From: Oleg Nesterov <oleg@redhat.com>
To: Andy Lutomirski <luto@amacapital.net>
Cc: Roman Pen <roman.penyaev@profitbricks.com>,
Andy Lutomirski <luto@kernel.org>,
Peter Zijlstra <peterz@infradead.org>,
Thomas Gleixner <tglx@linutronix.de>,
Ingo Molnar <mingo@redhat.com>, Tejun Heo <tj@kernel.org>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v3 1/1] kthread: allocate kthread structure using kmalloc
Date: Wed, 26 Oct 2016 16:14:00 +0200 [thread overview]
Message-ID: <20161026141359.GA6893@redhat.com> (raw)
In-Reply-To: <CALCETrVmczve=knhYGrcXkeAXRQR_oY4ac2BSc7V0TEnx6-GFA@mail.gmail.com>
On 10/25, Andy Lutomirski wrote:
>
> Would it perhaps make sense to do something like Roman's patch for 4.9
> and then consider further changes down the road?
OK, lets make it kmalloc'ed. Please see my old patch below, slightly changed.
I'll send it after I do some testing and write the changelog. No separate
refcounting, no complications.
Some notes right now. Of course, with this patch we are ready to remove
put_task_stack() from kthread.c right now. The next change should kill
to_live_kthread() altogether. And stop using ->vfork_done.
And. With this patch we do not need another "workqueue: ignore dead tasks
in a workqueue sleep hook" fix from Roman.
> Roman's patch
> appears to fix a real bug,
Well, it fixes the additional problems if we already have a bug, but I
agree this is a problem anyway.
> and I think that, while not really ideal,
> the code is an incredible mess right now and Roman's patch (assuming
> it's correct) makes it considerably nicer.
This is where I disagree with you and Roman. Yes, it needs cleanups and
only because of kthread on stack. But _IMO_ Roman's patch makes it much,
much worse and adds a lot of unnecessary complications.
Could you please look at the patch below?
Oleg.
---
diff --git a/include/linux/kthread.h b/include/linux/kthread.h
index a6e82a6..c1c3e63 100644
--- a/include/linux/kthread.h
+++ b/include/linux/kthread.h
@@ -48,6 +48,7 @@ struct task_struct *kthread_create_on_cpu(int (*threadfn)(void *data),
__k; \
})
+void free_kthread_struct(struct task_struct *k);
void kthread_bind(struct task_struct *k, unsigned int cpu);
void kthread_bind_mask(struct task_struct *k, const struct cpumask *mask);
int kthread_stop(struct task_struct *k);
diff --git a/kernel/fork.c b/kernel/fork.c
index 623259f..663c6a7 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -351,6 +351,8 @@ void free_task(struct task_struct *tsk)
ftrace_graph_exit_task(tsk);
put_seccomp_filter(tsk);
arch_release_task_struct(tsk);
+ if (tsk->flags & PF_KTHREAD)
+ free_kthread_struct(tsk);
free_task_struct(tsk);
}
EXPORT_SYMBOL(free_task);
diff --git a/kernel/kthread.c b/kernel/kthread.c
index be2cc1f..c6adbde 100644
--- a/kernel/kthread.c
+++ b/kernel/kthread.c
@@ -53,14 +53,34 @@ enum KTHREAD_BITS {
KTHREAD_IS_PARKED,
};
-#define __to_kthread(vfork) \
- container_of(vfork, struct kthread, exited)
+static inline void set_kthread_struct(void *kthread)
+{
+ /*
+ * We abuse ->set_child_tid to avoid the new member and because it
+ * can't be wrongly copied by copy_process(). We also rely on fact
+ * that the caller can't exec, so PF_KTHREAD can't be cleared.
+ */
+ current->set_child_tid = (__force void __user *)kthread;
+}
static inline struct kthread *to_kthread(struct task_struct *k)
{
- return __to_kthread(k->vfork_done);
+ WARN_ON(!(k->flags & PF_KTHREAD));
+ return (__force void *)k->set_child_tid;
}
+void free_kthread_struct(struct task_struct *k)
+{
+ kfree(to_kthread(k)); /* can be NULL if kmalloc() failed */
+}
+
+#define __to_kthread(vfork) \
+ container_of(vfork, struct kthread, exited)
+
+/*
+ * TODO: kill it and use to_kthread(). But we still need the users
+ * like kthread_stop() which has to sync with the exiting kthread.
+ */
static struct kthread *to_live_kthread(struct task_struct *k)
{
struct completion *vfork = ACCESS_ONCE(k->vfork_done);
@@ -181,14 +201,11 @@ static int kthread(void *_create)
int (*threadfn)(void *data) = create->threadfn;
void *data = create->data;
struct completion *done;
- struct kthread self;
+ struct kthread *self;
int ret;
- self.flags = 0;
- self.data = data;
- init_completion(&self.exited);
- init_completion(&self.parked);
- current->vfork_done = &self.exited;
+ self = kmalloc(sizeof(*self), GFP_KERNEL);
+ set_kthread_struct(self);
/* If user was SIGKILLed, I release the structure. */
done = xchg(&create->done, NULL);
@@ -196,6 +213,19 @@ static int kthread(void *_create)
kfree(create);
do_exit(-EINTR);
}
+
+ if (!self) {
+ create->result = ERR_PTR(-ENOMEM);
+ complete(done);
+ do_exit(-ENOMEM);
+ }
+
+ self->flags = 0;
+ self->data = data;
+ init_completion(&self->exited);
+ init_completion(&self->parked);
+ current->vfork_done = &self->exited;
+
/* OK, tell user we're spawned, wait for stop or wakeup */
__set_current_state(TASK_UNINTERRUPTIBLE);
create->result = current;
@@ -203,12 +233,10 @@ static int kthread(void *_create)
schedule();
ret = -EINTR;
-
- if (!test_bit(KTHREAD_SHOULD_STOP, &self.flags)) {
- __kthread_parkme(&self);
+ if (!test_bit(KTHREAD_SHOULD_STOP, &self->flags)) {
+ __kthread_parkme(self);
ret = threadfn(data);
}
- /* we can't just return, we must preserve "self" on stack */
do_exit(ret);
}
next prev parent reply other threads:[~2016-10-26 14:14 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-10-25 11:05 [PATCH v3 1/1] kthread: allocate kthread structure using kmalloc Roman Pen
2016-10-25 14:03 ` Oleg Nesterov
2016-10-25 15:43 ` Oleg Nesterov
2016-10-25 16:08 ` Roman Penyaev
2016-10-25 16:17 ` Oleg Nesterov
2016-10-25 16:52 ` Andy Lutomirski
2016-10-26 14:14 ` Oleg Nesterov [this message]
2016-10-26 14:57 ` Thomas Gleixner
2016-10-26 15:51 ` Oleg Nesterov
2016-10-26 18:31 ` Thomas Gleixner
2016-10-28 16:11 ` [PATCH 0/2] kthread: make struct kthread kmalloc'ed Oleg Nesterov
2016-10-28 16:11 ` [PATCH 1/2] " Oleg Nesterov
2016-10-28 18:48 ` Thomas Gleixner
2016-10-28 16:12 ` [PATCH 2/2] Revert "kthread: Pin the stack via try_get_task_stack()/put_task_stack() in to_live_kthread() function" Oleg Nesterov
2016-10-28 18:49 ` Thomas Gleixner
2016-10-28 18:44 ` [PATCH 0/2] kthread: make struct kthread kmalloc'ed Thomas Gleixner
2016-10-31 20:07 ` [PATCH 0/2] kthread: kill to_live_kthread() Oleg Nesterov
2016-10-31 20:07 ` [PATCH 1/2] kthread: don't use to_live_kthread() in kthread_stop() Oleg Nesterov
2016-11-09 7:58 ` Thomas Gleixner
2016-10-31 20:08 ` [PATCH 2/2] kthread: don't use to_live_kthread() in kthread_park() and kthread_unpark() Oleg Nesterov
2016-11-09 8:45 ` Thomas Gleixner
2016-11-09 17:27 ` Oleg Nesterov
2016-11-10 17:19 ` [PATCH 0/1] kthread: don't abuse kthread_create_on_cpu() in __kthread_create_worker() Oleg Nesterov
2016-11-10 17:20 ` [PATCH 1/1] " Oleg Nesterov
2016-11-14 11:12 ` Petr Mladek
2016-11-14 11:09 ` [PATCH 0/1] " Petr Mladek
2016-11-07 18:23 ` [PATCH 0/2] kthread: kill to_live_kthread() Andy Lutomirski
2016-10-26 16:13 ` [PATCH v3 1/1] kthread: allocate kthread structure using kmalloc Oleg Nesterov
2016-10-27 2:56 ` Josh Poimboeuf
2016-10-27 13:10 ` Oleg Nesterov
2016-10-25 15:46 ` Roman Penyaev
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=20161026141359.GA6893@redhat.com \
--to=oleg@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=luto@amacapital.net \
--cc=luto@kernel.org \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=roman.penyaev@profitbricks.com \
--cc=tglx@linutronix.de \
--cc=tj@kernel.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 NNTP newsgroup(s).