Arm Linux系統(tǒng)調(diào)用流程詳細(xì)解析
加入技術(shù)交流群
系統(tǒng)調(diào)用是操作系統(tǒng)提供的服務(wù)，用戶程序通過各種系統(tǒng)調(diào)用，來引用內(nèi)核提供的各種服務(wù)，系統(tǒng)調(diào)用的執(zhí)行讓用戶程序陷入內(nèi)核，該陷入動(dòng)作由swi軟中斷完成。
本文引用地址：http://m.butianyuan.cn/article/201611/318013.htm1、用戶可以通過兩種方式使用系統(tǒng)調(diào)用：
第一種方式是通過C庫函數(shù)，包括系統(tǒng)調(diào)用在C庫中的封裝函數(shù)和其他普通函數(shù)。
第二種方式是使用_syscall宏。2.6.18版本之前的內(nèi)核，在include/asm-i386/unistd.h文件中定義有7個(gè)_syscall宏，分別是：
_syscall0(type,name) _syscall1(type,name,type1,arg1) _syscall2(type,name,type1,arg1,type2,arg2) _syscall3(type,name,type1,arg1,type2,arg2,type3,arg3) _syscall4(type,name,type1,arg1,type2,arg2,type3,arg3,type4,arg4) _syscall5(type,name,type1,arg1,type2,arg2,type3,arg3,type4,arg4,type5,arg5) _syscall6(type,name,type1,arg1,type2,arg2,type3,arg3,type4,arg4,type5,arg5,type6,arg6)
其中，type表示所生成系統(tǒng)調(diào)用的返回值類型，name表示該系統(tǒng)調(diào)用的名稱，typeN、argN分別表示第N個(gè)參數(shù)的類型和名稱，它們的數(shù)目和_syscall后面的數(shù)字一樣大。
這些宏的作用是創(chuàng)建名為name的函數(shù)，_syscall后面跟的數(shù)字指明了該函數(shù)的參數(shù)的個(gè)數(shù)。
比如sysinfo系統(tǒng)調(diào)用用于獲取系統(tǒng)總體統(tǒng)計(jì)信息，使用_syscall宏定義為：
_syscall1(int, sysinfo, struct sysinfo *, info);
展開后的形式為：
int sysinfo(struct sysinfo * info) { long __res; __asm__ volatile("int $0x80" : "=a" (__res) : "0" (116),"b" ((long)(info))); do { if ((unsigned long)(__res) >= (unsigned long)(-(128 + 1))) { errno = -(__res); __res = -1; } return (int) (__res); } while (0); } 可以看出，_syscall1(int, sysinfo, struct sysinfo *, info)展開成一個(gè)名為sysinfo的函數(shù)，原參數(shù)int就是函數(shù)的返回類型，原參數(shù)struct sysinfo *和info分別構(gòu)成新函數(shù)的參數(shù)。
在程序文件里使用_syscall宏定義需要的系統(tǒng)調(diào)用，就可以在接下來的代碼中通過系統(tǒng)調(diào)用名稱直接調(diào)用該系統(tǒng)調(diào)用。下面是一個(gè)使用sysinfo系統(tǒng)調(diào)用的實(shí)例。
代碼清單5.1 sysinfo系統(tǒng)調(diào)用使用實(shí)例
#include#include #include #include /* for struct sysinfo */ _syscall1(int, sysinfo, struct sysinfo *, info); int main(void) { struct sysinfo s_info; int error; error = sysinfo(&s_info); printf("code error = %d/n", error); printf("Uptime = %lds/nLoad: 1 min %lu / 5 min %lu / 15 min %lu/n" "RAM: total %lu / free %lu / shared %lu/n" "Memory in buffers = %lu/nSwap: total %lu / free %lu/n" "Number of processes = %d/n", s_info.uptime, s_info.loads[0], s_info.loads[1], s_info.loads[2], s_info.totalram, s_info.freeram, s_info.sharedram, s_info.bufferram, s_info.totalswap, s_info.freeswap, s_info.procs); exit(EXIT_SUCCESS); }
但是自2.6.19版本開始，_syscall宏被廢除，我們需要使用syscall函數(shù)，通過指定系統(tǒng)調(diào)用號(hào)和一組參數(shù)來調(diào)用系統(tǒng)調(diào)用。
syscall函數(shù)原型為：
int syscall(int number, ...);
其中number是系統(tǒng)調(diào)用號(hào)，number后面應(yīng)順序接上該系統(tǒng)調(diào)用的所有參數(shù)。下面是gettid系統(tǒng)調(diào)用的調(diào)用實(shí)例。
代碼清單5.2 gettid系統(tǒng)調(diào)用使用實(shí)例
#include#include #include #define __NR_gettid 224 int main(int argc, char *argv[]) { pid_t tid; tid = syscall(__NR_gettid); }
大部分系統(tǒng)調(diào)用都包括了一個(gè)SYS_符號(hào)常量來指定自己到系統(tǒng)調(diào)用號(hào)的映射，因此上面第10行可重寫為：
tid = syscall(SYS_gettid);
2 系統(tǒng)調(diào)用與應(yīng)用編程接口（API）區(qū)別
應(yīng)用編程接口（API）與系統(tǒng)調(diào)用的不同在于，前者只是一個(gè)函數(shù)定義，說明了如何獲得一個(gè)給定的服務(wù)，而后者是通過軟件中斷向內(nèi)核發(fā)出的一個(gè)明確的請求。POSIX標(biāo)準(zhǔn)針對API，而不針對系統(tǒng)調(diào)用。Unix系統(tǒng)給程序員提供了很多API庫函數(shù)。libc的標(biāo)準(zhǔn)c庫所定義的一些API引用了封裝例程（wrapper routine）(其唯一目的就是發(fā)布系統(tǒng)調(diào)用)。通常情況下，每個(gè)系統(tǒng)調(diào)用對應(yīng)一個(gè)封裝例程，而封裝例程定義了應(yīng)用程序使用的API。反之則不然，一個(gè)API沒必要對應(yīng)一個(gè)特定的系統(tǒng)調(diào)用。從編程者的觀點(diǎn)看，API和系統(tǒng)調(diào)用之間的差別是沒有關(guān)系的：唯一相關(guān)的事情就是函數(shù)名、參數(shù)類型及返回代碼的含義。然而，從內(nèi)核設(shè)計(jì)者的觀點(diǎn)看，這種差別確實(shí)有關(guān)系，因?yàn)橄到y(tǒng)調(diào)用屬于內(nèi)核，而用戶態(tài)的庫函數(shù)不屬于內(nèi)核。
大部分封裝例程返回一個(gè)整數(shù)，其值的含義依賴于相應(yīng)的系統(tǒng)調(diào)用。返回-1通常表示內(nèi)核不能滿足進(jìn)程的請求。系統(tǒng)調(diào)用處理程序的失敗可能是由無效參數(shù)引起的，也可能是因?yàn)槿狈捎觅Y源，或硬件出了問題等等。在libd庫中定義的errno變量包含特定的出錯(cuò)碼。每個(gè)出錯(cuò)碼定義為一個(gè)常量宏。
