Date:   Wed, 14 Nov 2018 15:59:51 -0700
From:   Jordan Crouse <jcrouse@...eaurora.org>
To:     Jonathan Marek <jonathan@...ek.ca>
Cc:     freedreno@...ts.freedesktop.org, Rob Clark <robdclark@...il.com>,
        David Airlie <airlied@...ux.ie>,
        Nicolas Dechesne <nicolas.dechesne@...aro.org>,
        Arnd Bergmann <arnd@...db.de>,
        Daniel Vetter <daniel.vetter@...ll.ch>,
        Kees Cook <keescook@...omium.org>,
        Bjorn Andersson <bjorn.andersson@...aro.org>,
        open list <linux-kernel@...r.kernel.org>,
        "open list:DRM DRIVER FOR MSM ADRENO GPU" 
        <linux-arm-msm@...r.kernel.org>,
        "open list:DRM DRIVER FOR MSM ADRENO GPU" 
        <dri-devel@...ts.freedesktop.org>
Subject: Re: [PATCH 6/9] drm/msm/adreno: add a2xx

On Wed, Nov 14, 2018 at 05:24:13PM -0500, Jonathan Marek wrote:
> derived from the a3xx driver and tested on the following hardware:
> imx51-zii-rdu1 (a200 with 128kb gmem)
> imx53-qsrb (a200)
> msm8060-tenderloin (a220)

This looks sane but it has been quite a while since I've thought about 2xx.

If this works on real world hardware thats is a pretty good endorsement.

Only nit - I would turn the /* XXX */ comments into real comments. Since KGSL
dropped 2xx support some time in the 3.4 timeframe I think it is a good idea
to record what the old code did for posterity.

With that:
Reviewed-by: Jordan Crouse <jcrouse@...eaurora.org>


> Signed-off-by: Jonathan Marek <jonathan@...ek.ca>
> ---
>  drivers/gpu/drm/msm/Makefile               |   1 +
>  drivers/gpu/drm/msm/adreno/a2xx_gpu.c      | 445 +++++++++++++++++++++
>  drivers/gpu/drm/msm/adreno/a2xx_gpu.h      |  21 +
>  drivers/gpu/drm/msm/adreno/adreno_device.c |  33 ++
>  drivers/gpu/drm/msm/adreno/adreno_gpu.c    |  27 +-
>  drivers/gpu/drm/msm/adreno/adreno_gpu.h    |  15 +
>  6 files changed, 534 insertions(+), 8 deletions(-)
>  create mode 100644 drivers/gpu/drm/msm/adreno/a2xx_gpu.c
>  create mode 100644 drivers/gpu/drm/msm/adreno/a2xx_gpu.h
> 
> diff --git a/drivers/gpu/drm/msm/Makefile b/drivers/gpu/drm/msm/Makefile
> index 19ab521d4c3a..0170766393ea 100644
> --- a/drivers/gpu/drm/msm/Makefile
> +++ b/drivers/gpu/drm/msm/Makefile
> @@ -6,6 +6,7 @@ ccflags-$(CONFIG_DRM_MSM_DSI) += -Idrivers/gpu/drm/msm/dsi
>  msm-y := \
>  	adreno/adreno_device.o \
>  	adreno/adreno_gpu.o \
> +	adreno/a2xx_gpu.o \
>  	adreno/a3xx_gpu.o \
>  	adreno/a4xx_gpu.o \
>  	adreno/a5xx_gpu.o \
> diff --git a/drivers/gpu/drm/msm/adreno/a2xx_gpu.c b/drivers/gpu/drm/msm/adreno/a2xx_gpu.c
> new file mode 100644
> index 000000000000..65b2352408fa
> --- /dev/null
> +++ b/drivers/gpu/drm/msm/adreno/a2xx_gpu.c
> @@ -0,0 +1,445 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/* Copyright (c) 2018 The Linux Foundation. All rights reserved. */
> +
> +#include "a2xx_gpu.h"
> +
> +extern bool hang_debug;
> +
> +static void a2xx_dump(struct msm_gpu *gpu);
> +static bool a2xx_idle(struct msm_gpu *gpu);
> +
> +static bool a2xx_me_init(struct msm_gpu *gpu)
> +{
> +	struct msm_ringbuffer *ring = gpu->rb[0];
