[PATCH v4] drm/amdkfd: Use partial migrations in GPU page faults

Chen, Xiaogang xiaogang.chen at amd.com
Wed Oct 4 21:47:53 UTC 2023


On 10/4/2023 1:47 PM, Felix Kuehling wrote:
>
> On 2023-10-03 19:31, Xiaogang.Chen wrote:
>> From: Xiaogang Chen <xiaogang.chen at amd.com>
>>
>> This patch implements partial migration in gpu page fault according 
>> to migration
>> granularity(default 2MB) and not split svm range in cpu page fault 
>> handling.
>> A svm range may include pages from both system ram and vram of one 
>> gpu now.
>> These chagnes are expected to improve migration performance and 
>> reduce mmu
>> callback and TLB flush workloads.
>>
>> Signed-off-by: Xiaogang Chen<Xiaogang.Chen at amd.com>
>
> Minor (mostly cosemtic) nit-picks inline. With those fixed, the patch is
>
> Reviewed-by: Felix Kuehling <Felix.Kuehling at amd.com>
>
Thanks for the review. These indentations was due to my editor on Linux 
that it does not show some special characters correctly. I changed these 
by vi.

I need use a different editor now.

Regards

Xiaogang

>
>> ---
>>   drivers/gpu/drm/amd/amdkfd/kfd_migrate.c | 156 +++++++++++++----------
>>   drivers/gpu/drm/amd/amdkfd/kfd_migrate.h |   6 +-
>>   drivers/gpu/drm/amd/amdkfd/kfd_svm.c     |  83 +++++++++---
>>   drivers/gpu/drm/amd/amdkfd/kfd_svm.h     |   6 +-
>>   4 files changed, 162 insertions(+), 89 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c 
>> b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
>> index 6c25dab051d5..6a059e4aff86 100644
>> --- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
>> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
>> @@ -442,10 +442,10 @@ svm_migrate_vma_to_vram(struct kfd_node *node, 
>> struct svm_range *prange,
>>           goto out_free;
>>       }
>>       if (cpages != npages)
>> -        pr_debug("partial migration, 0x%lx/0x%llx pages migrated\n",
>> +        pr_debug("partial migration, 0x%lx/0x%llx pages collected\n",
>>                cpages, npages);
>>       else
>> -        pr_debug("0x%lx pages migrated\n", cpages);
>> +        pr_debug("0x%lx pages collected\n", cpages);
>>         r = svm_migrate_copy_to_vram(node, prange, &migrate, &mfence, 
>> scratch, ttm_res_offset);
>>       migrate_vma_pages(&migrate);
>> @@ -479,6 +479,8 @@ svm_migrate_vma_to_vram(struct kfd_node *node, 
>> struct svm_range *prange,
>>    * svm_migrate_ram_to_vram - migrate svm range from system to device
>>    * @prange: range structure
>>    * @best_loc: the device to migrate to
>> + * @start_mgr: start page to migrate
>> + * @last_mgr: last page to migrate
>>    * @mm: the process mm structure
>>    * @trigger: reason of migration
>>    *
>> @@ -489,6 +491,7 @@ svm_migrate_vma_to_vram(struct kfd_node *node, 
>> struct svm_range *prange,
>>    */
>>   static int
>>   svm_migrate_ram_to_vram(struct svm_range *prange, uint32_t best_loc,
>> +            unsigned long start_mgr, unsigned long last_mgr,
>>               struct mm_struct *mm, uint32_t trigger)
>>   {
>>       unsigned long addr, start, end;
>> @@ -498,23 +501,30 @@ svm_migrate_ram_to_vram(struct svm_range 
>> *prange, uint32_t best_loc,
>>       unsigned long cpages = 0;
>>       long r = 0;
>>   -    if (prange->actual_loc == best_loc) {
>> -        pr_debug("svms 0x%p [0x%lx 0x%lx] already on best_loc 0x%x\n",
>> -             prange->svms, prange->start, prange->last, best_loc);
>> +    if (!best_loc) {
>> +        pr_debug("svms 0x%p [0x%lx 0x%lx] migrate to sys ram\n",
>> +            prange->svms, start_mgr, last_mgr);
>>           return 0;
>>       }
>>   +    if (start_mgr < prange->start || last_mgr > prange->last) {
>> +        pr_debug("range [0x%lx 0x%lx] out prange [0x%lx 0x%lx]\n",
>> +                 start_mgr, last_mgr, prange->start, prange->last);
>> +        return -EFAULT;
