fs/proc: Report eip/esp in /prod/PID/stat for coredumping
authorJohn Ogness <john.ogness@linutronix.de>
Thu, 14 Sep 2017 09:42:17 +0000 (11:42 +0200)
committerThomas Gleixner <tglx@linutronix.de>
Fri, 15 Sep 2017 21:31:16 +0000 (23:31 +0200)
Commit 0a1eb2d474ed ("fs/proc: Stop reporting eip and esp in
/proc/PID/stat") stopped reporting eip/esp because it is
racy and dangerous for executing tasks. The comment adds:

    As far as I know, there are no use programs that make any
    material use of these fields, so just get rid of them.

However, existing userspace core-dump-handler applications (for
example, minicoredumper) are using these fields since they
provide an excellent cross-platform interface to these valuable
pointers. So that commit introduced a user space visible
regression.

Partially revert the change and make the readout possible for
tasks with the proper permissions and only if the target task
has the PF_DUMPCORE flag set.

Fixes: 0a1eb2d474ed ("fs/proc: Stop reporting eip and esp in> /proc/PID/stat")
Reported-by: Marco Felsch <marco.felsch@preh.de>
Signed-off-by: John Ogness <john.ogness@linutronix.de>
Reviewed-by: Andy Lutomirski <luto@kernel.org>
Cc: Tycho Andersen <tycho.andersen@canonical.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: stable@vger.kernel.org
Cc: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Linux API <linux-api@vger.kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/87poatfwg6.fsf@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
fs/proc/array.c

index 88c355574aa0afe3b1a62b371b048af8d948070e..525157ca25cbe6a706ecb05c631b2b6d0bcb59a9 100644 (file)
@@ -62,6 +62,7 @@
 #include <linux/mman.h>
 #include <linux/sched/mm.h>
 #include <linux/sched/numa_balancing.h>
+#include <linux/sched/task_stack.h>
 #include <linux/sched/task.h>
 #include <linux/sched/cputime.h>
 #include <linux/proc_fs.h>
@@ -421,7 +422,15 @@ static int do_task_stat(struct seq_file *m, struct pid_namespace *ns,
                 * esp and eip are intentionally zeroed out.  There is no
                 * non-racy way to read them without freezing the task.
                 * Programs that need reliable values can use ptrace(2).
+                *
+                * The only exception is if the task is core dumping because
+                * a program is not able to use ptrace(2) in that case. It is
+                * safe because the task has stopped executing permanently.
                 */
+               if (permitted && (task->flags & PF_DUMPCORE)) {
+                       eip = KSTK_EIP(task);
+                       esp = KSTK_ESP(task);
+               }
        }
 
        get_task_comm(tcomm, task);