diff options
Diffstat (limited to 'drivers/gpu/drm/xe')
24 files changed, 229 insertions, 286 deletions
diff --git a/drivers/gpu/drm/xe/Makefile b/drivers/gpu/drm/xe/Makefile index 5a428ca00f10..c29a850859ad 100644 --- a/drivers/gpu/drm/xe/Makefile +++ b/drivers/gpu/drm/xe/Makefile @@ -172,8 +172,8 @@ subdir-ccflags-$(CONFIG_DRM_XE_DISPLAY) += \ -Ddrm_i915_gem_object=xe_bo \ -Ddrm_i915_private=xe_device -CFLAGS_i915-display/intel_fbdev.o = $(call cc-disable-warning, override-init) -CFLAGS_i915-display/intel_display_device.o = $(call cc-disable-warning, override-init) +CFLAGS_i915-display/intel_fbdev.o = -Wno-override-init +CFLAGS_i915-display/intel_display_device.o = -Wno-override-init # Rule to build SOC code shared with i915 $(obj)/i915-soc/%.o: $(srctree)/drivers/gpu/drm/i915/soc/%.c FORCE diff --git a/drivers/gpu/drm/xe/tests/xe_migrate.c b/drivers/gpu/drm/xe/tests/xe_migrate.c index a6523df0f1d3..c347e2c29f81 100644 --- a/drivers/gpu/drm/xe/tests/xe_migrate.c +++ b/drivers/gpu/drm/xe/tests/xe_migrate.c @@ -114,21 +114,21 @@ static void test_copy(struct xe_migrate *m, struct xe_bo *bo, region | XE_BO_NEEDS_CPU_ACCESS); if (IS_ERR(remote)) { - KUNIT_FAIL(test, "Failed to allocate remote bo for %s: %li\n", - str, PTR_ERR(remote)); + KUNIT_FAIL(test, "Failed to allocate remote bo for %s: %pe\n", + str, remote); return; } err = xe_bo_validate(remote, NULL, false); if (err) { - KUNIT_FAIL(test, "Failed to validate system bo for %s: %li\n", + KUNIT_FAIL(test, "Failed to validate system bo for %s: %i\n", str, err); goto out_unlock; } err = xe_bo_vmap(remote); if (err) { - KUNIT_FAIL(test, "Failed to vmap system bo for %s: %li\n", + KUNIT_FAIL(test, "Failed to vmap system bo for %s: %i\n", str, err); goto out_unlock; } diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c index 76dfaf1cd200..9c0837b6fdfc 100644 --- a/drivers/gpu/drm/xe/xe_bo.c +++ b/drivers/gpu/drm/xe/xe_bo.c @@ -144,9 +144,6 @@ static void try_add_system(struct xe_device *xe, struct xe_bo *bo, .mem_type = XE_PL_TT, }; *c += 1; - - if (bo->props.preferred_mem_type == XE_BO_PROPS_INVALID) - bo->props.preferred_mem_type = XE_PL_TT; } } @@ -181,25 +178,15 @@ static void add_vram(struct xe_device *xe, struct xe_bo *bo, } places[*c] = place; *c += 1; - - if (bo->props.preferred_mem_type == XE_BO_PROPS_INVALID) - bo->props.preferred_mem_type = mem_type; } static void try_add_vram(struct xe_device *xe, struct xe_bo *bo, u32 bo_flags, u32 *c) { - if (bo->props.preferred_gt == XE_GT1) { - if (bo_flags & XE_BO_CREATE_VRAM1_BIT) - add_vram(xe, bo, bo->placements, bo_flags, XE_PL_VRAM1, c); - if (bo_flags & XE_BO_CREATE_VRAM0_BIT) - add_vram(xe, bo, bo->placements, bo_flags, XE_PL_VRAM0, c); - } else { - if (bo_flags & XE_BO_CREATE_VRAM0_BIT) - add_vram(xe, bo, bo->placements, bo_flags, XE_PL_VRAM0, c); - if (bo_flags & XE_BO_CREATE_VRAM1_BIT) - add_vram(xe, bo, bo->placements, bo_flags, XE_PL_VRAM1, c); - } + if (bo_flags & XE_BO_CREATE_VRAM0_BIT) + add_vram(xe, bo, bo->placements, bo_flags, XE_PL_VRAM0, c); + if (bo_flags & XE_BO_CREATE_VRAM1_BIT) + add_vram(xe, bo, bo->placements, bo_flags, XE_PL_VRAM1, c); } static void try_add_stolen(struct xe_device *xe, struct xe_bo *bo, @@ -223,17 +210,8 @@ static int __xe_bo_placement_for_flags(struct xe_device *xe, struct xe_bo *bo, { u32 c = 0; - bo->props.preferred_mem_type = XE_BO_PROPS_INVALID; - - /* The order of placements should indicate preferred location */ - - if (bo->props.preferred_mem_class == DRM_XE_MEM_REGION_CLASS_SYSMEM) { - try_add_system(xe, bo, bo_flags, &c); - try_add_vram(xe, bo, bo_flags, &c); - } else { - try_add_vram(xe, bo, bo_flags, &c); - try_add_system(xe, bo, bo_flags, &c); - } + try_add_vram(xe, bo, bo_flags, &c); + try_add_system(xe, bo, bo_flags, &c); try_add_stolen(xe, bo, bo_flags, &c); if (!c) @@ -736,7 +714,7 @@ static int xe_bo_move(struct ttm_buffer_object *ttm_bo, bool evict, migrate = xe->tiles[0].migrate; xe_assert(xe, migrate); - trace_xe_bo_move(bo, new_mem->mem_type, old_mem_type); + trace_xe_bo_move(bo, new_mem->mem_type, old_mem_type, move_lacks_source); xe_device_mem_access_get(xe); if (xe_bo_is_pinned(bo) && !xe_bo_is_user(bo)) { @@ -1126,13 +1104,6 @@ static void xe_gem_object_close(struct drm_gem_object *obj, } } -static bool should_migrate_to_system(struct xe_bo *bo) -{ - struct xe_device *xe = xe_bo_device(bo); - - return xe_device_in_fault_mode(xe) && bo->props.cpu_atomic; -} - static vm_fault_t xe_gem_fault(struct vm_fault *vmf) { struct