AMD-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Alex Deucher <alexander.deucher@amd.com>
To: <amd-gfx@lists.freedesktop.org>
Cc: "Alex Deucher" <alexander.deucher@amd.com>,
	"Jack Xiao" <Jack.Xiao@amd.com>,
	"Christian König" <christian.koenig@amd.com>,
	"Hawking Zhang" <Hawking.Zhang@amd.com>
Subject: [PATCH 39/73] drm/amdgpu/mes: manage mes doorbell allocation
Date: Fri, 29 Apr 2022 13:45:50 -0400	[thread overview]
Message-ID: <20220429174624.459475-40-alexander.deucher@amd.com> (raw)
In-Reply-To: <20220429174624.459475-1-alexander.deucher@amd.com>

From: Jack Xiao <Jack.Xiao@amd.com>

It is used to manage the doorbell allocation of mes processes and queues.
Driver calls into process doorbell allocation to get the slice doorbell
for the process, then the doorbell for a queue is allocated from the
process doorbell slice.

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/Makefile     |   1 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c | 133 ++++++++++++++++++++++++
 2 files changed, 134 insertions(+)
 create mode 100644 drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c

diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile
index 7a1b13fabebb..803e7f5dc458 100644
--- a/drivers/gpu/drm/amd/amdgpu/Makefile
+++ b/drivers/gpu/drm/amd/amdgpu/Makefile
@@ -144,6 +144,7 @@ amdgpu-y += \
 
 # add MES block
 amdgpu-y += \
+	amdgpu_mes.o \
 	mes_v10_1.o
 
 # add UVD block
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
new file mode 100644
index 000000000000..1c591cb45fd9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
@@ -0,0 +1,133 @@
+/*
+ * Copyright 2019 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_mes.h"
+#include "amdgpu.h"
+#include "soc15_common.h"
+#include "amdgpu_mes_ctx.h"
+
+#define AMDGPU_MES_MAX_NUM_OF_QUEUES_PER_PROCESS 1024
+#define AMDGPU_ONE_DOORBELL_SIZE 8
+
+static int amdgpu_mes_doorbell_process_slice(struct amdgpu_device *adev)
+{
+	return roundup(AMDGPU_ONE_DOORBELL_SIZE *
+		       AMDGPU_MES_MAX_NUM_OF_QUEUES_PER_PROCESS,
+		       PAGE_SIZE);
+}
+
+static int amdgpu_mes_alloc_process_doorbells(struct amdgpu_device *adev,
+				      struct amdgpu_mes_process *process)
+{
+	int r = ida_simple_get(&adev->mes.doorbell_ida, 2,
+			       adev->mes.max_doorbell_slices,
+			       GFP_KERNEL);
+	if (r > 0)
+		process->doorbell_index = r;
+
+	return r;
+}
+
+static void amdgpu_mes_free_process_doorbells(struct amdgpu_device *adev,
+				      struct amdgpu_mes_process *process)
+{
+	if (process->doorbell_index)
+		ida_simple_remove(&adev->mes.doorbell_ida,
+				  process->doorbell_index);
+}
+
+static int amdgpu_mes_queue_doorbell_get(struct amdgpu_device *adev,
+					 struct amdgpu_mes_process *process,
+					 int ip_type, uint64_t *doorbell_index)
+{
+	unsigned int offset, found;
+
+	if (ip_type == AMDGPU_RING_TYPE_SDMA) {
+		offset = adev->doorbell_index.sdma_engine[0];
+		found = find_next_zero_bit(process->doorbell_bitmap,
+					   AMDGPU_MES_MAX_NUM_OF_QUEUES_PER_PROCESS,
+					   offset);
+	} else {
+		found = find_first_zero_bit(process->doorbell_bitmap,
+					    AMDGPU_MES_MAX_NUM_OF_QUEUES_PER_PROCESS);
+	}
+
+	if (found >= AMDGPU_MES_MAX_NUM_OF_QUEUES_PER_PROCESS) {
+		DRM_WARN("No doorbell available\n");
+		return -ENOSPC;
+	}
+
+	set_bit(found, process->doorbell_bitmap);
+
+	*doorbell_index =
+		(process->doorbell_index *
+		 amdgpu_mes_doorbell_process_slice(adev)) / sizeof(u32) +
+		found * 2;
+
+	return 0;
+}
+
+static void amdgpu_mes_queue_doorbell_free(struct amdgpu_device *adev,
+					   struct amdgpu_mes_process *process,
+					   uint32_t doorbell_index)
+{
+	unsigned int old, doorbell_id;
+
+	doorbell_id = doorbell_index -
+		(process->doorbell_index *
+		 amdgpu_mes_doorbell_process_slice(adev)) / sizeof(u32);
+	doorbell_id /= 2;
+
+	old = test_and_clear_bit(doorbell_id, process->doorbell_bitmap);
+	WARN_ON(!old);
+}
+
+static int amdgpu_mes_doorbell_init(struct amdgpu_device *adev)
+{
+	size_t doorbell_start_offset;
+	size_t doorbell_aperture_size;
+	size_t doorbell_process_limit;
+
+	doorbell_start_offset = (adev->doorbell_index.max_assignment+1) * sizeof(u32);
+	doorbell_start_offset =
+		roundup(doorbell_start_offset,
+			amdgpu_mes_doorbell_process_slice(adev));
+
+	doorbell_aperture_size = adev->doorbell.size;
+	doorbell_aperture_size =
+			rounddown(doorbell_aperture_size,
+				  amdgpu_mes_doorbell_process_slice(adev));
+
+	if (doorbell_aperture_size > doorbell_start_offset)
+		doorbell_process_limit =
+			(doorbell_aperture_size - doorbell_start_offset) /
+			amdgpu_mes_doorbell_process_slice(adev);
+	else
+		return -ENOSPC;
+
+	adev->mes.doorbell_id_offset = doorbell_start_offset / sizeof(u32);
+	adev->mes.max_doorbell_slices = doorbell_process_limit;
+
+	DRM_INFO("max_doorbell_slices=%ld\n", doorbell_process_limit);
+	return 0;
+}
-- 
2.35.1


  parent reply	other threads:[~2022-04-29 17:47 UTC|newest]

Thread overview: 74+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-04-29 17:45 [PATCH 00/73] MES support Alex Deucher
2022-04-29 17:45 ` [PATCH 01/73] drm/amdgpu: define MQD abstract layer for hw ip Alex Deucher
2022-04-29 17:45 ` [PATCH 02/73] drm/amdgpu: add helper function to initialize mqd from ring v4 Alex Deucher
2022-04-29 17:45 ` [PATCH 03/73] drm/amdgpu: add the per-context meta data v3 Alex Deucher
2022-04-29 17:45 ` [PATCH 04/73] drm/amdgpu: add mes ctx data in amdgpu_ring Alex Deucher
2022-04-29 17:45 ` [PATCH 05/73] drm/amdgpu: define ring structure to access rptr/wptr/fence Alex Deucher
2022-04-29 17:45 ` [PATCH 06/73] drm/amdgpu: use ring structure to access rptr/wptr v2 Alex Deucher
2022-04-29 17:45 ` [PATCH 07/73] drm/amdgpu: initialize/finalize the ring for mes queue Alex Deucher
2022-04-29 17:45 ` [PATCH 08/73] drm/amdgpu: assign the cpu/gpu address of fence from ring Alex Deucher
2022-04-29 17:45 ` [PATCH 09/73] drm/amdgpu/gfx10: implement mqd functions of gfx/compute eng v2 Alex Deucher
2022-04-29 17:45 ` [PATCH 10/73] drm/amdgpu/gfx10: use per ctx CSA for ce metadata Alex Deucher
2022-04-29 17:45 ` [PATCH 11/73] drm/amdgpu/gfx10: use per ctx CSA for de metadata Alex Deucher
2022-04-29 17:45 ` [PATCH 12/73] drm/amdgpu/gfx10: associate mes queue id with fence v2 Alex Deucher
2022-04-29 17:45 ` [PATCH 13/73] drm/amdgpu/gfx10: inherit vmid from mqd Alex Deucher
2022-04-29 17:45 ` [PATCH 14/73] drm/amdgpu/gfx10: use INVALIDATE_TLBS to invalidate TLBs v2 Alex Deucher
2022-04-29 17:45 ` [PATCH 15/73] drm/amdgpu/gmc10: skip emitting pasid mapping packet Alex Deucher
2022-04-29 17:45 ` [PATCH 16/73] drm/amdgpu: use the whole doorbell space for mes Alex Deucher
2022-04-29 17:45 ` [PATCH 17/73] drm/amdgpu: update mes process/gang/queue definitions Alex Deucher
2022-04-29 17:45 ` [PATCH 18/73] drm/amdgpu: add mes_kiq module parameter v2 Alex Deucher
2022-04-29 17:45 ` [PATCH 19/73] drm/amdgpu: allocate doorbell index for mes kiq Alex Deucher
2022-04-29 17:45 ` [PATCH 20/73] drm/amdgpu/mes: extend mes framework to support multiple mes pipes Alex Deucher
2022-04-29 17:45 ` [PATCH 21/73] drm/amdgpu/gfx10: add mes queue fence handling Alex Deucher
2022-04-29 17:45 ` [PATCH 22/73] drm/amdgpu/gfx10: add mes support for gfx ib test Alex Deucher
2022-04-29 17:45 ` [PATCH 23/73] drm/amdgpu: don't use kiq to flush gpu tlb if mes enabled Alex Deucher
2022-04-29 17:45 ` [PATCH 24/73] drm/amdgpu/sdma: use per-ctx sdma csa address for mes sdma queue Alex Deucher
2022-04-29 17:45 ` [PATCH 25/73] drm/amdgpu/sdma5.2: initialize sdma mqd Alex Deucher
2022-04-29 17:45 ` [PATCH 26/73] drm/amdgpu/sdma5.2: associate mes queue id with fence Alex Deucher
2022-04-29 17:45 ` [PATCH 27/73] drm/amdgpu/sdma5.2: add mes queue fence handling Alex Deucher
2022-04-29 17:45 ` [PATCH 28/73] drm/amdgpu/sdma5.2: add mes support for sdma ring test Alex Deucher
2022-04-29 17:45 ` [PATCH 29/73] drm/amdgpu/sdma5.2: add mes support for sdma ib test Alex Deucher
2022-04-29 17:45 ` [PATCH 30/73] drm/amdgpu/sdma5: initialize sdma mqd Alex Deucher
2022-04-29 17:45 ` [PATCH 31/73] drm/amdgpu/sdma5: associate mes queue id with fence Alex Deucher
2022-04-29 17:45 ` [PATCH 32/73] drm/amdgpu/sdma5: add mes queue fence handling Alex Deucher
2022-04-29 17:45 ` [PATCH 33/73] drm/amdgpu/sdma5: add mes support for sdma ring test Alex Deucher
2022-04-29 17:45 ` [PATCH 34/73] drm/amdgpu/sdma5: add mes support for sdma ib test Alex Deucher
2022-04-29 17:45 ` [PATCH 35/73] drm/amdgpu: add mes kiq PSP GFX FW type Alex Deucher
2022-04-29 17:45 ` [PATCH 36/73] drm/amdgpu/mes: add mes kiq callback Alex Deucher
2022-04-29 17:45 ` [PATCH 37/73] drm/amdgpu: add mes kiq frontdoor loading support Alex Deucher
2022-04-29 17:45 ` [PATCH 38/73] drm/amdgpu: enable mes kiq N-1 test on sienna cichlid Alex Deucher
2022-04-29 17:45 ` Alex Deucher [this message]
2022-04-29 17:45 ` [PATCH 40/73] drm/amdgpu: add mes queue id mask v2 Alex Deucher
2022-04-29 17:45 ` [PATCH 41/73] drm/amdgpu/mes: initialize/finalize common mes structure v2 Alex Deucher
2022-04-29 17:45 ` [PATCH 42/73] drm/amdgpu/mes: relocate status_fence slot allocation Alex Deucher
2022-04-29 17:45 ` [PATCH 43/73] drm/amdgpu/mes10.1: call general mes initialization Alex Deucher
2022-04-29 17:45 ` [PATCH 44/73] drm/amdgpu/mes10.1: add delay after mes engine enable Alex Deucher
2022-04-29 17:45 ` [PATCH 45/73] drm/amdgpu/mes10.1: implement the suspend/resume routine Alex Deucher
2022-04-29 17:45 ` [PATCH 46/73] drm/amdgpu/mes: implement creating mes process v2 Alex Deucher
2022-04-29 17:45 ` [PATCH 47/73] drm/amdgpu/mes: implement destroying mes process Alex Deucher
2022-04-29 17:45 ` [PATCH 48/73] drm/amdgpu/mes: implement adding mes gang Alex Deucher
2022-04-29 17:46 ` [PATCH 49/73] drm/amdgpu/mes: implement removing " Alex Deucher
2022-04-29 17:46 ` [PATCH 50/73] drm/amdgpu/mes: implement suspending all gangs Alex Deucher
