diff options
Diffstat (limited to 'drivers/gpu/drm/msm/adreno/a5xx_gpu.c')
| -rw-r--r-- | drivers/gpu/drm/msm/adreno/a5xx_gpu.c | 839 |
1 files changed, 559 insertions, 280 deletions
diff --git a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c index d5f5e56422f5..56eaff2ee4e4 100644 --- a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c @@ -1,23 +1,11 @@ +// SPDX-License-Identifier: GPL-2.0-only /* Copyright (c) 2016-2017 The Linux Foundation. All rights reserved. - * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License version 2 and - * only version 2 as published by the Free Software Foundation. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * */ #include <linux/kernel.h> #include <linux/types.h> #include <linux/cpumask.h> -#include <linux/qcom_scm.h> -#include <linux/dma-mapping.h> -#include <linux/of_address.h> -#include <linux/soc/qcom/mdt_loader.h> +#include <linux/firmware/qcom/qcom_scm.h> #include <linux/pm_opp.h> #include <linux/nvmem-consumer.h> #include <linux/slab.h> @@ -30,102 +18,34 @@ static void a5xx_dump(struct msm_gpu *gpu); #define GPU_PAS_ID 13 -static int zap_shader_load_mdt(struct msm_gpu *gpu, const char *fwname) +static void update_shadow_rptr(struct msm_gpu *gpu, struct msm_ringbuffer *ring) { - struct device *dev = &gpu->pdev->dev; - const struct firmware *fw; - struct device_node *np; - struct resource r; - phys_addr_t mem_phys; - ssize_t mem_size; - void *mem_region = NULL; - int ret; - - if (!IS_ENABLED(CONFIG_ARCH_QCOM)) - return -EINVAL; - - np = of_get_child_by_name(dev->of_node, "zap-shader"); - if (!np) - return -ENODEV; - - np = of_parse_phandle(np, "memory-region", 0); - if (!np) - return -EINVAL; - - ret = of_address_to_resource(np, 0, &r); - if (ret) - return ret; - - mem_phys = r.start; - mem_size = resource_size(&r); - - /* Request the MDT file for the firmware */ - fw = adreno_request_fw(to_adreno_gpu(gpu), fwname); - if (IS_ERR(fw)) { - DRM_DEV_ERROR(dev, "Unable to load %s\n", fwname); - return PTR_ERR(fw); - } - - /* Figure out how much memory we need */ - mem_size = qcom_mdt_get_size(fw); - if (mem_size < 0) { - ret = mem_size; - goto out; - } - - /* Allocate memory for the firmware image */ - mem_region = memremap(mem_phys, mem_size, MEMREMAP_WC); - if (!mem_region) { - ret = -ENOMEM; - goto out; - } - - /* - * Load the rest of the MDT - * - * Note that we could be dealing with two different paths, since - * with upstream linux-firmware it would be in a qcom/ subdir.. - * adreno_request_fw() handles this, but qcom_mdt_load() does - * not. But since we've already gotten thru adreno_request_fw() - * we know which of the two cases it is: - */ - if (to_adreno_gpu(gpu)->fwloc == FW_LOCATION_LEGACY) { - ret = qcom_mdt_load(dev, fw, fwname, GPU_PAS_ID, - mem_region, mem_phys, mem_size, NULL); - } else { - char *newname; - - newname = kasprintf(GFP_KERNEL, "qcom/%s", fwname); + struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); + struct a5xx_gpu *a5xx_gpu = to_a5xx_gpu(adreno_gpu); - ret = qcom_mdt_load(dev, fw, newname, GPU_PAS_ID, - mem_region, mem_phys, mem_size, NULL); - kfree(newname); + if (a5xx_gpu->has_whereami) { + OUT_PKT7(ring, CP_WHERE_AM_I, 2); + OUT_RING(ring, lower_32_bits(shadowptr(a5xx_gpu, ring))); + OUT_RING(ring, upper_32_bits(shadowptr(a5xx_gpu, ring))); } - if (ret) - goto out; - - /* Send the image to the secure world */ - ret = qcom_scm_pas_auth_and_reset(GPU_PAS_ID); - if (ret) - DRM_DEV_ERROR(dev, "Unable to authorize the image\n"); - -out: - if (mem_region) - memunmap(mem_region); - - release_firmware(fw); - - return ret; } -static void a5xx_flush(struct msm_gpu *gpu, struct msm_ringbuffer *ring) +void a5xx_flush(struct msm_gpu *gpu, struct msm_ringbuffer *ring, + bool sync) { struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); struct a5xx_gpu *a5xx_gpu = to_a5xx_gpu(adreno_gpu); uint32_t wptr; unsigned long flags; - spin_lock_irqsave(&ring->lock, flags); + /* + * Most flush operations need to issue a WHERE_AM_I opcode to sync up + * the rptr shadow + */ + if (sync) + update_shadow_rptr(gpu, ring); + + spin_lock_irqsave(&ring->preempt_lock, flags); /* Copy the shadow to the actual register */ ring->cur = ring->next; @@ -133,7 +53,7 @@ static void a5xx_flush(struct msm_gpu *gpu, struct msm_ringbuffer *ring) /* Make sure to wrap wptr if we need to */ wptr = get_wptr(ring); - spin_unlock_irqrestore(&ring->lock, flags); + spin_unlock_irqrestore(&ring->preempt_lock, flags); /* Make sure everything is posted before making a decision */ mb(); @@ -143,12 +63,12 @@ static void a5xx_flush(struct msm_gpu *gpu, struct msm_ringbuffer *ring) gpu_write(gpu, REG_A5XX_CP_RB_WPTR, wptr); } -static void a5xx_submit_in_rb(struct msm_gpu *gpu, struct msm_gem_submit *submit, - struct msm_file_private *ctx) +static void a5xx_submit_in_rb(struct msm_gpu *gpu, struct msm_gem_submit *submit) { - struct msm_drm_private *priv = gpu->dev->dev_private; + struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); + struct a5xx_gpu *a5xx_gpu = to_a5xx_gpu(adreno_gpu); struct msm_ringbuffer *ring = submit->ring; - struct msm_gem_object *obj; + struct drm_gem_object *obj; uint32_t *ptr, dwords; unsigned int i; @@ -157,20 +77,21 @@ static void a5xx_submit_in_rb(struct msm_gpu *gpu, struct msm_gem_submit *submit case MSM_SUBMIT_CMD_IB_TARGET_BUF: break; case MSM_SUBMIT_CMD_CTX_RESTORE_BUF: - if (priv->lastctx == ctx) + if (ring->cur_ctx_seqno == submit->queue->ctx->seqno) break; + fallthrough; case MSM_SUBMIT_CMD_BUF: /* copy commands into RB: */ obj = submit->bos[submit->cmd[i].idx].obj; dwords = submit->cmd[i].size; - ptr = msm_gem_get_vaddr(&obj->base); + ptr = msm_gem_get_vaddr(obj); /* _get_vaddr() shouldn't fail at this point, * since we've already mapped it