From: Andrei Vagin <avagin@gmail.com>
To: Florian Weimer <fweimer@redhat.com>
Cc: Christian Brauner <brauner@kernel.org>,
Adhemerval Zanella <adhemerval.zanella@linaro.org>,
libc-alpha@sourceware.org, Alexey Izbyshev <izbyshev@ispras.ru>,
Carlos O'Donell <carlos@redhat.com>,
Dmitry Safonov <0x7f454c46@gmail.com>
Subject: Re: [PATCH v4 0/3] Linux: Fix posix_spawn when user with time namespaces
Date: Fri, 27 May 2022 17:03:14 -0700 [thread overview]
Message-ID: <YpFmwi3aUUzHjVBA@gmail.com> (raw)
In-Reply-To: <CANaxB-zHZMYpXRqbc85k7Af9QQCWORw23mgYLjH9ow7S8QrhfQ@mail.gmail.com>
On Fri, May 27, 2022 at 08:53:54AM -0700, Andrei Vagin wrote:
> On Wed, May 25, 2022 at 5:24 AM Florian Weimer <fweimer@redhat.com> wrote:
> >
> > * Christian Brauner:
> >
> > > On Tue, May 10, 2022 at 09:18:46PM +0200, Florian Weimer wrote:
> > >> * Adhemerval Zanella:
> > >>
> > >> > The patchset adds some support to tests the fallback code to
> > >> > use only use CLONE_VFORK. It uses unshare directly because
> > >> > it simpler than add container support.
> > >> >
> > >> > Adhemerval Zanella (3):
> > >> > linux: Add CLONE_NEWTIME from Linux 5.6 to bits/sched.h
> > >> > support: Add support_enter_time_namespace
> > >> > linux: Add fallback for clone failure on posix_spawn (BZ #29115)
> > >>
> > >> Christan, how likely is it that we'd get another time namespace variant
> > >> that would only become effective after execve (when the DSO is remapped
> > >> anyway)?
> > >
> > > Not unlikely if it helps you avoid a lot of complexity. I will need some
> > > time to track down Andrei and others to discuss though.
> >
> > Any progress with that? (I hope I guessed the right Andrei.)
>
> I think this is the right me. Have I missed something?
>
> >
> > Breaking vfork is really a bit of a hassle for us, and the workaround
> > code is quite non-trivial and will have to implemented across many
> > projects (not just glibc). An unshare request that takes effect on
> > execve only would really help.
>
> Is the problem that vfork fails if a process has half-entered a time namespace?
I like the idea of entering a time namespace on exec and don't fail
vfork. The only worry is that the behavior becomes more complicated and
less obvious.
Here is the untested patch:
diff --git a/fs/exec.c b/fs/exec.c
index 14b4b3755580..96f650ec7533 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -65,6 +65,7 @@
#include <linux/io_uring.h>
#include <linux/syscall_user_dispatch.h>
#include <linux/coredump.h>
+#include <linux/time_namespace.h>
#include <linux/uaccess.h>
#include <asm/mmu_context.h>
@@ -1266,6 +1267,7 @@ int begin_new_exec(struct linux_binprm * bprm)
if (retval)
goto out;
+ timens_on_fork(me->nsproxy, me);
/*
* Cancel any io_uring activity across execve
*/
diff --git a/kernel/fork.c b/kernel/fork.c
index 124829ed0163..653b80524a54 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -2030,15 +2030,6 @@ static __latent_entropy struct task_struct
*copy_process(
return ERR_PTR(-EINVAL);
}
- /*
- * If the new process will be in a different time namespace
- * do not allow it to share VM or a thread group with the
forking task.
- */
- if (clone_flags & (CLONE_THREAD | CLONE_VM)) {
- if (nsp->time_ns != nsp->time_ns_for_children)
- return ERR_PTR(-EINVAL);
- }
-
if (clone_flags & CLONE_PIDFD) {
/*
* - CLONE_DETACHED is blocked so that we can
* potentially
diff --git a/kernel/nsproxy.c b/kernel/nsproxy.c
index eec72ca962e2..b4cbb406bc28 100644
--- a/kernel/nsproxy.c
+++ b/kernel/nsproxy.c
@@ -179,7 +179,8 @@ int copy_namespaces(unsigned long flags, struct
task_struct *tsk)
if (IS_ERR(new_ns))
return PTR_ERR(new_ns);
- timens_on_fork(new_ns, tsk);
+ if ((flags & CLONE_VM) == 0)
+ timens_on_fork(new_ns, tsk);
tsk->nsproxy = new_ns;
return 0;
If this is what we want, I can prepare a final patch. But I will be on
vacation next week, so it will happen after it.
Thanks,
Andrei
next prev parent reply other threads:[~2022-05-28 0:03 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-05-10 19:11 Adhemerval Zanella
2022-05-10 19:11 ` [PATCH v4 1/3] linux: Add CLONE_NEWTIME from Linux 5.6 to bits/sched.h Adhemerval Zanella
2022-05-23 13:45 ` Adhemerval Zanella
2022-05-10 19:11 ` [PATCH v4 2/3] support: Add support_enter_time_namespace Adhemerval Zanella
2022-05-10 19:11 ` [PATCH v4 3/3] linux: Add fallback for clone failure on posix_spawn (BZ #29115) Adhemerval Zanella
2022-05-10 19:18 ` [PATCH v4 0/3] Linux: Fix posix_spawn when user with time namespaces Florian Weimer
2022-05-11 9:21 ` Christian Brauner
2022-05-11 16:01 ` Alexey Izbyshev
2022-05-25 12:24 ` Florian Weimer
2022-05-27 15:53 ` Andrei Vagin
2022-05-28 0:03 ` Andrei Vagin [this message]
2022-05-29 14:33 ` Christian Brauner
2022-05-30 12:58 ` Florian Weimer
2022-06-08 7:21 ` Andrei Vagin
2022-06-08 8:35 ` Florian Weimer
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=YpFmwi3aUUzHjVBA@gmail.com \
--to=avagin@gmail.com \
--cc=0x7f454c46@gmail.com \
--cc=adhemerval.zanella@linaro.org \
--cc=brauner@kernel.org \
--cc=carlos@redhat.com \
--cc=fweimer@redhat.com \
--cc=izbyshev@ispras.ru \
--cc=libc-alpha@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).