>> +    }
>> +
>>       node = svm_range_get_node_by_id(prange, best_loc);
>>       if (!node) {
>>           pr_debug("failed to get kfd node by id 0x%x\n", best_loc);
>>           return -ENODEV;
>>       }
>>   -    pr_debug("svms 0x%p [0x%lx 0x%lx] to gpu 0x%x\n", prange->svms,
>> -         prange->start, prange->last, best_loc);
>> +    pr_debug("svms 0x%p [0x%lx 0x%lx] in [0x%lx 0x%lx] to gpu 0x%x\n",
>> +        prange->svms, start_mgr, last_mgr, prange->start, prange->last,
>> +        best_loc);
>>   -    start = prange->start << PAGE_SHIFT;
>> -    end = (prange->last + 1) << PAGE_SHIFT;
>> +    start = start_mgr << PAGE_SHIFT;
>> +    end = (last_mgr + 1) << PAGE_SHIFT;
>>         r = svm_range_vram_node_new(node, prange, true);
>>       if (r) {
>> @@ -544,8 +554,11 @@ svm_migrate_ram_to_vram(struct svm_range 
>> *prange, uint32_t best_loc,
>>         if (cpages) {
>>           prange->actual_loc = best_loc;
>> -        svm_range_dma_unmap(prange);
>> -    } else {
>> +        prange->vram_pages = prange->vram_pages + cpages;
>> +    } else if (!prange->actual_loc) {
>> +        /* if no page migrated and all pages from prange are at
>> +         * sys ram drop svm_bo got from svm_range_vram_node_new
>> +         */
>>           svm_range_vram_node_free(prange);
>>       }
>>   @@ -663,19 +676,19 @@ svm_migrate_copy_to_ram(struct amdgpu_device 
>> *adev, struct svm_range *prange,
>>    * Context: Process context, caller hold mmap read lock, 
>> prange->migrate_mutex
>>    *
>>    * Return:
>> - *   0 - success with all pages migrated
>>    *   negative values - indicate error
>> - *   positive values - partial migration, number of pages not migrated
>> + *   positive values or zero - number of pages got migrated
>>    */
>>   static long
>>   svm_migrate_vma_to_ram(struct kfd_node *node, struct svm_range 
>> *prange,
>> -               struct vm_area_struct *vma, uint64_t start, uint64_t 
>> end,
>> -               uint32_t trigger, struct page *fault_page)
>> +            struct vm_area_struct *vma, uint64_t start, uint64_t end,
>> +            uint32_t trigger, struct page *fault_page)
>
> This indentation change is unnecessary and for the worse. Alignment 
> with the open parenthesis is preferred.
>
>
>>   {
>>       struct kfd_process *p = container_of(prange->svms, struct 
>> kfd_process, svms);
>>       uint64_t npages = (end - start) >> PAGE_SHIFT;
>>       unsigned long upages = npages;
>>       unsigned long cpages = 0;
>> +    unsigned long mpages = 0;
>>       struct amdgpu_device *adev = node->adev;
>>       struct kfd_process_device *pdd;
>>       struct dma_fence *mfence = NULL;
>> @@ -725,10 +738,10 @@ svm_migrate_vma_to_ram(struct kfd_node *node, 
>> struct svm_range *prange,
>>           goto out_free;
>>       }
>>       if (cpages != npages)
>> -        pr_debug("partial migration, 0x%lx/0x%llx pages migrated\n",
>> +        pr_debug("partial migration, 0x%lx/0x%llx pages collected\n",
>>                cpages, npages);
>>       else
>> -        pr_debug("0x%lx pages migrated\n", cpages);
>> +        pr_debug("0x%lx pages collected\n", cpages);
>>         r = svm_migrate_copy_to_ram(adev, prange, &migrate, &mfence,
>>                       scratch, npages);
>> @@ -751,17 +764,21 @@ svm_migrate_vma_to_ram(struct kfd_node *node, 
>> struct svm_range *prange,
>>       kvfree(buf);
>>   out:
>>       if (!r && cpages) {
>> +        mpages = cpages - upages;
>>           pdd = svm_range_get_pdd_by_node(prange, node);
>>           if (pdd)
>> -            WRITE_ONCE(pdd->page_out, pdd->page_out + cpages);
>> +            WRITE_ONCE(pdd->page_out, pdd->page_out + mpages);
>>       }
>> -    return r ? r : upages;
>> +
>> +    return r ? r : mpages;
>>   }
>>     /**
>>    * svm_migrate_vram_to_ram - migrate svm range from device to system
>>    * @prange: range structure
>>    * @mm: process mm, use current->mm if NULL
>> + * @start_mgr: start page need be migrated to sys ram
>> + * @last_mgr: last page need be migrated to sys ram
>>    * @trigger: reason of migration
>>    * @fault_page: is from vmf->page, svm_migrate_to_ram(), this is 
>> CPU page fault callback
>>    *
>> @@ -771,33 +788,41 @@ svm_migrate_vma_to_ram(struct kfd_node *node, 
>> struct svm_range *prange,
>>    * 0 - OK, otherwise error code
>>    */
>>   int svm_migrate_vram_to_ram(struct svm_range *prange, struct 
>> mm_struct *mm,
>> -                uint32_t trigger, struct page *fault_page)
>> +            unsigned long start_mgr, unsigned long last_mgr,
>> +            uint32_t trigger, struct page *fault_page)
>>   {
>>       struct kfd_node *node;
>>       struct vm_area_struct *vma;
>>       unsigned long addr;
>>       unsigned long start;
>>       unsigned long end;
>> -    unsigned long upages = 0;
>> +    unsigned long mpages = 0;
>>       long r = 0;
>>   +    /* this pragne has no any vram page to migrate to sys ram */
>>       if (!prange->actual_loc) {
>>           pr_debug("[0x%lx 0x%lx] already migrated to ram\n",
>>                prange->start, prange->last);
>>           return 0;
>>       }
>>   +    if (start_mgr < prange->start || last_mgr > prange->last) {
>> +        pr_debug("range [0x%lx 0x%lx] out prange [0x%lx 0x%lx]\n",
>> +                 start_mgr, last_mgr, prange->start, prange->last);
>> +        return -EFAULT;
>> +    }
>> +
>>       node = svm_range_get_node_by_id(prange, prange->actual_loc);
>>       if (!node) {
>>           pr_debug("failed to get kfd node by id 0x%x\n", 
>> prange->actual_loc);
>>           return -ENODEV;
>>       }
>>       pr_debug("svms 0x%p prange 0x%p [0x%lx 0x%lx] from gpu 0x%x to 
>> ram\n",
>> -         prange->svms, prange, prange->start, prange->last,
>> +         prange->svms, prange, start_mgr, last_mgr,
>>            prange->actual_loc);
>>   -    start = prange->start << PAGE_SHIFT;
>> -    end = (prange->last + 1) << PAGE_SHIFT;
>> +    start = start_mgr << PAGE_SHIFT;
>> +    end = (last_mgr + 1) << PAGE_SHIFT;
>>         for (addr = start; addr < end;) {
>>           unsigned long next;
>> @@ -816,14 +841,21 @@ int svm_migrate_vram_to_ram(struct svm_range 
>> *prange, struct mm_struct *mm,
>>               pr_debug("failed %ld to migrate prange %p\n", r, prange);
>>               break;
>>           } else {
>> -            upages += r;
>> +            mpages += r;
>>           }
>>           addr = next;
>>       }
>>   -    if (r >= 0 && !upages) {
>> -        svm_range_vram_node_free(prange);
>> -        prange->actual_loc = 0;
>> +    if (r >= 0) {
>> +        prange->vram_pages -= mpages;
>> +
>> +        /* prange does not have vram page set its actual_loc to system
>> +         * and drop its svm_bo ref
>> +         */
>> +        if (prange->vram_pages == 0 && prange->ttm_res) {
>> +            prange->actual_loc = 0;
>> +            svm_range_vram_node_free(prange);
>> +        }
>>       }
>>         return r < 0 ? r : 0;
>> @@ -833,17 +865,23 @@ int svm_migrate_vram_to_ram(struct svm_range 
>> *prange, struct mm_struct *mm,
>>    * svm_migrate_vram_to_vram - migrate svm range from device to device
>>    * @prange: range structure
>>    * @best_loc: the device to migrate to
>> + * @start: start page need be migrated to sys ram
>> + * @last: last page need be migrated to sys ram
>>    * @mm: process mm, use current->mm if NULL
>>    * @trigger: reason of migration
>>    *
>>    * Context: Process context, caller hold mmap read lock, svms lock, 
>> prange lock
>>    *
>> + * migrate all vram pages in prange to sys ram, then migrate
>> + * [start, last] pages from sys ram to gpu node best_loc.
>> + *
>>    * Return:
>>    * 0 - OK, otherwise error code
>>    */
>>   static int
>>   svm_migrate_vram_to_vram(struct svm_range *prange, uint32_t best_loc,
>> -             struct mm_struct *mm, uint32_t trigger)
>> +            unsigned long start, unsigned long last,
>> +            struct mm_struct *mm, uint32_t trigger)
>>   {
>>       int r, retries = 3;
>>   @@ -855,7 +893,8 @@ svm_migrate_vram_to_vram(struct svm_range 
>> *prange, uint32_t best_loc,
>>       pr_debug("from gpu 0x%x to gpu 0x%x\n", prange->actual_loc, 
>> best_loc);
>>         do {
>> -        r = svm_migrate_vram_to_ram(prange, mm, trigger, NULL);
>> +        r = svm_migrate_vram_to_ram(prange, mm, prange->start, 
>> prange->last,
>> +                                trigger, NULL);
>>           if (r)
>>               return r;
>>       } while (prange->actual_loc && --retries);
>> @@ -863,17 +902,21 @@ svm_migrate_vram_to_vram(struct svm_range 
>> *prange, uint32_t best_loc,
>>       if (prange->actual_loc)
>>           return -EDEADLK;
>>   -    return svm_migrate_ram_to_vram(prange, best_loc, mm, trigger);
>> +    return svm_migrate_ram_to_vram(prange, best_loc, start, last, 
>> mm, trigger);
>>   }
>>     int
>>   svm_migrate_to_vram(struct svm_range *prange, uint32_t best_loc,
>> -            struct mm_struct *mm, uint32_t trigger)
>> +            unsigned long start, unsigned long last,
>> +            struct mm_struct *mm, uint32_t trigger)
>>   {
>> -    if  (!prange->actual_loc)
>> -        return svm_migrate_ram_to_vram(prange, best_loc, mm, trigger);
>> +    if  (!prange->actual_loc || prange->actual_loc == best_loc)
>> +        return svm_migrate_ram_to_vram(prange, best_loc, start, last,
>> +                        mm, trigger);
>> +
>>       else
>> -        return svm_migrate_vram_to_vram(prange, best_loc, mm, trigger);
>> +        return svm_migrate_vram_to_vram(prange, best_loc, start, last,
>> +                        mm, trigger);
>>     }
>>   @@ -889,10 +932,9 @@ svm_migrate_to_vram(struct svm_range *prange, 
>> uint32_t best_loc,
>>    */
>>   static vm_fault_t svm_migrate_to_ram(struct vm_fault *vmf)
>>   {
>> +    unsigned long start, last, size;
>>       unsigned long addr = vmf->address;
>>       struct svm_range_bo *svm_bo;
>> -    enum svm_work_list_ops op;
>> -    struct svm_range *parent;
>>       struct svm_range *prange;
>>       struct kfd_process *p;
>>       struct mm_struct *mm;
>> @@ -929,51 +971,31 @@ static vm_fault_t svm_migrate_to_ram(struct 
>> vm_fault *vmf)
>>         mutex_lock(&p->svms.lock);
>>   -    prange = svm_range_from_addr(&p->svms, addr, &parent);
>> +    prange = svm_range_from_addr(&p->svms, addr, NULL);
>>       if (!prange) {
>>           pr_debug("failed get range svms 0x%p addr 0x%lx\n", 
>> &p->svms, addr);
>>           r = -EFAULT;
>>           goto out_unlock_svms;
>>       }
>>   -    mutex_lock(&parent->migrate_mutex);
>> -    if (prange != parent)
>> -        mutex_lock_nested(&prange->migrate_mutex, 1);
>> +    mutex_lock(&prange->migrate_mutex);
>>         if (!prange->actual_loc)
>>           goto out_unlock_prange;
>>   -    svm_range_lock(parent);
>> -    if (prange != parent)
>> -        mutex_lock_nested(&prange->lock, 1);
>> -    r = svm_range_split_by_granularity(p, mm, addr, parent, prange);
>> -    if (prange != parent)
>> -        mutex_unlock(&prange->lock);
>> -    svm_range_unlock(parent);
>> -    if (r) {
>> -        pr_debug("failed %d to split range by granularity\n", r);
>> -        goto out_unlock_prange;
>> -    }
>> +    /* Align migration range start and size to granularity size */
>> +    size = 1UL << prange->granularity;
>> +    start = max(ALIGN_DOWN(addr, size), prange->start);
>> +    last = min(ALIGN(addr + 1, size) - 1, prange->last);
>>   -    r = svm_migrate_vram_to_ram(prange, vmf->vma->vm_mm,
>> -                    KFD_MIGRATE_TRIGGER_PAGEFAULT_CPU,
>> -                    vmf->page);
>> +    r = svm_migrate_vram_to_ram(prange, vmf->vma->vm_mm, start, last,
>> +                        KFD_MIGRATE_TRIGGER_PAGEFAULT_CPU, vmf->page);
>>       if (r)
>>           pr_debug("failed %d migrate svms 0x%p range 0x%p [0x%lx 
>> 0x%lx]\n",
>> -             r, prange->svms, prange, prange->start, prange->last);
>> -
>> -    /* xnack on, update mapping on GPUs with ACCESS_IN_PLACE */
>> -    if (p->xnack_enabled && parent == prange)
>> -        op = SVM_OP_UPDATE_RANGE_NOTIFIER_AND_MAP;
>> -    else
>> -        op = SVM_OP_UPDATE_RANGE_NOTIFIER;
>> -    svm_range_add_list_work(&p->svms, parent, mm, op);
>> -    schedule_deferred_list_work(&p->svms);
>> +            r, prange->svms, prange, start, last);
>>     out_unlock_prange:
>> -    if (prange != parent)
>> -        mutex_unlock(&prange->migrate_mutex);
>> -    mutex_unlock(&parent->migrate_mutex);
>> +    mutex_unlock(&prange->migrate_mutex);
>>   out_unlock_svms:
>>       mutex_unlock(&p->svms.lock);
>>   out_unref_process:
>> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h 
>> b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h
>> index 487f26368164..9e48d10e848e 100644
>> --- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h
>> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h
>> @@ -41,9 +41,13 @@ enum MIGRATION_COPY_DIR {
>>   };
>>     int svm_migrate_to_vram(struct svm_range *prange,  uint32_t 
>> best_loc,
>> +            unsigned long start, unsigned long last,
>>               struct mm_struct *mm, uint32_t trigger);
>> +
>>   int svm_migrate_vram_to_ram(struct svm_range *prange, struct 
>> mm_struct *mm,
>> -                uint32_t trigger, struct page *fault_page);
>> +            unsigned long start, unsigned long last,
>> +            uint32_t trigger, struct page *fault_page);
>> +
>
> Indentation: Alignment with the open parenthesis is preferred.
>
>
>>   unsigned long
>>   svm_migrate_addr_to_pfn(struct amdgpu_device *adev, unsigned long 
>> addr);
>>   diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c 
>> b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
>> index 649f9685e1ec..5a15866b52ee 100644
>> --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
>> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
>> @@ -156,7 +156,7 @@ svm_is_valid_dma_mapping_addr(struct device *dev, 
>> dma_addr_t dma_addr)
>>   static int
>>   svm_range_dma_map_dev(struct amdgpu_device *adev, struct svm_range 
>> *prange,
>>                 unsigned long offset, unsigned long npages,
>> -              unsigned long *hmm_pfns, uint32_t gpuidx)
>> +              unsigned long *hmm_pfns, uint32_t gpuidx, uint64_t 
>> *vram_pages)
>>   {
>>       enum dma_data_direction dir = DMA_BIDIRECTIONAL;
>>       dma_addr_t *addr = prange->dma_addr[gpuidx];
>> @@ -171,6 +171,7 @@ svm_range_dma_map_dev(struct amdgpu_device *adev, 
>> struct svm_range *prange,
>>           prange->dma_addr[gpuidx] = addr;
>>       }
>>   +    *vram_pages = 0;
>
> I'd prefer to use a local variable here and only assign to the pointer 
> in the end. That can be optimized much better by the compiler (e.g. by 
> allocating the local variable in a register).
>
>
>>       addr += offset;
>>       for (i = 0; i < npages; i++) {
>>           if (svm_is_valid_dma_mapping_addr(dev, addr[i]))
>> @@ -180,6 +181,7 @@ svm_range_dma_map_dev(struct amdgpu_device *adev, 
>> struct svm_range *prange,
>>           if (is_zone_device_page(page)) {
>>               struct amdgpu_device *bo_adev = 
>> prange->svm_bo->node->adev;
>>   +            (*vram_pages)++;
>
> Local variable.
>
>
>>               addr[i] = (hmm_pfns[i] << PAGE_SHIFT) +
>>                      bo_adev->vm_manager.vram_base_offset -
>>                      bo_adev->kfd.pgmap.range.start;
>> @@ -202,7 +204,7 @@ svm_range_dma_map_dev(struct amdgpu_device *adev, 
>> struct svm_range *prange,
>>   static int
>>   svm_range_dma_map(struct svm_range *prange, unsigned long *bitmap,
>>             unsigned long offset, unsigned long npages,
>> -          unsigned long *hmm_pfns)
>> +          unsigned long *hmm_pfns, uint64_t *vram_pages)
>>   {
>>       struct kfd_process *p;
>>       uint32_t gpuidx;
>> @@ -221,7 +223,7 @@ svm_range_dma_map(struct svm_range *prange, 
>> unsigned long *bitmap,
>>           }
>>             r = svm_range_dma_map_dev(pdd->dev->adev, prange, offset, 
>> npages,
>> -                      hmm_pfns, gpuidx);
>> +                      hmm_pfns, gpuidx, vram_pages);
>>           if (r)
>>               break;
>>       }
>> @@ -347,6 +349,7 @@ svm_range *svm_range_new(struct svm_range_list 
>> *svms, uint64_t start,
>>       INIT_LIST_HEAD(&prange->child_list);
>>       atomic_set(&prange->invalid, 0);
>>       prange->validate_timestamp = 0;
>> +    prange->vram_pages = 0;
>>       mutex_init(&prange->migrate_mutex);
>>       mutex_init(&prange->lock);
>>   @@ -393,6 +396,8 @@ static void svm_range_bo_release(struct kref 
>> *kref)
>>                prange->start, prange->last);
>>           mutex_lock(&prange->lock);
>>           prange->svm_bo = NULL;
>> +        /* prange should not hold vram page now */
>> +        WARN_ON(prange->actual_loc);
>>           mutex_unlock(&prange->lock);
>>             spin_lock(&svm_bo->list_lock);
>> @@ -973,6 +978,11 @@ svm_range_split_nodes(struct svm_range *new, 
>> struct svm_range *old,
>>       new->svm_bo = svm_range_bo_ref(old->svm_bo);
>>       new->ttm_res = old->ttm_res;
>>   +    /* set new's vram_pages as old range's now, the acurate 
>> vram_pages
>> +     * will be updated during mapping
>> +     */
>> +    new->vram_pages = min(old->vram_pages, new->npages);
>> +
>>       spin_lock(&new->svm_bo->list_lock);
>>       list_add(&new->svm_bo_list, &new->svm_bo->range_list);
>>       spin_unlock(&new->svm_bo->list_lock);
>> @@ -1617,6 +1627,7 @@ static int svm_range_validate_and_map(struct 
>> mm_struct *mm,
>>       struct svm_validate_context *ctx;
>>       unsigned long start, end, addr;
>>       struct kfd_process *p;
>> +    uint64_t vram_pages;
>>       void *owner;
>>       int32_t idx;
>>       int r = 0;
>> @@ -1685,11 +1696,13 @@ static int svm_range_validate_and_map(struct 
>> mm_struct *mm,
>>           }
>>       }
>>   +    vram_pages = 0;
>>       start = prange->start << PAGE_SHIFT;
>>       end = (prange->last + 1) << PAGE_SHIFT;
>>       for (addr = start; !r && addr < end; ) {
>>           struct hmm_range *hmm_range;
>>           struct vm_area_struct *vma;
>> +        uint64_t vram_pages_vma;
>>           unsigned long next = 0;
>>           unsigned long offset;
>>           unsigned long npages;
>> @@ -1718,9 +1731,11 @@ static int svm_range_validate_and_map(struct 
>> mm_struct *mm,
>>           if (!r) {
>>               offset = (addr - start) >> PAGE_SHIFT;
>>               r = svm_range_dma_map(prange, ctx->bitmap, offset, npages,
>> -                          hmm_range->hmm_pfns);
>> +                          hmm_range->hmm_pfns, &vram_pages_vma);
>>               if (r)
>>                   pr_debug("failed %d to dma map range\n", r);
>> +            else
>> +                vram_pages += vram_pages_vma;
>>           }
>>             svm_range_lock(prange);
>> @@ -1746,6 +1761,19 @@ static int svm_range_validate_and_map(struct 
>> mm_struct *mm,
>>           addr = next;
>>       }
>>   +    if (addr == end) {
>> +        prange->vram_pages = vram_pages;
>> +
>> +        /* if prange does not include any vram page and it
>> +         * has not released svm_bo drop its svm_bo reference
>> +         * and set its actaul_loc to sys ram
>> +         */
>> +        if (!vram_pages && prange->ttm_res) {
>> +            prange->actual_loc = 0;
