add amdgpu from linux 4.19.44 for recent AMD Radeon parts

committing now so this can be worked on in tree

Thanks to the OpenBSD Foundation for sponsoring this work and kettenis@
for helping.

1 files changed, 344 insertions, 0 deletions

diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_gfx.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gfx.c
new file mode 100644
index 00000000000..35e7da3a443
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gfx.c
@@ -0,0 +1,344 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * 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 <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+
+/*
+ * GPU scratch registers helpers function.
+ */
+/**
+ * amdgpu_gfx_scratch_get - Allocate a scratch register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: scratch register mmio offset
+ *
+ * Allocate a CP scratch register for use by the driver (all asics).
+ * Returns 0 on success or -EINVAL on failure.
+ */
+int amdgpu_gfx_scratch_get(struct amdgpu_device *adev, uint32_t *reg)
+{
+	int i;
+
+	i = ffs(adev->gfx.scratch.free_mask);
+	if (i != 0 && i <= adev->gfx.scratch.num_reg) {
+		i--;
+		adev->gfx.scratch.free_mask &= ~(1u << i);
+		*reg = adev->gfx.scratch.reg_base + i;
+		return 0;
+	}
+	return -EINVAL;
+}
+
+/**
+ * amdgpu_gfx_scratch_free - Free a scratch register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: scratch register mmio offset
+ *
+ * Free a CP scratch register allocated for use by the driver (all asics)
+ */
+void amdgpu_gfx_scratch_free(struct amdgpu_device *adev, uint32_t reg)
+{
+	adev->gfx.scratch.free_mask |= 1u << (reg - adev->gfx.scratch.reg_base);
+}
+
+/**
+ * amdgpu_gfx_parse_disable_cu - Parse the disable_cu module parameter
+ *
+ * @mask: array in which the per-shader array disable masks will be stored
+ * @max_se: number of SEs
+ * @max_sh: number of SHs
+ *
+ * The bitmask of CUs to be disabled in the shader array determined by se and
+ * sh is stored in mask[se * max_sh + sh].
+ */
+void amdgpu_gfx_parse_disable_cu(unsigned *mask, unsigned max_se, unsigned max_sh)
+{
+	unsigned se, sh, cu;
+	const char *p;
+
+	memset(mask, 0, sizeof(*mask) * max_se * max_sh);
+
+	if (!amdgpu_disable_cu || !*amdgpu_disable_cu)
+		return;
+
+#ifdef notyet
+	p = amdgpu_disable_cu;
+	for (;;) {
+		char *next;
+		int ret = sscanf(p, "%u.%u.%u", &se, &sh, &cu);
+		if (ret < 3) {
+			DRM_ERROR("amdgpu: could not parse disable_cu\n");
+			return;
+		}
+
+		if (se < max_se && sh < max_sh && cu < 16) {
+			DRM_INFO("amdgpu: disabling CU %u.%u.%u\n", se, sh, cu);
+			mask[se * max_sh + sh] |= 1u << cu;
+		} else {
+			DRM_ERROR("amdgpu: disable_cu %u.%u.%u is out of range\n",
+				  se, sh, cu);
+		}
+
+		next = strchr(p, ',');
+		if (!next)
+			break;
+		p = next + 1;
+	}
+#endif
+}
+
+static bool amdgpu_gfx_is_multipipe_capable(struct amdgpu_device *adev)
+{
+	if (amdgpu_compute_multipipe != -1) {
+		DRM_INFO("amdgpu: forcing compute pipe policy %d\n",
+			 amdgpu_compute_multipipe);
+		return amdgpu_compute_multipipe == 1;
+	}
+
+	/* FIXME: spreading the queues across pipes causes perf regressions
+	 * on POLARIS11 compute workloads */
+	if (adev->asic_type == CHIP_POLARIS11)
+		return false;
+
+	return adev->gfx.mec.num_mec > 1;
+}
+
+void amdgpu_gfx_compute_queue_acquire(struct amdgpu_device *adev)
+{
+	int i, queue, pipe, mec;
+	bool multipipe_policy = amdgpu_gfx_is_multipipe_capable(adev);
+
+	/* policy for amdgpu compute queue ownership */
+	for (i = 0; i < AMDGPU_MAX_COMPUTE_QUEUES; ++i) {
+		queue = i % adev->gfx.mec.num_queue_per_pipe;
+		pipe = (i / adev->gfx.mec.num_queue_per_pipe)
+			% adev->gfx.mec.num_pipe_per_mec;
+		mec = (i / adev->gfx.mec.num_queue_per_pipe)
+			/ adev->gfx.mec.num_pipe_per_mec;
+
+		/* we've run out of HW */
+		if (mec >= adev->gfx.mec.num_mec)
+			break;
+
+		if (multipipe_policy) {
+			/* policy: amdgpu owns the first two queues of the first MEC */
+			if (mec == 0 && queue < 2)
+				set_bit(i, adev->gfx.mec.queue_bitmap);
+		} else {
+			/* policy: amdgpu owns all queues in the first pipe */
+			if (mec == 0 && pipe == 0)
+				set_bit(i, adev->gfx.mec.queue_bitmap);
+		}
+	}
+
+	/* update the number of active compute rings */
+	adev->gfx.num_compute_rings =
+		bitmap_weight(adev->gfx.mec.queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+
+	/* If you hit this case and edited the policy, you probably just
+	 * need to increase AMDGPU_MAX_COMPUTE_RINGS */
+	if (WARN_ON(adev->gfx.num_compute_rings > AMDGPU_MAX_COMPUTE_RINGS))
+		adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS;
+}
+
+static int amdgpu_gfx_kiq_acquire(struct amdgpu_device *adev,
+				  struct amdgpu_ring *ring)
+{
+	int queue_bit;
+	int mec, pipe, queue;
+
+	queue_bit = adev->gfx.mec.num_mec
+		    * adev->gfx.mec.num_pipe_per_mec
+		    * adev->gfx.mec.num_queue_per_pipe;
+
+	while (queue_bit-- >= 0) {
+		if (test_bit(queue_bit, adev->gfx.mec.queue_bitmap))
+			continue;
+
+		amdgpu_gfx_bit_to_queue(adev, queue_bit, &mec, &pipe, &queue);
+
+		/*
+		 * 1. Using pipes 2/3 from MEC 2 seems cause problems.
+		 * 2. It must use queue id 0, because CGPG_IDLE/SAVE/LOAD/RUN
+		 * only can be issued on queue 0.
+		 */
+		if ((mec == 1 && pipe > 1) || queue != 0)
+			continue;
+
+		ring->me = mec + 1;
+		ring->pipe = pipe;
+		ring->queue = queue;
+
+		return 0;
+	}
+
+	dev_err(adev->dev, "Failed to find a queue for KIQ\n");
+	return -EINVAL;
+}
+
+int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev,
+			     struct amdgpu_ring *ring,
+			     struct amdgpu_irq_src *irq)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+	int r = 0;
+
+	mtx_init(&kiq->ring_lock, IPL_TTY);
+
+	r = amdgpu_device_wb_get(adev, &adev->virt.reg_val_offs);
+	if (r)
+		return r;
+
+	ring->adev = NULL;
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = AMDGPU_DOORBELL_KIQ;
+
+	r = amdgpu_gfx_kiq_acquire(adev, ring);
+	if (r)
+		return r;
+
+	ring->eop_gpu_addr = kiq->eop_gpu_addr;
+	snprintf(ring->name, sizeof(ring->name), "kiq_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+	r = amdgpu_ring_init(adev, ring, 1024,
+			     irq, AMDGPU_CP_KIQ_IRQ_DRIVER0);
+	if (r)
+		dev_warn(adev->dev, "(%d) failed to init kiq ring\n", r);
+
+	return r;
+}
+
+void amdgpu_gfx_kiq_free_ring(struct amdgpu_ring *ring,
+			      struct amdgpu_irq_src *irq)
+{
+	amdgpu_device_wb_free(ring->adev, ring->adev->virt.reg_val_offs);
+	amdgpu_ring_fini(ring);
+}
+
+void amdgpu_gfx_kiq_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+
+	amdgpu_bo_free_kernel(&kiq->eop_obj, &kiq->eop_gpu_addr, NULL);
+}
+
+int amdgpu_gfx_kiq_init(struct amdgpu_device *adev,
+			unsigned hpd_size)
+{
+	int r;
+	u32 *hpd;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+
+	r = amdgpu_bo_create_kernel(adev, hpd_size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_GTT, &kiq->eop_obj,
+				    &kiq->eop_gpu_addr, (void **)&hpd);
+	if (r) {
+		dev_warn(adev->dev, "failed to create KIQ bo (%d).\n", r);
+		return r;
+	}
+
+	memset(hpd, 0, hpd_size);
+
+	r = amdgpu_bo_reserve(kiq->eop_obj, true);
+	if (unlikely(r != 0))
+		dev_warn(adev->dev, "(%d) reserve kiq eop bo failed\n", r);
+	amdgpu_bo_kunmap(kiq->eop_obj);
+	amdgpu_bo_unreserve(kiq->eop_obj);
+
+	return 0;
+}
+
+/* create MQD for each compute queue */
+int amdgpu_gfx_compute_mqd_sw_init(struct amdgpu_device *adev,
+				   unsigned mqd_size)
+{
+	struct amdgpu_ring *ring = NULL;
+	int r, i;
+
+	/* create MQD for KIQ */
+	ring = &adev->gfx.kiq.ring;
+	if (!ring->mqd_obj) {
+		/* originaly the KIQ MQD is put in GTT domain, but for SRIOV VRAM domain is a must
+		 * otherwise hypervisor trigger SAVE_VF fail after driver unloaded which mean MQD
+		 * deallocated and gart_unbind, to strict diverage we decide to use VRAM domain for
+		 * KIQ MQD no matter SRIOV or Bare-metal
+		 */
+		r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_VRAM, &ring->mqd_obj,
+					    &ring->mqd_gpu_addr, &ring->mqd_ptr);
+		if (r) {
+			dev_warn(adev->dev, "failed to create ring mqd ob (%d)", r);
+			return r;
+		}
+
+		/* prepare MQD backup */
+		adev->gfx.mec.mqd_backup[AMDGPU_MAX_COMPUTE_RINGS] = kmalloc(mqd_size, GFP_KERNEL);
+		if (!adev->gfx.mec.mqd_backup[AMDGPU_MAX_COMPUTE_RINGS])
+				dev_warn(adev->dev, "no memory to create MQD backup for ring %s\n", ring->name);
+	}
+
+	/* create MQD for each KCQ */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+		if (!ring->mqd_obj) {
+			r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
+						    AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
+						    &ring->mqd_gpu_addr, &ring->mqd_ptr);
+			if (r) {
+				dev_warn(adev->dev, "failed to create ring mqd ob (%d)", r);
+				return r;
+			}
+
+			/* prepare MQD backup */
+			adev->gfx.mec.mqd_backup[i] = kmalloc(mqd_size, GFP_KERNEL);
+			if (!adev->gfx.mec.mqd_backup[i])
+				dev_warn(adev->dev, "no memory to create MQD backup for ring %s\n", ring->name);
+		}
+	}
+
+	return 0;
+}
+
+void amdgpu_gfx_compute_mqd_sw_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = NULL;
+	int i;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+		kfree(adev->gfx.mec.mqd_backup[i]);
+		amdgpu_bo_free_kernel(&ring->mqd_obj,
+				      &ring->mqd_gpu_addr,
+				      &ring->mqd_ptr);
+	}
+
+	ring = &adev->gfx.kiq.ring;
+	kfree(adev->gfx.mec.mqd_backup[AMDGPU_MAX_COMPUTE_RINGS]);
+	amdgpu_bo_free_kernel(&ring->mqd_obj,
+			      &ring->mqd_gpu_addr,
+			      &ring->mqd_ptr);
+}