diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 543 |
1 files changed, 361 insertions, 182 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index c0ddbe7d6f0b..4c0c77eafbd1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -33,6 +33,7 @@ #include <drm/drm_vblank.h> #include <linux/cc_platform.h> +#include <linux/console.h> #include <linux/dynamic_debug.h> #include <linux/module.h> #include <linux/mmu_notifier.h> @@ -51,6 +52,8 @@ #include "amdgpu_reset.h" #include "amdgpu_sched.h" #include "amdgpu_xgmi.h" +#include "amdgpu_userq.h" +#include "amdgpu_userq_fence.h" #include "../amdxcp/amdgpu_xcp_drv.h" /* @@ -121,9 +124,12 @@ * - 3.59.0 - Cleared VRAM * - 3.60.0 - Add AMDGPU_TILING_GFX12_DCC_WRITE_COMPRESS_DISABLE (Vulkan requirement) * - 3.61.0 - Contains fix for RV/PCO compute queues + * - 3.62.0 - Add AMDGPU_IDS_FLAGS_MODE_PF, AMDGPU_IDS_FLAGS_MODE_VF & AMDGPU_IDS_FLAGS_MODE_PT + * - 3.63.0 - GFX12 display DCC supports 256B max compressed block size + * - 3.64.0 - Userq IP support query */ #define KMS_DRIVER_MAJOR 3 -#define KMS_DRIVER_MINOR 61 +#define KMS_DRIVER_MINOR 64 #define KMS_DRIVER_PATCHLEVEL 0 /* @@ -136,6 +142,11 @@ enum AMDGPU_DEBUG_MASK { AMDGPU_DEBUG_USE_VRAM_FW_BUF = BIT(3), AMDGPU_DEBUG_ENABLE_RAS_ACA = BIT(4), AMDGPU_DEBUG_ENABLE_EXP_RESETS = BIT(5), + AMDGPU_DEBUG_DISABLE_GPU_RING_RESET = BIT(6), + AMDGPU_DEBUG_SMU_POOL = BIT(7), + AMDGPU_DEBUG_VM_USERPTR = BIT(8), + AMDGPU_DEBUG_DISABLE_RAS_CE_LOG = BIT(9), + AMDGPU_DEBUG_ENABLE_CE_CS = BIT(10) }; unsigned int amdgpu_vram_limit = UINT_MAX; @@ -172,7 +183,8 @@ uint amdgpu_pg_mask = 0xffffffff; uint amdgpu_sdma_phase_quantum = 32; char *amdgpu_disable_cu; char *amdgpu_virtual_display; -bool enforce_isolation; +int amdgpu_enforce_isolation = -1; +int amdgpu_modeset = -1; /* Specifies the default granularity for SVM, used in buffer * migration and restoration of backing memory when handling @@ -211,9 +223,7 @@ uint amdgpu_dc_visual_confirm; int amdgpu_async_gfx_ring = 1; int amdgpu_mcbp = -1; int amdgpu_discovery = -1; -int amdgpu_mes; int amdgpu_mes_log_enable = 0; -int amdgpu_mes_kiq; int amdgpu_uni_mes = 1; int amdgpu_noretry = -1; int amdgpu_force_asic_type = -1; @@ -233,6 +243,10 @@ int amdgpu_agp = -1; /* auto */ int amdgpu_wbrf = -1; int amdgpu_damage_clips = -1; /* auto */ int amdgpu_umsch_mm_fwlog; +int amdgpu_rebar = -1; /* auto */ +int amdgpu_user_queue = -1; +uint amdgpu_hdmi_hpd_debounce_delay_ms; +int amdgpu_ptl = -1; /* auto */ DECLARE_DYNDBG_CLASSMAP(drm_debug_classes, DD_CLASS_TYPE_DISJOINT_BITS, 0, "DRM_UT_CORE", @@ -283,6 +297,7 @@ module_param_named(gartsize, amdgpu_gart_size, uint, 0600); * DOC: gttsize (int) * Restrict the size of GTT domain (for userspace use) in MiB for testing. * The default is -1 (Use value specified by TTM). + * This parameter is deprecated and will be removed in the future. */ MODULE_PARM_DESC(gttsize, "Size of the GTT userspace domain in megabytes (-1 = auto)"); module_param_named(gttsize, amdgpu_gtt_size, int, 0600); @@ -298,7 +313,7 @@ module_param_named(moverate, amdgpu_moverate, int, 0600); * DOC: audio (int) * Set HDMI/DPAudio. Only affects non-DC display handling. The default is -1 (Enabled), set 0 to disabled it. */ -MODULE_PARM_DESC(audio, "Audio enable (-1 = auto, 0 = disable, 1 = enable)"); +MODULE_PARM_DESC(audio, "HDMI/DP Audio enable for non DC displays (-1 = auto, 0 = disable, 1 = enable)"); module_param_named(audio, amdgpu_audio, int, 0444); /** @@ -340,22 +355,16 @@ module_param_named(svm_default_granularity, amdgpu_svm_default_granularity, uint * DOC: lockup_timeout (string) * Set GPU scheduler timeout value in ms. * - * The format can be [Non-Compute] or [GFX,Compute,SDMA,Video]. That is there can be one or - * multiple values specified. 0 and negative values are invalidated. They will be adjusted - * to the default timeout. + * The format can be [single value] for setting all timeouts at once or + * [GFX,Compute,SDMA,Video] to set individual timeouts. + * Negative values mean infinity. * - * - With one value specified, the setting will apply to all non-compute jobs. - * - With multiple values specified, the first one will be for GFX. - * The second one is for Compute. The third and fourth ones are - * for SDMA and Video. - * - * By default(with no lockup_timeout settings), the timeout for all non-compute(GFX, SDMA and Video) - * jobs is 10000. The timeout for compute is 60000. + * By default(with no lockup_timeout settings), the timeout for all queues is 2000. */ -MODULE_PARM_DESC(lockup_timeout, "GPU lockup timeout in ms (default: for bare metal 10000 for non-compute jobs and 60000 for compute jobs; " - "for passthrough or sriov, 10000 for all jobs. 0: keep default value. negative: infinity timeout), format: for bare metal [Non-Compute] or [GFX,Compute,SDMA,Video]; " - "for passthrough or sriov [all jobs] or [GFX,Compute,SDMA,Video]."); -module_param_string(lockup_timeout, amdgpu_lockup_timeout, sizeof(amdgpu_lockup_timeout), 0444); +MODULE_PARM_DESC(lockup_timeout, + "GPU lockup timeout in ms (default: 2000. 0: keep default value. negative: infinity timeout), format: [single value for all] or [GFX,Compute,SDMA,Video]."); +module_param_string(lockup_timeout, amdgpu_lockup_timeout, + sizeof(amdgpu_lockup_timeout), 0444); /** * DOC: dpm (int) @@ -610,39 +619,37 @@ module_param_named(timeout_period, amdgpu_watchdog_timer.period, uint, 0644); /** * DOC: si_support (int) - * Set SI support driver. This parameter works after set config CONFIG_DRM_AMDGPU_SI. For SI asic, when radeon driver is enabled, - * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available, - * otherwise using amdgpu driver. - */ + * 1 = enabled, 0 = disabled, -1 = default + * + * SI (Southern Islands) are first generation GCN GPUs, supported by both + * drivers: radeon (old) and amdgpu (new). This parameter controls whether + * amdgpu should support SI. + * By default, SI dedicated GPUs are supported by amdgpu. + * Only relevant when CONFIG_DRM_AMDGPU_SI is enabled to build SI support in amdgpu. + * See also radeon.si_support which should be disabled when amdgpu.si_support is + * enabled, and vice versa. + */ +int amdgpu_si_support = -1; #ifdef CONFIG_DRM_AMDGPU_SI - -#if IS_ENABLED(CONFIG_DRM_RADEON) || IS_ENABLED(CONFIG_DRM_RADEON_MODULE) -int amdgpu_si_support; -MODULE_PARM_DESC(si_support, "SI support (1 = enabled, 0 = disabled (default))"); -#else -int amdgpu_si_support = 1; -MODULE_PARM_DESC(si_support, "SI support (1 = enabled (default), 0 = disabled)"); -#endif - +MODULE_PARM_DESC(si_support, "SI support (1 = enabled, 0 = disabled, -1 = default)"); module_param_named(si_support, amdgpu_si_support, int, 0444); #endif /** * DOC: cik_support (int) - * Set CIK support driver. This parameter works after set config CONFIG_DRM_AMDGPU_CIK. For CIK asic, when radeon driver is enabled, - * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available, - * otherwise using amdgpu driver. - */ + * 1 = enabled, 0 = disabled, -1 = default + * + * CIK (Sea Islands) are second generation GCN GPUs, supported by both + * drivers: radeon (old) and amdgpu (new). This parameter controls whether + * amdgpu should support CIK. + * By default, CIK dedicated GPUs and APUs are supported by amdgpu. + * Only relevant when CONFIG_DRM_AMDGPU_CIK is enabled to build CIK support in amdgpu. + * See also radeon.cik_support which should be disabled when amdgpu.cik_support is + * enabled, and vice versa. + */ +int amdgpu_cik_support = -1; #ifdef CONFIG_DRM_AMDGPU_CIK - -#if IS_ENABLED(CONFIG_DRM_RADEON) || IS_ENABLED(CONFIG_DRM_RADEON_MODULE) -int amdgpu_cik_support; -MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled, 0 = disabled (default))"); -#else -int amdgpu_cik_support = 1; -MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled (default), 0 = disabled)"); -#endif - +MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled, 0 = disabled, -1 = default)"); module_param_named(cik_support, amdgpu_cik_support, int, 0444); #endif @@ -681,15 +688,6 @@ MODULE_PARM_DESC(discovery, module_param_named(discovery, amdgpu_discovery, int, 0444); /** - * DOC: mes (int) - * Enable Micro Engine Scheduler. This is a new hw scheduling engine for gfx, sdma, and compute. - * (0 = disabled (default), 1 = enabled) - */ -MODULE_PARM_DESC(mes, - "Enable Micro Engine Scheduler (0 = disabled (default), 1 = enabled)"); -module_param_named(mes, amdgpu_mes, int, 0444); - -/** * DOC: mes_log_enable (int) * Enable Micro Engine Scheduler log. This is used to enable/disable MES internal log. * (0 = disabled (default), 1 = enabled) @@ -699,15 +697,6 @@ MODULE_PARM_DESC(mes_log_enable, module_param_named(mes_log_enable, amdgpu_mes_log_enable, int, 0444); /** - * DOC: mes_kiq (int) - * Enable Micro Engine Scheduler KIQ. This is a new engine pipe for kiq. - * (0 = disabled (default), 1 = enabled) - */ -MODULE_PARM_DESC(mes_kiq, - "Enable Micro Engine Scheduler KIQ (0 = disabled (default), 1 = enabled)"); -module_param_named(mes_kiq, amdgpu_mes_kiq, int, 0444); - -/** * DOC: uni_mes (int) * Enable Unified Micro Engine Scheduler. This is a new engine pipe for unified scheduler. * (0 = disabled (default), 1 = enabled) @@ -849,8 +838,8 @@ module_param_named_unsafe(no_queue_eviction_on_vm_fault, amdgpu_no_queue_evictio /** * DOC: mtype_local (int) */ -int amdgpu_mtype_local; -MODULE_PARM_DESC(mtype_local, "MTYPE for local memory (0 = MTYPE_RW (default), 1 = MTYPE_NC, 2 = MTYPE_CC)"); +int amdgpu_mtype_local = -1; +MODULE_PARM_DESC(mtype_local, "MTYPE for local memory (default: ASIC dependent, 0 = MTYPE_RW, 1 = MTYPE_NC, 2 = MTYPE_CC)"); module_param_named_unsafe(mtype_local, amdgpu_mtype_local, int, 0444); /** @@ -873,7 +862,7 @@ module_param_named(dcfeaturemask, amdgpu_dc_feature_mask, uint, 0444); /** * DOC: dcdebugmask (uint) - * Override display features enabled. See enum DC_DEBUG_MASK in drivers/gpu/drm/amd/include/amd_shared.h. + * Display debug options. See enum DC_DEBUG_MASK in drivers/gpu/drm/amd/include/amd_shared.h. */ MODULE_PARM_DESC(dcdebugmask, "all debug options disabled (default))"); module_param_named(dcdebugmask, amdgpu_dc_debug_mask, uint, 0444); @@ -947,7 +936,7 @@ module_param_named(tmz, amdgpu_tmz, int, 0444); */ MODULE_PARM_DESC( freesync_video, - "Enable freesync modesetting optimization feature (0 = off (default), 1 = on)"); + "Adds additional modes via VRR for refresh changes without a full modeset (0 = off (default), 1 = on)"); module_param_named(freesync_video, amdgpu_freesync_vid_mode, uint, 0444); /** @@ -963,7 +952,7 @@ module_param_named_unsafe(reset_method, amdgpu_reset_method, int, 0644); * result in the GPU entering bad status when the number of total * faulty pages by ECC exceeds the threshold value. */ -MODULE_PARM_DESC(bad_page_threshold, "Bad page threshold(-1 = ignore threshold (default value), 0 = disable bad page retirement, -2 = driver sets threshold)"); +MODULE_PARM_DESC(bad_page_threshold, "Bad page threshold(-1 = ignore threshold (default value), 0 = disable bad page retirement, -2 = threshold determined by a formula, 0 < threshold < max records, user-defined threshold)"); module_param_named(bad_page_threshold, amdgpu_bad_page_threshold, int, 0444); MODULE_PARM_DESC(num_kcq, "number of kernel compute queue user want to setup (8 if set to greater than 8 or less than 0, only affect gfx 8+)"); @@ -1027,11 +1016,20 @@ module_param_named(user_partt_mode, amdgpu_user_partt_mode, uint, 0444); /** - * DOC: enforce_isolation (bool) - * enforce process isolation between graphics and compute via using the same reserved vmid. + * DOC: enforce_isolation (int) + * enforce process isolation between graphics and compute. + * (-1 = auto, 0 = disable, 1 = enable, 2 = enable legacy mode, 3 = enable without cleaner shader) */ -module_param(enforce_isolation, bool, 0444); -MODULE_PARM_DESC(enforce_isolation, "enforce process isolation between graphics and compute . enforce_isolation = on"); +module_param_named(enforce_isolation, amdgpu_enforce_isolation, int, 0444); +MODULE_PARM_DESC(enforce_isolation, +"enforce process isolation between graphics and compute. (-1 = auto, 0 = disable, 1 = enable, 2 = enable legacy mode, 3 = enable without cleaner shader)"); + +/** + * DOC: modeset (int) + * Override nomodeset (1 = override, -1 = auto). The default is -1 (auto). + */ +MODULE_PARM_DESC(modeset, "Override nomodeset (1 = enable, -1 = auto)"); +module_param_named(modeset, amdgpu_modeset, int, 0444); /** * DOC: seamless (int) @@ -1049,6 +1047,11 @@ module_param_named(seamless, amdgpu_seamless, int, 0444); * limits the VRAM size reported to ROCm applications to the visible * size, usually 256MB. * - 0x4: Disable GPU soft recovery, always do a full reset + * - 0x8: Use VRAM for firmware loading + * - 0x10: Enable ACA based RAS logging + * - 0x20: Enable experimental resets + * - 0x40: Disable ring resets + * - 0x80: Use VRAM for SMU pool */ MODULE_PARM_DESC(debug_mask, "debug options for amdgpu, disabled by default"); module_param_named_unsafe(debug_mask, amdgpu_debug_mask, uint, 0444); @@ -1078,6 +1081,50 @@ MODULE_PARM_DESC(wbrf, "Enable Wifi RFI interference mitigation (0 = disabled, 1 = enabled, -1 = auto(default)"); module_param_named(wbrf, amdgpu_wbrf, int, 0444); +/** + * DOC: rebar (int) + * Allow BAR resizing. Disable this to prevent the driver from attempting + * to resize the BAR if the GPU supports it and there is available MMIO space. + * Note that this just prevents the driver from resizing the BAR. The BIOS + * may have already resized the BAR at boot time. + */ +MODULE_PARM_DESC(rebar, "Resizable BAR (-1 = auto (default), 0 = disable, 1 = enable)"); +module_param_named(rebar, amdgpu_rebar, int, 0444); + +/** + * DOC: user_queue (int) + * Enable user queues on systems that support user queues. Possible values: + * + * - -1 = auto (ASIC specific default) + * - 0 = user queues disabled + * - 1 = user queues enabled and kernel queues enabled (if supported) + * - 2 = user queues enabled and kernel queues disabled + */ +MODULE_PARM_DESC(user_queue, "Enable user queues (-1 = auto (default), 0 = disable, 1 = enable, 2 = enable UQs and disable KQs)"); +module_param_named(user_queue, amdgpu_user_queue, int, 0444); + +/* + * DOC: hdmi_hpd_debounce_delay_ms (uint) + * HDMI HPD disconnect debounce delay in milliseconds. + * + * Used to filter short disconnect->reconnect HPD toggles some HDMI sinks + * generate while entering/leaving power save. Set to 0 to disable by default. + */ +MODULE_PARM_DESC(hdmi_hpd_debounce_delay_ms, "HDMI HPD disconnect debounce delay in milliseconds (0 to disable (by default), 1500 is common)"); +module_param_named(hdmi_hpd_debounce_delay_ms, amdgpu_hdmi_hpd_debounce_delay_ms, uint, 0644); + +/** + * DOC: ptl (int) + * Enable PTL feature at boot time. Possible values: + * + * - -1 = auto (ASIC specific default) + * - 0 = disable PTL (default) + * - 1 = enable PTL + * - 2 = permanently disable PTL (cannot be re-enabled at runtime) + */ +MODULE_PARM_DESC(ptl, "Enable PTL (-1 = auto, 0 = disable (default), 1 = enable, 2 = permanently disable)"); +module_param_named(ptl, amdgpu_ptl, int, 0444); + /* These devices are not supported by amdgpu. * They are supported by the mach64, r128, radeon drivers */ @@ -1791,7 +1838,6 @@ static const u16 amdgpu_unsupported_pciidlist[] = { }; static const struct pci_device_id pciidlist[] = { -#ifdef CONFIG_DRM_AMDGPU_SI {0x1002, 0x6780, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI}, {0x1002, 0x6784, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI}, {0x1002, 0x6788, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI}, @@ -1864,8 +1910,6 @@ static const struct pci_device_id pciidlist[] = { {0x1002, 0x6665, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY}, {0x1002, 0x6667, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY}, {0x1002, 0x666F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY}, -#endif -#ifdef CONFIG_DRM_AMDGPU_CIK /* Kaveri */ {0x1002, 0x1304, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU}, {0x1002, 0x1305, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU}, @@ -1948,7 +1992,6 @@ static const struct pci_device_id pciidlist[] = { {0x1002, 0x985D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU}, {0x1002, 0x985E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU}, {0x1002, 0x985F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU}, -#endif /* topaz */ {0x1002, 0x6900, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ}, {0x1002, 0x6901, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ}, @@ -2127,6 +2170,11 @@ static const struct pci_device_id pciidlist[] = { {0x1002, 0x7410, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN}, /* CYAN_SKILLFISH */ + {0x1002, 0x13DB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU}, + {0x1002, 0x13F9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU}, + {0x1002, 0x13FA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU}, + {0x1002, 0x13FB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU}, + {0x1002, 0x13FC, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU}, {0x1002, 0x13FE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU}, {0x1002, 0x143F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU}, @@ -2183,7 +2231,6 @@ static void amdgpu_get_secondary_funcs(struct amdgpu_device *adev) adev->pdev->bus->number, i); if (p) { pm_runtime_get_sync(&p->dev); - pm_runtime_mark_last_busy(&p->dev); pm_runtime_put_autosuspend(&p->dev); pci_dev_put(p); } @@ -2221,6 +2268,29 @@ static void amdgpu_init_debug_options(struct amdgpu_device *adev) pr_info("debug: enable experimental reset features\n"); adev->debug_exp_resets = true; } + + if (amdgpu_debug_mask & AMDGPU_DEBUG_DISABLE_GPU_RING_RESET) { + pr_info("debug: ring reset disabled\n"); + adev->debug_disable_gpu_ring_reset = true; + } + if (amdgpu_debug_mask & AMDGPU_DEBUG_SMU_POOL) { + pr_info("debug: use vram for smu pool\n"); + adev->pm.smu_debug_mask |= SMU_DEBUG_POOL_USE_VRAM; + } + if (amdgpu_debug_mask & AMDGPU_DEBUG_VM_USERPTR) { + pr_info("debug: VM mode debug for userptr is enabled\n"); + adev->debug_vm_userptr = true; + } + + if (amdgpu_debug_mask & AMDGPU_DEBUG_DISABLE_RAS_CE_LOG) { + pr_info("debug: disable kernel logs of correctable errors\n"); + adev->debug_disable_ce_logs = true; + } + + if (amdgpu_debug_mask & AMDGPU_DEBUG_ENABLE_CE_CS) { + pr_info("debug: allowing command submission to CE engine\n"); + adev->debug_enable_ce_cs = true; + } } static unsigned long amdgpu_fix_asic_type(struct pci_dev *pdev, unsigned long flags) @@ -2239,6 +2309,71 @@ static unsigned long amdgpu_fix_asic_type(struct pci_dev *pdev, unsigned long fl return flags; } +static bool amdgpu_support_enabled(struct device *dev, + const enum amd_asic_type family) +{ + const char *gen; + const char *param; + int module_param = -1; + bool radeon_support_built = IS_ENABLED(CONFIG_DRM_RADEON); + bool amdgpu_support_built = false; + bool support_by_default = false; + + switch (family) { + case CHIP_TAHITI: + case CHIP_PITCAIRN: + case CHIP_VERDE: + case CHIP_OLAND: + case CHIP_HAINAN: + gen = "SI"; + param = "si_support"; + module_param = amdgpu_si_support; + amdgpu_support_built = IS_ENABLED(CONFIG_DRM_AMDGPU_SI); + support_by_default = true; + break; + + case CHIP_BONAIRE: + case CHIP_HAWAII: + case CHIP_KAVERI: + case CHIP_KABINI: + case CHIP_MULLINS: + gen = "CIK"; + param = "cik_support"; + module_param = amdgpu_cik_support; + amdgpu_support_built = IS_ENABLED(CONFIG_DRM_AMDGPU_CIK); + support_by_default = true; + break; + + default: + /* All other chips are supported by amdgpu only */ + return true; + } + + if (!amdgpu_support_built) { + dev_info(dev, "amdgpu built without %s support\n", gen); + return false; + } + + if ((module_param == -1 && (support_by_default || !radeon_support_built)) || + module_param == 1) { + if (radeon_support_built) + dev_info(dev, "%s support provided by amdgpu.\n" + "Use radeon.%s=1 amdgpu.%s=0 to override.\n", + gen, param, param); + + return true; + } + + if (radeon_support_built) + dev_info(dev, "%s support provided by radeon.\n" + "Use radeon.%s=0 amdgpu.%s=1 to override.\n", + gen, param, param); + else if (module_param == 0) + dev_info(dev, "%s support disabled by module param\n", gen); + + return false; +} + static int amdgpu_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) { @@ -2248,21 +2383,24 @@ static int amdgpu_pci_probe(struct pci_dev *pdev, int ret, retry = 0, i; bool supports_atomic = false; + if ((pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA || + (pdev->class >> 8) == PCI_CLASS_DISPLAY_OTHER) { + if (drm_firmware_drivers_only() && amdgpu_modeset == -1) + return -EINVAL; + } + /* skip devices which are owned by radeon */ for (i = 0; i < ARRAY_SIZE(amdgpu_unsupported_pciidlist); i++) { if (amdgpu_unsupported_pciidlist[i] == pdev->device) return -ENODEV; } - if (amdgpu_aspm == -1 && !pcie_aspm_enabled(pdev)) - amdgpu_aspm = 0; - if (amdgpu_virtual_display || - amdgpu_device_asic_has_dc_support(flags & AMD_ASIC_MASK)) + amdgpu_device_asic_has_dc_support(pdev, flags & AMD_ASIC_MASK)) supports_atomic = true; if ((flags & AMD_EXP_HW_SUPPORT) && !amdgpu_exp_hw_support) { - DRM_INFO("This hardware requires experimental hardware support.\n" + dev_info(&pdev->dev, "This hardware requires experimental hardware support.\n" "See modparam exp_hw_support\n"); return -ENODEV; } @@ -2280,40 +2418,8 @@ static int amdgpu_pci_probe(struct pci_dev *pdev, return -ENOTSUPP; } -#ifdef CONFIG_DRM_AMDGPU_SI - if (!amdgpu_si_support) { - switch (flags & AMD_ASIC_MASK) { - case CHIP_TAHITI: - case CHIP_PITCAIRN: - case CHIP_VERDE: - case CHIP_OLAND: - case CHIP_HAINAN: - dev_info(&pdev->dev, - "SI support provided by radeon.\n"); - dev_info(&pdev->dev, - "Use radeon.si_support=0 amdgpu.si_support=1 to override.\n" - ); - return -ENODEV; - } - } -#endif -#ifdef CONFIG_DRM_AMDGPU_CIK - if (!amdgpu_cik_support) { - switch (flags & AMD_ASIC_MASK) { - case CHIP_KAVERI: - case CHIP_BONAIRE: - case CHIP_HAWAII: - case CHIP_KABINI: - case CHIP_MULLINS: - dev_info(&pdev->dev, - "CIK support provided by radeon.\n"); - dev_info(&pdev->dev, - "Use radeon.cik_support=0 amdgpu.cik_support=1 to override.