[PATCH v4 02/13] modesetting: scanout_pixmap_gpu one call only

[Alex Goins]

Subsequent commits will involve multiple calls to
rrCrtcSetScanoutPixmap(ppix) without a call to rrCrtcSetScanoutPixmap(NULL)
in between. This causes drmmode_set_scanout_pixmap_gpu() to track damage
multiple times but ultimately only stop tracking on one, which could lead
to dangling pointers.

I don't intend to support PRIME synchronization with reverse PRIME at this
time, so just fail in this case.

It's probably a bad idea to rely on randr setting
randr_crtc->scanout_pixmap in order to know which pixmap to stop damage
tracking on, but I'll keep it that way for now to stay consistent.

v1: N/A
v2: N/A
v3: N/A
v4: Initial commit

Signed-off-by: Alex Goins <agoins at nvidia.com>
---
 hw/xfree86/drivers/modesetting/drmmode_display.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/hw/xfree86/drivers/modesetting/drmmode_display.c b/hw/xfree86/drivers/modesetting/drmmode_display.c
index 0d34ca1..ed67dbe 100644
--- a/hw/xfree86/drivers/modesetting/drmmode_display.c
+++ b/hw/xfree86/drivers/modesetting/drmmode_display.c
@@ -586,6 +586,11 @@ drmmode_set_scanout_pixmap_gpu(xf86CrtcPtr crtc, PixmapPtr ppix)
         drmmode_crtc->prime_pixmap_x = 0;
         return TRUE;
     }
+
+    /* Currently supports only one scanout pixmap */
+    if (crtc->randr_crtc->scanout_pixmap)
+        return FALSE;
+
     /* iterate over all the attached crtcs to work out the bounding box */
     for (c = 0; c < xf86_config->num_crtc; c++) {
         xf86CrtcPtr iter = xf86_config->crtc[c];
-- 
1.9.1