Merge tag 'drm-fixes-2026-03-07' of https://gitlab.freedesktop.org/drm/kernel

Pull drm fixes from Dave Airlie:
 "Weekly fixes pull.

  There is one mm fix in here for a HMM livelock triggered by the xe
  driver tests. Otherwise it's a pretty wide range of fixes across the
  board, ttm UAF regression fix, amdgpu fixes, nouveau doesn't crash my
  laptop anymore fix, and a fair bit of misc.

  Seems about right for rc3.

  mm:
   - mm: Fix a hmm_range_fault() livelock / starvation problem

  pagemap:
   - Revert "drm/pagemap: Disable device-to-device migration"

  ttm:
   - fix function return breaking reclaim
   - fix build failure on PREEMPT_RT
   - fix bo->resource UAF

  dma-buf:
   - include ioctl.h in uapi header

  sched:
   - fix kernel doc warning

  amdgpu:
   - LUT fixes
   - VCN5 fix
   - Dispclk fix
   - SMU 13.x fix
   - Fix race in VM acquire
   - PSP 15.x fix
   - UserQ fix

  amdxdna:
   - fix invalid payload for failed command
   - fix NULL ptr dereference
   - fix major fw version check
   - avoid inconsistent fw state on error

  i915/display:
   - Fix for Lenovo T14 G7 display not refreshing

  xe:
   - Do not preempt fence signaling CS instructions
   - Some leak and finalization fixes
   - Workaround fix

  nouveau:
   - avoid runtime suspend oops when using dp aux

  panthor:
   - fix gem_sync argument ordering

  solomon:
   - fix incorrect display output

  renesas:
   - fix DSI divider programming

  ethosu:
   - fix job submit error clean-up refcount
   - fix NPU_OP_ELEMENTWISE validation
   - handle possible underflows in IFM size calcs"

* tag 'drm-fixes-2026-03-07' of https://gitlab.freedesktop.org/drm/kernel: (38 commits)
  accel: ethosu: Handle possible underflow in IFM size calculations
  accel: ethosu: Fix NPU_OP_ELEMENTWISE validation with scalar
  accel: ethosu: Fix job submit error clean-up refcount underflows
  accel/amdxdna: Split mailbox channel create function
  drm/panthor: Correct the order of arguments passed to gem_sync
  Revert "drm/syncobj: Fix handle <-> fd ioctls with dirty stack"
  drm/ttm: Fix bo resource use-after-free
  nouveau/dpcd: return EBUSY for aux xfer if the device is asleep
  accel/amdxdna: Fix major version check on NPU1 platform
  drm/amdgpu/userq: refcount userqueues to avoid any race conditions
  drm/amdgpu/userq: Consolidate wait ioctl exit path
  drm/amdgpu/psp: Use Indirect access address for GFX to PSP mailbox
  drm/amdgpu: Fix use-after-free race in VM acquire
  drm/amd/pm: remove invalid gpu_metrics.energy_accumulator on smu v13.0.x
  drm/xe: Fix memory leak in xe_vm_madvise_ioctl
  drm/xe/reg_sr: Fix leak on xa_store failure
  drm/xe/xe2_hpg: Correct implementation of Wa_16025250150
  drm/xe/gsc: Fix GSC proxy cleanup on early initialization failure
  Revert "drm/pagemap: Disable device-to-device migration"
  drm/i915/psr: Fix for Panel Replay X granularity DPCD register handling
  ...

Author: torvalds

drivers/accel/amdxdna/aie2_ctx.c

@@ -186,13 +186,13 @@ aie2_sched_resp_handler(void *handle, void __iomem *data, size_t size)
 	cmd_abo = job->cmd_bo;
 
 	if (unlikely(job->job_timeout)) {
-		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_TIMEOUT);
+		amdxdna_cmd_set_error(cmd_abo, job, 0, ERT_CMD_STATE_TIMEOUT);
 		ret = -EINVAL;
 		goto out;
 	}
 
 	if (unlikely(!data) || unlikely(size != sizeof(u32))) {
-		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_ABORT);
+		amdxdna_cmd_set_error(cmd_abo, job, 0, ERT_CMD_STATE_ABORT);
 		ret = -EINVAL;
 		goto out;
 	}
@@ -202,7 +202,7 @@ aie2_sched_resp_handler(void *handle, void __iomem *data, size_t size)
 	if (status == AIE2_STATUS_SUCCESS)
 		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_COMPLETED);
 	else
-		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_ERROR);
+		amdxdna_cmd_set_error(cmd_abo, job, 0, ERT_CMD_STATE_ERROR);
 
 out:
 	aie2_sched_notify(job);
@@ -244,13 +244,13 @@ aie2_sched_cmdlist_resp_handler(void *handle, void __iomem *data, size_t size)
 	cmd_abo = job->cmd_bo;
 
 	if (unlikely(job->job_timeout)) {
-		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_TIMEOUT);
+		amdxdna_cmd_set_error(cmd_abo, job, 0, ERT_CMD_STATE_TIMEOUT);
 		ret = -EINVAL;
 		goto out;
 	}
 
 	if (unlikely(!data) || unlikely(size != sizeof(u32) * 3)) {
-		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_ABORT);
+		amdxdna_cmd_set_error(cmd_abo, job, 0, ERT_CMD_STATE_ABORT);
 		ret = -EINVAL;
 		goto out;
 	}
@@ -270,19 +270,12 @@ aie2_sched_cmdlist_resp_handler(void *handle, void __iomem *data, size_t size)
 		 fail_cmd_idx, fail_cmd_status);
 
 	if (fail_cmd_status == AIE2_STATUS_SUCCESS) {
-		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_ABORT);
+		amdxdna_cmd_set_error(cmd_abo, job, fail_cmd_idx, ERT_CMD_STATE_ABORT);
 		ret = -EINVAL;
-		goto out;
+	} else {
+		amdxdna_cmd_set_error(cmd_abo, job, fail_cmd_idx, ERT_CMD_STATE_ERROR);
 	}
-	amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_ERROR);
 
-	if (amdxdna_cmd_get_op(cmd_abo) == ERT_CMD_CHAIN) {
-		struct amdxdna_cmd_chain *cc = amdxdna_cmd_get_payload(cmd_abo, NULL);
-
-		cc->error_index = fail_cmd_idx;
-		if (cc->error_index >= cc->command_count)
-			cc->error_index = 0;
-	}
 out:
 	aie2_sched_notify(job);
 	return ret;

drivers/accel/amdxdna/aie2_message.c

@@ -40,11 +40,8 @@ static int aie2_send_mgmt_msg_wait(struct amdxdna_dev_hdl *ndev,
 		return -ENODEV;
 
 	ret = xdna_send_msg_wait(xdna, ndev->mgmt_chann, msg);
-	if (ret == -ETIME) {
-		xdna_mailbox_stop_channel(ndev->mgmt_chann);
-		xdna_mailbox_destroy_channel(ndev->mgmt_chann);
-		ndev->mgmt_chann = NULL;
-	}
+	if (ret == -ETIME)
+		aie2_destroy_mgmt_chann(ndev);
 
 	if (!ret && *hdl->status != AIE2_STATUS_SUCCESS) {
 		XDNA_ERR(xdna, "command opcode 0x%x failed, status 0x%x",
@@ -296,20 +293,29 @@ int aie2_create_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwct
 	}
 
 	intr_reg = i2x.mb_head_ptr_reg + 4;
-	hwctx->priv->mbox_chann = xdna_mailbox_create_channel(ndev->mbox, &x2i, &i2x,
-							      intr_reg, ret);
+	hwctx->priv->mbox_chann = xdna_mailbox_alloc_channel(ndev->mbox);
 	if (!hwctx->priv->mbox_chann) {
 		XDNA_ERR(xdna, "Not able to create channel");
 		ret = -EINVAL;
 		goto del_ctx_req;
 	}
+
+	ret = xdna_mailbox_start_channel(hwctx->priv->mbox_chann, &x2i, &i2x,
+					 intr_reg, ret);
+	if (ret) {
+		XDNA_ERR(xdna, "Not able to create channel");
+		ret = -EINVAL;
+		goto free_channel;
+	}
 	ndev->hwctx_num++;
 
 	XDNA_DBG(xdna, "Mailbox channel irq: %d, msix_id: %d", ret, resp.msix_id);
 	XDNA_DBG(xdna, "Created fw ctx %d pasid %d", hwctx->fw_ctx_id, hwctx->client->pasid);
 
 	return 0;
 
+free_channel:
+	xdna_mailbox_free_channel(hwctx->priv->mbox_chann);
 del_ctx_req:
 	aie2_destroy_context_req(ndev, hwctx->fw_ctx_id);
 	return ret;
@@ -325,7 +331,7 @@ int aie2_destroy_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwc
 
 	xdna_mailbox_stop_channel(hwctx->priv->mbox_chann);
 	ret = aie2_destroy_context_req(ndev, hwctx->fw_ctx_id);
-	xdna_mailbox_destroy_channel(hwctx->priv->mbox_chann);
+	xdna_mailbox_free_channel(hwctx->priv->mbox_chann);
 	XDNA_DBG(xdna, "Destroyed fw ctx %d", hwctx->fw_ctx_id);
 	hwctx->priv->mbox_chann = NULL;
 	hwctx->fw_ctx_id = -1;
@@ -914,6 +920,20 @@ void aie2_msg_init(struct amdxdna_dev_hdl *ndev)
 		ndev->exec_msg_ops = &legacy_exec_message_ops;
 }
 