> +
> +	OUT_PKT3(ring, CP_ME_INIT, 18);
> +
> +	/* All fields present (bits 9:0) */
> +	OUT_RING(ring, 0x000003ff);
> +	/* Disable/Enable Real-Time Stream processing (present but ignored) */
> +	OUT_RING(ring, 0x00000000);
> +	/* Enable (2D <-> 3D) implicit synchronization (present but ignored) */
> +	OUT_RING(ring, 0x00000000);
> +
> +	OUT_RING(ring, REG_A2XX_RB_SURFACE_INFO - 0x2000);
> +	OUT_RING(ring, REG_A2XX_PA_SC_WINDOW_OFFSET - 0x2000);
> +	OUT_RING(ring, REG_A2XX_VGT_MAX_VTX_INDX - 0x2000);
> +	OUT_RING(ring, REG_A2XX_SQ_PROGRAM_CNTL - 0x2000);
> +	OUT_RING(ring, REG_A2XX_RB_DEPTHCONTROL - 0x2000);
> +	OUT_RING(ring, REG_A2XX_PA_SU_POINT_SIZE - 0x2000);
> +	OUT_RING(ring, REG_A2XX_PA_SC_LINE_CNTL - 0x2000);
> +	OUT_RING(ring, REG_A2XX_PA_SU_POLY_OFFSET_FRONT_SCALE - 0x2000);
> +
> +	/* Vertex and Pixel Shader Start Addresses in instructions
> +	 * (3 DWORDS per instruction) */
> +	OUT_RING(ring, 0x80000180);
> +	/* Maximum Contexts */
> +	OUT_RING(ring, 0x00000001);
> +	/* Write Confirm Interval and The CP will wait the
> +	 * wait_interval * 16 clocks between polling  */
> +	OUT_RING(ring, 0x00000000);
> +	/* NQ and External Memory Swap */
> +	OUT_RING(ring, 0x00000000);
> +	/* protected mode error checking (0x1f2 is REG_AXXX_CP_INT_CNTL) */
> +	OUT_RING(ring, 0x200001f2);
> +	/* Disable header dumping and Header dump address */
> +	OUT_RING(ring, 0x00000000);
> +	/* Header dump size */
> +	OUT_RING(ring, 0x00000000);
> +
> +	/* enable protected mode */
> +	OUT_PKT3(ring, CP_SET_PROTECTED_MODE, 1);
> +	OUT_RING(ring, 1);
> +
> +	gpu->funcs->flush(gpu, ring);
> +	return a2xx_idle(gpu);
> +}
> +
> +static int a2xx_hw_init(struct msm_gpu *gpu)
> +{
> +	struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu);
> +	uint32_t *ptr, len;
> +	int i, ret;
> +
> +	DBG("%s", gpu->name);
> +
> +	gpu_write(gpu, REG_A2XX_RBBM_PM_OVERRIDE1, 0xfffffffe);
> +	gpu_write(gpu, REG_A2XX_RBBM_PM_OVERRIDE2, 0xffffffff);
> +
> +	/* XXX kgsl uses 0x00000001 after first reset on a22x */
> +	gpu_write(gpu, REG_A2XX_RBBM_SOFT_RESET, 0xffffffff);
> +	msleep(30);
> +	gpu_write(gpu, REG_A2XX_RBBM_SOFT_RESET, 0x00000000);
> +
> +	if (adreno_is_a225(adreno_gpu))
> +		gpu_write(gpu, REG_A2XX_SQ_FLOW_CONTROL, 0x18000000);
> +
> +	/* XXX kgsl uses 0x0000ffff for a20x */
> +	gpu_write(gpu, REG_A2XX_RBBM_CNTL, 0x00004442);
> +
> +	gpu_write(gpu, REG_A2XX_MH_MMU_CONFIG, 0);
> +	gpu_write(gpu, REG_A2XX_MH_MMU_MPU_BASE, 0);
> +	gpu_write(gpu, REG_A2XX_MH_MMU_MPU_END, 0xfffff000);
> +	gpu_write(gpu, REG_A2XX_MH_ARBITER_CONFIG,
> +		A2XX_MH_ARBITER_CONFIG_SAME_PAGE_LIMIT(16) |
> +		A2XX_MH_ARBITER_CONFIG_L1_ARB_ENABLE |
> +		A2XX_MH_ARBITER_CONFIG_L1_ARB_HOLD_ENABLE |
> +		A2XX_MH_ARBITER_CONFIG_PAGE_SIZE(1) |
> +		A2XX_MH_ARBITER_CONFIG_TC_REORDER_ENABLE |
> +		A2XX_MH_ARBITER_CONFIG_TC_ARB_HOLD_ENABLE |
> +		A2XX_MH_ARBITER_CONFIG_IN_FLIGHT_LIMIT_ENABLE |
> +		A2XX_MH_ARBITER_CONFIG_IN_FLIGHT_LIMIT(8) |
> +		A2XX_MH_ARBITER_CONFIG_CP_CLNT_ENABLE |
> +		A2XX_MH_ARBITER_CONFIG_VGT_CLNT_ENABLE |
