Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Jaewon Kim <jaewon31.kim@samsung.com>
To:     willy@infradead.org, walken@google.com, bp@suse.de,
        akpm@linux-foundation.org, srostedt@vmware.com
Cc:     linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        jaewon31.kim@gmail.com, Jaewon Kim <jaewon31.kim@samsung.com>
Subject: [PATCH v3 2/2] mm: mmap: add trace point of vm_unmapped_area
Date:   Fri, 20 Mar 2020 14:58:23 +0900
Message-Id: <20200320055823.27089-3-jaewon31.kim@samsung.com>
In-Reply-To: <20200320055823.27089-1-jaewon31.kim@samsung.com>
Content-Type: text/plain; charset="utf-8"
CMS-TYPE: 101P
DLP-Filter: Pass
References: <20200320055823.27089-1-jaewon31.kim@samsung.com>
        <CGME20200320055839epcas1p189100549687530619d8a19919e8b5de0@epcas1p1.samsung.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Even on 64 bit kernel, the mmap failure can happen for a 32 bit task.
Virtual memory space shortage of a task on mmap is reported to userspace
as -ENOMEM. It can be confused as physical memory shortage of overall
system.

The vm_unmapped_area can be called to by some drivers or other kernel
core system like filesystem. In my platform, GPU driver calls to
vm_unmapped_area and the driver returns -ENOMEM even in GPU side
shortage. It can be hard to distinguish which code layer returns the
-ENOMEM.

Create mmap trace file and add trace point of vm_unmapped_area.

i.e.)
277.156599: vm_unmapped_area: addr=77e0d03000 err=0 total_vm=0x17014b flags=0x1 len=0x400000 lo=0x8000 hi=0x7878c27000 mask=0x0 ofs=0x1
342.838740: vm_unmapped_area: addr=0 err=-12 total_vm=0xffb08 flags=0x0 len=0x100000 lo=0x40000000 hi=0xfffff000 mask=0x0 ofs=0x22

Signed-off-by: Jaewon Kim <jaewon31.kim@samsung.com>
---
v3: reduce fast_assign and print format
v2: use trace point rather than printk with ratelimit
v1: use printk with ratelimit 
---
 include/trace/events/mmap.h | 48 +++++++++++++++++++++++++++++++++++++++++++++
 mm/mmap.c                   | 12 ++++++++++--
 2 files changed, 58 insertions(+), 2 deletions(-)
 create mode 100644 include/trace/events/mmap.h

diff --git a/include/trace/events/mmap.h b/include/trace/events/mmap.h
new file mode 100644
index 000000000000..986a41b6cfa9
--- /dev/null
+++ b/include/trace/events/mmap.h
@@ -0,0 +1,48 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mmap
+
+#if !defined(_TRACE_MMAP_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_MMAP_H
+
+#include <linux/tracepoint.h>
+
+TRACE_EVENT(vm_unmapped_area,
+
+	TP_PROTO(unsigned long addr, struct vm_unmapped_area_info *info),
+
+	TP_ARGS(addr, info),
+
+	TP_STRUCT__entry(
+		__field(unsigned long,	addr)
+		__field(unsigned long,	total_vm)
+		__field(unsigned long,	flags)
+		__field(unsigned long,	length)
+		__field(unsigned long,	low_limit)
+		__field(unsigned long,	high_limit)
+		__field(unsigned long,	align_mask)
+		__field(unsigned long,	align_offset)
+	),
+
+	TP_fast_assign(
+		__entry->addr = addr;
+		__entry->total_vm = current->mm->total_vm;
+		__entry->flags = info->flags;
+		__entry->length = info->length;
+		__entry->low_limit = info->low_limit;
+		__entry->high_limit = info->high_limit;
+		__entry->align_mask = info->align_mask;
+		__entry->align_offset = info->align_offset;
+	),
+
+	TP_printk("addr=%lx err=%ld total_vm=0x%lx flags=0x%lx len=0x%lx lo=0x%lx hi=0x%lx mask=0x%lx ofs=0x%lx\n",
+		IS_ERR_VALUE(__entry->addr) ? 0 : __entry->addr,
+		IS_ERR_VALUE(__entry->addr) ? __entry->addr : 0,
+		__entry->total_vm, __entry->flags, __entry->length,
+		__entry->low_limit, __entry->high_limit, __entry->align_mask,
+		__entry->align_offset)
+);
+#endif
+
+/* This part must be outside protection */
+#include <trace/define_trace.h>
diff --git a/mm/mmap.c b/mm/mmap.c
index ba990c20ecc2..94ae18398c59 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -53,6 +53,9 @@
 #include <asm/tlb.h>
 #include <asm/mmu_context.h>
 
+#define CREATE_TRACE_POINTS
+#include <trace/events/mmap.h>
+
 #include "internal.h"
 
 #ifndef arch_mmap_check
@@ -2061,10 +2064,15 @@ static unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info)
  */
 unsigned long vm_unmapped_area(struct vm_unmapped_area_info *info)
 {
+	unsigned long addr;
+
 	if (info->flags & VM_UNMAPPED_AREA_TOPDOWN)
-		return unmapped_area_topdown(info);
+		addr = unmapped_area_topdown(info);
 	else
-		return unmapped_area(info);
+		addr = unmapped_area(info);
+
+	trace_vm_unmapped_area(addr, info);
+	return addr;
 }
 
 #ifndef arch_get_mmap_end
-- 
2.13.7