mirrors/linux
1
0
Fork 1
mirror of https://github.com/torvalds/linux.git synced 2025-11-04 10:40:15 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

drm/amdgpu: Use mmu_interval_insert instead of hmm_mirror

Browse source 
Remove the interval tree in the driver and rely on the tree maintained by
the mmu_notifier for delivering mmu_notifier invalidation callbacks.

For some reason amdgpu has a very complicated arrangement where it tries
to prevent duplicate entries in the interval_tree, this is not necessary,
each amdgpu_bo can be its own stand alone entry. interval_tree already
allows duplicates and overlaps in the tree.

Also, there is no need to remove entries upon a release callback, the
mmu_interval API safely allows objects to remain registered beyond the
lifetime of the mm. The driver only has to stop touching the pages during
release.

Link: https://lore.kernel.org/r/20191112202231.3856-12-jgg@ziepe.ca
Reviewed-by: Philip Yang <Philip.Yang@amd.com>
Tested-by: Philip Yang <Philip.Yang@amd.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
This commit is contained in:
Jason Gunthorpe 2019-11-12 16:22:28 -04:00
parent a9ae8731e6
commit 62914a99de
6 changed files with 77 additions and 281 deletions
repo.diff.show_diff_stats Download patch file Download diff file Expand all files Collapse all files

2
drivers/gpu/drm/amd/amdgpu/amdgpu.h
View file

@ -1006,6 +1006,8 @@ struct amdgpu_device {
struct mutex lock_reset; struct mutex lock_reset;
struct amdgpu_doorbell_index doorbell_index; struct amdgpu_doorbell_index doorbell_index;
struct mutex notifier_lock;
int asic_reset_res; int asic_reset_res;
struct work_struct xgmi_reset_work; struct work_struct xgmi_reset_work;

5
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
View file

@ -481,8 +481,7 @@ static void remove_kgd_mem_from_kfd_bo_list(struct kgd_mem *mem,
* *
* Returns 0 for success, negative errno for errors. * Returns 0 for success, negative errno for errors.
*/ */
static int init_user_pages(struct kgd_mem *mem, struct mm_struct *mm, static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr)
uint64_t user_addr)
{ {
struct amdkfd_process_info *process_info = mem->process_info; struct amdkfd_process_info *process_info = mem->process_info;
struct amdgpu_bo *bo = mem->bo; struct amdgpu_bo *bo = mem->bo;
@ -1195,7 +1194,7 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
add_kgd_mem_to_kfd_bo_list(*mem, avm->process_info, user_addr); add_kgd_mem_to_kfd_bo_list(*mem, avm->process_info, user_addr);
if (user_addr) { if (user_addr) {
ret = init_user_pages(*mem, current->mm, user_addr); ret = init_user_pages(*mem, user_addr);
if (ret) if (ret)
goto allocate_init_user_pages_failed; goto allocate_init_user_pages_failed;
} }

1
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
View file

@ -2633,6 +2633,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
mutex_init(&adev->virt.vf_errors.lock); mutex_init(&adev->virt.vf_errors.lock);
hash_init(adev->mn_hash); hash_init(adev->mn_hash);
mutex_init(&adev->lock_reset); mutex_init(&adev->lock_reset);
mutex_init(&adev->notifier_lock);
mutex_init(&adev->virt.dpm_mutex); mutex_init(&adev->virt.dpm_mutex);
mutex_init(&adev->psp.mutex); mutex_init(&adev->psp.mutex);

339
drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
View file

@ -50,66 +50,6 @@
#include "amdgpu.h" #include "amdgpu.h"
#include "amdgpu_amdkfd.h" #include "amdgpu_amdkfd.h"
/**
* struct amdgpu_mn_node
*
* @it: interval node defining start-last of the affected address range
* @bos: list of all BOs in the affected address range
*
* Manages all BOs which are affected of a certain range of address space.
*/
struct amdgpu_mn_node {
struct interval_tree_node it;
struct list_head bos;
};
/**
* amdgpu_mn_destroy - destroy the HMM mirror
*
* @work: previously sheduled work item
*
* Lazy destroys the notifier from a work item
*/
static void amdgpu_mn_destroy(struct work_struct *work)
{
struct amdgpu_mn *amn = container_of(work, struct amdgpu_mn, work);
struct amdgpu_device *adev = amn->adev;
struct amdgpu_mn_node *node, *next_node;
struct amdgpu_bo *bo, *next_bo;
mutex_lock(&adev->mn_lock);
down_write(&amn->lock);
hash_del(&amn->node);
rbtree_postorder_for_each_entry_safe(node, next_node,
&amn->objects.rb_root, it.rb) {
list_for_each_entry_safe(bo, next_bo, &node->bos, mn_list) {
bo->mn = NULL;
list_del_init(&bo->mn_list);
}
kfree(node);
}
up_write(&amn->lock);
mutex_unlock(&adev->mn_lock);
hmm_mirror_unregister(&amn->mirror);
kfree(amn);
}
/**
* amdgpu_hmm_mirror_release - callback to notify about mm destruction
*
* @mirror: the HMM mirror (mm) this callback is about
*
* Shedule a work item to lazy destroy HMM mirror.
*/
static void amdgpu_hmm_mirror_release(struct hmm_mirror *mirror)
{
struct amdgpu_mn *amn = container_of(mirror, struct amdgpu_mn, mirror);
INIT_WORK(&amn->work, amdgpu_mn_destroy);
schedule_work(&amn->work);
}
/** /**
* amdgpu_mn_lock - take the write side lock for this notifier * amdgpu_mn_lock - take the write side lock for this notifier
* *
@ -133,157 +73,80 @@ void amdgpu_mn_unlock(struct amdgpu_mn *mn)
} }
/** /**
* amdgpu_mn_read_lock - take the read side lock for this notifier * amdgpu_mn_invalidate_gfx - callback to notify about mm change
* *
* @amn: our notifier * @mni: the range (mm) is about to update
*/ * @range: details on the invalidation
static int amdgpu_mn_read_lock(struct amdgpu_mn *amn, bool blockable) * @cur_seq: Value to pass to mmu_interval_set_seq()
{
if (blockable)
down_read(&amn->lock);
else if (!down_read_trylock(&amn->lock))
return -EAGAIN;
return 0;
}
/**
* amdgpu_mn_read_unlock - drop the read side lock for this notifier
*
* @amn: our notifier
*/
static void amdgpu_mn_read_unlock(struct amdgpu_mn *amn)
{
up_read(&amn->lock);
}
/**
* amdgpu_mn_invalidate_node - unmap all BOs of a node
*
* @node: the node with the BOs to unmap
* @start: start of address range affected
* @end: end of address range affected
* *
* Block for operations on BOs to finish and mark pages as accessed and * Block for operations on BOs to finish and mark pages as accessed and
* potentially dirty. * potentially dirty.
*/ */
static void amdgpu_mn_invalidate_node(struct amdgpu_mn_node *node, static bool amdgpu_mn_invalidate_gfx(struct mmu_interval_notifier *mni,
unsigned long start, const struct mmu_notifier_range *range,
unsigned long end) unsigned long cur_seq)
{ {
struct amdgpu_bo *bo; struct amdgpu_bo *bo = container_of(mni, struct amdgpu_bo, notifier);
struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
long r; long r;
list_for_each_entry(bo, &node->bos, mn_list) { if (!mmu_notifier_range_blockable(range))
return false;
if (!amdgpu_ttm_tt_affect_userptr(bo->tbo.ttm, start, end)) mutex_lock(&adev->notifier_lock);
continue; r = dma_resv_wait_timeout_rcu(bo->tbo.base.resv, true, false,
MAX_SCHEDULE_TIMEOUT);
r = dma_resv_wait_timeout_rcu(bo->tbo.base.resv, mutex_unlock(&adev->notifier_lock);
true, false, MAX_SCHEDULE_TIMEOUT);
if (r <= 0) if (r <= 0)
DRM_ERROR("(%ld) failed to wait for user bo\n", r); DRM_ERROR("(%ld) failed to wait for user bo\n", r);
} return true;
} }
static const struct mmu_interval_notifier_ops amdgpu_mn_gfx_ops = {
.invalidate = amdgpu_mn_invalidate_gfx,
};
/** /**
* amdgpu_mn_sync_pagetables_gfx - callback to notify about mm change * amdgpu_mn_invalidate_hsa - callback to notify about mm change
* *
* @mirror: the hmm_mirror (mm) is about to update * @mni: the range (mm) is about to update
* @update: the update start, end address * @range: details on the invalidation
* @cur_seq: Value to pass to mmu_interval_set_seq()
* *
* Block for operations on BOs to finish and mark pages as accessed and * We temporarily evict the BO attached to this range. This necessitates
* potentially dirty. * evicting all user-mode queues of the process.
*/ */
static int static bool amdgpu_mn_invalidate_hsa(struct mmu_interval_notifier *mni,
amdgpu_mn_sync_pagetables_gfx(struct hmm_mirror *mirror, const struct mmu_notifier_range *range,
unsigned long cur_seq)
{
struct amdgpu_bo *bo = container_of(mni, struct amdgpu_bo, notifier);
struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
if (!mmu_notifier_range_blockable(range))
return false;
mutex_lock(&adev->notifier_lock);
amdgpu_amdkfd_evict_userptr(bo->kfd_bo, bo->notifier.mm);
mutex_unlock(&adev->notifier_lock);
return true;
}
static const struct mmu_interval_notifier_ops amdgpu_mn_hsa_ops = {
.invalidate = amdgpu_mn_invalidate_hsa,
};
static int amdgpu_mn_sync_pagetables(struct hmm_mirror *mirror,
const struct mmu_notifier_range *update) const struct mmu_notifier_range *update)
{ {
struct amdgpu_mn *amn = container_of(mirror, struct amdgpu_mn, mirror); struct amdgpu_mn *amn = container_of(mirror, struct amdgpu_mn, mirror);
unsigned long start = update->start;
unsigned long end = update->end;
bool blockable = mmu_notifier_range_blockable(update);
struct interval_tree_node *it;
/* notification is exclusive, but interval is inclusive */ if (!mmu_notifier_range_blockable(update))
end -= 1;
/* TODO we should be able to split locking for interval tree and
* amdgpu_mn_invalidate_node
*/
if (amdgpu_mn_read_lock(amn, blockable))
return -EAGAIN; return -EAGAIN;
it = interval_tree_iter_first(&amn->objects, start, end); down_read(&amn->lock);
while (it) { up_read(&amn->lock);
struct amdgpu_mn_node *node;
if (!blockable) {
amdgpu_mn_read_unlock(amn);
return -EAGAIN;
}
node = container_of(it, struct amdgpu_mn_node, it);
it = interval_tree_iter_next(it, start, end);
amdgpu_mn_invalidate_node(node, start, end);
}
amdgpu_mn_read_unlock(amn);
return 0;
}
/**
* amdgpu_mn_sync_pagetables_hsa - callback to notify about mm change
*
* @mirror: the hmm_mirror (mm) is about to update
* @update: the update start, end address
*
* We temporarily evict all BOs between start and end. This
* necessitates evicting all user-mode queues of the process. The BOs
* are restorted in amdgpu_mn_invalidate_range_end_hsa.
*/
static int
amdgpu_mn_sync_pagetables_hsa(struct hmm_mirror *mirror,
const struct mmu_notifier_range *update)
{
struct amdgpu_mn *amn = container_of(mirror, struct amdgpu_mn, mirror);
unsigned long start = update->start;
unsigned long end = update->end;
bool blockable = mmu_notifier_range_blockable(update);
struct interval_tree_node *it;
/* notification is exclusive, but interval is inclusive */
end -= 1;
if (amdgpu_mn_read_lock(amn, blockable))
return -EAGAIN;
it = interval_tree_iter_first(&amn->objects, start, end);
while (it) {
struct amdgpu_mn_node *node;
struct amdgpu_bo *bo;
if (!blockable) {
amdgpu_mn_read_unlock(amn);
return -EAGAIN;
}
node = container_of(it, struct amdgpu_mn_node, it);
it = interval_tree_iter_next(it, start, end);
list_for_each_entry(bo, &node->bos, mn_list) {
struct kgd_mem *mem = bo->kfd_bo;
if (amdgpu_ttm_tt_affect_userptr(bo->tbo.ttm,
start, end))
amdgpu_amdkfd_evict_userptr(mem, amn->mm);
}
}
amdgpu_mn_read_unlock(amn);
return 0; return 0;
} }
@ -295,12 +158,10 @@ amdgpu_mn_sync_pagetables_hsa(struct hmm_mirror *mirror,
static struct hmm_mirror_ops amdgpu_hmm_mirror_ops[] = { static struct hmm_mirror_ops amdgpu_hmm_mirror_ops[] = {
[AMDGPU_MN_TYPE_GFX] = { [AMDGPU_MN_TYPE_GFX] = {
.sync_cpu_device_pagetables = amdgpu_mn_sync_pagetables_gfx, .sync_cpu_device_pagetables = amdgpu_mn_sync_pagetables,
.release = amdgpu_hmm_mirror_release
}, },
[AMDGPU_MN_TYPE_HSA] = { [AMDGPU_MN_TYPE_HSA] = {
.sync_cpu_device_pagetables = amdgpu_mn_sync_pagetables_hsa, .sync_cpu_device_pagetables = amdgpu_mn_sync_pagetables,
.release = amdgpu_hmm_mirror_release
}, },
}; };
@ -327,7 +188,8 @@ struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
} }
hash_for_each_possible(adev->mn_hash, amn, node, key) hash_for_each_possible(adev->mn_hash, amn, node, key)
if (AMDGPU_MN_KEY(amn->mm, amn->type) == key) if (AMDGPU_MN_KEY(amn->mirror.hmm->mmu_notifier.mm,
amn->type) == key)
goto release_locks; goto release_locks;
amn = kzalloc(sizeof(*amn), GFP_KERNEL); amn = kzalloc(sizeof(*amn), GFP_KERNEL);
@ -337,10 +199,8 @@ struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
} }
amn->adev = adev; amn->adev = adev;
amn->mm = mm;
init_rwsem(&amn->lock); init_rwsem(&amn->lock);
amn->type = type; amn->type = type;
amn->objects = RB_ROOT_CACHED;
amn->mirror.ops = &amdgpu_hmm_mirror_ops[type]; amn->mirror.ops = &amdgpu_hmm_mirror_ops[type];
r = hmm_mirror_register(&amn->mirror, mm); r = hmm_mirror_register(&amn->mirror, mm);
@ -369,100 +229,33 @@ struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
* @bo: amdgpu buffer object * @bo: amdgpu buffer object
* @addr: userptr addr we should monitor * @addr: userptr addr we should monitor
* *
* Registers an HMM mirror for the given BO at the specified address. * Registers a mmu_notifier for the given BO at the specified address.
* Returns 0 on success, -ERRNO if anything goes wrong. * Returns 0 on success, -ERRNO if anything goes wrong.
*/ */
int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr) int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr)
{ {
unsigned long end = addr + amdgpu_bo_size(bo) - 1; if (bo->kfd_bo)
struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); bo->notifier.ops = &amdgpu_mn_hsa_ops;
enum amdgpu_mn_type type =
bo->kfd_bo ? AMDGPU_MN_TYPE_HSA : AMDGPU_MN_TYPE_GFX;
struct amdgpu_mn *amn;
struct amdgpu_mn_node *node = NULL, *new_node;
struct list_head bos;
struct interval_tree_node *it;
amn = amdgpu_mn_get(adev, type);
if (IS_ERR(amn))
return PTR_ERR(amn);
new_node = kmalloc(sizeof(*new_node), GFP_KERNEL);
if (!new_node)
return -ENOMEM;
INIT_LIST_HEAD(&bos);
down_write(&amn->lock);
while ((it = interval_tree_iter_first(&amn->objects, addr, end))) {
kfree(node);
node = container_of(it, struct amdgpu_mn_node, it);
interval_tree_remove(&node->it, &amn->objects);
addr = min(it->start, addr);
end = max(it->last, end);
list_splice(&node->bos, &bos);
}
if (!node)
node = new_node;
else else
kfree(new_node); bo->notifier.ops = &amdgpu_mn_gfx_ops;
bo->mn = amn; return mmu_interval_notifier_insert(&bo->notifier, addr,
amdgpu_bo_size(bo), current->mm);
node->it.start = addr;
node->it.last = end;
INIT_LIST_HEAD(&node->bos);
list_splice(&bos, &node->bos);
list_add(&bo->mn_list, &node->bos);
interval_tree_insert(&node->it, &amn->objects);
up_write(&amn->lock);
return 0;
} }
/** /**
* amdgpu_mn_unregister - unregister a BO for HMM mirror updates * amdgpu_mn_unregister - unregister a BO for notifier updates
* *
* @bo: amdgpu buffer object * @bo: amdgpu buffer object
* *
* Remove any registration of HMM mirror updates from the buffer object. * Remove any registration of mmu notifier updates from the buffer object.
*/ */
void amdgpu_mn_unregister(struct amdgpu_bo *bo) void amdgpu_mn_unregister(struct amdgpu_bo *bo)
{ {
struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); if (!bo->notifier.mm)
struct amdgpu_mn *amn;
struct list_head *head;
mutex_lock(&adev->mn_lock);
amn = bo->mn;
if (amn == NULL) {
mutex_unlock(&adev->mn_lock);
return; return;
} mmu_interval_notifier_remove(&bo->notifier);
bo->notifier.mm = NULL;
down_write(&amn->lock);
/* save the next list entry for later */
head = bo->mn_list.next;
bo->mn = NULL;
list_del_init(&bo->mn_list);
if (list_empty(head)) {
struct amdgpu_mn_node *node;
node = container_of(head, struct amdgpu_mn_node, bos);
interval_tree_remove(&node->it, &amn->objects);
kfree(node);
}
up_write(&amn->lock);
mutex_unlock(&adev->mn_lock);
} }
/* flags used by HMM internal, not related to CPU/GPU PTE flags */ /* flags used by HMM internal, not related to CPU/GPU PTE flags */

4
drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h
View file

@ -39,12 +39,10 @@ enum amdgpu_mn_type {
* struct amdgpu_mn * struct amdgpu_mn
* *
* @adev: amdgpu device pointer * @adev: amdgpu device pointer
* @mm: process address space
* @type: type of MMU notifier * @type: type of MMU notifier
* @work: destruction work item * @work: destruction work item
* @node: hash table node to find structure by adev and mn * @node: hash table node to find structure by adev and mn
* @lock: rw semaphore protecting the notifier nodes * @lock: rw semaphore protecting the notifier nodes
* @objects: interval tree containing amdgpu_mn_nodes
* @mirror: HMM mirror function support * @mirror: HMM mirror function support
* *
* Data for each amdgpu device and process address space. * Data for each amdgpu device and process address space.
@ -52,7 +50,6 @@ enum amdgpu_mn_type {
struct amdgpu_mn { struct amdgpu_mn {
/* constant after initialisation */ /* constant after initialisation */
struct amdgpu_device *adev; struct amdgpu_device *adev;
struct mm_struct *mm;
enum amdgpu_mn_type type; enum amdgpu_mn_type type;
/* only used on destruction */ /* only used on destruction */
@ -63,7 +60,6 @@ struct amdgpu_mn {
/* objects protected by lock */ /* objects protected by lock */
struct rw_semaphore lock; struct rw_semaphore lock;
struct rb_root_cached objects;
#ifdef CONFIG_HMM_MIRROR #ifdef CONFIG_HMM_MIRROR
/* HMM mirror */ /* HMM mirror */

11
drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
View file

@ -30,6 +30,9 @@
#include <drm/amdgpu_drm.h> #include <drm/amdgpu_drm.h>
#include "amdgpu.h" #include "amdgpu.h"
#ifdef CONFIG_MMU_NOTIFIER
#include <linux/mmu_notifier.h>
#endif
#define AMDGPU_BO_INVALID_OFFSET LONG_MAX #define AMDGPU_BO_INVALID_OFFSET LONG_MAX
#define AMDGPU_BO_MAX_PLACEMENTS 3 #define AMDGPU_BO_MAX_PLACEMENTS 3
@ -100,10 +103,12 @@ struct amdgpu_bo {
struct ttm_bo_kmap_obj dma_buf_vmap; struct ttm_bo_kmap_obj dma_buf_vmap;
struct amdgpu_mn *mn; struct amdgpu_mn *mn;
union {
struct list_head mn_list; #ifdef CONFIG_MMU_NOTIFIER
struct mmu_interval_notifier notifier;
#endif
struct list_head shadow_list; struct list_head shadow_list;
};
struct kgd_mem *kfd_bo; struct kgd_mem *kfd_bo;
}; };