drm/msm: add support to take dpu snapshot

Add the msm_disp_snapshot module which adds supports to dump dpu
registers and capture the drm atomic state which can be used in
case of error conditions.

changes in v5:
 - start storing disp_state in msm_kms instead of dpu_kms
 - get rid of MSM_DISP_SNAPSHOT_IN_* enum by simplifying the functions
 - move snprintf inside the snapshot core by using varargs
 - get rid of some stale code comments
 - allow snapshot module for non-DPU targets

Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Abhinav Kumar <abhinavk@codeaurora.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/1618606645-19695-3-git-send-email-abhinavk@codeaurora.org
Signed-off-by: Rob Clark <robdclark@chromium.org>

drivers/gpu/drm/msm/Makefile

@@ -77,6 +77,8 @@ msm-y := \
 	disp/dpu1/dpu_plane.o \
 	disp/dpu1/dpu_rm.o \
 	disp/dpu1/dpu_vbif.o \
+	disp/msm_disp_snapshot.o \
+	disp/msm_disp_snapshot_util.o \
 	msm_atomic.o \
 	msm_atomic_tracepoints.o \
 	msm_debugfs.o \

drivers/gpu/drm/msm/disp/dpu1/dpu_hw_catalog.h

@@ -1,5 +1,5 @@
 /* SPDX-License-Identifier: GPL-2.0-only */
-/* Copyright (c) 2015-2018, The Linux Foundation. All rights reserved.
+/* Copyright (c) 2015-2018, 2020 The Linux Foundation. All rights reserved.
  */
 
 #ifndef _DPU_HW_CATALOG_H

drivers/gpu/drm/msm/disp/msm_disp_snapshot.c

@@ -0,0 +1,161 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2020-2021, The Linux Foundation. All rights reserved.
+ */
+
+#define pr_fmt(fmt)	"[drm:%s:%d] " fmt, __func__, __LINE__
+
+#include "msm_disp_snapshot.h"
+
+#ifdef CONFIG_DEV_COREDUMP
+static ssize_t disp_devcoredump_read(char *buffer, loff_t offset,
+		size_t count, void *data, size_t datalen)
+{
+	struct drm_print_iterator iter;
+	struct drm_printer p;
+	struct msm_disp_state *disp_state;
+
+	disp_state = data;
+
+	iter.data = buffer;
+	iter.offset = 0;
+	iter.start = offset;
+	iter.remain = count;
+
+	p = drm_coredump_printer(&iter);
+
+	msm_disp_state_print(disp_state, &p);
+
+	return count - iter.remain;
+}
+
+static void disp_devcoredump_free(void *data)
+{
+	struct msm_disp_state *disp_state;
+
+	disp_state = data;
+
+	msm_disp_state_free(disp_state);
+
+	disp_state->coredump_pending = false;
+}
+#endif /* CONFIG_DEV_COREDUMP */
+
+static void _msm_disp_snapshot_work(struct kthread_work *work)
+{
+	struct msm_disp_state *disp_state = container_of(work, struct msm_disp_state, dump_work);
+	struct drm_printer p;
+
+	mutex_lock(&disp_state->mutex);
+
+	msm_disp_snapshot_capture_state(disp_state);
+
+	if (MSM_DISP_SNAPSHOT_DUMP_IN_CONSOLE) {
+		p = drm_info_printer(disp_state->drm_dev->dev);
+		msm_disp_state_print(disp_state, &p);
+	}
+
+	/*
+	 * if devcoredump is not defined free the state immediately
+	 * otherwise it will be freed in the free handler.
+	 */
+#ifdef CONFIG_DEV_COREDUMP
+	dev_coredumpm(disp_state->dev, THIS_MODULE, disp_state, 0, GFP_KERNEL,
+			disp_devcoredump_read, disp_devcoredump_free);
+	disp_state->coredump_pending = true;
+#else
+	msm_disp_state_free(disp_state);
+#endif
+
+	mutex_unlock(&disp_state->mutex);
+}
+
+void msm_disp_snapshot_state(struct drm_device *drm_dev)
+{
+	struct msm_drm_private *priv;
+	struct msm_kms *kms;
+	struct msm_disp_state *disp_state;
+
+	if (!drm_dev) {
+		DRM_ERROR("invalid params\n");
+		return;
+	}
+
+	priv = drm_dev->dev_private;
+	kms = priv->kms;
+	disp_state = kms->disp_state;
+
+	if (!disp_state) {
+		DRM_ERROR("invalid params\n");
+		return;
+	}
+
+	/*
+	 * if there is a coredump pending return immediately till dump
+	 * if read by userspace or timeout happens
+	 */
+	if (disp_state->coredump_pending) {
+		DRM_DEBUG("coredump is pending read\n");
+		return;
+	}
+
+	kthread_queue_work(disp_state->dump_worker,
+			&disp_state->dump_work);
+}
+
+int msm_disp_snapshot_init(struct drm_device *drm_dev)
+{
+	struct msm_drm_private *priv;
+	struct msm_disp_state *disp_state;
+	struct msm_kms *kms;
+
+	if (!drm_dev) {
+		DRM_ERROR("invalid params\n");
+		return -EINVAL;
+	}
+
+	priv = drm_dev->dev_private;
+	kms = priv->kms;
+
+	disp_state = devm_kzalloc(drm_dev->dev, sizeof(struct msm_disp_state), GFP_KERNEL);
+
+	mutex_init(&disp_state->mutex);
+
+	disp_state->dev = drm_dev->dev;
+	disp_state->drm_dev = drm_dev;
+
+	INIT_LIST_HEAD(&disp_state->blocks);
+
+	disp_state->dump_worker = kthread_create_worker(0, "%s", "disp_snapshot");
+	if (IS_ERR(disp_state->dump_worker))
+		DRM_ERROR("failed to create disp state task\n");
+
+	kthread_init_work(&disp_state->dump_work, _msm_disp_snapshot_work);
+
+	kms->disp_state = disp_state;
+
+	return 0;
+}
+
+void msm_disp_snapshot_destroy(struct drm_device *drm_dev)
+{
+	struct msm_kms *kms;
+	struct msm_drm_private *priv;
+	struct msm_disp_state *disp_state;
+
+	if (!drm_dev) {
+		DRM_ERROR("invalid params\n");
+		return;
+	}
+
+	priv = drm_dev->dev_private;
+	kms = priv->kms;
+	disp_state = kms->disp_state;
+
+	if (disp_state->dump_worker)
+		kthread_destroy_worker(disp_state->dump_worker);
+
+	list_del(&disp_state->blocks);
+
+	mutex_destroy(&disp_state->mutex);
+}

drivers/gpu/drm/msm/disp/msm_disp_snapshot.h

@@ -0,0 +1,153 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2020-2021, The Linux Foundation. All rights reserved.
+ */
+
+#ifndef MSM_DISP_SNAPSHOT_H_
+#define MSM_DISP_SNAPSHOT_H_
+
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_device.h>
+#include "../../../drm_crtc_internal.h"
+#include <drm/drm_print.h>
+#include <drm/drm_atomic.h>
+#include <linux/debugfs.h>
+#include <linux/list.h>
+#include <linux/delay.h>
+#include <linux/spinlock.h>
+#include <linux/ktime.h>
+#include <linux/debugfs.h>
+#include <linux/uaccess.h>
+#include <linux/dma-buf.h>
+#include <linux/slab.h>
+#include <linux/list_sort.h>
+#include <linux/pm.h>
+#include <linux/pm_runtime.h>
+#include <linux/kthread.h>
+#include <linux/devcoredump.h>
+#include <stdarg.h>
+#include "msm_kms.h"
+
+#define MSM_DISP_SNAPSHOT_MAX_BLKS		10
+
+/* debug option to print the registers in logs */
+#define MSM_DISP_SNAPSHOT_DUMP_IN_CONSOLE 0
+
+/* print debug ranges in groups of 4 u32s */
+#define REG_DUMP_ALIGN		16
+
+/**
+ * struct msm_disp_state - structure to store current dpu state
+ * @dev: device pointer
+ * @drm_dev: drm device pointer
+ * @mutex: mutex to serialize access to serialze dumps, debugfs access
+ * @coredump_pending: coredump is pending read from userspace
+ * @atomic_state: atomic state duplicated at the time of the error
+ * @dump_worker: kworker thread which runs the dump work
+ * @dump_work: kwork which dumps the registers and drm state
+ * @timestamp: timestamp at which the coredump was captured
+ */
+struct msm_disp_state {
+	struct device *dev;
+	struct drm_device *drm_dev;
+	struct mutex mutex;
+
+	bool coredump_pending;
+
+	struct list_head blocks;
+
+	struct drm_atomic_state *atomic_state;
+
+	struct kthread_worker *dump_worker;
+	struct kthread_work dump_work;
+	ktime_t timestamp;
+};
+
+/**
+ * struct msm_disp_state_block - structure to store each hardware block state
+ * @name: name of the block
+ * @drm_dev: handle to the linked list head
+ * @size: size of the register space of this hardware block
+ * @state: array holding the register dump of this hardware block
+ * @base_addr: starting address of this hardware block's register space
+ */
+struct msm_disp_state_block {
+	char name[SZ_128];
+	struct list_head node;
+	unsigned int size;
+	u32 *state;
+	void __iomem *base_addr;
+};
+
+/**
+ * msm_disp_snapshot_init - initialize display snapshot
+ * @drm_dev:	drm device handle
+ *
+ * Returns:		0 or -ERROR
+ */
+int msm_disp_snapshot_init(struct drm_device *drm_dev);
+
+/**
+ * msm_disp_snapshot_destroy - destroy the display snapshot
+ * @drm_dev:    drm device handle
+ *
+ * Returns:	none
+ */
+void msm_disp_snapshot_destroy(struct drm_device *drm_dev);
+
+/**
+ * msm_disp_snapshot_state - trigger to dump the display snapshot
+ * @drm_dev:	handle to drm device
+
+ * Returns:	none
+ */
+void msm_disp_snapshot_state(struct drm_device *drm_dev);
+
+/**
+ * msm_disp_state_get - get the handle to msm_disp_state struct from the drm device
+ * @drm:	    handle to drm device
+
+ * Returns:	handle to the msm_disp_state struct
+ */
+struct msm_disp_state *msm_disp_state_get(struct drm_device *drm);
+
+/**
+ * msm_disp_state_print - print out the current dpu state
+ * @disp_state:	    handle to drm device
+ * @p:	    handle to drm printer
+ *
+ * Returns:	none
+ */
+void msm_disp_state_print(struct msm_disp_state *disp_state, struct drm_printer *p);
+
+/**
+ * msm_disp_snapshot_capture_state - utility to capture atomic state and hw registers
+ * @disp_state:	    handle to msm_disp_state struct
+
+ * Returns:	none
+ */
+void msm_disp_snapshot_capture_state(struct msm_disp_state *disp_state);
+
+/**
+ * msm_disp_state_free - free the memory after the coredump has been read
+ * @disp_state:	    handle to struct msm_disp_state
+
+ * Returns: none
+ */
+void msm_disp_state_free(struct msm_disp_state *disp_state);
+
+/**
+ * msm_disp_snapshot_add_block - add a hardware block with its register dump
+ * @disp_state:	    handle to struct msm_disp_state
+ * @name:           name of the hardware block
+ * @len:            size of the register space of the hardware block
+ * @base_addr:      starting address of the register space of the hardware block
+ * @fmt:            format in which the block names need to be printed
+ *
+ * Returns: none
+ */
+__printf(4, 5)
+void msm_disp_snapshot_add_block(struct msm_disp_state *disp_state, u32 len,
+		void __iomem *base_addr, const char *fmt, ...);
+
+#endif /* MSM_DISP_SNAPSHOT_H_ */