> +		A2XX_MH_ARBITER_CONFIG_TC_CLNT_ENABLE |
> +		A2XX_MH_ARBITER_CONFIG_RB_CLNT_ENABLE |
> +		A2XX_MH_ARBITER_CONFIG_PA_CLNT_ENABLE);
> +	if (!adreno_is_a20x(adreno_gpu))
> +		gpu_write(gpu, REG_A2XX_MH_CLNT_INTF_CTRL_CONFIG1, 0x00032f07);
> +
> +	gpu_write(gpu, REG_A2XX_SQ_VS_PROGRAM, 0x00000000);
> +	gpu_write(gpu, REG_A2XX_SQ_PS_PROGRAM, 0x00000000);
> +
> +	gpu_write(gpu, REG_A2XX_RBBM_PM_OVERRIDE1, 0); /* 0x200 for msm8960? */
> +	gpu_write(gpu, REG_A2XX_RBBM_PM_OVERRIDE2, 0); /* 0x80/0x1a0 for a22x? */
> +
> +	/* XXX gsl doesn't set this */
> +	gpu_write(gpu, REG_A2XX_RBBM_DEBUG, 0x00080000);
> +
> +	gpu_write(gpu, REG_A2XX_RBBM_INT_CNTL, 0);
> +	gpu_write(gpu, REG_AXXX_CP_INT_CNTL, 0x80000000); /* RB INT */
> +	gpu_write(gpu, REG_A2XX_SQ_INT_CNTL, 0);
> +
> +	for (i = 3; i <= 5; i++)
> +		if ((SZ_16K << i) == adreno_gpu->gmem)
> +			break;
> +	gpu_write(gpu, REG_A2XX_RB_EDRAM_INFO, i);
> +
> +	ret = adreno_hw_init(gpu);
> +	if (ret)
> +		return ret;
> +
> +	/* NOTE: PM4/micro-engine firmware registers look to be the same
> +	 * for a2xx and a3xx.. we could possibly push that part down to
> +	 * adreno_gpu base class.  Or push both PM4 and PFP but
> +	 * parameterize the pfp ucode addr/data registers..
> +	 */
> +
> +	/* Load PM4: */
> +	ptr = (uint32_t *)(adreno_gpu->fw[ADRENO_FW_PM4]->data);
> +	len = adreno_gpu->fw[ADRENO_FW_PM4]->size / 4;
> +	DBG("loading PM4 ucode version: %x", ptr[1]);
> +
> +	gpu_write(gpu, REG_AXXX_CP_DEBUG, 0x02000000);
> +	gpu_write(gpu, REG_AXXX_CP_ME_RAM_WADDR, 0);
> +	for (i = 1; i < len; i++)
> +		gpu_write(gpu, REG_AXXX_CP_ME_RAM_DATA, ptr[i]);
> +
> +	/* Load PFP: */
> +	ptr = (uint32_t *)(adreno_gpu->fw[ADRENO_FW_PFP]->data);
> +	len = adreno_gpu->fw[ADRENO_FW_PFP]->size / 4;
> +	DBG("loading PFP ucode version: %x", ptr[5]);
> +
> +	gpu_write(gpu, REG_A2XX_CP_PFP_UCODE_ADDR, 0);
> +	for (i = 1; i < len; i++)
> +		gpu_write(gpu, REG_A2XX_CP_PFP_UCODE_DATA, ptr[i]);
> +
> +	gpu_write(gpu, REG_AXXX_CP_QUEUE_THRESHOLDS, 0x000C0804);
> +
> +	/* clear ME_HALT to start micro engine */
> +	gpu_write(gpu, REG_AXXX_CP_ME_CNTL, 0);
> +
> +	return a2xx_me_init(gpu) ? 0 : -EINVAL;
> +}
> +
> +static void a2xx_recover(struct msm_gpu *gpu)
> +{
> +	int i;
> +
> +	adreno_dump_info(gpu);
> +
> +	for (i = 0; i < 8; i++) {
> +		printk("CP_SCRATCH_REG%d: %u\n", i,
> +			gpu_read(gpu, REG_AXXX_CP_SCRATCH_REG0 + i));
> +	}
> +
> +	/* dump registers before resetting gpu, if enabled: */
> +	if (hang_debug)
> +		a2xx_dump(gpu);
> +
> +	gpu_write(gpu, REG_A2XX_RBBM_SOFT_RESET, 1);
> +	gpu_read(gpu, REG_A2XX_RBBM_SOFT_RESET);
> +	gpu_write(gpu, REG_A2XX_RBBM_SOFT_RESET, 0);
> +	adreno_recover(gpu);
> +}
> +
> +static void a2xx_destroy(struct msm_gpu *gpu)
> +{
> +	struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu);
> +	struct a2xx_gpu *a2xx_gpu = to_a2xx_gpu(adreno_gpu);
> +
> +	DBG("%s", gpu->name);
> +
> +	adreno_gpu_cleanup(adreno_gpu);
> +
> +	kfree(a2xx_gpu);
> +}
> +
> +static bool a2xx_idle(struct msm_gpu *gpu)
> +{
> +	/* wait for ringbuffer to drain: */
> +	if (!adreno_idle(gpu, gpu->rb[0]))
> +		return false;
> +
> +	/* then wait for GPU to finish: */
> +	if (spin_until(!(gpu_read(gpu, REG_A2XX_RBBM_STATUS) &
> +			A2XX_RBBM_STATUS_GUI_ACTIVE))) {
> +		DRM_ERROR("%s: timeout waiting for GPU to idle!\n", gpu->name);
> +
> +		/* TODO maybe we need to reset GPU here to recover from hang? */
> +		return false;
> +	}
> +
> +	return true;
> +}
> +
> +static irqreturn_t a2xx_irq(struct msm_gpu *gpu)
> +{
> +
> +	uint32_t mstatus, status;
> +
> +	mstatus = gpu_read(gpu, REG_A2XX_MASTER_INT_SIGNAL);
> +
> +	if (mstatus & A2XX_MASTER_INT_SIGNAL_MH_INT_STAT) {
> +		status = gpu_read(gpu, REG_A2XX_MH_INTERRUPT_STATUS);
> +
> +		dev_warn(gpu->dev->dev, "MH_INT: %08X\n", status);
> +		dev_warn(gpu->dev->dev, "MMU_PAGE_FAULT: %08X\n",
> +			gpu_read(gpu, REG_A2XX_MH_MMU_PAGE_FAULT));
> +		gpu_write(gpu, REG_A2XX_MH_INTERRUPT_CLEAR, status);
> +	}
> +
> +	if (mstatus & A2XX_MASTER_INT_SIGNAL_CP_INT_STAT) {
> +		status = gpu_read(gpu, REG_AXXX_CP_INT_STATUS);
> +
> +		/* only RB_INT is expected */
> +		if (status & ~AXXX_CP_INT_CNTL_RB_INT_MASK)
> +			dev_warn(gpu->dev->dev, "CP_INT: %08X\n", status);
> +
> +		gpu_write(gpu, REG_AXXX_CP_INT_ACK, status);
> +	}
> +
> +	if (mstatus & A2XX_MASTER_INT_SIGNAL_RBBM_INT_STAT) {
> +		status = gpu_read(gpu, REG_A2XX_RBBM_INT_STATUS);
> +
> +		dev_warn(gpu->dev->dev, "RBBM_INT: %08X\n", status);
> +
> +		gpu_write(gpu, REG_A2XX_RBBM_INT_ACK, status);
> +	}
> +
> +	msm_gpu_retire(gpu);
> +
> +	return IRQ_HANDLED;
> +}
> +
> +static const unsigned int a200_registers[] = {
> +	0x0000, 0x0002, 0x0004, 0x000B, 0x003B, 0x003D, 0x0040, 0x0044,
> +	0x0046, 0x0047, 0x01C0, 0x01C1, 0x01C3, 0x01C8, 0x01D5, 0x01D9,
> +	0x01DC, 0x01DD, 0x01EA, 0x01EA, 0x01EE, 0x01F3, 0x01F6, 0x01F7,
> +	0x01FC, 0x01FF, 0x0391, 0x0392, 0x039B, 0x039E, 0x03B2, 0x03B5,
> +	0x03B7, 0x03B7, 0x03F8, 0x03FB, 0x0440, 0x0440, 0x0443, 0x0444,
> +	0x044B, 0x044B, 0x044D, 0x044F, 0x0452, 0x0452, 0x0454, 0x045B,
> +	0x047F, 0x047F, 0x0578, 0x0587, 0x05C9, 0x05C9, 0x05D0, 0x05D0,
> +	0x0601, 0x0604, 0x0606, 0x0609, 0x060B, 0x060E, 0x0613, 0x0614,
> +	0x0A29, 0x0A2B, 0x0A2F, 0x0A31, 0x0A40, 0x0A43, 0x0A45, 0x0A45,
> +	0x0A4E, 0x0A4F, 0x0C2C, 0x0C2C, 0x0C30, 0x0C30, 0x0C38, 0x0C3C,
> +	0x0C40, 0x0C40, 0x0C44, 0x0C44, 0x0C80, 0x0C86, 0x0C88, 0x0C94,
> +	0x0C99, 0x0C9A, 0x0CA4, 0x0CA5, 0x0D00, 0x0D03, 0x0D06, 0x0D06,
> +	0x0D08, 0x0D0B, 0x0D34, 0x0D35, 0x0DAE, 0x0DC1, 0x0DC8, 0x0DD4,
> +	0x0DD8, 0x0DD9, 0x0E00, 0x0E00, 0x0E02, 0x0E04, 0x0E17, 0x0E1E,
> +	0x0EC0, 0x0EC9, 0x0ECB, 0x0ECC, 0x0ED0, 0x0ED0, 0x0ED4, 0x0ED7,
> +	0x0EE0, 0x0EE2, 0x0F01, 0x0F02, 0x0F0C, 0x0F0C, 0x0F0E, 0x0F12,
> +	0x0F26, 0x0F2A, 0x0F2C, 0x0F2C, 0x2000, 0x2002, 0x2006, 0x200F,
> +	0x2080, 0x2082, 0x2100, 0x2109, 0x210C, 0x2114, 0x2180, 0x2184,
> +	0x21F5, 0x21F7, 0x2200, 0x2208, 0x2280, 0x2283, 0x2293, 0x2294,
> +	0x2300, 0x2308, 0x2312, 0x2312, 0x2316, 0x231D, 0x2324, 0x2326,
> +	0x2380, 0x2383, 0x2400, 0x2402, 0x2406, 0x240F, 0x2480, 0x2482,
> +	0x2500, 0x2509, 0x250C, 0x2514, 0x2580, 0x2584, 0x25F5, 0x25F7,
> +	0x2600, 0x2608, 0x2680, 0x2683, 0x2693, 0x2694, 0x2700, 0x2708,
> +	0x2712, 0x2712, 0x2716, 0x271D, 0x2724, 0x2726, 0x2780, 0x2783,
> +	0x4000, 0x4003, 0x4800, 0x4805, 0x4900, 0x4900, 0x4908, 0x4908,
> +	~0   /* sentinel */
> +};
> +
> +static const unsigned int a220_registers[] = {
> +	0x0000, 0x0002, 0x0004, 0x000B, 0x003B, 0x003D, 0x0040, 0x0044,
> +	0x0046, 0x0047, 0x01C0, 0x01C1, 0x01C3, 0x01C8, 0x01D5, 0x01D9,
> +	0x01DC, 0x01DD, 0x01EA, 0x01EA, 0x01EE, 0x01F3, 0x01F6, 0x01F7,
> +	0x01FC, 0x01FF, 0x0391, 0x0392, 0x039B, 0x039E, 0x03B2, 0x03B5,
> +	0x03B7, 0x03B7, 0x03F8, 0x03FB, 0x0440, 0x0440, 0x0443, 0x0444,
> +	0x044B, 0x044B, 0x044D, 0x044F, 0x0452, 0x0452, 0x0454, 0x045B,
> +	0x047F, 0x047F, 0x0578, 0x0587, 0x05C9, 0x05C9, 0x05D0, 0x05D0,
> +	0x0601, 0x0604, 0x0606, 0x0609, 0x060B, 0x060E, 0x0613, 0x0614,
> +	0x0A29, 0x0A2B, 0x0A2F, 0x0A31, 0x0A40, 0x0A40, 0x0A42, 0x0A43,
> +	0x0A45, 0x0A45, 0x0A4E, 0x0A4F, 0x0C30, 0x0C30, 0x0C38, 0x0C39,
> +	0x0C3C, 0x0C3C, 0x0C80, 0x0C81, 0x0C88, 0x0C93, 0x0D00, 0x0D03,
> +	0x0D05, 0x0D06, 0x0D08, 0x0D0B, 0x0D34, 0x0D35, 0x0DAE, 0x0DC1,
> +	0x0DC8, 0x0DD4, 0x0DD8, 0x0DD9, 0x0E00, 0x0E00, 0x0E02, 0x0E04,
> +	0x0E17, 0x0E1E, 0x0EC0, 0x0EC9, 0x0ECB, 0x0ECC, 0x0ED0, 0x0ED0,
> +	0x0ED4, 0x0ED7, 0x0EE0, 0x0EE2, 0x0F01, 0x0F02, 0x2000, 0x2002,
> +	0x2006, 0x200F, 0x2080, 0x2082, 0x2100, 0x2102, 0x2104, 0x2109,
> +	0x210C, 0x2114, 0x2180, 0x2184, 0x21F5, 0x21F7, 0x2200, 0x2202,
> +	0x2204, 0x2204, 0x2208, 0x2208, 0x2280, 0x2282, 0x2294, 0x2294,
> +	0x2300, 0x2308, 0x2309, 0x230A, 0x2312, 0x2312, 0x2316, 0x2316,
> +	0x2318, 0x231D, 0x2324, 0x2326, 0x2380, 0x2383, 0x2400, 0x2402,
> +	0x2406, 0x240F, 0x2480, 0x2482, 0x2500, 0x2502, 0x2504, 0x2509,
> +	0x250C, 0x2514, 0x2580, 0x2584, 0x25F5, 0x25F7, 0x2600, 0x2602,
> +	0x2604, 0x2606, 0x2608, 0x2608, 0x2680, 0x2682, 0x2694, 0x2694,
> +	0x2700, 0x2708, 0x2712, 0x2712, 0x2716, 0x2716, 0x2718, 0x271D,
> +	0x2724, 0x2726, 0x2780, 0x2783, 0x4000, 0x4003, 0x4800, 0x4805,
> +	0x4900, 0x4900, 0x4908, 0x4908,
> +	~0   /* sentinel */
> +};
> +
> +static const unsigned int a225_registers[] = {
> +	0x0000, 0x0002, 0x0004, 0x000B, 0x003B, 0x003D, 0x0040, 0x0044,
> +	0x0046, 0x0047, 0x013C, 0x013C, 0x0140, 0x014F, 0x01C0, 0x01C1,
> +	0x01C3, 0x01C8, 0x01D5, 0x01D9, 0x01DC, 0x01DD, 0x01EA, 0x01EA,
> +	0x01EE, 0x01F3, 0x01F6, 0x01F7, 0x01FC, 0x01FF, 0x0391, 0x0392,
> +	0x039B, 0x039E, 0x03B2, 0x03B5, 0x03B7, 0x03B7, 0x03F8, 0x03FB,
> +	0x0440, 0x0440, 0x0443, 0x0444, 0x044B, 0x044B, 0x044D, 0x044F,
> +	0x0452, 0x0452, 0x0454, 0x045B, 0x047F, 0x047F, 0x0578, 0x0587,
> +	0x05C9, 0x05C9, 0x05D0, 0x05D0, 0x0601, 0x0604, 0x0606, 0x0609,
> +	0x060B, 0x060E, 0x0613, 0x0614, 0x0A29, 0x0A2B, 0x0A2F, 0x0A31,
> +	0x0A40, 0x0A40, 0x0A42, 0x0A43, 0x0A45, 0x0A45, 0x0A4E, 0x0A4F,
> +	0x0C01, 0x0C1D, 0x0C30, 0x0C30, 0x0C38, 0x0C39, 0x0C3C, 0x0C3C,
> +	0x0C80, 0x0C81, 0x0C88, 0x0C93, 0x0D00, 0x0D03, 0x0D05, 0x0D06,
> +	0x0D08, 0x0D0B, 0x0D34, 0x0D35, 0x0DAE, 0x0DC1, 0x0DC8, 0x0DD4,
