Skip to contentSkip to conversation

drm/meson: add AFBC support

PATCH In review Neil Armstrong 2 participants 13 messages Oct 21, 2019 – Dec 10, 2019

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

v3 cover drm/meson: add AFBC support

This adds support for the ARM Framebuffer Compression decoders found in the Amlogic GXM and G12A SoCs.

This patchset is a merge of v2 "drm/meson: add AFBC support" at [3] and v2 "drm/meson: implement RDMA for AFBC reset on vsync" at [4].

The VPU embeds a "Register DMA" that can write a sequence of registers on the VPU AHB bus, either manually or triggered by an internal IRQ event like VSYNC or a line input counter.

The Amlogic GXM and G12A AFBC decoder are totally different, the GXM only handling only the AFBC v1.0 modes and the G12A decoder handling the AFBC v1.2 modes.

The G12A AFBC decoder is an external IP integrated in the video pipeline, and the GXM AFBC decoder seems to the an Amlogic custom decoder more tighly integrated in the video pipeline.

The GXM AFBC decoder can handle only one AFBC plane for 2 available OSD planes available in HW, and the G12A AFBC decoder can handle up to 4 AFBC planes for up to 3 OSD planes available in HW.

The Amlogic GXM supports 16x16 SPARSE and 16x16 SPLIT AFBC buffers up to 4k.

On the other side, for G12A SPLIT is mandatory in 16x16 block mode, but for 4k modes 32x8+SPLIT AFBC buffers is manadatory for performances reasons.

The Amlogic GXM and G12A AFBC decoders are integrated very differently.

The Amlogic GXM has a direct output path to the OSD1 VIU pixel input, because the GXM AFBC decoder seem to be a custom IP developed by Amlogic.

On the other side, the Amlogic G12A AFBC decoder seems to be an external IP that emit pixels on an AXI master hooked to a "Mali Unpack" block feeding the OSD1 VIU pixel input.
This uses a weird "0x1000000" internal HW physical address on both sides to transfer the pixels.

For Amlogic GXM, the supported pixel formats are the same as the normal linear OSD1 mode.

On the other side, Amlogic added support for all AFBC v1.2 formats for the G12A AFBC integration.

The initial RDMA implementation handles a single channel (over 8), triggered by the VSYNC irq and does not handle the RDMA irq.

The RDMA will be usefull to reset and program the AFBC decoder unit on each vsync without involving the interrupt handler that can be masked for a long period of time, producing display glitches.

For this we use the meson_rdma_writel_sync() which adds the register write tuple (VPU register offset and register value) to the RDMA buffer and write the value to the HW.

When enabled, the RDMA is enabled to rewritte the same sequence at the next VSYNC event, until a new buffer is committed to the OSD plane.

For testing, the only available AFBC buffer generation is the Android Yukawa Dvalin Android Mali blobs found at [1].

Both SoCs has been tested using buffers generated under AOSP, but only G12A was tested using a runtime stream of AFBC buffers, GXM was only tested using static buffers loaded from files.

Changes since v2 of [3] and [4]:
- Merge of RDMA patchset
- Fix suspend/resume when AFBC is enabled
- Re-enabled vsync_force otherwise RDMA is also stopped

Changes since v1 at [2]:
- Discards XRGB modes with YTR, only allow XBGR with YTR
- Add non-YTR modifiers for G12A
- Disable completely XRGB modes for GXM, until we find how to disable YTR
- Add proper argb remap for XRGB modes
- Enable OSD_STAT sync signal

[1] https://android.googlesource.com/device/amlogic/yukawa/+/refs/heads/master/gpu/
[2] https://patchwork.freedesktop.org/series/67832/#rev1
[3] https://patchwork.freedesktop.org/series/67832/#rev2
[4] https://patchwork.freedesktop.org/series/68021/#rev2

Neil Armstrong (9):
drm/meson: add AFBC decoder registers for GXM and G12A drm/meson: add RDMA register bits defines
drm/meson: store the framebuffer width for plane commit drm/meson: add RDMA module driver
drm/meson: Add AFBCD module driver
drm/meson: plane: add support for AFBC mode for OSD1 plane drm/meson: viu: add AFBC modules routing functions drm/meson: hold 32 lines after vsync to give time for AFBC start drm/meson: crtc: add OSD1 plane AFBC commit

 drivers/gpu/drm/meson/Makefile          |   1 +
 drivers/gpu/drm/meson/meson_crtc.c      |  79 ++++-
 drivers/gpu/drm/meson/meson_drv.c       |  50 ++-
 drivers/gpu/drm/meson/meson_drv.h       |  23 ++
 drivers/gpu/drm/meson/meson_osd_afbcd.c | 389 ++++++++++++++++++++++++
 drivers/gpu/drm/meson/meson_osd_afbcd.h |  28 ++
 drivers/gpu/drm/meson/meson_plane.c     | 229 ++++++++++++--
 drivers/gpu/drm/meson/meson_rdma.c      | 135 ++++++++
 drivers/gpu/drm/meson/meson_rdma.h      |  21 ++
 drivers/gpu/drm/meson/meson_registers.h | 110 +++++++
 drivers/gpu/drm/meson/meson_viu.c       |  83 ++++-
 drivers/gpu/drm/meson/meson_viu.h       |  19 ++
 12 files changed, 1123 insertions(+), 44 deletions(-)
 create mode 100644 drivers/gpu/drm/meson/meson_osd_afbcd.c
 create mode 100644 drivers/gpu/drm/meson/meson_osd_afbcd.h
 create mode 100644 drivers/gpu/drm/meson/meson_rdma.c
 create mode 100644 drivers/gpu/drm/meson/meson_rdma.h
Signature · 6 lines
2.22.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#1 lore raw

PATCH v3 1/9 drm/meson: add AFBC decoder registers for GXM and G12A

Add the registers used to program the ARM Framebuffer Compression decoders used in the Amlogic GXM and G12A SoCs families.

This also adds the routing and pipeline configuration bits and registers needed to enable AFBC support.

---
 drivers/gpu/drm/meson/meson_registers.h | 62 +++++++++++++++++++++++++
 drivers/gpu/drm/meson/meson_viu.h       | 15 ++++++
 2 files changed, 77 insertions(+)
drivers/gpu/drm/meson/meson_registers.h +62 -0
index 05fce48ceee0..547bee04b46e 100644
@@ -138,19 +138,25 @@
 #define VIU_ADDR_START 0x1a00
 #define VIU_ADDR_END 0x1aff
 #define VIU_SW_RESET 0x1a01
+#define		VIU_SW_RESET_OSD1_AFBCD		BIT(31)
+#define		VIU_SW_RESET_G12A_OSD1_AFBCD	BIT(21)
+#define		VIU_SW_RESET_G12A_AFBC_ARB	BIT(19)
 #define		VIU_SW_RESET_OSD1               BIT(0)
 #define VIU_MISC_CTRL0 0x1a06
 #define		VIU_CTRL0_VD1_AFBC_MASK         0x170000
 #define VIU_MISC_CTRL1 0x1a07
+#define		MALI_AFBC_MISC			GENMASK(15, 8)
 #define D2D3_INTF_LENGTH 0x1a08
 #define D2D3_INTF_CTRL0 0x1a09
 #define VIU_OSD1_CTRL_STAT 0x1a10
 #define		VIU_OSD1_OSD_BLK_ENABLE         BIT(0)
+#define		VIU_OSD1_OSD_MEM_MODE_LINEAR	BIT(2)
 #define		VIU_OSD1_POSTBLD_SRC_VD1        (1 << 8)
 #define		VIU_OSD1_POSTBLD_SRC_VD2        (2 << 8)
 #define		VIU_OSD1_POSTBLD_SRC_OSD1       (3 << 8)
 #define		VIU_OSD1_POSTBLD_SRC_OSD2       (4 << 8)
 #define		VIU_OSD1_OSD_ENABLE             BIT(21)
+#define		VIU_OSD1_CFG_SYN_EN             BIT(31)
 #define VIU_OSD1_CTRL_STAT2 0x1a2d
 #define VIU_OSD1_COLOR_ADDR 0x1a11
 #define VIU_OSD1_COLOR 0x1a12
@@ -181,6 +187,16 @@
 #define VIU_OSD1_FIFO_CTRL_STAT 0x1a2b
 #define VIU_OSD1_TEST_RDDATA 0x1a2c
 #define VIU_OSD1_PROT_CTRL 0x1a2e
+#define VIU_OSD1_MALI_UNPACK_CTRL 0x1a2f
+#define		VIU_OSD1_MALI_UNPACK_EN		BIT(31)
+#define		VIU_OSD1_MALI_AFBCD_R_REORDER	GENMASK(15, 12)
+#define		VIU_OSD1_MALI_AFBCD_G_REORDER	GENMASK(11, 8)
+#define		VIU_OSD1_MALI_AFBCD_B_REORDER	GENMASK(7, 4)
+#define		VIU_OSD1_MALI_AFBCD_A_REORDER	GENMASK(3, 0)
+#define		VIU_OSD1_MALI_REORDER_R		1
+#define		VIU_OSD1_MALI_REORDER_G		2
+#define		VIU_OSD1_MALI_REORDER_B		3
+#define		VIU_OSD1_MALI_REORDER_A		4
 #define VIU_OSD2_CTRL_STAT 0x1a30
 #define VIU_OSD2_CTRL_STAT2 0x1a4d
 #define VIU_OSD2_COLOR_ADDR 0x1a31
@@ -1595,15 +1611,33 @@
 
 /* osd afbcd on gxtvbb */
 #define OSD1_AFBCD_ENABLE 0x31a0
+#define		OSD1_AFBCD_ID_FIFO_THRD			GENMASK(15, 9)
+#define		OSD1_AFBCD_DEC_ENABLE			BIT(8)
+#define		OSD1_AFBCD_FRM_START			BIT(0)
 #define OSD1_AFBCD_MODE 0x31a1
+#define		OSD1_AFBCD_SOFT_RESET			BIT(31)
+#define		OSD1_AFBCD_AXI_REORDER_MODE		BIT(28)
+#define		OSD1_AFBCD_MIF_URGENT			GENMASK(25, 24)
+#define		OSD1_AFBCD_HOLD_LINE_NUM		GENMASK(22, 16)
+#define		OSD1_AFBCD_RGBA_EXCHAN_CTRL		GENMASK(15, 8)
+#define		OSD1_AFBCD_HREG_BLOCK_SPLIT		BIT(6)
+#define		OSD1_AFBCD_HREG_HALF_BLOCK		BIT(5)
+#define		OSD1_AFBCD_HREG_PIXEL_PACKING_FMT	GENMASK(4, 0)
 #define OSD1_AFBCD_SIZE_IN 0x31a2
+#define		OSD1_AFBCD_HREG_VSIZE_IN		GENMASK(31, 16)
+#define		OSD1_AFBCD_HREG_HSIZE_IN		GENMASK(15, 0)
 #define OSD1_AFBCD_HDR_PTR 0x31a3
 #define OSD1_AFBCD_FRAME_PTR 0x31a4
 #define OSD1_AFBCD_CHROMA_PTR 0x31a5
 #define OSD1_AFBCD_CONV_CTRL 0x31a6
+#define		OSD1_AFBCD_CONV_LBUF_LEN		GENMASK(15, 0)
 #define OSD1_AFBCD_STATUS 0x31a8
 #define OSD1_AFBCD_PIXEL_HSCOPE 0x31a9
+#define		OSD1_AFBCD_DEC_PIXEL_BGN_H		GENMASK(31, 16)
+#define		OSD1_AFBCD_DEC_PIXEL_END_H		GENMASK(15, 0)
 #define OSD1_AFBCD_PIXEL_VSCOPE 0x31aa
+#define		OSD1_AFBCD_DEC_PIXEL_BGN_V		GENMASK(31, 16)
+#define		OSD1_AFBCD_DEC_PIXEL_END_V		GENMASK(15, 0)
 
 /* add for gxm and 962e dv core2 */
 #define DOLBY_CORE2A_SWAP_CTRL1	0x3434
@@ -1615,12 +1649,34 @@
 #define VPU_MAFBC_IRQ_CLEAR 0x3a02
 #define VPU_MAFBC_IRQ_MASK 0x3a03
 #define VPU_MAFBC_IRQ_STATUS 0x3a04
+#define		VPU_MAFBC_IRQ_SECURE_ID_ERROR		BIT(5)
+#define		VPU_MAFBC_IRQ_AXI_ERROR			BIT(4)
+#define		VPU_MAFBC_IRQ_DETILING_ERROR		BIT(3)
+#define		VPU_MAFBC_IRQ_DECODE_ERROR		BIT(2)
+#define		VPU_MAFBC_IRQ_CONFIGURATION_SWAPPED	BIT(1)
+#define		VPU_MAFBC_IRQ_SURFACES_COMPLETED	BIT(0)
 #define VPU_MAFBC_COMMAND 0x3a05
+#define		VPU_MAFBC_PENDING_SWAP	BIT(1)
+#define		VPU_MAFBC_DIRECT_SWAP	BIT(0)
 #define VPU_MAFBC_STATUS 0x3a06
+#define		VPU_MAFBC_ERROR		BIT(2)
+#define		VPU_MAFBC_SWAPPING	BIT(1)
+#define		VPU_MAFBC_ACTIVE	BIT(0)
 #define VPU_MAFBC_SURFACE_CFG 0x3a07
+#define		VPU_MAFBC_CONTINUOUS_DECODING_ENABLE	BIT(16)
+#define		VPU_MAFBC_S3_ENABLE			BIT(3)
+#define		VPU_MAFBC_S2_ENABLE			BIT(2)
+#define		VPU_MAFBC_S1_ENABLE			BIT(1)
+#define		VPU_MAFBC_S0_ENABLE			BIT(0)
 #define VPU_MAFBC_HEADER_BUF_ADDR_LOW_S0 0x3a10
 #define VPU_MAFBC_HEADER_BUF_ADDR_HIGH_S0 0x3a11
 #define VPU_MAFBC_FORMAT_SPECIFIER_S0 0x3a12
+#define		VPU_MAFBC_PAYLOAD_LIMIT_EN	BIT(19)
+#define		VPU_MAFBC_TILED_HEADER_EN	BIT(18)
+#define		VPU_MAFBC_SUPER_BLOCK_ASPECT	GENMASK(17, 16)
+#define		VPU_MAFBC_BLOCK_SPLIT		BIT(9)
+#define		VPU_MAFBC_YUV_TRANSFORM		BIT(8)
+#define		VPU_MAFBC_PIXEL_FORMAT		GENMASK(3, 0)
 #define VPU_MAFBC_BUFFER_WIDTH_S0 0x3a13
 #define VPU_MAFBC_BUFFER_HEIGHT_S0 0x3a14
 #define VPU_MAFBC_BOUNDING_BOX_X_START_S0 0x3a15
@@ -1631,6 +1687,8 @@
 #define VPU_MAFBC_OUTPUT_BUF_ADDR_HIGH_S0 0x3a1a
 #define VPU_MAFBC_OUTPUT_BUF_STRIDE_S0 0x3a1b
 #define VPU_MAFBC_PREFETCH_CFG_S0 0x3a1c
+#define		VPU_MAFBC_PREFETCH_READ_DIRECTION_Y	BIT(1)
+#define		VPU_MAFBC_PREFETCH_READ_DIRECTION_X	BIT(0)
 
