Merge tag 'amd-drm-fixes-6.15-2025-04-09' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.15-2025-04-09:

amdgpu:
- MES FW version caching fixes
- Only use GTT as a fallback if we already have a backing store
- dma_buf fix
- IP discovery fix
- Replay and PSR with VRR fix
- DC FP fixes
- eDP fixes
- KIQ TLB invalidate fix
- Enable dmem groups support
- Allow pinning VRAM dma bufs if imports can do P2P
- Workload profile fixes
- Prevent possible division by 0 in fan handling

amdkfd:
- Queue reset fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Alex Deucher <alexander.deucher@amd.com>
Link: https://lore.kernel.org/r/20250409165238.1180153-1-alexander.deucher@amd.com

Author: airlied

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -353,7 +353,6 @@ enum amdgpu_kiq_irq {
 	AMDGPU_CP_KIQ_IRQ_DRIVER0 = 0,
 	AMDGPU_CP_KIQ_IRQ_LAST
 };
-#define SRIOV_USEC_TIMEOUT  1200000 /* wait 12 * 100ms for SRIOV */
 #define MAX_KIQ_REG_WAIT       5000 /* in usecs, 5ms */
 #define MAX_KIQ_REG_BAILOUT_INTERVAL   5 /* in msecs, 5ms */
 #define MAX_KIQ_REG_TRY 1000

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

@@ -3643,6 +3643,13 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
 			    adev, adev->ip_blocks[i].version->type))
 			continue;
 
+		/* Since we skip suspend for S0i3, we need to cancel the delayed
+		 * idle work here as the suspend callback never gets called.
+		 */
+		if (adev->in_s0ix &&
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX &&
+		    amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 0, 0))
+			cancel_delayed_work_sync(&adev->gfx.idle_work);
 		/* skip suspend of gfx/mes and psp for S0ix
 		 * gfx is in gfxoff state, so on resume it will exit gfxoff just
 		 * like at runtime. PSP is also part of the always on hardware

drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c

@@ -120,6 +120,8 @@ MODULE_FIRMWARE("amdgpu/vega20_ip_discovery.bin");
 MODULE_FIRMWARE("amdgpu/raven_ip_discovery.bin");
 MODULE_FIRMWARE("amdgpu/raven2_ip_discovery.bin");
 MODULE_FIRMWARE("amdgpu/picasso_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/arcturus_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/aldebaran_ip_discovery.bin");
 
 #define mmIP_DISCOVERY_VERSION  0x16A00
 #define mmRCC_CONFIG_MEMSIZE	0xde3

drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c

@@ -75,11 +75,25 @@ static int amdgpu_dma_buf_attach(struct dma_buf *dmabuf,
  */
 static int amdgpu_dma_buf_pin(struct dma_buf_attachment *attach)
 {
-	struct drm_gem_object *obj = attach->dmabuf->priv;
-	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	struct dma_buf *dmabuf = attach->dmabuf;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(dmabuf->priv);
+	u32 domains = bo->preferred_domains;
 
-	/* pin buffer into GTT */
-	return amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT);
+	dma_resv_assert_held(dmabuf->resv);
+
+	/*
+	 * Try pinning into VRAM to allow P2P with RDMA NICs without ODP
+	 * support if all attachments can do P2P. If any attachment can't do
+	 * P2P just pin into GTT instead.
+	 */
+	list_for_each_entry(attach, &dmabuf->attachments, node)
+		if (!attach->peer2peer)
+			domains &= ~AMDGPU_GEM_DOMAIN_VRAM;
+
+	if (domains & AMDGPU_GEM_DOMAIN_VRAM)
+		bo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+
+	return amdgpu_bo_pin(bo, domains);
 }
 
 /**
@@ -134,9 +148,6 @@ static struct sg_table *amdgpu_dma_buf_map(struct dma_buf_attachment *attach,
 		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 		if (r)
 			return ERR_PTR(r);
-
-	} else if (bo->tbo.resource->mem_type != TTM_PL_TT) {
-		return ERR_PTR(-EBUSY);
 	}
 
 	switch (bo->tbo.resource->mem_type) {
@@ -184,7 +195,7 @@ static void amdgpu_dma_buf_unmap(struct dma_buf_attachment *attach,
 				 struct sg_table *sgt,
 				 enum dma_data_direction dir)
 {
-	if (sgt->sgl->page_link) {
+	if (sg_page(sgt->sgl)) {
 		dma_unmap_sgtable(attach->dev, sgt, dir, 0);
 		sg_free_table(sgt);
 		kfree(sgt);

drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c

@@ -699,12 +699,10 @@ int amdgpu_gmc_flush_gpu_tlb_pasid(struct amdgpu_device *adev, uint16_t pasid,
 				   uint32_t flush_type, bool all_hub,
 				   uint32_t inst)
 {
-	u32 usec_timeout = amdgpu_sriov_vf(adev) ? SRIOV_USEC_TIMEOUT :
-		adev->usec_timeout;
 	struct amdgpu_ring *ring = &adev->gfx.kiq[inst].ring;
 	struct amdgpu_kiq *kiq = &adev->gfx.kiq[inst];
 	unsigned int ndw;
-	int r;
+	int r, cnt = 0;
 	uint32_t seq;
 
 	/*
@@ -761,10 +759,21 @@ int amdgpu_gmc_flush_gpu_tlb_pasid(struct amdgpu_device *adev, uint16_t pasid,
 
 		amdgpu_ring_commit(ring);
 		spin_unlock(&adev->gfx.kiq[inst].ring_lock);
-		if (amdgpu_fence_wait_polling(ring, seq, usec_timeout) < 1) {
+
+		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+		might_sleep();
+		while (r < 1 && cnt++ < MAX_KIQ_REG_TRY &&
+		       !amdgpu_reset_pending(adev->reset_domain)) {
+			msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+			r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+		}
+
+		if (cnt > MAX_KIQ_REG_TRY) {
 			dev_err(adev->dev, "timeout waiting for kiq fence\n");
 			r = -ETIME;
-		}
+		} else
+			r = 0;
 	}
 
 error_unlock_reset:

drivers/gpu/drm/amd/amdgpu/amdgpu_object.c

@@ -163,8 +163,8 @@ void amdgpu_bo_placement_from_domain(struct amdgpu_bo *abo, u32 domain)
 		 * When GTT is just an alternative to VRAM make sure that we
 		 * only use it as fallback and still try to fill up VRAM first.
 		 */
-		if (domain & abo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM &&
-		    !(adev->flags & AMD_IS_APU))
+		if (abo->tbo.resource && !(adev->flags & AMD_IS_APU) &&
+		    domain & abo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM)
 			places[c].flags |= TTM_PL_FLAG_FALLBACK;
 		c++;
 	}

drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c

@@ -24,6 +24,7 @@
 
 #include <linux/dma-mapping.h>
 #include <drm/ttm/ttm_range_manager.h>
+#include <drm/drm_drv.h>
 
 #include "amdgpu.h"
 #include "amdgpu_vm.h"
@@ -907,6 +908,9 @@ int amdgpu_vram_mgr_init(struct amdgpu_device *adev)
 	struct ttm_resource_manager *man = &mgr->manager;
 	int err;
 
+	man->cg = drmm_cgroup_register_region(adev_to_drm(adev), "vram", adev->gmc.real_vram_size);
+	if (IS_ERR(man->cg))
+		return PTR_ERR(man->cg);
 	ttm_resource_manager_init(man, &adev->mman.bdev,
 				  adev->gmc.real_vram_size);
 

drivers/gpu/drm/amd/amdgpu/mes_v11_0.c

@@ -894,6 +894,10 @@ static void mes_v11_0_get_fw_version(struct amdgpu_device *adev)
 {
 	int pipe;
 
+	/* return early if we have already fetched these */
+	if (adev->mes.sched_version && adev->mes.kiq_version)
+		return;
+
 	/* get MES scheduler/KIQ versions */
 	mutex_lock(&adev->srbm_mutex);
 

drivers/gpu/drm/amd/amdgpu/mes_v12_0.c

@@ -1392,17 +1392,20 @@ static int mes_v12_0_queue_init(struct amdgpu_device *adev,
 		mes_v12_0_queue_init_register(ring);
 	}
 
-	/* get MES scheduler/KIQ versions */
-	mutex_lock(&adev->srbm_mutex);
-	soc21_grbm_select(adev, 3, pipe, 0, 0);
+	if (((pipe == AMDGPU_MES_SCHED_PIPE) && !adev->mes.sched_version) ||
+	    ((pipe == AMDGPU_MES_KIQ_PIPE) && !adev->mes.kiq_version)) {
+		/* get MES scheduler/KIQ versions */
+		mutex_lock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, 3, pipe, 0, 0);
 
-	if (pipe == AMDGPU_MES_SCHED_PIPE)
-		adev->mes.sched_version = RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
-	else if (pipe == AMDGPU_MES_KIQ_PIPE && adev->enable_mes_kiq)
-		adev->mes.kiq_version = RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
+		if (pipe == AMDGPU_MES_SCHED_PIPE)
+			adev->mes.sched_version = RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
+		else if (pipe == AMDGPU_MES_KIQ_PIPE && adev->enable_mes_kiq)
+			adev->mes.kiq_version = RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
 
-	soc21_grbm_select(adev, 0, 0, 0, 0);
-	mutex_unlock(&adev->srbm_mutex);
+		soc21_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	}
 
 	return 0;
 }

drivers/gpu/drm/amd/amdkfd/kfd_topology.c

@@ -1983,9 +1983,6 @@ static void kfd_topology_set_capabilities(struct kfd_topology_device *dev)
 	if (kfd_dbg_has_ttmps_always_setup(dev->gpu))
 		dev->node_props.debug_prop |= HSA_DBG_DISPATCH_INFO_ALWAYS_VALID;
 
-	if (dev->gpu->adev->sdma.supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE)
-		dev->node_props.capability2 |= HSA_CAP2_PER_SDMA_QUEUE_RESET_SUPPORTED;
-
 	if (KFD_GC_VERSION(dev->gpu) < IP_VERSION(10, 0, 0)) {
 		if (KFD_GC_VERSION(dev->gpu) == IP_VERSION(9, 4, 3) ||
 		    KFD_GC_VERSION(dev->gpu) == IP_VERSION(9, 4, 4))
@@ -2001,7 +1998,11 @@ static void kfd_topology_set_capabilities(struct kfd_topology_device *dev)
 			dev->node_props.capability |=
 				HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED;
 
-		dev->node_props.capability |= HSA_CAP_PER_QUEUE_RESET_SUPPORTED;
+		if (!amdgpu_sriov_vf(dev->gpu->adev))
+			dev->node_props.capability |= HSA_CAP_PER_QUEUE_RESET_SUPPORTED;
+
+		if (dev->gpu->adev->sdma.supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE)
+			dev->node_props.capability2 |= HSA_CAP2_PER_SDMA_QUEUE_RESET_SUPPORTED;
 	} else {
 		dev->node_props.debug_prop |= HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX10 |
 					HSA_DBG_WATCH_ADDR_MASK_HI_BIT;