Skip to contentSkip to conversation

drm/i915: Avoid div by zero when pixel clock is large

PATCH Applied [email protected] 3 participants 4 messages Feb 14, 2014 – Mar 6, 2014

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/i915: Avoid div by zero when pixel clock is large

From: Ville Syrjälä <[email protected]>

Make sure the line_time_us isn't zero in the gmch watermarks code as that would cause a div by zero. This can be triggered by specifying a very fast pixel clock for the mode.

At some point we should probably just switch over to using the same math we use on PCH platforms which avoids such intermediate rounded results.

Also we should verify the user provided mode much more rigorously. At the moment we accept pretty much anything.

---
 drivers/gpu/drm/i915/intel_pm.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/i915/intel_pm.c +4 -4
index e4a0c9c..b84836f 100644
@@ -1131,7 +1131,7 @@ static bool g4x_compute_wm0(struct drm_device *dev,
 		*plane_wm = display->max_wm;
 
 	/* Use the large buffer method to calculate cursor watermark */
-	line_time_us = ((htotal * 1000) / clock);
+	line_time_us = max(htotal * 1000 / clock, 1);
 	line_count = (cursor_latency_ns / line_time_us + 1000) / 1000;
 	entries = line_count * 64 * pixel_size;
 	tlb_miss = cursor->fifo_size*cursor->cacheline_size - hdisplay * 8;
@@ -1207,7 +1207,7 @@ static bool g4x_compute_srwm(struct drm_device *dev,
 	hdisplay = to_intel_crtc(crtc)->config.pipe_src_w;
 	pixel_size = crtc->fb->bits_per_pixel / 8;
 
-	line_time_us = (htotal * 1000) / clock;
+	line_time_us = max(htotal * 1000 / clock, 1);
 	line_count = (latency_ns / line_time_us + 1000) / 1000;
 	line_size = hdisplay * pixel_size;
 
@@ -1440,7 +1440,7 @@ static void i965_update_wm(struct drm_crtc *unused_crtc)
 		unsigned long line_time_us;
 		int entries;
 
-		line_time_us = ((htotal * 1000) / clock);
+		line_time_us = max(htotal * 1000 / clock, 1);
 
 		/* Use ns/us then divide to preserve precision */
 		entries = (((sr_latency_ns / line_time_us) + 1000) / 1000) *
@@ -1566,7 +1566,7 @@ static void i9xx_update_wm(struct drm_crtc *unused_crtc)
 		unsigned long line_time_us;
 		int entries;
 
-		line_time_us = (htotal * 1000) / clock;
+		line_time_us = max(htotal * 1000 / clock, 1);
 
 		/* Use ns/us then divide to preserve precision */
 		entries = (((sr_latency_ns / line_time_us) + 1000) / 1000) *
Signature · 6 lines
1.8.3.2

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v1 1/1
#1 lore raw
Quoted 5 lines · On Fri, Feb 14, 2014 at 02:18:57PM +0200, [email protected] wrote:

From: Ville Syrjälä <[email protected]>

Make sure the line_time_us isn't zero in the gmch watermarks code as that would cause a div by zero. This can be triggered by specifying a very fast pixel clock for the mode.

Very fast but valid? If it was just very fast, we should have rejected the mode line. On the other hand, if it was valid, maybe we should adjust the bias (* 1000 / 1000) so that the computation remains reasonably accurate because we will need a larger FIFO to accommodate the higher clock rate.

Anyway, even with the larger bias, such a safe guard is sensisble. I can't argue against this patch, but I just wonder if you opened a can of worms.

-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
#2 lore raw
Quoted 8 lines · Chris Wilson
Quoted 5 lines · On Fri, Feb 14, 2014 at 02:18:57PM +0200, [email protected] wrote:

From: Ville Syrjälä <[email protected]>

Make sure the line_time_us isn't zero in the gmch watermarks code as that would cause a div by zero. This can be triggered by specifying a very fast pixel clock for the mode.

Very fast but valid?

74.25 GHz. I'm not sure which way you would classify that :)

Quoted 2 lines

If it was just very fast, we should have rejected
the mode line.

Except we do more or less zero checks on the mode supplied to the setcrtc ioctl. Something we should really get fixed at some point.

Signature · 2 lines
Ville Syrjälä
Intel OTC
#3 lore raw
Quoted 11 lines · Ville Syrjälä
Quoted 8 lines · Chris Wilson
Quoted 5 lines · On Fri, Feb 14, 2014 at 02:18:57PM +0200, [email protected] wrote:

From: Ville Syrjälä <[email protected]>

Make sure the line_time_us isn't zero in the gmch watermarks code as that would cause a div by zero. This can be triggered by specifying a very fast pixel clock for the mode.

Very fast but valid?

74.25 GHz. I'm not sure which way you would classify that :)

I've added this to the commit message ...

Quoted 5 lines
Quoted 2 lines

If it was just very fast, we should have rejected
the mode line.

Except we do more or less zero checks on the mode supplied to the setcrtc ioctl. Something we should really get fixed at some point.

Queued for -next, thanks for the patch.
-Daniel

Signature · 3 lines
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch