Message-ID: <alpine.DEB.2.22.394.2211301657200.4039@ubuntu-linux-20-04-desktop>
Date:   Wed, 30 Nov 2022 17:08:06 -0800 (PST)
From:   Stefano Stabellini <sstabellini@...nel.org>
To:     Roger Pau Monne <roger.pau@...rix.com>
cc:     linux-kernel@...r.kernel.org, xen-devel@...ts.xenproject.org,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Jiri Slaby <jirislaby@...nel.org>,
        Juergen Gross <jgross@...e.com>,
        Jan Beulich <jbeulich@...e.com>,
        Boris Ostrovsky <boris.ostrovsky@...cle.com>,
        Stefano Stabellini <sstabellini@...nel.org>,
        Chris Wright <chrisw@...s-sol.org>,
        Ingo Molnar <mingo@...e.hu>,
        Jeremy Fitzhardinge <jeremy@...source.com>,
        Olof Johansson <olof@...om.net>, linuxppc-dev@...ts.ozlabs.org
Subject: Re: [PATCH v2] hvc/xen: prevent concurrent accesses to the shared
 ring

On Wed, 30 Nov 2022, Roger Pau Monne wrote:
> The hvc machinery registers both a console and a tty device based on
> the hv ops provided by the specific implementation.  Those two
> interfaces however have different locks, and there's no single locks
> that's shared between the tty and the console implementations, hence
> the driver needs to protect itself against concurrent accesses.
> Otherwise concurrent calls using the split interfaces are likely to
> corrupt the ring indexes, leaving the console unusable.
> 
> Introduce a lock to xencons_info to serialize accesses to the shared
> ring.  This is only required when using the shared memory console,
> concurrent accesses to the hypercall based console implementation are
> not an issue.
> 
> Note the conditional logic in domU_read_console() is slightly modified
> so the notify_daemon() call can be done outside of the locked region:
> it's an hypercall and there's no need for it to be done with the lock
> held.

For domU_read_console: I don't mean to block this patch but we need to
be sure about the semantics of hv_ops.get_chars. Either it is expected
to be already locked, then we definitely shouldn't add another lock to
domU_read_console. Or it is not expected to be already locked, then we
should add the lock.

My impression is that it is expected to be already locked, but I think
we need Greg or Jiri to confirm one way or the other.

Aside from that the rest looks fine.



> Fixes: b536b4b96230 ('xen: use the hvc console infrastructure for Xen console')
> Signed-off-by: Roger Pau Monné <roger.pau@...rix.com>
> ---
> Changes since v1:
>  - Properly initialize the introduced lock in all paths.
> ---
>  drivers/tty/hvc/hvc_xen.c | 19 +++++++++++++++++--
>  1 file changed, 17 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/tty/hvc/hvc_xen.c b/drivers/tty/hvc/hvc_xen.c
> index 7c23112dc923..e63c1761a361 100644
> --- a/drivers/tty/hvc/hvc_xen.c
> +++ b/drivers/tty/hvc/hvc_xen.c
> @@ -43,6 +43,7 @@ struct xencons_info {
>  	int irq;
>  	int vtermno;
>  	grant_ref_t gntref;
> +	spinlock_t ring_lock;
>  };
>  
>  static LIST_HEAD(xenconsoles);
> @@ -84,12 +85,15 @@ static int __write_console(struct xencons_info *xencons,
>  	XENCONS_RING_IDX cons, prod;
>  	struct xencons_interface *intf = xencons->intf;
>  	int sent = 0;
> +	unsigned long flags;
>  
> +	spin_lock_irqsave(&xencons->ring_lock, flags);
>  	cons = intf->out_cons;
>  	prod = intf->out_prod;
>  	mb();			/* update queue values before going on */
>  
>  	if ((prod - cons) > sizeof(intf->out)) {
> +		spin_unlock_irqrestore(&xencons->ring_lock, flags);
>  		pr_err_once("xencons: Illegal ring page indices");
>  		return -EINVAL;
>  	}
> @@ -99,6 +103,7 @@ static int __write_console(struct xencons_info *xencons,
>  
>  	wmb();			/* write ring before updating pointer */
>  	intf->out_prod = prod;
> +	spin_unlock_irqrestore(&xencons->ring_lock, flags);
>  
>  	if (sent)
>  		notify_daemon(xencons);
> @@ -141,16 +146,19 @@ static int domU_read_console(uint32_t vtermno, char *buf, int len)
>  	int recv = 0;
>  	struct xencons_info *xencons = vtermno_to_xencons(vtermno);
>  	unsigned int eoiflag = 0;
> +	unsigned long flags;
>  
>  	if (xencons == NULL)
>  		return -EINVAL;
>  	intf = xencons->intf;
>  
> +	spin_lock_irqsave(&xencons->ring_lock, flags);
>  	cons = intf->in_cons;
>  	prod = intf->in_prod;
>  	mb();			/* get pointers before reading ring */
>  
>  	if ((prod - cons) > sizeof(intf->in)) {
> +		spin_unlock_irqrestore(&xencons->ring_lock, flags);
>  		pr_err_once("xencons: Illegal ring page indices");
>  		return -EINVAL;
>  	}
> @@ -174,10 +182,13 @@ static int domU_read_console(uint32_t vtermno, char *buf, int len)
>  		xencons->out_cons = intf->out_cons;
>  		xencons->out_cons_same = 0;
>  	}
> +	if (!recv && xencons->out_cons_same++ > 1) {
> +		eoiflag = XEN_EOI_FLAG_SPURIOUS;
> +	}
> +	spin_unlock_irqrestore(&xencons->ring_lock, flags);
> +
>  	if (recv) {
>  		notify_daemon(xencons);
> -	} else if (xencons->out_cons_same++ > 1) {
> -		eoiflag = XEN_EOI_FLAG_SPURIOUS;
>  	}
>  
>  	xen_irq_lateeoi(xencons->irq, eoiflag);
> @@ -234,6 +245,7 @@ static int xen_hvm_console_init(void)
>  		info = kzalloc(sizeof(struct xencons_info), GFP_KERNEL);
>  		if (!info)
>  			return -ENOMEM;
> +		spin_lock_init(&info->ring_lock);
>  	} else if (info->intf != NULL) {
>  		/* already configured */
>  		return 0;
> @@ -270,6 +282,7 @@ static int xen_hvm_console_init(void)
>  
>  static int xencons_info_pv_init(struct xencons_info *info, int vtermno)
>  {
> +	spin_lock_init(&info->ring_lock);
>  	info->evtchn = xen_start_info->console.domU.evtchn;
>  	/* GFN == MFN for PV guest */
>  	info->intf = gfn_to_virt(xen_start_info->console.domU.mfn);
> @@ -318,6 +331,7 @@ static int xen_initial_domain_console_init(void)
>  		info = kzalloc(sizeof(struct xencons_info), GFP_KERNEL);
>  		if (!info)
>  			return -ENOMEM;
> +		spin_lock_init(&info->ring_lock);
>  	}
>  
>  	info->irq = bind_virq_to_irq(VIRQ_CONSOLE, 0, false);
> @@ -472,6 +486,7 @@ static int xencons_probe(struct xenbus_device *dev,
>  	info = kzalloc(sizeof(struct xencons_info), GFP_KERNEL);
>  	if (!info)
>  		return -ENOMEM;
> +	spin_lock_init(&info->ring_lock);
>  	dev_set_drvdata(&dev->dev, info);
>  	info->xbdev = dev;
>  	info->vtermno = xenbus_devid_to_vtermno(devid);
> -- 
> 2.37.3
> 

Powered by blists - more mailing lists