Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Mon, 27 Nov 2017 17:24:42 +0100
From:   Daniel Vetter <daniel@ffwll.ch>
To:     Arnd Bergmann <arnd@arndb.de>
Cc:     David Airlie <airlied@linux.ie>, linux-kernel@vger.kernel.org,
        dri-devel@lists.freedesktop.org
Subject: Re: [PATCH] drm/via: use monotonic time for VIA_WAIT_IRQ
Message-ID: <20171127162442.brlncy5wq3kntm45@phenom.ffwll.local>
Mail-Followup-To: Arnd Bergmann <arnd@arndb.de>,
        David Airlie <airlied@linux.ie>, linux-kernel@vger.kernel.org,
        dri-devel@lists.freedesktop.org
References: <20171127111728.1710742-1-arnd@arndb.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20171127111728.1710742-1-arnd@arndb.de>
User-Agent: NeoMutt/20170609 (1.8.3)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Nov 27, 2017 at 12:17:03PM +0100, Arnd Bergmann wrote:
> The normal DRM vblank interrupt events started using monotonic times in
> commit c61eef726a78 ("drm: add support for monotonic vblank timestamps"),
> which is useful for a number of reasons, including the possible y2038
> overflow.
> 
> The VIA_WAIT_IRQ ioctl uses the same drm_wait_vblank_reply as
> DRM_IOCTL_WAIT_VBLANK, but still uses wall-clock time.
> 
> This converts it to using monotonic time as well, which is more
> consistent, and avoids problems with the y2038 overflow as well
> as synchronization issues when the real time skips.
> 
> I could not find the matching user space that calls the VIA_WAIT_IRQ
> ioctl to verify that this is safe, but it very likely is. Please
> either test or review the user space side before applying this.

You need to check out _seriously_ old versions of mesa and X (plus it's
drivers). Git ftw.

The only caller I've found (in the Xv implementation of the via/openchrome
ddx) doesn't even care about the timestamp. I'm mildly tempted to just
report 0 always, but via isn't worth that much bother really.

Applied, thanks for the patch.
-Daniel

> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  drivers/gpu/drm/via/via_irq.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/via/via_irq.c b/drivers/gpu/drm/via/via_irq.c
> index 24e71578af4d..c96830ccc0ec 100644
> --- a/drivers/gpu/drm/via/via_irq.c
> +++ b/drivers/gpu/drm/via/via_irq.c
> @@ -343,7 +343,7 @@ void via_driver_irq_uninstall(struct drm_device *dev)
>  int via_wait_irq(struct drm_device *dev, void *data, struct drm_file *file_priv)
>  {
>  	drm_via_irqwait_t *irqwait = data;
> -	struct timeval now;
> +	struct timespec64 now;
>  	int ret = 0;
>  	drm_via_private_t *dev_priv = (drm_via_private_t *) dev->dev_private;
>  	drm_via_irq_t *cur_irq = dev_priv->via_irqs;
> @@ -377,9 +377,9 @@ int via_wait_irq(struct drm_device *dev, void *data, struct drm_file *file_priv)
>  
>  	ret = via_driver_irq_wait(dev, irqwait->request.irq, force_sequence,
>  				  &irqwait->request.sequence);
> -	do_gettimeofday(&now);
> +	ktime_get_ts64(&now);
>  	irqwait->reply.tval_sec = now.tv_sec;
> -	irqwait->reply.tval_usec = now.tv_usec;
> +	irqwait->reply.tval_usec = now.tv_nsec / NSEC_PER_USEC;
>  
>  	return ret;
>  }
> -- 
> 2.9.0
> 
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

From 1585217844977779198@xxx Mon Nov 27 11:19:04 +0000 2017
X-GM-THRID: 1585217844977779198
X-Gmail-Labels: Inbox,Category Forums,HistoricalUnread