System V AMD64 ABI calling conventions

The calling convention of the System V AMD64 ABI is followed on Solaris, Linux, FreeBSD, Mac OS X, and other UNIX-like or POSIX-compliant operating systems. The first six integer or pointer arguments are passed in registers RDI, RSI, RDX, RCX, R8, and R9, while XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6 and XMM7 are used for floating point arguments. For system calls, R10 is used instead of RCX. As in the Microsoft x64 calling convention, additional arguments are passed on the stack and the return value is stored in RAX.

Registers RBP, RBX, and R12-R15 are callee-save registers; all others must be saved by the caller if they wish to preserve their values.

Unlike the Microsoft calling convention, a shadow space is not provided; on function entry, the return address is adjacent to the seventh integer argument on the stack.


Linux 不使用 chroot 临时替换C运行时库

有些时候我们需要在老旧的 Linux 系统上运行一些依赖于较新版本 C 库的应用程序或库,应用程序会因为系统中安装的C库缺少符号还启动失败。解决方法之一就是临时替换使用非系统安装的C运行时库。使用临时C库需要做些什么配置及会带来哪些问题呢?

1. 下载与目标应用程序版本相匹配的临时C库,解压缩到临时位置 A。
2. 需要设置 LD_LIBRARY_PATH 环境变量指向目标临时C库的存储位置 A。
3. 需要通过与临时C库匹配的 启用应用程序。因为应用程序默认是链接了一个绝对路径的,如 x86_64 是 /lib64/

使用临时C库的 启动的应用程序执行系统标准命令的子进程出错,原因是因为环境变量 LD_LIBRARY_PATH 被子进程继承,从而导致子进程在执行系统C库的ld.so中加载了版本不匹配的临时C库。

在合适的时机清除环境变量 LD_LIBRARY_PATH,最合适的时机应用就是执行目标应用程序 main 函数之前啦。这里又要用到了之前写过的方法 => Linux 平台一种进程代码注入方法

/* fakemain.c
 * Heiher <>
#include <stdio.h>
#include <stdlib.h>
#define __USE_GNU
#include <dlfcn.h>
__libc_start_main(int (*main)(int, char **, char **),
			int argc, char **ubp_av, void (*init)(void),
			void (*fini)(void), void (*rtld_fini)(void),
			void (*stack_end))
	int (*__libc_start_main_real)(int (*main) (int, char **, char **),
				int argc, char **ubp_av, void (*init)(void),
				void (*fini)(void), void (*rtld_fini)(void),
				void (*stack_end));
	unsetenv ("LD_PRELOAD");
	unsetenv ("LD_LIBRARY_PATH");
	__libc_start_main_real = dlsym(RTLD_NEXT, "__libc_start_main");
	return __libc_start_main_real(main, argc, ubp_av, init, fini,
				rtld_fini, stack_end);
gcc -fPIC -O3 -shared -o fakemain.c -ldl

设置环境变量 LD_PRELOAD=/xxx/,运行目标应用程序在执行 main 之前即会清除 LD_PRELOAD 和 LD_LIBRARY_PATH 变量。

为了方便使用我还写了个 wrapper,使用方法是将真实的目标应用程序 xxx 重命令为 xxx.bin,然后创建个符号链接 xxx 指向 wrapper,执行时直接执行 xxx,wrapper 会自动设置所需要的环境变量。

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
main (int argc, char *argv[])
	int i;
	char buf[1024], path[1024];
	char *str, *root, *args[512];
	/* get FAKE_ROOT */
	root = getenv ("FAKE_ROOT");
	if (!root) {
		fprintf (stderr, "Please set environment FAKE_ROOT!\n");
		return -1;
	/* export PATH */
	str = getenv ("PATH");
	if (!str) {
		fprintf (stderr, "Get environment PATH failed!\n");
		return -2;
	if (NULL == strstr (str, root)) {
		strcpy (buf, root);
		strcat (buf, "/bin:");
		strcat (buf, str);
		if (0 != setenv ("PATH", buf, 1)) {
			fprintf (stderr, "Set environment PATH failed!\n");
			return -3;
	/* export LD_PRELOAD */
	strcpy (buf, root);
	strcat (buf, "/lib64/");
	if (0 != setenv ("LD_PRELOAD", buf, 1)) {
		fprintf (stderr, "Set environment LD_PRELOAD failed!\n");
		return -4;
	/* export LD_LIBRARY_PATH */
	strcpy (buf, root);
	strcat (buf, "/lib64");
	if (0 != setenv ("LD_LIBRARY_PATH", buf, 1)) {
		fprintf (stderr, "Set environment LD_LIBRARY_PATH failed!\n");
		return -5;
	/* set new path */
	strcpy (path, root);
	strcat (path, "/lib64/");
	args[0] = path;
	/* set real program path */
	strcpy (buf, root);
	strcat (buf, "/bin/");
	strcat (buf, argv[0]);
	strcat (buf, ".bin");
	args[1] = buf;
	/* copy arguments */
	for (i=1; i<argc; i++)
	      args[i+1] = argv[i];
	args[i+1] = NULL;
	/* run real program */
	return execv (path, args);;
gcc -O3 -o wrapper wrapper.c


mips64el toolchain for x86_64

mips64el toolchain 是用于在 x86_64 平台交叉编译 mips64el 目标程序的工具集,该工具集分为两种大版本:odd-spreg 和 no-odd-spreg,其中龙芯仅适用 no-odd-spreg 版本。系统库包含 mips64el o32, n32 和 n64 多种版本的库,分别有依赖于 Linux 2.6 内核和 Linux 3.4 内核的两种版本。另外还有支持 x86_64 交叉编译 Mozilla JS 引擎的支持包。

Source: mips64el-toolchain-2.src.tar.xz
toolchain: mips64el-toolchain-2.x64.tar.xz
system libaries (Linux 2.6): mips64el-toolchain-linux-2.6-2.x64.tar.xz
system libaries (Linux 3.4): mips64el-toolchain-linux-3.4-2.x64.tar.xz
system libaries (Linux 3.4 MozJS): mips64el-toolchain-linux-3.4-mozjs-2.x64.tar.xz
toolchain: mips64el-toolchain-2.x64.tar.xz
system libaries (Linux 2.6): mips64el-toolchain-linux-2.6-2.x64.tar.xz
system libaries (Linux 3.4): mips64el-toolchain-linux-3.4-2.x64.tar.xz


sudo tar --numeric-owner -xf xxxx -C /


export PATH=${PATH}:/opt/mips64el-toolchain/bin


sudo ln -s -f linux-2.6 /opt/mips64el-toolchain/platforms/current


# MIPS32 o32
mips64el-unknown-linux-gnu-gcc -march=mips32r2 -mabi=32 -o test test.c
# MIPS64 n32
mips64el-unknown-linux-gnu-gcc -march=mips64r2 -mabi=n32 -o test test.c
# MIPS64 n64
mips64el-unknown-linux-gnu-gcc -march=mips64r2 -mabi=64 -o test test.c


How to connect to a WPA/WPA2 WiFi network using Linux command line

This is a step-to-step guide for connecting to a WPA/WPA2 WiFi network via the Linux command line interface. The tools are:
1. wpa_supplicant
2. iw
3. ip
4. ping

iw is the basic tool for WiFi network-related tasks, such as finding the WiFi device name, and scanning access points. wpa_supplicant is the wireless tool for connecting to a WPA/WPA2 network. ip is used for enabling/disabling devices, and finding out general network interface information.

The steps for connecting to a WPA/WPA2 network are:

1. Find out the wireless device name.

    $ /sbin/iw dev
    	Interface wlan0
    		ifindex 3
    		type managed

The above output showed that the system has 1 physical WiFi card, designated as phy#0. The device name is wlan0. The type specifies the operation mode of the wireless device. managed means the device is a WiFi station or client that connects to an access point.

2. Check that the wireless device is up.

    $ ip link show wlan0
    3: wlan0: (BROADCAST,MULTICAST) mtu 1500 qdisc noop state DOWN mode DEFAULT qlen 1000
        link/ether 74:e5:43:a1:ce:65 brd ff:ff:ff:ff:ff:ff

Look for the word “UP” inside the brackets in the first line of the output.

In the above example, wlan0 is not UP. Execute the following command to bring it up:

    $ sudo ip link set wlan0 up  
    [sudo] password for peter:

Note: you need root privilege for the above operation.

If you run the show link command again, you can tell that wlan0 is now UP.

    $ ip link show wlan0
    3: wlan0: (NO-CARRIER,BROADCAST,MULTICAST,UP) mtu 1500 qdisc mq state DOWN mode DEFAULT qlen 1000
        link/ether 74:e5:43:a1:ce:65 brd ff:ff:ff:ff:ff:ff

3. Check the connection status.

    $ /sbin/iw wlan0 link
    Not connected.

The above output shows that you are not connected to any network.

4. Scan to find out what WiFi network(s) are detected

    $ sudo /sbin/iw wlan0 scan
    BSS 00:14:d1:9c:1f:c8 (on wlan0)
            ... sniped ...
    	freq: 2412
    	SSID: gorilla
    	RSN:	 * Version: 1
    		 * Group cipher: CCMP
    		 * Pairwise ciphers: CCMP
    		 * Authentication suites: PSK
    		 * Capabilities: (0x0000)
            ... sniped ...

The 2 important pieces of information from the above are the SSID and the security protocol (WPA/WPA2 vs WEP). The SSID from the above example is gorilla. The security protocol is RSN, also commonly referred to as WPA2. The security protocol is important because it determines what tool you use to connect to the network.

5. Connect to WPA/WPA2 WiFi network.
This is a 2 step process. First, you generate a configuration file for wpa_supplicant that contains the pre-shared key (“passphrase”) for the WiFi network.

    $ sudo -s
    [sudo] password for peter: 
    $ wpa_passphrase gorilla >> /etc/wpa_supplicant.conf 
    ...type in the passphrase and hit enter...

wpa_passphrase takes the SSID as the single argument. You must type in the passphrase for the WiFi network gorilla after you run the command. Using that information, wpa_passphrase will output the necessary configuration statements to the standard output. Those statements are appended to the wpa_supplicant configuration file located at /etc/wpa_supplicant.conf.

Note: you need root privilege to write to /etc/wpa_supplicant.conf.

    $ cat /etc/wpa_supplicant.conf 
    # reading passphrase from stdin

The second step is to run wpa_supplicant with the new configuration file.

    $ sudo wpa_supplicant -B -D wext -i wlan0 -c /etc/wpa_supplicant.conf
    -B means run wpa_supplicant in the background.
    -D specifies the wireless driver. wext is the generic driver.
    -c specifies the path for the configuration file.

Use the iw command to verify that you are indeed connected to the SSID.

    $ /sbin/iw wlan0 link
    Connected to 00:14:d1:9c:1f:c8 (on wlan0)
    	SSID: gorilla
    	freq: 2412
    	RX: 63825 bytes (471 packets)
    	TX: 1344 bytes (12 packets)
    	signal: -27 dBm
    	tx bitrate: 6.5 MBit/s MCS 0
    	bss flags:	short-slot-time
    	dtim period:	0
    	beacon int:	100

6. Obtain IP address by DHCP

    $ sudo dhclient wlan0

Use the ip command to verify the IP address assigned by DHCP. The IP address is from below.

    $ ip addr show wlan0
    3: wlan0:  mtu 1500 qdisc mq state UP qlen 1000
        link/ether 74:e5:43:a1:ce:65 brd ff:ff:ff:ff:ff:ff
        inet brd scope global wlan0
        inet6 fe80::76e5:43ff:fea1:ce65/64 scope link 
           valid_lft forever preferred_lft forever

7. Add default routing rule.
The last configuration step is to make sure that you have the proper routing rules.

    $ ip route show dev wlan0  proto kernel  scope link  src

The above routing table contains only 1 rule which redirects all traffic destined for the local subnet (192.168.1.x) to the wlan0 interface. You may want to add a default routing rule to pass all other traffic through wlan0 as well.

    $ sudo ip route add default via dev wlan0
    $ ip route show
    default via dev wlan0 dev wlan0  proto kernel  scope link  src

8. ping external ip address to test connectivity

    $ ping
    PING ( 56(84) bytes of data.
    64 bytes from icmp_req=1 ttl=48 time=135 ms
    64 bytes from icmp_req=2 ttl=48 time=135 ms
    64 bytes from icmp_req=3 ttl=48 time=134 ms
    --- ping statistics ---
    3 packets transmitted, 3 received, 0% packet loss, time 2000ms
    rtt min/avg/max/mdev = 134.575/134.972/135.241/0.414 ms

The above series of steps is a very verbose explanation of how to connect a WPA/WPA2 WiFi network. Some steps can be skipped as you connect to the same access point for a second time. For instance, you already know the WiFi device name, and the configuration file is already set up for the network. The process needs to be tailored according to your situation.


How to disable auto suspend when I close laptop lid?

Edit /etc/systemd/logind.conf and make sure you have,


which will make it ignore the lid being closed. (You may need to also undo the other changes you’ve made).

Full details over at the archlinux Wiki.

The man page for logind.conf also has the relevant information,

   HandlePowerKey=, HandleSuspendKey=, HandleHibernateKey=,
       Controls whether logind shall handle the system power and sleep
       keys and the lid switch to trigger actions such as system power-off
       or suspend. Can be one of ignore, poweroff, reboot, halt, kexec,
       suspend, hibernate, hybrid-sleep and lock. If ignore logind will
       never handle these keys. If lock all running sessions will be
       screen locked. Otherwise the specified action will be taken in the
       respective event. Only input devices with the power-switch udev tag
       will be watched for key/lid switch events.  HandlePowerKey=
       defaults to poweroff.  HandleSuspendKey= and HandleLidSwitch=
       default to suspend.  HandleHibernateKey= defaults to hibernate.


x86 pslldq to Loongson psllq

x86 pslldq 指令逻辑左移字节为单位的数据,而转换成龙芯的MMI只能使用 dsll 和 dsrl 指令模拟实现,需要特别注意的是 dsll 和 dsrl 指令移动的数据是以位为单位的。

/* SSE: pslldq (bytes) */
#define _mm_psllq(_D, _d, _s, _s64, _tf)                    \
        "subu %["#_tf"], %["#_s64"], %["#_s"] \n\t"         \
        "dsrl %["#_tf"], %["#_d"l], %["#_tf"] \n\t"         \
        "dsll %["#_D"h], %["#_d"h], %["#_s"] \n\t"          \
        "dsll %["#_D"l], %["#_d"l], %["#_s"] \n\t"          \
        "or %["#_D"h], %["#_D"h], %["#_tf"] \n\t"
pslldq $4, %xmm0 => mm_psllq(d, d, s32, s64, t)


看龙芯3A的 dmtc1 指令有多慢!

龙芯2F和3A处理器都实现了与 x86 MMX 基本兼容的 SIMD,即 MMI,该 ASE 是在浮点部件中的实现的,并且复用了 64-bit 的浮点寄存器(FPR)。在使用 MMI 时不可避免的会使用到通用寄存器向浮点器移动数据的情况,那么 dmtc1 的效率如何呢?

GPR 向 FPR 移动数据的指令共有3种:
mtc1 : 从 GPR 向 FPR 移动 32-bit 的数据,64-bit 平台上目标 FPR 的高 32-bit 清 0。
mthc1 : 从 GPR (低 32-bit)向 FPR 的高 32-bit 移动 32-bit 的数据,目标 FPR 的低 32-bit 数据保留。
dmtc1 : 从 GPR 向 FPR 移动 64-bit 数据。

从上面的说明可以看出, dmtc1 的功能是可以使用 mtc1 与 mthc1 模拟实现的,那么我们就设计个实验程序来验证一下这两条方式的时间开销分别如何吧。

for (i=0; i<100000000; i++) {
#if 0
    move $2, $3
    mtc1 $3, $f31
    dsra $3, 32
    mthc1 $3, $f31
    move $3, $2
    dmtc1 $3, $f31
    dmtc1 $3, $f31

在 MIPS64 系统上,每个循环中做8次GPR2FPR的数据移动,其 dmtc1 实现时间大概为 0m4.463s,而 mtc1 与 mthc1 组合实现为 0m3.857s,后者如不做寄存器的保存恢复,开销仅为 0m1.791s。