> +	0x0DD8, 0x0DD9, 0x0E00, 0x0E00, 0x0E02, 0x0E04, 0x0E17, 0x0E1E,
> +	0x0EC0, 0x0EC9, 0x0ECB, 0x0ECC, 0x0ED0, 0x0ED0, 0x0ED4, 0x0ED7,
> +	0x0EE0, 0x0EE2, 0x0F01, 0x0F02, 0x2000, 0x200F, 0x2080, 0x2082,
> +	0x2100, 0x2109, 0x210C, 0x2114, 0x2180, 0x2184, 0x21F5, 0x21F7,
> +	0x2200, 0x2202, 0x2204, 0x2206, 0x2208, 0x2210, 0x2220, 0x2222,
> +	0x2280, 0x2282, 0x2294, 0x2294, 0x2297, 0x2297, 0x2300, 0x230A,
> +	0x2312, 0x2312, 0x2315, 0x2316, 0x2318, 0x231D, 0x2324, 0x2326,
> +	0x2340, 0x2357, 0x2360, 0x2360, 0x2380, 0x2383, 0x2400, 0x240F,
> +	0x2480, 0x2482, 0x2500, 0x2509, 0x250C, 0x2514, 0x2580, 0x2584,
> +	0x25F5, 0x25F7, 0x2600, 0x2602, 0x2604, 0x2606, 0x2608, 0x2610,
> +	0x2620, 0x2622, 0x2680, 0x2682, 0x2694, 0x2694, 0x2697, 0x2697,
> +	0x2700, 0x270A, 0x2712, 0x2712, 0x2715, 0x2716, 0x2718, 0x271D,
> +	0x2724, 0x2726, 0x2740, 0x2757, 0x2760, 0x2760, 0x2780, 0x2783,
> +	0x4000, 0x4003, 0x4800, 0x4806, 0x4808, 0x4808, 0x4900, 0x4900,
> +	0x4908, 0x4908,
> +	~0   /* sentinel */
> +};
> +
> +/* would be nice to not have to duplicate the _show() stuff with printk(): */
> +static void a2xx_dump(struct msm_gpu *gpu)
> +{
> +	printk("status:   %08x\n",
> +			gpu_read(gpu, REG_A2XX_RBBM_STATUS));
> +	adreno_dump(gpu);
> +}
> +
> +static struct msm_gpu_state *a2xx_gpu_state_get(struct msm_gpu *gpu)
> +{
> +	struct msm_gpu_state *state = kzalloc(sizeof(*state), GFP_KERNEL);
> +
> +	if (!state)
> +		return ERR_PTR(-ENOMEM);
> +
> +	adreno_gpu_state_get(gpu, state);
> +
> +	state->rbbm_status = gpu_read(gpu, REG_A2XX_RBBM_STATUS);
> +
> +	return state;
> +}
> +
> +/* Register offset defines for A2XX - copy of A3XX */
> +static const unsigned int a2xx_register_offsets[REG_ADRENO_REGISTER_MAX] = {
> +	REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_BASE, REG_AXXX_CP_RB_BASE),
> +	REG_ADRENO_SKIP(REG_ADRENO_CP_RB_BASE_HI),
> +	REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_RPTR_ADDR, REG_AXXX_CP_RB_RPTR_ADDR),
> +	REG_ADRENO_SKIP(REG_ADRENO_CP_RB_RPTR_ADDR_HI),
> +	REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_RPTR, REG_AXXX_CP_RB_RPTR),
> +	REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_WPTR, REG_AXXX_CP_RB_WPTR),
> +	REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_CNTL, REG_AXXX_CP_RB_CNTL),
> +};
> +
> +static const struct adreno_gpu_funcs funcs = {
> +	.base = {
> +		.get_param = adreno_get_param,
> +		.hw_init = a2xx_hw_init,
> +		.pm_suspend = msm_gpu_pm_suspend,
> +		.pm_resume = msm_gpu_pm_resume,
> +		.recover = a2xx_recover,
> +		.submit = adreno_submit,
> +		.flush = adreno_flush,
> +		.active_ring = adreno_active_ring,
> +		.irq = a2xx_irq,
> +		.destroy = a2xx_destroy,
> +#if defined(CONFIG_DEBUG_FS) || defined(CONFIG_DEV_COREDUMP)
> +		.show = adreno_show,
> +#endif
> +		.gpu_state_get = a2xx_gpu_state_get,
> +		.gpu_state_put = adreno_gpu_state_put,
> +	},
> +};
> +
> +static const struct msm_gpu_perfcntr perfcntrs[] = {
> +/* TODO */
> +};
> +
> +struct msm_gpu *a2xx_gpu_init(struct drm_device *dev)
> +{
