public inbox for [email protected]
 help / color / mirror / Atom feed
From: "Michael S. Tsirkin" <[email protected]>
To: Christoph Hellwig <[email protected]>
Cc: Linus Torvalds <[email protected]>,
	Andrew Morton <[email protected]>,
	Al Viro <[email protected]>,
	Felix Kuehling <[email protected]>,
	Alex Deucher <[email protected]>,
	Zhenyu Wang <[email protected]>,
	Zhi Wang <[email protected]>, Felipe Balbi <[email protected]>,
	Jason Wang <[email protected]>, Jens Axboe <[email protected]>,
	[email protected], [email protected],
	[email protected],
	[email protected], [email protected],
	[email protected],
	[email protected], [email protected],
	[email protected]
Subject: Re: [PATCH 6/6] kernel: set USER_DS in kthread_use_mm
Date: Mon, 6 Apr 2020 17:49:34 -0400	[thread overview]
Message-ID: <[email protected]> (raw)
In-Reply-To: <[email protected]>

On Sat, Apr 04, 2020 at 11:41:01AM +0200, Christoph Hellwig wrote:
> Some architectures like arm64 and s390 require USER_DS to be set for
> kernel threads to access user address space, which is the whole purpose
> of kthread_use_mm, but other like x86 don't.  That has lead to a huge
> mess where some callers are fixed up once they are tested on said
> architectures, while others linger around and yet other like io_uring
> try to do "clever" optimizations for what usually is just a trivial
> asignment to a member in the thread_struct for most architectures.
> 
> Make kthread_use_mm set USER_DS, and kthread_unuse_mm restore to the
> previous value instead.
> 
> Signed-off-by: Christoph Hellwig <[email protected]>

I'm ok with vhost bits:

Acked-by: Michael S. Tsirkin <[email protected]>

> ---
>  drivers/usb/gadget/function/f_fs.c | 4 ----
>  drivers/vhost/vhost.c              | 3 ---
>  fs/io-wq.c                         | 8 ++------
>  fs/io_uring.c                      | 4 ----
>  kernel/kthread.c                   | 6 ++++++
>  5 files changed, 8 insertions(+), 17 deletions(-)
> 
> diff --git a/drivers/usb/gadget/function/f_fs.c b/drivers/usb/gadget/function/f_fs.c
> index d9e48bd7c692..a1198f4c527c 100644
> --- a/drivers/usb/gadget/function/f_fs.c
> +++ b/drivers/usb/gadget/function/f_fs.c
> @@ -824,13 +824,9 @@ static void ffs_user_copy_worker(struct work_struct *work)
>  	bool kiocb_has_eventfd = io_data->kiocb->ki_flags & IOCB_EVENTFD;
>  
>  	if (io_data->read && ret > 0) {
> -		mm_segment_t oldfs = get_fs();
> -
> -		set_fs(USER_DS);
>  		kthread_use_mm(io_data->mm);
>  		ret = ffs_copy_to_iter(io_data->buf, ret, &io_data->data);
>  		kthread_unuse_mm(io_data->mm);
> -		set_fs(oldfs);
>  	}
>  
>  	io_data->kiocb->ki_complete(io_data->kiocb, ret, ret);
> diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
> index 1787d426a956..b5229ae01d3b 100644
> --- a/drivers/vhost/vhost.c
> +++ b/drivers/vhost/vhost.c
> @@ -333,9 +333,7 @@ static int vhost_worker(void *data)
>  	struct vhost_dev *dev = data;
>  	struct vhost_work *work, *work_next;
>  	struct llist_node *node;
> -	mm_segment_t oldfs = get_fs();
>  
> -	set_fs(USER_DS);
>  	kthread_use_mm(dev->mm);
>  
>  	for (;;) {
> @@ -365,7 +363,6 @@ static int vhost_worker(void *data)
>  		}
>  	}
>  	kthread_unuse_mm(dev->mm);
> -	set_fs(oldfs);
>  	return 0;
>  }
>  
> diff --git a/fs/io-wq.c b/fs/io-wq.c
> index 83c2868eff2a..75cc2f31816d 100644
> --- a/fs/io-wq.c
> +++ b/fs/io-wq.c
> @@ -168,7 +168,6 @@ static bool __io_worker_unuse(struct io_wqe *wqe, struct io_worker *worker)
>  			dropped_lock = true;
>  		}
>  		__set_current_state(TASK_RUNNING);
> -		set_fs(KERNEL_DS);
>  		kthread_unuse_mm(worker->mm);
>  		mmput(worker->mm);
>  		worker->mm = NULL;
> @@ -420,14 +419,11 @@ static void io_wq_switch_mm(struct io_worker *worker, struct io_wq_work *work)
>  		mmput(worker->mm);
>  		worker->mm = NULL;
>  	}
> -	if (!work->mm) {
> -		set_fs(KERNEL_DS);
> +	if (!work->mm)
>  		return;
> -	}
> +
>  	if (mmget_not_zero(work->mm)) {
>  		kthread_use_mm(work->mm);
> -		if (!worker->mm)
> -			set_fs(USER_DS);
>  		worker->mm = work->mm;
>  		/* hang on to this mm */
>  		work->mm = NULL;
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index 367406381044..c332a34e8b34 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -5871,15 +5871,12 @@ static int io_sq_thread(void *data)
>  	struct io_ring_ctx *ctx = data;
>  	struct mm_struct *cur_mm = NULL;
>  	const struct cred *old_cred;
> -	mm_segment_t old_fs;
>  	DEFINE_WAIT(wait);
>  	unsigned long timeout;
>  	int ret = 0;
>  
>  	complete(&ctx->completions[1]);
>  
> -	old_fs = get_fs();
> -	set_fs(USER_DS);
>  	old_cred = override_creds(ctx->creds);
>  
>  	timeout = jiffies + ctx->sq_thread_idle;
> @@ -5985,7 +5982,6 @@ static int io_sq_thread(void *data)
>  	if (current->task_works)
>  		task_work_run();
>  
> -	set_fs(old_fs);
>  	if (cur_mm) {
>  		kthread_unuse_mm(cur_mm);
>  		mmput(cur_mm);
> diff --git a/kernel/kthread.c b/kernel/kthread.c
> index 316db17f6b4f..9e27d01b6d78 100644
> --- a/kernel/kthread.c
> +++ b/kernel/kthread.c
> @@ -52,6 +52,7 @@ struct kthread {
>  	unsigned long flags;
>  	unsigned int cpu;
>  	void *data;
> +	mm_segment_t oldfs;
>  	struct completion parked;
>  	struct completion exited;
>  #ifdef CONFIG_BLK_CGROUP
> @@ -1235,6 +1236,9 @@ void kthread_use_mm(struct mm_struct *mm)
>  
>  	if (active_mm != mm)
>  		mmdrop(active_mm);
> +
> +	to_kthread(tsk)->oldfs = get_fs();
> +	set_fs(USER_DS);
>  }
>  EXPORT_SYMBOL_GPL(kthread_use_mm);
>  
> @@ -1249,6 +1253,8 @@ void kthread_unuse_mm(struct mm_struct *mm)
>  	WARN_ON_ONCE(!(tsk->flags & PF_KTHREAD));
>  	WARN_ON_ONCE(!tsk->mm);
>  
> +	set_fs(to_kthread(tsk)->oldfs);
> +
>  	task_lock(tsk);
>  	sync_mm_rss(mm);
>  	tsk->mm = NULL;
> -- 
> 2.25.1


      reply	other threads:[~2020-04-06 21:49 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-04-04  9:40 improve use_mm / unuse_mm Christoph Hellwig
2020-04-04  9:40 ` [PATCH 1/6] amdgpu: a NULL ->mm does not mean a thread is a kthread Christoph Hellwig
2020-04-06 16:07   ` Felix Kuehling
2020-04-04  9:40 ` [PATCH 2/6] i915/gvt/kvm: " Christoph Hellwig
2020-04-04 10:05   ` Sergei Shtylyov
2020-04-07  3:08   ` Yan Zhao
2020-04-13 13:27     ` Christoph Hellwig
2020-04-14  0:04       ` Yan Zhao
2020-04-14  7:00         ` Christoph Hellwig
2020-04-14  7:03           ` Yan Zhao
2020-04-04  9:40 ` [PATCH 3/6] i915/gvt: remove unused xen bits Christoph Hellwig
2020-04-08  1:44   ` Zhenyu Wang
2020-04-13 13:08     ` Christoph Hellwig
2020-04-14  3:04       ` Zhenyu Wang
2020-04-04  9:40 ` [PATCH 4/6] kernel: move use_mm/unuse_mm to kthread.c Christoph Hellwig
2020-04-06 16:09   ` Felix Kuehling
2020-04-04  9:41 ` [PATCH 5/6] kernel: better document the use_mm/unuse_mm API contract Christoph Hellwig
2020-04-04 13:07   ` [Intel-gfx] " Pavel Begunkov
2020-04-06 16:10   ` Felix Kuehling
2020-04-04  9:41 ` [PATCH 6/6] kernel: set USER_DS in kthread_use_mm Christoph Hellwig
2020-04-06 21:49   ` Michael S. Tsirkin [this message]

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 \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    [email protected] \
    /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