ttm_buffer_object *tbo = vmf->vma->vm_private_data; @@ -1141,7 +1112,7 @@ static vm_fault_t xe_gem_fault(struct vm_fault *vmf) struct xe_bo *bo = ttm_to_xe_bo(tbo); bool needs_rpm = bo->flags & XE_BO_CREATE_VRAM_MASK; vm_fault_t ret; - int idx, r = 0; + int idx; if (needs_rpm) xe_device_mem_access_get(xe); @@ -1153,17 +1124,8 @@ static vm_fault_t xe_gem_fault(struct vm_fault *vmf) if (drm_dev_enter(ddev, &idx)) { trace_xe_bo_cpu_fault(bo); - if (should_migrate_to_system(bo)) { - r = xe_bo_migrate(bo, XE_PL_TT); - if (r == -EBUSY || r == -ERESTARTSYS || r == -EINTR) - ret = VM_FAULT_NOPAGE; - else if (r) - ret = VM_FAULT_SIGBUS; - } - if (!ret) - ret = ttm_bo_vm_fault_reserved(vmf, - vmf->vma->vm_page_prot, - TTM_BO_VM_NUM_PREFAULT); + ret = ttm_bo_vm_fault_reserved(vmf, vmf->vma->vm_page_prot, + TTM_BO_VM_NUM_PREFAULT); drm_dev_exit(idx); } else { ret = ttm_bo_vm_dummy_page(vmf, vmf->vma->vm_page_prot); @@ -1291,9 +1253,6 @@ struct xe_bo *___xe_bo_create_locked(struct xe_device *xe, struct xe_bo *bo, bo->flags = flags; bo->cpu_caching = cpu_caching; bo->ttm.base.funcs = &xe_gem_object_funcs; - bo->props.preferred_mem_class = XE_BO_PROPS_INVALID; - bo->props.preferred_gt = XE_BO_PROPS_INVALID; - bo->props.preferred_mem_type = XE_BO_PROPS_INVALID; bo->ttm.priority = XE_BO_PRIORITY_NORMAL; INIT_LIST_HEAD(&bo->pinned_link); #ifdef CONFIG_PROC_FS diff --git a/drivers/gpu/drm/xe/xe_bo_types.h b/drivers/gpu/drm/xe/xe_bo_types.h index 14ef13b7b421..86422e113d39 100644 --- a/drivers/gpu/drm/xe/xe_bo_types.h +++ b/drivers/gpu/drm/xe/xe_bo_types.h @@ -56,25 +56,6 @@ struct xe_bo { */ struct list_head client_link; #endif - /** @props: BO user controlled properties */ - struct { - /** @preferred_mem: preferred memory class for this BO */ - s16 preferred_mem_class; - /** @prefered_gt: preferred GT for this BO */ - s16 preferred_gt; - /** @preferred_mem_type: preferred memory type */ - s32 preferred_mem_type; - /** - * @cpu_atomic: the CPU expects to do atomics operations to - * this BO - */ - bool cpu_atomic; - /** - * @device_atomic: the device expects to do atomics operations - * to this BO - */ - bool device_atomic; - } props; /** @freed: List node for delayed put. */ struct llist_node freed; /** @created: Whether the bo has passed initial creation */ diff --git a/drivers/gpu/drm/xe/xe_device.h b/drivers/gpu/drm/xe/xe_device.h index 14be34d9f543..d413bc2c6be5 100644 --- a/drivers/gpu/drm/xe/xe_device.h +++ b/drivers/gpu/drm/xe/xe_device.h @@ -58,7 +58,7 @@ static inline struct xe_tile *xe_device_get_root_tile(struct xe_device *xe) static inline struct xe_gt *xe_tile_get_gt(struct xe_tile *tile, u8 gt_id) { - if (drm_WARN_ON(&tile_to_xe(tile)->drm, gt_id > XE_MAX_GT_PER_TILE)) + if (drm_WARN_ON(&tile_to_xe(tile)->drm, gt_id >= XE_MAX_GT_PER_TILE)) gt_id = 0; return gt_id ? tile->media_gt : tile->primary_gt; @@ -79,7 +79,7 @@ static inline struct xe_gt *xe_device_get_gt(struct xe_device *xe, u8 gt_id) if (MEDIA_VER(xe) >= 13) { gt = xe_tile_get_gt(root_tile, gt_id); } else { - if (drm_WARN_ON(&xe->drm, gt_id > XE_MAX_TILES_PER_DEVICE)) + if (drm_WARN_ON(&xe->drm, gt_id >= XE_MAX_TILES_PER_DEVICE)) gt_id = 0; gt = xe->tiles[gt_id].primary_gt; diff --git a/drivers/gpu/drm/xe/xe_exec.c b/drivers/gpu/drm/xe/xe_exec.c index 952496c6260d..826c8b389672 100644 --- a/drivers/gpu/drm/xe/xe_exec.c +++ b/drivers/gpu/drm/xe/xe_exec.c @@ -235,6 +235,29 @@ retry: goto err_unlock_list; } + if (!args->num_batch_buffer) { + err = xe_vm_lock(vm, true); + if (err) + goto err_unlock_list; + + if (!xe_vm_in_lr_mode(vm)) { + struct dma_fence *fence; + + fence = xe_sync_in_fence_get(syncs, num_syncs, q, vm); + if (IS_ERR(fence)) { + err = PTR_ERR(fence); + goto err_unlock_list; + } + for (i = 0; i < num_syncs; i++) + xe_sync_entry_signal(&syncs[i], NULL, fence); + xe_exec_queue_last_fence_set(q, vm, fence); + dma_fence_put(fence); + } + + xe_vm_unlock(vm); + goto err_unlock_list; + } + vm_exec.vm = &vm->gpuvm; vm_exec.flags = DRM_EXEC_INTERRUPTIBLE_WAIT; if (xe_vm_in_lr_mode(vm)) { @@ -254,24 +277,6 @@ retry: goto err_exec; } - if (!args->num_batch_buffer) { - if (!xe_vm_in_lr_mode(vm)) { - struct dma_fence *fence; - - fence = xe_sync_in_fence_get(syncs, num_syncs, q, vm); - if (IS_ERR(fence)) { - err = PTR_ERR(fence); - goto err_exec; - } - for (i = 0; i < num_syncs; i++) - xe_sync_entry_signal(&syncs[i], NULL, fence); - xe_exec_queue_last_fence_set(q, vm, fence); - dma_fence_put(fence); - } - - goto err_exec; - } - if (xe_exec_queue_is_lr(q) && xe_exec_queue_ring_full(q)) { err = -EWOULDBLOCK; /* Aliased to -EAGAIN */ skip_retry = true; diff --git a/drivers/gpu/drm/xe/xe_exec_queue.c b/drivers/gpu/drm/xe/xe_exec_queue.c index 4bb8f897bf15..ead25d5e723e 100644 --- a/drivers/gpu/drm/xe/xe_exec_queue.c +++ b/drivers/gpu/drm/xe/xe_exec_queue.c @@ -354,91 +354,6 @@ static int exec_queue_set_timeslice(struct xe_device *xe, struct xe_exec_queue * return 0; } -static int exec_queue_set_preemption_timeout(struct xe_device *xe, - struct xe_exec_queue *q, u64 value, - bool create) -{ - u32 min = 0, max = 0; - - xe_exec_queue_get_prop_minmax(q->hwe->eclass, - XE_EXEC_QUEUE_PREEMPT_TIMEOUT, &min, &max); - - if (xe_exec_queue_enforce_schedule_limit() && - !xe_hw_engine_timeout_in_range(value, min, max)) - return -EINVAL; - - if (!create) - return q->ops->set_preempt_timeout(q, value); - - q->sched_props.preempt_timeout_us = value; - return 0; -} - -static int exec_queue_set_job_timeout(struct xe_device *xe, struct xe_exec_queue *q, - u64 value, bool create) -{ - u32 min = 0, max = 0; - - if (XE_IOCTL_DBG(xe, !create)) - return -EINVAL; - - xe_exec_queue_get_prop_minmax(q->hwe->eclass, - XE_EXEC_QUEUE_JOB_TIMEOUT, &min, &max); - - if (xe_exec_queue_enforce_schedule_limit() && - !xe_hw_engine_timeout_in_range(value, min, max)) - return -EINVAL; - - q->sched_props.job_timeout_ms = value; - - return 0; -} - -static int exec_queue_set_acc_trigger(struct xe_device *xe, struct xe_exec_queue *q, - u64 value, bool create) -{ - if (XE_IOCTL_DBG(xe, !create)) - return -EINVAL; - - if (XE_IOCTL_DBG(xe, !xe->info.has_usm)) - return -EINVAL; - - q->usm.acc_trigger = value; - - return 0; -} - -static int exec_queue_set_acc_notify(struct xe_device *xe, struct xe_exec_queue *q, - u64 value, bool create) -{ - if (XE_IOCTL_DBG(xe, !create)) - return -EINVAL; - - if (XE_IOCTL_DBG(xe, !xe->info.has_usm)) - return -EINVAL; - - q->usm.acc_notify = value; - - return 0; -} - -static int exec_queue_set_acc_granularity(struct xe_device *xe, struct xe_exec_queue *q, - u64 value, bool create) -{ - if (XE_IOCTL_DBG(xe, !create)) - return -EINVAL; - - if (XE_IOCTL_DBG(xe, !xe->info.has_usm)) - return -EINVAL; - - if (value > DRM_XE_ACC_GRANULARITY_64M) - return -EINVAL; - - q->usm.acc_granularity = value; - - return 0; -} - typedef int (*xe_exec_queue_set_property_fn)(struct xe_device *xe, struct xe_exec_queue *q, u64 value, bool create); @@ -446,11 +361,6 @@ typedef int (*xe_exec_queue_set_property_fn)(struct xe_device *xe, static const xe_exec_queue_set_property_fn exec_queue_set_property_funcs[] = { [DRM_XE_EXEC_QUEUE_SET_PROPERTY_PRIORITY] = exec_queue_set_priority, [DRM_XE_EXEC_QUEUE_SET_PROPERTY_TIMESLICE] = exec_queue_set_timeslice, - [DRM_XE_EXEC_QUEUE_SET_PROPERTY_PREEMPTION_TIMEOUT] = exec_queue_set_preemption_timeout, - [DRM_XE_EXEC_QUEUE_SET_PROPERTY_JOB_TIMEOUT] = exec_queue_set_job_timeout, - [DRM_XE_EXEC_QUEUE_SET_PROPERTY_ACC_TRIGGER] = exec_queue_set_acc_trigger, - [DRM_XE_EXEC_QUEUE_SET_PROPERTY_ACC_NOTIFY] = exec_queue_set_acc_notify, - [DRM_XE_EXEC_QUEUE_SET_PROPERTY_ACC_GRANULARITY] = exec_queue_set_acc_granularity, }; static int exec_queue_user_ext_set_property(struct xe_device *xe, @@ -469,7 +379,9 @@ static int exec_queue_user_ext_set_property(struct xe_device *xe, if (XE_IOCTL_DBG(xe, ext.property >= ARRAY_SIZE(exec_queue_set_property_funcs)) || - XE_IOCTL_DBG(xe, ext.pad)) + XE_IOCTL_DBG(xe, ext.pad) || + XE_IOCTL_DBG(xe, ext.property != DRM_XE_EXEC_QUEUE_SET_PROPERTY_PRIORITY && + ext.property != DRM_XE_EXEC_QUEUE_SET_PROPERTY_TIMESLICE)) return -EINVAL; idx = array_index_nospec(ext.property, ARRAY_SIZE(exec_queue_set_property_funcs)); @@ -536,7 +448,7 @@ find_hw_engine(struct xe_device *xe, { u32 idx; - if (eci.engine_class > ARRAY_SIZE(user_to_xe_engine_class)) + if (eci.engine_class >= ARRAY_SIZE(user_to_xe_engine_class)) return NULL; if (eci.gt_id >= xe->info.gt_count) diff --git a/drivers/gpu/drm/xe/xe_exec_queue_types.h b/drivers/gpu/drm/xe/xe_exec_queue_types.h index c40240e88068..62b3d9d1d7cd 100644 --- a/drivers/gpu/drm/xe/xe_exec_queue_types.h +++ b/drivers/gpu/drm/xe/xe_exec_queue_types.h @@ -141,16 +141,6 @@ struct xe_exec_queue { spinlock_t lock; } compute; - /** @usm: unified shared memory state */ - struct { - /** @usm.acc_trigger: access counter trigger */ - u32 acc_trigger; - /** @usm.acc_notify: access counter notify */ - u32 acc_notify; - /** @usm.acc_granularity: access counter granularity */ - u32 acc_granularity; - } usm; - /** @ops: submission backend exec queue operations */ const struct xe_exec_queue_ops *ops; diff --git a/drivers/gpu/drm/xe/xe_execlist.c b/drivers/gpu/drm/xe/xe_execlist.c index 1788e78caf5c..dece2785933c 100644 --- a/drivers/gpu/drm/xe/xe_execlist.c +++ b/drivers/gpu/drm/xe/xe_execlist.c @@ -212,7 +212,7 @@ static void xe_execlist_port_wake_locked(struct xe_execlist_port *port, static void xe_execlist_make_active(struct xe_execlist_exec_queue *exl) { struct xe_execlist_port *port = exl->port; - enum xe_exec_queue_priority priority = exl->active_priority; + enum xe_exec_queue_priority priority = exl->q->sched_props.priority; XE_WARN_ON(priority == XE_EXEC_QUEUE_PRIORITY_UNSET); XE_WARN_ON(priority < 0); diff --git a/drivers/gpu/drm/xe/xe_gt_pagefault.c b/drivers/gpu/drm/xe/xe_gt_pagefault.c index 73c535193a98..241c294270d9 100644 --- a/drivers/gpu/drm/xe/xe_gt_pagefault.c +++ b/drivers/gpu/drm/xe/xe_gt_pagefault.c @@ -69,7 +69,7 @@ static bool access_is_atomic(enum access_type access_type) static bool vma_is_valid(struct xe_tile *tile, struct xe_vma *vma) { return BIT(tile->id) & vma->tile_present && - !(BIT(tile->id) & vma->usm.tile_invalidated); + !(BIT(tile->id) & vma->tile_invalidated); } static bool vma_matches(struct xe_vma *vma, u64 page_addr) @@ -226,7 +226,7 @@ retry_userptr: if (xe_vma_is_userptr(vma)) ret = xe_vma_userptr_check_repin(to_userptr_vma(vma)); - vma->usm.tile_invalidated &= ~BIT(tile->id); + vma->tile_invalidated &= ~BIT(tile->id); unlock_dma_resv: drm_exec_fini(&exec); diff --git a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c index e3a4131ebb58..f03e077f81a0 100644 --- a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c +++ b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c @@ -248,6 +248,14 @@ int xe_gt_tlb_invalidation_vma(struct xe_gt *gt, xe_gt_assert(gt, vma); + /* Execlists not supported */ + if (gt_to_xe(gt)->info.force_execlist) { + if (fence) + __invalidation_fence_signal(fence); + + return 0; + } + action[len++] = XE_GUC_ACTION_TLB_INVALIDATION; action[len++] = 0; /* seqno, replaced in send_tlb_invalidation */ if (!xe->info.has_range_tlb_invalidation) { @@ -316,6 +324,10 @@ int xe_gt_tlb_invalidation_wait(struct xe_gt *gt, int seqno) struct xe_guc *guc = >->uc.guc; int ret; + /* Execlists not supported */ + if (gt_to_xe(gt)->info.force_execlist) + return 0; + /* * XXX: See above, this algorithm only works if seqno are always in * order diff --git a/drivers/gpu/drm/xe/xe_guc_submit.c b/drivers/gpu/drm/xe/xe_guc_submit.c index ff77bc8da1b2..e2a4c3b5e9ff 100644 --- a/drivers/gpu/drm/xe/xe_guc_submit.c +++ b/drivers/gpu/drm/xe/xe_guc_submit.c @@ -1220,7 +1220,7 @@ static int guc_exec_queue_init(struct xe_exec_queue *q) init_waitqueue_head(&ge->suspend_wait); timeout = (q->vm && xe_vm_in_lr_mode(q->vm)) ? MAX_SCHEDULE_TIMEOUT : - q->sched_props.job_timeout_ms; + msecs_to_jiffies(q->sched_props.job_timeout_ms); err = xe_sched_init(&ge->sched, &drm_sched_ops, &xe_sched_ops, get_submit_wq(guc), q->lrc[0].ring.size / MAX_JOB_SIZE_BYTES, 64, diff --git a/drivers/gpu/drm/xe/xe_lrc.c b/drivers/gpu/drm/xe/xe_lrc.c index 8c85e90220de..1426febe86eb 100644 --- a/drivers/gpu/drm/xe/xe_lrc.c +++ b/drivers/gpu/drm/xe/xe_lrc.c @@ -97,7 +97,6 @@ static void set_offsets(u32 *regs, #define REG16(x) \ (((x) >> 9) | BIT(7) | BUILD_BUG_ON_ZERO(x >= 0x10000)), \ (((x) >> 2) & 0x7f) -#define END 0 { const u32 base = hwe->mmio_base; @@ -168,7 +167,7 @@ static const u8 gen12_xcs_offsets[] = { REG16(0x274), REG16(0x270), - END + 0 }; static const u8 dg2_xcs_offsets[] = { @@ -202,7 +201,7 @@ static const u8 dg2_xcs_offsets[] = { REG16(0x274), REG16(0x270), - END + 0 }; static const u8 gen12_rcs_offsets[] = { @@ -298,7 +297,7 @@ static const u8 gen12_rcs_offsets[] = { REG(0x084), NOP(1), - END + 0 }; static const u8 xehp_rcs_offsets[] = { @@ -339,7 +338,7 @@ static const u8 xehp_rcs_offsets[] = { LRI(1, 0), REG(0x0c8), - END + 0 }; static const u8 dg2_rcs_offsets[] = { @@ -382,7 +381,7 @@ static const u8 dg2_rcs_offsets[] = { LRI(1, 0), REG(0x0c8), - END + 0 }; static const u8 mtl_rcs_offsets[] = { @@ -425,7 +424,7 @@ static const u8 mtl_rcs_offsets[] = { LRI(1, 0), REG(0x0c8), - END + 0 }; #define XE2_CTX_COMMON \ @@ -471,7 +470,7 @@ static const u8 xe2_rcs_offsets[] = { LRI(1, 0), /* [0x47] */ REG(0x0c8), /* [0x48] R_PWR_CLK_STATE */ - END + 0 }; static const u8 xe2_bcs_offsets[] = { @@ -482,16 +481,15 @@ static const u8 