 #define VPU_MAFBC_HEADER_BUF_ADDR_LOW_S1 0x3a30
 #define VPU_MAFBC_HEADER_BUF_ADDR_HIGH_S1 0x3a31
@@ -1677,7 +1735,11 @@
 #define DOLBY_PATH_CTRL 0x1a0c
 #define		DOLBY_BYPASS_EN(val)            (val & 0xf)
 #define OSD_PATH_MISC_CTRL 0x1a0e
+#define		OSD_PATH_OSD_AXI_SEL_OSD1_AFBCD	BIT(4)
+#define		OSD_PATH_OSD_AXI_SEL_OSD2_AFBCD	BIT(5)
+#define		OSD_PATH_OSD_AXI_SEL_OSD3_AFBCD	BIT(6)
 #define MALI_AFBCD_TOP_CTRL 0x1a0f
+#define		MALI_AFBCD_MANUAL_RESET		BIT(23)
 
 #define VIU_OSD_BLEND_CTRL 0x39b0
 #define		VIU_OSD_BLEND_REORDER(dest, src)      ((src) << (dest * 4))
drivers/gpu/drm/meson/meson_viu.h +15 -0
index a112e8d18850..e297772d967f 100644
@@ -10,6 +10,8 @@
 #define __MESON_VIU_H
 
 /* OSDx_BLKx_CFG */
+#define OSD_MALI_SRC_EN		BIT(30)
+
 #define OSD_CANVAS_SEL		16
 
 #define OSD_ENDIANNESS_LE	BIT(15)
@@ -33,19 +35,32 @@
 #define OSD_COLOR_MATRIX_16_RGB655	(0x00 << 2)
 #define OSD_COLOR_MATRIX_16_RGB565	(0x04 << 2)
 
+#define OSD_MALI_COLOR_MODE_R8		(0 << 8)
+#define OSD_MALI_COLOR_MODE_YUV422	(1 << 8)
+#define OSD_MALI_COLOR_MODE_RGB565	(2 << 8)
+#define OSD_MALI_COLOR_MODE_RGBA5551	(3 << 8)
+#define OSD_MALI_COLOR_MODE_RGBA4444	(4 << 8)
+#define OSD_MALI_COLOR_MODE_RGBA8888	(5 << 8)
+#define OSD_MALI_COLOR_MODE_RGB888	(7 << 8)
+#define OSD_MALI_COLOR_MODE_YUV422_10B	(8 << 8)
+#define OSD_MALI_COLOR_MODE_RGBA1010102	(9 << 8)
+
 #define OSD_INTERLACE_ENABLED	BIT(1)
 #define OSD_INTERLACE_ODD	BIT(0)
 #define OSD_INTERLACE_EVEN	(0)
 
 /* OSDx_CTRL_STAT */
 #define OSD_ENABLE		BIT(21)
+#define OSD_MEM_LINEAR_ADDR	BIT(2)
 #define OSD_BLK0_ENABLE		BIT(0)
 
 #define OSD_GLOBAL_ALPHA_SHIFT	12
 
 /* OSDx_CTRL_STAT2 */
+#define OSD_DPATH_MALI_AFBCD	BIT(15)
 #define OSD_REPLACE_EN		BIT(14)
 #define OSD_REPLACE_SHIFT	6
+#define OSD_PENDING_STAT_CLEAN	BIT(1)
 
 void meson_viu_osd1_reset(struct meson_drm *priv);
 void meson_viu_init(struct meson_drm *priv);
Signature · 6 lines
2.22.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#2 lore raw

PATCH v3 2/9 drm/meson: add RDMA register bits defines

In reply to Neil Armstrong

The Amlogic VPU embeds a "Register DMA" that can write a sequence of registers on the VPU AHB bus, either manually or triggered by an internal IRQ event like VSYNC or a line input counter.

This adds the register defines.

---
 drivers/gpu/drm/meson/meson_registers.h | 48 +++++++++++++++++++++++++
 1 file changed, 48 insertions(+)
drivers/gpu/drm/meson/meson_registers.h +48 -0
index 547bee04b46e..8ea00546cd4e 100644
@@ -1211,11 +1211,59 @@
 #define RDMA_AHB_START_ADDR_7 0x110e
 #define RDMA_AHB_END_ADDR_7 0x110f
 #define RDMA_ACCESS_AUTO 0x1110
+#define		RDMA_ACCESS_TRIGGER_CHAN3	GENMASK(31, 24)
+#define		RDMA_ACCESS_TRIGGER_CHAN2	GENMASK(23, 16)
+#define		RDMA_ACCESS_TRIGGER_CHAN1	GENMASK(15, 8)
+#define		RDMA_ACCESS_TRIGGER_STOP	0
+#define		RDMA_ACCESS_TRIGGER_VSYNC	1
+#define		RDMA_ACCESS_TRIGGER_LINE	32
+#define		RDMA_ACCESS_RW_FLAG_CHAN3	BIT(7)
+#define		RDMA_ACCESS_RW_FLAG_CHAN2	BIT(6)
+#define		RDMA_ACCESS_RW_FLAG_CHAN1	BIT(5)
+#define		RDMA_ACCESS_ADDR_INC_CHAN3	BIT(3)
+#define		RDMA_ACCESS_ADDR_INC_CHAN2	BIT(2)
+#define		RDMA_ACCESS_ADDR_INC_CHAN1	BIT(1)
 #define RDMA_ACCESS_AUTO2 0x1111
+#define		RDMA_ACCESS_RW_FLAG_CHAN7	BIT(7)
+#define		RDMA_ACCESS_RW_FLAG_CHAN6	BIT(6)
+#define		RDMA_ACCESS_RW_FLAG_CHAN5	BIT(5)
+#define		RDMA_ACCESS_RW_FLAG_CHAN4	BIT(4)
+#define		RDMA_ACCESS_ADDR_INC_CHAN7	BIT(3)
+#define		RDMA_ACCESS_ADDR_INC_CHAN6	BIT(2)
+#define		RDMA_ACCESS_ADDR_INC_CHAN5	BIT(1)
+#define		RDMA_ACCESS_ADDR_INC_CHAN4	BIT(0)
 #define RDMA_ACCESS_AUTO3 0x1112
+#define		RDMA_ACCESS_TRIGGER_CHAN7	GENMASK(31, 24)
+#define		RDMA_ACCESS_TRIGGER_CHAN6	GENMASK(23, 16)
+#define		RDMA_ACCESS_TRIGGER_CHAN5	GENMASK(15, 8)
+#define		RDMA_ACCESS_TRIGGER_CHAN4	GENMASK(7, 0)
 #define RDMA_ACCESS_MAN 0x1113
+#define		RDMA_ACCESS_MAN_RW_FLAG		BIT(2)
+#define		RDMA_ACCESS_MAN_ADDR_INC	BIT(1)
+#define		RDMA_ACCESS_MAN_START		BIT(0)
 #define RDMA_CTRL 0x1114
+#define		RDMA_IRQ_CLEAR_CHAN7	BIT(31)
+#define		RDMA_IRQ_CLEAR_CHAN6	BIT(30)
+#define		RDMA_IRQ_CLEAR_CHAN5	BIT(29)
+#define		RDMA_IRQ_CLEAR_CHAN4	BIT(28)
+#define		RDMA_IRQ_CLEAR_CHAN3	BIT(27)
+#define		RDMA_IRQ_CLEAR_CHAN2	BIT(26)
+#define		RDMA_IRQ_CLEAR_CHAN1	BIT(25)
+#define		RDMA_IRQ_CLEAR_CHAN_MAN	BIT(24)
+#define		RDMA_DEFAULT_CONFIG	(BIT(7) | BIT(6))
+#define		RDMA_CTRL_AHB_WR_BURST	GENMASK(5, 4)
+#define		RDMA_CTRL_AHB_RD_BURST	GENMASK(3, 2)
+#define		RDMA_CTRL_SW_RESET	BIT(1)
+#define		RDMA_CTRL_FREE_CLK_EN	BIT(0)
 #define RDMA_STATUS 0x1115
+#define		RDMA_IRQ_STAT_CHAN7	BIT(31)
+#define		RDMA_IRQ_STAT_CHAN6	BIT(30)
+#define		RDMA_IRQ_STAT_CHAN5	BIT(29)
+#define		RDMA_IRQ_STAT_CHAN4	BIT(28)
+#define		RDMA_IRQ_STAT_CHAN3	BIT(27)
+#define		RDMA_IRQ_STAT_CHAN2	BIT(26)
+#define		RDMA_IRQ_STAT_CHAN1	BIT(25)
+#define		RDMA_IRQ_STAT_CHAN_MAN	BIT(24)
 #define RDMA_STATUS2 0x1116
 #define RDMA_STATUS3 0x1117
 #define L_GAMMA_CNTL_PORT 0x1400
Signature · 6 lines
2.22.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#3 lore raw

