patches/mm-migrate_device-add-tracepoints-for-debugging.patch

From: Balbir Singh <balbirs@nvidia.com>
Subject: mm/migrate_device: add tracepoints for debugging
Date: Thu, 16 Oct 2025 16:46:19 +1100

Add tracepoints for debugging device migration flow in migrate_device.c. 
This is helpful in debugging how long migration took (time can be tracked
backwards from migrate_device_finalize to migrate_vma_setup).

A combination of these events along with existing thp:*, exceptions:* and
migrate:* is very useful for debugging issues related to migration.

Link: https://lkml.kernel.org/r/20251016054619.3174997-1-balbirs@nvidia.com
Signed-off-by: Balbir Singh <balbirs@nvidia.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Zi Yan <ziy@nvidia.com>
Cc: Joshua Hahn <joshua.hahnjy@gmail.com>
Cc: Rakie Kim <rakie.kim@sk.com>
Cc: Byungchul Park <byungchul@sk.com>
Cc: Gregory Price <gourry@gourry.net>
Cc: Ying Huang <ying.huang@linux.alibaba.com>
Cc: Alistair Popple <apopple@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 include/trace/events/migrate_device.h |  196 ++++++++++++++++++++++++
 mm/migrate_device.c                   |   11 +
 2 files changed, 207 insertions(+)

diff --git a/include/trace/events/migrate_device.h a/include/trace/events/migrate_device.h
new file mode 100644
--- /dev/null
+++ a/include/trace/events/migrate_device.h
@@ -0,0 +1,196 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2025, NVIDIA CORPORATION & AFFILIATES
+ */
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM migrate_device
+
+#if !defined(_TRACE_MIGRATE_DEVICE_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_MIGRATE_DEVICE_H
+
+#include <linux/tracepoint.h>
+#include <linux/migrate.h>
+
+TRACE_EVENT(migrate_vma_setup,
+
+	TP_PROTO(unsigned long start, unsigned long end, unsigned long nr_pages),
+
+	TP_ARGS(start, end, nr_pages),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, start)
+		__field(unsigned long, end)
+		__field(unsigned long, nr_pages)
+	),
+
+	TP_fast_assign(
+		__entry->start = start;
+		__entry->end = end;
+		__entry->nr_pages = nr_pages;
+	),
+
+	TP_printk("start=0x%lx end=0x%lx nr_pages=%lu",
+		__entry->start, __entry->end, __entry->nr_pages)
+);
+
+TRACE_EVENT(migrate_vma_collect,
+
+	TP_PROTO(unsigned long start, unsigned long end, unsigned long npages),
+
+	TP_ARGS(start, end, npages),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, start)
+		__field(unsigned long, end)
+		__field(unsigned long, npages)
+	),
+
+	TP_fast_assign(
+		__entry->start = start;
+		__entry->end = end;
+		__entry->npages = npages;
+	),
+
+	TP_printk("start=0x%lx end=0x%lx npages=%lu",
+		__entry->start, __entry->end, __entry->npages)
+);
+
+TRACE_EVENT(migrate_vma_collect_skip,
+
+	TP_PROTO(unsigned long start, unsigned long end),
+
+	TP_ARGS(start, end),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, start)
+		__field(unsigned long, end)
+	),
+
+	TP_fast_assign(
+		__entry->start = start;
+		__entry->end = end;
+	),
+
+	TP_printk("start=0x%lx end=0x%lx", __entry->start, __entry->end)
+);
+
+TRACE_EVENT(migrate_vma_collect_hole,
+
+	TP_PROTO(unsigned long start, unsigned long end, unsigned long npages),
+
+	TP_ARGS(start, end, npages),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, start)
+		__field(unsigned long, end)
+		__field(unsigned long, npages)
+	),
+
+	TP_fast_assign(
+		__entry->start = start;
+		__entry->end = end;
+		__entry->npages = npages;
+	),
+
+	TP_printk("start=0x%lx end=0x%lx npages=%lu",
+		__entry->start, __entry->end, __entry->npages)
+);
+
+TRACE_EVENT(migrate_vma_unmap,
+
+	TP_PROTO(unsigned long npages, unsigned long cpages),
+
+	TP_ARGS(npages, cpages),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, npages)
+		__field(unsigned long, cpages)
+	),
+
+	TP_fast_assign(
+		__entry->npages = npages;
+		__entry->cpages = cpages;
+	),
+
+	TP_printk("npages=%lu cpages=%lu",
+		__entry->npages, __entry->cpages)
+);
+
+TRACE_EVENT(migrate_device_pages,
+
+	TP_PROTO(unsigned long npages, unsigned long migrated),
+
+	TP_ARGS(npages, migrated),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, npages)
+		__field(unsigned long, migrated)
+	),
+
+	TP_fast_assign(
+		__entry->npages = npages;
+		__entry->migrated = migrated;
+	),
+
+	TP_printk("npages=%lu migrated=%lu",
+		__entry->npages, __entry->migrated)
+);
+
+TRACE_EVENT(migrate_vma_pages,
+
+	TP_PROTO(unsigned long npages, unsigned long start, unsigned long end),
+
+	TP_ARGS(npages, start, end),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, npages)
+		__field(unsigned long, start)
+		__field(unsigned long, end)
+	),
+
+	TP_fast_assign(
+		__entry->npages = npages;
+		__entry->start = start;
+		__entry->end = end;
+	),
+
+	TP_printk("npages=%lu start=0x%lx end=0x%lx",
+		__entry->npages, __entry->start, __entry->end)
+);
+
+TRACE_EVENT(migrate_device_finalize,
+
+	TP_PROTO(unsigned long npages),
+
+	TP_ARGS(npages),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, npages)
+	),
+
+	TP_fast_assign(
+		__entry->npages = npages;
+	),
+
+	TP_printk("npages=%lu", __entry->npages)
+);
+
+TRACE_EVENT(migrate_vma_finalize,
+
+	TP_PROTO(unsigned long npages),
+
+	TP_ARGS(npages),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, npages)
+	),
+
+	TP_fast_assign(
+		__entry->npages = npages;
+	),
+
+	TP_printk("npages=%lu", __entry->npages)
+);
+#endif /* _TRACE_MIGRATE_DEVICE_H */
+
+#include <trace/define_trace.h>
--- a/mm/migrate_device.c~mm-migrate_device-add-tracepoints-for-debugging
+++ a/mm/migrate_device.c
@@ -18,6 +18,9 @@
 #include <asm/tlbflush.h>
 #include "internal.h"
 
+#define CREATE_TRACE_POINTS
+#include <trace/events/migrate_device.h>
+
 static int migrate_vma_collect_skip(unsigned long start,
 				    unsigned long end,
 				    struct mm_walk *walk)
@@ -25,6 +28,8 @@ static int migrate_vma_collect_skip(unsi
 	struct migrate_vma *migrate = walk->private;
 	unsigned long addr;
 
+	trace_migrate_vma_collect_skip(start, end);
+
 	for (addr = start; addr < end; addr += PAGE_SIZE) {
 		migrate->dst[migrate->npages] = 0;
 		migrate->src[migrate->npages++] = 0;
@@ -69,6 +74,7 @@ static int migrate_vma_collect_hole(unsi
 		migrate->cpages++;
 	}
 
+	trace_migrate_vma_collect_hole(start, end, migrate->npages);
 	return 0;
 }
 
@@ -517,6 +523,7 @@ static void migrate_vma_collect(struct m
 
 	mmu_notifier_invalidate_range_end(&range);
 	migrate->end = migrate->start + (migrate->npages << PAGE_SHIFT);
+	trace_migrate_vma_collect(migrate->start, migrate->end, migrate->npages);
 }
 
 /*
@@ -748,6 +755,8 @@ int migrate_vma_setup(struct migrate_vma
 	if (args->fault_page && !PageLocked(args->fault_page))
 		return -EINVAL;
 
+	trace_migrate_vma_setup(args->start, args->end, nr_pages);
+
 	memset(args->src, 0, sizeof(*args->src) * nr_pages);
 	args->cpages = 0;
 	args->npages = 0;
@@ -1259,6 +1268,7 @@ EXPORT_SYMBOL(migrate_device_pages);
 void migrate_vma_pages(struct migrate_vma *migrate)
 {
 	__migrate_device_pages(migrate->src, migrate->dst, migrate->npages, migrate);
+	trace_migrate_device_pages(migrate->npages, migrate->npages);
 }
 EXPORT_SYMBOL(migrate_vma_pages);
 
@@ -1312,6 +1322,7 @@ static void __migrate_device_finalize(un
 			folio_put(dst);
 		}
 	}
+	trace_migrate_vma_finalize(npages);
 }
 
 /*
_