-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathfreezer-for-linux-5.0-rc1.patch
311 lines (296 loc) · 12.6 KB
/
freezer-for-linux-5.0-rc1.patch
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
From 01f15501b8a1e3a6eb5329170533ac52614694d4 Mon Sep 17 00:00:00 2001
From: Dongli Zhang <[email protected]>
Date: Sat, 19 Jan 2019 21:02:31 +0800
Subject: [PATCH 1/1] freezer for linux-5.0-rc1
Signed-off-by: Dongli Zhang <[email protected]>
---
include/linux/freezer.h | 61 ++++++++++++++++++++++++++++++++
include/linux/sched.h | 16 +++++++++
kernel/freezer.c | 94 +++++++++++++++++++++++++++++++++++++++++++++++++
kernel/power/process.c | 9 +++++
4 files changed, 180 insertions(+)
diff --git a/include/linux/freezer.h b/include/linux/freezer.h
index 21f5aa0..06f0176 100644
--- a/include/linux/freezer.h
+++ b/include/linux/freezer.h
@@ -32,10 +32,33 @@ extern bool freezing_slow_path(struct task_struct *p);
/*
* Check if there is a request to freeze a process
*/
+/*
+ * 很多地方在调用
+ *
+ * 如果当前没有suspend (比如没有调用xen的do_suspend() )就返回false
+ * 否则调用freezing_slow_path()来slow path for testing whether a task needs to be frozen
+ * 比如如果设置了PF_NOFREEZE就返回false
+ *
+ * 就是检查系统是否处于freezing状态
+ */
static inline bool freezing(struct task_struct *p)
{
+ /*
+ * 除了cgroup,主要在freeze_processes()增加, 在thaw_processes()减少
+ *
+ * 就是说在以下增加:
+ * - drivers/xen/manage.c|103| <<do_suspend>> err = freeze_processes();
+ * - kernel/kexec_core.c|1134| <<kernel_kexec>> error = freeze_processes();
+ * - kernel/power/hibernate.c|716| <<hibernate>> error = freeze_processes();
+ * - kernel/power/hibernate.c|892| <<software_resume>> error = freeze_processes();
+ * - kernel/power/power.h|256| <<suspend_freeze_processes>> error = freeze_processes();
+ * - kernel/power/user.c|235| <<snapshot_ioctl>> error = freeze_processes();
+ */
if (likely(!atomic_read(&system_freezing_cnt)))
return false;
+ /*
+ * slow path for testing whether a task needs to be frozen
+ */
return freezing_slow_path(p);
}
@@ -52,18 +75,56 @@ extern void thaw_kernel_threads(void);
* DO NOT ADD ANY NEW CALLERS OF THIS FUNCTION
* If try_to_freeze causes a lockdep warning it means the caller may deadlock
*/
+/*
+ * called by:
+ * - include/linux/freezer.h|67| <<try_to_freeze>> return try_to_freeze_unsafe();
+ * - include/linux/freezer.h|136| <<freezer_count_unsafe>> try_to_freeze_unsafe();
+ *
+ * 调用freezing():
+ * 如果当前没有suspend (比如没有调用xen的do_suspend() )就返回false
+ * 否则调用freezing_slow_path()来slow path for testing whether a task needs to be frozen
+ * 比如如果设置了PF_NOFREEZE就返回false
+ * 如果freezing()返回true, 则调用__refrigerator()尝试冻结这个线程
+ */
static inline bool try_to_freeze_unsafe(void)
{
might_sleep();
+ /*
+ * freezing():
+ * 如果当前没有suspend (比如没有调用xen的do_suspend() )就返回false
+ * 否则调用freezing_slow_path()来slow path for testing whether a task needs to be frozen
+ * 比如如果设置了PF_NOFREEZE就返回false
+ *
+ * 就是检查系统是否处于freezing状态
+ */
if (likely(!freezing(current)))
return false;
+ /*
+ * 主动进入冻结
+ */
return __refrigerator(false);
}
+/*
+ * 调用freezing():
+ * 如果当前没有suspend (比如没有调用xen的do_suspend() )就返回false
+ * 否则调用freezing_slow_path()来slow path for testing whether a task needs to be frozen
+ * 比如如果设置了PF_NOFREEZE就返回false
+ * 如果freezing()返回true, 则调用__refrigerator()尝试冻结这个线程
+ *
+ * 很多内核线程在用,比如jbd2
+ */
static inline bool try_to_freeze(void)
{
if (!(current->flags & PF_NOFREEZE))
debug_check_no_locks_held();
+ /*
+ * 调用freezing():
+ * 如果当前没有suspend (比如没有调用xen的do_suspend() )就返回false
+ * 否则调用freezing_slow_path()来slow path for testing whether a task needs to be frozen
+ * 比如如果设置了PF_NOFREEZE就返回false
+ * 如果freezing()返回true, 则调用__refrigerator()尝试冻结这个线程
+ */
return try_to_freeze_unsafe();
}
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 89541d2..9110660 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1396,7 +1396,23 @@ extern struct pid *cad_pid;
#define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
#define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
#define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */
+/*
+ * 在以下被使用:
+ * - drivers/block/aoe/aoecmd.c|1234| <<kthread>> current->flags |= PF_NOFREEZE;
+ * - drivers/scsi/lpfc/lpfc_hbadisc.c|745| <<lpfc_do_work>> current->flags |= PF_NOFREEZE;
+ * - fs/exec.c|1293| <<flush_old_exec>> PF_NOFREEZE | PF_NO_SETAFFINITY);
+ * - include/linux/freezer.h|65| <<try_to_freeze>> if (!(current->flags & PF_NOFREEZE))
+ * - kernel/freezer.c|48| <<freezing_slow_path>> if (p->flags & (PF_NOFREEZE | PF_SUSPEND_TASK))
+ * - kernel/freezer.c|180| <<set_freezable>> current->flags &= ~PF_NOFREEZE;
+ * - kernel/kthread.c|569| <<kthreadd>> current->flags |= PF_NOFREEZE;
+ * - kernel/sched/core.c|3487| <<do_task_dead>> current->flags |= PF_NOFREEZE;
+ */
#define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
+/*
+ * 在以下被设置:
+ * - kernel/freezer.c|78| <<__refrigerator>> current->flags |= PF_FROZEN;
+ * - kernel/freezer.c|81| <<__refrigerator>> current->flags &= ~PF_FROZEN;
+ */
#define PF_FROZEN 0x00010000 /* Frozen for system suspend */
#define PF_KSWAPD 0x00020000 /* I am kswapd */
#define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */
diff --git a/kernel/freezer.c b/kernel/freezer.c
index b162b74..83349fb 100644
--- a/kernel/freezer.c
+++ b/kernel/freezer.c
@@ -11,6 +11,50 @@
#include <linux/freezer.h>
#include <linux/kthread.h>
+/*
+ * 假设没有冻结技术,进程可以在任意可调度的点暂停,而且直到
+ * cpu_down才会暂停并迁移.这会给系统带来很多问题.
+ *
+ * (1) 有可能破坏文件系统.在系统创建hibernate image到cpu down之间,如果有进程
+ * 还在修改文件系统的内容,这将会导致系统恢复之后无法完全恢复文件系统.
+ *
+ * (2) 有可能导致创建hibernation image失败.创建hibernation image需要足够的内存
+ * 空间,但是在这期间如果还有进程在申请内存,就可能导致创建失败.
+ *
+ * (3) 有可能干扰设备的suspend和resume.在cpu down之前,device suspend期间,如果
+ * 进程还在访问设备,尤其是访问竞争资源,就有可能引起设备suspend异常.
+ *
+ * (4) 有可能导致进程感知系统休眠.系统休眠的理想状态是所有任务对休眠过程无感知,睡
+ * 醒之后全部自动恢复工作,但是有些进程,比如某个进程需要所有cpu online才能正常
+ * 工作,如果进程不冻结,那么在休眠过程中将会工作异常.
+ */
+
+/*
+ * 标记系统freeze状态的有三个重要的全局变量:
+ * - pm_freezing
+ * - system_freezing_cnt
+ * - pm_nosig_freezing
+ *
+ * 如果全为0,表示系统未进入冻结;
+ * system_freezing_cnt>0表示系统进入冻结,pm_freezing=true表示冻结用户进程,pm_nosig_freezing=true表示冻结内核线程和workqueue.
+ * 它们会在freeze_processes和freeze_kernel_threads中置位,在thaw_processes和thaw_kernel_threads中清零.
+ *
+ * fake_signal_wake_up函数巧妙的利用了信号处理机制,只设置任务的
+ * TIF_SIGPENDING位,但不传递任何信号,然后唤醒任务;这样任务在返
+ * 回用户态时会进入信号处理流程,检查系统的freeze状态,并做相应处理.
+ */
+
+/*
+ * 在以下被使用:
+ * - kernel/freezer.c|15| <<global>> atomic_t system_freezing_cnt = ATOMIC_INIT(0);
+ * - include/linux/freezer.h|37| <<freezing>> if (likely(!atomic_read(&system_freezing_cnt)))
+ * - kernel/cgroup/freezer.c|116| <<freezer_css_online>> atomic_inc(&system_freezing_cnt);
+ * - kernel/cgroup/freezer.c|137| <<freezer_css_offline>> atomic_dec(&system_freezing_cnt);
+ * - kernel/cgroup/freezer.c|360| <<freezer_apply_state>> atomic_inc(&system_freezing_cnt);
+ * - kernel/cgroup/freezer.c|370| <<freezer_apply_state>> atomic_dec(&system_freezing_cnt);
+ * - kernel/power/process.c|135| <<freeze_processes>> atomic_inc(&system_freezing_cnt);
+ * - kernel/power/process.c|196| <<thaw_processes>> atomic_dec(&system_freezing_cnt);
+ */
/* total number of freezing conditions in effect */
atomic_t system_freezing_cnt = ATOMIC_INIT(0);
EXPORT_SYMBOL(system_freezing_cnt);
@@ -18,7 +62,25 @@ EXPORT_SYMBOL(system_freezing_cnt);
/* indicate whether PM freezing is in effect, protected by
* system_transition_mutex
*/
+/*
+ * 被以下使用:
+ * - drivers/ata/libata-scsi.c|4807| <<ata_scsi_hotplug>> while (pm_freezing)
+ * - kernel/freezer.c|86| <<freezing_slow_path>> if (pm_freezing && !(p->flags & PF_KTHREAD))
+ * - kernel/power/process.c|143| <<freeze_processes>> if (!pm_freezing)
+ * - kernel/power/process.c|148| <<freeze_processes>> pm_freezing = true;
+ * - kernel/power/process.c|204| <<thaw_processes>> if (pm_freezing)
+ * - kernel/power/process.c|206| <<thaw_processes>> pm_freezing = false;
+ *
+ * 只在freeze_processes()和thaw_processes()被修改
+ */
bool pm_freezing;
+/*
+ * 在以下被使用:
+ * - kernel/freezer.c|94| <<freezing_slow_path>> if (pm_nosig_freezing || cgroup_freezing(p))
+ * - kernel/power/process.c|185| <<freeze_kernel_threads>> pm_nosig_freezing = true;
+ * - kernel/power/process.c|207| <<thaw_processes>> pm_nosig_freezing = false;
+ * - kernel/power/process.c|240| <<thaw_kernel_threads>> pm_nosig_freezing = false;
+ */
bool pm_nosig_freezing;
/*
@@ -39,17 +101,30 @@ static DEFINE_SPINLOCK(freezer_lock);
* called under any context. The freezers are responsible for ensuring the
* target tasks see the updated state.
*/
+/*
+ * called only by:
+ * - include/linux/freezer.h|39| <<freezing>> return freezing_slow_path(p);
+ */
bool freezing_slow_path(struct task_struct *p)
{
+ /*
+ * 当前进程是否允许冻结
+ */
if (p->flags & (PF_NOFREEZE | PF_SUSPEND_TASK))
return false;
if (test_tsk_thread_flag(p, TIF_MEMDIE))
return false;
+ /*
+ * 系统冻结kernel threads
+ */
if (pm_nosig_freezing || cgroup_freezing(p))
return true;
+ /*
+ * 系统冻结用户进程
+ */
if (pm_freezing && !(p->flags & PF_KTHREAD))
return true;
@@ -58,6 +133,11 @@ bool freezing_slow_path(struct task_struct *p)
EXPORT_SYMBOL(freezing_slow_path);
/* Refrigerator is place where frozen processes are stored :-). */
+/*
+ * called by:
+ * - include/linux/freezer.h|86| <<try_to_freeze_unsafe>> return __refrigerator(false);
+ * - kernel/kthread.c|137| <<kthread_freezable_should_stop>> frozen = __refrigerator(true);
+ */
bool __refrigerator(bool check_kthr_stop)
{
/* Hmm, should we be allowed to suspend when there are realtime
@@ -72,6 +152,17 @@ bool __refrigerator(bool check_kthr_stop)
spin_lock_irq(&freezer_lock);
current->flags |= PF_FROZEN;
+ /*
+ * freezing():
+ * 如果当前没有suspend (比如没有调用xen的do_suspend() )就返回false
+ * 否则调用freezing_slow_path()来slow path for testing whether a task needs to be frozen
+ * 比如如果设置了PF_NOFREEZE就返回false
+ *
+ * if语句只要返回true就不设置PF_FROZEN
+ * 这个kthread就不会被放入refrigerator
+ *
+ * 判断系统是否还处于冻结, 如果系统已解冻，则取消冻结状态
+ */
if (!freezing(current) ||
(check_kthr_stop && kthread_should_stop()))
current->flags &= ~PF_FROZEN;
@@ -80,6 +171,9 @@ bool __refrigerator(bool check_kthr_stop)
if (!(current->flags & PF_FROZEN))
break;
was_frozen = true;
+ /*
+ * 这里就hang了 直到resume!!!
+ */
schedule();
}
diff --git a/kernel/power/process.c b/kernel/power/process.c
index 4b6a54d..6464f97 100644
--- a/kernel/power/process.c
+++ b/kernel/power/process.c
@@ -120,6 +120,15 @@ static int try_to_freeze_tasks(bool user_only)
*
* On success, returns 0. On failure, -errno and system is fully thawed.
*/
+/*
+ * called by:
+ * - drivers/xen/manage.c|103| <<do_suspend>> err = freeze_processes();
+ * - kernel/kexec_core.c|1134| <<kernel_kexec>> error = freeze_processes();
+ * - kernel/power/hibernate.c|716| <<hibernate>> error = freeze_processes();
+ * - kernel/power/hibernate.c|892| <<software_resume>> error = freeze_processes();
+ * - kernel/power/power.h|256| <<suspend_freeze_processes>> error = freeze_processes();
+ * - kernel/power/user.c|235| <<snapshot_ioctl>> error = freeze_processes();
+ */
int freeze_processes(void)
{
int error;
--
2.7.4