kk Blog —— 通用基础


date [-d @int|str] [+%s|"+%F %T"]
netstat -ltunp

abrt 应用程序core dump

一、安装

1
yum install abrt

二、设置

1
2
ulimit -c
ulimit -c unlimited

三、常见错误

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
1、ERROR
$ tail -f /var/log/message
abrtd: Package 'XXX' isn't signed with proper key

$ vim /etc/abrt/abrt.conf
OR
$ vim /etc/abrt/abrt-action-save-package-data.conf
OpenGPGCheck = no

2、ERROR
tail -f /var/log/message
abrtd: Duplicate: UUID

Whenever a problem is detected, ABRT compares it with all 
existing problem data and determines whether that same problem 
has been recorded. If it has been, the existing problem data 
is updated and the most recent (duplicate) problem is not recorded again.

3、
ProcessUnpackaged = <yes/no>
This directive tells ABRT whether to process crashes 
in executables that do not belong to any package. 

abrt

http://docs.fedoraproject.org/en-US/Fedora/14/html/Deployment_Guide/configuring.html
https://fedorahosted.org/releases/a/b/abrt/Deployment_Guide.html

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
21.6. Configuring ABRT

ABRT's main configuration file is /etc/abrt/abrt.conf. 
ABRT plugins can be configured through their config files, 
located in the /etc/abrt/plugins/ directory.

After changing and saving the abrt.conf configuration file, 
you must restart the abrtd daemon—as root—for the new settings to take effect:

~]# service abrtd restart

The following configuration directives are currently supported in /etc/abrt/abrt.conf.

[ Common ] Section DirectivesOpenGPGCheck = <yes/no>

Setting the OpenGPGCheck directive to yes (the default setting) tells 
ABRT to only analyze and handle crashes in applications provided by 
packages which are signed by the GPG keys whose locations are listed 
in the /etc/abrt/gpg_keys file. Setting OpenGPGCheck to no tells 
ABRT to catch crashes in all programs.

BlackList = nspluginwrapper, valgrind, strace, avant-window-navigator, [<additional_packages> ]

Crashes in packages and binaries listed after the BlackList directive 
will not be handled by ABRT. If you want ABRT to ignore other packages 
and binaries, list them here separated by commas.

ProcessUnpackaged = <yes/no>

This directive tells ABRT whether to process crashes in executables 
that do not belong to any package.    

BlackListedPaths = /usr/share/doc/*, */example*

Crashes in executables in these paths will be ignored by ABRT.

Database = SQLite3

This directive instructs ABRT to store its crash data in the SQLite3 database. 
Other databases are not currently supported. However, 
ABRT's plugin architecture allows for future support for alternative databases.

#WatchCrashdumpArchiveDir = /var/spool/abrt-upload/

This directive is commented out by default. 
Enable (uncomment) it if you want abrtd to auto-unpack crashdump tarballs 
which appear in the specified directory — in this case /var/spool/abrt-upload/ — 
(for example, uploaded via ftp, scp, etc.). You must ensure that whatever 
directory you specify in this directive exists and is writable for abrtd. 
abrtd will not create it automatically.

MaxCrashReportsSize = <size_in_megabytes>

This option sets the amount of storage space, in megabytes, 
used by ABRT to store all crash information from all users. 
The default setting is 1000 MB. Once the quota specified here has been met, 
ABRT will continue catching crashes, and in order to make room for the new crash dumps, 
it will delete the oldest and largest ones.

ActionsAndReporters = SOSreport, [<additional_plugins> ]

This option tells ABRT to run the specified plugin(s) immediately 
after a crash is detected and saved. For example, the SOSreport plugin runs 
the sosreport tool which adds the data collected by it to the created crash dump. 
You can turn this behavior off by commenting out this line. For further fine-tuning,
 you can add SOSreport (or any other specified plugin) to either the CCpp or 
Python options to make ABRT run sosreport (or any other specified plugin) after 
any C and C++ or Python applications crash, respectively. For more information 
on various Action and Reporter plugins, refer to Section 21.3, “ ABRT Plugins”

[ AnalyzerActionsAndReporters ] Section Directives

This section allows you to associate certain analyzer actions and reporter 
actions to run when ABRT catches kernel oopses or crashes in C, C++ or Python programs. 
The actions and reporters specified in any of the directives below will run only 
if you run abrt-gui or abrt-cli and report the crash that occurred. 
If you do not specify any actions and reporters in these directives, 
you will not be able to report a crash via abrt-gui or abrt-cli. 
The order of actions and reporters is important. Commenting out a directive, 
will cause ABRT not to catch the crashes associated with that directive. 
For example, commenting out the Kerneloops line will cause ABRT not to catch kernel oopses.

Kerneloops = RHTSupport, Logger

This directive specifies that, for kernel oopses, 
both the RHTSupport and Logger reporters will be run.

CCpp = RHTSupport, Logger

This directive specifies that, when C or C++ program crashes occur, 
both the RHTSupport and Logger reporters will be run.

Python = RHTSupport, Logger

This directive specifies that, when Python program crashes occur, 
both the RHTSupport and Logger reporters will be run.

Each of these destinations' details can be specified in the corresponding 
plugins/*.conf file. For example, plugins/RHTSupport.conf specifies 
which RHTSupport URL to use (set to https://api.access.redhat.com/rs by default), 
the user's login name, password for logging in to the RHTSupport site, 
etc. All these options can also be configured through the abrt-gui application
 (for more information on plugin configuration refer to Section 21.3, “ ABRT Plugins”).

[ Cron ] Section Directives <time> = <action_to_run>

The [ Cron ] section of abrt.conf allows you to specify the exact time, 
or elapsed amount of time between, when ABRT should run a certain action, 
such as scanning for kernel oopses or performing file transfers. 
You can list further actions to run by appending them to the end of this section.

Example 21.1. [ Cron ] section of /etc/abrt/abrt.conf

# Which Action plugins to run repeatedly
[ Cron ]
# h:m - at h:m
# s - every s seconds
120 = KerneloopsScanner
#02:00 = FileTransfer


The format for an entry is either 
<time_in_seconds> = <action_to_run> or <hh:mm> = <action_to_run> , 
where hh (hour) is in the range 00-23 
(all hours less than 10 should be zero-filled, i.e. preceded by a 0), 
and mm (minute) is 00-59, zero-filled likewise. 

centos系统各种包下载

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git

https://buildlogs.centos.org/c7.2009.u.x86_64/kernel/


0 centos 系統原包

爬取el7所有版本

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
import re
import urllib

def getHtml(url):
	page = urllib.urlopen(url)
	html = page.read()
	return html

def getHref(html, reg):
	reg = re.compile(reg)
	reslist = re.findall(reg, html)
	return reslist

URL = "https://buildlogs.centos.org/"
html = getHtml(URL)
c7Href = getHref(html, r'href="(c7.+)/"')
for ver in c7Href:
	if '.a32' in ver or '.a64' in ver or '.p32' in ver or '.i386' in ver:
		continue
	url1 = URL + ver + "/kernel/"
	print url1
	html = getHtml(url1)
	dateHref = getHref(html, r'href="(20............)/"')
	for date in dateHref:
		url2 = url1 + date + "/"
		html = getHtml(url2)
		kernelHref = getHref(html, r'href="(.+el7.x86_64)/"')
		for kver in kernelHref:
			print url2 + kver

https://buildlogs.centos.org/c7-dotnet/kernel/
https://buildlogs.centos.org/c7-epel/kernel/
https://buildlogs.centos.org/c7-extras.x86_64/kernel/
https://buildlogs.centos.org/c7-plus.x86_64/kernel/
https://buildlogs.centos.org/c7-plus/kernel/
https://buildlogs.centos.org/c7-rt/kernel/
https://buildlogs.centos.org/c7-updates/kernel/
https://buildlogs.centos.org/c7.00.02/kernel/
https://buildlogs.centos.org/c7.00.02/kernel/20140529190808/3.10.0-121.el7.x86_64
https://buildlogs.centos.org/c7.00.03/kernel/
https://buildlogs.centos.org/c7.00.03/kernel/20140609184350/3.10.0-121.el7.x86_64
https://buildlogs.centos.org/c7.00.04/kernel/
https://buildlogs.centos.org/c7.00.04/kernel/20140612172658/3.10.0-123.el7.x86_64
https://buildlogs.centos.org/c7.00.04/kernel/20140619231033/3.10.0-123.el7.x86_64
https://buildlogs.centos.org/c7.01.00/kernel/
https://buildlogs.centos.org/c7.01.00/kernel/20150306113403/3.10.0-229.el7.x86_64
https://buildlogs.centos.org/c7.01.u/kernel/
https://buildlogs.centos.org/c7.01.u/kernel/20150327030147/3.10.0-229.1.2.el7.x86_64
https://buildlogs.centos.org/c7.01.u/kernel/20150513100324/3.10.0-229.4.2.el7.x86_64
https://buildlogs.centos.org/c7.01.u/kernel/20150623220331/3.10.0-229.7.2.el7.x86_64
https://buildlogs.centos.org/c7.01.u/kernel/20150806010338/3.10.0-229.11.1.el7.x86_64
https://buildlogs.centos.org/c7.01.u/kernel/20150915124206/3.10.0-229.14.1.el7.x86_64
https://buildlogs.centos.org/c7.01.u/kernel/20150915150313/3.10.0-229.14.1.el7.x86_64
https://buildlogs.centos.org/c7.01.u/kernel/20151103190728/3.10.0-229.20.1.el7.x86_64
https://buildlogs.centos.org/c7.1511.00/kernel/
https://buildlogs.centos.org/c7.1511.00/kernel/20151119220809/3.10.0-327.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/
https://buildlogs.centos.org/c7.1511.exp/kernel/20151016161452/4.2.0-1.centos.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20151016163253/4.2.0-1.centos.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20151016164628/4.2.0-1.centos.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160321183722/4.3.3-200.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160324145107/4.4.6-301.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160324192831/4.4.6-301.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160325232209/4.4.6-301.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160415133359/4.4.7-301.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160506113850/4.4.9-301.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160601130532/4.4.11-301.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160602142804/4.4.12-301.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160608070903/4.4.13-301.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160620154312/4.4.13-303.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160625132228/4.4.14-201.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160625133615/4.4.14-201.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160815150500/4.4.17-201.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160815161333/4.4.17-201.el7.x86_64
https://buildlogs.centos.org/c7.1511.exp/kernel/20160817141019/4.4.18-201.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/
https://buildlogs.centos.org/c7.1511.u/kernel/20151209124337/3.10.0-327.3.1.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20151209140627/3.10.0-327.3.1.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20160105150501/3.10.0-327.4.4.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20160125220424/3.10.0-327.4.5.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20160217024115/3.10.0-327.10.1.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20160331160950/3.10.0-327.13.1.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20160512110105/3.10.0-327.18.2.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20160623161521/3.10.0-327.22.2.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20160802204906/3.10.0-327.28.2.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20160818163946/3.10.0-327.28.3.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20160918123639/3.10.0-327.36.1.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20161010214658/3.10.0-327.36.2.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20161010215511/3.10.0-327.36.2.el7.x86_64
https://buildlogs.centos.org/c7.1511.u/kernel/20161024152721/3.10.0-327.36.3.el7.x86_64
https://buildlogs.centos.org/c7.1611.00/kernel/
https://buildlogs.centos.org/c7.1611.01/kernel/
https://buildlogs.centos.org/c7.1611.01/kernel/20161117160457/3.10.0-514.el7.x86_64
https://buildlogs.centos.org/c7.1611.exp/kernel/
https://buildlogs.centos.org/c7.1611.exp/kernel/20171018140113/4.9.57-204.el7.x86_64
https://buildlogs.centos.org/c7.1611.exp/kernel/20171120151900/4.9.63-204.el7.x86_64
https://buildlogs.centos.org/c7.1611.u/kernel/
https://buildlogs.centos.org/c7.1611.u/kernel/20161207134106/3.10.0-514.2.2.el7.x86_64
https://buildlogs.centos.org/c7.1611.u/kernel/20170118010633/3.10.0-514.6.1.el7.x86_64
https://buildlogs.centos.org/c7.1611.u/kernel/20170223034721/3.10.0-514.2.2.el7.x86_64
https://buildlogs.centos.org/c7.1611.u/kernel/20170303004149/3.10.0-514.10.2.el7.x86_64
https://buildlogs.centos.org/c7.1611.u/kernel/20170412150118/3.10.0-514.16.1.el7.x86_64
https://buildlogs.centos.org/c7.1611.u/kernel/20170525170145/3.10.0-514.21.1.el7.x86_64
https://buildlogs.centos.org/c7.1611.u/kernel/20170620122143/3.10.0-514.21.2.el7.x86_64
https://buildlogs.centos.org/c7.1611.u/kernel/20170620132051/3.10.0-514.21.2.el7.x86_64
https://buildlogs.centos.org/c7.1611.u/kernel/20170628200657/3.10.0-514.26.1.el7.x86_64
https://buildlogs.centos.org/c7.1611.u/kernel/20170704132018/3.10.0-514.26.2.el7.x86_64
https://buildlogs.centos.org/c7.1708.00/kernel/
https://buildlogs.centos.org/c7.1708.00/kernel/20170822030048/3.10.0-693.el7.x86_64
https://buildlogs.centos.org/c7.1708.exp.x86_64/kernel/
https://buildlogs.centos.org/c7.1708.u.x86_64/kernel/
https://buildlogs.centos.org/c7.1708.u.x86_64/kernel/20170823130501/3.10.0-693.1.1.el7.x86_64
https://buildlogs.centos.org/c7.1708.u.x86_64/kernel/20170906160426/3.10.0-693.2.1.el7.x86_64
https://buildlogs.centos.org/c7.1708.u.x86_64/kernel/20170913001530/3.10.0-693.2.2.el7.x86_64
https://buildlogs.centos.org/c7.1708.u.x86_64/kernel/20171023132245/3.10.0-693.5.2.el7.x86_64
https://buildlogs.centos.org/c7.1708.u.x86_64/kernel/20171204203818/3.10.0-693.11.1.el7.x86_64
https://buildlogs.centos.org/c7.1708.u/kernel/
https://buildlogs.centos.org/c7.1708.u/kernel/20170823130501/3.10.0-693.1.1.el7.x86_64
https://buildlogs.centos.org/c7.1708.u/kernel/20170906160426/3.10.0-693.2.1.el7.x86_64
https://buildlogs.centos.org/c7.1708.u/kernel/20170913001530/3.10.0-693.2.2.el7.x86_64
https://buildlogs.centos.org/c7.1708.u/kernel/20171023132245/3.10.0-693.5.2.el7.x86_64
https://buildlogs.centos.org/c7.1708.u/kernel/20171204203818/3.10.0-693.11.1.el7.x86_64
https://buildlogs.centos.org/c7.common/kernel/

