/* * Copyright 2008 Advanced Micro Devices, Inc. * Copyright 2008 Red Hat Inc. * Copyright 2009 Jerome Glisse. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be included in * all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. * * Authors: Dave Airlie * Alex Deucher * Jerome Glisse */#include <linux/fence-array.h>#include <drm/drmP.h>#include <drm/amdgpu_drm.h>#include "amdgpu.h"#include "amdgpu_trace.h" /* * GPUVM * GPUVM is similar to the legacy gart on older asics, however * rather than there being a single global gart table * for the entire GPU, there are multiple VM page tables active * at any given time. The VM page tables can contain a mix * vram pages and system memory pages and system memory pages * can be mapped as snooped (cached system pages) or unsnooped * (uncached system pages). * Each VM has an ID associated with it and there is a page table * associated with each VMID. When execting a command buffer, * the kernel tells the the ring what VMID to use for that command * buffer. VMIDs are allocated dynamically as commands are submitted. * The userspace drivers maintain their own address space and the kernel * sets up their pages tables accordingly when they submit their * command buffers and a VMID is assigned. * Cayman/Trinity support up to 8 active VMs at any given time; * SI supports 16. */ /* Special value that no flush is necessary */#define AMDGPU_VM_NO_FLUSH (~0ll) /* Local structure. Encapsulate some VM table update parameters to reduce * the number of function parameters */struct amdgpu_vm_update_params { /* address where to copy page table entries from */ uint64_t src; /* DMA addresses to use for mapping */ dma_addr_t *pages_addr; /* indirect buffer to fill with commands */ struct amdgpu_ib *ib;}; /** * amdgpu_vm_num_pde - return the number of page directory entries * * @adev: amdgpu_device pointer * * Calculate the number of page directory entries. */static unsigned amdgpu_vm_num_pdes(struct amdgpu_device *adev){ return adev->vm_manager.max_pfn >> amdgpu_vm_block_size;} /** * amdgpu_vm_directory_size - returns the size of the page directory in bytes * * @adev: amdgpu_device pointer * * Calculate the size of the page directory in bytes. */static unsigned amdgpu_vm_directory_size(struct amdgpu_device *adev){ return AMDGPU_GPU_PAGE_ALIGN(amdgpu_vm_num_pdes(adev) * 8);} /** * amdgpu_vm_get_pd_bo - add the VM PD to a validation list * * @vm: vm providing the BOs * @validated: head of validation list * @entry: entry to add * * Add the page directory to the list of BOs to * validate for command submission. */void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm, struct list_head *validated, struct amdgpu_bo_list_entry *entry){ entry->robj = vm->page_directory; entry->priority = 0; entry->tv.bo = &vm->page_directory->tbo; entry->tv.shared = true; entry->user_pages = NULL; list_add(&entry->tv.head, validated);} /** * amdgpu_vm_get_bos - add the vm BOs to a duplicates list * * @adev: amdgpu device pointer * @vm: vm providing the BOs * @duplicates: head of duplicates list * * Add the page directory to the BO duplicates list * for command submission. */void amdgpu_vm_get_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm, struct list_head *duplicates){ uint64_t num_evictions; unsigned i; /* We only need to validate the page tables * if they aren't already valid. */ num_evictions = atomic64_read(&adev->num_evictions); if (num_evictions == vm->last_eviction_counter) return; /* add the vm page table to the list */ for (i = 0; i <= vm->max_pde_used; ++i) { struct amdgpu_bo_list_entry *entry = &vm->page_tables[i].entry; if (!entry->robj) continue; list_add(&entry->tv.head, duplicates); } } /** * amdgpu_vm_move_pt_bos_in_lru - move the PT BOs to the LRU tail * * @adev: amdgpu device instance * @vm: vm providing the BOs * * Move the PT BOs to the tail of the LRU. */void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev, struct amdgpu_vm *vm){ struct ttm_bo_global *glob = adev->mman.bdev.glob; unsigned i; spin_lock(&glob->lru_lock); for (i = 0; i <= vm->max_pde_used; ++i) { struct amdgpu_bo_list_entry *entry = &vm->page_tables[i].entry; if (!entry->robj) continue; ttm_bo_move_to_lru_tail(&entry->robj->tbo); } spin_unlock(&glob->lru_lock);} static bool amdgpu_vm_is_gpu_reset(struct amdgpu_device *adev, struct amdgpu_vm_id *id){ return id->current_gpu_reset_count != atomic_read(&adev->gpu_reset_counter) ? true : false;} /** * amdgpu_vm_grab_id - allocate the next free VMID * * @vm: vm to allocate id for * @ring: ring we want to submit job to * @sync: sync object where we add dependencies * @fence: fence protecting ID from reuse * * Allocate an id for the vm, adding fences to the sync obj as necessary. */int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, struct amdgpu_sync *sync, struct fence *fence, struct amdgpu_job *job){ struct amdgpu_device *adev = ring->adev; uint64_t fence_context = adev->fence_context + ring->idx; struct fence *updates = sync->last_vm_update; struct amdgpu_vm_id *id, *idle; struct fence **fences; unsigned i; int r = 0; fences = kmalloc_array(sizeof(void *), adev->vm_manager.num_ids, GFP_KERNEL); if (!fences) return -ENOMEM; mutex_lock(&adev->vm_manager.lock); /* Check if we have an idle VMID */ i = 0; list_for_each_entry(idle, &adev->vm_manager.ids_lru, list) { fences[i] = amdgpu_sync_peek_fence(&idle->active, ring); if (!fences[i]) break; ++i; } /* If we can't find a idle VMID to use, wait till one becomes available */ if (&idle->list == &adev->vm_manager.ids_lru) { u64 fence_context = adev->vm_manager.fence_context + ring->idx; unsigned seqno = ++adev->vm_manager.seqno[ring->idx]; struct fence_array *array; unsigned j; for (j = 0; j < i; ++j) fence_get(fences[j]); array = fence_array_create(i, fences, fence_context, seqno, true); if (!array) { for (j = 0; j < i; ++j) fence_put(fences[j]); kfree(fences); r = -ENOMEM; goto error; } r = amdgpu_sync_fence(ring->adev, sync, &array->base); fence_put(&array->base); if (r) goto error; mutex_unlock(&adev->vm_manager.lock); return 0; } kfree(fences); job->vm_needs_flush = true; /* Check if we can use a VMID already assigned to this VM */ i = ring->idx; do { struct fence *flushed; id = vm->ids[i++]; if (i == AMDGPU_MAX_RINGS) i = 0; /* Check all the prerequisites to using this VMID */ if (!id) continue; if (amdgpu_vm_is_gpu_reset(adev, id)) continue; if (atomic64_read(&id->owner) != vm->client_id) continue; if (job->vm_pd_addr != id->pd_gpu_addr) continue; if (!id->last_flush) continue; if (id->last_flush->context != fence_context && !fence_is_signaled(id->last_flush)) continue; flushed = id->flushed_updates; if (updates && (!flushed || fence_is_later(updates, flushed))) continue; /* Good we can use this VMID. Remember this submission as * user of the VMID. */ r = amdgpu_sync_fence(ring->adev, &id->active, fence); if (r) goto error; id->current_gpu_reset_count = atomic_read(&adev->gpu_reset_counter); list_move_tail(&id->list, &adev->vm_manager.ids_lru); vm->ids[ring->idx] = id; job->vm_id = id - adev->vm_manager.ids; job->vm_needs_flush = false; trace_amdgpu_vm_grab_id(vm, ring->idx, job); mutex_unlock(&adev->vm_manager.lock); return 0; } while (i != ring->idx); /* Still no ID to use? Then use the idle one found earlier */ id = idle; /* Remember this submission as user of the VMID */ r = amdgpu_sync_fence(ring->adev, &id->active, fence); if (r) goto error; fence_put(id->first); id->first = fence_get(fence); fence_put(id->last_flush); id->last_flush = NULL; fence_put(id->flushed_updates); id->flushed_updates = fence_get(updates); id->pd_gpu_addr = job->vm_pd_addr; id->current_gpu_reset_count = atomic_read(&adev->gpu_reset_counter); list_move_tail(&id->list, &adev->vm_manager.ids_lru); atomic64_set(&id->owner, vm->client_id); vm->ids[ring->idx] = id; job->vm_id = id - adev->vm_manager.ids; trace_amdgpu_vm_grab_id(vm, ring->idx, job); error: mutex_unlock(&adev->vm_manager.lock); return r;} static bool amdgpu_vm_ring_has_compute_vm_bug(struct amdgpu_ring *ring){ struct amdgpu_device *adev = ring->adev; const struct amdgpu_ip_block_version *ip_block; if (ring->type != AMDGPU_RING_TYPE_COMPUTE) /* only compute rings */ return false; ip_block = amdgpu_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX); if (!ip_block) return false; if (ip_block->major <= 7) { /* gfx7 has no workaround */ return true; } else if (ip_block->major == 8) { if (adev->gfx.mec_fw_version >= 673) /* gfx8 is fixed in MEC firmware 673 */ return false; else return true; } return false;} /** * amdgpu_vm_flush - hardware flush the vm * * @ring: ring to use for flush * @vm_id: vmid number to use * @pd_addr: address of the page directory * * Emit a VM flush when it is necessary. */int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job){ struct amdgpu_device *adev = ring->adev; struct amdgpu_vm_id *id = &adev->vm_manager.ids[job->vm_id]; bool gds_switch_needed = ring->funcs->emit_gds_switch && ( id->gds_base != job->gds_base || id->gds_size != job->gds_size || id->gws_base != job->gws_base || id->gws_size != job->gws_size || id->oa_base != job->oa_base || id->oa_size != job->oa_size); int r; if (ring->funcs->emit_pipeline_sync && ( job->vm_needs_flush || gds_switch_needed || amdgpu_vm_ring_has_compute_vm_bug(ring))) amdgpu_ring_emit_pipeline_sync(ring); if (ring->funcs->emit_vm_flush && (job->vm_needs_flush || amdgpu_vm_is_gpu_reset(adev, id))) { struct fence *fence; trace_amdgpu_vm_flush(job->vm_pd_addr, ring->idx, job->vm_id); amdgpu_ring_emit_vm_flush(ring, job->vm_id, job->vm_pd_addr); r = amdgpu_fence_emit(ring, &fence); if (r) return r; mutex_lock(&adev->vm_manager.lock); fence_put(id->last_flush); id->last_flush = fence; mutex_unlock(&adev->vm_manager.lock); } if (gds_switch_needed) { id->gds_base = job->gds_base; id->gds_size = job->gds_size; id->gws_base = job->gws_base; id->gws_size = job->gws_size; id->oa_base = job->oa_base; id->oa_size = job->oa_size; amdgpu_ring_emit_gds_switch(ring, job->vm_id, job->gds_base, job->gds_size, job->gws_base, job->gws_size, job->oa_base, job->oa_size); } return 0;} /** * amdgpu_vm_reset_id - reset VMID to zero * * @adev: amdgpu device structure * @vm_id: vmid number to use * * Reset saved GDW, GWS and OA to force switch on next flush. */void amdgpu_vm_reset_id(struct amdgpu_device *adev, unsigned vm_id){ struct amdgpu_vm_id *id = &adev->vm_manager.ids[vm_id]; id->gds_base = 0; id->gds_size = 0; id->gws_base = 0; id->gws_size = 0; id->oa_base = 0; id->oa_size = 0;} /** * amdgpu_vm_bo_find - find the bo_va for a specific vm & bo * * @vm: requested vm * @bo: requested buffer object * * Find @bo inside the requested vm. * Search inside the @bos vm list for the requested vm * Returns the found bo_va or NULL if none is found * * Object has to be reserved! */struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm, struct amdgpu_bo *bo){ struct amdgpu_bo_va *bo_va; list_for_each_entry(bo_va, &bo->va, bo_list) { if (bo_va->vm == vm) { return bo_va; } } return NULL;} /** * amdgpu_vm_update_pages - helper to call the right asic function * * @adev: amdgpu_device pointer * @vm_update_params: see amdgpu_vm_update_params definition * @pe: addr of the page entry * @addr: dst addr to write into pe * @count: number of page entries to update * @incr: increase next addr by incr bytes * @flags: hw access flags * * Traces the parameters and calls the right asic functions * to setup the page table using the DMA. */static void amdgpu_vm_update_pages(struct amdgpu_device *adev, struct amdgpu_vm_update_params *vm_update_params, uint64_t pe, uint64_t addr, unsigned count, uint32_t incr, uint32_t flags){ trace_amdgpu_vm_set_page(pe, addr, count, incr, flags); if (vm_update_params->src) { amdgpu_vm_copy_pte(adev, vm_update_params->ib, pe, (vm_update_params->src + (addr >> 12) * 8), count); } else if (vm_update_params->pages_addr) { amdgpu_vm_write_pte(adev, vm_update_params->ib, vm_update_params->pages_addr, pe, addr, count, incr, flags); Lines 1–500 of 1658. Load more
/* * Copyright 2008 Advanced Micro Devices, Inc. * Copyright 2008 Red Hat Inc. * Copyright 2009 Jerome Glisse. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be included in * all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. * * Authors: Dave Airlie * Alex Deucher * Jerome Glisse */ #include <linux/fence-array.h> #include <drm/drmP.h> #include <drm/amdgpu_drm.h> #include "amdgpu.h" #include "amdgpu_trace.h" /* * GPUVM * GPUVM is similar to the legacy gart on older asics, however * rather than there being a single global gart table * for the entire GPU, there are multiple VM page tables active * at any given time. The VM page tables can contain a mix * vram pages and system memory pages and system memory pages * can be mapped as snooped (cached system pages) or unsnooped * (uncached system pages). * Each VM has an ID associated with it and there is a page table * associated with each VMID. When execting a command buffer, * the kernel tells the the ring what VMID to use for that command * buffer. VMIDs are allocated dynamically as commands are submitted. * The userspace drivers maintain their own address space and the kernel * sets up their pages tables accordingly when they submit their * command buffers and a VMID is assigned. * Cayman/Trinity support up to 8 active VMs at any given time; * SI supports 16. */ /* Special value that no flush is necessary */ #define AMDGPU_VM_NO_FLUSH (~0ll) /* Local structure. Encapsulate some VM table update parameters to reduce * the number of function parameters */ struct amdgpu_vm_update_params { /* address where to copy page table entries from */ uint64_t src; /* DMA addresses to use for mapping */ dma_addr_t *pages_addr; /* indirect buffer to fill with commands */ struct amdgpu_ib *ib; }; /** * amdgpu_vm_num_pde - return the number of page directory entries * * @adev: amdgpu_device pointer * * Calculate the number of page directory entries. */ static unsigned amdgpu_vm_num_pdes(struct amdgpu_device *adev) { return adev->vm_manager.max_pfn >> amdgpu_vm_block_size; } /** * amdgpu_vm_directory_size - returns the size of the page directory in bytes * * @adev: amdgpu_device pointer * * Calculate the size of the page directory in bytes. */ static unsigned amdgpu_vm_directory_size(struct amdgpu_device *adev) { return AMDGPU_GPU_PAGE_ALIGN(amdgpu_vm_num_pdes(adev) * 8); } /** * amdgpu_vm_get_pd_bo - add the VM PD to a validation list * * @vm: vm providing the BOs * @validated: head of validation list * @entry: entry to add * * Add the page directory to the list of BOs to * validate for command submission. */ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm, struct list_head *validated, struct amdgpu_bo_list_entry *entry) { entry->robj = vm->page_directory; entry->priority = 0; entry->tv.bo = &vm->page_directory->tbo; entry->tv.shared = true; entry->user_pages = NULL; list_add(&entry->tv.head, validated); } /** * amdgpu_vm_get_bos - add the vm BOs to a duplicates list * * @adev: amdgpu device pointer * @vm: vm providing the BOs * @duplicates: head of duplicates list * * Add the page directory to the BO duplicates list * for command submission. */ void amdgpu_vm_get_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm, struct list_head *duplicates) { uint64_t num_evictions; unsigned i; /* We only need to validate the page tables * if they aren't already valid. */ num_evictions = atomic64_read(&adev->num_evictions); if (num_evictions == vm->last_eviction_counter) return; /* add the vm page table to the list */ for (i = 0; i <= vm->max_pde_used; ++i) { struct amdgpu_bo_list_entry *entry = &vm->page_tables[i].entry; if (!entry->robj) continue; list_add(&entry->tv.head, duplicates); } } /** * amdgpu_vm_move_pt_bos_in_lru - move the PT BOs to the LRU tail * * @adev: amdgpu device instance * @vm: vm providing the BOs * * Move the PT BOs to the tail of the LRU. */ void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev, struct amdgpu_vm *vm) { struct ttm_bo_global *glob = adev->mman.bdev.glob; unsigned i; spin_lock(&glob->lru_lock); for (i = 0; i <= vm->max_pde_used; ++i) { struct amdgpu_bo_list_entry *entry = &vm->page_tables[i].entry; if (!entry->robj) continue; ttm_bo_move_to_lru_tail(&entry->robj->tbo); } spin_unlock(&glob->lru_lock); } static bool amdgpu_vm_is_gpu_reset(struct amdgpu_device *adev, struct amdgpu_vm_id *id) { return id->current_gpu_reset_count != atomic_read(&adev->gpu_reset_counter) ? true : false; } /** * amdgpu_vm_grab_id - allocate the next free VMID * * @vm: vm to allocate id for * @ring: ring we want to submit job to * @sync: sync object where we add dependencies * @fence: fence protecting ID from reuse * * Allocate an id for the vm, adding fences to the sync obj as necessary. */ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, struct amdgpu_sync *sync, struct fence *fence, struct amdgpu_job *job) { struct amdgpu_device *adev = ring->adev; uint64_t fence_context = adev->fence_context + ring->idx; struct fence *updates = sync->last_vm_update; struct amdgpu_vm_id *id, *idle; struct fence **fences; unsigned i; int r = 0; fences = kmalloc_array(sizeof(void *), adev->vm_manager.num_ids, GFP_KERNEL); if (!fences) return -ENOMEM; mutex_lock(&adev->vm_manager.lock); /* Check if we have an idle VMID */ i = 0; list_for_each_entry(idle, &adev->vm_manager.ids_lru, list) { fences[i] = amdgpu_sync_peek_fence(&idle->active, ring); if (!fences[i]) break; ++i; } /* If we can't find a idle VMID to use, wait till one becomes available */ if (&idle->list == &adev->vm_manager.ids_lru) { u64 fence_context = adev->vm_manager.fence_context + ring->idx; unsigned seqno = ++adev->vm_manager.seqno[ring->idx]; struct fence_array *array; unsigned j; for (j = 0; j < i; ++j) fence_get(fences[j]); array = fence_array_create(i, fences, fence_context, seqno, true); if (!array) { for (j = 0; j < i; ++j) fence_put(fences[j]); kfree(fences); r = -ENOMEM; goto error; } r = amdgpu_sync_fence(ring->adev, sync, &array->base); fence_put(&array->base); if (r) goto error; mutex_unlock(&adev->vm_manager.lock); return 0; } kfree(fences); job->vm_needs_flush = true; /* Check if we can use a VMID already assigned to this VM */ i = ring->idx; do { struct fence *flushed; id = vm->ids[i++]; if (i == AMDGPU_MAX_RINGS) i = 0; /* Check all the prerequisites to using this VMID */ if (!id) continue; if (amdgpu_vm_is_gpu_reset(adev, id)) continue; if (atomic64_read(&id->owner) != vm->client_id) continue; if (job->vm_pd_addr != id->pd_gpu_addr) continue; if (!id->last_flush) continue; if (id->last_flush->context != fence_context && !fence_is_signaled(id->last_flush)) continue; flushed = id->flushed_updates; if (updates && (!flushed || fence_is_later(updates, flushed))) continue; /* Good we can use this VMID. Remember this submission as * user of the VMID. */ r = amdgpu_sync_fence(ring->adev, &id->active, fence); if (r) goto error; id->current_gpu_reset_count = atomic_read(&adev->gpu_reset_counter); list_move_tail(&id->list, &adev->vm_manager.ids_lru); vm->ids[ring->idx] = id; job->vm_id = id - adev->vm_manager.ids; job->vm_needs_flush = false; trace_amdgpu_vm_grab_id(vm, ring->idx, job); mutex_unlock(&adev->vm_manager.lock); return 0; } while (i != ring->idx); /* Still no ID to use? Then use the idle one found earlier */ id = idle; /* Remember this submission as user of the VMID */ r = amdgpu_sync_fence(ring->adev, &id->active, fence); if (r) goto error; fence_put(id->first); id->first = fence_get(fence); fence_put(id->last_flush); id->last_flush = NULL; fence_put(id->flushed_updates); id->flushed_updates = fence_get(updates); id->pd_gpu_addr = job->vm_pd_addr; id->current_gpu_reset_count = atomic_read(&adev->gpu_reset_counter); list_move_tail(&id->list, &adev->vm_manager.ids_lru); atomic64_set(&id->owner, vm->client_id); vm->ids[ring->idx] = id; job->vm_id = id - adev->vm_manager.ids; trace_amdgpu_vm_grab_id(vm, ring->idx, job); error: mutex_unlock(&adev->vm_manager.lock); return r; } static bool amdgpu_vm_ring_has_compute_vm_bug(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; const struct amdgpu_ip_block_version *ip_block; if (ring->type != AMDGPU_RING_TYPE_COMPUTE) /* only compute rings */ return false; ip_block = amdgpu_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX); if (!ip_block) return false; if (ip_block->major <= 7) { /* gfx7 has no workaround */ return true; } else if (ip_block->major == 8) { if (adev->gfx.mec_fw_version >= 673) /* gfx8 is fixed in MEC firmware 673 */ return false; else return true; } return false; } /** * amdgpu_vm_flush - hardware flush the vm * * @ring: ring to use for flush * @vm_id: vmid number to use * @pd_addr: address of the page directory * * Emit a VM flush when it is necessary. */ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job) { struct amdgpu_device *adev = ring->adev; struct amdgpu_vm_id *id = &adev->vm_manager.ids[job->vm_id]; bool gds_switch_needed = ring->funcs->emit_gds_switch && ( id->gds_base != job->gds_base || id->gds_size != job->gds_size || id->gws_base != job->gws_base || id->gws_size != job->gws_size || id->oa_base != job->oa_base || id->oa_size != job->oa_size); int r; if (ring->funcs->emit_pipeline_sync && ( job->vm_needs_flush || gds_switch_needed || amdgpu_vm_ring_has_compute_vm_bug(ring))) amdgpu_ring_emit_pipeline_sync(ring); if (ring->funcs->emit_vm_flush && (job->vm_needs_flush || amdgpu_vm_is_gpu_reset(adev, id))) { struct fence *fence; trace_amdgpu_vm_flush(job->vm_pd_addr, ring->idx, job->vm_id); amdgpu_ring_emit_vm_flush(ring, job->vm_id, job->vm_pd_addr); r = amdgpu_fence_emit(ring, &fence); if (r) return r; mutex_lock(&adev->vm_manager.lock); fence_put(id->last_flush); id->last_flush = fence; mutex_unlock(&adev->vm_manager.lock); } if (gds_switch_needed) { id->gds_base = job->gds_base; id->gds_size = job->gds_size; id->gws_base = job->gws_base; id->gws_size = job->gws_size; id->oa_base = job->oa_base; id->oa_size = job->oa_size; amdgpu_ring_emit_gds_switch(ring, job->vm_id, job->gds_base, job->gds_size, job->gws_base, job->gws_size, job->oa_base, job->oa_size); } return 0; } /** * amdgpu_vm_reset_id - reset VMID to zero * * @adev: amdgpu device structure * @vm_id: vmid number to use * * Reset saved GDW, GWS and OA to force switch on next flush. */ void amdgpu_vm_reset_id(struct amdgpu_device *adev, unsigned vm_id) { struct amdgpu_vm_id *id = &adev->vm_manager.ids[vm_id]; id->gds_base = 0; id->gds_size = 0; id->gws_base = 0; id->gws_size = 0; id->oa_base = 0; id->oa_size = 0; } /** * amdgpu_vm_bo_find - find the bo_va for a specific vm & bo * * @vm: requested vm * @bo: requested buffer object * * Find @bo inside the requested vm. * Search inside the @bos vm list for the requested vm * Returns the found bo_va or NULL if none is found * * Object has to be reserved! */ struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm, struct amdgpu_bo *bo) { struct amdgpu_bo_va *bo_va; list_for_each_entry(bo_va, &bo->va, bo_list) { if (bo_va->vm == vm) { return bo_va; } } return NULL; } /** * amdgpu_vm_update_pages - helper to call the right asic function * * @adev: amdgpu_device pointer * @vm_update_params: see amdgpu_vm_update_params definition * @pe: addr of the page entry * @addr: dst addr to write into pe * @count: number of page entries to update * @incr: increase next addr by incr bytes * @flags: hw access flags * * Traces the parameters and calls the right asic functions * to setup the page table using the DMA. */ static void amdgpu_vm_update_pages(struct amdgpu_device *adev, struct amdgpu_vm_update_params *vm_update_params, uint64_t pe, uint64_t addr, unsigned count, uint32_t incr, uint32_t flags) { trace_amdgpu_vm_set_page(pe, addr, count, incr, flags); if (vm_update_params->src) { amdgpu_vm_copy_pte(adev, vm_update_params->ib, pe, (vm_update_params->src + (addr >> 12) * 8), count); } else if (vm_update_params->pages_addr) { amdgpu_vm_write_pte(adev, vm_update_params->ib, vm_update_params->pages_addr, pe, addr, count, incr, flags);