+void aie2_destroy_mgmt_chann(struct amdxdna_dev_hdl *ndev)
+{
+	struct amdxdna_dev *xdna = ndev->xdna;
+
+	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
+
+	if (!ndev->mgmt_chann)
+		return;
+
+	xdna_mailbox_stop_channel(ndev->mgmt_chann);
+	xdna_mailbox_free_channel(ndev->mgmt_chann);
+	ndev->mgmt_chann = NULL;
+}
+
 static inline struct amdxdna_gem_obj *
 aie2_cmdlist_get_cmd_buf(struct amdxdna_sched_job *job)
 {

drivers/accel/amdxdna/aie2_pci.c

@@ -330,9 +330,7 @@ static void aie2_hw_stop(struct amdxdna_dev *xdna)
 
 	aie2_runtime_cfg(ndev, AIE2_RT_CFG_CLK_GATING, NULL);
 	aie2_mgmt_fw_fini(ndev);
-	xdna_mailbox_stop_channel(ndev->mgmt_chann);
-	xdna_mailbox_destroy_channel(ndev->mgmt_chann);
-	ndev->mgmt_chann = NULL;
+	aie2_destroy_mgmt_chann(ndev);
 	drmm_kfree(&xdna->ddev, ndev->mbox);
 	ndev->mbox = NULL;
 	aie2_psp_stop(ndev->psp_hdl);
@@ -363,10 +361,29 @@ static int aie2_hw_start(struct amdxdna_dev *xdna)
 	}
 	pci_set_master(pdev);
 
+	mbox_res.ringbuf_base = ndev->sram_base;
+	mbox_res.ringbuf_size = pci_resource_len(pdev, xdna->dev_info->sram_bar);
+	mbox_res.mbox_base = ndev->mbox_base;
+	mbox_res.mbox_size = MBOX_SIZE(ndev);
+	mbox_res.name = "xdna_mailbox";
+	ndev->mbox = xdnam_mailbox_create(&xdna->ddev, &mbox_res);
+	if (!ndev->mbox) {
+		XDNA_ERR(xdna, "failed to create mailbox device");
+		ret = -ENODEV;
+		goto disable_dev;
+	}
+
+	ndev->mgmt_chann = xdna_mailbox_alloc_channel(ndev->mbox);
+	if (!ndev->mgmt_chann) {
+		XDNA_ERR(xdna, "failed to alloc channel");
+		ret = -ENODEV;
+		goto disable_dev;
+	}
+
 	ret = aie2_smu_init(ndev);
 	if (ret) {
 		XDNA_ERR(xdna, "failed to init smu, ret %d", ret);
-		goto disable_dev;
+		goto free_channel;
 	}
 
 	ret = aie2_psp_start(ndev->psp_hdl);
@@ -381,18 +398,6 @@ static int aie2_hw_start(struct amdxdna_dev *xdna)
 		goto stop_psp;
 	}
 
-	mbox_res.ringbuf_base = ndev->sram_base;
-	mbox_res.ringbuf_size = pci_resource_len(pdev, xdna->dev_info->sram_bar);
-	mbox_res.mbox_base = ndev->mbox_base;
-	mbox_res.mbox_size = MBOX_SIZE(ndev);
-	mbox_res.name = "xdna_mailbox";
-	ndev->mbox = xdnam_mailbox_create(&xdna->ddev, &mbox_res);
-	if (!ndev->mbox) {
-		XDNA_ERR(xdna, "failed to create mailbox device");
-		ret = -ENODEV;
-		goto stop_psp;
-	}
-
 	mgmt_mb_irq = pci_irq_vector(pdev, ndev->mgmt_chan_idx);
 	if (mgmt_mb_irq < 0) {
 		ret = mgmt_mb_irq;
@@ -401,52 +406,55 @@ static int aie2_hw_start(struct amdxdna_dev *xdna)
 	}
 
 	xdna_mailbox_intr_reg = ndev->mgmt_i2x.mb_head_ptr_reg + 4;
-	ndev->mgmt_chann = xdna_mailbox_create_channel(ndev->mbox,
-						       &ndev->mgmt_x2i,
-						       &ndev->mgmt_i2x,
-						       xdna_mailbox_intr_reg,
-						       mgmt_mb_irq);
-	if (!ndev->mgmt_chann) {
-		XDNA_ERR(xdna, "failed to create management mailbox channel");
+	ret = xdna_mailbox_start_channel(ndev->mgmt_chann,
+					 &ndev->mgmt_x2i,
+					 &ndev->mgmt_i2x,
+					 xdna_mailbox_intr_reg,
+					 mgmt_mb_irq);
+	if (ret) {
+		XDNA_ERR(xdna, "failed to start management mailbox channel");
 		ret = -EINVAL;
 		goto stop_psp;
 	}
 
 	ret = aie2_mgmt_fw_init(ndev);
 	if (ret) {
 		XDNA_ERR(xdna, "initial mgmt firmware failed, ret %d", ret);
-		goto destroy_mgmt_chann;
+		goto stop_fw;
 	}
 
 	ret = aie2_pm_init(ndev);
 	if (ret) {
 		XDNA_ERR(xdna, "failed to init pm, ret %d", ret);
-		goto destroy_mgmt_chann;
+		goto stop_fw;
 	}
 
 	ret = aie2_mgmt_fw_query(ndev);
 	if (ret) {
 		XDNA_ERR(xdna, "failed to query fw, ret %d", ret);
-		goto destroy_mgmt_chann;
+		goto stop_fw;
 	}
 
 	ret = aie2_error_async_events_alloc(ndev);
 	if (ret) {
 		XDNA_ERR(xdna, "Allocate async events failed, ret %d", ret);
-		goto destroy_mgmt_chann;
+		goto stop_fw;
 	}
 
 	ndev->dev_status = AIE2_DEV_START;
 
 	return 0;
 
-destroy_mgmt_chann:
+stop_fw:
+	aie2_suspend_fw(ndev);
 	xdna_mailbox_stop_channel(ndev->mgmt_chann);
-	xdna_mailbox_destroy_channel(ndev->mgmt_chann);
 stop_psp:
 	aie2_psp_stop(ndev->psp_hdl);
 fini_smu:
 	aie2_smu_fini(ndev);
+free_channel:
+	xdna_mailbox_free_channel(ndev->mgmt_chann);
+	ndev->mgmt_chann = NULL;
 disable_dev:
 	pci_disable_device(pdev);
 

drivers/accel/amdxdna/aie2_pci.h

@@ -303,6 +303,7 @@ int aie2_get_array_async_error(struct amdxdna_dev_hdl *ndev,
 
 /* aie2_message.c */
 void aie2_msg_init(struct amdxdna_dev_hdl *ndev);
+void aie2_destroy_mgmt_chann(struct amdxdna_dev_hdl *ndev);
 int aie2_suspend_fw(struct amdxdna_dev_hdl *ndev);
 int aie2_resume_fw(struct amdxdna_dev_hdl *ndev);
 int aie2_set_runtime_cfg(struct amdxdna_dev_hdl *ndev, u32 type, u64 value);

drivers/accel/amdxdna/amdxdna_ctx.c

@@ -135,6 +135,33 @@ u32 amdxdna_cmd_get_cu_idx(struct amdxdna_gem_obj *abo)
 	return INVALID_CU_IDX;
 }
 
+int amdxdna_cmd_set_error(struct amdxdna_gem_obj *abo,
+			  struct amdxdna_sched_job *job, u32 cmd_idx,
+			  enum ert_cmd_state error_state)
+{
+	struct amdxdna_client *client = job->hwctx->client;
+	struct amdxdna_cmd *cmd = abo->mem.kva;
+	struct amdxdna_cmd_chain *cc = NULL;
+
+	cmd->header &= ~AMDXDNA_CMD_STATE;
+	cmd->header |= FIELD_PREP(AMDXDNA_CMD_STATE, error_state);
+
+	if (amdxdna_cmd_get_op(abo) == ERT_CMD_CHAIN) {
+		cc = amdxdna_cmd_get_payload(abo, NULL);
+		cc->error_index = (cmd_idx < cc->command_count) ? cmd_idx : 0;
+		abo = amdxdna_gem_get_obj(client, cc->data[0], AMDXDNA_BO_CMD);
+		if (!abo)
+			return -EINVAL;
+		cmd = abo->mem.kva;
+	}
+
+	memset(cmd->data, 0xff, abo->mem.size - sizeof(*cmd));
+	if (cc)
+		amdxdna_gem_put_obj(abo);
+
+	return 0;
+}
+
 /*
  * This should be called in close() and remove(). DO NOT call in other syscalls.
  * This guarantee that when hwctx and resources will be released, if user

drivers/accel/amdxdna/amdxdna_ctx.h

@@ -167,6 +167,9 @@ amdxdna_cmd_get_state(struct amdxdna_gem_obj *abo)
 
 void *amdxdna_cmd_get_payload(struct amdxdna_gem_obj *abo, u32 *size);
 u32 amdxdna_cmd_get_cu_idx(struct amdxdna_gem_obj *abo);
+int amdxdna_cmd_set_error(struct amdxdna_gem_obj *abo,
+			  struct amdxdna_sched_job *job, u32 cmd_idx,
+			  enum ert_cmd_state error_state);
 
 void amdxdna_sched_job_cleanup(struct amdxdna_sched_job *job);
 void amdxdna_hwctx_remove_all(struct amdxdna_client *client);