PATCH v3 3/9 drm/meson: store the framebuffer width for plane commit

In reply to Neil Armstrong

Also store the framebuffer width in the private common struct to be used by the AFBC decoder module driver when committing the AFBC plane.

---
 drivers/gpu/drm/meson/meson_drv.h   | 1 +
 drivers/gpu/drm/meson/meson_plane.c | 1 +
 2 files changed, 2 insertions(+)
drivers/gpu/drm/meson/meson_drv.h +1 -0
index 820d07bdd42a..3287282450a9 100644
@@ -52,6 +52,7 @@ struct meson_drm {
 		uint32_t osd1_addr;
 		uint32_t osd1_stride;
 		uint32_t osd1_height;
+		uint32_t osd1_width;
 		uint32_t osd_sc_ctrl0;
 		uint32_t osd_sc_i_wh_m1;
 		uint32_t osd_sc_o_h_start_end;
drivers/gpu/drm/meson/meson_plane.c +1 -0
index ed543227b00d..5e798c276037 100644
@@ -305,6 +305,7 @@ static void meson_plane_atomic_update(struct drm_plane *plane,
 	priv->viu.osd1_addr = gem->paddr;
 	priv->viu.osd1_stride = fb->pitches[0];
 	priv->viu.osd1_height = fb->height;
+	priv->viu.osd1_width = fb->width;
 
 	if (!meson_plane->enabled) {
 		/* Reset OSD1 before enabling it on GXL+ SoCs */
Signature · 6 lines
2.22.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#4 lore raw

PATCH v3 4/9 drm/meson: add RDMA module driver

In reply to Neil Armstrong

The VPU embeds a "Register DMA" that can write a sequence of registers on the VPU AHB bus, either manually or triggered by an internal IRQ event like VSYNC or a line input counter.

The initial implementation handles a single channel (over 8), triggered by the VSYNC irq and does not handle the RDMA irq.

The RDMA will be usefull to reset and program the AFBC decoder unit on each vsync without involving the interrupt handler that can be masked for a log period of time, producing display glitches.

---
 drivers/gpu/drm/meson/Makefile     |   1 +
 drivers/gpu/drm/meson/meson_drv.h  |   6 ++
 drivers/gpu/drm/meson/meson_rdma.c | 135 +++++++++++++++++++++++++++++
 drivers/gpu/drm/meson/meson_rdma.h |  21 +++++
 4 files changed, 163 insertions(+)
 create mode 100644 drivers/gpu/drm/meson/meson_rdma.c
 create mode 100644 drivers/gpu/drm/meson/meson_rdma.h
drivers/gpu/drm/meson/Makefile +1 -0
index c389e2399133..fb7f5440fe57 100644
@@ -1,6 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0-only
 meson-drm-y := meson_drv.o meson_plane.o meson_crtc.o meson_venc_cvbs.o
 meson-drm-y += meson_viu.o meson_vpp.o meson_venc.o meson_vclk.o meson_overlay.o
+meson-drm-y += meson_rdma.o
 
 obj-$(CONFIG_DRM_MESON) += meson-drm.o
 obj-$(CONFIG_DRM_MESON_DW_HDMI) += meson_dw_hdmi.o
drivers/gpu/drm/meson/meson_drv.h +6 -0
index 3287282450a9..ea39be913c0f 100644
@@ -123,6 +123,12 @@ struct meson_drm {
 		bool venc_repeat;
 		bool hdmi_use_enci;
 	} venc;
+
+	struct {
+		dma_addr_t addr_phys;
+		uint32_t *addr;
+		unsigned int offset;
+	} rdma;
 };
 
 static inline int meson_vpu_is_compatible(struct meson_drm *priv,
drivers/gpu/drm/meson/meson_rdma.c new file +135 -0

135 lines not on this page · raw message

drivers/gpu/drm/meson/meson_rdma.h new file +21 -0
new file mode 100644
index 000000000000..3870bff7b47f
@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (C) 2019 BayLibre, SAS
+ * Author: Neil Armstrong <[email protected]>
+ */
+
+#ifndef __MESON_RDMA_H
+#define __MESON_RDMA_H
+
+#include "meson_drv.h"
+
+int meson_rdma_init(struct meson_drm *priv);
+void meson_rdma_free(struct meson_drm *priv);
+void meson_rdma_setup(struct meson_drm *priv);
+void meson_rdma_reset(struct meson_drm *priv);
+void meson_rdma_stop(struct meson_drm *priv);
+
+void meson_rdma_writel_sync(struct meson_drm *priv, uint32_t val, uint32_t reg);
+void meson_rdma_flush(struct meson_drm *priv);
+
+#endif /* __MESON_RDMA_H */
Signature · 6 lines
2.22.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#5 lore raw

PATCH v3 5/9 drm/meson: Add AFBCD module driver

In reply to Neil Armstrong

This adds the driver for the ARM Framebuffer Compression decoders found in the Amlogic GXM and G12A SoCs.

The Amlogic GXM and G12A AFBC decoder are totally different, the GXM only handling only the AFBC v1.0 modes and the G12A decoder handling the AFBC v1.2 modes.

The G12A AFBC decoder is an external IP integrated in the video pipeline, and the GXM AFBC decoder seems to the an Amlogic custom decoder more tighly integrated in the video pipeline.

The GXM AFBC decoder can handle only one AFBC plane for 2 available OSD planes available in HW, and the G12A AFBC decoder can handle up to 4 AFBC planes for up to 3 OSD planes available in HW.

The Amlogic GXM supports 16x16 SPARSE and 16x16 SPLIT AFBC buffers up to 4k.

On the other side, for G12A SPLIT is mandatory in 16x16 block mode, but for 4k modes 32x8+SPLIT AFBC buffers is manadatory for performances reasons.

The RDMA is used here to reset and program the AFBC decoder unit on each vsync without involving the interrupt handler that can be masked for a long period of time, producing display glitches.

For this we use the meson_rdma_writel_sync() which adds the register write tuple (VPU register offset and register value) to the RDMA buffer and write the value to the HW.

When enabled, the RDMA is enabled to rewrite the same sequence at the next VSYNC event, until a new buffer is committed to the OSD plane.

The the Amlogic G12A is switched to RDMA, the Amlogic GXM Decoder doesn't need a reset/reprogram at each vsync, but needs to keep the vsync interrupt enabled to trigger the RDMA module.

---
 drivers/gpu/drm/meson/Makefile          |   2 +-
 drivers/gpu/drm/meson/meson_drv.c       |  50 ++-
 drivers/gpu/drm/meson/meson_drv.h       |  12 +
 drivers/gpu/drm/meson/meson_osd_afbcd.c | 389 ++++++++++++++++++++++++
 drivers/gpu/drm/meson/meson_osd_afbcd.h |  28 ++
 5 files changed, 473 insertions(+), 8 deletions(-)
 create mode 100644 drivers/gpu/drm/meson/meson_osd_afbcd.c
 create mode 100644 drivers/gpu/drm/meson/meson_osd_afbcd.h
drivers/gpu/drm/meson/Makefile +1 -1
index fb7f5440fe57..28a519cdf66b 100644
@@ -1,7 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0-only
 meson-drm-y := meson_drv.o meson_plane.o meson_crtc.o meson_venc_cvbs.o
 meson-drm-y += meson_viu.o meson_vpp.o meson_venc.o meson_vclk.o meson_overlay.o
-meson-drm-y += meson_rdma.o
+meson-drm-y += meson_rdma.o meson_osd_afbcd.o
 
 obj-$(CONFIG_DRM_MESON) += meson-drm.o
 obj-$(CONFIG_DRM_MESON_DW_HDMI) += meson_dw_hdmi.o
drivers/gpu/drm/meson/meson_drv.c +43 -7

112 lines not on this page · raw message

drivers/gpu/drm/meson/meson_drv.h +12 -0
index ea39be913c0f..f1515381bd02 100644
@@ -16,6 +16,7 @@ struct drm_crtc;
 struct drm_device;
 struct drm_plane;
 struct meson_drm;
+struct meson_afbcd_ops;
 
 enum vpu_compatible {
 	VPU_COMPATIBLE_GXBB = 0,
@@ -24,6 +25,11 @@ enum vpu_compatible {
 	VPU_COMPATIBLE_G12A = 3,
 };
 
+struct meson_drm_match_data {
+	enum vpu_compatible compat;
+	struct meson_afbcd_ops *afbcd_ops;
+};
+
 struct meson_drm {
 	struct device *dev;
 	enum vpu_compatible compat;
@@ -129,6 +135,12 @@ struct meson_drm {
 		uint32_t *addr;
 		unsigned int offset;
 	} rdma;
+
+	struct {
+		struct meson_afbcd_ops *ops;
+		u64 modifier;
+		u32 format;
+	} afbcd;
 };
 
 static inline int meson_vpu_is_compatible(struct meson_drm *priv,
drivers/gpu/drm/meson/meson_osd_afbcd.c new file +389 -0

389 lines not on this page · raw message

drivers/gpu/drm/meson/meson_osd_afbcd.h new file +28 -0
new file mode 100644
index 000000000000..5e5523304f42
@@ -0,0 +1,28 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (C) 2019 BayLibre, SAS
+ * Author: Neil Armstrong <[email protected]>
+ */
+
+#ifndef __MESON_OSD_AFBCD_H
+#define __MESON_OSD_AFBCD_H
+
+#include "meson_drv.h"
+
+/* This is an internal address used to transfer pixel from AFBC to the VIU */
+#define MESON_G12A_AFBCD_OUT_ADDR	0x1000000
+
+struct meson_afbcd_ops {
+	int (*init)(struct meson_drm *priv);
+	int (*reset)(struct meson_drm *priv);
+	int (*enable)(struct meson_drm *priv);
+	int (*disable)(struct meson_drm *priv);
+	int (*setup)(struct meson_drm *priv);
+	int (*fmt_to_blk_mode)(u64 modifier, uint32_t format);
+	bool (*supported_fmt)(u64 modifier, uint32_t format);
+};
+
+extern struct meson_afbcd_ops meson_afbcd_gxm_ops;
+extern struct meson_afbcd_ops meson_afbcd_g12a_ops;
+
+#endif /* __MESON_OSD_AFBCD_H */
Signature · 6 lines
2.22.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#6 lore raw

PATCH v3 6/9 drm/meson: plane: add support for AFBC mode for OSD1 plane

In reply to Neil Armstrong

This adds all the OSD configuration plumbing to support the AFBC decoders path to display of the OSD1 plane.

The Amlogic GXM and G12A AFBC decoders are integrated very differently.

The Amlogic GXM has a direct output path to the OSD1 VIU pixel input, because the GXM AFBC decoder seem to be a custom IP developed by Amlogic.

On the other side, the Amlogic G12A AFBC decoder seems to be an external IP that emit pixels on an AXI master hooked to a "Mali Unpack" block feeding the OSD1 VIU pixel input.
This uses a weird "0x1000000" internal HW physical address on both sides to transfer the pixels.

For Amlogic GXM, the supported pixel formats are the same as the normal linear OSD1 mode.

On the other side, Amlogic added support for all AFBC v1.2 formats for the G12A AFBC integration.

For simplicity, we stick to the already supported formats for now.

---
 drivers/gpu/drm/meson/meson_crtc.c  |   2 +
 drivers/gpu/drm/meson/meson_drv.h   |   4 +
 drivers/gpu/drm/meson/meson_plane.c | 228 ++++++++++++++++++++++++----
 3 files changed, 203 insertions(+), 31 deletions(-)
drivers/gpu/drm/meson/meson_crtc.c +2 -0
index 57ae1c13d1e6..d478fa232951 100644
@@ -281,6 +281,8 @@ void meson_crtc_irq(struct meson_drm *priv)
 	if (priv->viu.osd1_enabled && priv->viu.osd1_commit) {
 		writel_relaxed(priv->viu.osd1_ctrl_stat,
 				priv->io_base + _REG(VIU_OSD1_CTRL_STAT));
+		writel_relaxed(priv->viu.osd1_ctrl_stat2,
+				priv->io_base + _REG(VIU_OSD1_CTRL_STAT2));
 		writel_relaxed(priv->viu.osd1_blk0_cfg[0],
 				priv->io_base + _REG(VIU_OSD1_BLK0_CFG_W0));
 		writel_relaxed(priv->viu.osd1_blk0_cfg[1],
drivers/gpu/drm/meson/meson_drv.h +4 -0
index f1515381bd02..f9a0c8e9d4d0 100644
@@ -53,8 +53,12 @@ struct meson_drm {
 		bool osd1_enabled;
 		bool osd1_interlace;
 		bool osd1_commit;
+		bool osd1_afbcd;
 		uint32_t osd1_ctrl_stat;
+		uint32_t osd1_ctrl_stat2;
 		uint32_t osd1_blk0_cfg[5];
+		uint32_t osd1_blk1_cfg4;
+		uint32_t osd1_blk2_cfg4;
 		uint32_t osd1_addr;
 		uint32_t osd1_stride;
 		uint32_t osd1_height;
drivers/gpu/drm/meson/meson_plane.c +197 -31

317 lines not on this page · raw message

Signature · 6 lines
2.22.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#7 lore raw

PATCH v3 7/9 drm/meson: viu: add AFBC modules routing functions

In reply to Neil Armstrong

The Amlogic G12A AFBC Decoder pixel input need to be routed diferently than the Amlogic GXM AFBC decoder, this adds support for routing the VIU OSD1 pixel source to the AFBC "Mali Unpack" module.

This "Mali Unpack" module is also configured with a static RGBA mapping for now until we support more pixel formats.

---
 drivers/gpu/drm/meson/meson_viu.c | 81 +++++++++++++++++++++++++++++++
 drivers/gpu/drm/meson/meson_viu.h |  4 ++
 2 files changed, 85 insertions(+)
drivers/gpu/drm/meson/meson_viu.c +81 -0
index 68cf2c2eca5f..fc246248226b 100644
@@ -7,6 +7,9 @@
  */
 
 #include <linux/export.h>
+#include <linux/bitfield.h>
+
+#include <drm/drm_fourcc.h>
 
 #include "meson_drv.h"
 #include "meson_viu.h"
@@ -335,6 +338,79 @@ void meson_viu_osd1_reset(struct meson_drm *priv)
 	meson_viu_load_matrix(priv);
 }
 
+#define OSD1_MALI_ORDER_ABGR				\
+	(FIELD_PREP(VIU_OSD1_MALI_AFBCD_A_REORDER,	\
+		    VIU_OSD1_MALI_REORDER_A) |		\
+	 FIELD_PREP(VIU_OSD1_MALI_AFBCD_B_REORDER,	\
+		    VIU_OSD1_MALI_REORDER_B) |		\
+	 FIELD_PREP(VIU_OSD1_MALI_AFBCD_G_REORDER,	\
+		    VIU_OSD1_MALI_REORDER_G) |		\
+	 FIELD_PREP(VIU_OSD1_MALI_AFBCD_R_REORDER,	\
+		    VIU_OSD1_MALI_REORDER_R))
+
+#define OSD1_MALI_ORDER_ARGB				\
+	(FIELD_PREP(VIU_OSD1_MALI_AFBCD_A_REORDER,	\
+		    VIU_OSD1_MALI_REORDER_A) |		\
+	 FIELD_PREP(VIU_OSD1_MALI_AFBCD_B_REORDER,	\
+		    VIU_OSD1_MALI_REORDER_R) |		\
+	 FIELD_PREP(VIU_OSD1_MALI_AFBCD_G_REORDER,	\
+		    VIU_OSD1_MALI_REORDER_G) |		\
+	 FIELD_PREP(VIU_OSD1_MALI_AFBCD_R_REORDER,	\
+		    VIU_OSD1_MALI_REORDER_B))
+
+void meson_viu_g12a_enable_osd1_afbc(struct meson_drm *priv)
+{
+	u32 afbc_order = OSD1_MALI_ORDER_ARGB;
+
+	/* Enable Mali AFBC Unpack */
+	writel_bits_relaxed(VIU_OSD1_MALI_UNPACK_EN,
+			    VIU_OSD1_MALI_UNPACK_EN,
+			    priv->io_base + _REG(VIU_OSD1_MALI_UNPACK_CTRL));
+
+	switch (priv->afbcd.format) {
+	case DRM_FORMAT_XBGR8888:
+	case DRM_FORMAT_ABGR8888:
+		afbc_order = OSD1_MALI_ORDER_ABGR;
+		break;
+	}
+
+	/* Setup RGBA Reordering */
+	writel_bits_relaxed(VIU_OSD1_MALI_AFBCD_A_REORDER |
+			    VIU_OSD1_MALI_AFBCD_B_REORDER |
+			    VIU_OSD1_MALI_AFBCD_G_REORDER |
+			    VIU_OSD1_MALI_AFBCD_R_REORDER,
+			    afbc_order,
+			    priv->io_base + _REG(VIU_OSD1_MALI_UNPACK_CTRL));
+
+	/* Select AFBCD path for OSD1 */
+	writel_bits_relaxed(OSD_PATH_OSD_AXI_SEL_OSD1_AFBCD,
+			    OSD_PATH_OSD_AXI_SEL_OSD1_AFBCD,
+			    priv->io_base + _REG(OSD_PATH_MISC_CTRL));
+}
+
+void meson_viu_g12a_disable_osd1_afbc(struct meson_drm *priv)
+{
+	/* Disable AFBCD path for OSD1 */
+	writel_bits_relaxed(OSD_PATH_OSD_AXI_SEL_OSD1_AFBCD, 0,
+			    priv->io_base + _REG(OSD_PATH_MISC_CTRL));
+
+	/* Disable AFBCD unpack */
+	writel_bits_relaxed(VIU_OSD1_MALI_UNPACK_EN, 0,
+			    priv->io_base + _REG(VIU_OSD1_MALI_UNPACK_CTRL));
+}
+
+void meson_viu_gxm_enable_osd1_afbc(struct meson_drm *priv)
+{
+	writel_bits_relaxed(MALI_AFBC_MISC, FIELD_PREP(MALI_AFBC_MISC, 0x90),
+			    priv->io_base + _REG(VIU_MISC_CTRL1));
+}
+
+void meson_viu_gxm_disable_osd1_afbc(struct meson_drm *priv)
+{
+	writel_bits_relaxed(MALI_AFBC_MISC, FIELD_PREP(MALI_AFBC_MISC, 0x00),
+			    priv->io_base + _REG(VIU_MISC_CTRL1));
+}
+
 static inline uint32_t meson_viu_osd_burst_length_reg(uint32_t length)
 {
 	uint32_t val = (((length & 0x80) % 24) / 12);
@@ -420,8 +496,13 @@ void meson_viu_init(struct meson_drm *priv)
 
 		writel_bits_relaxed(DOLBY_BYPASS_EN(0xc), DOLBY_BYPASS_EN(0xc),
 				    priv->io_base + _REG(DOLBY_PATH_CTRL));
+
+		meson_viu_g12a_disable_osd1_afbc(priv);
 	}
 
+	if (meson_vpu_is_compatible(priv, VPU_COMPATIBLE_GXM))
+		meson_viu_gxm_disable_osd1_afbc(priv);
+
 	priv->viu.osd1_enabled = false;
 	priv->viu.osd1_commit = false;
 	priv->viu.osd1_interlace = false;
drivers/gpu/drm/meson/meson_viu.h +4 -0
index e297772d967f..e4a2f24d7c38 100644
@@ -63,6 +63,10 @@
 #define OSD_PENDING_STAT_CLEAN	BIT(1)
 
 void meson_viu_osd1_reset(struct meson_drm *priv);
+void meson_viu_g12a_enable_osd1_afbc(struct meson_drm *priv);
+void meson_viu_g12a_disable_osd1_afbc(struct meson_drm *priv);
+void meson_viu_gxm_enable_osd1_afbc(struct meson_drm *priv);
+void meson_viu_gxm_disable_osd1_afbc(struct meson_drm *priv);
 void meson_viu_init(struct meson_drm *priv);
 
 #endif /* __MESON_VIU_H */
Signature · 6 lines
2.22.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#8 lore raw

PATCH v3 8/9 drm/meson: hold 32 lines after vsync to give time for AFBC start

In reply to Neil Armstrong

When using an AFBC encoded frame, the AFBC Decoder must be resetted, configured and enabled at each vsync IRQ.

To leave time for that, use the maximum lines hold time to give time for AFBC setup and avoid visual glitches.

---
 drivers/gpu/drm/meson/meson_viu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/meson/meson_viu.c +1 -1
index fc246248226b..304f8ff1339c 100644
@@ -438,7 +438,7 @@ void meson_viu_init(struct meson_drm *priv)
 
 	/* Initialize OSD1 fifo control register */
 	reg = VIU_OSD_DDR_PRIORITY_URGENT |
-		VIU_OSD_HOLD_FIFO_LINES(4) |
+		VIU_OSD_HOLD_FIFO_LINES(31) |
 		VIU_OSD_FIFO_DEPTH_VAL(32) | /* fifo_depth_val: 32*8=256 */
 		VIU_OSD_WORDS_PER_BURST(4) | /* 4 words in 1 burst */
 		VIU_OSD_FIFO_LIMITS(2);      /* fifo_lim: 2*16=32 */
Signature · 6 lines
2.22.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
maintainer Reviewed-by on v3 8/9
#9 lore raw

Re: [PATCH v3 8/9] drm/meson: hold 32 lines after vsync to give time for AFBC start

Quoted 1 line · Neil Armstrong

When using an AFBC encoded frame, the AFBC Decoder must be resetted,

minor grammar nit: s/resetted/reset/

Quoted 6 lines

configured and enabled at each vsync IRQ.

To leave time for that, use the maximum lines hold time to give time for AFBC setup and avoid visual glitches.

otherwise...

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#10 lore raw

PATCH v3 9/9 drm/meson: crtc: add OSD1 plane AFBC commit

In reply to Neil Armstrong

Finally, setup the VIU registers and start the AFBC decoder to support displaying AFBC encoded buffers on Amlogic GXM and G12A SoCs.

The RDMA is used here to reset and program the AFBC decoder unit on each vsync without involving the interrupt handler that can be masked for a long period of time, producing display glitches.

The vsync irq must still be left enabled otherwise the RDMA modules isn't trigerred when the interrupt line is masked.

---
 drivers/gpu/drm/meson/meson_crtc.c | 77 ++++++++++++++++++++++++++++--
 1 file changed, 72 insertions(+), 5 deletions(-)
drivers/gpu/drm/meson/meson_crtc.c +72 -5
index d478fa232951..8366a0ed37af 100644
@@ -23,7 +23,9 @@
 #include "meson_registers.h"
 #include "meson_venc.h"
 #include "meson_viu.h"
+#include "meson_rdma.h"
 #include "meson_vpp.h"
+#include "meson_osd_afbcd.h"
 
 #define MESON_G12A_VIU_OFFSET	0x5ec0
 
@@ -35,7 +37,11 @@ struct meson_crtc {
 	struct meson_drm *priv;
 	void (*enable_osd1)(struct meson_drm *priv);
 	void (*enable_vd1)(struct meson_drm *priv);
+	void (*enable_osd1_afbc)(struct meson_drm *priv);
+	void (*disable_osd1_afbc)(struct meson_drm *priv);
 	unsigned int viu_offset;
+	bool vsync_forced;
+	bool vsync_disabled;
 };
 #define to_meson_crtc(x) container_of(x, struct meson_crtc, base)
 
@@ -46,6 +52,7 @@ static int meson_crtc_enable_vblank(struct drm_crtc *crtc)
 	struct meson_crtc *meson_crtc = to_meson_crtc(crtc);
 	struct meson_drm *priv = meson_crtc->priv;
 
+	meson_crtc->vsync_disabled = false;
 	meson_venc_enable_vsync(priv);
 
 	return 0;
@@ -56,7 +63,10 @@ static void meson_crtc_disable_vblank(struct drm_crtc *crtc)
 	struct meson_crtc *meson_crtc = to_meson_crtc(crtc);
 	struct meson_drm *priv = meson_crtc->priv;
 
-	meson_venc_disable_vsync(priv);
+	if (!meson_crtc->vsync_forced) {
+		meson_crtc->vsync_disabled = true;
+		meson_venc_disable_vsync(priv);
+	}
 }
 
 static const struct drm_crtc_funcs meson_crtc_funcs = {
@@ -236,6 +246,26 @@ static void meson_crtc_enable_osd1(struct meson_drm *priv)
 			    priv->io_base + _REG(VPP_MISC));
 }
 
+static void meson_crtc_g12a_enable_osd1_afbc(struct meson_drm *priv)
+{
+	writel_relaxed(priv->viu.osd1_blk2_cfg4,
+		       priv->io_base + _REG(VIU_OSD1_BLK2_CFG_W4));
+
+	writel_bits_relaxed(OSD_MEM_LINEAR_ADDR, OSD_MEM_LINEAR_ADDR,
+			    priv->io_base + _REG(VIU_OSD1_CTRL_STAT));
+
+	writel_relaxed(priv->viu.osd1_blk1_cfg4,
+		       priv->io_base + _REG(VIU_OSD1_BLK1_CFG_W4));
+
+	meson_viu_g12a_enable_osd1_afbc(priv);
+
+	writel_bits_relaxed(OSD_MEM_LINEAR_ADDR, OSD_MEM_LINEAR_ADDR,
+			    priv->io_base + _REG(VIU_OSD1_CTRL_STAT));
+
+	writel_bits_relaxed(OSD_MALI_SRC_EN, OSD_MALI_SRC_EN,
+			    priv->io_base + _REG(VIU_OSD1_BLK0_CFG_W0));
+}
+
 static void meson_g12a_crtc_enable_osd1(struct meson_drm *priv)
 {
 	writel_relaxed(priv->viu.osd_blend_din0_scope_h,
@@ -293,6 +323,20 @@ void meson_crtc_irq(struct meson_drm *priv)
 				priv->io_base + _REG(VIU_OSD1_BLK0_CFG_W3));
 		writel_relaxed(priv->viu.osd1_blk0_cfg[4],
 				priv->io_base + _REG(VIU_OSD1_BLK0_CFG_W4));
+
+		if (priv->viu.osd1_afbcd) {
+			if (meson_crtc->enable_osd1_afbc)
+				meson_crtc->enable_osd1_afbc(priv);
+		} else {
+			if (meson_crtc->disable_osd1_afbc)
+				meson_crtc->disable_osd1_afbc(priv);
+			if (priv->afbcd.ops) {
+				priv->afbcd.ops->reset(priv);
+				priv->afbcd.ops->disable(priv);
+			}
+			meson_crtc->vsync_forced = false;
+		}
+
 		writel_relaxed(priv->viu.osd_sc_ctrl0,
 				priv->io_base + _REG(VPP_OSD_SC_CTRL0));
 		writel_relaxed(priv->viu.osd_sc_i_wh_m1,
@@ -314,15 +358,25 @@ void meson_crtc_irq(struct meson_drm *priv)
 		writel_relaxed(priv->viu.osd_sc_v_ctrl0,
 				priv->io_base + _REG(VPP_OSD_VSC_CTRL0));
 
-		meson_canvas_config(priv->canvas, priv->canvas_id_osd1,
-				priv->viu.osd1_addr, priv->viu.osd1_stride,
-				priv->viu.osd1_height, MESON_CANVAS_WRAP_NONE,
-				MESON_CANVAS_BLKMODE_LINEAR, 0);
+		if (!priv->viu.osd1_afbcd)
+			meson_canvas_config(priv->canvas, priv->canvas_id_osd1,
+					    priv->viu.osd1_addr,
+					    priv->viu.osd1_stride,
+					    priv->viu.osd1_height,
+					    MESON_CANVAS_WRAP_NONE,
+					    MESON_CANVAS_BLKMODE_LINEAR, 0);
 
 		/* Enable OSD1 */
 		if (meson_crtc->enable_osd1)
 			meson_crtc->enable_osd1(priv);
 
+		if (priv->viu.osd1_afbcd) {
+			priv->afbcd.ops->reset(priv);
+			priv->afbcd.ops->setup(priv);
+			priv->afbcd.ops->enable(priv);
+			meson_crtc->vsync_forced = true;
+		}
+
 		priv->viu.osd1_commit = false;
 	}
 
@@ -545,6 +599,9 @@ void meson_crtc_irq(struct meson_drm *priv)
 		priv->viu.vd1_commit = false;
 	}
 