> +	struct a2xx_gpu *a2xx_gpu = NULL;
> +	struct adreno_gpu *adreno_gpu;
> +	struct msm_gpu *gpu;
> +	struct msm_drm_private *priv = dev->dev_private;
> +	struct platform_device *pdev = priv->gpu_pdev;
> +	int ret;
> +
> +	if (!pdev) {
> +		dev_err(dev->dev, "no a2xx device\n");
> +		ret = -ENXIO;
> +		goto fail;
> +	}
> +
> +	a2xx_gpu = kzalloc(sizeof(*a2xx_gpu), GFP_KERNEL);
> +	if (!a2xx_gpu) {
> +		ret = -ENOMEM;
> +		goto fail;
> +	}
> +
> +	adreno_gpu = &a2xx_gpu->base;
> +	gpu = &adreno_gpu->base;
> +
> +	gpu->perfcntrs = perfcntrs;
> +	gpu->num_perfcntrs = ARRAY_SIZE(perfcntrs);
> +
> +	if (adreno_is_a20x(adreno_gpu))
> +		adreno_gpu->registers = a200_registers;
> +	else if (adreno_is_a225(adreno_gpu))
> +		adreno_gpu->registers = a225_registers;
> +	else
> +		adreno_gpu->registers = a220_registers;
> +
> +	adreno_gpu->reg_offsets = a2xx_register_offsets;
> +
> +	ret = adreno_gpu_init(dev, pdev, adreno_gpu, &funcs, 1);
> +	if (ret)
> +		goto fail;
> +
> +	if (!gpu->aspace) {
> +		/* MMU is not implemented...  */
> +		dev_err(dev->dev, "No memory protection without MMU\n");
> +	}
> +
> +	return gpu;
> +
> +fail:
> +	if (a2xx_gpu)
> +		a2xx_destroy(&a2xx_gpu->base.base);
> +
> +	return ERR_PTR(ret);
> +}
> diff --git a/drivers/gpu/drm/msm/adreno/a2xx_gpu.h b/drivers/gpu/drm/msm/adreno/a2xx_gpu.h
> new file mode 100644
> index 000000000000..02fba2cb8932
> --- /dev/null
> +++ b/drivers/gpu/drm/msm/adreno/a2xx_gpu.h
> @@ -0,0 +1,21 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/* Copyright (c) 2018 The Linux Foundation. All rights reserved. */
> +
> +#ifndef __A2XX_GPU_H__
> +#define __A2XX_GPU_H__
> +
> +#include "adreno_gpu.h"
> +
> +/* arrg, somehow fb.h is getting pulled in: */
> +#undef ROP_COPY
> +#undef ROP_XOR
> +
> +#include "a2xx.xml.h"
> +
> +struct a2xx_gpu {
> +	struct adreno_gpu base;
> +	bool pm_enabled;
> +};
> +#define to_a2xx_gpu(x) container_of(x, struct a2xx_gpu, base)
> +
> +#endif /* __A2XX_GPU_H__ */
> diff --git a/drivers/gpu/drm/msm/adreno/adreno_device.c b/drivers/gpu/drm/msm/adreno/adreno_device.c
> index 86abdb2b3a9c..6e3f78d4da5d 100644
> --- a/drivers/gpu/drm/msm/adreno/adreno_device.c
> +++ b/drivers/gpu/drm/msm/adreno/adreno_device.c
> @@ -27,6 +27,39 @@ module_param_named(hang_debug, hang_debug, bool, 0600);
>  
>  static const struct adreno_info gpulist[] = {
>  	{
> +		.rev   = ADRENO_REV(2, 0, 0, 0),
> +		.revn  = 200,
> +		.name  = "A200",
> +		.fw = {
> +			[ADRENO_FW_PM4] = "yamato_pm4.fw",
> +			[ADRENO_FW_PFP] = "yamato_pfp.fw",
> +		},
> +		.gmem  = SZ_256K,
> +		.inactive_period = DRM_MSM_INACTIVE_PERIOD,
> +		.init  = a2xx_gpu_init,
> +	}, { /* a200 on i.mx51 has only 128kib gmem */
> +		.rev   = ADRENO_REV(2, 0, 0, 1),
> +		.revn  = 201,
> +		.name  = "A200",
> +		.fw = {
> +			[ADRENO_FW_PM4] = "yamato_pm4.fw",
> +			[ADRENO_FW_PFP] = "yamato_pfp.fw",
> +		},
> +		.gmem  = SZ_128K,
> +		.inactive_period = DRM_MSM_INACTIVE_PERIOD,
> +		.init  = a2xx_gpu_init,
> +	}, {
> +		.rev   = ADRENO_REV(2, 2, 0, ANY_ID),
> +		.revn  = 220,
> +		.name  = "A220",
> +		.fw = {
> +			[ADRENO_FW_PM4] = "leia_pm4_470.fw",
> +			[ADRENO_FW_PFP] = "leia_pfp_470.fw",
> +		},
> +		.gmem  = SZ_512K,
> +		.inactive_period = DRM_MSM_INACTIVE_PERIOD,
> +		.init  = a2xx_gpu_init,
> +	}, {
>  		.rev   = ADRENO_REV(3, 0, 5, ANY_ID),
>  		.revn  = 305,
>  		.name  = "A305",
> diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
> index 93d70f4a2154..e09671324b05 100644
> --- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c
> +++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
> @@ -319,16 +319,27 @@ void adreno_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit,
>  		 */
>  		OUT_PKT3(ring, CP_EVENT_WRITE, 1);
>  		OUT_RING(ring, HLSQ_FLUSH);
> -
> -		OUT_PKT3(ring, CP_WAIT_FOR_IDLE, 1);
> -		OUT_RING(ring, 0x00000000);
>  	}
>  
> -	/* BIT(31) of CACHE_FLUSH_TS triggers CACHE_FLUSH_TS IRQ from GPU */
> -	OUT_PKT3(ring, CP_EVENT_WRITE, 3);
> -	OUT_RING(ring, CACHE_FLUSH_TS | BIT(31));
> -	OUT_RING(ring, rbmemptr(ring, fence));
> -	OUT_RING(ring, submit->seqno);
> +	/* wait for idle before cache flush/interrupt */
> +	OUT_PKT3(ring, CP_WAIT_FOR_IDLE, 1);
> +	OUT_RING(ring, 0x00000000);
> +
> +	if (!adreno_is_a2xx(adreno_gpu)) {
> +		/* BIT(31) of CACHE_FLUSH_TS triggers CACHE_FLUSH_TS IRQ from GPU */
> +		OUT_PKT3(ring, CP_EVENT_WRITE, 3);
> +		OUT_RING(ring, CACHE_FLUSH_TS | BIT(31));
> +		OUT_RING(ring, rbmemptr(ring, fence));
> +		OUT_RING(ring, submit->seqno);
> +	} else {
> +		/* BIT(31) means something else on a2xx */
> +		OUT_PKT3(ring, CP_EVENT_WRITE, 3);
> +		OUT_RING(ring, CACHE_FLUSH_TS);
> +		OUT_RING(ring, rbmemptr(ring, fence));
> +		OUT_RING(ring, submit->seqno);
> +		OUT_PKT3(ring, CP_INTERRUPT, 1);
> +		OUT_RING(ring, 0x80000000);
> +	}
>  
>  #if 0
>  	if (adreno_is_a3xx(adreno_gpu)) {
> diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.h b/drivers/gpu/drm/msm/adreno/adreno_gpu.h
> index de6e6ee42fba..4adc97a91e1d 100644
> --- a/drivers/gpu/drm/msm/adreno/adreno_gpu.h
> +++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.h
> @@ -154,6 +154,20 @@ struct adreno_platform_config {
>  	__ret;                                             \
>  })
>  
> +static inline bool adreno_is_a2xx(struct adreno_gpu *gpu)
> +{
> +	return (gpu->revn < 300);
> +}
> +
> +static inline bool adreno_is_a20x(struct adreno_gpu *gpu)
> +{
> +	return (gpu->revn < 210);
> +}
> +
> +static inline bool adreno_is_a225(struct adreno_gpu *gpu)
> +{
> +	return gpu->revn == 225;
> +}
>  
>  static inline bool adreno_is_a3xx(struct adreno_gpu *gpu)
>  {
> @@ -334,6 +348,7 @@ static inline void adreno_gpu_write(struct adreno_gpu *gpu,
>  		gpu_write(&gpu->base, reg - 1, data);
>  }
>  
> +struct msm_gpu *a2xx_gpu_init(struct drm_device *dev);
>  struct msm_gpu *a3xx_gpu_init(struct drm_device *dev);
>  struct msm_gpu *a4xx_gpu_init(struct drm_device *dev);
>  struct msm_gpu *a5xx_gpu_init(struct drm_device *dev);
> -- 
> 2.17.1
> 

-- 
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

Powered by blists - more mailing lists