1、系统包

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
http://mirror.centos.org/centos/6.5/os/x86_64/Packages/
国内地址
http://isoredirect.centos.org/centos/6.5/isos/x86_64/  
ex:
	http://mirror.symnds.com/distributions/CentOS-vault/5.5/isos/x86_64/  
	http://mirrors.stuhome.net/centos/6.5/isos/x86_64/  
	http://mirrors.neusoft.edu.cn/centos/6.5/isos/x86_64/
	http://mirrors.163.com/centos/6.5/isos/x86_64/
	http://mirrors.hust.edu.cn/centos/6.5/isos/x86_64/
	http://centos.ustc.edu.cn/centos/6.5/isos/x86_64/
	http://mirror.bit.edu.cn/centos/6.5/isos/x86_64/
	http://mirrors.tuna.tsinghua.edu.cn/centos/6.5/isos/x86_64/
	http://mirrors.grandcloud.cn/centos/6.5/isos/x86_64/
	http://mirror.neu.edu.cn/centos/6.5/isos/x86_64/
	http://mirrors.btte.net/centos/6.5/isos/x86_64/
	http://mirrors.hustunique.com/centos/6.5/isos/x86_64/
	http://mirrors.aliyun.com/centos/6.5/isos/x86_64/

2、debuginfo包:

1
http://debuginfo.centos.org/6/x86_64/

3、src.prm包

1
2
3
4
ftp://ftp.redhat.com/pub/redhat/linux/enterprise
ftp://ftp.redhat.com/pub/redhat/linux/enterprise/5Client/en/os/SRPMS/kexec-tools-1.102pre-154.el5.src.rpm
ftp://ftp.redhat.com/pub/redhat/linux/enterprise/5Client/en/os/SRPMS/kexec-tools-1.102pre-164.el5.src.rpm
http://vault.centos.org/5.11/os/SRPMS/kexec-tools-1.102pre-165.el5.src.rpm

4、各种包

1
pkgs/org

tsc时钟初始化

tsc时钟源初始化
1
2
3
4
5
6
7
8
9
//    调用路径:time_init->tsc_init
//    函数任务:
//        1.矫正tsc,获取tsc频率,设置cpu频率等于tsc频率
//        2.初始化基于tsc的延迟函数
//        3.检查tsc的特性
//            3.1 tsc之间是否同步
//                3.1.1 如果tsc之间不同步,标记tsc不稳定,设置rating=0
//            3.2 tsc是否稳定
//        4.注册tsc时钟源设备
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
void __init tsc_init(void)
{
	u64 lpj;
	int cpu;

	//矫正tsc,获取tsc频率
	tsc_khz = x86_platform.calibrate_tsc();
	//cpu频率等于tsc频率
	cpu_khz = tsc_khz;
	//计算辅助cycle到ns转换的辅助参数scale
	for_each_possible_cpu(cpu)
	    set_cyc2ns_scale(cpu_khz, cpu);
	//初始化基于tsc的延迟函数,ndely,udelay,mdelay
	use_tsc_delay();
	//检查cpu之间tsc是否同步
	if (unsynchronized_tsc())
	    mark_tsc_unstable("TSCs unsynchronized");
	//检查tsc是否可靠
	check_system_tsc_reliable();
	//注册tsc时钟源设备
	init_tsc_clocksource();
}
延迟函数ndelay,udelay,mdelay

通过tsc实现短延迟

1
2
3
4
5
void use_tsc_delay(void)
{
	//通过tsc进行短延迟
	delay_fn = delay_tsc;
}
tsc延迟函数

通过rep_nop实现轮询时的短延迟,查询tsc时禁止内核抢占,确保不受不同cpu间影响。

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
static void delay_tsc(unsigned long loops)
{
	unsigned long bclock, now;
	int cpu;
	//短延迟,禁止内核抢占
	preempt_disable();
	//delay_tsc当前运行的cpu
	cpu = smp_processor_id();
	rdtsc_barrier();
	rdtscl(bclock);
	for (;;) {
	    rdtsc_barrier();
	    rdtscl(now);
	    if ((now - bclock) >= loops)
	        break;
	    //允许rt策略进程运行
	    preempt_enable();
	    //空操作
	    rep_nop();
	    preempt_disable();

	    //delay_tsc在运行过程中,可能会迁移到不同的cpu
	    //tsc
	    if (unlikely(cpu != smp_processor_id())) {
	        loops -= (now - bclock);
	        cpu = smp_processor_id();
	        rdtsc_barrier();
	        rdtscl(bclock);
	    }
	}
	preempt_enable();
}
检查tsc是否同步
1
2
3
4
5
6
//    调用路径:tsc_init->unsynchronized_tsc
//    检查办法:
//        1.如果apic在多块板卡,则tsc不同步
//        2.如果cpuid显示具有稳定的tsc,则tsc同步
//        3.intel cpu的tsc都是同步的
//        4.默认其他品牌的多核的tsc不同步
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
__cpuinit int unsynchronized_tsc(void)
{
	//如果apic分布在多块板卡上,tsc可能不同步
	if (apic_is_clustered_box())
	    return 1;
	//cpu具有稳定的tsc
	if (boot_cpu_has(X86_FEATURE_CONSTANT_TSC))
	    return 0;
	//intel cpu的tsc都是同步的
	if (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL) {
	    //非intel cpu,如果cpu个数>1,则认为不同步
	    if (num_possible_cpus() > 1)
	        tsc_unstable = 1;
	}
	return tsc_unstable;
}
标记tsc不稳定
1
2
3
4
//    调用路径:tsc_init->mark_tsc_unstable
//    函数任务:
//        1.如果tsc时钟已经注册,异步设置tsc的rating=0,标识其不稳定
//        2.如果tsc时钟还未注册,同步设置tsc的rating=0,标识其不稳定
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
void mark_tsc_unstable(char *reason)
{
	if (!tsc_unstable) {
	    tsc_unstable = 1;
	    sched_clock_stable = 0;
	    //tsc已经注册,
	    if (clocksource_tsc.mult)
	    {
	        clocksource_mark_unstable(&clocksource_tsc);
	    }
	    //如果tsc时钟源未注册,修改rating为最低,从而不会被当做最佳的时钟源
	    else {
	        clocksource_tsc.flags |= CLOCK_SOURCE_UNSTABLE;
	        clocksource_tsc.rating = 0;
	    }
	}
}
注册tsc时钟源
1
2
3
4
5
6
//    函数任务:
//        1.计算tsc的mult
//        2.检查tsc是否稳定
//            2.1 如果tsc不稳定,降低其rating,清除时钟源连续标志
//        3.向系统注册tsc clocksource
//    调用路径:tsc_init->init_tsc_clocksource
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
static void __init init_tsc_clocksource(void)
{
	// 计算tsc的mult
	clocksource_tsc.mult = clocksource_khz2mult(tsc_khz,
	        clocksource_tsc.shift);
	// 如果tsc的可靠性已经验证,则清除 必须验证 标记
	if (tsc_clocksource_reliable)
	    clocksource_tsc.flags &= ~CLOCK_SOURCE_MUST_VERIFY;
	
	// 检查tsc是否稳定
	// 在tsc_init前通过全局变量标记tsc是否稳定,可靠
	if (check_tsc_unstable()) {
	    // 如果tsc不稳定,则降低rating最低,清除连续标记
	    clocksource_tsc.rating = 0;
	    clocksource_tsc.flags &= ~CLOCK_SOURCE_IS_CONTINUOUS;
	}
	// 向系统注册tsc clocksource
	clocksource_register(&clocksource_tsc);
}

TSC时间错误

arch/x86/kernel/tsc.c:
开机初始化会调用tsc_init() -> set_cyc2ns_scale() 设置per_cpu变量cyc2ns、cyc2ns_offset。以供后面shced_clock()->native_sched_clock()->__cycles_2_ns()调用。

在cpufreq_tsc()中如果
//cpu具有稳定的tsc
if (boot_cpu_has(X86_FEATURE_CONSTANT_TSC))
return 0;
所以一般不会注册time_cpufreq_notifier函数,也就不会再调用set_cyc2ns_scale。

  • 现象:top、ps出来的TIME和CPU的值非常异常。
1
2
3
4
5
6
7
8
9
10
11
// 查看TSC寄存器的值
#include <stdio.h>

int main()
{
	    unsigned long low, high, val;
	    asm volatile("rdtsc": "=a" (low), "=d" (high));
	    val = ((low) | ((unsigned long)(high) << 32));
	    printf("%lu\n", val);
	    return 0;
}

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=733043

Xeon E5 has a bug, it doesn’t reset TSC on warm reboot, just keep it instead. see “BT81. X X X No Fix TSC is Not Affected by Warm Reset” http://www.intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-e5-family-spec-update.pdf

And also kernel 2.6.32 has a bug.
Xeon bug + kernel bug = hung after warm reboot (or kexec) after 208.5 days
since booting. So, administrators should shutdown it once at all, then
boot it again because “shutdown -r” causes hang up.

Red Hat has released a fix for this as kernel 2.6.32-220, 2.6.32-279
and 2.6.32-358 series (RHEL6.x) https://access.redhat.com/site/solutions/433883 (for detail subscriber only :-(

Attached patch is based on upstream patch.
see http://kernel.opensuse.org/cgit/kernel/patch/?id=9993bc635d01a6ee7f6b833b4ee65ce7c06350b1


Red Hat Enterprise Linux 6.1 (kernel-2.6.32-131.26.1.el6 and newer)
Red Hat Enterprise Linux 6.2 (kernel-2.6.32-220.4.2.el6 and newer)
Red Hat Enterprise Linux 6.3 (kernel-2.6.32-279 series)
Red Hat Enterprise Linux 6.4 (kernel-2.6.32-358 series)
Any Intel® Xeon® E5, Intel® Xeon® E5 v2, or Intel® Xeon® E7 v2 series processor


From 9993bc635d01a6ee7f6b833b4ee65ce7c06350b1 Mon Sep 17 00:00:00 2001
From: Salman Qazi <sqazi@google.com>
Date: Sat, 10 Mar 2012 00:41:01 +0000
Subject: sched/x86: Fix overflow in cyc2ns_offset

When a machine boots up, the TSC generally gets reset. However, when kexec is used to boot into a kernel, the TSC value would be carried over from the previous kernel. The computation of cycns_offset in set_cyc2ns_scale is prone to an overflow, if the machine has been up more than 208 days prior to the kexec. The overflow happens when we multiply *scale, even though there is enough room to store the final answer.

We fix this issue by decomposing tsc_now into the quotient and remainder of division by CYC2NS_SCALE_FACTOR and then performing the multiplication separately on the two components.