當(dāng)用戶態(tài)的進(jìn)程調(diào)用一個(gè)系統(tǒng)調(diào)用時(shí)，CPU切換到內(nèi)核態(tài)并開始執(zhí)行一個(gè)內(nèi)核函數(shù)。因?yàn)閮?nèi)核實(shí)現(xiàn)了很多不同的系統(tǒng)調(diào)用，因此進(jìn)程必須傳遞一個(gè)名為系統(tǒng)調(diào)用號(hào)（system call number）的參數(shù)來識(shí)別所需的系統(tǒng)調(diào)用。所有的系統(tǒng)調(diào)用都返回一個(gè)整數(shù)值。這些返回值與封裝例程返回值的約定是不同的。在內(nèi)核中，整數(shù)或0表示系統(tǒng)調(diào)用成功結(jié)束，而負(fù)數(shù)表示一個(gè)出錯(cuò)條件。在后一種情況下，這個(gè)值就是存放在errno變量中必須返回給應(yīng)用程序的負(fù)出錯(cuò)碼。
3 系統(tǒng)調(diào)用執(zhí)行過程
ARM Linux系統(tǒng)利用SWI指令來從用戶空間進(jìn)入內(nèi)核空間，還是先讓我們了解下這個(gè)SWI指令吧。SWI指令用于產(chǎn)生軟件中斷，從而實(shí)現(xiàn)從用戶模式變換到管理模式，CPSR保存到管理模式的SPSR，執(zhí)行轉(zhuǎn)移到SWI向量。在其他模式下也可使用SWI指令，處理器同樣地切換到管理模式。指令格式如下：
SWI{cond} immed_24其中：
immed_2424位立即數(shù)，值為從0——16215之間的整數(shù)。
使用SWI指令時(shí)，通常使用一下兩種方法進(jìn)行參數(shù)傳遞，SWI異常處理程序可以提供相關(guān)的服務(wù)，這兩種方法均是用戶軟件協(xié)定。SWI異常中斷處理程序要通過讀取引起軟件中斷的SWI指令，以取得24為立即數(shù)。
1）、指令中24位的立即數(shù)指定了用戶請求的服務(wù)類型，參數(shù)通過通用寄存器傳遞。如：
MOV R0,#34SWI 12
2）、指令中的24位立即數(shù)被忽略，用戶請求的服務(wù)類型有寄存器R0的只決定，參數(shù)通過其他的通用寄存器傳遞。如：
MOV R0, #12MOV R1, #34SWI 0
在SWI異常處理程序中，去除SWI立即數(shù)的步驟為：首先確定一起軟中斷的SWI指令時(shí)ARM指令還是Thumb指令，這可通過對SPSR訪問得到；然后取得該SWI指令的地址，這可通過訪問LR寄存器得到；接著讀出指令，分解出立即數(shù)（低24位）。
由用戶空間進(jìn)入系統(tǒng)調(diào)用
通常情況下，我們寫的代碼都是通過封裝的C lib來調(diào)用系統(tǒng)調(diào)用的。以0.9.30版uClibc中的open為例，來追蹤一下這個(gè)封裝的函數(shù)是如何一步一步的調(diào)用系統(tǒng)調(diào)用的。在include/fcntl.h中有定義：
# define open open64
open實(shí)際上只是open64的一個(gè)別名而已。
在libc/sysdeps/linux/common/open64.c中可以看到：
extern __typeof(open64) __libc_open64;extern __typeof(open) __libc_open;
可見open64也只不過是__libc_open64的別名，而__libc_open64函數(shù)在同一個(gè)文件中定義：
libc_hidden_proto(__libc_open64)int __libc_open64 (const char *file, int oflag, ...){mode_t mode = 0;if (oflag & O_CREAT){va_list arg;va_start (arg, oflag);mode = va_arg (arg, mode_t);va_end (arg);}return __libc_open(file, oflag O_LARGEFILE, mode);}libc_hidden_def(__libc_open64)最終__libc_open64又調(diào)用了__libc_open函數(shù)，這個(gè)函數(shù)在文件libc/sysdeps/linux/common/open.c中定義：
libc_hidden_proto(__libc_open)int __libc_open(const char *file, int oflag, ...){mode_t mode = 0;if (oflag & O_CREAT) {va_list arg;va_start (arg, oflag);mode = va_arg (arg, mode_t);va_end (arg);}return __syscall_open(file, oflag, mode);}libc_hidden_def(__libc_open)__syscall_open在同一個(gè)文件中定義：
static __inline__ _syscall3(int, __syscall_open, const char *, file, int, flags, __kernel_mode_t, mode)
在文件libc/sysdeps/linux/arm/bits/syscalls.h文件中可以看到：
#undef _syscall3#define _syscall3(type,name,type1,arg1,type2,arg2,type3,arg3) type name(type1 arg1,type2 arg2,type3 arg3) { return (type) (INLINE_SYSCALL(name, 3, arg1, arg2, arg3)); }這個(gè)宏實(shí)際上完成定義一個(gè)函數(shù)的工作，這個(gè)宏的第一個(gè)參數(shù)是函數(shù)的返回值類型，第二個(gè)參數(shù)是函數(shù)名，之后的參數(shù)就如同它的參數(shù)名所表明的那樣，分別是函數(shù)的參數(shù)類型及參數(shù)名。__syscall_open實(shí)際上為：
int __syscall_open (const char * file,int flags, __kernel_mode_t mode){return (int) (INLINE_SYSCALL(__syscall_open, 3, file, flags, mode));}INLINE_SYSCALL為同一個(gè)文件中定義的宏：
#undef INLINE_SYSCALL#define INLINE_SYSCALL(name, nr, args...) ({ unsigned int _inline_sys_result = INTERNAL_SYSCALL (name, , nr, args); if (__builtin_expect (INTERNAL_SYSCALL_ERROR_P (_inline_sys_result, ), 0)) { __set_errno (INTERNAL_SYSCALL_ERRNO (_inline_sys_result, )); _inline_sys_result = (unsigned int) -1; } (int) _inline_sys_result; })#undef INTERNAL_SYSCALL#if !defined(__thumb__)#if defined(__ARM_EABI__)#define INTERNAL_SYSCALL(name, err, nr, args...) ({unsigned int __sys_result; { register int _a1 __asm__ ("r0"), _nr __asm__ ("r7"); LOAD_ARGS_##nr (args) _nr = SYS_ify(name); __asm__ __volatile__ ("swi 0x0 @ syscall " #name : "=r" (_a1) : "r" (_nr) ASM_ARGS_##nr : "memory"); __sys_result = _a1; } (int) __sys_result; })#else /* defined(__ARM_EABI__) */#define INTERNAL_SYSCALL(name, err, nr, args...) ({ unsigned int __sys_result; { register int _a1 __asm__ ("a1"); LOAD_ARGS_##nr (args) __asm__ __volatile__ ("swi %1 @ syscall " #name : "=r" (_a1) : "i" (SYS_ify(name)) ASM_ARGS_##nr : "memory"); __sys_result = _a1; } (int) __sys_result; })#endif#else /* !defined(__thumb__) *//* We cant use push/pop inside the asm because that breaksunwinding (ie. thread cancellation).*/#define INTERNAL_SYSCALL(name, err, nr, args...) ({ unsigned int __sys_result; { int _sys_buf[2]; register int _a1 __asm__ ("a1"); register int *_v3 __asm__ ("v3") = _sys_buf; *_v3 = (int) (SYS_ify(name)); LOAD_ARGS_##nr (args) __asm__ __volatile__ ("str r7, [v3, #4]n" "tldr r7, [v3]n" "tswi 0 @ syscall " #name "n" "tldr r7, [v3, #4]" : "=r" (_a1) : "r" (_v3) ASM_ARGS_##nr : "memory"); __sys_result = _a1; } (int) __sys_result; })#endif /*!defined(__thumb__)*/這里也將同文件中的LOAD_ARGS宏的定義貼出來：
#define LOAD_ARGS_0()#define ASM_ARGS_0#define LOAD_ARGS_1(a1) _a1 = (int) (a1); LOAD_ARGS_0 ()#define ASM_ARGS_1 ASM_ARGS_0, "r" (_a1)#define LOAD_ARGS_2(a1, a2) register int _a2 __asm__ ("a2") = (int) (a2); LOAD_ARGS_1 (a1)#define ASM_ARGS_2 ASM_ARGS_1, "r" (_a2)#define LOAD_ARGS_3(a1, a2, a3) register int _a3 __asm__ ("a3") = (int) (a3); LOAD_ARGS_2 (a1, a2)這項(xiàng)宏用來在相應(yīng)的寄存器中加載相應(yīng)的參數(shù)。SYS_ify宏獲得系統(tǒng)調(diào)用號(hào)
#define SYS_ify(syscall_name) (__NR_##syscall_name)
也就是__NR___syscall_open，在libc/sysdeps/linux/common/open.c中可以看到這個(gè)宏的定義：
#define __NR___syscall_open __NR_open
__NR_open在內(nèi)核代碼的頭文件中有定義。在r7寄存器中存放系統(tǒng)調(diào)用號(hào)，而參數(shù)傳遞似乎和普通的函數(shù)調(diào)用的參數(shù)傳遞也沒有什么區(qū)別。
在這個(gè)地方，得注意那個(gè)EABI，EABI是什么東西呢？ABI，Application Binary Interface，應(yīng)用二進(jìn)制接口。在較新的EABI規(guī)范中，是將系統(tǒng)調(diào)用號(hào)壓入寄存器r7中，而在老的OABI中則是執(zhí)行的swi中斷號(hào)的方式，也就是說原來的調(diào)用方式(Old ABI)是通過跟隨在swi指令中的調(diào)用號(hào)來進(jìn)行的。同時(shí)這兩種調(diào)用方式的系統(tǒng)調(diào)用號(hào)也是存在這區(qū)別的，在內(nèi)核的文件arch/arm/inclue/asm/unistd.h中可以看到：
#define __NR_OABI_SYSCALL_BASE0x900#if defined(__thumb__) defined(__ARM_EABI__)#define __NR_SYSCALL_BASE 0#else#define __NR_SYSCALL_BASE __NR_OABI_SYSCALL_BASE#endif/** This file contains the system call numbers.*/#define __NR_restart_syscall (__NR_SYSCALL_BASE+ 0)#define __NR_exit (__NR_SYSCALL_BASE+ 1)#define __NR_fork (__NR_SYSCALL_BASE+ 2)#define __NR_read (__NR_SYSCALL_BASE+ 3)#define __NR_write (__NR_SYSCALL_BASE+ 4)#define __NR_open (__NR_SYSCALL_BASE+ 5)……
接下來來看操作系統(tǒng)對系統(tǒng)調(diào)用的處理。我們回到ARM Linux的異常向量表，因?yàn)楫?dāng)執(zhí)行swi時(shí)，會(huì)從異常向量表中取例程的地址從而跳轉(zhuǎn)到相應(yīng)的處理程序中。在文件arch/arm/kernel/entry-armv.S中：
/** We group all the following data together to optimise* for CPUs with separate I & D caches.*/.align 5.LCvswi:.word vector_swi.globl __stubs_end__stubs_end:.equ stubs_offset, __vectors_start + 0x200 - __stubs_start.globl __vectors_start__vectors_start:ARM( swi SYS_ERROR0 )THUMB( svc #0 )THUMB( nop )W(b) vector_und + stubs_offsetW(ldr) pc, .LCvswi + stubs_offsetW(b) vector_pabt + stubs_offsetW(b) vector_dabt + stubs_offsetW(b) vector_addrexcptn + stubs_offsetW(b) vector_irq + stubs_offsetW(b) vector_fiq + stubs_offset.globl __vectors_end__vectors_end:
而.LCvswi在同一個(gè)文件中定義為：
.LCvswi:.word vector_swi
也就是最終會(huì)執(zhí)行例程vector_swi來完成對系統(tǒng)調(diào)用的處理，接下來我們來看下在arch/arm/kernel/entry-common.S中定義的vector_swi例程：
/*=============================================================================* SWI handler*--*//* If were optimising for StrongARM the resulting code wont run on an ARM7 and we can save a couple of instructions. --pb */#ifdef CONFIG_CPU_ARM710#define A710(code...) code.Larm710bug:ldmia sp, {r0 - lr}^ @ Get calling r0 - lrmov r0, r0add sp, sp, #S_FRAME_SIZEsubs pc, lr, #4#else#define A710(code...)#endif.align 5ENTRY(vector_swi)sub sp, sp, #S_FRAME_SIZEstmia sp, {r0 - r12} @ Calling r0 - r12ARM( add r8, sp, #S_PC )ARM( stmdb r8, {sp, lr}^ ) @ Calling sp, lrTHUMB( mov r8, sp )THUMB( store_user_sp_lr r8, r10, S_SP ) @ calling sp, lrmrs r8, spsr @ called from non-FIQ mode, so ok.str lr, [sp, #S_PC] @ Save calling PCstr r8, [sp, #S_PSR] @ Save CPSRstr r0, [sp, #S_OLD_R0] @ Save OLD_R0zero_fp/** Get the system call number.*/#if defined(CONFIG_OABI_COMPAT)/** If we have CONFIG_OABI_COMPAT then we need to look at the swi* value to determine if it is an EABI or an old ABI call.*/#ifdef CONFIG_ARM_THUMBtst r8, #PSR_T_BITmovne r10, #0 @ no thumb OABI emulationldreq r10, [lr, #-4] @ get SWI instruction#elseldr r10, [lr, #-4] @ get SWI instructionA710( and ip, r10, #0x0f @ check for SWI )A710( teq ip, #0x0f )A710( bne .Larm710bug )#endif#ifdef CONFIG_CPU_ENDIAN_BE8rev r10, r10 @ little endian instruction#endif#elif defined(CONFIG_AEABI)/** Pure EABI user space always put syscall number into scno (r7).*/A710( ldr ip, [lr, #-4] @ get SWI instruction )A710( and ip, ip, #0x0f @ check for SWI )A710( teq ip, #0x0f )A710( bne .Larm710bug )#elif defined(CONFIG_ARM_THUMB)/* Legacy ABI only, possibly thumb mode. */tst r8, #PSR_T_BIT @ this is SPSR from save_user_regsaddne scno, r7, #__NR_SYSCALL_BASE @ put OS number inldreq scno, [lr, #-4]#else/* Legacy ABI only. */ldr scno, [lr, #-4] @ get SWI instructionA710( and ip, scno, #0x0f @ check for SWI )A710( teq ip, #0x0f )A710( bne .Larm710bug )#endif#ifdef CONFIG_ALIGNMENT_TRAPldr ip, __cr_alignmentldr ip, [ip]mcr p15, 0, ip, c1, c0 @ update control register#endifenable_irq//tsk 是寄存器r9的別名，在arch/arm/kernel/entry-header.S中定義：// tsk .req r9 @current thread_info
// 獲得線程對象的基地址。
get_thread_info tsk
// tbl是r8寄存器的別名，在arch/arm/kernel/entry-header.S中定義：
// tbl .req r8 @syscall table pointer，
// 用來存放系統(tǒng)調(diào)用表的指針，系統(tǒng)調(diào)用表在后面調(diào)用
adr tbl, sys_call_table @ load syscall table pointer#if defined(CONFIG_OABI_COMPAT)/** If the swi argument is zero, this is an EABI call and we do nothing.** If this is an old ABI call, get the syscall number into scno and* get the old ABI syscall table address.*/bics r10, r10, #0xffeorne scno, r10, #__NR_OABI_SYSCALL_BASEldrne tbl, =sys_oabi_call_table#elif !defined(CONFIG_AEABI) // scno是寄存器r7的別名bic scno, scno, #0xff @ mask off SWI op-codeeor scno, scno, #__NR_SYSCALL_BASE @ check OS number#endifldr r10, [tsk, #TI_FLAGS] @ check for syscall tracingstmdb sp!, {r4, r5} @ push fifth and sixth args#ifdef CONFIG_SECCOMPtst r10, #_TIF_SECCOMPbeq 1fmov r0, scnobl __secure_computing add r0, sp, #S_R0 + S_OFF @ pointer to regsldmia r0, {r0 - r3} @ have to reload r0 - r31:#endiftst r10, #_TIF_SYSCALL_TRACE @ are we tracing syscalls?bne __sys_tracecmp scno, #NR_syscalls @ check upper syscall limitadr lr, BSYM(ret_fast_syscall) @ return addressldrcc pc, [tbl, scno, lsl #2] @ call sys_* routineadd r1, sp, #S_OFF// why也是r8寄存器的別名
2: mov why, #0@ no longer a real syscall
cmp scno, #(__ARM_NR_BASE - __NR_SYSCALL_BASE)eor r0, scno, #__NR_SYSCALL_BASE @ put OS number backbcs arm_syscall b sys_ni_syscall @ not private funcENDPROC(vector_swi)/** This is the really slow path. Were going to be doing* context switches, and waiting for our parent to respond.*/__sys_trace:mov r2, scnoadd r1, sp, #S_OFFmov r0, #0 @ trace entry [IP = 0]bl syscall_traceadr lr, BSYM(__sys_trace_return) @ return addressmov scno, r0 @ syscall number (possibly new)add r1, sp, #S_R0 + S_OFF @ pointer to regscmp scno, #NR_syscalls @ check upper syscall limitldmccia r1, {r0 - r3} @ have to reload r0 - r3ldrcc pc, [tbl, scno, lsl #2] @ call sys_* routineb 2b__sys_trace_return:str r0, [sp, #S_R0 + S_OFF]! @ save returned r0mov r2, scnomov r1, spmov r0, #1 @ trace exit [IP = 1]bl syscall_traceb ret_slow_syscall.align 5#ifdef CONFIG_ALIGNMENT_TRAP.type __cr_alignment, #object__cr_alignment:.word cr_alignment#endif.ltorg/** This is the syscall table declaration for native ABI syscalls.* With EABI a couple syscalls are obsolete and defined as sys_ni_syscall.*/#define ABI(native, compat) native#ifdef CONFIG_AEABI#define OBSOLETE(syscall) sys_ni_syscall#else#define OBSOLETE(syscall) syscall#endif.type sys_call_table, #objectENTRY(sys_call_table)#include "calls.S"#undef ABI#undef OBSOLETE上面的zero_fp是一個(gè)宏，在arch/arm/kernel/entry-header.S中定義：
.macro zero_fp#ifdef CONFIG_FRAME_POINTERmov fp, #0#endif.endm//而fp位寄存器r11。
像每一個(gè)異常處理程序一樣，要做的第一件事當(dāng)然就是保護(hù)現(xiàn)場了。緊接著是獲得系統(tǒng)調(diào)用的系統(tǒng)調(diào)用號(hào)。
然后以系統(tǒng)調(diào)用號(hào)作為索引來查找系統(tǒng)調(diào)用表，如果系統(tǒng)調(diào)用號(hào)正常的話，就會(huì)調(diào)用相應(yīng)的處理例程來處理，就是上面的那個(gè)ldrccpc, [tbl, scno, lsl #2]語句，然后通過例程ret_fast_syscall來返回。
在這個(gè)地方我們接著來討論ABI的問題?，F(xiàn)在，我們首先來看兩個(gè)宏，一個(gè)是CONFIG_OABI_COMPAT意思是說與old ABI兼容，另一個(gè)是CONFIG_AEABI意思是說指定現(xiàn)在的方式為EABI。這兩個(gè)宏可以同時(shí)配置，也可以都不配，也可以配置任何一種。我們來看一下內(nèi)核是怎么處理這一問題的。我們知道，sys_call_table在內(nèi)核中是個(gè)跳轉(zhuǎn)表，這個(gè)表中存儲(chǔ)的是一系列的函數(shù)指針，這些指針就是系統(tǒng)調(diào)用函數(shù)的指針，如(sys_open)。系統(tǒng)調(diào)用是根據(jù)一個(gè)系統(tǒng)調(diào)用號(hào)（通常就是表的索引）找到實(shí)際該調(diào)用內(nèi)核哪個(gè)函數(shù)，然后通過運(yùn)行該函數(shù)完成的。
首先，對于old ABI，內(nèi)核給出的處理是為它建立一個(gè)單獨(dú)的system call table,叫sys_oabi_call_table，這樣，兼容方式下就會(huì)有兩個(gè)system call table,以old ABI方式的系統(tǒng)調(diào)用會(huì)執(zhí)行old_syscall_table表中的系統(tǒng)調(diào)用函數(shù)，EABI方式的系統(tǒng)調(diào)用會(huì)用sys_call_table中的函數(shù)指針。
配置無外乎以下4中：
第一、兩個(gè)宏都配置行為就是上面說的那樣。
第二、只配置CONFIG_OABI_COMPAT，那么以old ABI方式調(diào)用的會(huì)用sys_oabi_call_table，以EABI方式調(diào)用的用sys_call_table，和1實(shí)質(zhì)上是相同的。只是情況1更加明確。
第三、只配置CONFIG_AEABI系統(tǒng)中不存在sys_oabi_call_table，對old ABI方式調(diào)用不兼容。只能 以EABI方式調(diào)用，用sys_call_table。
第四、兩個(gè)都沒有配置，系統(tǒng)默認(rèn)會(huì)只允許old ABI方式，但是不存在old_syscall_table，最終會(huì)通過sys_call_table完成函數(shù)調(diào)用
系統(tǒng)會(huì)根據(jù)ABI的不同而將相應(yīng)的系統(tǒng)調(diào)用表的基地址加載進(jìn)tbl寄存器，也就是r8寄存器。接下來來看系統(tǒng)調(diào)用表，如前面所說的那樣，有兩個(gè)，同樣都在文件arch/arm/kernel/entry-common.S中：
/** This is the syscall table declaration for native ABI syscalls.* With EABI a couple syscalls are obsolete and defined as sys_ni_syscall.*/#define ABI(native, compat) native#ifdef CONFIG_AEABI#define OBSOLETE(syscall) sys_ni_syscall#else#define OBSOLETE(syscall) syscall#endif.type sys_call_table, #objectENTRY(sys_call_table)#include "calls.S"#undef ABI#undef OBSOLETE
另外一個(gè)為：
/** This is the syscall table declaration for native ABI syscalls.* With EABI a couple syscalls are obsolete and defined as sys_ni_syscall.*/#define ABI(native, compat) native#ifdef CONFIG_AEABI#define OBSOLETE(syscall) sys_ni_syscall#else#define OBSOLETE(syscall) syscall#endif.type sys_call_table, #objectENTRY(sys_call_table)#include "calls.S"#undef ABI#undef OBSOLETE
這樣看來貌似兩個(gè)系統(tǒng)調(diào)用表是完全一樣的。這里預(yù)處理指令include的獨(dú)特用法也挺有意思，在系統(tǒng)調(diào)用表的內(nèi)容就是整個(gè)arch/arm/kernel/calls.S文件的內(nèi)容這個(gè)文件的內(nèi)容如下（由于太長，這里就不全部列出了）：
/** linux/arch/arm/kernel/calls.S** Copyright (C) 1995-2005 Russell King** This program is free software; you can redistribute it and/or modify* it under the terms of the GNU General Public License version 2 as* published by the Free Software Foundation.** This file is included thrice in entry-common.S*//* 0 */ CALL(sys_restart_syscall)CALL(sys_exit)CALL(sys_fork_wrapper)CALL(sys_read)CALL(sys_write)/* 5 */ CALL(sys_open)CALL(sys_close)CALL(sys_ni_syscall) /* was sys_waitpid */CALL(sys_creat)CALL(sys_link)...
這個(gè)是同樣在文件arch/arm/kernel/entry-common.S中的宏CALL()的定義：
.equ NR_syscalls,0#define CALL(x) .equ NR_syscalls,NR_syscalls+1#include "calls.S"#undef CALL#define CALL(x) .long x
最后再羅嗦一點(diǎn)，如果用sys_open來搜的話，是搜不到系統(tǒng)調(diào)用open的定義的，系統(tǒng)調(diào)用函數(shù)都是用宏來定義的，比如對于open，在文件fs/open.c文件中這樣定義：
SYSCALL_DEFINE3(open, const char __user *, filename, int, flags, int, mode){long ret;if (force_o_largefile())flags = O_LARGEFILE;ret = do_sys_open(AT_FDCWD, filename, flags, mode);/* avoid REGPARM breakage on x86: */asmlinkage_protect(3, ret, filename, flags, mode);return ret;}繼續(xù)回到vector_swi，而如果系統(tǒng)調(diào)用號(hào)不正確，則會(huì)調(diào)用arm_syscall函數(shù)來進(jìn)行處理，這個(gè)函數(shù)在文件arch/arm/kernel/traps.c中定義：
/** Handle all unrecognised system calls.* 0x9f0 - 0x9fffff are some more esoteric system calls*/#define NR(x) ((__ARM_NR_##x) - __ARM_NR_BASE)asmlinkage int arm_syscall(int no, struct pt_regs *regs){struct thread_info *thread = current_thread_info();siginfo_t info;if ((no >> 16) != (__ARM_NR_BASE>> 16))return bad_syscall(no, regs);switch (no & 0xffff) {case 0: /* branch through 0 */info.si_signo = SIGSEGV;info.si_errno = 0;info.si_code = SEGV_MAPERR;info.si_addr = NULL;arm_notify_die("branch through zero", regs, &info, 0, 0);return 0;case NR(breakpoint): /* SWI BREAK_POINT */regs->ARM_pc -= thumb_mode(regs) ? 2 : 4;ptrace_break(current, regs);return regs->ARM_r0;/** Flush a region from virtual address r0 to virtual address r1* _exclusive_. There is no alignment requirement on either address;* user space does not need to know the hardware cache layout.** r2 contains flags. It should ALWAYS be passed as ZERO until it* is defined to be something else. For now we ignore it, but may* the fires of hell burn in your belly if you break this rule. ;)** (at a later date, we may want to allow this call to not flush* various aspects of the cache. Passing 0 will guarantee that* everything necessary gets flushed to maintain consistency in* the specified region).*/case NR(cacheflush):do_cache_op(regs->ARM_r0, regs->ARM_r1, regs->ARM_r2);return 0;case NR(usr26):if (!(elf_hwcap & HWCAP_26BIT))break;regs->ARM_cpsr &= ~MODE32_BIT;return regs->ARM_r0;case NR(usr32):if (!(elf_hwcap & HWCAP_26BIT))break;regs->ARM_cpsr = MODE32_BIT;return regs->ARM_r0;case NR(set_tls):thread->tp_value = regs->ARM_r0;if (tls_emu)return 0;if (has_tls_reg) {asm ("mcr p15, 0, %0, c13, c0, 3": : "r" (regs->ARM_r0));} else {/** User space must never try to access this directly.* Expect your app to break eventually if you do so.* The user helper at 0xffff0fe0 must be used instead.* (see entry-armv.S for details)*/*((unsigned int *)0xffff0ff0) = regs->ARM_r0;}return 0;#ifdef CONFIG_NEEDS_SYSCALL_FOR_CMPXCHG/** Atomically store r1 in *r2 if *r2 is equal to r0 for user space.* Return zero in r0 if *MEM was changed or non-zero if no exchange* happened. Also set the user C flag accordingly.* If access permissions have to be fixed up then non-zero is* returned and the operation has to be re-attempted.** *NOTE*: This is a ghost syscall private to the kernel. Only the* __kuser_cmpxchg code in entry-armv.S should be aware of its* existence. Dont ever use this from user code.*/case NR(cmpxchg):for (;;) {extern void do_DataAbort(unsigned long addr, unsigned int fsr,struct pt_regs *regs);unsigned long val;unsigned long addr = regs->ARM_r2;struct mm_struct *mm = current->mm;pgd_t *pgd; pmd_t *pmd; pte_t *pte;spinlock_t *ptl;regs->ARM_cpsr &= ~PSR_C_BIT;down_read(&mm->mmap_sem);pgd = pgd_offset(mm, addr);if (!pgd_present(*pgd))goto bad_access;pmd = pmd_offset(pgd, addr);if (!pmd_present(*pmd))goto bad_access;pte = pte_offset_map_lock(mm, pmd, addr, &ptl);if (!pte_present(*pte) !pte_dirty(*pte)) {pte_unmap_unlock(pte, ptl);goto bad_access;}val = *(unsigned long *)addr;val -= regs->ARM_r0;if (val == 0) {*(unsigned long *)addr = regs->ARM_r1;regs->ARM_cpsr = PSR_C_BIT;}pte_unmap_unlock(pte, ptl);up_read(&mm->mmap_sem);return val;bad_access:up_read(&mm->mmap_sem);/* simulate a write access fault */do_DataAbort(addr, 15 + (1 << 11), regs);}#endifdefault:/* Calls 9f00xx..9f07ff are defined to return -ENOSYSif not implemented, rather than raising SIGILL. Thisway the calling program can gracefully determine whethera feature is supported. */if ((no & 0xffff) <= 0x7ff)return -ENOSYS;break;}#ifdef CONFIG_DEBUG_USER/** experience shows that these seem to indicate that* something catastrophic has happened*/if (user_debug & UDBG_SYSCALL) {printk("[%d] %s: arm syscall %dn",task_pid_nr(current), current->comm, no);dump_instr("", regs);if (user_mode(regs)) {__show_regs(regs);c_backtrace(regs->ARM_fp, processor_mode(regs));}}#endifinfo.si_signo = SIGILL;info.si_errno = 0;info.si_code = ILL_ILLTRP;info.si_addr = (void __user *)instruction_pointer(regs) -(thumb_mode(regs) ? 