>> +            svm_range_vram_node_free(prange);
>> +        }
>> +    }
>> +
>>       svm_range_unreserve_bos(ctx);
>>       if (!r)
>>           prange->validate_timestamp = ktime_get_boottime();
>> @@ -1998,6 +2026,7 @@ static struct svm_range *svm_range_clone(struct 
>> svm_range *old)
>>       new->actual_loc = old->actual_loc;
>>       new->granularity = old->granularity;
>>       new->mapped_to_gpu = old->mapped_to_gpu;
>> +    new->vram_pages = old->vram_pages;
>>       bitmap_copy(new->bitmap_access, old->bitmap_access, 
>> MAX_GPU_INSTANCE);
>>       bitmap_copy(new->bitmap_aip, old->bitmap_aip, MAX_GPU_INSTANCE);
>>   @@ -2116,7 +2145,7 @@ svm_range_add(struct kfd_process *p, uint64_t 
>> start, uint64_t size,
>>           next_start = min(node->last, last) + 1;
>>             if (svm_range_is_same_attrs(p, prange, nattr, attrs) &&
>> -            prange->mapped_to_gpu) {
>> +            prange->mapped_to_gpu) {
>
> Unnecessary indentation change. Also the previous indentation looked 
> more consistent (aligned with the open parenthesis).
>
>
>>               /* nothing to do */
>>           } else if (node->start < start || node->last > last) {
>>               /* node intersects the update range and its attributes
>> @@ -2906,6 +2935,7 @@ svm_range_restore_pages(struct amdgpu_device 
>> *adev, unsigned int pasid,
>>               uint32_t vmid, uint32_t node_id,
>>               uint64_t addr, bool write_fault)
>>   {
>> +    unsigned long start, last, size;
>>       struct mm_struct *mm = NULL;
>>       struct svm_range_list *svms;
>>       struct svm_range *prange;
>> @@ -3041,32 +3071,35 @@ svm_range_restore_pages(struct amdgpu_device 
>> *adev, unsigned int pasid,
>>       kfd_smi_event_page_fault_start(node, p->lead_thread->pid, addr,
>>                          write_fault, timestamp);
>>   -    if (prange->actual_loc != best_loc) {
>> +    if (prange->actual_loc != 0 || best_loc != 0) {
>>           migration = true;
>> +        /* Align migration range start and size to granularity size */
>> +        size = 1UL << prange->granularity;
>> +        start = max_t(unsigned long, ALIGN_DOWN(addr, size), 
>> prange->start);
>> +        last = min_t(unsigned long, ALIGN(addr + 1, size) - 1, 
>> prange->last);
>> +
>>           if (best_loc) {
>> -            r = svm_migrate_to_vram(prange, best_loc, mm,
>> -                    KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU);
>> +            r = svm_migrate_to_vram(prange, best_loc, start, last,
>> +                    mm, KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU);
>>               if (r) {
>>                   pr_debug("svm_migrate_to_vram failed (%d) at %llx, 
>> falling back to system memory\n",
>>                        r, addr);
>>                   /* Fallback to system memory if migration to
>>                    * VRAM failed
>>                    */
>> -                if (prange->actual_loc)
>> -                    r = svm_migrate_vram_to_ram(prange, mm,
>> -                       KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU,
>> -                       NULL);
>> +                if (prange->actual_loc && prange->actual_loc != 
>> best_loc)
>> +                    r = svm_migrate_vram_to_ram(prange, mm, start, 
>> last,
>> +                        KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU, NULL);
>>                   else
>>                       r = 0;
>>               }
>>           } else {
>> -            r = svm_migrate_vram_to_ram(prange, mm,
>> -                    KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU,
>> -                    NULL);
>> +            r = svm_migrate_vram_to_ram(prange, mm, start, last,
>> +                    KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU, NULL);
>>           }
>>           if (r) {
>>               pr_debug("failed %d to migrate svms %p [0x%lx 0x%lx]\n",
>> -                 r, svms, prange->start, prange->last);
>> +                r, svms, start, last);
>
> Unnecessary indentation change, the previous indentation aligned with 
> the open parenthesis is preferred.
>
>
>>               goto out_unlock_range;
>>           }
>>       }
>> @@ -3420,18 +3453,24 @@ svm_range_trigger_migration(struct mm_struct 
>> *mm, struct svm_range *prange,
>>       *migrated = false;
>>       best_loc = svm_range_best_prefetch_location(prange);
>>   -    if (best_loc == KFD_IOCTL_SVM_LOCATION_UNDEFINED ||
>> -        best_loc == prange->actual_loc)
>> +    /* when best_loc is a gpu node and same as prange->actual_loc
>> +     * we still need do migration as prange->actual_loc !=0 does
>> +     * not mean all pages in prange are vram. hmm migrate will pick
>> +     * up right pages during migration.
>> +     */
>> +    if ((best_loc == KFD_IOCTL_SVM_LOCATION_UNDEFINED) ||
>> +        (best_loc == 0 && prange->actual_loc == 0))
>
> Indentation.
>
>
>>           return 0;
>>         if (!best_loc) {
>> -        r = svm_migrate_vram_to_ram(prange, mm,
>> +        r = svm_migrate_vram_to_ram(prange, mm, prange->start, 
>> prange->last,
>>                       KFD_MIGRATE_TRIGGER_PREFETCH, NULL);
>>           *migrated = !r;
>>           return r;
>>       }
>>   -    r = svm_migrate_to_vram(prange, best_loc, mm, 
>> KFD_MIGRATE_TRIGGER_PREFETCH);
>> +    r = svm_migrate_to_vram(prange, best_loc, prange->start, 
>> prange->last,
>> +                    mm, KFD_MIGRATE_TRIGGER_PREFETCH);
>
> Aligning the second line with the open parenthesis is preferred.
>
> Regards,
>   Felix
>
>
>>       *migrated = !r;
>>         return r;
>> @@ -3486,7 +3525,11 @@ static void 
>> svm_range_evict_svm_bo_worker(struct work_struct *work)
>>             mutex_lock(&prange->migrate_mutex);
>>           do {
>> +            /* migrate all vram pages in this prange to sys ram
>> +             * after that prange->actual_loc should be zero
>> +             */
>>               r = svm_migrate_vram_to_ram(prange, mm,
>> +                    prange->start, prange->last,
>>                       KFD_MIGRATE_TRIGGER_TTM_EVICTION, NULL);
>>           } while (!r && prange->actual_loc && --retries);
>>   diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h 
>> b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
>> index c528df1d0ba2..be11ba0c4289 100644
>> --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
>> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
>> @@ -78,6 +78,7 @@ struct svm_work_list_item {
>>    * @update_list:link list node used to add to update_list
>>    * @mapping:    bo_va mapping structure to create and update GPU 
>> page table
>>    * @npages:     number of pages
>> + * @vram_pages: vram pages number in this svm_range
>>    * @dma_addr:   dma mapping address on each GPU for system memory 
>> physical page
>>    * @ttm_res:    vram ttm resource map
>>    * @offset:     range start offset within mm_nodes
>> @@ -88,7 +89,9 @@ struct svm_work_list_item {
>>    * @flags:      flags defined as KFD_IOCTL_SVM_FLAG_*
>>    * @perferred_loc: perferred location, 0 for CPU, or GPU id
>>    * @perfetch_loc: last prefetch location, 0 for CPU, or GPU id
>> - * @actual_loc: the actual location, 0 for CPU, or GPU id
>> + * @actual_loc: this svm_range location. 0: all pages are from sys ram;
>> + *              GPU id: this svm_range may include vram pages from 
>> GPU with
>> + *              id actual_loc.
>>    * @granularity:migration granularity, log2 num pages
>>    * @invalid:    not 0 means cpu page table is invalidated
>>    * @validate_timestamp: system timestamp when range is validated
>> @@ -112,6 +115,7 @@ struct svm_range {
>>       struct list_head        list;
>>       struct list_head        update_list;
>>       uint64_t            npages;
>> +    uint64_t            vram_pages;
>>       dma_addr_t            *dma_addr[MAX_GPU_INSTANCE];
>>       struct ttm_resource        *ttm_res;
>>       uint64_t            offset;


More information about the amd-gfx mailing list