once in * submit_reloc() */ - if (WARN_ON(!ptr)) + if (WARN_ON(IS_ERR_OR_NULL(ptr))) return; for (i = 0; i < dwords; i++) { @@ -184,13 +105,14 @@ static void a5xx_submit_in_rb(struct msm_gpu *gpu, struct msm_gem_submit *submit OUT_RING(ring, ptr[i]); } - msm_gem_put_vaddr(&obj->base); + msm_gem_put_vaddr(obj); break; } } - a5xx_flush(gpu, ring); + a5xx_gpu->last_seqno[ring->id] = submit->seqno; + a5xx_flush(gpu, ring, true); a5xx_preempt_trigger(gpu); /* we might not necessarily have a cmd from userspace to @@ -202,18 +124,18 @@ static void a5xx_submit_in_rb(struct msm_gpu *gpu, struct msm_gem_submit *submit msm_gpu_retire(gpu); } -static void a5xx_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit, - struct msm_file_private *ctx) +static void a5xx_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit) { struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); struct a5xx_gpu *a5xx_gpu = to_a5xx_gpu(adreno_gpu); - struct msm_drm_private *priv = gpu->dev->dev_private; struct msm_ringbuffer *ring = submit->ring; unsigned int i, ibs = 0; + adreno_check_and_reenable_stall(adreno_gpu); + if (IS_ENABLED(CONFIG_DRM_MSM_GPU_SUDO) && submit->in_rb) { - priv->lastctx = NULL; - a5xx_submit_in_rb(gpu, submit, ctx); + ring->cur_ctx_seqno = 0; + a5xx_submit_in_rb(gpu, submit); return; } @@ -233,9 +155,13 @@ static void a5xx_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit, OUT_PKT7(ring, CP_SET_PROTECTED_MODE, 1); OUT_RING(ring, 1); - /* Enable local preemption for finegrain preemption */ - OUT_PKT7(ring, CP_PREEMPT_ENABLE_GLOBAL, 1); - OUT_RING(ring, 0x02); + /* + * Disable local preemption by default because it requires + * user-space to be aware of it and provide additional handling + * to restore rendering state or do various flushes on switch. + */ + OUT_PKT7(ring, CP_PREEMPT_ENABLE_LOCAL, 1); + OUT_RING(ring, 0x0); /* Allow CP_CONTEXT_SWITCH_YIELD packets in the IB2 */ OUT_PKT7(ring, CP_YIELD_ENABLE, 1); @@ -247,8 +173,9 @@ static void a5xx_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit, case MSM_SUBMIT_CMD_IB_TARGET_BUF: break; case MSM_SUBMIT_CMD_CTX_RESTORE_BUF: - if (priv->lastctx == ctx) + if (ring->cur_ctx_seqno == submit->queue->ctx->seqno) break; + fallthrough; case MSM_SUBMIT_CMD_BUF: OUT_PKT7(ring, CP_INDIRECT_BUFFER_PFE, 3); OUT_RING(ring, lower_32_bits(submit->cmd[i].iova)); @@ -257,6 +184,16 @@ static void a5xx_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit, ibs++; break; } + + /* + * Periodically update shadow-wptr if needed, so that we + * can see partial progress of submits with large # of + * cmds.. otherwise we could needlessly stall waiting for + * ringbuffer state, simply due to looking at a shadow + * rptr value that has not been updated + */ + if ((ibs % 32) == 0) + update_shadow_rptr(gpu, ring); } /* @@ -278,13 +215,15 @@ static void a5xx_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit, /* Write the fence to the scratch register */ OUT_PKT4(ring, REG_A5XX_CP_SCRATCH_REG(2), 1); OUT_RING(ring, submit->seqno); + a5xx_gpu->last_seqno[ring->id] = submit->seqno; /* * Execute a CACHE_FLUSH_TS event. This will ensure that the * timestamp is written to the memory and then triggers the interrupt */ OUT_PKT7(ring, CP_EVENT_WRITE, 4); - OUT_RING(ring, CACHE_FLUSH_TS | (1 << 31)); + OUT_RING(ring, CP_EVENT_WRITE_0_EVENT(CACHE_FLUSH_TS) | + CP_EVENT_WRITE_0_IRQ); OUT_RING(ring, lower_32_bits(rbmemptr(ring, fence))); OUT_RING(ring, upper_32_bits(rbmemptr(ring, fence))); OUT_RING(ring, submit->seqno); @@ -303,13 +242,14 @@ static void a5xx_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit, /* Set bit 0 to trigger an interrupt on preempt complete */ OUT_RING(ring, 0x01); - a5xx_flush(gpu, ring); + /* A WHERE_AM_I packet is not needed after a YIELD */ + a5xx_flush(gpu, ring, false); /* Check to see if we need to start preemption */ a5xx_preempt_trigger(gpu); } -static const struct { +static const struct adreno_five_hwcg_regs { u32 offset; u32 value; } a5xx_hwcg[] = { @@ -405,15 +345,130 @@ static const struct { {REG_A5XX_RBBM_CLOCK_DELAY_TSE_RAS_RBBM, 0x00004000}, {REG_A5XX_RBBM_CLOCK_DELAY_GPC, 0x00000200}, {REG_A5XX_RBBM_CLOCK_DELAY_VFD, 0x00002222} +}, a50x_hwcg[] = { + {REG_A5XX_RBBM_CLOCK_CNTL_SP0, 0x02222222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_SP0, 0x02222220}, + {REG_A5XX_RBBM_CLOCK_HYST_SP0, 0x0000F3CF}, + {REG_A5XX_RBBM_CLOCK_DELAY_SP0, 0x00000080}, + {REG_A5XX_RBBM_CLOCK_CNTL_TP0, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_TP0, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL3_TP0, 0x00002222}, + {REG_A5XX_RBBM_CLOCK_HYST_TP0, 0x77777777}, + {REG_A5XX_RBBM_CLOCK_HYST2_TP0, 0x77777777}, + {REG_A5XX_RBBM_CLOCK_HYST3_TP0, 0x00007777}, + {REG_A5XX_RBBM_CLOCK_DELAY_TP0, 0x11111111}, + {REG_A5XX_RBBM_CLOCK_DELAY2_TP0, 0x11111111}, + {REG_A5XX_RBBM_CLOCK_DELAY3_TP0, 0x00001111}, + {REG_A5XX_RBBM_CLOCK_CNTL2_UCHE, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL3_UCHE, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL4_UCHE, 0x00222222}, + {REG_A5XX_RBBM_CLOCK_CNTL_UCHE, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_HYST_UCHE, 0x00FFFFF4}, + {REG_A5XX_RBBM_CLOCK_DELAY_UCHE, 0x00000002}, + {REG_A5XX_RBBM_CLOCK_CNTL_RB0, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_RB0, 0x00222222}, + {REG_A5XX_RBBM_CLOCK_CNTL_CCU0, 0x00022220}, + {REG_A5XX_RBBM_CLOCK_CNTL_RAC, 0x05522222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_RAC, 0x00505555}, + {REG_A5XX_RBBM_CLOCK_HYST_RB_CCU0, 0x04040404}, + {REG_A5XX_RBBM_CLOCK_HYST_RAC, 0x07444044}, + {REG_A5XX_RBBM_CLOCK_DELAY_RB_CCU_L1_0, 0x00000002}, + {REG_A5XX_RBBM_CLOCK_DELAY_RAC, 0x00010011}, + {REG_A5XX_RBBM_CLOCK_CNTL_TSE_RAS_RBBM, 0x04222222}, + {REG_A5XX_RBBM_CLOCK_MODE_GPC, 0x02222222}, + {REG_A5XX_RBBM_CLOCK_MODE_VFD, 0x00002222}, + {REG_A5XX_RBBM_CLOCK_HYST_TSE_RAS_RBBM, 0x00000000}, + {REG_A5XX_RBBM_CLOCK_HYST_GPC, 0x04104004}, + {REG_A5XX_RBBM_CLOCK_HYST_VFD, 0x00000000}, + {REG_A5XX_RBBM_CLOCK_DELAY_HLSQ, 0x00000000}, + {REG_A5XX_RBBM_CLOCK_DELAY_TSE_RAS_RBBM, 0x00004000}, + {REG_A5XX_RBBM_CLOCK_DELAY_GPC, 0x00000200}, + {REG_A5XX_RBBM_CLOCK_DELAY_VFD, 0x00002222}, +}, a512_hwcg[] = { + {REG_A5XX_RBBM_CLOCK_CNTL_SP0, 0x02222222}, + {REG_A5XX_RBBM_CLOCK_CNTL_SP1, 0x02222222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_SP0, 0x02222220}, + {REG_A5XX_RBBM_CLOCK_CNTL2_SP1, 0x02222220}, + {REG_A5XX_RBBM_CLOCK_HYST_SP0, 0x0000F3CF}, + {REG_A5XX_RBBM_CLOCK_HYST_SP1, 0x0000F3CF}, + {REG_A5XX_RBBM_CLOCK_DELAY_SP0, 0x00000080}, + {REG_A5XX_RBBM_CLOCK_DELAY_SP1, 0x00000080}, + {REG_A5XX_RBBM_CLOCK_CNTL_TP0, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL_TP1, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_TP0, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_TP1, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL3_TP0, 0x00002222}, + {REG_A5XX_RBBM_CLOCK_CNTL3_TP1, 0x00002222}, + {REG_A5XX_RBBM_CLOCK_HYST_TP0, 0x77777777}, + {REG_A5XX_RBBM_CLOCK_HYST_TP1, 0x77777777}, + {REG_A5XX_RBBM_CLOCK_HYST2_TP0, 0x77777777}, + {REG_A5XX_RBBM_CLOCK_HYST2_TP1, 0x77777777}, + {REG_A5XX_RBBM_CLOCK_HYST3_TP0, 0x00007777}, + {REG_A5XX_RBBM_CLOCK_HYST3_TP1, 0x00007777}, + {REG_A5XX_RBBM_CLOCK_DELAY_TP0, 0x11111111}, + {REG_A5XX_RBBM_CLOCK_DELAY_TP1, 0x11111111}, + {REG_A5XX_RBBM_CLOCK_DELAY2_TP0, 0x11111111}, + {REG_A5XX_RBBM_CLOCK_DELAY2_TP1, 0x11111111}, + {REG_A5XX_RBBM_CLOCK_DELAY3_TP0, 0x00001111}, + {REG_A5XX_RBBM_CLOCK_DELAY3_TP1, 0x00001111}, + {REG_A5XX_RBBM_CLOCK_CNTL_UCHE, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_UCHE, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL3_UCHE, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL4_UCHE, 0x00222222}, + {REG_A5XX_RBBM_CLOCK_HYST_UCHE, 0x00444444}, + {REG_A5XX_RBBM_CLOCK_DELAY_UCHE, 0x00000002}, + {REG_A5XX_RBBM_CLOCK_CNTL_RB0, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL_RB1, 0x22222222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_RB0, 0x00222222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_RB1, 0x00222222}, + {REG_A5XX_RBBM_CLOCK_CNTL_CCU0, 0x00022220}, + {REG_A5XX_RBBM_CLOCK_CNTL_CCU1, 0x00022220}, + {REG_A5XX_RBBM_CLOCK_CNTL_RAC, 0x05522222}, + {REG_A5XX_RBBM_CLOCK_CNTL2_RAC, 0x00505555}, + {REG_A5XX_RBBM_CLOCK_HYST_RB_CCU0, 0x04040404}, + {REG_A5XX_RBBM_CLOCK_HYST_RB_CCU1, 0x04040404}, + {REG_A5XX_RBBM_CLOCK_HYST_RAC, 0x07444044}, + {REG_A5XX_RBBM_CLOCK_DELAY_RB_CCU_L1_0, 0x00000002}, + {REG_A5XX_RBBM_CLOCK_DELAY_RB_CCU_L1_1, 0x00000002}, + {REG_A5XX_RBBM_CLOCK_DELAY_RAC, 0x00010011}, + {REG_A5XX_RBBM_CLOCK_CNTL_TSE_RAS_RBBM, 0x04222222}, + {REG_A5XX_RBBM_CLOCK_MODE_GPC, 0x02222222}, + {REG_A5XX_RBBM_CLOCK_MODE_VFD, 0x00002222}, + {REG_A5XX_RBBM_CLOCK_HYST_TSE_RAS_RBBM, 0x00000000}, + {REG_A5XX_RBBM_CLOCK_HYST_GPC, 0x04104004}, + {REG_A5XX_RBBM_CLOCK_HYST_VFD, 0x00000000}, + {REG_A5XX_RBBM_CLOCK_DELAY_HLSQ, 0x00000000}, + {REG_A5XX_RBBM_CLOCK_DELAY_TSE_RAS_RBBM, 0x00004000}, + {REG_A5XX_RBBM_CLOCK_DELAY_GPC, 0x00000200}, + {REG_A5XX_RBBM_CLOCK_DELAY_VFD, 0x00002222}, }; void a5xx_set_hwcg(struct msm_gpu *gpu, bool state) { - unsigned int i; + struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); + const struct adreno_five_hwcg_regs *regs; + unsigned int i, sz; + + if (adreno_is_a505(adreno_gpu) || adreno_is_a506(adreno_gpu) || + adreno_is_a508(adreno_gpu)) { + regs = a50x_hwcg; + sz = ARRAY_SIZE(a50x_hwcg); + } else if (adreno_is_a509(adreno_gpu) || adreno_is_a512(adreno_gpu)) { + regs = a512_hwcg; + sz = ARRAY_SIZE(a512_hwcg); + } else { + regs = a5xx_hwcg; + sz = ARRAY_SIZE(a5xx_hwcg); + } - for (i = 0; i < ARRAY_SIZE(a5xx_hwcg); i++) - gpu_write(gpu, a5xx_hwcg[i].offset, - state ? a5xx_hwcg[i].value : 0); + for (i = 0; i < sz; i++) + gpu_write(gpu, regs[i].offset, + state ? regs[i].value : 0); + + if (adreno_is_a540(adreno_gpu)) { + gpu_write(gpu, REG_A5XX_RBBM_CLOCK_DELAY_GPMU, state ? 0x00000770 : 0); + gpu_write(gpu, REG_A5XX_RBBM_CLOCK_HYST_GPMU, state ? 0x00000004 : 0); + } gpu_write(gpu, REG_A5XX_RBBM_CLOCK_CNTL, state ? 0xAAA8AA00 : 0); gpu_write(gpu, REG_A5XX_RBBM_ISDB_CNT, state ? 0x182 : 0x180); @@ -439,12 +494,16 @@ static int a5xx_me_init(struct msm_gpu *gpu) OUT_RING(ring, 0x00000000); /* Specify workarounds for various microcode issues */ - if (adreno_is_a530(adreno_gpu)) { + if (adreno_is_a505(adreno_gpu) || adreno_is_a506(adreno_gpu) || + adreno_is_a530(adreno_gpu)) { /* Workaround for token end syncs * Force a WFI after every direct-render 3D mode draw and every * 2D mode 3 draw */ OUT_RING(ring, 0x0000000B); + } else if (adreno_is_a510(adreno_gpu)) { + /* Workaround for token and syncs */ + OUT_RING(ring, 0x00000001); } else { /* No workarounds enabled */ OUT_RING(ring, 0x00000000); @@ -453,7 +512,7 @@ static int a5xx_me_init(struct msm_gpu *gpu) OUT_RING(ring, 0x00000000); OUT_RING(ring, 0x00000000); - gpu->funcs->flush(gpu, ring); + a5xx_flush(gpu, ring, true); return a5xx_idle(gpu, ring) ? 0 : -EINVAL; } @@ -495,12 +554,32 @@ static int a5xx_preempt_start(struct msm_gpu *gpu) OUT_RING(ring, 0x01); OUT_RING(ring, 0x01); - gpu->funcs->flush(gpu, ring); + /* The WHERE_AMI_I packet is not needed after a YIELD is issued */ + a5xx_flush(gpu, ring, false); return a5xx_idle(gpu, ring) ? 0 : -EINVAL; } -static int a5xx_ucode_init(struct msm_gpu *gpu) +static void a5xx_ucode_check_version(struct a5xx_gpu *a5xx_gpu, + struct drm_gem_object *obj) +{ + u32 *buf = msm_gem_get_vaddr(obj); + + if (IS_ERR(buf)) + return; + + /* + * If the lowest nibble is 0xa that is an indication that this microcode + * has been patched. The actual version is in dword [3] but we only care + * about the patchlevel which is the lowest nibble of dword [3] + */ + if (((buf[0] & 0xf) == 0xa) && (buf[2] & 0xf) >= 1) + a5xx_gpu->has_whereami = true; + + msm_gem_put_vaddr(obj); +} + +static int a5xx_ucode_load(struct msm_gpu *gpu) { struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); struct a5xx_gpu *a5xx_gpu = to_a5xx_gpu(adreno_gpu); @@ -535,13 +614,27 @@ static int a5xx_ucode_init(struct msm_gpu *gpu) } msm_gem_object_set_name(a5xx_gpu->pfp_bo, "pfpfw"); + a5xx_ucode_check_version(a5xx_gpu, a5xx_gpu->pfp_bo); } - gpu_write64(gpu, REG_A5XX_CP_ME_INSTR_BASE_LO, - REG_A5XX_CP_ME_INSTR_BASE_HI, a5xx_gpu->pm4_iova); + if (a5xx_gpu->has_whereami) { + if (!a5xx_gpu->shadow_bo) { + a5xx_gpu->shadow = msm_gem_kernel_new(gpu->dev, + sizeof(u32) * gpu->nr_rings, + MSM_BO_WC | MSM_BO_MAP_PRIV, + gpu->vm, &a5xx_gpu->shadow_bo, + &a5xx_gpu->shadow_iova); - gpu_write64(gpu, REG_A5XX_CP_PFP_INSTR_BASE_LO, - REG_A5XX_CP_PFP_INSTR_BASE_HI, a5xx_gpu->pfp_iova); + if (IS_ERR(a5xx_gpu->shadow)) + return PTR_ERR(a5xx_gpu->shadow); + + msm_gem_object_set_name(a5xx_gpu->shadow_bo, "shadow"); + } + } else if (gpu->nr_rings > 1) { + /* Disable preemption if WHERE_AM_I isn't available */ + a5xx_preempt_fini(gpu); + gpu->nr_rings = 1; + } return 0; } @@ -550,8 +643,16 @@ static int a5xx_ucode_init(struct msm_gpu *gpu) static int a5xx_zap_shader_resume(struct msm_gpu *gpu) { + struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); int ret; + /* + * Adreno 506 have CPZ Retention feature and doesn't require + * to resume zap shader + */ + if (adreno_is_a506(adreno_gpu)) + return 0; + ret = qcom_scm_set_remote_state(SCM_GPU_ZAP_SHADER_RESUME, GPU_PAS_ID); if (ret) DRM_ERROR("%s: zap-shader resume failed: %d\n", @@ -563,8 +664,6 @@ static int a5xx_zap_shader_resume(struct msm_gpu *gpu) static int a5xx_zap_shader_init(struct msm_gpu *gpu) { static bool loaded; - struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); - struct platform_device *pdev = gpu->pdev; int ret; /* @@ -574,23 +673,9 @@ static int a5xx_zap_shader_init(struct msm_gpu *gpu) if (loaded) return a5xx_zap_shader_resume(gpu); - /* We need SCM to be able to load the firmware */ - if (!qcom_scm_is_available()) { - DRM_DEV_ERROR(&pdev->dev, "SCM is not available\n"); - return -EPROBE_DEFER; - } - - /* Each GPU has a target specific zap shader firmware name to use */ - if (!adreno_gpu->info->zapfw) { - DRM_DEV_ERROR(&pdev->dev, - "Zap shader firmware file not specified for this target\n"); - return -ENODEV; - } - - ret = zap_shader_load_mdt(gpu, adreno_gpu->info->zapfw); + ret = adreno_zap_shader_load(gpu, GPU_PAS_ID); loaded = !ret; - return ret; } @@ -610,10 +695,16 @@ static int a5xx_zap_shader_init(struct msm_gpu *gpu) static int a5xx_hw_init(struct msm_gpu *gpu) { struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); + struct a5xx_gpu *a5xx_gpu = to_a5xx_gpu(adreno_gpu); + u32 hbb; int ret; gpu_write(gpu, REG_A5XX_VBIF_ROUND_ROBIN_QOS_ARB, 0x00000003); + if (adreno_is_a509(adreno_gpu) || adreno_is_a512(adreno_gpu) || + adreno_is_a540(adreno_gpu)) + gpu_write(gpu, REG_A5XX_VBIF_GATE_OFF_WRREQ_EN, 0x00000009); + /* Make all blocks contribute to the GPU BUSY perf counter */ gpu_write(gpu, REG_A5XX_RBBM_PERFCTR_GPU_BUSY_MASKED, 0xFFFFFFFF); @@ -661,29 +752,64 @@ static int a5xx_hw_init(struct msm_gpu *gpu) gpu_write(gpu, REG_A5XX_UCHE_CACHE_WAYS, 0x02); /* Disable L2 bypass in the UCHE */ - gpu_write(gpu, REG_A5XX_UCHE_TRAP_BASE_LO, 0xFFFF0000); - gpu_write(gpu, REG_A5XX_UCHE_TRAP_BASE_HI, 0x0001FFFF); - gpu_write(gpu, REG_A5XX_UCHE_WRITE_THRU_BASE_LO, 0xFFFF0000); - gpu_write(gpu, REG_A5XX_UCHE_WRITE_THRU_BASE_HI, 0x0001FFFF); + gpu_write(gpu, REG_A5XX_UCHE_TRAP_BASE_LO, lower_32_bits(adreno_gpu->uche_trap_base)); + gpu_write(gpu, REG_A5XX_UCHE_TRAP_BASE_HI, upper_32_bits(adreno_gpu->uche_trap_base)); + gpu_write(gpu, REG_A5XX_UCHE_WRITE_THRU_BASE_LO, lower_32_bits(adreno_gpu->uche_trap_base)); + gpu_write(gpu, REG_A5XX_UCHE_WRITE_THRU_BASE_HI, upper_32_bits(adreno_gpu->uche_trap_base)); /* Set the GMEM VA range (0 to gpu->gmem) */ gpu_write(gpu, REG_A5XX_UCHE_GMEM_RANGE_MIN_LO, 0x00100000); gpu_write(gpu, REG_A5XX_UCHE_GMEM_RANGE_MIN_HI, 0x00000000); gpu_write(gpu, REG_A5XX_UCHE_GMEM_RANGE_MAX_LO, - 0x00100000 + adreno_gpu->gmem - 1); + 0x00100000 + adreno_gpu->info->gmem - 1); gpu_write(gpu, REG_A5XX_UCHE_GMEM_RANGE_MAX_HI, 0x00000000); - gpu_write(gpu, REG_A5XX_CP_MEQ_THRESHOLDS, 0x40); - gpu_write(gpu, REG_A5XX_CP_MERCIU_SIZE, 0x40); - gpu_write(gpu, REG_A5XX_CP_ROQ_THRESHOLDS_2, 0x80000060); - gpu_write(gpu, REG_A5XX_CP_ROQ_THRESHOLDS_1, 0x40201B16); + if (adreno_is_a505(adreno_gpu) || adreno_is_a506(adreno_gpu) || + adreno_is_a508(adreno_gpu) || adreno_is_a510(adreno_gpu)) { + gpu_write(gpu, REG_A5XX_CP_MEQ_THRESHOLDS, 0x20); + if (adreno_is_a505(adreno_gpu) || adreno_is_a506(adreno_gpu) || + adreno_is_a508(adreno_gpu)) + gpu_write(gpu, REG_A5XX_CP_MERCIU_SIZE, 0x400); + else + gpu_write(gpu, REG_A5XX_CP_MERCIU_SIZE, 0x20); + gpu_write(gpu, REG_A5XX_CP_ROQ_THRESHOLDS_2, 0x40000030); + gpu_write(gpu, REG_A5XX_CP_ROQ_THRESHOLDS_1, 0x20100D0A); + } else { + gpu_write(gpu, REG_A5XX_CP_MEQ_THRESHOLDS, 0x40); + if (adreno_is_a530(adreno_gpu)) + gpu_write(gpu, REG_A5XX_CP_MERCIU_SIZE, 0x40); + else + gpu_write(gpu, REG_A5XX_CP_MERCIU_SIZE, 0x400); + gpu_write(gpu, REG_A5XX_CP_ROQ_THRESHOLDS_2, 0x80000060); + gpu_write(gpu, REG_A5XX_CP_ROQ_THRESHOLDS_1, 0x40201B16); + } - gpu_write(gpu, REG_A5XX_PC_DBG_ECO_CNTL, (0x400 << 11 | 0x300 << 22)); + if (adreno_is_a505(adreno_gpu) || adreno_is_a506(adreno_gpu) || + adreno_is_a508(adreno_gpu)) + gpu_write(gpu, REG_A5XX_PC_DBG_ECO_CNTL, + (0x100 << 11 | 0x100 << 22)); + else if (adreno_is_a509(adreno_gpu) || adreno_is_a510(adreno_gpu) || + adreno_is_a512(adreno_gpu)) + gpu_write(gpu, REG_A5XX_PC_DBG_ECO_CNTL, + (0x200 << 11 | 0x200 << 22)); + else + gpu_write(gpu, REG_A5XX_PC_DBG_ECO_CNTL, + (0x400 << 11 | 0x300 << 22)); if (adreno_gpu->info->quirks & ADRENO_QUIRK_TWO_PASS_USE_WFI) gpu_rmw(gpu, REG_A5XX_PC_DBG_ECO_CNTL, 0, (1 << 8)); - gpu_write(gpu, REG_A5XX_PC_DBG_ECO_CNTL, 0xc0200100); + /* + * Disable the RB sampler datapath DP2 clock gating optimization + * for 1-SP GPUs, as it is enabled by default. + */ + if (adreno_is_a505(adreno_gpu) || adreno_is_a506(adreno_gpu) || + adreno_is_a508(adreno_gpu) || adreno_is_a509(adreno_gpu) || + adreno_is_a512(adreno_gpu)) + gpu_rmw(gpu, REG_A5XX_RB_DBG_ECO_CNTL, 0, (1 << 9)); + + /* Disable UCHE global filter as SP can invalidate/flush independently */ + gpu_write(gpu, REG_A5XX_UCHE_MODE_CNTL, BIT(29)); /* Enable USE_RETENTION_FLOPS */ gpu_write(gpu, REG_A5XX_CP_CHICKEN_DBG, 0x02000000); @@ -691,14 +817,36 @@ static int a5xx_hw_init(struct msm_gpu *gpu) /* Enable ME/PFP split notification */ gpu_write(gpu, REG_A5XX_RBBM_AHB_CNTL1, 0xA6FFFFFF); + /* + * In A5x, CCU can send context_done event of a particular context to + * UCHE which ultimately reaches CP even when there is valid + * transaction of that context inside CCU. This can let CP to program + * config registers, which will make the "valid transaction" inside + * CCU to be interpreted differently. This can cause gpu fault. This + * bug is fixed in latest A510 revision. To enable this bug fix - + * bit[11] of RB_DBG_ECO_CNTL need to be set to 0, default is 1 + * (disable). For older A510 version this bit is unused. + */ + if (adreno_is_a510(adreno_gpu)) + gpu_rmw(gpu, REG_A5XX_RB_DBG_ECO_CNTL, (1 << 11), 0); + /* Enable HWCG */ a5xx_set_hwcg(gpu, true); gpu_write(gpu, REG_A5XX_RBBM_AHB_CNTL2, 0x0000003F); - /* Set the highest bank bit */ - gpu_write(gpu, REG_A5XX_TPL1_MODE_CNTL, 2 << 7); - gpu_write(gpu, REG_A5XX_RB_MODE_CNTL, 2 << 1); + BUG_ON(adreno_gpu->ubwc_config->highest_bank_bit < 13); + hbb = adreno_gpu->ubwc_config->highest_bank_bit - 13; + + gpu_write(gpu, REG_A5XX_TPL1_MODE_CNTL, hbb << 7); + gpu_write(gpu, REG_A5XX_RB_MODE_CNTL, hbb << 1); + + if (adreno_is_a509(adreno_gpu) || adreno_is_a512(adreno_gpu) || + adreno_is_a540(adreno_gpu)) + gpu_write(gpu, REG_A5XX_UCHE_DBG_ECO_CNTL_2, hbb); + + /* Disable All flat shading optimization (ALLFLATOPTDIS) */ + gpu_rmw(gpu, REG_A5XX_VPC_DBG_ECO_CNTL, 0, (1 << 10)); /* Protect registers from the CP */ gpu_write(gpu, REG_A5XX_CP_PROTECT_CNTL, 0x00000007); @@ -730,13 +878,13 @@ static int a5xx_hw_init(struct msm_gpu *gpu) /* VPC */ gpu_write(gpu, REG_A5XX_CP_PROTECT(14), ADRENO_PROTECT_RW(0xE68, 8)); - gpu_write(gpu, REG_A5XX_CP_PROTECT(15), ADRENO_PROTECT_RW(0xE70, 4)); + gpu_write(gpu, REG_A5XX_CP_PROTECT(15), ADRENO_PROTECT_RW(0xE70, 16)); /* UCHE */ gpu_write(gpu, REG_A5XX_CP_PROTECT(16), ADRENO_PROTECT_RW(0xE80, 16)); - if (adreno_is_a530(adreno_gpu)) - gpu_write(gpu, REG_A5XX_CP_PROTECT(17), + /* SMMU */ + gpu_write(gpu, REG_A5XX_CP_PROTECT(17), ADRENO_PROTECT_RW(0x10000, 0x8000)); gpu_write(gpu, REG_A5XX_RBBM_SECVID_TSB_CNTL, 0); @@ -745,21 +893,62 @@ static int a5xx_hw_init(struct msm_gpu *gpu) * memory rendering at this point in time and we don't want to block off * part of the virtual memory space. */ - gpu_write64(gpu, REG_A5XX_RBBM_SECVID_TSB_TRUSTED_BASE_LO, - REG_A5XX_RBBM_SECVID_TSB_TRUSTED_BASE_HI, 0x00000000); + gpu_write64(gpu, REG_A5XX_RBBM_SECVID_TSB_TRUSTED_BASE_LO, 0x00000000); gpu_write(gpu, REG_A5XX_RBBM_SECVID_TSB_TRUSTED_SIZE, 0x00000000); + /* Put the GPU into 64 bit by default */ + gpu_write(gpu, REG_A5XX_CP_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_VSC_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_GRAS_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_RB_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_PC_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_HLSQ_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_VFD_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_VPC_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_UCHE_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_SP_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_TPL1_ADDR_MODE_CNTL, 0x1); + gpu_write(gpu, REG_A5XX_RBBM_SECVID_TSB_ADDR_MODE_CNTL, 0x1); + + /* + * VPC corner case with local memory load kill leads to corrupt + * internal state. Normal Disable does not work for all a5x chips. + * So do the following setting to disable it. + */ + if (adreno_gpu->info->quirks & ADRENO_QUIRK_LMLOADKILL_DISABLE) { + gpu_rmw(gpu, REG_A5XX_VPC_DBG_ECO_CNTL, 0, BIT(23)); + gpu_rmw(gpu, REG_A5XX_HLSQ_DBG_ECO_CNTL, BIT(18), 0); + } + ret = adreno_hw_init(gpu); if (ret) return ret; - a5xx_preempt_hw_init(gpu); + if (adreno_is_a530(adreno_gpu) || adreno_is_a540(adreno_gpu)) + a5xx_gpmu_ucode_init(gpu); - a5xx_gpmu_ucode_init(gpu); + gpu_write64(gpu, REG_A5XX_CP_ME_INSTR_BASE_LO, a5xx_gpu->pm4_iova); + gpu_write64(gpu, REG_A5XX_CP_PFP_INSTR_BASE_LO, a5xx_gpu->pfp_iova); - ret = a5xx_ucode_init(gpu); - if (ret) - return ret; + /* Set the ringbuffer address */ + gpu_write64(gpu, REG_A5XX_CP_RB_BASE, gpu->rb[0]->iova); + + /* + * If the microcode supports the WHERE_AM_I opcode then we can use that + * in lieu of the RPTR shadow and enable preemption. Otherwise, we + * can't safely use the RPTR shadow or preemption. In either case, the + * RPTR shadow should be disabled in hardware. + */ + gpu_write(gpu, REG_A5XX_CP_RB_CNTL, + MSM_GPU_RB_CNTL_DEFAULT | AXXX_CP_RB_CNTL_NO_UPDATE); + + /* Configure the RPTR shadow if needed: */ + if (a5xx_gpu->shadow_bo) { + gpu_write64(gpu, REG_A5XX_CP_RB_RPTR_ADDR, + shadowptr(a5xx_gpu, gpu->rb[0])); + } + + a5xx_preempt_hw_init(gpu); /* Disable the interrupts through the initial bringup stage */ gpu_write(gpu, REG_A5XX_RBBM_INT_0_MASK, A5XX_INT_MASK); @@ -780,15 +969,16 @@ static int a5xx_hw_init(struct msm_gpu *gpu) */ if (adreno_is_a530(adreno_gpu)) { OUT_PKT7(gpu->rb[0], CP_EVENT_WRITE, 1); - OUT_RING(gpu->rb[0], 0x0F); + OUT_RING(gpu->rb[0], CP_EVENT_WRITE_0_EVENT(STAT_EVENT)); - gpu->funcs->flush(gpu, gpu->rb[0]); + a5xx_flush(gpu, gpu->rb[0], true); if (!a5xx_idle(gpu, gpu->rb[0])) return -EINVAL; } /* - * Try to load a zap shader into the secure world. If successful + * If the chip that we are using does support loading one, then + * try to load a zap shader into the secure world. If successful * we can use the CP to switch out of secure mode. If not then we * have no resource but to try to switch ourselves out manually. If we * guessed wrong then access to the RBBM_SECVID_TRUST_CNTL register will @@ -799,14 +989,21 @@ static int a5xx_hw_init(struct msm_gpu *gpu) OUT_PKT7(gpu->rb[0], CP_SET_SECURE_MODE, 1); OUT_RING(gpu->rb[0], 0x00000000); - gpu->funcs->flush(gpu, gpu->rb[0]); + a5xx_flush(gpu, gpu->rb[0], true); if (!a5xx_idle(gpu, gpu->rb[0])) return -EINVAL; - } else { - /* Print a warning so if we die, we know why */ + } else if (ret == -ENODEV) { + /* + * This device does not use zap shader (but print a warning + * just in case someone got their dt wrong.. hopefully they + * have a debug UART to realize the error of their ways... + * if you mess this up you are about to crash horribly) + */ dev_warn_once(gpu->dev->dev, "Zap shader not enabled - using SECVID_TRUST_CNTL instead\n"); gpu_write(gpu, REG_A5XX_RBBM_SECVID_TRUST_CNTL, 0x0); + } else { + return ret; } /* Last step - yield the ringbuffer */ @@ -845,18 +1042,23 @@ static void a5xx_destroy(struct msm_gpu *gpu) a5xx_preempt_fini(gpu); if (a5xx_gpu->pm4_bo) { - msm_gem_unpin_iova(a5xx_gpu->pm4_bo, gpu->aspace); - drm_gem_object_put_unlocked(a5xx_gpu->pm4_bo); + msm_gem_unpin_iova(a5xx_gpu->pm4_bo, gpu->vm); + drm_gem_object_put(a5xx_gpu->pm4_bo); } if (a5xx_gpu->pfp_bo) { - msm_gem_unpin_iova(a5xx_gpu->pfp_bo, gpu->aspace); - drm_gem_object_put_unlocked(a5xx_gpu->pfp_bo); + msm_gem_unpin_iova(a5xx_gpu->pfp_bo, gpu->vm); + drm_gem_object_put(a5xx_gpu->pfp_bo); } if (a5xx_gpu->gpmu_bo) { - msm_gem_unpin_iova(a5xx_gpu->gpmu_bo, gpu->aspace); - drm_gem_object_put_unlocked(a5xx_gpu->gpmu_bo); + msm_gem_unpin_iova(a5xx_gpu->gpmu_bo, gpu->vm); + drm_gem_object_put(a5xx_gpu->gpmu_bo); + } + + if (a5xx_gpu->shadow_bo) { + msm_gem_unpin_iova(a5xx_gpu->shadow_bo, gpu->vm); + drm_gem_object_put(a5xx_gpu->shadow_bo); } adreno_gpu_cleanup(adreno_gpu); @@ -903,17 +1105,22 @@ bool a5xx_idle(struct msm_gpu *gpu, struct msm_ringbuffer *ring) return true; } -static int a5xx_fault_handler(void *arg, unsigned long iova, int flags) +static int a5xx_fault_handler(void *arg, unsigned long iova, int flags, void *data) { struct msm_gpu *gpu = arg; - pr_warn_ratelimited("*** gpu fault: iova=%08lx, flags=%d (%u,%u,%u,%u)\n", - iova, flags, + struct adreno_smmu_fault_info *info = data; + char block[12] = "unknown"; + u32 scratch[] = { gpu_read(gpu, REG_A5XX_CP_SCRATCH_REG(4)), gpu_read(gpu, REG_A5XX_CP_SCRATCH_REG(5)), gpu_read(gpu, REG_A5XX_CP_SCRATCH_REG(6)), - gpu_read(gpu, REG_A5XX_CP_SCRATCH_REG(7))); + gpu_read(gpu, REG_A5XX_CP_SCRATCH_REG(7)), + }; + + if (info) + snprintf(block, sizeof(block), "%x", info->fsynr1); - return -EFAULT; + return adreno_fault_handler(gpu, iova, flags, info, block, scratch); } static void a5xx_cp_err_irq(struct msm_gpu *gpu) @@ -1026,23 +1233,31 @@ static void a5xx_gpmu_err_irq(struct msm_gpu *gpu) static void a5xx_fault_detect_irq(struct msm_gpu *gpu) { struct drm_device *dev = gpu->dev; - struct msm_drm_private *priv = dev->dev_private; struct msm_ringbuffer *ring = gpu->funcs->active_ring(gpu); + /* + * If stalled on SMMU fault, we could trip the GPU's hang detection, + * but the fault handler will trigger the devcore dump, and we want + * to otherwise resume normally rather than killing the submit, so + * just bail. + */ + if (gpu_read(gpu, REG_A5XX_RBBM_STATUS3) & BIT(24)) + return; + DRM_DEV_ERROR(dev->dev, "gpu fault ring %d fence %x status %8.8X rb %4.4x/%4.4x ib1 %16.16llX/%4.4x ib2 %16.16llX/%4.4x\n", - ring ? ring->id : -1, ring ? ring->seqno : 0, + ring ? ring->id : -1, ring ? ring->fctx->last_fence : 0, gpu_read(gpu, REG_A5XX_RBBM_STATUS), gpu_read(gpu, REG_A5XX_CP_RB_RPTR), gpu_read(gpu, REG_A5XX_CP_RB_WPTR), - gpu_read64(gpu, REG_A5XX_CP_IB1_BASE, REG_A5XX_CP_IB1_BASE_HI), + gpu_read64(gpu, REG_A5XX_CP_IB1_BASE), gpu_read(gpu, REG_A5XX_CP_IB1_BUFSZ), - gpu_read64(gpu, REG_A5XX_CP_IB2_BASE, REG_A5XX_CP_IB2_BASE_HI), + gpu_read64(gpu, REG_A5XX_CP_IB2_BASE), gpu_read(gpu, REG_A5XX_CP_IB2_BUFSZ)); /* Turn off the hangcheck timer to keep it from bothering us */ - del_timer(&gpu->hangcheck_timer); + timer_delete(&gpu->hangcheck_timer); - queue_work(priv->wq, &gpu->recover_work); + kthread_queue_work(gpu->worker, &gpu->recover_work); } #define RBBM_ERROR_MASK \ @@ -1055,6 +1270,7 @@ static void a5xx_fault_detect_irq(struct msm_gpu *gpu) static irqreturn_t a5xx_irq(struct msm_gpu *gpu) { + struct msm_drm_private *priv = gpu->dev->dev_private; u32 status = gpu_read(gpu, REG_A5XX_RBBM_INT_0_STATUS); /* @@ -1064,6 +1280,11 @@ static irqreturn_t a5xx_irq(struct msm_gpu *gpu) gpu_write(gpu, REG_A5XX_RBBM_INT_CLEAR_CMD, status & ~A5XX_RBBM_INT_0_MASK_RBBM_AHB_ERROR); + if (priv->disable_err_irq) { + status &= A5XX_RBBM_INT_0_MASK_CP_CACHE_FLUSH_TS | + A5XX_RBBM_INT_0_MASK_CP_SW; + } + /* Pass status to a5xx_rbbm_err_irq because we've already cleared it */ if (status & RBBM_ERROR_MASK) a5xx_rbbm_err_irq(gpu, status); @@ -1091,17 +1312,6 @@ static irqreturn_t a5xx_irq(struct msm_gpu *gpu) return IRQ_HANDLED; } -static const u32 a5xx_register_offsets[REG_ADRENO_REGISTER_MAX] = { - REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_BASE, REG_A5XX_CP_RB_BASE), - REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_BASE_HI, REG_A5XX_CP_RB_BASE_HI), - REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_RPTR_ADDR, REG_A5XX_CP_RB_RPTR_ADDR), - REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_RPTR_ADDR_HI, - REG_A5XX_CP_RB_RPTR_ADDR_HI), - REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_RPTR, REG_A5XX_CP_RB_RPTR), - REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_WPTR, REG_A5XX_CP_RB_WPTR), - REG_ADRENO_DEFINE(REG_ADRENO_CP_RB_CNTL, REG_A5XX_CP_RB_CNTL), -}; - static const u32 a5xx_registers[] = { 0x0000, 0x0002, 0x0004, 0x0020, 0x0022, 0x0026, 0x0029, 0x002B, 0x002E, 0x0035, 0x0038, 0x0042, 0x0044, 0x0044, 0x0047, 0x0095, @@ -1142,6 +1352,7 @@ static void a5xx_dump(struct msm_gpu *gpu) static int a5xx_pm_resume(struct msm_gpu *gpu) { + struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); int ret; /* Turn on the core power */ @@ -1149,6 +1360,16 @@ static int a5xx_pm_resume(struct msm_gpu *gpu) if (ret) return ret; + /* Adreno 505, 506, 508, 509, 510, 512 needs manual RBBM sus/res control */ + if (!(adreno_is_a530(adreno_gpu) || adreno_is_a540(adreno_gpu))) { + /* Halt the sp_input_clk at HM level */ + gpu_write(gpu, REG_A5XX_RBBM_CLOCK_CNTL, 0x00000055); + a5xx_set_hwcg(gpu, true); + /* Turn on sp_input_clk at HM level */ + gpu_rmw(gpu, REG_A5XX_RBBM_CLOCK_CNTL, 0xff, 0); + return 0; + } + /* Turn the RBCCU domain first to limit the chances of voltage droop */ gpu_write(gpu, REG_A5XX_GPMU_RBCCU_POWER_CNTL, 0x778000); @@ -1177,26 +1398,46 @@ static int a5xx_pm_resume(struct msm_gpu *gpu) static int a5xx_pm_suspend(struct msm_gpu *gpu) { + struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); + struct a5xx_gpu *a5xx_gpu = to_a5xx_gpu(adreno_gpu); + u32 mask = 0xf; + int i, ret; + + /* A505, A506, A508, A510 have 3 XIN ports in VBIF */ + if (adreno_is_a505(adreno_gpu) || adreno_is_a506(adreno_gpu) || + adreno_is_a508(adreno_gpu) || adreno_is_a510(adreno_gpu)) + mask = 0x7; + /* Clear the VBIF pipe before shutting down */ - gpu_write(gpu, REG_A5XX_VBIF_XIN_HALT_CTRL0, 0xF); - spin_until((gpu_read(gpu, REG_A5XX_VBIF_XIN_HALT_CTRL1) & 0xF) == 0xF); + gpu_write(gpu, REG_A5XX_VBIF_XIN_HALT_CTRL0, mask); + spin_until((gpu_read(gpu, REG_A5XX_VBIF_XIN_HALT_CTRL1) & + mask) == mask); gpu_write(gpu, REG_A5XX_VBIF_XIN_HALT_CTRL0, 0); /* * Reset the VBIF before power collapse to avoid issue with FIFO - * entries + * entries on Adreno A510 and A530 (the others will tend to lock up) */ - gpu_write(gpu, REG_A5XX_RBBM_BLOCK_SW_RESET_CMD, 0x003C0000); - gpu_write(gpu, REG_A5XX_RBBM_BLOCK_SW_RESET_CMD, 0x00000000); + if (adreno_is_a510(adreno_gpu) || adreno_is_a530(adreno_gpu)) { + gpu_write(gpu, REG_A5XX_RBBM_BLOCK_SW_RESET_CMD, 0x003C0000); + gpu_write(gpu, REG_A5XX_RBBM_BLOCK_SW_RESET_CMD, 0x00000000); + } + + ret = msm_gpu_pm_suspend(gpu); + if (ret) + return ret; + + if (a5xx_gpu->has_whereami) + for (i = 0; i < gpu->nr_rings; i++) + a5xx_gpu->shadow[i] = 0; - return msm_gpu_pm_suspend(gpu); + return 0; } static int a5xx_get_timestamp(struct msm_gpu *gpu, uint64_t *value) { - *value = gpu_read64(gpu, REG_A5XX_RBBM_PERFCTR_CP_0_LO, - REG_A5XX_RBBM_PERFCTR_CP_0_HI); + *value = gpu_read64(gpu, REG_A5XX_RBBM_ALWAYSON_COUNTER_LO); return 0; } @@ -1215,8 +1456,8 @@ struct a5xx_gpu_state { static int a5xx_crashdumper_init(struct msm_gpu *gpu, struct a5xx_crashdumper *dumper) { - dumper->ptr = msm_gem_kernel_new_locked(gpu->dev, - SZ_1M, MSM_BO_UNCACHED, gpu->aspace, + dumper->ptr = msm_gem_kernel_new(gpu->dev, + SZ_1M, MSM_BO_WC, gpu->vm, &dumper->bo, &dumper->iova); if (!IS_ERR(dumper->ptr)) @@ -1233,8 +1474,7 @@ static int a5xx_crashdumper_run(struct msm_gpu *gpu, if (IS_ERR_OR_NULL(dumper->ptr)) return -EINVAL; - gpu_write64(gpu, REG_A5XX_CP_CRASH_SCRIPT_BASE_LO, - REG_A5XX_CP_CRASH_SCRIPT_BASE_HI, dumper->iova); + gpu_write64(gpu, REG_A5XX_CP_CRASH_SCRIPT_BASE_LO, dumper->iova); gpu_write(gpu, REG_A5XX_CP_CRASH_DUMP_CNTL, 1); @@ -1317,7 +1557,7 @@ static void a5xx_gpu_state_get_hlsq_regs(struct msm_gpu *gpu, if (a5xx_crashdumper_run(gpu, &dumper)) { kfree(a5xx_state->hlsqregs); - msm_gem_kernel_put(dumper.bo, gpu->aspace, true); + msm_gem_kernel_put(dumper.bo, gpu->vm); return; } @@ -1325,13 +1565,14 @@ static void a5xx_gpu_state_get_hlsq_regs(struct msm_gpu *gpu, memcpy(a5xx_state->hlsqregs, dumper.ptr + (256 * SZ_1K), count * sizeof(u32)); - msm_gem_kernel_put(dumper.bo, gpu->aspace, true); + msm_gem_kernel_put(dumper.bo, gpu->vm); } static struct msm_gpu_state *a5xx_gpu_state_get(struct msm_gpu *gpu) { struct a5xx_gpu_state *a5xx_state = kzalloc(sizeof(*a5xx_state), GFP_KERNEL); + bool stalled = !!(gpu_read(gpu, REG_A5XX_RBBM_STATUS3) & BIT(24)); if (!a5xx_state) return ERR_PTR(-ENOMEM); @@ -1344,8 +1585,13 @@ static struct msm_gpu_state *a5xx_gpu_state_get(struct msm_gpu *gpu) a5xx_state->base.rbbm_status = gpu_read(gpu, REG_A5XX_RBBM_STATUS); - /* Get the HLSQ regs with the help of the crashdumper */ - a5xx_gpu_state_get_hlsq_regs(gpu, a5xx_state); + /* + * Get the HLSQ regs with the help of the crashdumper, but only if + * we are not stalled in an iommu fault (in which case the crashdumper + * would not have access to memory) + */ + if (!stalled) + a5xx_gpu_state_get_hlsq_regs(gpu, a5xx_state); a5xx_set_hwcg(gpu, true); @@ -1365,7 +1611,7 @@ static void a5xx_gpu_state_destroy(struct kref *kref) kfree(a5xx_state); } -int a5xx_gpu_state_put(struct msm_gpu_state *state) +static int a5xx_gpu_state_put(struct msm_gpu_state *state) { if (IS_ERR_OR_NULL(state)) return 