+	if (meson_crtc->vsync_disabled)
+		return;
+
 	drm_crtc_handle_vblank(priv->crtc);
 
 	spin_lock_irqsave(&priv->drm->event_lock, flags);
@@ -581,10 +638,20 @@ int meson_crtc_create(struct meson_drm *priv)
 		meson_crtc->enable_osd1 = meson_g12a_crtc_enable_osd1;
 		meson_crtc->enable_vd1 = meson_g12a_crtc_enable_vd1;
 		meson_crtc->viu_offset = MESON_G12A_VIU_OFFSET;
+		meson_crtc->enable_osd1_afbc =
+					meson_crtc_g12a_enable_osd1_afbc;
+		meson_crtc->disable_osd1_afbc =
+					meson_viu_g12a_disable_osd1_afbc;
 		drm_crtc_helper_add(crtc, &meson_g12a_crtc_helper_funcs);
 	} else {
 		meson_crtc->enable_osd1 = meson_crtc_enable_osd1;
 		meson_crtc->enable_vd1 = meson_crtc_enable_vd1;
+		if (meson_vpu_is_compatible(priv, VPU_COMPATIBLE_GXM)) {
+			meson_crtc->enable_osd1_afbc =
+					meson_viu_gxm_enable_osd1_afbc;
+			meson_crtc->disable_osd1_afbc =
+					meson_viu_gxm_disable_osd1_afbc;
+		}
 		drm_crtc_helper_add(crtc, &meson_crtc_helper_funcs);
 	}
 
Signature · 6 lines
2.22.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
maintainer Reviewed-by on the v3 series
#11 lore raw
In reply to Neil Armstrong
Quoted 5 lines · Neil Armstrong

This adds support for the ARM Framebuffer Compression decoders found in the Amlogic GXM and G12A SoCs.

This patchset is a merge of v2 "drm/meson: add AFBC support" at [3] and v2 "drm/meson: implement RDMA for AFBC reset on vsync" at [4].

Oops, replied to the wrong series earlier...

Quoted 53 lines

The VPU embeds a "Register DMA" that can write a sequence of registers on the VPU AHB bus, either manually or triggered by an internal IRQ event like VSYNC or a line input counter.

The Amlogic GXM and G12A AFBC decoder are totally different, the GXM only handling only the AFBC v1.0 modes and the G12A decoder handling the AFBC v1.2 modes.

The G12A AFBC decoder is an external IP integrated in the video pipeline, and the GXM AFBC decoder seems to the an Amlogic custom decoder more tighly integrated in the video pipeline.

The GXM AFBC decoder can handle only one AFBC plane for 2 available OSD planes available in HW, and the G12A AFBC decoder can handle up to 4 AFBC planes for up to 3 OSD planes available in HW.

The Amlogic GXM supports 16x16 SPARSE and 16x16 SPLIT AFBC buffers up to 4k.

On the other side, for G12A SPLIT is mandatory in 16x16 block mode, but for 4k modes 32x8+SPLIT AFBC buffers is manadatory for performances reasons.

The Amlogic GXM and G12A AFBC decoders are integrated very differently.

The Amlogic GXM has a direct output path to the OSD1 VIU pixel input, because the GXM AFBC decoder seem to be a custom IP developed by Amlogic.

On the other side, the Amlogic G12A AFBC decoder seems to be an external IP that emit pixels on an AXI master hooked to a "Mali Unpack" block feeding the OSD1 VIU pixel input.
This uses a weird "0x1000000" internal HW physical address on both sides to transfer the pixels.

For Amlogic GXM, the supported pixel formats are the same as the normal linear OSD1 mode.

On the other side, Amlogic added support for all AFBC v1.2 formats for the G12A AFBC integration.

The initial RDMA implementation handles a single channel (over 8), triggered by the VSYNC irq and does not handle the RDMA irq.

The RDMA will be usefull to reset and program the AFBC decoder unit on each vsync without involving the interrupt handler that can be masked for a long period of time, producing display glitches.

For this we use the meson_rdma_writel_sync() which adds the register write tuple (VPU register offset and register value) to the RDMA buffer and write the value to the HW.

When enabled, the RDMA is enabled to rewritte the same sequence at the next VSYNC event, until a new buffer is committed to the OSD plane.

Quoted 6 lines

For testing, the only available AFBC buffer generation is the Android Yukawa Dvalin Android Mali blobs found at [1].

Both SoCs has been tested using buffers generated under AOSP, but only G12A was tested using a runtime stream of AFBC buffers, GXM was only tested using static buffers loaded from files.

Kevin
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#12 lore raw
Quoted 66 lines · Kevin Hilman
Quoted 5 lines · Neil Armstrong

This adds support for the ARM Framebuffer Compression decoders found in the Amlogic GXM and G12A SoCs.

This patchset is a merge of v2 "drm/meson: add AFBC support" at [3] and v2 "drm/meson: implement RDMA for AFBC reset on vsync" at [4].

Oops, replied to the wrong series earlier...

Quoted 56 lines

The VPU embeds a "Register DMA" that can write a sequence of registers on the VPU AHB bus, either manually or triggered by an internal IRQ event like VSYNC or a line input counter.

The Amlogic GXM and G12A AFBC decoder are totally different, the GXM only handling only the AFBC v1.0 modes and the G12A decoder handling the AFBC v1.2 modes.

The G12A AFBC decoder is an external IP integrated in the video pipeline, and the GXM AFBC decoder seems to the an Amlogic custom decoder more tighly integrated in the video pipeline.

The GXM AFBC decoder can handle only one AFBC plane for 2 available OSD planes available in HW, and the G12A AFBC decoder can handle up to 4 AFBC planes for up to 3 OSD planes available in HW.

The Amlogic GXM supports 16x16 SPARSE and 16x16 SPLIT AFBC buffers up to 4k.

On the other side, for G12A SPLIT is mandatory in 16x16 block mode, but for 4k modes 32x8+SPLIT AFBC buffers is manadatory for performances reasons.

The Amlogic GXM and G12A AFBC decoders are integrated very differently.

The Amlogic GXM has a direct output path to the OSD1 VIU pixel input, because the GXM AFBC decoder seem to be a custom IP developed by Amlogic.

On the other side, the Amlogic G12A AFBC decoder seems to be an external IP that emit pixels on an AXI master hooked to a "Mali Unpack" block feeding the OSD1 VIU pixel input.
This uses a weird "0x1000000" internal HW physical address on both sides to transfer the pixels.

For Amlogic GXM, the supported pixel formats are the same as the normal linear OSD1 mode.

On the other side, Amlogic added support for all AFBC v1.2 formats for the G12A AFBC integration.

The initial RDMA implementation handles a single channel (over 8), triggered by the VSYNC irq and does not handle the RDMA irq.

The RDMA will be usefull to reset and program the AFBC decoder unit on each vsync without involving the interrupt handler that can be masked for a long period of time, producing display glitches.

For this we use the meson_rdma_writel_sync() which adds the register write tuple (VPU register offset and register value) to the RDMA buffer and write the value to the HW.

When enabled, the RDMA is enabled to rewritte the same sequence at the next VSYNC event, until a new buffer is committed to the OSD plane.

For testing, the only available AFBC buffer generation is the Android Yukawa Dvalin Android Mali blobs found at [1].

Quoted 8 lines
Quoted 3 lines

Both SoCs has been tested using buffers generated under AOSP, but only G12A was tested using a runtime stream of AFBC buffers, GXM was only tested using static buffers loaded from files.

Kevin

Thanks,
Applied to drm-misc-next with typo fixup in patches 8 & 9 and review tags

Neil
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel