* [PATCH 1/2] drm/ttm: drop size from resource manager base class.
2020-08-06 4:39 [rfc] further ttm cleanups Dave Airlie
@ 2020-08-06 4:39 ` Dave Airlie
2020-08-06 5:21 ` Dave Airlie
2020-08-06 4:39 ` [PATCH 2/2] drm/ttm/nouveau: move io_lru storage into driver Dave Airlie
1 sibling, 1 reply; 4+ messages in thread
From: Dave Airlie @ 2020-08-06 4:39 UTC (permalink / raw)
To: dri-devel
From: Dave Airlie <airlied@redhat.com>
This is a bit more involved that it looked, the range manager
needs accessors adding and amdgpu needs a bit of a refactor.
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c | 17 +++++++--------
drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 5 +++--
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 13 ++++++------
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 2 +-
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h | 12 +++++++++++
drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 14 +++++--------
drivers/gpu/drm/nouveau/nouveau_ttm.c | 5 ++---
drivers/gpu/drm/radeon/radeon_gem.c | 2 +-
drivers/gpu/drm/radeon/radeon_ttm.c | 3 +--
drivers/gpu/drm/ttm/ttm_bo.c | 7 ++-----
drivers/gpu/drm/ttm/ttm_range_manager.c | 21 +++++++++++++++++--
drivers/gpu/drm/vmwgfx/vmwgfx_gmrid_manager.c | 2 +-
drivers/gpu/drm/vmwgfx/vmwgfx_thp.c | 9 ++++----
include/drm/ttm/ttm_bo_api.h | 4 +---
include/drm/ttm/ttm_bo_driver.h | 20 ++++++++++++++++--
15 files changed, 84 insertions(+), 52 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
index ccc8b8196d26..db5c0d34daa1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
@@ -24,11 +24,6 @@
#include "amdgpu.h"
-static inline struct amdgpu_gtt_mgr *to_gtt_mgr(struct ttm_resource_manager *man)
-{
- return container_of(man, struct amdgpu_gtt_mgr, manager);
-}
-
struct amdgpu_gtt_node {
struct drm_mm_node node;
struct ttm_buffer_object *tbo;
@@ -48,8 +43,9 @@ static ssize_t amdgpu_mem_info_gtt_total_show(struct device *dev,
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT);
+ struct amdgpu_gtt_mgr *gtt_mgr = to_gtt_mgr(man);
return snprintf(buf, PAGE_SIZE, "%llu\n",
- man->size * PAGE_SIZE);
+ gtt_mgr->size * PAGE_SIZE);
}
/**
@@ -96,13 +92,14 @@ int amdgpu_gtt_mgr_init(struct amdgpu_device *adev, uint64_t gtt_size)
man->available_caching = TTM_PL_MASK_CACHING;
man->default_caching = TTM_PL_FLAG_CACHED;
- ttm_resource_manager_init(man, gtt_size >> PAGE_SHIFT);
+ mgr->size = gtt_size >> PAGE_SHIFT;
+ ttm_resource_manager_init(man);
start = AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GTT_NUM_TRANSFER_WINDOWS;
size = (adev->gmc.gart_size >> PAGE_SHIFT) - start;
drm_mm_init(&mgr->mm, start, size);
spin_lock_init(&mgr->lock);
- atomic64_set(&mgr->available, gtt_size >> PAGE_SHIFT);
+ atomic64_set(&mgr->available, mgr->size);
ret = device_create_file(adev->dev, &dev_attr_mem_info_gtt_total);
if (ret) {
@@ -262,7 +259,7 @@ static void amdgpu_gtt_mgr_del(struct ttm_resource_manager *man,
uint64_t amdgpu_gtt_mgr_usage(struct ttm_resource_manager *man)
{
struct amdgpu_gtt_mgr *mgr = to_gtt_mgr(man);
- s64 result = man->size - atomic64_read(&mgr->available);
+ s64 result = mgr->size - atomic64_read(&mgr->available);
return (result > 0 ? result : 0) * PAGE_SIZE;
}
@@ -304,7 +301,7 @@ static void amdgpu_gtt_mgr_debug(struct ttm_resource_manager *man,
spin_unlock(&mgr->lock);
drm_printf(printer, "man size:%llu pages, gtt available:%lld pages, usage:%lluMB\n",
- man->size, (u64)atomic64_read(&mgr->available),
+ mgr->size, (u64)atomic64_read(&mgr->available),
amdgpu_gtt_mgr_usage(man) >> 20);
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index 2763bca163e3..8dcee654d7f5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -623,7 +623,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
min(adev->gmc.visible_vram_size -
atomic64_read(&adev->visible_pin_size),
vram_gtt.vram_size);
- vram_gtt.gtt_size = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT)->size;
+ vram_gtt.gtt_size = to_gtt_mgr(ttm_manager_type(&adev->mman.bdev, TTM_PL_TT))->size;
vram_gtt.gtt_size *= PAGE_SIZE;
vram_gtt.gtt_size -= atomic64_read(&adev->gart_pin_size);
return copy_to_user(out, &vram_gtt,
@@ -635,6 +635,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM);
struct ttm_resource_manager *gtt_man =
ttm_manager_type(&adev->mman.bdev, TTM_PL_TT);
+ struct amdgpu_gtt_mgr *gtt_mgr = to_gtt_mgr(gtt_man);
memset(&mem, 0, sizeof(mem));
mem.vram.total_heap_size = adev->gmc.real_vram_size;
mem.vram.usable_heap_size = adev->gmc.real_vram_size -
@@ -655,7 +656,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
mem.cpu_accessible_vram.max_allocation =
mem.cpu_accessible_vram.usable_heap_size * 3 / 4;
- mem.gtt.total_heap_size = gtt_man->size;
+ mem.gtt.total_heap_size = gtt_mgr->size;
mem.gtt.total_heap_size *= PAGE_SIZE;
mem.gtt.usable_heap_size = mem.gtt.total_heap_size -
atomic64_read(&adev->gart_pin_size);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 43f4966331dd..33957d829197 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -442,25 +442,24 @@ void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
static bool amdgpu_bo_validate_size(struct amdgpu_device *adev,
unsigned long size, u32 domain)
{
- struct ttm_resource_manager *man = NULL;
-
+ unsigned long mgr_size = 0;
/*
* If GTT is part of requested domains the check must succeed to
* allow fall back to GTT
*/
if (domain & AMDGPU_GEM_DOMAIN_GTT) {
- man = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT);
+ mgr_size = to_gtt_mgr(ttm_manager_type(&adev->mman.bdev, TTM_PL_TT))->size;
- if (size < (man->size << PAGE_SHIFT))
+ if (size < (mgr_size << PAGE_SHIFT))
return true;
else
goto fail;
}
if (domain & AMDGPU_GEM_DOMAIN_VRAM) {
- man = ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM);
+ mgr_size = to_vram_mgr(ttm_manager_type(&adev->mman.bdev, TTM_PL_TT))->size;
- if (size < (man->size << PAGE_SHIFT))
+ if (size < (mgr_size << PAGE_SHIFT))
return true;
else
goto fail;
@@ -472,7 +471,7 @@ static bool amdgpu_bo_validate_size(struct amdgpu_device *adev,
fail:
DRM_DEBUG("BO size %lu > total memory in domain: %llu\n", size,
- man->size << PAGE_SHIFT);
+ mgr_size << PAGE_SHIFT);
return false;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index 682172d59f60..8004acd2292a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -2063,7 +2063,7 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
size = adev->gmc.real_vram_size;
else
size = adev->gmc.visible_vram_size;
- man->size = size >> PAGE_SHIFT;
+ to_vram_mgr(man)->size = size >> PAGE_SHIFT;
adev->mman.buffer_funcs_enabled = enable;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
index 1fd7ed43be43..3db950ef2b86 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
@@ -46,16 +46,28 @@ struct amdgpu_gtt_mgr {
struct drm_mm mm;
spinlock_t lock;
atomic64_t available;
+ unsigned long size;
};
+static inline struct amdgpu_gtt_mgr *to_gtt_mgr(struct ttm_resource_manager *man)
+{
+ return container_of(man, struct amdgpu_gtt_mgr, manager);
+}
+
struct amdgpu_vram_mgr {
struct ttm_resource_manager manager;
struct drm_mm mm;
spinlock_t lock;
atomic64_t usage;
atomic64_t vis_usage;
+ unsigned long size;
};
+static inline struct amdgpu_vram_mgr *to_vram_mgr(struct ttm_resource_manager *man)
+{
+ return container_of(man, struct amdgpu_vram_mgr, manager);
+}
+
struct amdgpu_mman {
struct ttm_bo_device bdev;
bool mem_global_referenced;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
index d2f04abe3f81..adf582bbffed 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
@@ -28,11 +28,6 @@
#include "amdgpu_atomfirmware.h"
#include "atom.h"
-static inline struct amdgpu_vram_mgr *to_vram_mgr(struct ttm_resource_manager *man)
-{
- return container_of(man, struct amdgpu_vram_mgr, manager);
-}
-
static inline struct amdgpu_device *man_to_adev(struct ttm_resource_manager *man)
{
return container_of(man, struct amdgpu_device, mman.vram_mgr.manager);
@@ -180,11 +175,12 @@ int amdgpu_vram_mgr_init(struct amdgpu_device *adev)
man->available_caching = TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_WC;
man->default_caching = TTM_PL_FLAG_WC;
- ttm_resource_manager_init(man, adev->gmc.real_vram_size >> PAGE_SHIFT);
+ ttm_resource_manager_init(man);
man->func = &amdgpu_vram_mgr_func;
- drm_mm_init(&mgr->mm, 0, man->size);
+ mgr->size = adev->gmc.real_vram_size >> PAGE_SHIFT;
+ drm_mm_init(&mgr->mm, 0, mgr->size);
spin_lock_init(&mgr->lock);
/* Add the two VRAM-related sysfs files */
@@ -325,7 +321,7 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man,
lpfn = place->lpfn;
if (!lpfn)
- lpfn = man->size;
+ lpfn = mgr->size;
max_bytes = adev->gmc.mc_vram_size;
if (tbo->type != ttm_bo_type_kernel)
@@ -599,7 +595,7 @@ static void amdgpu_vram_mgr_debug(struct ttm_resource_manager *man,
spin_unlock(&mgr->lock);
drm_printf(printer, "man size:%llu pages, ram usage:%lluMB, vis usage:%lluMB\n",
- man->size, amdgpu_vram_mgr_usage(man) >> 20,
+ mgr->size, amdgpu_vram_mgr_usage(man) >> 20,
amdgpu_vram_mgr_vis_usage(man) >> 20);
}
diff --git a/drivers/gpu/drm/nouveau/nouveau_ttm.c b/drivers/gpu/drm/nouveau/nouveau_ttm.c
index 9e96b6ff24cf..2aba2e245548 100644
--- a/drivers/gpu/drm/nouveau/nouveau_ttm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_ttm.c
@@ -175,8 +175,7 @@ nouveau_ttm_init_vram(struct nouveau_drm *drm)
man->func = &nouveau_vram_manager;
man->use_io_reserve_lru = true;
- ttm_resource_manager_init(man,
- drm->gem.vram_available >> PAGE_SHIFT);
+ ttm_resource_manager_init(man);
ttm_set_driver_manager(&drm->ttm.bdev, TTM_PL_VRAM, man);
ttm_resource_manager_set_used(man, true);
return 0;
@@ -237,7 +236,7 @@ nouveau_ttm_init_gtt(struct nouveau_drm *drm)
man->available_caching = available_caching;
man->default_caching = default_caching;
man->use_tt = true;
- ttm_resource_manager_init(man, size_pages);
+ ttm_resource_manager_init(man);
ttm_set_driver_manager(&drm->ttm.bdev, TTM_PL_TT, man);
ttm_resource_manager_set_used(man, true);
return 0;
diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c
index 7f5dfe04789e..00258d158cfe 100644
--- a/drivers/gpu/drm/radeon/radeon_gem.c
+++ b/drivers/gpu/drm/radeon/radeon_gem.c
@@ -228,7 +228,7 @@ int radeon_gem_info_ioctl(struct drm_device *dev, void *data,
man = ttm_manager_type(&rdev->mman.bdev, TTM_PL_VRAM);
- args->vram_size = (u64)man->size << PAGE_SHIFT;
+ args->vram_size = (u64)ttm_range_man_size(man) << PAGE_SHIFT;
args->vram_visible = rdev->mc.visible_vram_size;
args->vram_visible -= rdev->vram_pin_size;
args->gart_size = rdev->mc.gtt_size;
diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c
index a068d6960c23..1b96872cc756 100644
--- a/drivers/gpu/drm/radeon/radeon_ttm.c
+++ b/drivers/gpu/drm/radeon/radeon_ttm.c
@@ -841,9 +841,8 @@ void radeon_ttm_set_active_vram_size(struct radeon_device *rdev, u64 size)
if (!rdev->mman.initialized)
return;
- man = ttm_manager_type(&rdev->mman.bdev, TTM_PL_VRAM);
/* this just adjusts TTM size idea, which sets lpfn to the correct value */
- man->size = size >> PAGE_SHIFT;
+ ttm_range_man_adjust_size(man, size >> PAGE_SHIFT);
}
static vm_fault_t radeon_ttm_fault(struct vm_fault *vmf)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index 1ea5de976ec3..74b7def25f91 100644
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -82,7 +82,6 @@ void ttm_resource_manager_debug(struct ttm_resource_manager *man,
{
drm_printf(p, " use_type: %d\n", man->use_type);
drm_printf(p, " use_tt: %d\n", man->use_tt);
- drm_printf(p, " size: %llu\n", man->size);
drm_printf(p, " available_caching: 0x%08X\n", man->available_caching);
drm_printf(p, " default_caching: 0x%08X\n", man->default_caching);
if (man->func && man->func->debug)
@@ -1470,8 +1469,7 @@ int ttm_bo_evict_mm(struct ttm_bo_device *bdev, unsigned mem_type)
}
EXPORT_SYMBOL(ttm_bo_evict_mm);
-void ttm_resource_manager_init(struct ttm_resource_manager *man,
- unsigned long p_size)
+void ttm_resource_manager_init(struct ttm_resource_manager *man)
{
unsigned i;
@@ -1479,7 +1477,6 @@ void ttm_resource_manager_init(struct ttm_resource_manager *man,
mutex_init(&man->io_reserve_mutex);
spin_lock_init(&man->move_lock);
INIT_LIST_HEAD(&man->io_reserve_lru);
- man->size = p_size;
for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i)
INIT_LIST_HEAD(&man->lru[i]);
@@ -1592,7 +1589,7 @@ static void ttm_bo_init_sysman(struct ttm_bo_device *bdev)
man->available_caching = TTM_PL_MASK_CACHING;
man->default_caching = TTM_PL_FLAG_CACHED;
- ttm_resource_manager_init(man, 0);
+ ttm_resource_manager_init(man);
ttm_set_driver_manager(bdev, TTM_PL_SYSTEM, man);
ttm_resource_manager_set_used(man, true);
}
diff --git a/drivers/gpu/drm/ttm/ttm_range_manager.c b/drivers/gpu/drm/ttm/ttm_range_manager.c
index 7442d811f867..0194967a7ec5 100644
--- a/drivers/gpu/drm/ttm/ttm_range_manager.c
+++ b/drivers/gpu/drm/ttm/ttm_range_manager.c
@@ -47,6 +47,7 @@ struct ttm_range_manager {
struct ttm_resource_manager manager;
struct drm_mm mm;
spinlock_t lock;
+ uint64_t size;
};
static inline struct ttm_range_manager *to_range_manager(struct ttm_resource_manager *man)
@@ -68,7 +69,7 @@ static int ttm_range_man_get_node(struct ttm_resource_manager *man,
lpfn = place->lpfn;
if (!lpfn)
- lpfn = man->size;
+ lpfn = rman->size;
node = kzalloc(sizeof(*node), GFP_KERNEL);
if (!node)
@@ -133,8 +134,9 @@ int ttm_range_man_init(struct ttm_bo_device *bdev,
man->func = &ttm_range_manager_func;
- ttm_resource_manager_init(man, p_size);
+ ttm_resource_manager_init(man);
+ rman->size = p_size;
drm_mm_init(&rman->mm, 0, p_size);
spin_lock_init(&rman->lock);
@@ -185,3 +187,18 @@ static const struct ttm_resource_manager_func ttm_range_manager_func = {
.put_node = ttm_range_man_put_node,
.debug = ttm_range_man_debug
};
+
+void ttm_range_man_adjust_size(struct ttm_resource_manager *man,
+ unsigned long size)
+{
+ struct ttm_range_manager *rman = to_range_manager(man);
+ rman->size = size;
+}
+EXPORT_SYMBOL(ttm_range_man_adjust_size);
+
+unsigned long ttm_range_man_size(struct ttm_resource_manager *man)
+{
+ struct ttm_range_manager *rman = to_range_manager(man);
+ return rman->size;
+}
+EXPORT_SYMBOL(ttm_range_man_size);
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_gmrid_manager.c b/drivers/gpu/drm/vmwgfx/vmwgfx_gmrid_manager.c
index 3fea7a6c7cfa..a7a2deae2277 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_gmrid_manager.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_gmrid_manager.c
@@ -116,7 +116,7 @@ int vmw_gmrid_man_init(struct vmw_private *dev_priv, int type)
man->default_caching = TTM_PL_FLAG_CACHED;
/* TODO: This is most likely not correct */
man->use_tt = true;
- ttm_resource_manager_init(man, 0);
+ ttm_resource_manager_init(man);
spin_lock_init(&gman->lock);
gman->used_gmr_pages = 0;
ida_init(&gman->gmr_ida);
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_thp.c b/drivers/gpu/drm/vmwgfx/vmwgfx_thp.c
index f594e2e6ab7e..74cdee48b191 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_thp.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_thp.c
@@ -19,6 +19,7 @@ struct vmw_thp_manager {
struct ttm_resource_manager manager;
struct drm_mm mm;
spinlock_t lock;
+ uint64_t size;
};
static struct vmw_thp_manager *to_thp_manager(struct ttm_resource_manager *man)
@@ -63,7 +64,7 @@ static int vmw_thp_get_node(struct ttm_resource_manager *man,
lpfn = place->lpfn;
if (!lpfn)
- lpfn = man->size;
+ lpfn = rman->size;
mode = DRM_MM_INSERT_BEST;
if (place->flags & TTM_PL_FLAG_TOPDOWN)
@@ -134,10 +135,10 @@ int vmw_thp_init(struct vmw_private *dev_priv)
man->available_caching = TTM_PL_FLAG_CACHED;
man->default_caching = TTM_PL_FLAG_CACHED;
- ttm_resource_manager_init(man,
- dev_priv->vram_size >> PAGE_SHIFT);
+ ttm_resource_manager_init(man);
- drm_mm_init(&rman->mm, 0, man->size);
+ rman->size = dev_priv->vram_size >> PAGE_SHIFT;
+ drm_mm_init(&rman->mm, 0, rman->size);
spin_lock_init(&rman->lock);
ttm_set_driver_manager(&dev_priv->bdev, TTM_PL_VRAM, &rman->manager);
diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h
index 247d4f803443..9570ed7d01bf 100644
--- a/include/drm/ttm/ttm_bo_api.h
+++ b/include/drm/ttm/ttm_bo_api.h
@@ -537,12 +537,10 @@ int ttm_bo_create(struct ttm_bo_device *bdev, unsigned long size,
* ttm_resource_manager_init
*
* @man: memory manager object to init
- * @p_size: size managed area in pages.
*
* Initialise core parts of a manager object.
*/
-void ttm_resource_manager_init(struct ttm_resource_manager *man,
- unsigned long p_size);
+void ttm_resource_manager_init(struct ttm_resource_manager *man);
/**
* ttm_bo_evict_mm
diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h
index 576c91c85a6b..dc57445f90e6 100644
--- a/include/drm/ttm/ttm_bo_driver.h
+++ b/include/drm/ttm/ttm_bo_driver.h
@@ -116,7 +116,6 @@ struct ttm_resource_manager_func {
* managed by this memory type.
* @gpu_offset: If used, the GPU offset of the first managed page of
* fixed memory or the first managed location in an aperture.
- * @size: Size of the managed region.
* @available_caching: A mask of available caching types, TTM_PL_FLAG_XX,
* as defined in ttm_placement_common.h
* @default_caching: The default caching policy used for a buffer object
@@ -142,7 +141,6 @@ struct ttm_resource_manager {
*/
bool use_type;
bool use_tt;
- uint64_t size;
uint32_t available_caching;
uint32_t default_caching;
const struct ttm_resource_manager_func *func;
@@ -872,4 +870,22 @@ int ttm_range_man_fini(struct ttm_bo_device *bdev,
void ttm_resource_manager_debug(struct ttm_resource_manager *man,
struct drm_printer *p);
+/**
+ * ttm_range_man_adjust_size
+ *
+ * @man: manager to adjust size for
+ * @p_size: new size.
+ */
+void ttm_range_man_adjust_size(struct ttm_resource_manager *man,
+ unsigned long p_size);
+
+/**
+ * ttm_range_man_size
+ *
+ * @man: manager to get size for
+ * Returns:
+ * size of range manager.
+ */
+unsigned long ttm_range_man_size(struct ttm_resource_manager *man);
+
#endif
--
2.27.0
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel
^ permalink raw reply related [flat|nested] 4+ messages in thread* [PATCH 2/2] drm/ttm/nouveau: move io_lru storage into driver.
2020-08-06 4:39 [rfc] further ttm cleanups Dave Airlie
2020-08-06 4:39 ` [PATCH 1/2] drm/ttm: drop size from resource manager base class Dave Airlie
@ 2020-08-06 4:39 ` Dave Airlie
1 sibling, 0 replies; 4+ messages in thread
From: Dave Airlie @ 2020-08-06 4:39 UTC (permalink / raw)
To: dri-devel
From: Dave Airlie <airlied@redhat.com>
This moves the io lru tracking into the driver allocated structure.
Probably need to consider if we can move more stuff in there around the
nouveau only io_lru functionality.
---
drivers/gpu/drm/nouveau/nouveau_ttm.c | 24 +++++++++++++++++++++---
drivers/gpu/drm/ttm/ttm_bo.c | 11 ++++++++---
drivers/gpu/drm/ttm/ttm_bo_util.c | 24 +++++++++++++++---------
include/drm/ttm/ttm_bo_api.h | 3 +++
include/drm/ttm/ttm_bo_driver.h | 19 +++++++++++--------
5 files changed, 58 insertions(+), 23 deletions(-)
diff --git a/drivers/gpu/drm/nouveau/nouveau_ttm.c b/drivers/gpu/drm/nouveau/nouveau_ttm.c
index 2aba2e245548..5e0008d98b07 100644
--- a/drivers/gpu/drm/nouveau/nouveau_ttm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_ttm.c
@@ -31,6 +31,16 @@
#include <core/tegra.h>
+struct nouveau_resource_manager {
+ struct ttm_resource_manager manager;
+ struct ttm_resource_manager_io_lru io_lru;
+};
+
+static inline struct nouveau_resource_manager *to_mgr(struct ttm_resource_manager *man)
+{
+ return container_of(man, struct nouveau_resource_manager, manager);
+}
+
static void
nouveau_manager_del(struct ttm_resource_manager *man, struct ttm_resource *reg)
{
@@ -63,9 +73,15 @@ nouveau_vram_manager_new(struct ttm_resource_manager *man,
return 0;
}
+static struct ttm_resource_manager_io_lru *nouveau_resource_io_lru(struct ttm_resource_manager *man)
+{
+ return &to_mgr(man)->io_lru;
+}
+
const struct ttm_resource_manager_func nouveau_vram_manager = {
.get_node = nouveau_vram_manager_new,
.put_node = nouveau_manager_del,
+ .io_lru = nouveau_resource_io_lru,
};
static int
@@ -160,7 +176,8 @@ nouveau_ttm_init_vram(struct nouveau_drm *drm)
if (drm->client.device.info.family >= NV_DEVICE_INFO_V0_TESLA) {
/* Some BARs do not support being ioremapped WC */
const u8 type = mmu->type[drm->ttm.type_vram].type;
- struct ttm_resource_manager *man = kzalloc(sizeof(struct ttm_resource_manager), GFP_KERNEL);
+ struct nouveau_resource_manager *nman = kzalloc(sizeof(*nman), GFP_KERNEL);
+ struct ttm_resource_manager *man = &nman->manager;
if (!man)
return -ENOMEM;
@@ -173,9 +190,9 @@ nouveau_ttm_init_vram(struct nouveau_drm *drm)
}
man->func = &nouveau_vram_manager;
- man->use_io_reserve_lru = true;
ttm_resource_manager_init(man);
+ ttm_resource_manager_io_lru_init(&nman->io_lru);
ttm_set_driver_manager(&drm->ttm.bdev, TTM_PL_VRAM, man);
ttm_resource_manager_set_used(man, true);
return 0;
@@ -193,11 +210,12 @@ nouveau_ttm_fini_vram(struct nouveau_drm *drm)
struct ttm_resource_manager *man = ttm_manager_type(&drm->ttm.bdev, TTM_PL_VRAM);
if (drm->client.device.info.family >= NV_DEVICE_INFO_V0_TESLA) {
+ struct nouveau_resource_manager *nman = to_mgr(man);
ttm_resource_manager_set_used(man, false);
ttm_resource_manager_force_list_clean(&drm->ttm.bdev, man);
ttm_resource_manager_cleanup(man);
ttm_set_driver_manager(&drm->ttm.bdev, TTM_PL_VRAM, NULL);
- kfree(man);
+ kfree(nman);
} else
ttm_range_man_fini(&drm->ttm.bdev, TTM_PL_VRAM);
}
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index 74b7def25f91..e4aa6eb60a50 100644
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -1469,14 +1469,19 @@ int ttm_bo_evict_mm(struct ttm_bo_device *bdev, unsigned mem_type)
}
EXPORT_SYMBOL(ttm_bo_evict_mm);
+void ttm_resource_manager_io_lru_init(struct ttm_resource_manager_io_lru *io_lru)
+{
+ mutex_init(&io_lru->io_reserve_mutex);
+ INIT_LIST_HEAD(&io_lru->io_reserve_lru);
+}
+EXPORT_SYMBOL(ttm_resource_manager_io_lru_init);
+
void ttm_resource_manager_init(struct ttm_resource_manager *man)
{
unsigned i;
- man->use_io_reserve_lru = false;
- mutex_init(&man->io_reserve_mutex);
+
spin_lock_init(&man->move_lock);
- INIT_LIST_HEAD(&man->io_reserve_lru);
for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i)
INIT_LIST_HEAD(&man->lru[i]);
diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c
index 496158acd5b9..137aab6c53ae 100644
--- a/drivers/gpu/drm/ttm/ttm_bo_util.c
+++ b/drivers/gpu/drm/ttm/ttm_bo_util.c
@@ -93,29 +93,33 @@ EXPORT_SYMBOL(ttm_bo_move_ttm);
int ttm_mem_io_lock(struct ttm_resource_manager *man, bool interruptible)
{
- if (likely(!man->use_io_reserve_lru))
+ struct ttm_resource_manager_io_lru *io_lru;
+ if (!man->func && !man->func->io_lru)
return 0;
+ io_lru = man->func->io_lru(man);
if (interruptible)
- return mutex_lock_interruptible(&man->io_reserve_mutex);
+ return mutex_lock_interruptible(&io_lru->io_reserve_mutex);
- mutex_lock(&man->io_reserve_mutex);
+ mutex_lock(&io_lru->io_reserve_mutex);
return 0;
}
void ttm_mem_io_unlock(struct ttm_resource_manager *man)
{
- if (likely(!man->use_io_reserve_lru))
+ struct ttm_resource_manager_io_lru *io_lru;
+ if (!man->func && !man->func->io_lru)
return;
- mutex_unlock(&man->io_reserve_mutex);
+ io_lru = man->func->io_lru(man);
+ mutex_unlock(&io_lru->io_reserve_mutex);
}
static int ttm_mem_io_evict(struct ttm_resource_manager *man)
{
struct ttm_buffer_object *bo;
-
- bo = list_first_entry_or_null(&man->io_reserve_lru,
+ struct ttm_resource_manager_io_lru *io_lru = man->func->io_lru(man);
+ bo = list_first_entry_or_null(&io_lru->io_reserve_lru,
struct ttm_buffer_object,
io_reserve_lru);
if (!bo)
@@ -173,9 +177,11 @@ int ttm_mem_io_reserve_vm(struct ttm_buffer_object *bo)
if (unlikely(ret != 0))
return ret;
mem->bus.io_reserved_vm = true;
- if (man->use_io_reserve_lru)
+ if (man->func && man->func->io_lru) {
+ struct ttm_resource_manager_io_lru *io_lru = man->func->io_lru(man);
list_add_tail(&bo->io_reserve_lru,
- &man->io_reserve_lru);
+ &io_lru->io_reserve_lru);
+ }
return 0;
}
diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h
index 9570ed7d01bf..0861de24335c 100644
--- a/include/drm/ttm/ttm_bo_api.h
+++ b/include/drm/ttm/ttm_bo_api.h
@@ -56,6 +56,8 @@ struct ttm_lru_bulk_move;
struct ttm_resource_manager;
+struct ttm_resource_manager_io_lru;
+
/**
* struct ttm_bus_placement
*
@@ -542,6 +544,7 @@ int ttm_bo_create(struct ttm_bo_device *bdev, unsigned long size,
*/
void ttm_resource_manager_init(struct ttm_resource_manager *man);
+void ttm_resource_manager_io_lru_init(struct ttm_resource_manager_io_lru *io_lru);
/**
* ttm_bo_evict_mm
*
diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h
index dc57445f90e6..7408ee01b6f3 100644
--- a/include/drm/ttm/ttm_bo_driver.h
+++ b/include/drm/ttm/ttm_bo_driver.h
@@ -46,6 +46,7 @@
#define TTM_MAX_BO_PRIORITY 4U
struct ttm_resource_manager;
+struct ttm_resource_manager_io_lru;
struct ttm_resource_manager_func {
/**
@@ -106,6 +107,8 @@ struct ttm_resource_manager_func {
*/
void (*debug)(struct ttm_resource_manager *man,
struct drm_printer *printer);
+
+ struct ttm_resource_manager_io_lru *(*io_lru)(struct ttm_resource_manager *man);
};
/**
@@ -133,7 +136,14 @@ struct ttm_resource_manager_func {
* This structure is used to identify and manage memory types for a device.
*/
+struct ttm_resource_manager_io_lru {
+ /*
+ * Protected by @io_reserve_mutex:
+ */
+ struct list_head io_reserve_lru;
+ struct mutex io_reserve_mutex;
+};
struct ttm_resource_manager {
/*
@@ -144,15 +154,8 @@ struct ttm_resource_manager {
uint32_t available_caching;
uint32_t default_caching;
const struct ttm_resource_manager_func *func;
- struct mutex io_reserve_mutex;
- bool use_io_reserve_lru;
- spinlock_t move_lock;
- /*
- * Protected by @io_reserve_mutex:
- */
-
- struct list_head io_reserve_lru;
+ spinlock_t move_lock;
/*
* Protected by the global->lru_lock.
--
2.27.0
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel
^ permalink raw reply related [flat|nested] 4+ messages in thread