2022-04-29 17:46 ` [PATCH 51/73] drm/amdgpu/mes: implement resuming " Alex Deucher
2022-04-29 17:46 ` [PATCH 52/73] drm/amdgpu/mes: initialize mqd from queue properties Alex Deucher
2022-04-29 17:46 ` [PATCH 53/73] drm/amdgpu/mes: implement adding mes queue Alex Deucher
2022-04-29 17:46 ` [PATCH 54/73] drm/amdgpu/mes: implement removing " Alex Deucher
2022-04-29 17:46 ` [PATCH 55/73] drm/amdgpu/mes: add helper function to convert ring to queue property Alex Deucher
2022-04-29 17:46 ` [PATCH 56/73] drm/amdgpu/mes: add helper function to get the ctx meta data offset Alex Deucher
2022-04-29 17:46 ` [PATCH 57/73] drm/amdgpu/mes: use ring for kernel queue submission Alex Deucher
2022-04-29 17:46 ` [PATCH 58/73] drm/amdgpu/mes: implement removing mes ring Alex Deucher
2022-04-29 17:46 ` [PATCH 59/73] drm/amdgpu/mes: add helper functions to alloc/free ctx metadata Alex Deucher
2022-04-29 17:46 ` [PATCH 60/73] drm/amdgpu: skip kfd routines when mes enabled Alex Deucher
2022-04-29 17:46 ` [PATCH 61/73] drm/amdgpu: Enable KFD with MES enabled Alex Deucher
2022-04-29 17:46 ` [PATCH 62/73] drm/amdgpu: skip some checking for mes queue ib submission Alex Deucher
2022-04-29 17:46 ` [PATCH 63/73] drm/amdgpu: skip kiq ib tests if mes enabled Alex Deucher
2022-04-29 17:46 ` [PATCH 64/73] drm/amdgpu: skip gds switch for mes queue Alex Deucher
2022-04-29 17:46 ` [PATCH 65/73] drm/amdgpu: kiq takes charge of all queues Alex Deucher
2022-04-29 17:46 ` [PATCH 66/73] drm/amdgpu/mes: map ctx metadata for mes self test Alex Deucher
2022-04-29 17:46 ` [PATCH 67/73] drm/amdgpu/mes: create gang and queues " Alex Deucher
2022-04-29 17:46 ` [PATCH 68/73] drm/amdgpu/mes: add ring/ib test " Alex Deucher
2022-04-29 17:46 ` [PATCH 69/73] drm/amdgpu/mes: implement " Alex Deucher
2022-04-29 17:46 ` [PATCH 70/73] drm/amdgpu/mes10.1: add mes self test in late init Alex Deucher
2022-04-29 17:46 ` [PATCH 71/73] drm/amdgpu/mes: fix vm csa update issue Alex Deucher
2022-04-29 17:46 ` [PATCH 72/73] drm/amdgpu/mes: disable mes sdma queue test Alex Deucher
2022-04-29 17:46 ` [PATCH 73/73] drm/amdgpu/mes: Update the doorbell function signatures Alex Deucher

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=20220429174624.459475-40-alexander.deucher@amd.com \
    --to=alexander.deucher@amd.com \
    --cc=Hawking.Zhang@amd.com \
    --cc=Jack.Xiao@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    --cc=christian.koenig@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