On Mon, 26 May 2014 14:46:23 +0300 ville.syrjala@linux.intel.com wrote:
From: Ville Syrjälä ville.syrjala@linux.intel.com
Another vblank series with the following features:
- Plug a race between drm_vblank_off() and marking the crtc inactive
- Don't send zeroed vblank evens to userspace at drm_vblank_off()
- Have the user visible vblank counter account the entire time when the crtc was active, regardless of how long vblank interrupts were enabled
- Avoid random jumps in the user visible vblank counter if the hardware counter gets reset
- Allow disabling vblank interrupts immediately at drm_vblank_put()
- Some polish via coccinelle
While setting drm_vblank_offdelay to 0 is now possible, I'm not sure if we should set it 0 automatically in the i915 driver. If there are multiple GPUs in the system that setting will affect them all, which might have bad consequences if the other GPU doesn't have a hardware frame counter, or if it's just buggy. So perhaps we should move that option to be per-driver?
Ville Syrjälä (9): drm: Always reject drm_vblank_get() after drm_vblank_off() drm/i915: Warn if drm_vblank_get() still works after drm_vblank_off() drm: Don't clear vblank timestamps when vblank interrupt is disabled drm: Move drm_update_vblank_count() drm: Have the vblank counter account for the time between vblank irq disable and drm_vblank_off() drm: Avoid random vblank counter jumps if the hardware counter has been reset drm: Disable vblank interrupt immediately when drm_vblank_offdelay==0 drm: Reduce the amount of dev->vblank[crtc] in the code drm/i915: Leave interrupts enabled while disabling crtcs during suspend
Here's one that may be fixed by this series, needs testing though: https://bugs.freedesktop.org/show_bug.cgi?id=79054