From: "Sharma, Shashank" <shashank.sharma@amd.com>
To: "Christian König" <christian.koenig@amd.com>,
dri-devel@lists.freedesktop.org, amd-gfx@lists.freedesktop.org
Cc: Alex Deucher <alexander.deucher@amd.com>,
Marek Olsak <marek.olsak@amd.com>,
Amarnath Somalapuram <amaranath.somalapuram@amd.com>
Subject: Re: [PATCH] amdgpu: add context creation flags in CS IOCTL
Date: Mon, 8 Aug 2022 13:46:25 +0200 [thread overview]
Message-ID: <57c92713-b06a-e68a-9079-72f44c2ca3b9@amd.com> (raw)
In-Reply-To: <acc970f2-632f-0372-8082-6bd54f0c7cf5@amd.com>
On 8/8/2022 12:59 PM, Christian König wrote:
> Am 02.08.22 um 15:55 schrieb Shashank Sharma:
>> This patch adds:
>> - A new input parameter "flags" in the amdgpu_ctx_create2 call.
>> - Some new flags defining workload type hints.
>> - Some change in the caller function of amdgpu_ctx_create2, to
>> accomodate this new parameter.
>>
>> The idea is to pass the workload hints while context creation,
>
> Bad idea.
>
> Please take AMDGPU_CTX_OP_SET_STABLE_PSTATE and
> AMDGPU_CTX_OP_GET_STABLE_PSTATE as blueprint for this and don't add
> extra flags to the context creation.
>
> Regards,
> Christian.
Hey Christian,
Noted, let me have a look at AMDGPU_CTX_OP_GET/SET_STABLE_PSTATE
implementation.
- Shashank
>
>
>> so
>> that kernel GPU scheduler can pass this information to GPU FW, which in
>> turn can adjust the GPU characterstics as per the workload type.
>>
>> Signed-off-by: Shashank Sharma <shashank.sharma@amd.com>
>> Cc: Alex Deucher <alexander.deucher@amd.com>
>> Cc: Marek Olsak <marek.olsak@amd.com>
>> Cc: Christian Koenig <christian.koenig@amd.com>
>> Cc: Amarnath Somalapuram <amaranath.somalapuram@amd.com>
>> ---
>> amdgpu/amdgpu.h | 2 ++
>> amdgpu/amdgpu_cs.c | 5 ++++-
>> include/drm/amdgpu_drm.h | 10 +++++++++-
>> 3 files changed, 15 insertions(+), 2 deletions(-)
>>
>> diff --git a/amdgpu/amdgpu.h b/amdgpu/amdgpu.h
>> index b118dd48..1ebb46e6 100644
>> --- a/amdgpu/amdgpu.h
>> +++ b/amdgpu/amdgpu.h
>> @@ -874,6 +874,7 @@ int amdgpu_bo_list_update(amdgpu_bo_list_handle
>> handle,
>> *
>> * \param dev - \c [in] Device handle. See
>> #amdgpu_device_initialize()
>> * \param priority - \c [in] Context creation flags. See
>> AMDGPU_CTX_PRIORITY_*
>> + * \param flags - \c [in] Context flags. See AMDGPU_CTX_FLAGS_*
>> * \param context - \c [out] GPU Context handle
>> *
>> * \return 0 on success\n
>> @@ -884,6 +885,7 @@ int amdgpu_bo_list_update(amdgpu_bo_list_handle
>> handle,
>> */
>> int amdgpu_cs_ctx_create2(amdgpu_device_handle dev,
>> uint32_t priority,
>> + uint32_t flags,
>> amdgpu_context_handle *context);
>> /**
>> * Create GPU execution Context
>> diff --git a/amdgpu/amdgpu_cs.c b/amdgpu/amdgpu_cs.c
>> index fad484bf..d4723ea5 100644
>> --- a/amdgpu/amdgpu_cs.c
>> +++ b/amdgpu/amdgpu_cs.c
>> @@ -44,12 +44,14 @@ static int
>> amdgpu_cs_reset_sem(amdgpu_semaphore_handle sem);
>> *
>> * \param dev - \c [in] Device handle. See
>> #amdgpu_device_initialize()
>> * \param priority - \c [in] Context creation flags. See
>> AMDGPU_CTX_PRIORITY_*
>> + * \param flags - \c [in] Context flags. See AMDGPU_CTX_FLAGS_*
>> * \param context - \c [out] GPU Context handle
>> *
>> * \return 0 on success otherwise POSIX Error code
>> */
>> drm_public int amdgpu_cs_ctx_create2(amdgpu_device_handle dev,
>> uint32_t priority,
>> + uint32_t flags,
>> amdgpu_context_handle *context)
>> {
>> struct amdgpu_context *gpu_context;
>> @@ -74,6 +76,7 @@ drm_public int
>> amdgpu_cs_ctx_create2(amdgpu_device_handle dev,
>> memset(&args, 0, sizeof(args));
>> args.in.op = AMDGPU_CTX_OP_ALLOC_CTX;
>> args.in.priority = priority;
>> + args.in.flags = flags;
>> r = drmCommandWriteRead(dev->fd, DRM_AMDGPU_CTX, &args,
>> sizeof(args));
>> if (r)
>> @@ -97,7 +100,7 @@ error:
>> drm_public int amdgpu_cs_ctx_create(amdgpu_device_handle dev,
>> amdgpu_context_handle *context)
>> {
>> - return amdgpu_cs_ctx_create2(dev, AMDGPU_CTX_PRIORITY_NORMAL,
>> context);
>> + return amdgpu_cs_ctx_create2(dev, AMDGPU_CTX_PRIORITY_NORMAL, 0,
>> context);
>> }
>> /**
>> diff --git a/include/drm/amdgpu_drm.h b/include/drm/amdgpu_drm.h
>> index 0cbd1540..d9fb1f20 100644
>> --- a/include/drm/amdgpu_drm.h
>> +++ b/include/drm/amdgpu_drm.h
>> @@ -238,10 +238,18 @@ union drm_amdgpu_bo_list {
>> #define AMDGPU_CTX_PRIORITY_HIGH 512
>> #define AMDGPU_CTX_PRIORITY_VERY_HIGH 1023
>> +/* GPU context workload hint bitmask */
>> +#define AMDGPU_CTX_FLAGS_WORKLOAD_HINT_MASK 0xFF
>> +#define AMDGPU_CTX_FLAGS_WORKLOAD_HINT_NONE 0
>> +#define AMDGPU_CTX_FLAGS_WORKLOAD_HINT_3D (1 << 1)
>> +#define AMDGPU_CTX_FLAGS_WORKLOAD_HINT_VIDEO (1 << 2)
>> +#define AMDGPU_CTX_FLAGS_WORKLOAD_HINT_VR (1 << 3)
>> +#define AMDGPU_CTX_FLAGS_WORKLOAD_HINT_COMPUTE (1 << 4)
>> +
>> struct drm_amdgpu_ctx_in {
>> /** AMDGPU_CTX_OP_* */
>> __u32 op;
>> - /** For future use, no flags defined so far */
>> + /** AMDGPU_CTX_FLAGS_* */
>> __u32 flags;
>> __u32 ctx_id;
>> /** AMDGPU_CTX_PRIORITY_* */
>
prev parent reply other threads:[~2022-08-08 11:46 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-02 13:55 [PATCH] amdgpu: add context creation flags in CS IOCTL Shashank Sharma
2022-08-02 15:58 ` Michel Dänzer
2022-08-02 16:00 ` Sharma, Shashank
2022-08-08 6:51 ` Somalapuram, Amaranath
2022-08-08 10:59 ` Christian König
2022-08-08 11:46 ` Sharma, Shashank [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=57c92713-b06a-e68a-9079-72f44c2ca3b9@amd.com \
--to=shashank.sharma@amd.com \
--cc=alexander.deucher@amd.com \
--cc=amaranath.somalapuram@amd.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=christian.koenig@amd.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=marek.olsak@amd.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox