Re: [PATCH 5/7] drm: Don't compute obj counts expensively in get_resources

From: Chris Wilson <chris@chris-wilson.co.uk>
To: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Intel Graphics Development <intel-gfx@lists.freedesktop.org>,
	DRI Development <dri-devel@lists.freedesktop.org>
Subject: Re: [PATCH 5/7] drm: Don't compute obj counts expensively in get_resources
Date: Fri, 9 Dec 2016 21:47:56 +0000	[thread overview]
Message-ID: <20161209214756.GD26328@nuc-i3427.alporthouse.com> (raw)
In-Reply-To: <20161209141944.22121-5-daniel.vetter@ffwll.ch>

On Fri, Dec 09, 2016 at 03:19:42PM +0100, Daniel Vetter wrote:
> Looping when we keep track of this is silly. Only thing we have to
> be careful is with sampling the connector count. To avoid inconsisten
> results due to gcc re-computing this, use READ_ONCE.

Later on in the function we take the mutex that should prevent the
values from changing.

If each block was like

mutex_lock(&mode_config->lockc);

count = dev->mode_config.num_crtc;
if (card_res->count_crtcs >= count) {
	u32 __user id = u64_to_user_ptr(card_res->crtc_id_ptr);
	unsigned int copied = 0;

	drm_for_each_crtc(crtc, dev) {
		if (copied >= count)
			break;

		if (put_user(crtc->base.id, id + copied)) {
			ret = -EFAULT;
			goto out;
		}

		copied++;
	}

	count = copied;
}
card_res->count_crtcs = count;

...

mutex_unlock(&mode_config->lockc);

it would look a bit neater.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx