+ tsk->restore_sigmask = false;
+}
static inline void clear_restore_sigmask(void)
{
current->restore_sigmask = false;
@@ -447,6 +461,10 @@ static inline bool test_restore_sigmask(void)
{
return current->restore_sigmask;
}
+static inline bool test_tsk_restore_sigmask(struct task_struct *tsk)
+{
+ return tsk->restore_sigmask;
+}
static inline bool test_and_clear_restore_sigmask(void)
{
if (!current->restore_sigmask)
diff --git a/kernel/ptrace.c b/kernel/ptrace.c
index 771e93f9c43f..6f357f4fc859 100644
--- a/kernel/ptrace.c
+++ b/kernel/ptrace.c
@@ -29,6 +29,7 @@
#include <linux/hw_breakpoint.h>
#include <linux/cn_proc.h>
#include <linux/compat.h>
+#include <linux/sched/signal.h>
/*
* Access another process' address space via ptrace.
@@ -924,18 +925,26 @@ int ptrace_request(struct task_struct *child, long request,
ret = ptrace_setsiginfo(child, &siginfo);
break;
- case PTRACE_GETSIGMASK:
+ case PTRACE_GETSIGMASK: {
+ sigset_t *mask;
+
if (addr != sizeof(sigset_t)) {
ret = -EINVAL;
break;
}
- if (copy_to_user(datavp, &child->blocked, sizeof(sigset_t)))
+ if (test_tsk_restore_sigmask(child))
+ mask = &child->saved_sigmask;
+ else
+ mask = &child->blocked;
+
+ if (copy_to_user(datavp, mask, sizeof(sigset_t)))
ret = -EFAULT;
else
ret = 0;
break;
+ }
case PTRACE_SETSIGMASK: {
sigset_t new_set;
@@ -961,6 +970,8 @@ int ptrace_request(struct task_struct *child, long request,
child->blocked = new_set;
spin_unlock_irq(&child->sighand->siglock);
+ clear_tsk_restore_sigmask(child);
+
ret = 0;
break;
}
--
2.20.1
On Tue, Mar 19, 2019 at 12:19:57PM -0700, Andrei Vagin wrote:
> There are a few system calls (pselect, ppoll, etc) which replace a task
> sigmask while they are running in a kernel-space
>
> When a task calls one of these syscalls, the kernel saves a current
> sigmask in task->saved_sigmask and sets a syscall sigmask.
>
> On syscall-exit-stop, ptrace traps a task before restoring the
> saved_sigmask, so PTRACE_GETSIGMASK returns the syscall sigmask and
> PTRACE_SETSIGMASK does nothing, because its sigmask is replaced by
> saved_sigmask, when the task returns to user-space.
>
> This patch fixes this problem. PTRACE_GET_SIGMASK returns saved_sigmask
> is it's set. PTRACE_SETSIGMASK drops the TIF_RESTORE_SIGMASK flag.
If it's not too late, could somebody tweak the commit message so that
PTRACE_GET_SIGMASK becomes PTRACE_GETSIGMASK and "is it's set" is changed
to "if it's set", please?
--
ldv
On Wed, 20 Mar 2019 01:32:53 +0300 "Dmitry V. Levin" <[email protected]> wrote:
> On Tue, Mar 19, 2019 at 12:19:57PM -0700, Andrei Vagin wrote:
> > There are a few system calls (pselect, ppoll, etc) which replace a task
> > sigmask while they are running in a kernel-space
> >
> > When a task calls one of these syscalls, the kernel saves a current
> > sigmask in task->saved_sigmask and sets a syscall sigmask.
> >
> > On syscall-exit-stop, ptrace traps a task before restoring the
> > saved_sigmask, so PTRACE_GETSIGMASK returns the syscall sigmask and
> > PTRACE_SETSIGMASK does nothing, because its sigmask is replaced by
> > saved_sigmask, when the task returns to user-space.
> >
> > This patch fixes this problem. PTRACE_GET_SIGMASK returns saved_sigmask
> > is it's set. PTRACE_SETSIGMASK drops the TIF_RESTORE_SIGMASK flag.
>
> If it's not too late, could somebody tweak the commit message so that
> PTRACE_GET_SIGMASK becomes PTRACE_GETSIGMASK and "is it's set" is changed
> to "if it's set", please?
I made those changes to my copy.
On 03/19, Andrei Vagin wrote:
>
> There are a few system calls (pselect, ppoll, etc) which replace a task
> sigmask while they are running in a kernel-space
>
> When a task calls one of these syscalls, the kernel saves a current
> sigmask in task->saved_sigmask and sets a syscall sigmask.
>
> On syscall-exit-stop, ptrace traps a task before restoring the
> saved_sigmask, so PTRACE_GETSIGMASK returns the syscall sigmask and
> PTRACE_SETSIGMASK does nothing, because its sigmask is replaced by
> saved_sigmask, when the task returns to user-space.
>
> This patch fixes this problem. PTRACE_GET_SIGMASK returns saved_sigmask
> is it's set. PTRACE_SETSIGMASK drops the TIF_RESTORE_SIGMASK flag.
Acked-by: Oleg Nesterov <[email protected]>