Refactor code to share the calculation with the previous fix in __cycles_2_ns().

Signed-off-by: Salman Qazi <sqazi@google.com>
Acked-by: John Stultz <john.stultz@linaro.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Turner <pjt@google.com>
Cc: john stultz <johnstul@us.ibm.com>
Link: http://lkml.kernel.org/r/20120310004027.19291.88460.stgit@dungbeetle.mtv.corp.google.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>


patch: http://kernel.opensuse.org/cgit/kernel/patch/?id=9993bc635d01a6ee7f6b833b4ee65ce7c06350b1

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
diff --git a/arch/x86/include/asm/timer.h b/arch/x86/include/asm/timer.h
index 431793e..34baa0e 100644
--- a/arch/x86/include/asm/timer.h
+++ b/arch/x86/include/asm/timer.h
@@ -57,14 +57,10 @@ DECLARE_PER_CPU(unsigned long long, cyc2ns_offset);
 
 static inline unsigned long long __cycles_2_ns(unsigned long long cyc)
 {
- unsigned long long quot;
- unsigned long long rem;
  int cpu = smp_processor_id();
  unsigned long long ns = per_cpu(cyc2ns_offset, cpu);
- quot = (cyc >> CYC2NS_SCALE_FACTOR);
- rem = cyc & ((1ULL << CYC2NS_SCALE_FACTOR) - 1);
- ns += quot * per_cpu(cyc2ns, cpu) +
-     ((rem * per_cpu(cyc2ns, cpu)) >> CYC2NS_SCALE_FACTOR);
+ ns += mult_frac(cyc, per_cpu(cyc2ns, cpu),
+         (1UL << CYC2NS_SCALE_FACTOR));
  return ns;
 }
 
diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index a62c201..183c592 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -620,7 +620,8 @@ static void set_cyc2ns_scale(unsigned long cpu_khz, int cpu)
 
  if (cpu_khz) {
      *scale = (NSEC_PER_MSEC << CYC2NS_SCALE_FACTOR)/cpu_khz;
-     *offset = ns_now - (tsc_now * *scale >> CYC2NS_SCALE_FACTOR);
+     *offset = ns_now - mult_frac(tsc_now, *scale,
+                      (1UL << CYC2NS_SCALE_FACTOR));
  }
 
  sched_clock_idle_wakeup_event(0);
diff --git a/include/linux/kernel.h b/include/linux/kernel.h
index e834342..d801acb 100644
--- a/include/linux/kernel.h
+++ b/include/linux/kernel.h
@@ -85,6 +85,19 @@
 }                            \
 )
 
+/*
+ * Multiplies an integer by a fraction, while avoiding unnecessary
+ * overflow or loss of precision.
+ */
+#define mult_frac(x, numer, denom)(          \
+{                            \
+ typeof(x) quot = (x) / (denom);         \
+ typeof(x) rem  = (x) % (denom);         \
+ (quot * (numer)) + ((rem * (numer)) / (denom)); \
+}                            \
+)
+
+
 #define _RET_IP_     (unsigned long)__builtin_return_address(0)
 #define _THIS_IP_  ({ __label__ __here; __here: (unsigned long)&&__here; }) 

mark problem

C语言中宽字符和多字节字符 MB_CUR_MAX

stdlib.h
MB_CUR_MAX 当前locale中多字节字符的最大字节数目
如果把字符串当成多字节字符处理会慢很多
sort命令的一些版本中会根据locale中的LANG来设置比较函数,如果没设置好会导致sort跑的很慢。

内核栈溢出

因为内核栈中栈底保存thread_info,所以如果内核栈溢出会破坏thread_info,这样当进程发生睡眠、中断、抢占等调度时就会出错。
容易报try_to_wakeup+XXX错误,等等

定时器 timer 注意

如果mod_timer设置成马上执行,然后下面又是setup_timer。
有可能的情况是:进入了中断执行过程,可是setup_timer又改变timer结构,导致继续执行时用到timer结构出错。

查看内核栈

用 echo t > /proc/sysrq-trigger 把内核栈整个打出来。
可以看/proc/{pid}/wchan,里面是该进程阻塞位置的内核函数名,在所有办法都没戏的时候可以看它。

mark

修复ext4日志(jbd2)bug
rhel6再次发现jbd2的bug
rhel6 的软RAID问题
stable pages
追踪CPU跑满