Re: [PATCH] drm/ttm: fix busy memory to fail other user v5

["Christian König" <ckoenig.leichtzumerken@gmail.com>]

Ping!

Marek is going to need this for its GDS patches as well.

Christina.

Am 30.04.19 um 11:10 schrieb Christian König:
> Am 30.04.19 um 09:01 schrieb Chunming Zhou:
>> heavy gpu job could occupy memory long time, which lead other user 
>> fail to get memory.
>>
>> basically pick up Christian idea:
>>
>> 1. Reserve the BO in DC using a ww_mutex ticket (trivial).
>> 2. If we then run into this EBUSY condition in TTM check if the BO we 
>> need memory for (or rather the ww_mutex of its reservation object) 
>> has a ticket assigned.
>> 3. If we have a ticket we grab a reference to the first BO on the 
>> LRU, drop the LRU lock and try to grab the reservation lock with the 
>> ticket.
>> 4. If getting the reservation lock with the ticket succeeded we check 
>> if the BO is still the first one on the LRU in question (the BO could 
>> have moved).
>> 5. If the BO is still the first one on the LRU in question we try to 
>> evict it as we would evict any other BO.
>> 6. If any of the "If's" above fail we just back off and return -EBUSY.
>>
>> v2: fix some minor check
>> v3: address Christian v2 comments.
>> v4: fix some missing
>> v5: handle first_bo unlock and bo_get/put
>>
>> Change-Id: I21423fb922f885465f13833c41df1e134364a8e7
>> Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
>> ---
>>   drivers/gpu/drm/amd/amdgpu/amdgpu_object.c    |  7 +-
>>   .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 22 +++--
>>   drivers/gpu/drm/ttm/ttm_bo.c                  | 81 +++++++++++++++++--
>>   3 files changed, 99 insertions(+), 11 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c 
>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
>> index affde72b44db..523773e85284 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
>> @@ -811,7 +811,12 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo 
>> *bo, u32 domain,
>>                    u64 min_offset, u64 max_offset)
>>   {
>>       struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
>> -    struct ttm_operation_ctx ctx = { false, false };
>> +    struct ttm_operation_ctx ctx = {
>> +        .interruptible = false,
>> +        .no_wait_gpu = false,
>> +        .resv = bo->tbo.resv,
>> +        .flags = 0
>> +    };
>>       int r, i;
>>         if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm))
>> diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c 
>> b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> index a5cacf846e1b..cc3677c4a4c2 100644
>> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> @@ -4101,6 +4101,9 @@ static int dm_plane_helper_prepare_fb(struct 
>> drm_plane *plane,
>>       struct amdgpu_device *adev;
>>       struct amdgpu_bo *rbo;
>>       struct dm_plane_state *dm_plane_state_new, *dm_plane_state_old;
>> +    struct list_head list, duplicates;
>> +    struct ttm_validate_buffer tv;
>> +    struct ww_acquire_ctx ticket;
>>       uint64_t tiling_flags;
>>       uint32_t domain;
>>       int r;
>> @@ -4117,9 +4120,18 @@ static int dm_plane_helper_prepare_fb(struct 
>> drm_plane *plane,
>>       obj = new_state->fb->obj[0];
>>       rbo = gem_to_amdgpu_bo(obj);
>>       adev = amdgpu_ttm_adev(rbo->tbo.bdev);
>> -    r = amdgpu_bo_reserve(rbo, false);
>> -    if (unlikely(r != 0))
>> +    INIT_LIST_HEAD(&list);
>> +    INIT_LIST_HEAD(&duplicates);
>> +
>> +    tv.bo = &rbo->tbo;
>> +    tv.num_shared = 1;
>> +    list_add(&tv.head, &list);
>> +
>> +    r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates);
>> +    if (r) {
>> +        dev_err(adev->dev, "fail to reserve bo (%d)\n", r);
>>           return r;
>> +    }
>>         if (plane->type != DRM_PLANE_TYPE_CURSOR)
>>           domain = amdgpu_display_supported_domains(adev);
>> @@ -4130,21 +4142,21 @@ static int dm_plane_helper_prepare_fb(struct 
>> drm_plane *plane,
>>       if (unlikely(r != 0)) {
>>           if (r != -ERESTARTSYS)
>>               DRM_ERROR("Failed to pin framebuffer with error %d\n", r);
>> -        amdgpu_bo_unreserve(rbo);
>> +        ttm_eu_backoff_reservation(&ticket, &list);
>>           return r;
>>       }
>>         r = amdgpu_ttm_alloc_gart(&rbo->tbo);
>>       if (unlikely(r != 0)) {
>>           amdgpu_bo_unpin(rbo);
>> -        amdgpu_bo_unreserve(rbo);
>> +        ttm_eu_backoff_reservation(&ticket, &list);
>>           DRM_ERROR("%p bind failed\n", rbo);
>>           return r;
>>       }
>>         amdgpu_bo_get_tiling_flags(rbo, &tiling_flags);
>>   -    amdgpu_bo_unreserve(rbo);
>> +    ttm_eu_backoff_reservation(&ticket, &list);
>
> Well I can only repeat myself, please put that into a separate patch!
>
>>         afb->address = amdgpu_bo_gpu_offset(rbo);
>>   diff --git a/drivers/gpu/drm/ttm/ttm_bo.c 
>> b/drivers/gpu/drm/ttm/ttm_bo.c
>> index 8502b3ed2d88..2c4963e105d9 100644
>> --- a/drivers/gpu/drm/ttm/ttm_bo.c
>> +++ b/drivers/gpu/drm/ttm/ttm_bo.c
>> @@ -766,11 +766,13 @@ EXPORT_SYMBOL(ttm_bo_eviction_valuable);
>>    * b. Otherwise, trylock it.
>>    */
>>   static bool ttm_bo_evict_swapout_allowable(struct ttm_buffer_object 
>> *bo,
>> -            struct ttm_operation_ctx *ctx, bool *locked)
>> +            struct ttm_operation_ctx *ctx, bool *locked, bool *busy)
>>   {
>>       bool ret = false;
>>         *locked = false;
>> +    if (busy)
>> +        *busy = false;
>>       if (bo->resv == ctx->resv) {
>>           reservation_object_assert_held(bo->resv);
>>           if (ctx->flags & TTM_OPT_FLAG_ALLOW_RES_EVICT
>> @@ -779,6 +781,8 @@ static bool ttm_bo_evict_swapout_allowable(struct 
>> ttm_buffer_object *bo,
>>       } else {
>>           *locked = reservation_object_trylock(bo->resv);
>>           ret = *locked;
>> +        if (!ret && busy)
>> +            *busy = true;
>>       }
>>         return ret;
>> @@ -791,7 +795,7 @@ static int ttm_mem_evict_first(struct 
>> ttm_bo_device *bdev,
>>   {
>>       struct ttm_bo_global *glob = bdev->glob;
>>       struct ttm_mem_type_manager *man = &bdev->man[mem_type];
>> -    struct ttm_buffer_object *bo = NULL;
>> +    struct ttm_buffer_object *bo = NULL, *first_bo = NULL;
>>       bool locked = false;
>>       unsigned i;
>>       int ret;
>> @@ -799,8 +803,15 @@ static int ttm_mem_evict_first(struct 
>> ttm_bo_device *bdev,
>>       spin_lock(&glob->lru_lock);
>>       for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) {
>>           list_for_each_entry(bo, &man->lru[i], lru) {
>> -            if (!ttm_bo_evict_swapout_allowable(bo, ctx, &locked))
>> +            bool busy = false;
>> +            if (!ttm_bo_evict_swapout_allowable(bo, ctx, &locked,
>> +                                &busy)) {
>> +                if (!first_bo && busy) {
>> +                    ttm_bo_get(bo);
>> +                    first_bo = bo;
>> +                }
>>                   continue;
>> +            }
>>                 if (place && !bdev->driver->eviction_valuable(bo,
>>                                         place)) {
>> @@ -808,6 +819,7 @@ static int ttm_mem_evict_first(struct 
>> ttm_bo_device *bdev,
>>                       reservation_object_unlock(bo->resv);
>>                   continue;
>>               }
>> +
>>               break;
>>           }
>>   @@ -820,7 +832,65 @@ static int ttm_mem_evict_first(struct 
>> ttm_bo_device *bdev,
>>         if (!bo) {
>>           spin_unlock(&glob->lru_lock);
>> -        return -EBUSY;
>> +        /* check if other user occupy memory too long time */
>> +        if (!first_bo || !ctx || !ctx->resv || !ctx->resv->lock.ctx) {
>> +            if (first_bo)
>> +                ttm_bo_put(first_bo);
>> +            return -EBUSY;
>> +        }
>> +        if (ctx->interruptible)
>> +            ret = ww_mutex_lock_interruptible(&first_bo->resv->lock,
>> +                              ctx->resv->lock.ctx);
>> +        else
>> +            ret = ww_mutex_lock(&first_bo->resv->lock, 
>> ctx->resv->lock.ctx);
>> +        if (ret) {
>> +            ttm_bo_put(first_bo);
>> +            return ret;
>> +        }
>> +        spin_lock(&glob->lru_lock);
>> +        for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) {
>> +            /* previous busy resv lock is held by above, idle now,
>> +             * so let them evictable.
>> +             */
>> +            struct ttm_operation_ctx busy_ctx = {
>> +                .interruptible = ctx->interruptible,
>> +                .no_wait_gpu   = ctx->no_wait_gpu,
>> +                .resv           = first_bo->resv,
>> +                .flags           = TTM_OPT_FLAG_ALLOW_RES_EVICT
>> +            };
>> +            list_for_each_entry(bo, &man->lru[i], lru) {
>> +                if (!ttm_bo_evict_swapout_allowable(bo,
>> +                                    &busy_ctx,
>> +                                    &locked,
>> +                                    NULL))
>> +                    continue;
>> +
>> +                if (place && !bdev->driver->eviction_valuable(bo,
>> +                                      place)) {
>> +                    if (locked)
>> +                        reservation_object_unlock(bo->resv);
>> +                    continue;
>> +                }
>> +                break;
>> +            }
>> +            /* If the inner loop terminated early, we have our 
>> candidate */
>> +            if (&bo->lru != &man->lru[i])
>> +                break;
>> +            bo = NULL;
>> +        }
>
> So we are now searching for any BO again which matches the criteria? 
> Good idea.
>
> Can we factor those two loops out into a separate function? I mean 
> they are mostly identical now to the one above, aren't they?
>
>> +        if (bo && (bo->resv == first_bo->resv))
>> +            locked = true;
>> +        else if (bo)
>> +            ww_mutex_unlock(&first_bo->resv->lock);
>> +        ttm_bo_put(first_bo);
>> +        first_bo = NULL;
>> +        if (!bo) {
>> +            spin_unlock(&glob->lru_lock);
>> +            return -EBUSY;
>> +        }
>> +    } else {
>> +        if (first_bo)
>> +            ttm_bo_put(first_bo);
>
> This must come a bit later, since we can't call ttm_bo_put() while 
> holding the spinlock.
>
> Regards,
> Christian.
>
>>       }
>>         kref_get(&bo->list_kref);
>> @@ -1784,7 +1854,8 @@ int ttm_bo_swapout(struct ttm_bo_global *glob, 
>> struct ttm_operation_ctx *ctx)
>>       spin_lock(&glob->lru_lock);
>>       for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) {
>>           list_for_each_entry(bo, &glob->swap_lru[i], swap) {
>> -            if (ttm_bo_evict_swapout_allowable(bo, ctx, &locked)) {
>> +            if (ttm_bo_evict_swapout_allowable(bo, ctx, &locked,
>> +                               NULL)) {
>>                   ret = 0;
>>                   break;
>>               }
>

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel