drm/radeon: Support DRM_MODE_PAGE_FLIP_ASYNC

When this flag is set, we program the hardware to execute the flip
during horizontal blank (i.e. for the next scanline) instead of during
vertical blank (i.e. for the next frame).

Currently this is only supported on ASICs which have a page flip
completion interrupt (>= R600), and only if the use_pflipirq parameter
has value 2 (the default).

Reviewed-by: Christian König <[e-mail]>
Signed-off-by: Michel Dänzer <[e-mail]>
Signed-off-by: Alex Deucher <[e-mail]>

drivers/gpu/drm/radeon/atombios_crtc.c

···
 		break;
 	}
 
+	/* Make sure surface address is updated at vertical blank rather than
+	 * horizontal blank
+	 */
+	WREG32(EVERGREEN_GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset, 0);
+
 	WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + radeon_crtc->crtc_offset,
 	       upper_32_bits(fb_location));
 	WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + radeon_crtc->crtc_offset,
···
 	WREG32(EVERGREEN_VIEWPORT_SIZE + radeon_crtc->crtc_offset,
 	       (viewport_w << 16) | viewport_h);
 
-	/* pageflip setup */
-	/* make sure flip is at vb rather than hb */
-	tmp = RREG32(EVERGREEN_GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset);
-	tmp &= ~EVERGREEN_GRPH_SURFACE_UPDATE_H_RETRACE_EN;
-	WREG32(EVERGREEN_GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset, tmp);
-
 	/* set pageflip to happen only at start of vblank interval (front porch) */
 	WREG32(EVERGREEN_MASTER_UPDATE_MODE + radeon_crtc->crtc_offset, 3);
 
···
 	uint64_t fb_location;
 	uint32_t fb_format, fb_pitch_pixels, tiling_flags;
 	u32 fb_swap = R600_D1GRPH_SWAP_ENDIAN_NONE;
-	u32 tmp, viewport_w, viewport_h;
+	u32 viewport_w, viewport_h;
 	int r;
 	bool bypass_lut = false;
 
···
 	else
 		WREG32(AVIVO_D2VGA_CONTROL, 0);
 
+	/* Make sure surface address is update at vertical blank rather than
+	 * horizontal blank
+	 */
+	WREG32(AVIVO_D1GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset, 0);
+
 	if (rdev->family >= CHIP_RV770) {
 		if (radeon_crtc->crtc_id) {
 			WREG32(R700_D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, upper_32_bits(fb_location));
···
 	viewport_h = (crtc->mode.vdisplay + 1) & ~1;
 	WREG32(AVIVO_D1MODE_VIEWPORT_SIZE + radeon_crtc->crtc_offset,
 	       (viewport_w << 16) | viewport_h);
-
-	/* pageflip setup */
-	/* make sure flip is at vb rather than hb */
-	tmp = RREG32(AVIVO_D1GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset);
-	tmp &= ~AVIVO_D1GRPH_SURFACE_UPDATE_H_RETRACE_EN;
-	WREG32(AVIVO_D1GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset, tmp);
 
 	/* set pageflip to happen only at start of vblank interval (front porch) */
 	WREG32(AVIVO_D1MODE_MASTER_UPDATE_MODE + radeon_crtc->crtc_offset, 3);

drivers/gpu/drm/radeon/evergreen.c

···
  * Triggers the actual pageflip by updating the primary
  * surface base address (evergreen+).
  */
-void evergreen_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base)
+void evergreen_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base,
+			 bool async)
 {
 	struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id];
 
 	/* update the scanout addresses */
+	WREG32(EVERGREEN_GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset,
+	       async ? EVERGREEN_GRPH_SURFACE_UPDATE_H_RETRACE_EN : 0);
 	WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + radeon_crtc->crtc_offset,
 	       upper_32_bits(crtc_base));
 	WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset,

drivers/gpu/drm/radeon/r100.c

···
  * bit to go high, when it does, we release the lock, and allow the
  * double buffered update to take place.
  */
-void r100_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base)
+void r100_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base, bool async)
 {
 	struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id];
 	u32 tmp = ((u32)crtc_base) | RADEON_CRTC_OFFSET__OFFSET_LOCK;

drivers/gpu/drm/radeon/radeon.h

···
 	struct drm_pending_vblank_event *event;
 	struct radeon_bo		*old_rbo;
 	struct fence			*fence;
+	bool				async;
 };
 
 struct r500_irq_stat_regs {
···
 	} dpm;
 	/* pageflipping */
 	struct {
-		void (*page_flip)(struct radeon_device *rdev, int crtc, u64 crtc_base);
+		void (*page_flip)(struct radeon_device *rdev, int crtc, u64 crtc_base, bool async);
 		bool (*page_flip_pending)(struct radeon_device *rdev, int crtc);
 	} pflip;
 };
···
 #define radeon_pm_finish(rdev) (rdev)->asic->pm.finish((rdev))
 #define radeon_pm_init_profile(rdev) (rdev)->asic->pm.init_profile((rdev))
 #define radeon_pm_get_dynpm_state(rdev) (rdev)->asic->pm.get_dynpm_state((rdev))
-#define radeon_page_flip(rdev, crtc, base) (rdev)->asic->pflip.page_flip((rdev), (crtc), (base))
+#define radeon_page_flip(rdev, crtc, base, async) (rdev)->asic->pflip.page_flip((rdev), (crtc), (base), (async))
 #define radeon_page_flip_pending(rdev, crtc) (rdev)->asic->pflip.page_flip_pending((rdev), (crtc))
 #define radeon_wait_for_vblank(rdev, crtc) (rdev)->asic->display.wait_for_vblank((rdev), (crtc))
 #define radeon_mc_wait_for_idle(rdev) (rdev)->asic->mc_wait_for_idle((rdev))

drivers/gpu/drm/radeon/radeon_asic.h

···
 extern void r100_pm_init_profile(struct radeon_device *rdev);
 extern void r100_pm_get_dynpm_state(struct radeon_device *rdev);
 extern void r100_page_flip(struct radeon_device *rdev, int crtc,
-			   u64 crtc_base);
+			   u64 crtc_base, bool async);
 extern bool r100_page_flip_pending(struct radeon_device *rdev, int crtc);
 extern void r100_wait_for_vblank(struct radeon_device *rdev, int crtc);
 extern int r100_mc_wait_for_idle(struct radeon_device *rdev);
···
 extern void rs600_pm_prepare(struct radeon_device *rdev);
 extern void rs600_pm_finish(struct radeon_device *rdev);
 extern void rs600_page_flip(struct radeon_device *rdev, int crtc,
-			    u64 crtc_base);
+			    u64 crtc_base, bool async);
 extern bool rs600_page_flip_pending(struct radeon_device *rdev, int crtc);
 void rs600_set_safe_registers(struct radeon_device *rdev);
 extern void avivo_wait_for_vblank(struct radeon_device *rdev, int crtc);
···
 int rv770_suspend(struct radeon_device *rdev);
 int rv770_resume(struct radeon_device *rdev);
 void rv770_pm_misc(struct radeon_device *rdev);
-void rv770_page_flip(struct radeon_device *rdev, int crtc, u64 crtc_base);
+void rv770_page_flip(struct radeon_device *rdev, int crtc, u64 crtc_base,
+		     bool async);
 bool rv770_page_flip_pending(struct radeon_device *rdev, int crtc);
 void r700_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc);
 void r700_cp_stop(struct radeon_device *rdev);
···
 int sumo_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk);
 int evergreen_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk);
 extern void evergreen_page_flip(struct radeon_device *rdev, int crtc,
-				u64 crtc_base);
+				u64 crtc_base, bool async);
 extern bool evergreen_page_flip_pending(struct radeon_device *rdev, int crtc);
 extern void dce4_wait_for_vblank(struct radeon_device *rdev, int crtc);
 void evergreen_disable_interrupt_state(struct radeon_device *rdev);

drivers/gpu/drm/radeon/radeon_display.c

···
 				 vblank->linedur_ns / 1000, stat, vpos, hpos);
 
 	/* do the flip (mmio) */
-	radeon_page_flip(rdev, radeon_crtc->crtc_id, work->base);
+	radeon_page_flip(rdev, radeon_crtc->crtc_id, work->base, work->async);
 
 	radeon_crtc->flip_status = RADEON_FLIP_SUBMITTED;
 	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
···
 	work->rdev = rdev;
 	work->crtc_id = radeon_crtc->crtc_id;
 	work->event = event;
+	work->async = (page_flip_flags & DRM_MODE_PAGE_FLIP_ASYNC) != 0;
 
 	/* schedule unpin of the old buffer */
 	old_radeon_fb = to_radeon_framebuffer(crtc->primary->fb);
···
 	rdev->mode_info.mode_config_initialized = true;
 
 	rdev->ddev->mode_config.funcs = &radeon_mode_funcs;
+
+	if (radeon_use_pflipirq == 2 && rdev->family >= CHIP_R600)
+		rdev->ddev->mode_config.async_page_flip = true;
 
 	if (ASIC_IS_DCE5(rdev)) {
 		rdev->ddev->mode_config.max_width = 16384;

drivers/gpu/drm/radeon/rs600.c

···
 	}
 }
 
-void rs600_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base)
+void rs600_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base, bool async)
 {
 	struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id];
 	u32 tmp = RREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset);
···
 	WREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset, tmp);
 
 	/* update the scanout addresses */
+	WREG32(AVIVO_D1GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset,
+	       async ? AVIVO_D1GRPH_SURFACE_UPDATE_H_RETRACE_EN : 0);
 	WREG32(AVIVO_D1GRPH_SECONDARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset,
 	       (u32)crtc_base);
 	WREG32(AVIVO_D1GRPH_PRIMARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset,

drivers/gpu/drm/radeon/rv770.c

···
 	return reference_clock;
 }
 
-void rv770_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base)
+void rv770_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base, bool async)
 {
 	struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id];
 	u32 tmp = RREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset);
···
 	WREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset, tmp);
 
 	/* update the scanout addresses */
+	WREG32(AVIVO_D1GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset,
+	       async ? AVIVO_D1GRPH_SURFACE_UPDATE_H_RETRACE_EN : 0);
 	if (radeon_crtc->crtc_id) {
 		WREG32(D2GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, upper_32_bits(crtc_base));
 		WREG32(D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, upper_32_bits(crtc_base));