\n" - ); - return -ENODEV; - } - } -#endif + if (!amdgpu_support_enabled(&pdev->dev, flags & AMD_ASIC_MASK)) + return -ENODEV; adev = devm_drm_dev_alloc(&pdev->dev, &amdgpu_kms_driver, typeof(*adev), ddev); if (IS_ERR(adev)) @@ -2341,7 +2447,7 @@ static int amdgpu_pci_probe(struct pci_dev *pdev, retry_init: ret = drm_dev_register(ddev, flags); if (ret == -EAGAIN && ++retry <= 3) { - DRM_INFO("retry init %d\n", retry); + drm_info(adev_to_drm(adev), "retry init %d\n", retry); /* Don't request EX mode too frequently which is attacking */ msleep(5000); goto retry_init; @@ -2380,10 +2486,10 @@ retry_init: if (adev->pm.rpm_mode != AMDGPU_RUNPM_NONE) { /* only need to skip on ATPX */ - if (amdgpu_device_supports_px(ddev)) + if (amdgpu_device_supports_px(adev)) dev_pm_set_driver_flags(ddev->dev, DPM_FLAG_NO_DIRECT_COMPLETE); /* we want direct complete for BOCO */ - if (amdgpu_device_supports_boco(ddev)) + if (amdgpu_device_supports_boco(adev)) dev_pm_set_driver_flags(ddev->dev, DPM_FLAG_SMART_PREPARE | DPM_FLAG_SMART_SUSPEND | DPM_FLAG_MAY_SKIP_RESUME); @@ -2392,7 +2498,6 @@ retry_init: pm_runtime_allow(ddev->dev); - pm_runtime_mark_last_busy(ddev->dev); pm_runtime_put_autosuspend(ddev->dev); pci_wake_from_d3(pdev, TRUE); @@ -2416,9 +2521,9 @@ retry_init: * into D0 state. Then there will be a PMFW-aware D-state * transition(D0->D3) on runpm suspend. */ - if (amdgpu_device_supports_baco(ddev) && + if (amdgpu_device_supports_baco(adev) && !(adev->flags & AMD_IS_APU) && - (adev->asic_type >= CHIP_NAVI10)) + adev->asic_type >= CHIP_NAVI10) amdgpu_get_secondary_funcs(adev); } @@ -2435,6 +2540,7 @@ amdgpu_pci_remove(struct pci_dev *pdev) struct drm_device *dev = pci_get_drvdata(pdev); struct amdgpu_device *adev = drm_to_adev(dev); + amdgpu_ras_eeprom_check_and_recover(adev); amdgpu_xcp_dev_unplug(adev); amdgpu_gmc_prepare_nps_mode_change(adev); drm_dev_unplug(dev); @@ -2464,6 +2570,10 @@ amdgpu_pci_shutdown(struct pci_dev *pdev) if (amdgpu_ras_intr_triggered()) return; + /* device maybe not resumed here, return immediately in this case */ + if (adev->in_s4 && adev->in_suspend) + return; + /* if we are running in a VM, make sure the device * torn down properly on reboot/shutdown. * unfortunately we can't detect certain @@ -2471,7 +2581,8 @@ amdgpu_pci_shutdown(struct pci_dev *pdev) */ if (!amdgpu_passthrough(adev)) adev->mp1_state = PP_MP1_STATE_UNLOAD; - amdgpu_device_ip_suspend(adev); + amdgpu_device_prepare(dev); + amdgpu_device_suspend(dev, true); adev->mp1_state = PP_MP1_STATE_NONE; } @@ -2480,11 +2591,14 @@ static int amdgpu_pmops_prepare(struct device *dev) struct drm_device *drm_dev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(drm_dev); + /* device maybe not resumed here, return immediately in this case */ + if (adev->in_s4 && adev->in_suspend) + return 0; + /* Return a positive number here so * DPM_FLAG_SMART_SUSPEND works properly */ - if (amdgpu_device_supports_boco(drm_dev) && - pm_runtime_suspended(dev)) + if (amdgpu_device_supports_boco(adev) && pm_runtime_suspended(dev)) return 1; /* if we will not support s3 or s2i for the device @@ -2499,7 +2613,7 @@ static int amdgpu_pmops_prepare(struct device *dev) static void amdgpu_pmops_complete(struct device *dev) { - /* nothing to do */ + amdgpu_device_complete(dev_get_drvdata(dev)); } static int amdgpu_pmops_suspend(struct device *dev) @@ -2511,8 +2625,24 @@ static int amdgpu_pmops_suspend(struct device *dev) adev->in_s0ix = true; else if (amdgpu_acpi_is_s3_active(adev)) adev->in_s3 = true; - if (!adev->in_s0ix && !adev->in_s3) + if (!adev->in_s0ix && !adev->in_s3) { +#if IS_ENABLED(CONFIG_SUSPEND) + /* don't allow going deep first time followed by s2idle the next time */ + if (adev->last_suspend_state != PM_SUSPEND_ON && + adev->last_suspend_state != pm_suspend_target_state) { + drm_err_once(drm_dev, "Unsupported suspend state %d\n", + pm_suspend_target_state); + return -EINVAL; + } +#endif return 0; + } + +#if IS_ENABLED(CONFIG_SUSPEND) + /* cache the state last used for suspend */ + adev->last_suspend_state = pm_suspend_target_state; +#endif + return amdgpu_device_suspend(drm_dev, true); } @@ -2520,9 +2650,14 @@ static int amdgpu_pmops_suspend_noirq(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(drm_dev); + int r; - if (amdgpu_acpi_should_gpu_reset(adev)) - return amdgpu_asic_reset(adev); + if (amdgpu_acpi_should_gpu_reset(adev)) { + amdgpu_device_lock_reset_domain(adev->reset_domain); + r = amdgpu_asic_reset(adev); + amdgpu_device_unlock_reset_domain(adev->reset_domain); + return r; + } return 0; } @@ -2558,26 +2693,36 @@ static int amdgpu_pmops_freeze(struct device *dev) if (r) return r; - if (amdgpu_acpi_should_gpu_reset(adev)) - return amdgpu_asic_reset(adev); + if (amdgpu_acpi_should_gpu_reset(adev)) { + amdgpu_device_lock_reset_domain(adev->reset_domain); + r = amdgpu_asic_reset(adev); + amdgpu_device_unlock_reset_domain(adev->reset_domain); + return r; + } return 0; } static int amdgpu_pmops_thaw(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); - struct amdgpu_device *adev = drm_to_adev(drm_dev); - int r; - r = amdgpu_device_resume(drm_dev, true); - adev->in_s4 = false; + /* do not resume device if it's normal hibernation */ + if (console_suspend_enabled && + !pm_hibernate_is_recovering() && + !pm_hibernation_mode_is_suspend()) + return 0; - return r; + return amdgpu_device_resume(drm_dev, true); } static int amdgpu_pmops_poweroff(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); + struct amdgpu_device *adev = drm_to_adev(drm_dev); + + /* device maybe not resumed here, return immediately in this case */ + if (adev->in_s4 && adev->in_suspend) + return 0; return amdgpu_device_suspend(drm_dev, true); } @@ -2585,9 +2730,6 @@ static int amdgpu_pmops_poweroff(struct device *dev) static int amdgpu_pmops_restore(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); - struct amdgpu_device *adev = drm_to_adev(drm_dev); - - adev->in_s4 = false; return amdgpu_device_resume(drm_dev, true); } @@ -2659,12 +2801,20 @@ static int amdgpu_runtime_idle_check_display(struct device *dev) return 0; } -static int amdgpu_pmops_runtime_suspend(struct device *dev) +static int amdgpu_runtime_idle_check_userq(struct device *dev) { struct pci_dev *pdev = to_pci_dev(dev); struct drm_device *drm_dev = pci_get_drvdata(pdev); struct amdgpu_device *adev = drm_to_adev(drm_dev); - int ret, i; + + return xa_empty(&adev->userq_doorbell_xa) ? 0 : -EBUSY; +} + +static int amdgpu_pmops_runtime_checks(struct device *dev) +{ + struct drm_device *drm_dev = dev_get_drvdata(dev); + struct amdgpu_device *adev = drm_to_adev(drm_dev); + int ret; if (adev->pm.rpm_mode == AMDGPU_RUNPM_NONE) { pm_runtime_forbid(dev); @@ -2675,6 +2825,29 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev) if (ret) return ret; + return amdgpu_runtime_idle_check_userq(dev); +} + +static int amdgpu_pmops_runtime_idle(struct device *dev) +{ + int ret; + + ret = amdgpu_pmops_runtime_checks(dev); + pm_runtime_autosuspend(dev); + return ret; +} + +static int amdgpu_pmops_runtime_suspend(struct device *dev) +{ + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + struct amdgpu_device *adev = drm_to_adev(drm_dev); + int ret, i; + + ret = amdgpu_pmops_runtime_checks(dev); + if (ret) + return ret; + /* wait for all rings to drain before suspending */ for (i = 0; i < AMDGPU_MAX_RINGS; i++) { struct amdgpu_ring *ring = adev->rings[i]; @@ -2727,7 +2900,7 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev) /* nothing to do */ } else if ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) || (adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO)) { - amdgpu_device_baco_enter(drm_dev); + amdgpu_device_baco_enter(adev); } dev_dbg(&pdev->dev, "asic/device is runtime suspended\n"); @@ -2768,7 +2941,7 @@ static int amdgpu_pmops_runtime_resume(struct device *dev) pci_set_master(pdev); } else if ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) || (adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO)) { - amdgpu_device_baco_exit(drm_dev); + amdgpu_device_baco_exit(adev); } ret = amdgpu_device_resume(drm_dev, false); if (ret) { @@ -2783,22 +2956,23 @@ static int amdgpu_pmops_runtime_resume(struct device *dev) return 0; } -static int amdgpu_pmops_runtime_idle(struct device *dev) +static int amdgpu_drm_release(struct inode *inode, struct file *filp) { - struct drm_device *drm_dev = dev_get_drvdata(dev); - struct amdgpu_device *adev = drm_to_adev(drm_dev); - int ret; - - if (adev->pm.rpm_mode == AMDGPU_RUNPM_NONE) { - pm_runtime_forbid(dev); - return -EBUSY; + struct drm_file *file_priv = filp->private_data; + struct amdgpu_fpriv *fpriv = file_priv->driver_priv; + struct drm_device *dev = file_priv->minor->dev; + int idx; + + if (fpriv && drm_dev_enter(dev, &idx)) { + amdgpu_evf_mgr_shutdown(&fpriv->evf_mgr); + amdgpu_userq_mgr_cancel_resume(&fpriv->userq_mgr); + amdgpu_evf_mgr_flush_suspend(&fpriv->evf_mgr); + amdgpu_userq_mgr_fini(&fpriv->userq_mgr); + amdgpu_evf_mgr_fini(&fpriv->evf_mgr); + drm_dev_exit(idx); } - ret = amdgpu_runtime_idle_check_display(dev); - - pm_runtime_mark_last_busy(dev); - pm_runtime_autosuspend(dev); - return ret; + return drm_release(inode, filp); } long amdgpu_drm_ioctl(struct file *filp, @@ -2815,22 +2989,21 @@ long amdgpu_drm_ioctl(struct file *filp, ret = drm_ioctl(filp, cmd, arg); - pm_runtime_mark_last_busy(dev->dev); out: pm_runtime_put_autosuspend(dev->dev); return ret; } static const struct dev_pm_ops amdgpu_pm_ops = { - .prepare = amdgpu_pmops_prepare, - .complete = amdgpu_pmops_complete, - .suspend = amdgpu_pmops_suspend, - .suspend_noirq = amdgpu_pmops_suspend_noirq, - .resume = amdgpu_pmops_resume, - .freeze = amdgpu_pmops_freeze, - .thaw = amdgpu_pmops_thaw, - .poweroff = amdgpu_pmops_poweroff, - .restore = amdgpu_pmops_restore, + .prepare = pm_sleep_ptr(amdgpu_pmops_prepare), + .complete = pm_sleep_ptr(amdgpu_pmops_complete), + .suspend = pm_sleep_ptr(amdgpu_pmops_suspend), + .suspend_noirq = pm_sleep_ptr(amdgpu_pmops_suspend_noirq), + .resume = pm_sleep_ptr(amdgpu_pmops_resume), + .freeze = pm_sleep_ptr(amdgpu_pmops_freeze), + .thaw = pm_sleep_ptr(amdgpu_pmops_thaw), + .poweroff = pm_sleep_ptr(amdgpu_pmops_poweroff), + .restore = pm_sleep_ptr(amdgpu_pmops_restore), .runtime_suspend = amdgpu_pmops_runtime_suspend, .runtime_resume = amdgpu_pmops_runtime_resume, .runtime_idle = amdgpu_pmops_runtime_idle, @@ -2852,7 +3025,7 @@ static const struct file_operations amdgpu_driver_kms_fops = { .owner = THIS_MODULE, .open = drm_open, .flush = amdgpu_flush, - .release = drm_release, + .release = amdgpu_drm_release, .unlocked_ioctl = amdgpu_drm_ioctl, .mmap = drm_gem_mmap, .poll = drm_poll, @@ -2899,6 +3072,10 @@ const struct drm_ioctl_desc amdgpu_ioctls_kms[] = { DRM_IOCTL_DEF_DRV(AMDGPU_GEM_VA, amdgpu_gem_va_ioctl, DRM_AUTH|DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(AMDGPU_GEM_OP, amdgpu_gem_op_ioctl, DRM_AUTH|DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(AMDGPU_GEM_USERPTR, amdgpu_gem_userptr_ioctl, DRM_AUTH|DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(AMDGPU_USERQ, amdgpu_userq_ioctl, DRM_AUTH|DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(AMDGPU_USERQ_SIGNAL, amdgpu_userq_signal_ioctl, DRM_AUTH|DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(AMDGPU_USERQ_WAIT, amdgpu_userq_wait_ioctl, DRM_AUTH|DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(AMDGPU_GEM_LIST_HANDLES, amdgpu_gem_list_handles_ioctl, DRM_AUTH|DRM_RENDER_ALLOW), }; static const struct drm_driver amdgpu_kms_driver = { @@ -2972,7 +3149,7 @@ static struct pci_driver amdgpu_kms_pci_driver = { .probe = amdgpu_pci_probe, .remove = amdgpu_pci_remove, .shutdown = amdgpu_pci_shutdown, - .driver.pm = &amdgpu_pm_ops, + .driver.pm = pm_ptr(&amdgpu_pm_ops), .err_handler = &amdgpu_pci_err_handler, .dev_groups = amdgpu_sysfs_groups, }; @@ -2981,23 +3158,20 @@ static int __init amdgpu_init(void) { int r; - if (drm_firmware_drivers_only()) - return -EINVAL; + /* Train lockdep on correct lock ordering */ + amdgpu_lockdep_init(); r = amdgpu_sync_init(); if (r) - goto error_sync; - - r = amdgpu_fence_slab_init(); - if (r) - goto error_fence; + return r; - DRM_INFO("amdgpu kernel modesetting enabled.\n"); amdgpu_register_atpx_handler(); amdgpu_acpi_detect(); - /* Ignore KFD init failures. Normal when CONFIG_HSA_AMD is not set. */ - amdgpu_amdkfd_init(); + /* Ignore KFD init failures when CONFIG_HSA_AMD is not set. */ + r = amdgpu_amdkfd_init(); + if (r && r != -ENOENT) + goto error_fini_sync; if (amdgpu_pp_feature_mask & PP_OVERDRIVE_MASK) { add_taint(TAINT_CPU_OUT_OF_SPEC, LOCKDEP_STILL_OK); @@ -3008,10 +3182,8 @@ static int __init amdgpu_init(void) /* let modprobe override vga console setting */ return pci_register_driver(&amdgpu_kms_pci_driver); -error_fence: +error_fini_sync: amdgpu_sync_fini(); - -error_sync: return r; } @@ -3022,9 +3194,16 @@ static void __exit amdgpu_exit(void) amdgpu_unregister_atpx_handler(); amdgpu_acpi_release(); amdgpu_sync_fini(); - amdgpu_fence_slab_fini(); mmu_notifier_synchronize(); amdgpu_xcp_drv_release(); + + /* + * Flush outstanding call_rcu() callbacks before the + * module text is freed. Otherwise a grace period elapsing after + * unload invokes a callback in already-freed module memory and + * faults in rcu_do_batch(). + */ + rcu_barrier(); } module_init(amdgpu_init); |