diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_device.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 594 |
1 files changed, 379 insertions, 215 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index c2394c8b4d6b..0171d240fcb0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -25,6 +25,8 @@ * Alex Deucher * Jerome Glisse */ + +#include <linux/aperture.h> #include <linux/power_supply.h> #include <linux/kthread.h> #include <linux/module.h> @@ -35,10 +37,9 @@ #include <linux/pci-p2pdma.h> #include <linux/apple-gmux.h> -#include <drm/drm_aperture.h> #include <drm/drm_atomic_helper.h> +#include <drm/drm_client_event.h> #include <drm/drm_crtc_helper.h> -#include <drm/drm_fb_helper.h> #include <drm/drm_probe_helper.h> #include <drm/amdgpu_drm.h> #include <linux/device.h> @@ -144,6 +145,51 @@ const char *amdgpu_asic_name[] = { "LAST", }; +#define AMDGPU_IP_BLK_MASK_ALL GENMASK(AMDGPU_MAX_IP_NUM - 1, 0) +/* + * Default init level where all blocks are expected to be initialized. This is + * the level of initialization expected by default and also after a full reset + * of the device. + */ +struct amdgpu_init_level amdgpu_init_default = { + .level = AMDGPU_INIT_LEVEL_DEFAULT, + .hwini_ip_block_mask = AMDGPU_IP_BLK_MASK_ALL, +}; + +/* + * Minimal blocks needed to be initialized before a XGMI hive can be reset. This + * is used for cases like reset on initialization where the entire hive needs to + * be reset before first use. + */ +struct amdgpu_init_level amdgpu_init_minimal_xgmi = { + .level = AMDGPU_INIT_LEVEL_MINIMAL_XGMI, + .hwini_ip_block_mask = + BIT(AMD_IP_BLOCK_TYPE_GMC) | BIT(AMD_IP_BLOCK_TYPE_SMC) | + BIT(AMD_IP_BLOCK_TYPE_COMMON) | BIT(AMD_IP_BLOCK_TYPE_IH) | + BIT(AMD_IP_BLOCK_TYPE_PSP) +}; + +static inline bool amdgpu_ip_member_of_hwini(struct amdgpu_device *adev, + enum amd_ip_block_type block) +{ + return (adev->init_lvl->hwini_ip_block_mask & (1U << block)) != 0; +} + +void amdgpu_set_init_level(struct amdgpu_device *adev, + enum amdgpu_init_lvl_id lvl) +{ + switch (lvl) { + case AMDGPU_INIT_LEVEL_MINIMAL_XGMI: + adev->init_lvl = &amdgpu_init_minimal_xgmi; + break; + case AMDGPU_INIT_LEVEL_DEFAULT: + fallthrough; + default: + adev->init_lvl = &amdgpu_init_default; + break; + } +} + static inline void amdgpu_device_stop_pending_resets(struct amdgpu_device *adev); /** @@ -227,6 +273,42 @@ void amdgpu_reg_state_sysfs_fini(struct amdgpu_device *adev) sysfs_remove_bin_file(&adev->dev->kobj, &bin_attr_reg_state); } +int amdgpu_ip_block_suspend(struct amdgpu_ip_block *ip_block) +{ + int r; + + if (ip_block->version->funcs->suspend) { + r = ip_block->version->funcs->suspend(ip_block); + if (r) { + dev_err(ip_block->adev->dev, + "suspend of IP block <%s> failed %d\n", + ip_block->version->funcs->name, r); + return r; + } + } + + ip_block->status.hw = false; + return 0; +} + +int amdgpu_ip_block_resume(struct amdgpu_ip_block *ip_block) +{ + int r; + + if (ip_block->version->funcs->resume) { + r = ip_block->version->funcs->resume(ip_block); + if (r) { + dev_err(ip_block->adev->dev, + "resume of IP block <%s> failed %d\n", + ip_block->version->funcs->name, r); + return r; + } + } + + ip_block->status.hw = true; + return 0; +} + /** * DOC: board_info * @@ -1655,7 +1737,7 @@ bool amdgpu_device_need_post(struct amdgpu_device *adev) } /* Don't post if we need to reset whole hive on init */ - if (adev->gmc.xgmi.pending_reset) + if (adev->init_lvl->level == AMDGPU_INIT_LEVEL_MINIMAL_XGMI) return false; if (adev->has_hw_reset) { @@ -2159,9 +2241,12 @@ int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev, if (!adev->ip_blocks[i].status.valid) continue; if (adev->ip_blocks[i].version->type == block_type) { - r = adev->ip_blocks[i].version->funcs->wait_for_idle((void *)adev); - if (r) - return r; + if (adev->ip_blocks[i].version->funcs->wait_for_idle) { + r = adev->ip_blocks[i].version->funcs->wait_for_idle( + &adev->ip_blocks[i]); + if (r) + return r; + } break; } } @@ -2170,26 +2255,24 @@ int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev, } /** - * amdgpu_device_ip_is_idle - is the hardware IP idle + * amdgpu_device_ip_is_valid - is the hardware IP enabled * * @adev: amdgpu_device pointer * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.) * - * Check if the hardware IP is idle or not. - * Returns true if it the IP is idle, false if not. + * Check if the hardware IP is enable or not. + * Returns true if it the IP is enable, false if not. */ -bool amdgpu_device_ip_is_idle(struct amdgpu_device *adev, - enum amd_ip_block_type block_type) +bool amdgpu_device_ip_is_valid(struct amdgpu_device *adev, + enum amd_ip_block_type block_type) { int i; for (i = 0; i < adev->num_ip_blocks; i++) { - if (!adev->ip_blocks[i].status.valid) - continue; if (adev->ip_blocks[i].version->type == block_type) - return adev->ip_blocks[i].version->funcs->is_idle((void *)adev); + return adev->ip_blocks[i].status.valid; } - return true; + return false; } @@ -2271,6 +2354,8 @@ int amdgpu_device_ip_block_add(struct amdgpu_device *adev, DRM_INFO("add ip block number %d <%s>\n", adev->num_ip_blocks, ip_block_version->funcs->name); + adev->ip_blocks[adev->num_ip_blocks].adev = adev; + adev->ip_blocks[adev->num_ip_blocks++].version = ip_block_version; return 0; @@ -2566,25 +2651,25 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev) total = true; for (i = 0; i < adev->num_ip_blocks; i++) { + ip_block = &adev->ip_blocks[i]; + if ((amdgpu_ip_block_mask & (1 << i)) == 0) { DRM_WARN("disabled ip block: %d <%s>\n", i, adev->ip_blocks[i].version->funcs->name); adev->ip_blocks[i].status.valid = false; - } else { - if (adev->ip_blocks[i].version->funcs->early_init) { - r = adev->ip_blocks[i].version->funcs->early_init((void *)adev); - if (r == -ENOENT) { - adev->ip_blocks[i].status.valid = false; - } else if (r) { - DRM_ERROR("early_init of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); - total = false; - } else { - adev->ip_blocks[i].status.valid = true; - } + } else if (ip_block->version->funcs->early_init) { + r = ip_block->version->funcs->early_init(ip_block); + if (r == -ENOENT) { + adev->ip_blocks[i].status.valid = false; + } else if (r) { + DRM_ERROR("early_init of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, r); + total = false; } else { adev->ip_blocks[i].status.valid = true; } + } else { + adev->ip_blocks[i].status.valid = true; } /* get the vbios after the asic_funcs are set up */ if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON) { @@ -2633,10 +2718,13 @@ static int amdgpu_device_ip_hw_init_phase1(struct amdgpu_device *adev) continue; if (adev->ip_blocks[i].status.hw) continue; + if (!amdgpu_ip_member_of_hwini( + adev, adev->ip_blocks[i].version->type)) + continue; if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON || (amdgpu_sriov_vf(adev) && (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP)) || adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH) { - r = adev->ip_blocks[i].version->funcs->hw_init(adev); + r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]); if (r) { DRM_ERROR("hw_init of IP block <%s> failed %d\n", adev->ip_blocks[i].version->funcs->name, r); @@ -2658,7 +2746,10 @@ static int amdgpu_device_ip_hw_init_phase2(struct amdgpu_device *adev) continue; if (adev->ip_blocks[i].status.hw) continue; - r = adev->ip_blocks[i].version->funcs->hw_init(adev); + if (!amdgpu_ip_member_of_hwini( + adev, adev->ip_blocks[i].version->type)) + continue; + r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]); if (r) { DRM_ERROR("hw_init of IP block <%s> failed %d\n", adev->ip_blocks[i].version->funcs->name, r); @@ -2681,6 +2772,10 @@ static int amdgpu_device_fw_loading(struct amdgpu_device *adev) if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_PSP) continue; + if (!amdgpu_ip_member_of_hwini(adev, + AMD_IP_BLOCK_TYPE_PSP)) + break; + if (!adev->ip_blocks[i].status.sw) continue; @@ -2689,22 +2784,18 @@ static int amdgpu_device_fw_loading(struct amdgpu_device *adev) break; if (amdgpu_in_reset(adev) || adev->in_suspend) { - r = adev->ip_blocks[i].version->funcs->resume(adev); - if (r) { - DRM_ERROR("resume of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + r = amdgpu_ip_block_resume(&adev->ip_blocks[i]); + if (r) return r; - } } else { - r = adev->ip_blocks[i].version->funcs->hw_init(adev); + r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]); if (r) { DRM_ERROR("hw_init of IP block <%s> failed %d\n", adev->ip_blocks[i].version->funcs->name, r); return r; } + adev->ip_blocks[i].status.hw = true; } - - adev->ip_blocks[i].status.hw = true; break; } } @@ -2786,6 +2877,7 @@ static int amdgpu_device_init_schedulers(struct amdgpu_device *adev) */ static int amdgpu_device_ip_init(struct amdgpu_device *adev) { + bool init_badpage; int i, r; r = amdgpu_ras_init(adev); @@ -2795,17 +2887,23 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) for (i = 0; i < adev->num_ip_blocks; i++) { if (!adev->ip_blocks[i].status.valid) continue; - r = adev->ip_blocks[i].version->funcs->sw_init((void *)adev); - if (r) { - DRM_ERROR("sw_init of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); - goto init_failed; + if (adev->ip_blocks[i].version->funcs->sw_init) { + r = adev->ip_blocks[i].version->funcs->sw_init(&adev->ip_blocks[i]); + if (r) { + DRM_ERROR("sw_init of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, r); + goto init_failed; + } } adev->ip_blocks[i].status.sw = true; + if (!amdgpu_ip_member_of_hwini( + adev, adev->ip_blocks[i].version->type)) + continue; + if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON) { /* need to do common hw init early so everything is set up for gmc */ - r = adev->ip_blocks[i].version->funcs->hw_init((void *)adev); + r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]); if (r) { DRM_ERROR("hw_init %d failed %d\n", i, r); goto init_failed; @@ -2822,7 +2920,7 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) DRM_ERROR("amdgpu_mem_scratch_init failed %d\n", r); goto init_failed; } - r = adev->ip_blocks[i].version->funcs->hw_init((void *)adev); + r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]); if (r) { DRM_ERROR("hw_init %d failed %d\n", i, r); goto init_failed; @@ -2895,7 +2993,8 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) * Note: theoretically, this should be called before all vram allocations * to protect retired page from abusing */ - r = amdgpu_ras_recovery_init(adev); + init_badpage = (adev->init_lvl->level != AMDGPU_INIT_LEVEL_MINIMAL_XGMI); + r = amdgpu_ras_recovery_init(adev, init_badpage); if (r) goto init_failed; @@ -2935,7 +3034,7 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) amdgpu_ttm_set_buffer_funcs_status(adev, true); /* Don't init kfd if whole hive need to be reset during init */ - if (!adev->gmc.xgmi.pending_reset) { + if (adev->init_lvl->level != AMDGPU_INIT_LEVEL_MINIMAL_XGMI) { kgd2kfd_init_zone_device(adev); amdgpu_amdkfd_device_init(adev); } @@ -3135,7 +3234,7 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) if (!adev->ip_blocks[i].status.hw) continue; if (adev->ip_blocks[i].version->funcs->late_init) { - r = adev->ip_blocks[i].version->funcs->late_init((void *)adev); + r = adev->ip_blocks[i].version->funcs->late_init(&adev->ip_blocks[i]); if (r) { DRM_ERROR("late_init of IP block <%s> failed %d\n", adev->ip_blocks[i].version->funcs->name, r); @@ -3206,6 +3305,25 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) return 0; } +static void amdgpu_ip_block_hw_fini(struct amdgpu_ip_block *ip_block) +{ + int r; + + if (!ip_block->version->funcs->hw_fini) { + DRM_ERROR("hw_fini of IP block <%s> not defined\n", + ip_block->version->funcs->name); + } else { + r = ip_block->version->funcs->hw_fini(ip_block); + /* XXX handle errors */ + if (r) { + DRM_DEBUG("hw_fini of IP block <%s> failed %d\n", + ip_block->version->funcs->name, r); + } + } + + ip_block->status.hw = false; +} + /** * amdgpu_device_smu_fini_early - smu hw_fini wrapper * @@ -3215,7 +3333,7 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) */ static void amdgpu_device_smu_fini_early(struct amdgpu_device *adev) { - int i, r; + int i; if (amdgpu_ip_version(adev, GC_HWIP, 0) > IP_VERSION(9, 0, 0)) return; @@ -3224,13 +3342,7 @@ static void amdgpu_device_smu_fini_early(struct amdgpu_device *adev) if (!adev->ip_blocks[i].status.hw) continue; if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) { - r = adev->ip_blocks[i].version->funcs->hw_fini((void *)adev); - /* XXX handle errors */ - if (r) { - DRM_DEBUG("hw_fini of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); - } - adev->ip_blocks[i].status.hw = false; + amdgpu_ip_block_hw_fini(&adev->ip_blocks[i]); break; } } @@ -3244,7 +3356,7 @@ static int amdgpu_device_ip_fini_early(struct amdgpu_device *adev) if (!adev->ip_blocks[i].version->funcs->early_fini) continue; - r = adev->ip_blocks[i].version->funcs->early_fini((void *)adev); + r = adev->ip_blocks[i].version->funcs->early_fini(&adev->ip_blocks[i]); if (r) { DRM_DEBUG("early_fini of IP block <%s> failed %d\n", adev->ip_blocks[i].version->funcs->name, r); @@ -3263,14 +3375,7 @@ static int amdgpu_device_ip_fini_early(struct amdgpu_device *adev) if (!adev->ip_blocks[i].status.hw) continue; - r = adev->ip_blocks[i].version->funcs->hw_fini((void *)adev); - /* XXX handle errors */ - if (r) { - DRM_DEBUG("hw_fini of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); - } - - adev->ip_blocks[i].status.hw = false; + amdgpu_ip_block_hw_fini(&adev->ip_blocks[i]); } if (amdgpu_sriov_vf(adev)) { @@ -3316,12 +3421,13 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev) amdgpu_ib_pool_fini(adev); amdgpu_seq64_fini(adev); } - - r = adev->ip_blocks[i].version->funcs->sw_fini((void *)adev); - /* XXX handle errors */ - if (r) { - DRM_DEBUG("sw_fini of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + if (adev->ip_blocks[i].version->funcs->sw_fini) { + r = adev->ip_blocks[i].version->funcs->sw_fini(&adev->ip_blocks[i]); + /* XXX handle errors */ + if (r) { + DRM_DEBUG("sw_fini of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, r); + } } adev->ip_blocks[i].status.sw = false; adev->ip_blocks[i].status.valid = false; @@ -3331,7 +3437,7 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev) if (!adev->ip_blocks[i].status.late_initialized) continue; if (adev->ip_blocks[i].version->funcs->late_fini) - adev->ip_blocks[i].version->funcs->late_fini((void *)adev); + adev->ip_blocks[i].version->funcs->late_fini(&adev->ip_blocks[i]); adev->ip_blocks[i].status.late_initialized = false; } @@ -3403,15 +3509,9 @@ static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev) continue; /* XXX handle errors */ - r = adev->ip_blocks[i].version->funcs->suspend(adev); - /* XXX handle errors */ - if (r) { - DRM_ERROR("suspend of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]); + if (r) return r; - } - - adev->ip_blocks[i].status.hw = false; } return 0; @@ -3449,14 +3549,9 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) } /* skip unnecessary suspend if we do not initialize them yet */ - if (adev->gmc.xgmi.pending_reset && - !(adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC || - adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC || - adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON || - adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH)) { - adev->ip_blocks[i].status.hw = false; + if (!amdgpu_ip_member_of_hwini( + adev, adev->ip_blocks[i].version->type)) continue; - } /* skip suspend of gfx/mes and psp for S0ix * gfx is in gfxoff state, so on resume it will exit gfxoff just @@ -3490,13 +3585,9 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) continue; /* XXX handle errors */ - r = adev->ip_blocks[i].version->funcs->suspend(adev); - /* XXX handle errors */ - if (r) { - DRM_ERROR("suspend of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); - } + r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]); adev->ip_blocks[i].status.hw = false; + /* handle putting the SMC in the appropriate state */ if (!amdgpu_sriov_vf(adev)) { if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) { @@ -3570,7 +3661,7 @@ static int amdgpu_device_ip_reinit_early_sriov(struct amdgpu_device *adev) !block->status.valid) continue; - r = block->version->funcs->hw_init(adev); + r = block->version->funcs->hw_init(&adev->ip_blocks[i]); DRM_INFO("RE-INIT-early: %s %s\n", block->version->funcs->name, r?"failed":"succeeded"); if (r) return r; @@ -3609,15 +3700,19 @@ static int amdgpu_device_ip_reinit_late_sriov(struct amdgpu_device *adev) block->status.hw) continue; - if (block->version->type == AMD_IP_BLOCK_TYPE_SMC) - r = block->version->funcs->resume(adev); - else - r = block->version->funcs->hw_init(adev); - - DRM_INFO("RE-INIT-late: %s %s\n", block->version->funcs->name, r?"failed":"succeeded"); - if (r) - return r; - block->status.hw = true; + if (block->version->type == AMD_IP_BLOCK_TYPE_SMC) { + r = amdgpu_ip_block_resume(&adev->ip_blocks[i]); + if (r) + return r; + } else { + r = block->version->funcs->hw_init(&adev->ip_blocks[i]); + if (r) { + DRM_ERROR("hw_init of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, r); + return r; + } + block->status.hw = true; + } } } @@ -3648,13 +3743,9 @@ static int amdgpu_device_ip_resume_phase1(struct amdgpu_device *adev) adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH || (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP && amdgpu_sriov_vf(adev))) { - r = adev->ip_blocks[i].version->funcs->resume(adev); - if (r) { - DRM_ERROR("resume of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + r = amdgpu_ip_block_resume(&adev->ip_blocks[i]); + if (r) return r; - } - adev->ip_blocks[i].status.hw = true; } } @@ -3686,13 +3777,9 @@ static int amdgpu_device_ip_resume_phase2(struct amdgpu_device *adev) adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH || adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP) continue; - r = adev->ip_blocks[i].version->funcs->resume(adev); - if (r) { - DRM_ERROR("resume of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + r = amdgpu_ip_block_resume(&adev->ip_blocks[i]); + if (r) return r; - } - adev->ip_blocks[i].status.hw = true; } return 0; @@ -4149,7 +4236,10 @@ int amdgpu_device_init(struct amdgpu_device *adev, * for throttling interrupt) = 60 seconds. */ ratelimit_state_init(&adev->throttling_logging_rs, (60 - 1) * HZ, 1); + ratelimit_state_init(&adev->virt.ras_telemetry_rs, 5 * HZ, 1); + ratelimit_set_flags(&adev->throttling_logging_rs, RATELIMIT_MSG_ON_RELEASE); + ratelimit_set_flags(&adev->virt.ras_telemetry_rs, RATELIMIT_MSG_ON_RELEASE); /* Registers mapping */ /* TODO: block userspace mapping of io register */ @@ -4193,13 +4283,19 @@ int amdgpu_device_init(struct amdgpu_device *adev, amdgpu_device_set_mcbp(adev); + /* + * By default, use default mode where all blocks are expected to be + * initialized. At present a 'swinit' of blocks is required to be + * completed before the need for a different level is detected. + */ + amdgpu_set_init_level(adev, AMDGPU_INIT_LEVEL_DEFAULT); /* early init functions */ r = amdgpu_device_ip_early_init(adev); if (r) return r; /* Get rid of things like offb */ - r = drm_aperture_remove_conflicting_pci_framebuffers(adev->pdev, &amdgpu_kms_driver); + r = aperture_remove_conflicting_pci_devices(adev->pdev, amdgpu_kms_driver.name); if (r) return r; @@ -4265,20 +4361,8 @@ int amdgpu_device_init(struct amdgpu_device *adev, if (!amdgpu_sriov_vf(adev) && amdgpu_asic_need_reset_on_init(adev)) { if (adev->gmc.xgmi.num_physical_nodes) { dev_info(adev->dev, "Pending hive reset.\n"); - adev->gmc.xgmi.pending_reset = true; - /* Only need to init necessary block for SMU to handle the reset */ - for (i = 0; i < adev->num_ip_blocks; i++) { - if (!adev->ip_blocks[i].status.valid) - continue; - if (!(adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC || - adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON || - adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH || - adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC)) { - DRM_DEBUG("IP %s disabled for hw_init.\n", - adev->ip_blocks[i].version->funcs->name); - adev->ip_blocks[i].status.hw = true; - } - } + amdgpu_set_init_level(adev, + AMDGPU_INIT_LEVEL_MINIMAL_XGMI); } else if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 10) && !amdgpu_device_has_display_hardware(adev)) { r = psp_gpu_reset(adev); @@ -4386,7 +4470,7 @@ fence_driver_init: /* enable clockgating, etc. after ib tests, etc. since some blocks require * explicit gating rather than handling it automatically. */ - if (!adev->gmc.xgmi.pending_reset) { + if (adev->init_lvl->level != AMDGPU_INIT_LEVEL_MINIMAL_XGMI) { r = amdgpu_device_ip_late_init(adev); if (r) { dev_err(adev->dev, "amdgpu_device_ip_late_init failed\n"); @@ -4436,6 +4520,7 @@ fence_driver_init: amdgpu_fru_sysfs_init(adev); amdgpu_reg_state_sysfs_init(adev); + amdgpu_xcp_cfg_sysfs_init(adev); if (IS_ENABLED(CONFIG_PERF_EVENTS)) r = amdgpu_pmu_init(adev); @@ -4463,9 +4548,8 @@ fence_driver_init: if (px) vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain); - if (adev->gmc.xgmi.pending_reset) - queue_delayed_work(system_wq, &mgpu_info.delayed_reset_work, - msecs_to_jiffies(AMDGPU_RESUME_MS)); + if (adev->init_lvl->level == AMDGPU_INIT_LEVEL_MINIMAL_XGMI) + amdgpu_xgmi_reset_on_init(adev); amdgpu_device_check_iommu_direct_map(adev); @@ -4559,6 +4643,7 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev) amdgpu_fru_sysfs_fini(adev); amdgpu_reg_state_sysfs_fini(adev); + amdgpu_xcp_cfg_sysfs_fini(adev); /* disable ras feature must before hw fini */ amdgpu_ras_pre_fini(adev); @@ -4694,7 +4779,7 @@ int amdgpu_device_prepare(struct drm_device *dev) continue; if (!adev->ip_blocks[i].version->funcs->prepare_suspend) continue; - r = adev->ip_blocks[i].version->funcs->prepare_suspend((void *)adev); + r = adev->ip_blocks[i].version->funcs->prepare_suspend(&adev->ip_blocks[i]); if (r) goto unprepare; } @@ -4711,13 +4796,13 @@ unprepare: * amdgpu_device_suspend - initiate device suspend * * @dev: drm dev pointer - * @fbcon : notify the fbdev of suspend + * @notify_clients: notify in-kernel DRM clients * * Puts the hw in the suspend state (all asics). * Returns 0 for success or an error on failure. * Called at driver suspend. */ -int amdgpu_device_suspend(struct drm_device *dev, bool fbcon) +int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) { struct amdgpu_device *adev = drm_to_adev(dev); int r = 0; @@ -4737,8 +4822,8 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon) if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DEV_D3)) DRM_WARN("smart shift update failed\n"); - if (fbcon) - drm_fb_helper_set_suspend_unlocked(adev_to_drm(adev)->fb_helper, true); + if (notify_clients) + drm_client_dev_suspend(adev_to_drm(adev), false); cancel_delayed_work_sync(&adev->delayed_init_work); @@ -4773,13 +4858,13 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon) * amdgpu_device_resume - initiate device resume * * @dev: drm dev pointer - * @fbcon : notify the fbdev of resume + * @notify_clients: notify in-kernel DRM clients * * Bring the hw back to operating state (all asics). * Returns 0 for success or an error on failure. * Called at driver resume. */ -int amdgpu_device_resume(struct drm_device *dev, bool fbcon) +int amdgpu_device_resume(struct drm_device *dev, bool notify_clients) { struct amdgpu_device *adev = drm_to_adev(dev); int r = 0; @@ -4835,8 +4920,8 @@ exit: /* Make sure IB tests flushed */ flush_delayed_work(&adev->delayed_init_work); - if (fbcon) - drm_fb_helper_set_suspend_unlocked(adev_to_drm(adev)->fb_helper, false); + if (notify_clients) + drm_client_dev_resume(adev_to_drm(adev), false); amdgpu_ras_resume(adev); @@ -4898,7 +4983,8 @@ static bool amdgpu_device_ip_check_soft_reset(struct amdgpu_device *adev) continue; if (adev->ip_blocks[i].version->funcs->check_soft_reset) adev->ip_blocks[i].status.hang = - adev->ip_blocks[i].version->funcs->check_soft_reset(adev); + adev->ip_blocks[i].version->funcs->check_soft_reset( + &adev->ip_blocks[i]); if (adev->ip_blocks[i].status.hang) { dev_info(adev->dev, "IP block:%s is hung!\n", adev->ip_blocks[i].version->funcs->name); asic_hang = true; @@ -4927,7 +5013,7 @@ static int amdgpu_device_ip_pre_soft_reset(struct amdgpu_device *adev) continue; if (adev->ip_blocks[i].status.hang && adev->ip_blocks[i].version->funcs->pre_soft_reset) { - r = adev->ip_blocks[i].version->funcs->pre_soft_reset(adev); + r = adev->ip_blocks[i].version->funcs->pre_soft_reset(&adev->ip_blocks[i]); if (r) return r; } @@ -4989,7 +5075,7 @@ static int amdgpu_device_ip_soft_reset(struct amdgpu_device *adev) continue; if (adev->ip_blocks[i].status.hang && adev->ip_blocks[i].version->funcs->soft_reset) { - r = adev->ip_blocks[i].version->funcs->soft_reset(adev); + r = adev->ip_blocks[i].version->funcs->soft_reset(&adev->ip_blocks[i]); if (r) return r; } @@ -5018,7 +5104,7 @@ static int amdgpu_device_ip_post_soft_reset(struct amdgpu_device *adev) continue; if (adev->ip_blocks[i].status.hang && adev->ip_blocks[i].version->funcs->post_soft_reset) - r = adev->ip_blocks[i].version->funcs->post_soft_reset(adev); + r = adev->ip_blocks[i].version->funcs->post_soft_reset(&adev->ip_blocks[i]); if (r) return r; } @@ -5103,6 +5189,9 @@ static int amdgpu_device_reset_sriov(struct amdgpu_device *adev, amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) || amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3)) amdgpu_ras_resume(adev); + + amdgpu_virt_ras_telemetry_post_reset(adev); + return 0; } @@ -5309,7 +5398,7 @@ int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, for (i = 0; i < tmp_adev->num_ip_blocks; i++) if (tmp_adev->ip_blocks[i].version->funcs->dump_ip_state) tmp_adev->ip_blocks[i].version->funcs - ->dump_ip_state((void *)tmp_adev); + ->dump_ip_state((void *)&tmp_adev->ip_blocks[i]); dev_info(tmp_adev->dev, "Dumping IP State Completed\n"); } @@ -5325,74 +5414,25 @@ int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, return r; } -int amdgpu_do_asic_reset(struct list_head *device_list_handle, - struct amdgpu_reset_context *reset_context) +int amdgpu_device_reinit_after_reset(struct amdgpu_reset_context *reset_context) { - struct amdgpu_device *tmp_adev = NULL; - bool need_full_reset, skip_hw_reset, vram_lost = false; - int r = 0; - - /* Try reset handler method first */ - tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device, - reset_list); - - reset_context->reset_device_list = device_list_handle; - r = amdgpu_reset_perform_reset(tmp_adev, reset_context); - /* If reset handler not implemented, continue; otherwise return */ - if (r == -EOPNOTSUPP) - r = 0; - else - return r; - - /* Reset handler not implemented, use the default method */ - need_full_reset = - test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags); - skip_hw_reset = test_bit(AMDGPU_SKIP_HW_RESET, &reset_context->flags); - - /* - * ASIC reset has to be done on all XGMI hive nodes ASAP - * to allow proper links negotiation in FW (within 1 sec) - */ - if (!skip_hw_reset && need_full_reset) { - list_for_each_entry(tmp_adev, device_list_handle, reset_list) { - /* For XGMI run all resets in parallel to speed up the process */ - if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { - tmp_adev->gmc.xgmi.pending_reset = false; - if (!queue_work(system_unbound_wq, &tmp_adev->xgmi_reset_work)) - r = -EALREADY; - } else - r = amdgpu_asic_reset(tmp_adev); + struct list_head *device_list_handle; + bool full_reset, vram_lost = false; + struct amdgpu_device *tmp_adev; + int r; - if (r) { - dev_err(tmp_adev->dev, "ASIC reset failed with error, %d for drm dev, %s", - r, adev_to_drm(tmp_adev)->unique); - goto out; - } - } + device_list_handle = reset_context->reset_device_list; - /* For XGMI wait for all resets to complete before proceed */ - if (!r) { - list_for_each_entry(tmp_adev, device_list_handle, reset_list) { - if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { - flush_work(&tmp_adev->xgmi_reset_work); - r = tmp_adev->asic_reset_res; - if (r) - break; - } - } - } - } + if (!device_list_handle) + return -EINVAL; - if (!r && amdgpu_ras_intr_triggered()) { - list_for_each_entry(tmp_adev, device_list_handle, reset_list) { - amdgpu_ras_reset_error_count(tmp_adev, AMDGPU_RAS_BLOCK__MMHUB); - } - - amdgpu_ras_intr_cleared(); - } + full_reset = test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags); + r = 0; list_for_each_entry(tmp_adev, device_list_handle, reset_list) { - if (need_full_reset) { + /* After reset, it's default init level */ + amdgpu_set_init_level(tmp_adev, AMDGPU_INIT_LEVEL_DEFAULT); + if (full_reset) { /* post card */ amdgpu_ras_set_fed(tmp_adev, false); r = amdgpu_device_asic_init(tmp_adev); @@ -5448,7 +5488,7 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle, if (r) goto out; - drm_fb_helper_set_suspend_unlocked(adev_to_drm(tmp_adev)->fb_helper, false); + drm_client_dev_resume(adev_to_drm(tmp_adev), false); /* * The GPU enters bad state once faulty pages @@ -5482,7 +5522,6 @@ out: r = amdgpu_ib_ring_tests(tmp_adev); if (r) { dev_err(tmp_adev->dev, "ib ring test failed (%d).\n", r); - need_full_reset = true; r = -EAGAIN; goto end; } @@ -5493,10 +5532,85 @@ out: } end: - if (need_full_reset) + return r; +} + +int amdgpu_do_asic_reset(struct list_head *device_list_handle, + struct amdgpu_reset_context *reset_context) +{ + struct amdgpu_device *tmp_adev = NULL; + bool need_full_reset, skip_hw_reset; + int r = 0; + + /* Try reset handler method first */ + tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device, + reset_list); + + reset_context->reset_device_list = device_list_handle; + r = amdgpu_reset_perform_reset(tmp_adev, reset_context); + /* If reset handler not implemented, continue; otherwise return */ + if (r == -EOPNOTSUPP) + r = 0; + else + return r; + + /* Reset handler not implemented, use the default method */ + need_full_reset = + test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags); + skip_hw_reset = test_bit(AMDGPU_SKIP_HW_RESET, &reset_context->flags); + + /* + * ASIC reset has to be done on all XGMI hive nodes ASAP + * to allow proper links negotiation in FW (within 1 sec) + */ + if (!skip_hw_reset && need_full_reset) { + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { + /* For XGMI run all resets in parallel to speed up the process */ + if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { + if (!queue_work(system_unbound_wq, + &tmp_adev->xgmi_reset_work)) + r = -EALREADY; + } else + r = amdgpu_asic_reset(tmp_adev); + + if (r) { + dev_err(tmp_adev->dev, + "ASIC reset failed with error, %d for drm dev, %s", + r, adev_to_drm(tmp_adev)->unique); + goto out; + } + } + + /* For XGMI wait for all resets to complete before proceed */ + if (!r) { + list_for_each_entry(tmp_adev, device_list_handle, + reset_list) { + if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) { + flush_work(&tmp_adev->xgmi_reset_work); + r = tmp_adev->asic_reset_res; + if (r) + break; + } + } + } + } + + if (!r && amdgpu_ras_intr_triggered()) { + list_for_each_entry(tmp_adev, device_list_handle, reset_list) { + amdgpu_ras_reset_error_count(tmp_adev, + AMDGPU_RAS_BLOCK__MMHUB); + } + + amdgpu_ras_intr_cleared(); + } + + r = amdgpu_device_reinit_after_reset(reset_context); + if (r == -EAGAIN) set_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags); else clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags); + +out: return r; } @@ -5734,7 +5848,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, */ amdgpu_unregister_gpu_instance(tmp_adev); - drm_fb_helper_set_suspend_unlocked(adev_to_drm(tmp_adev)->fb_helper, true); + drm_client_dev_suspend(adev_to_drm(tmp_adev), false); /* disable ras on ALL IPs */ if (!need_emergency_restart && @@ -5824,7 +5938,7 @@ skip_hw_reset: if (!amdgpu_ring_sched_ready(ring)) continue; - drm_sched_start(&ring->sched); + drm_sched_start(&ring->sched, 0); } if (!drm_drv_uses_atomic_modeset(adev_to_drm(tmp_adev)) && !job_signaled) @@ -6092,6 +6206,9 @@ bool amdgpu_device_is_peer_accessible(struct amdgpu_device *adev, bool p2p_access = !adev->gmc.xgmi.connected_to_cpu && !(pci_p2pdma_distance(adev->pdev, peer_adev->dev, false) < 0); + if (!p2p_access) + dev_info(adev->dev, "PCIe P2P access from peer device %s is not supported by the chipset\n", + pci_name(peer_adev->pdev)); bool is_large_bar = adev->gmc.visible_vram_size && adev->gmc.real_vram_size == adev->gmc.visible_vram_size; @@ -6331,7 +6448,7 @@ void amdgpu_pci_resume(struct pci_dev *pdev) if (!amdgpu_ring_sched_ready(ring)) continue; - drm_sched_start(&ring->sched); + drm_sched_start(&ring->sched, 0); } amdgpu_device_unset_mp1_state(adev); @@ -6344,6 +6461,9 @@ bool amdgpu_device_cache_pci_state(struct pci_dev *pdev) struct amdgpu_device *adev = drm_to_adev(dev); int r; + if (amdgpu_sriov_vf(adev)) + return false; + r = pci_save_state(pdev); if (!r) { kfree(adev->pci_state); @@ -6604,3 +6724,47 @@ uint32_t amdgpu_device_wait_on_rreg(struct amdgpu_device *adev, } return ret; } + +ssize_t amdgpu_get_soft_full_reset_mask(struct amdgpu_ring *ring) +{ + ssize_t size = 0; + + if (!ring || !ring->adev) + return size; + + if (amdgpu_device_should_recover_gpu(ring->adev)) + size |= AMDGPU_RESET_TYPE_FULL; + + if (unlikely(!ring->adev->debug_disable_soft_recovery) && + !amdgpu_sriov_vf(ring->adev) && ring->funcs->soft_recovery) + size |= AMDGPU_RESET_TYPE_SOFT_RESET; + + return size; +} + +ssize_t amdgpu_show_reset_mask(char *buf, uint32_t supported_reset) +{ + ssize_t size = 0; + + if (supported_reset == 0) { + size += sysfs_emit_at(buf, size, "unsupported"); + size += sysfs_emit_at(buf, size, "\n"); + return size; + + } + + if (supported_reset & AMDGPU_RESET_TYPE_SOFT_RESET) + size += sysfs_emit_at(buf, size, "soft "); + + if (supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE) + size += sysfs_emit_at(buf, size, "queue "); + + if (supported_reset & AMDGPU_RESET_TYPE_PER_PIPE) + size += sysfs_emit_at(buf, size, "pipe "); + + if (supported_reset & AMDGPU_RESET_TYPE_FULL) + size += sysfs_emit_at(buf, size, "full "); + + size += sysfs_emit_at(buf, size, "\n"); + return size; +} |