2 : 4);arm_notify_die("Oops - bad syscall(2)", regs, &info, no, 0);return 0;}還有那個(gè)sys_ni_syscall，這個(gè)函數(shù)在kernel/sys_ni.c中定義，它的作用似乎也僅僅是要給用戶空間返回錯(cuò)誤碼ENOSYS。
/* we cant #includehere,but tell gcc to not warn with -Wmissing-prototypes */asmlinkage long sys_ni_syscall(void);/** Non-implemented system calls get redirected here.*/asmlinkage long sys_ni_syscall(void){return -ENOSYS;}
系統(tǒng)調(diào)用號(hào)正確也好不正確也好，最終都是通過ret_fast_syscall例程來返回，同樣在arch/arm/kernel/entry-common.S文件中：
.align 5/** This is the fast syscall return path. We do as little as* possible here, and this includes saving r0 back into the SVC* stack.*/ret_fast_syscall:UNWIND(.fnstart )UNWIND(.cantunwind )disable_irq @ disable interruptsldr r1, [tsk, #TI_FLAGS]tst r1, #_TIF_WORK_MASKbne fast_work_pending#if defined(CONFIG_IRQSOFF_TRACER)asm_trace_hardirqs_on#endif/* perform architecture specific actions before user return */arch_ret_to_user r1, lrrestore_user_regs fast = 1, offset = S_OFFUNWIND(.fnend )
四.聲明系統(tǒng)調(diào)用的相關(guān)宏
linux下的系統(tǒng)調(diào)用函數(shù)定義接口：
1.SYSCALL_DEFINE1~6（include/linux/syscalls.h）
#define SYSCALL_DEFINE1(name, ...) SYSCALL_DEFINEx(1, _##name, __VA_ARGS__)#define SYSCALL_DEFINE2(name, ...) SYSCALL_DEFINEx(2, _##name, __VA_ARGS__)#define SYSCALL_DEFINE3(name, ...) SYSCALL_DEFINEx(3, _##name, __VA_ARGS__)#define SYSCALL_DEFINE4(name, ...) SYSCALL_DEFINEx(4, _##name, __VA_ARGS__)#define SYSCALL_DEFINE5(name, ...) SYSCALL_DEFINEx(5, _##name, __VA_ARGS__)#define SYSCALL_DEFINE6(name, ...) SYSCALL_DEFINEx(6, _##name, __VA_ARGS__)
2.SYSCALL_DEFINEx
#ifdef CONFIG_FTRACE_SYSCALLS#define SYSCALL_DEFINEx(x, sname, ...) static const char *types_##sname[] = { __SC_STR_TDECL##x(__VA_ARGS__) }; static const char *args_##sname[] = { __SC_STR_ADECL##x(__VA_ARGS__) }; SYSCALL_METADATA(sname, x); __SYSCALL_DEFINEx(x, sname, __VA_ARGS__)#else#define SYSCALL_DEFINEx(x, sname, ...) __SYSCALL_DEFINEx(x, sname, __VA_ARGS__)#endif3.__SYSCALL_DEFINEx
#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS#define SYSCALL_DEFINE(name) static inline long SYSC_##name#define __SYSCALL_DEFINEx(x, name, ...) asmlinkage long sys##name(__SC_DECL##x(__VA_ARGS__)); static inline long SYSC##name(__SC_DECL##x(__VA_ARGS__)); asmlinkage long SyS##name(__SC_LONG##x(__VA_ARGS__)) { __SC_TEST##x(__VA_ARGS__); return (long) SYSC##name(__SC_CAST##x(__VA_ARGS__)); } SYSCALL_ALIAS(sys##name, SyS##name); static inline long SYSC##name(__SC_DECL##x(__VA_ARGS__))#else /* CONFIG_HAVE_SYSCALL_WRAPPERS */#define SYSCALL_DEFINE(name) asmlinkage long sys_##name#define __SYSCALL_DEFINEx(x, name, ...) asmlinkage long sys##name(__SC_DECL##x(__VA_ARGS__))#endif /* CONFIG_HAVE_SYSCALL_WRAPPERS */4.__SC_開頭的宏
#define __SC_DECL1(t1, a1) t1 a1#define __SC_DECL2(t2, a2, ...) t2 a2, __SC_DECL1(__VA_ARGS__)#define __SC_DECL3(t3, a3, ...) t3 a3, __SC_DECL2(__VA_ARGS__)#define __SC_DECL4(t4, a4, ...) t4 a4, __SC_DECL3(__VA_ARGS__)#define __SC_DECL5(t5, a5, ...) t5 a5, __SC_DECL4(__VA_ARGS__)#define __SC_DECL6(t6, a6, ...) t6 a6, __SC_DECL5(__VA_ARGS__)#define __SC_LONG1(t1, a1) long a1#define __SC_LONG2(t2, a2, ...) long a2, __SC_LONG1(__VA_ARGS__)#define __SC_LONG3(t3, a3, ...) long a3, __SC_LONG2(__VA_ARGS__)#define __SC_LONG4(t4, a4, ...) long a4, __SC_LONG3(__VA_ARGS__)#define __SC_LONG5(t5, a5, ...) long a5, __SC_LONG4(__VA_ARGS__)#define __SC_LONG6(t6, a6, ...) long a6, __SC_LONG5(__VA_ARGS__)#define __SC_CAST1(t1, a1) (t1) a1#define __SC_CAST2(t2, a2, ...) (t2) a2, __SC_CAST1(__VA_ARGS__)#define __SC_CAST3(t3, a3, ...) (t3) a3, __SC_CAST2(__VA_ARGS__)#define __SC_CAST4(t4, a4, ...) (t4) a4, __SC_CAST3(__VA_ARGS__)#define __SC_CAST5(t5, a5, ...) (t5) a5, __SC_CAST4(__VA_ARGS__)#define __SC_CAST6(t6, a6, ...) (t6) a6, __SC_CAST5(__VA_ARGS__)...
5.針對SYSCALL_DEFINE1(close, unsigned int, fd)來分析一下
SYSCALL_DEFINE1(close, unsigned int, fd)根據(jù)#define SYSCALL_DEFINE1(name, ...) SYSCALL_DEFINEx(1, _##name, __VA_ARGS__)
化簡SYSCALL_DEFINEx(1, _close, __VA_ARGS__)【 ##是連接符的意思】,根據(jù)SYSCALL_DEFINEx的定義
化簡__SYSCALL_DEFINEx(1, _close, __VA_ARGS__) 根據(jù)__SYSCALL_DEFINEx的定義
#define __SYSCALL_DEFINEx(1, _close, ...) asmlinkage long sys_close(__SC_DECL1(__VA_ARGS__)); static inline long SYSC_close(__SC_DECL1(__VA_ARGS__)); asmlinkage long SyS_close(__SC_LONG1(__VA_ARGS__)) { __SC_TEST1(__VA_ARGS__); return (long) SYSC_close(__SC_CAST1(__VA_ARGS__)); } SYSCALL_ALIAS(sys_close, SyS_close); static inline long SYSC_close(__SC_DECL1(__VA_ARGS__))這里__VA_ARGS__是可變參數(shù)宏,可以認(rèn)為等于unsigned int, fd
根據(jù)__SC_宏化簡
#define __SYSCALL_DEFINEx(1, _close, ...) asmlinkage long sys_close(unsigned int fd); static inline long SYSC_close(unsigned int fd); asmlinkage long SyS_close(long fd)) { BUILD_BUG_ON(sizeof(unsigned int) > sizeof(long)) return (long) SYSC_close((unsigned int)fd); } SYSCALL_ALIAS(sys_close, SyS_close); static inline long SYSC_close(unsigned int fd)聲明了sys_close函數(shù)
定義了SyS_close函數(shù),函數(shù)體調(diào)用SYSC_close函數(shù),并返回其返回值
SYSCALL_ALIAS宏
#define SYSCALL_ALIAS(alias, name) asm ("t.globl " #alias "nt.set " #alias ", " #name)插入?yún)R編代碼 讓執(zhí)行sys_close等同于執(zhí)行SYS_close
#define SYSCALL_ALIAS(alias, name) asm ("t.globl " #alias "nt.set " #alias ", " #name)【#是預(yù)處理的意思】
BUILD_BUG_ON宏是個(gè)錯(cuò)誤判斷檢測的功能
最后一句是SYSC_close的函數(shù)定義
所以在SYSCALL_DEFINE1宏定義后面緊跟的是{}包圍起來的函數(shù)體
6.根據(jù)5的解析可推斷出
SYSCALL_DEFINE1的1代表的是sys_close的參數(shù)個(gè)數(shù)為1
同理SYSCALL_DEFINE?的/代表的是sys_name的參數(shù)為?個(gè)
7.系統(tǒng)調(diào)用函數(shù)的定義用SYSCALL_DEFINE宏修飾
系統(tǒng)調(diào)用函數(shù)的外部聲明在include/linux/Syscalls.h頭文件中
5 添加新的系統(tǒng)調(diào)用
第一、打開arch/arm/kernel/calls.S，在最后添加系統(tǒng)調(diào)用的函數(shù)原型的指針，例如：
CALL(sys_set_senda)
補(bǔ)充說明一點(diǎn)關(guān)于NR_syscalls的東西，這個(gè)常量表示系統(tǒng)調(diào)用的總的個(gè)數(shù)，在較新版本的內(nèi)核中，文件arch/arm/kernel/entry-common.S中可以找到：
.equ NR_syscalls,0#define CALL(x) .equ NR_syscalls,NR_syscalls+1#include "calls.S"#undef CALL#define CALL(x) .long x
相當(dāng)?shù)那擅?，不是嗎？在系統(tǒng)調(diào)用表中每添加一個(gè)系統(tǒng)調(diào)用，NR_syscalls就自動(dòng)增加一。在這個(gè)地方先求出NR_syscalls，然后重新定義CALL(x)宏，這樣也可以不影響文件后面系統(tǒng)調(diào)用表的建立。
第二、打開include/asm-arm/unistd.h,添加系統(tǒng)調(diào)用號(hào)的宏，感覺這步可以省略，因?yàn)檫@個(gè)地方定義的系統(tǒng)調(diào)用號(hào)主要是個(gè)C庫，比如uClibc、Glibc用的。例如:
#define __NR_plan_set_senda (__NR_SYSCALL_BASE+365)
為了向后兼容，系統(tǒng)調(diào)用只能增加而不能減少，這里的編號(hào)添加時(shí)，也必須按順序來。否則會(huì)導(dǎo)致核心運(yùn)行錯(cuò)誤。
第三，實(shí)例化該系統(tǒng)調(diào)用，即編寫新添加系統(tǒng)調(diào)用的實(shí)現(xiàn)例如：
SYSCALL_DEFINE1(set_senda, int，iset){if(iset)UART_PUT_CR(&at91_port[2],AT91C_US_SENDA);elseUART_PUT_CR(&at91_port[2],AT91C_US_RSTSTA);return 0;}第四、打開include/linux/syscalls.h添加函數(shù)聲明
asmlinkage long sys_set_senda(int iset);
第五、在應(yīng)用程序中調(diào)用該系統(tǒng)調(diào)用，可以參考uClibc的實(shí)現(xiàn)。
第六、結(jié)束。
評論