xe2_bcs_offsets[] = { REG16(0x200), /* [0x42] BCS_SWCTRL */ REG16(0x204), /* [0x44] BLIT_CCTL */ - END + 0 }; static const u8 xe2_xcs_offsets[] = { XE2_CTX_COMMON, - END + 0 }; -#undef END #undef REG16 #undef REG #undef LRI @@ -706,8 +704,6 @@ static void xe_lrc_set_ppgtt(struct xe_lrc *lrc, struct xe_vm *vm) #define PVC_CTX_ASID (0x2e + 1) #define PVC_CTX_ACC_CTR_THOLD (0x2a + 1) -#define ACC_GRANULARITY_S 20 -#define ACC_NOTIFY_S 16 int xe_lrc_init(struct xe_lrc *lrc, struct xe_hw_engine *hwe, struct xe_exec_queue *q, struct xe_vm *vm, u32 ring_size) @@ -778,13 +774,7 @@ int xe_lrc_init(struct xe_lrc *lrc, struct xe_hw_engine *hwe, xe_lrc_write_ctx_reg(lrc, CTX_RING_CTL, RING_CTL_SIZE(lrc->ring.size) | RING_VALID); if (xe->info.has_asid && vm) - xe_lrc_write_ctx_reg(lrc, PVC_CTX_ASID, - (q->usm.acc_granularity << - ACC_GRANULARITY_S) | vm->usm.asid); - if (xe->info.has_usm && vm) - xe_lrc_write_ctx_reg(lrc, PVC_CTX_ACC_CTR_THOLD, - (q->usm.acc_notify << ACC_NOTIFY_S) | - q->usm.acc_trigger); + xe_lrc_write_ctx_reg(lrc, PVC_CTX_ASID, vm->usm.asid); lrc->desc = LRC_VALID; lrc->desc |= LRC_LEGACY_64B_CONTEXT << LRC_ADDRESSING_MODE_SHIFT; diff --git a/drivers/gpu/drm/xe/xe_mmio.c b/drivers/gpu/drm/xe/xe_mmio.c index e3db3a178760..7ba2477452d7 100644 --- a/drivers/gpu/drm/xe/xe_mmio.c +++ b/drivers/gpu/drm/xe/xe_mmio.c @@ -106,7 +106,7 @@ static void xe_resize_vram_bar(struct xe_device *xe) pci_bus_for_each_resource(root, root_res, i) { if (root_res && root_res->flags & (IORESOURCE_MEM | IORESOURCE_MEM_64) && - root_res->start > 0x100000000ull) + (u64)root_res->start > 0x100000000ul) break; } diff --git a/drivers/gpu/drm/xe/xe_pm.c b/drivers/gpu/drm/xe/xe_pm.c index ab283e9a8b4e..53b3b0b019ac 100644 --- a/drivers/gpu/drm/xe/xe_pm.c +++ b/drivers/gpu/drm/xe/xe_pm.c @@ -354,7 +354,7 @@ int xe_pm_runtime_put(struct xe_device *xe) int xe_pm_runtime_get_if_active(struct xe_device *xe) { - return pm_runtime_get_if_active(xe->drm.dev, true); + return pm_runtime_get_if_active(xe->drm.dev); } void xe_pm_assert_unbounded_bridge(struct xe_device *xe) diff --git a/drivers/gpu/drm/xe/xe_query.c b/drivers/gpu/drm/xe/xe_query.c index 92bb06c0586e..075f9eaef031 100644 --- a/drivers/gpu/drm/xe/xe_query.c +++ b/drivers/gpu/drm/xe/xe_query.c @@ -132,7 +132,7 @@ query_engine_cycles(struct xe_device *xe, return -EINVAL; eci = &resp.eci; - if (eci->gt_id > XE_MAX_GT_PER_TILE) + if (eci->gt_id >= XE_MAX_GT_PER_TILE) return -EINVAL; gt = xe_device_get_gt(xe, eci->gt_id); diff --git a/drivers/gpu/drm/xe/xe_sync.c b/drivers/gpu/drm/xe/xe_sync.c index aab92bee1d7c..02c9577fe418 100644 --- a/drivers/gpu/drm/xe/xe_sync.c +++ b/drivers/gpu/drm/xe/xe_sync.c @@ -19,7 +19,7 @@ #include "xe_macros.h" #include "xe_sched_job_types.h" -struct user_fence { +struct xe_user_fence { struct xe_device *xe; struct kref refcount; struct dma_fence_cb cb; @@ -27,31 +27,32 @@ struct user_fence { struct mm_struct *mm; u64 __user *addr; u64 value; + int signalled; }; static void user_fence_destroy(struct kref *kref) { - struct user_fence *ufence = container_of(kref, struct user_fence, + struct xe_user_fence *ufence = container_of(kref, struct xe_user_fence, refcount); mmdrop(ufence->mm); kfree(ufence); } -static void user_fence_get(struct user_fence *ufence) +static void user_fence_get(struct xe_user_fence *ufence) { kref_get(&ufence->refcount); } -static void user_fence_put(struct user_fence *ufence) +static void user_fence_put(struct xe_user_fence *ufence) { kref_put(&ufence->refcount, user_fence_destroy); } -static struct user_fence *user_fence_create(struct xe_device *xe, u64 addr, - u64 value) +static struct xe_user_fence *user_fence_create(struct xe_device *xe, u64 addr, + u64 value) { - struct user_fence *ufence; + struct xe_user_fence *ufence; ufence = kmalloc(sizeof(*ufence), GFP_KERNEL); if (!ufence) @@ -69,7 +70,7 @@ static struct user_fence *user_fence_create(struct xe_device *xe, u64 addr, static void user_fence_worker(struct work_struct *w) { - struct user_fence *ufence = container_of(w, struct user_fence, worker); + struct xe_user_fence *ufence = container_of(w, struct xe_user_fence, worker); if (mmget_not_zero(ufence->mm)) { kthread_use_mm(ufence->mm); @@ -80,10 +81,11 @@ static void user_fence_worker(struct work_struct *w) } wake_up_all(&ufence->xe->ufence_wq); + WRITE_ONCE(ufence->signalled, 1); user_fence_put(ufence); } -static void kick_ufence(struct user_fence *ufence, struct dma_fence *fence) +static void kick_ufence(struct xe_user_fence *ufence, struct dma_fence *fence) { INIT_WORK(&ufence->worker, user_fence_worker); queue_work(ufence->xe->ordered_wq, &ufence->worker); @@ -92,7 +94,7 @@ static void kick_ufence(struct user_fence *ufence, struct dma_fence *fence) static void user_fence_cb(struct dma_fence *fence, struct dma_fence_cb *cb) { - struct user_fence *ufence = container_of(cb, struct user_fence, cb); + struct xe_user_fence *ufence = container_of(cb, struct xe_user_fence, cb); kick_ufence(ufence, fence); } @@ -340,3 +342,39 @@ err_out: return ERR_PTR(-ENOMEM); } + +/** + * xe_sync_ufence_get() - Get user fence from sync + * @sync: input sync + * + * Get a user fence reference from sync. + * + * Return: xe_user_fence pointer with reference + */ +struct xe_user_fence *xe_sync_ufence_get(struct xe_sync_entry *sync) +{ + user_fence_get(sync->ufence); + + return sync->ufence; +} + +/** + * xe_sync_ufence_put() - Put user fence reference + * @ufence: user fence reference + * + */ +void xe_sync_ufence_put(struct xe_user_fence *ufence) +{ + user_fence_put(ufence); +} + +/** + * xe_sync_ufence_get_status() - Get user fence status + * @ufence: user fence + * + * Return: 1 if signalled, 0 not signalled, <0 on error + */ +int xe_sync_ufence_get_status(struct xe_user_fence *ufence) +{ + return READ_ONCE(ufence->signalled); +} diff --git a/drivers/gpu/drm/xe/xe_sync.h b/drivers/gpu/drm/xe/xe_sync.h index f43cdcaca6c5..0fd0d51208e6 100644 --- a/drivers/gpu/drm/xe/xe_sync.h +++ b/drivers/gpu/drm/xe/xe_sync.h @@ -38,4 +38,8 @@ static inline bool xe_sync_is_ufence(struct xe_sync_entry *sync) return !!sync->ufence; } +struct xe_user_fence *xe_sync_ufence_get(struct xe_sync_entry *sync); +void xe_sync_ufence_put(struct xe_user_fence *ufence); +int xe_sync_ufence_get_status(struct xe_user_fence *ufence); + #endif diff --git a/drivers/gpu/drm/xe/xe_sync_types.h b/drivers/gpu/drm/xe/xe_sync_types.h index 852db5e7884f..30ac3f51993b 100644 --- a/drivers/gpu/drm/xe/xe_sync_types.h +++ b/drivers/gpu/drm/xe/xe_sync_types.h @@ -18,7 +18,7 @@ struct xe_sync_entry { struct drm_syncobj *syncobj; struct dma_fence *fence; struct dma_fence_chain *chain_fence; - struct user_fence *ufence; + struct xe_user_fence *ufence; u64 addr; u64 timeline_value; u32 type; diff --git a/drivers/gpu/drm/xe/xe_tile.c b/drivers/gpu/drm/xe/xe_tile.c index 044c20881de7..0650b2fa75ef 100644 --- a/drivers/gpu/drm/xe/xe_tile.c +++ b/drivers/gpu/drm/xe/xe_tile.c @@ -167,9 +167,10 @@ int xe_tile_init_noalloc(struct xe_tile *tile) goto err_mem_access; tile->mem.kernel_bb_pool = xe_sa_bo_manager_init(tile, SZ_1M, 16); - if (IS_ERR(tile->mem.kernel_bb_pool)) + if (IS_ERR(tile->mem.kernel_bb_pool)) { err = PTR_ERR(tile->mem.kernel_bb_pool); - + goto err_mem_access; + } xe_wa_apply_tile_workarounds(tile); xe_tile_sysfs_init(tile); diff --git a/drivers/gpu/drm/xe/xe_trace.h b/drivers/gpu/drm/xe/xe_trace.h index 3b97633d81d8..846f14507d5f 100644 --- a/drivers/gpu/drm/xe/xe_trace.h +++ b/drivers/gpu/drm/xe/xe_trace.h @@ -102,14 +102,16 @@ DEFINE_EVENT(xe_bo, xe_bo_cpu_fault, ); TRACE_EVENT(xe_bo_move, - TP_PROTO(struct xe_bo *bo, uint32_t new_placement, uint32_t old_placement), - TP_ARGS(bo, new_placement, old_placement), + TP_PROTO(struct xe_bo *bo, uint32_t new_placement, uint32_t old_placement, + bool move_lacks_source), + TP_ARGS(bo, new_placement, old_placement, move_lacks_source), TP_STRUCT__entry( __field(struct xe_bo *, bo) __field(size_t, size) __field(u32, new_placement) __field(u32, old_placement) __array(char, device_id, 12) + __field(bool, move_lacks_source) ), TP_fast_assign( @@ -118,9 +120,11 @@ TRACE_EVENT(xe_bo_move, __entry->new_placement = new_placement; __entry->old_placement = old_placement; strscpy(__entry->device_id, dev_name(xe_bo_device(__entry->bo)->drm.dev), 12); + __entry->move_lacks_source = move_lacks_source; ), - TP_printk("migrate object %p [size %zu] from %s to %s device_id:%s", - __entry->bo, __entry->size, xe_mem_type_to_name[__entry->old_placement], + TP_printk("move_lacks_source:%s, migrate object %p [size %zu] from %s to %s device_id:%s", + __entry->move_lacks_source ? "yes" : "no", __entry->bo, __entry->size, + xe_mem_type_to_name[__entry->old_placement], xe_mem_type_to_name[__entry->new_placement], __entry->device_id) ); @@ -464,7 +468,7 @@ DEFINE_EVENT(xe_vma, xe_vma_userptr_invalidate, TP_ARGS(vma) ); -DEFINE_EVENT(xe_vma, xe_vma_usm_invalidate, +DEFINE_EVENT(xe_vma, xe_vma_invalidate, TP_PROTO(struct xe_vma *vma), TP_ARGS(vma) ); diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c index e3bde897f6e8..f88faef4142b 100644 --- a/drivers/gpu/drm/xe/xe_vm.c +++ b/drivers/gpu/drm/xe/xe_vm.c @@ -708,6 +708,7 @@ int xe_vm_userptr_pin(struct xe_vm *vm) int err = 0; LIST_HEAD(tmp_evict); + xe_assert(vm->xe, !xe_vm_in_fault_mode(vm)); lockdep_assert_held_write(&vm->lock); /* Collect invalidated userptrs */ @@ -724,11 +725,27 @@ int xe_vm_userptr_pin(struct xe_vm *vm) list_for_each_entry_safe(uvma, next, &vm->userptr.repin_list, userptr.repin_link) { err = xe_vma_userptr_pin_pages(uvma); - if (err < 0) - return err; + if (err == -EFAULT) { + list_del_init(&uvma->userptr.repin_link); + + /* Wait for pending binds */ + xe_vm_lock(vm, false); + dma_resv_wait_timeout(xe_vm_resv(vm), + DMA_RESV_USAGE_BOOKKEEP, + false, MAX_SCHEDULE_TIMEOUT); + + err = xe_vm_invalidate_vma(&uvma->vma); + xe_vm_unlock(vm); + if (err) + return err; + } else { + if (err < 0) + return err; - list_del_init(&uvma->userptr.repin_link); - list_move_tail(&uvma->vma.combined_links.rebind, &vm->rebind_list); + list_del_init(&uvma->userptr.repin_link); + list_move_tail(&uvma->vma.combined_links.rebind, + &vm->rebind_list); + } } return 0; @@ -903,6 +920,11 @@ static void xe_vma_destroy_late(struct xe_vma *vma) struct xe_device *xe = vm->xe; bool read_only = xe_vma_read_only(vma); + if (vma->ufence) { + xe_sync_ufence_put(vma->ufence); + vma->ufence = NULL; + } + if (xe_vma_is_userptr(vma)) { struct xe_userptr *userptr = &to_userptr_vma(vma)->userptr; @@ -1622,6 +1644,16 @@ xe_vm_unbind_vma(struct xe_vma *vma, struct xe_exec_queue *q, trace_xe_vma_unbind(vma); + if (vma->ufence) { + struct xe_user_fence * const f = vma->ufence; + + if (!xe_sync_ufence_get_status(f)) + return ERR_PTR(-EBUSY); + + vma->ufence = NULL; + xe_sync_ufence_put(f); + } + if (number_tiles > 1) { fences = kmalloc_array(number_tiles, sizeof(*fences), GFP_KERNEL); @@ -1755,6 +1787,21 @@ err_fences: return ERR_PTR(err); } +static struct xe_user_fence * +find_ufence_get(struct xe_sync_entry *syncs, u32 num_syncs) +{ + unsigned int i; + + for (i = 0; i < num_syncs; i++) { + struct xe_sync_entry *e = &syncs[i]; + + if (xe_sync_is_ufence(e)) + return xe_sync_ufence_get(e); + } + + return NULL; +} + static int __xe_vm_bind(struct xe_vm *vm, struct xe_vma *vma, struct xe_exec_queue *q, struct xe_sync_entry *syncs, u32 num_syncs, bool immediate, bool first_op, @@ -1762,9 +1809,16 @@ static int __xe_vm_bind(struct xe_vm *vm, struct xe_vma *vma, { struct dma_fence *fence; struct xe_exec_queue *wait_exec_queue = to_wait_exec_queue(vm, q); + struct xe_user_fence *ufence; xe_vm_assert_held(vm); + ufence = find_ufence_get(syncs, num_syncs); + if (vma->ufence && ufence) + xe_sync_ufence_put(vma->ufence); + + vma->ufence = ufence ?: vma->ufence; + if (immediate) { fence = xe_vm_bind_vma(vma, q, syncs, num_syncs, first_op, last_op); @@ -1987,7 +2041,7 @@ static int xe_vm_prefetch(struct xe_vm *vm, struct xe_vma *vma, return err; } - if (vma->tile_mask != (vma->tile_present & ~vma->usm.tile_invalidated)) { + if (vma->tile_mask != (vma->tile_present & ~vma->tile_invalidated)) { return xe_vm_bind(vm, vma, q, xe_vma_bo(vma), syncs, num_syncs, true, first_op, last_op); } else { @@ -2131,10 +2185,6 @@ vm_bind_ioctl_ops_create(struct xe_vm *vm, struct xe_bo *bo, struct xe_vma_op *op = gpuva_op_to_vma_op(__op); if (__op->op == DRM_GPUVA_OP_MAP) { - op->map.immediate = - flags & DRM_XE_VM_BIND_FLAG_IMMEDIATE; - op->map.read_only = - flags & DRM_XE_VM_BIND_FLAG_READONLY; op->map.is_null = flags & DRM_XE_VM_BIND_FLAG_NULL; op->map.dumpable = flags & DRM_XE_VM_BIND_FLAG_DUMPABLE; op->map.pat_index = pat_index; @@ -2329,8 +2379,6 @@ static int vm_bind_ioctl_ops_parse(struct xe_vm *vm, struct xe_exec_queue *q, switch (op->base.op) { case DRM_GPUVA_OP_MAP: { - flags |= op->map.read_only ? - VMA_CREATE_FLAG_READ_ONLY : 0; flags |= op->map.is_null ? VMA_CREATE_FLAG_IS_NULL : 0; flags |= op->map.dumpable ? @@ -2475,7 +2523,7 @@ static int op_execute(struct drm_exec *exec, struct xe_vm *vm, case DRM_GPUVA_OP_MAP: err = xe_vm_bind(vm, vma, op->q, xe_vma_bo(vma), op->syncs, op->num_syncs, - op->map.immediate || !xe_vm_in_fault_mode(vm), + !xe_vm_in_fault_mode(vm), op->flags & XE_VMA_OP_FIRST, op->flags & XE_VMA_OP_LAST); break; @@ -2750,15 +2798,11 @@ static int vm_bind_ioctl_ops_execute(struct xe_vm *vm, return 0; } -#define SUPPORTED_FLAGS \ - (DRM_XE_VM_BIND_FLAG_READONLY | \ - DRM_XE_VM_BIND_FLAG_IMMEDIATE | DRM_XE_VM_BIND_FLAG_NULL | \ +#define SUPPORTED_FLAGS (DRM_XE_VM_BIND_FLAG_NULL | \ DRM_XE_VM_BIND_FLAG_DUMPABLE) #define XE_64K_PAGE_MASK 0xffffull #define ALL_DRM_XE_SYNCS_FLAGS (DRM_XE_SYNCS_FLAG_WAIT_FOR_OP) -#define MAX_BINDS 512 /* FIXME: Picking random upper limit */ - static int vm_bind_ioctl_check_args(struct xe_device *xe, struct drm_xe_vm_bind *args, struct drm_xe_vm_bind_op **bind_ops) @@ -2770,16 +2814,16 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe, XE_IOCTL_DBG(xe, args->reserved[0] || args->reserved[1])) return -EINVAL; - if (XE_IOCTL_DBG(xe, args->extensions) || - XE_IOCTL_DBG(xe, args->num_binds > MAX_BINDS)) + if (XE_IOCTL_DBG(xe, args->extensions)) return -EINVAL; if (args->num_binds > 1) { u64 __user *bind_user = u64_to_user_ptr(args->vector_of_binds); - *bind_ops = kmalloc(sizeof(struct drm_xe_vm_bind_op) * - args->num_binds, GFP_KERNEL); + *bind_ops = kvmalloc_array(args->num_binds, + sizeof(struct drm_xe_vm_bind_op), + GFP_KERNEL | __GFP_ACCOUNT); if (!*bind_ops) return -ENOMEM; @@ -2869,7 +2913,7 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe, free_bind_ops: if (args->num_binds > 1) - kfree(*bind_ops); + kvfree(*bind_ops); return err; } @@ -2957,13 +3001,15 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file) } if (args->num_binds) { - bos = kcalloc(args->num_binds, sizeof(*bos), GFP_KERNEL); + bos = kvcalloc(args->num_binds, sizeof(*bos), + GFP_KERNEL | __GFP_ACCOUNT); if (!bos) { err = -ENOMEM; goto release_vm_lock; } - ops = kcalloc(args->num_binds, sizeof(*ops), GFP_KERNEL); + ops = kvcalloc(args->num_binds, sizeof(*ops), + GFP_KERNEL | __GFP_ACCOUNT); if (!ops) { err = -ENOMEM; goto release_vm_lock; @@ -3104,10 +3150,10 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file) for (i = 0; bos && i < args->num_binds; ++i) xe_bo_put(bos[i]); - kfree(bos); - kfree(ops); + kvfree(bos); + kvfree(ops); if (args->num_binds > 1) - kfree(bind_ops); + kvfree(bind_ops); return err; @@ -3131,10 +3177,10 @@ put_exec_queue: if (q) xe_exec_queue_put(q); free_objs: - kfree(bos); - kfree(ops); + kvfree(bos); + kvfree(ops); if (args->num_binds > 1) - kfree(bind_ops); + kvfree(bind_ops); return err; } @@ -3185,9 +3231,8 @@ int xe_vm_invalidate_vma(struct xe_vma *vma) u8 id; int ret; - xe_assert(xe, xe_vm_in_fault_mode(xe_vma_vm(vma))); xe_assert(xe, !xe_vma_is_null(vma)); - trace_xe_vma_usm_invalidate(vma); + trace_xe_vma_invalidate(vma); /* Check that we don't race with page-table updates */ if (IS_ENABLED(CONFIG_PROVE_LOCKING)) { @@ -3225,7 +3270,7 @@ int xe_vm_invalidate_vma(struct xe_vma *vma) } } - vma->usm.tile_invalidated = vma->tile_mask; + vma->tile_invalidated = vma->tile_mask; return 0; } diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h index 292f8cadb40f..ae5fb565f6bf 100644 --- a/drivers/gpu/drm/xe/xe_vm_types.h +++ b/drivers/gpu/drm/xe/xe_vm_types.h @@ -19,6 +19,7 @@ struct xe_bo; struct xe_sync_entry; +struct xe_user_fence; struct xe_vm; #define XE_VMA_READ_ONLY DRM_GPUVA_USERBITS @@ -83,11 +84,8 @@ struct xe_vma { struct work_struct destroy_work; }; - /** @usm: unified shared memory state */ - struct { - /** @tile_invalidated: VMA has been invalidated */ - u8 tile_invalidated; - } usm; + /** @tile_invalidated: VMA has been invalidated */ + u8 tile_invalidated; /** @tile_mask: Tile mask of where to create binding for this VMA */ u8 tile_mask; @@ -105,6 +103,12 @@ struct xe_vma { * @pat_index: The pat index to use when encoding the PTEs for this vma. */ u16 pat_index; + + /** + * @ufence: The user fence that was provided with MAP. + * Needs to be signalled before UNMAP can be processed. + */ + struct xe_user_fence *ufence; }; /** @@ -270,10 +274,6 @@ struct xe_vm { struct xe_vma_op_map { /** @vma: VMA to map */ struct xe_vma *vma; - /** @immediate: Immediate bind */ - bool immediate; - /** @read_only: Read only */ - bool read_only; /** @is_null: is NULL binding */ bool is_null; /** @dumpable: whether BO is dumped on GPU hang */ diff --git a/drivers/gpu/drm/xe/xe_vram_freq.c b/drivers/gpu/drm/xe/xe_vram_freq.c index 079cc283a186..c5f6b5a5d117 100644 --- a/drivers/gpu/drm/xe/xe_vram_freq.c +++ b/drivers/gpu/drm/xe/xe_vram_freq.c @@ -111,8 +111,10 @@ void xe_vram_freq_sysfs_init(struct xe_tile *tile) return; kobj = kobject_create_and_add("memory", tile->sysfs); - if (!kobj) + if (!kobj) { drm_warn(&xe->drm, "failed to add memory directory, err: %d\n", -ENOMEM); + return; + } err = sysfs_create_group(kobj, &freq_group_attrs); if (err) { |