1; @@ -1375,8 +1621,8 @@ int a5xx_gpu_state_put(struct msm_gpu_state *state) #if defined(CONFIG_DEBUG_FS) || defined(CONFIG_DEV_COREDUMP) -void a5xx_show(struct msm_gpu *gpu, struct msm_gpu_state *state, - struct drm_printer *p) +static void a5xx_show(struct msm_gpu *gpu, struct msm_gpu_state *state, + struct drm_printer *p) { int i, j; u32 pos = 0; @@ -1424,82 +1670,71 @@ static struct msm_ringbuffer *a5xx_active_ring(struct msm_gpu *gpu) return a5xx_gpu->cur_ring; } -static unsigned long a5xx_gpu_busy(struct msm_gpu *gpu) +static u64 a5xx_gpu_busy(struct msm_gpu *gpu, unsigned long *out_sample_rate) { - u64 busy_cycles, busy_time; + u64 busy_cycles; - busy_cycles = gpu_read64(gpu, REG_A5XX_RBBM_PERFCTR_RBBM_0_LO, - REG_A5XX_RBBM_PERFCTR_RBBM_0_HI); + busy_cycles = gpu_read64(gpu, REG_A5XX_RBBM_PERFCTR_RBBM_0_LO); + *out_sample_rate = clk_get_rate(gpu->core_clk); - busy_time = busy_cycles - gpu->devfreq.busy_cycles; - do_div(busy_time, clk_get_rate(gpu->core_clk) / 1000000); + return busy_cycles; +} - gpu->devfreq.busy_cycles = busy_cycles; +static uint32_t a5xx_get_rptr(struct msm_gpu *gpu, struct msm_ringbuffer *ring) +{ + struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); + struct a5xx_gpu *a5xx_gpu = to_a5xx_gpu(adreno_gpu); - if (WARN_ON(busy_time > ~0LU)) - return ~0LU; + if (a5xx_gpu->has_whereami) + return a5xx_gpu->shadow[ring->id]; - return (unsigned long)busy_time; + return ring->memptrs->rptr = gpu_read(gpu, REG_A5XX_CP_RB_RPTR); } -static const struct adreno_gpu_funcs funcs = { - .base = { - .get_param = adreno_get_param, - .hw_init = a5xx_hw_init, - .pm_suspend = a5xx_pm_suspend, - .pm_resume = a5xx_pm_resume, - .recover = a5xx_recover, - .submit = a5xx_submit, - .flush = a5xx_flush, - .active_ring = a5xx_active_ring, - .irq = a5xx_irq, - .destroy = a5xx_destroy, -#if defined(CONFIG_DEBUG_FS) || defined(CONFIG_DEV_COREDUMP) - .show = a5xx_show, -#endif -#if defined(CONFIG_DEBUG_FS) - .debugfs_init = a5xx_debugfs_init, -#endif - .gpu_busy = a5xx_gpu_busy, - .gpu_state_get = a5xx_gpu_state_get, - .gpu_state_put = a5xx_gpu_state_put, - }, - .get_timestamp = a5xx_get_timestamp, -}; - static void check_speed_bin(struct device *dev) { struct nvmem_cell *cell; - u32 bin, val; + u32 val; + + /* + * If the OPP table specifies a opp-supported-hw property then we have + * to set something with dev_pm_opp_set_supported_hw() or the table + * doesn't get populated so pick an arbitrary value that should + * ensure the default frequencies are selected but not conflict with any + * actual bins + */ + val = 0x80; cell = nvmem_cell_get(dev, "speed_bin"); - /* If a nvmem cell isn't defined, nothing to do */ - if (IS_ERR(cell)) - return; + if (!IS_ERR(cell)) { + void *buf = nvmem_cell_read(cell, NULL); - bin = *((u32 *) nvmem_cell_read(cell, NULL)); - nvmem_cell_put(cell); + if (!IS_ERR(buf)) { + u8 bin = *((u8 *) buf); - val = (1 << bin); + val = (1 << bin); + kfree(buf); + } - dev_pm_opp_set_supported_hw(dev, &val, 1); + nvmem_cell_put(cell); + } + + devm_pm_opp_set_supported_hw(dev, &val, 1); } -struct msm_gpu *a5xx_gpu_init(struct drm_device *dev) +static struct msm_gpu *a5xx_gpu_init(struct drm_device *dev) { struct msm_drm_private *priv = dev->dev_private; struct platform_device *pdev = priv->gpu_pdev; + struct adreno_platform_config *config = pdev->dev.platform_data; + const struct qcom_ubwc_cfg_data *common_cfg; struct a5xx_gpu *a5xx_gpu = NULL; struct adreno_gpu *adreno_gpu; struct msm_gpu *gpu; + unsigned int nr_rings; int ret; - if (!pdev) { - DRM_DEV_ERROR(dev->dev, "No A5XX device is defined\n"); - return ERR_PTR(-ENXIO); - } - a5xx_gpu = kzalloc(sizeof(*a5xx_gpu), GFP_KERNEL); if (!a5xx_gpu) return ERR_PTR(-ENOMEM); @@ -1508,23 +1743,67 @@ struct msm_gpu *a5xx_gpu_init(struct drm_device *dev) gpu = &adreno_gpu->base; adreno_gpu->registers = a5xx_registers; - adreno_gpu->reg_offsets = a5xx_register_offsets; a5xx_gpu->lm_leakage = 0x4E001A; check_speed_bin(&pdev->dev); - ret = adreno_gpu_init(dev, pdev, adreno_gpu, &funcs, 4); + nr_rings = 4; + + if (config->info->revn == 510) + nr_rings = 1; + + ret = adreno_gpu_init(dev, pdev, adreno_gpu, config->info->funcs, nr_rings); if (ret) { a5xx_destroy(&(a5xx_gpu->base.base)); return ERR_PTR(ret); } - if (gpu->aspace) - msm_mmu_set_fault_handler(gpu->aspace->mmu, gpu, a5xx_fault_handler); + msm_mmu_set_fault_handler(to_msm_vm(gpu->vm)->mmu, gpu, + a5xx_fault_handler); /* Set up the preemption specific bits and pieces for each ringbuffer */ a5xx_preempt_init(gpu); + /* Inherit the common config and make some necessary fixups */ + common_cfg = qcom_ubwc_config_get_data(); + if (IS_ERR(common_cfg)) + return ERR_CAST(common_cfg); + + /* Copy the data into the internal struct to drop the const qualifier (temporarily) */ + adreno_gpu->_ubwc_config = *common_cfg; + adreno_gpu->ubwc_config = &adreno_gpu->_ubwc_config; + + adreno_gpu->uche_trap_base = 0x0001ffffffff0000ull; + return gpu; } + +const struct adreno_gpu_funcs a5xx_gpu_funcs = { + .base = { + .get_param = adreno_get_param, + .set_param = adreno_set_param, + .hw_init = a5xx_hw_init, + .ucode_load = a5xx_ucode_load, + .pm_suspend = a5xx_pm_suspend, + .pm_resume = a5xx_pm_resume, + .recover = a5xx_recover, + .submit = a5xx_submit, + .active_ring = a5xx_active_ring, + .irq = a5xx_irq, + .destroy = a5xx_destroy, +#if defined(CONFIG_DEBUG_FS) || defined(CONFIG_DEV_COREDUMP) + .show = a5xx_show, +#endif +#if defined(CONFIG_DEBUG_FS) + .debugfs_init = a5xx_debugfs_init, +#endif + .gpu_busy = a5xx_gpu_busy, + .gpu_state_get = a5xx_gpu_state_get, + .gpu_state_put = a5xx_gpu_state_put, + .create_vm = adreno_create_vm, + .get_rptr = a5xx_get_rptr, + }, + .init = a5xx_gpu_init, + .get_timestamp = a5xx_get_timestamp, +}; |