Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <ea08865e-bacb-f377-7ac0-c191045fde72@amd.com>
Date:   Mon, 14 Aug 2023 14:05:55 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.13.0
Subject: Re: [PATCH 1/3] drm/amdgpu: Add new api to switch on/off power
 profile mode
Content-Language: en-US
To:     Arvind Yadav <Arvind.Yadav@amd.com>, alexander.deucher@amd.com,
        shashank.sharma@amd.com, Xinhui.Pan@amd.com, airlied@gmail.com,
        daniel@ffwll.ch, amd-gfx@lists.freedesktop.org,
        dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org
References: <20230814073438.10682-1-Arvind.Yadav@amd.com>
 <20230814073438.10682-2-Arvind.Yadav@amd.com>
From:   =?UTF-8?Q?Christian_K=c3=b6nig?= <christian.koenig@amd.com>
In-Reply-To: <20230814073438.10682-2-Arvind.Yadav@amd.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
MIME-Version: 1.0
X-MS-Exchange-AntiSpam-MessageData-ChunkCount: 1
X-MS-Exchange-AntiSpam-MessageData-0: =?utf-8?B?c054RDlOY0RubWdpQ1VYQzBCaXNySUFPWUJYSDBvOVpOSlI5UDJLSjJFNFA1?=
 =?utf-8?B?bGszampVdnBUWmg0NXZSU2QyVXBpYm1qYi9QbEhSZ2NVeVp1K1BvLzViYXE2?=
 =?utf-8?B?clJnV2NtbGpya0kzcU0vc1ZtTEY5bDAvajVwdXg0c2tiZ3FBa1pJZDliaXNK?=
 =?utf-8?B?a05uT3RKWVFUUS9uUUxKT3Azc2RlR1Z6VUJORktnZDQ3dWc0RmtIeWhHQTU1?=
 =?utf-8?B?VUpkNnNTaVIxRGJGMjZvaGE2dTQ2VkJ1S2o4Nmhrd0M4WU01U20xL3NEWmd3?=
 =?utf-8?B?b05YaFNpYTNtN0d3WjNkKzEzOGduSHJXM1U1YmFuWG9YQ2VoQVZuWXRBS080?=
 =?utf-8?B?Q1IvYXpKU2hyYWpzZm9LNnF1T3YwTDRFZlowTUJYVndoQzlDYlZTcGN5Y1hG?=
 =?utf-8?B?NTdKN0hmY1ZaL3FtbkV2dUc4UWdTUmJmWEN6c1EzNVJpQXZUbjhsT054RUlJ?=
 =?utf-8?B?YW5nWi9JbVBJWllnMkFQM2w0OFMrcVVwYkNLZHc4SGxVaXV6bzJRL2FTL0Fv?=
 =?utf-8?B?ekNFNU9GMFNTMFYveWhPM3libUFZN2d6K2RRM09uR2wwZXlKaUhhY3k3TThU?=
 =?utf-8?B?T2tUS1c1ZmQ4NUE5RU5tYnJUVUgvR1N3dHZTeEZhamY0bXhqMjg0VzFxR1Ny?=
 =?utf-8?B?cjBZVGhpZTVWNHFoaExHUUJyeGJFNHRGcWo1ZWhoeXlGZVVhTmxCTzFyUUZ4?=
 =?utf-8?B?WTIrN00zRkhUbzIxRTloU1R5N1BHTVE2OUtXS0R5QUJZWWtnY2ZqK21HN0JP?=
 =?utf-8?B?MStKTlpCMkhWRE8ySGFEalluZDJwS0xCMm5kbGVwYUxsd2RKcEdHOEM1UXFV?=
 =?utf-8?B?aW5rSU9KUUxOQi8yeXBuUEpPWk9sY1NKTk1GZ2hXM21GbXhOaW0wMm9SQmhD?=
 =?utf-8?B?blZlVjBqMytYTkhWWlNJdWFrS2ViTzFadUt0OUdjK2VDNGZvSno5bEphbkRT?=
 =?utf-8?B?L3QwdnRwTXp5VDBOTkxubDlWblBJYWlwSERyUWRJK3RJZ3N3THAxeWJqcWhQ?=
 =?utf-8?B?Y01qN1FPWXBrVVFVK0lSRTUvOVFiYzZqZWR3cVJsc2ROY0ZOUm56aGlNcHhN?=
 =?utf-8?B?UDVtK0hsZVIyeXkzNFdhMWcydlp5ZEJPNDRDS05HZklLUTFmVWxJRjU3RGor?=
 =?utf-8?B?NFp5WDVRcWV6SW5ZR2NNMkp1ZWdpS241Z3k5MVRDNXIvd0hsclhtdmo5YlVC?=
 =?utf-8?B?TnlXQjJwL1lMYkpac3U2TVdsZitXK0hZV0ZmTjU5cXNwZnJRQ3BxNmltWi9S?=
 =?utf-8?B?WGFicUZIK0hXVUpvQjNVTi9jODdia1hxeDBYWnVJTHp5d21lNzA3UG1DNVNw?=
 =?utf-8?B?UXBCRkJZOHQ2dVFHVHNVVktHcnhNZm8xbzVNRGxESDlOQjVXUGJMNEgvcUdP?=
 =?utf-8?B?SWJjRmM5cEVTeHVwcXFLT050YjZhWkdBODh2K2V4ZEd2Mk5DZE1aZXRKU1M4?=
 =?utf-8?B?ZXZxbUxNR3gxM3Fvd24yaFdnRWRLdG9JVFZib3MwZVM5dXRHYTlwSEpYQnJo?=
 =?utf-8?B?Y3RRRUMrUStMdUpyTEFCbjMwNnBzTS91V3hrbmYrYUZQU213L3htWkc3Y1R1?=
 =?utf-8?B?VG9TYTh2OVhyeXp5WHRlMCtiVVdNUXM1aHNYdmlER29KQXBjeHAycXhKYUQx?=
 =?utf-8?B?ZThPcitKOWM4TnpMbEJmdC83L1RlckJzeDRhanArTUE1UXp0SWNRc0QzZHJr?=
 =?utf-8?B?emN1T3o3b0ZyZ1Y2UDFtNnhMdlRVSXZKVG5yUjNqWU1KQ2dTU3RCd1JwWHJa?=
 =?utf-8?B?VVVIdGdzNjdrdXhuM3JXMFcycENCdnVmUUZxSVk4bjFaZnBZaWtGUkNOc0Qr?=
 =?utf-8?B?aE1YbU9CZTRDTjkrWDErZVdkek5lVU93WkJzd1dxOGFoeVVZT3hrV2NoT2Rs?=
 =?utf-8?B?R0R2MlErT3RrYXRZSGxPZE83ak9wckJuY2puMFB2TXVTYzA5V0d1aDlyYUpO?=
 =?utf-8?B?NjJjYlJIZDI4MmxJSHl5Qm9maDBiWUdUOC9FdGtuQytTZXBqaThQT2Q1Y3NB?=
 =?utf-8?B?SndzcUcvbElzRXlCTEpjbzZMdkRucHZUVlNHMW1CTkkvcU16N0tleExuTmdD?=
 =?utf-8?B?UkI5NXNPdG5TTTkxWG5XMkw5NkZrdkFFd0NJK0pFWnBrSU5saklTdXhMSUti?=
 =?utf-8?B?OTArczhBUlBMM3VXVU9IL1NpNUhsTUMyVUpvKzUvVXJHTHZHemZIa3VvdXVS?=
 =?utf-8?Q?02J4IqxCG2cWchYjoHxKw950+q7Xp1XZm7sBxmurFbnd?=
X-OriginatorOrg: amd.com
X-MS-Exchange-CrossTenant-Network-Message-Id: b27351cb-26a4-4913-9c4f-08db9cbed370
X-MS-Exchange-CrossTenant-AuthSource: BN8PR12MB3587.namprd12.prod.outlook.com
X-MS-Exchange-CrossTenant-AuthAs: Internal
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 14 Aug 2023 12:06:01.0160
 (UTC)
X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted
X-MS-Exchange-CrossTenant-Id: 3dd8961f-e488-4e60-8e11-a82d994e183d
X-MS-Exchange-CrossTenant-MailboxType: HOSTED
X-MS-Exchange-CrossTenant-UserPrincipalName: o/RZ9a98fksw4cljIznyD4x1u4zHKLsLLCK2KfRqRRV5nZcvcXtJ+1GaTVHwoySG
X-MS-Exchange-Transport-CrossTenantHeadersStamped: SJ2PR12MB8064
X-Spam-Status: No, score=-3.4 required=5.0 tests=BAYES_00,DKIMWL_WL_HIGH,
        DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,DKIM_VALID_EF,FORGED_SPF_HELO,
        NICE_REPLY_A,RCVD_IN_DNSWL_BLOCKED,RCVD_IN_MSPIKE_H2,SPF_HELO_PASS,
        SPF_NONE autolearn=ham autolearn_force=no version=3.4.6
X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on
        lindbergh.monkeyblade.net
Precedence: bulk
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

Am 14.08.23 um 09:34 schrieb Arvind Yadav:
> This patch adds a function which will allow to
> change the GPU power profile based on a submitted job.
> This can optimize the power performance when the
> workload is on.
>
> Cc: Shashank Sharma <shashank.sharma@amd.com>
> Cc: Christian Koenig <christian.koenig@amd.com>
> Cc: Alex Deucher <alexander.deucher@amd.com>
> Signed-off-by: Arvind Yadav <Arvind.Yadav@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/Makefile           |   2 +-
>   drivers/gpu/drm/amd/amdgpu/amdgpu.h           |   3 +
>   drivers/gpu/drm/amd/amdgpu/amdgpu_device.c    |   2 +
>   drivers/gpu/drm/amd/amdgpu/amdgpu_workload.c  | 156 ++++++++++++++++++
>   drivers/gpu/drm/amd/include/amdgpu_workload.h |  44 +++++
>   5 files changed, 206 insertions(+), 1 deletion(-)
>   create mode 100644 drivers/gpu/drm/amd/amdgpu/amdgpu_workload.c
>   create mode 100644 drivers/gpu/drm/amd/include/amdgpu_workload.h
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile
> index 415a7fa395c4..6a9e187d61e1 100644
> --- a/drivers/gpu/drm/amd/amdgpu/Makefile
> +++ b/drivers/gpu/drm/amd/amdgpu/Makefile
> @@ -60,7 +60,7 @@ amdgpu-y += amdgpu_device.o amdgpu_kms.o \
>   	amdgpu_umc.o smu_v11_0_i2c.o amdgpu_fru_eeprom.o amdgpu_rap.o \
>   	amdgpu_fw_attestation.o amdgpu_securedisplay.o \
>   	amdgpu_eeprom.o amdgpu_mca.o amdgpu_psp_ta.o amdgpu_lsdma.o \
> -	amdgpu_ring_mux.o
> +	amdgpu_ring_mux.o amdgpu_workload.o
>   
>   amdgpu-$(CONFIG_PROC_FS) += amdgpu_fdinfo.o
>   
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> index 02b827785e39..1939fa1af8a6 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> @@ -107,6 +107,7 @@
>   #include "amdgpu_fdinfo.h"
>   #include "amdgpu_mca.h"
>   #include "amdgpu_ras.h"
> +#include "amdgpu_workload.h"
>   
>   #define MAX_GPU_INSTANCE		16
>   
> @@ -1050,6 +1051,8 @@ struct amdgpu_device {
>   
>   	bool                            job_hang;
>   	bool                            dc_enabled;
> +
> +	struct amdgpu_smu_workload	smu_workload;
>   };
>   
>   static inline struct amdgpu_device *drm_to_adev(struct drm_device *ddev)
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index 5c7d40873ee2..0ec18b8fe29f 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -3672,6 +3672,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
>   
>   	INIT_WORK(&adev->xgmi_reset_work, amdgpu_device_xgmi_reset_func);
>   
> +	amdgpu_smu_workload_init(adev);
> +
>   	adev->gfx.gfx_off_req_count = 1;
>   	adev->gfx.gfx_off_residency = 0;
>   	adev->gfx.gfx_off_entrycount = 0;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_workload.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_workload.c
> new file mode 100644
> index 000000000000..ce0339d75c12
> --- /dev/null
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_workload.c
> @@ -0,0 +1,156 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright 2023 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 "amdgpu.h"
> +
> +/* 100 millsecond timeout */
> +#define SMU_IDLE_TIMEOUT	msecs_to_jiffies(100)
> +
> +static enum PP_SMC_POWER_PROFILE
> +ring_to_power_profile(uint32_t ring_type)
> +{
> +	switch (ring_type) {
> +	case AMDGPU_RING_TYPE_GFX:
> +		return PP_SMC_POWER_PROFILE_FULLSCREEN3D;
> +	case AMDGPU_RING_TYPE_COMPUTE:
> +		return PP_SMC_POWER_PROFILE_COMPUTE;
> +	case AMDGPU_RING_TYPE_UVD:
> +	case AMDGPU_RING_TYPE_VCE:
> +	case AMDGPU_RING_TYPE_UVD_ENC:
> +	case AMDGPU_RING_TYPE_VCN_DEC:
> +	case AMDGPU_RING_TYPE_VCN_ENC:
> +	case AMDGPU_RING_TYPE_VCN_JPEG:
> +		return PP_SMC_POWER_PROFILE_VIDEO;
> +	default:
> +		return PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT;
> +	}
> +}
> +
> +static void
> +amdgpu_power_profile_set(struct amdgpu_device *adev,
> +			 enum PP_SMC_POWER_PROFILE profile)
> +{
> +	int ret = amdgpu_dpm_switch_power_profile(adev, profile, true);
> +
> +	if (ret == 0) {

Please double check the coding style, stuff like this is usually 
complained about by the automated checkers.

Apart from that Alex needs to take a look if this here makes sense or not.

Regards,
Christian.

> +		/* Set the bit for the submitted workload profile */
> +		adev->smu_workload.submit_workload_status |= (1 << profile);
> +		atomic_inc(&adev->smu_workload.power_profile_ref[profile]);
> +	} else {
> +		DRM_ERROR("Failed to set power profile, error %d\n", ret);
> +	}
> +
> +}
> +
> +static void
> +amdgpu_power_profile_clear(struct amdgpu_device *adev,
> +			   enum PP_SMC_POWER_PROFILE profile)
> +{
> +	int ret = amdgpu_dpm_switch_power_profile(adev, profile, false);
> +
> +	if (ret == 0) {
> +		 /* Clear the bit for the submitted workload profile */
> +		adev->smu_workload.submit_workload_status &= ~(1 << profile);
> +	} else
> +		DRM_ERROR("Failed to clear power profile, error %d\n", ret);
> +
> +}
> +
> +static void amdgpu_smu_idle_work_handler(struct work_struct *work)
> +{
> +
> +	struct amdgpu_smu_workload *wl = container_of(work,
> +						      struct amdgpu_smu_workload,
> +						      smu_delayed_work.work);
> +	struct amdgpu_device *adev = wl->adev;
> +	bool reschedule = false;
> +
> +	mutex_lock(&adev->smu_workload.workload_lock);
> +	for (int index  = fls(adev->smu_workload.submit_workload_status);
> +	     index >= 0; index--) {
> +		if (!atomic_read(&adev->smu_workload.power_profile_ref[index]) &&
> +		    adev->smu_workload.submit_workload_status & (1 << index)) {
> +			amdgpu_power_profile_clear(adev, index);
> +		} else if (atomic_read(&adev->smu_workload.power_profile_ref[index]))
> +			reschedule = true;
> +	}
> +
> +	if (reschedule)
> +		schedule_delayed_work(&adev->smu_workload.smu_delayed_work,
> +				      SMU_IDLE_TIMEOUT);
> +
> +	mutex_unlock(&adev->smu_workload.workload_lock);
> +}
> +
> +void amdgpu_put_workload_profile(struct amdgpu_device *adev,
> +				 uint32_t ring_type)
> +{
> +
> +	enum PP_SMC_POWER_PROFILE profile = ring_to_power_profile(ring_type);
> +
> +	if (profile == PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT)
> +		return;
> +
> +	mutex_lock(&adev->smu_workload.workload_lock);
> +	atomic_dec(&adev->smu_workload.power_profile_ref[profile]);
> +	schedule_delayed_work(&adev->smu_workload.smu_delayed_work, SMU_IDLE_TIMEOUT);
> +	mutex_unlock(&adev->smu_workload.workload_lock);
> +}
> +
> +void amdgpu_set_workload_profile(struct amdgpu_device *adev,
> +				 uint32_t ring_type)
> +{
> +	enum PP_SMC_POWER_PROFILE profile = ring_to_power_profile(ring_type);
> +
> +	if (profile == PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT)
> +		return;
> +
> +	mutex_lock(&adev->smu_workload.workload_lock);
> +	cancel_delayed_work_sync(&adev->smu_workload.smu_delayed_work);
> +
> +	amdgpu_power_profile_set(adev, profile);
> +
> +	/* Clear the already finished jobs of higher power profile*/
> +	for (int index = fls(adev->smu_workload.submit_workload_status);
> +	     index > profile; index--) {
> +		if (!atomic_read(&adev->smu_workload.power_profile_ref[index]) &&
> +		    adev->smu_workload.submit_workload_status & (1 << index)) {
> +			amdgpu_power_profile_clear(adev, index);
> +		}
> +	}
> +
> +	mutex_unlock(&adev->smu_workload.workload_lock);
> +}
> +
> +void amdgpu_smu_workload_init(struct amdgpu_device *adev)
> +{
> +	struct amdgpu_smu_workload wl;
> +
> +	wl.adev = adev;
> +	wl.submit_workload_status = 0;
> +	adev->smu_workload = wl;
> +
> +	mutex_init(&adev->smu_workload.workload_lock);
> +	INIT_DELAYED_WORK(&adev->smu_workload.smu_delayed_work, amdgpu_smu_idle_work_handler);
> +}
> diff --git a/drivers/gpu/drm/amd/include/amdgpu_workload.h b/drivers/gpu/drm/amd/include/amdgpu_workload.h
> new file mode 100644
> index 000000000000..09804c3d2869
> --- /dev/null
> +++ b/drivers/gpu/drm/amd/include/amdgpu_workload.h
> @@ -0,0 +1,44 @@
> +/* SPDX-License-Identifier: MIT */
> +/*
> + * Copyright 2023 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.
> + *
> + */
> +
> +#ifndef _AMDGPU_WORKLOAD_H_
> +#define _AMDGPU_WORKLOAD_H_
> +
> +struct amdgpu_smu_workload {
> +	struct amdgpu_device	*adev;
> +	struct mutex		workload_lock;
> +	struct delayed_work	smu_delayed_work;
> +	uint32_t		submit_workload_status;
> +	atomic_t		power_profile_ref[PP_SMC_POWER_PROFILE_COUNT];
> +};
> +
> +void amdgpu_set_workload_profile(struct amdgpu_device *adev,
> +				 uint32_t ring_type);
> +
> +void amdgpu_put_workload_profile(struct amdgpu_device *adev,
> +				 uint32_t ring_type);
> +
> +void amdgpu_smu_workload_init(struct amdgpu_device *adev);
> +
> +#endif