Commit fb30fc59 authored by Shaoyun Liu's avatar Shaoyun Liu Committed by Alex Deucher

drm/amdgpu : Generate XGMI topology info from driver level

Driver will save an array of XGMI hive info, each hive will have a list of devices
that have the same hive ID.
Signed-off-by: default avatarShaoyun Liu <Shaoyun.Liu@amd.com>
Reviewed-by: default avatarFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: default avatarAlex Deucher <alexander.deucher@amd.com>
parent 78122127
...@@ -53,7 +53,7 @@ amdgpu-y += amdgpu_device.o amdgpu_kms.o \ ...@@ -53,7 +53,7 @@ amdgpu-y += amdgpu_device.o amdgpu_kms.o \
amdgpu_ucode.o amdgpu_bo_list.o amdgpu_ctx.o amdgpu_sync.o \ amdgpu_ucode.o amdgpu_bo_list.o amdgpu_ctx.o amdgpu_sync.o \
amdgpu_gtt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o amdgpu_atomfirmware.o \ amdgpu_gtt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o amdgpu_atomfirmware.o \
amdgpu_vf_error.o amdgpu_sched.o amdgpu_debugfs.o amdgpu_ids.o \ amdgpu_vf_error.o amdgpu_sched.o amdgpu_debugfs.o amdgpu_ids.o \
amdgpu_gmc.o amdgpu_gmc.o amdgpu_xgmi.o
# add asic specific block # add asic specific block
amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \ amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \
......
...@@ -1216,6 +1216,12 @@ void amdgpu_disable_vblank_kms(struct drm_device *dev, unsigned int pipe); ...@@ -1216,6 +1216,12 @@ void amdgpu_disable_vblank_kms(struct drm_device *dev, unsigned int pipe);
long amdgpu_kms_compat_ioctl(struct file *filp, unsigned int cmd, long amdgpu_kms_compat_ioctl(struct file *filp, unsigned int cmd,
unsigned long arg); unsigned long arg);
/*
* functions used by amdgpu_xgmi.c
*/
int amdgpu_xgmi_add_device(struct amdgpu_device *adev);
/* /*
* functions used by amdgpu_encoder.c * functions used by amdgpu_encoder.c
*/ */
......
...@@ -1586,6 +1586,7 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) ...@@ -1586,6 +1586,7 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
adev->ip_blocks[i].status.hw = true; adev->ip_blocks[i].status.hw = true;
} }
amdgpu_xgmi_add_device(adev);
amdgpu_amdkfd_device_init(adev); amdgpu_amdkfd_device_init(adev);
if (amdgpu_sriov_vf(adev)) if (amdgpu_sriov_vf(adev))
......
...@@ -97,6 +97,8 @@ struct amdgpu_xgmi { ...@@ -97,6 +97,8 @@ struct amdgpu_xgmi {
unsigned physical_node_id; unsigned physical_node_id;
/* number of nodes (0-4) */ /* number of nodes (0-4) */
unsigned num_physical_nodes; unsigned num_physical_nodes;
/* gpu list in the same hive */
struct list_head head;
}; };
struct amdgpu_gmc { struct amdgpu_gmc {
......
/*
* Copyright 2018 Advanced Micro Devices, Inc.
*
* 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.
*
*
*/
#include <linux/list.h>
#include "amdgpu.h"
#include "amdgpu_psp.h"
static DEFINE_MUTEX(xgmi_mutex);
#define AMDGPU_MAX_XGMI_HIVE 8
#define AMDGPU_MAX_XGMI_DEVICE_PER_HIVE 4
struct amdgpu_hive_info {
uint64_t hive_id;
struct list_head device_list;
};
static struct amdgpu_hive_info xgmi_hives[AMDGPU_MAX_XGMI_HIVE];
static unsigned hive_count = 0;
static struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev)
{
int i;
struct amdgpu_hive_info *tmp;
if (!adev->gmc.xgmi.hive_id)
return NULL;
for (i = 0 ; i < hive_count; ++i) {
tmp = &xgmi_hives[i];
if (tmp->hive_id == adev->gmc.xgmi.hive_id)
return tmp;
}
if (i >= AMDGPU_MAX_XGMI_HIVE)
return NULL;
/* initialize new hive if not exist */
tmp = &xgmi_hives[hive_count++];
tmp->hive_id = adev->gmc.xgmi.hive_id;
INIT_LIST_HEAD(&tmp->device_list);
return tmp;
}
int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
{
struct psp_xgmi_topology_info tmp_topology[AMDGPU_MAX_XGMI_DEVICE_PER_HIVE];
struct amdgpu_hive_info *hive;
struct amdgpu_xgmi *entry;
struct amdgpu_device *tmp_adev;
int count = 0, ret = -EINVAL;
if ((adev->asic_type < CHIP_VEGA20) ||
(adev->flags & AMD_IS_APU) )
return 0;
adev->gmc.xgmi.device_id = psp_xgmi_get_device_id(&adev->psp);
adev->gmc.xgmi.hive_id = psp_xgmi_get_hive_id(&adev->psp);
memset(&tmp_topology[0], 0, sizeof(tmp_topology));
mutex_lock(&xgmi_mutex);
hive = amdgpu_get_xgmi_hive(adev);
if (!hive)
goto exit;
list_add_tail(&adev->gmc.xgmi.head, &hive->device_list);
list_for_each_entry(entry, &hive->device_list, head)
tmp_topology[count++].device_id = entry->device_id;
ret = psp_xgmi_get_topology_info(&adev->psp, count, tmp_topology);
if (ret) {
dev_err(adev->dev,
"XGMI: Get topology failure on device %llx, hive %llx, ret %d",
adev->gmc.xgmi.device_id,
adev->gmc.xgmi.hive_id, ret);
goto exit;
}
/* Each psp need to set the latest topology */
list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
ret = psp_xgmi_set_topology_info(&tmp_adev->psp, count, tmp_topology);
if (ret) {
dev_err(tmp_adev->dev,
"XGMI: Set topology failure on device %llx, hive %llx, ret %d",
tmp_adev->gmc.xgmi.device_id,
tmp_adev->gmc.xgmi.hive_id, ret);
/* To do : continue with some node failed or disable the whole hive */
break;
}
}
if (!ret)
dev_info(adev->dev, "XGMI: Add node %d to hive 0x%llx.\n",
adev->gmc.xgmi.physical_node_id,
adev->gmc.xgmi.hive_id);
exit:
mutex_unlock(&xgmi_mutex);
return ret;
}
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment