Re: [PATCH 1/1] exec: do not sleep in TASK_TRACED under ->cred_guard_mutex

[Oleg Nesterov <oleg@redhat.com>]

On 09/04, David Howells wrote:
>
> Oleg Nesterov <oleg@redhat.com> wrote:
>
> > But I strongly believe we should blame another patch
> >
> > 	"CRED: Make execve() take advantage of copy-on-write credentials"
> > 	a6f76f23d297f70e2a6b3ec607f7aeeea9e37e8d
> >
> > The tracee must not sleep in TASK_TRACED holding this mutex (it was named
> > cred_exec_mutex). Even if we remove ->cred_guard_mutex from mm_for_maps()
> > and proc_pid_attr_write(), another task doing PTRACE_ATTACH should not
> > hang until it is killed or the tracee resumes.

(Argh. Sorry David, the changelog should have mentioned tracehook_report_exec()
 more explicitely).

So, David, do you agree with this patch? Do you think it can go to 2.6.31 ?


> Btw, should mm_for_maps() use mutex_lock_interruptible()?  There doesn't seem
> any point making it non-interruptible (except for kill signals) - unless that
> would muck up seqfile handling.

Perhaps, but we should change m_start() first, it should check IS_ERR() instead
of mm != NULL. Afaics, vfs_read()->seq_read() path will return ERESTART...
correctly.

I am not sure would be right though, short reads can confuse user space. And
this can't solve the problem, this only helps to react to signals.

Oleg.