ARM linux系统调用的实现原理

xiaoxiao2026-08-19  27

<script>function StorePage(){d=document;t=d.selection?(d.selection.type!='None'?d.selection.createRange().text:''):(d.getSelection?d.getSelection():'');void(keyit=window.open('http://www.365key.com/storeit.aspx?t='+escape(d.title)+'&u='+escape(d.location.href)+'&c='+escape(t),'keyit','scrollbars=no,width=475,height=575,left=75,top=20,status=no,resizable=yes'));keyit.focus();}</script>

作者:刘洪涛,华清远见嵌入式培训中心讲师。

大家都知道linux的应用程序要想访问内核必须使用系统调用从而实现从usr模式转到svc模式。下面咱们看看它的实现过程。

系统调用是os操作系统提供的服务,用户程序通过各种系统调用,来引用内核提供的各种服务,系统调用的执行让用户程序陷入内核,该陷入动作由swi软中断完成。

at91rm9200处理器对应的linux2.4.19内核系统调用对应的软中断定义如下:#if defined(__thumb__) //thumb模式#define __syscall(name) \"push {r7}\n\t" \"mov r7, #" __sys1(__NR_##name) "\n\t" \"swi 0\n\t" \"pop {r7}"#else //arm模式#define __syscall(name) "swi\t" __sys1(__NR_##name) "\n\t"#endif#define __sys2(x) #x#define __sys1(x) __sys2(x)#define __NR_SYSCALL_BASE 0x900000 //此为OS_NUMBER << 20运算值#define __NR_open (__NR_SYSCALL_BASE+ 5) //0x900005

举一个例子来说:open系统调用,库函数最终会调用__syscall(open),宏展开之后为swi #__NR_open,即,swi #0x900005触发中断,中断号0x900005存放在[lr,#-4]地址中,处理器跳转到arch/arm/kernel/entry-common.S中vector_swi读取[lr,#-4]地址中的中断号,之后查询arch/arm/kernel/entry-common.S中的sys_call_table系统调用表,该表内容在arch/arm/kernel/calls.S中定义,__NR_open在表中对应的顺序号为__syscall_start:....long SYMBOL_NAME(sys_open) //第5个...

将sys_call_table[5]中内容传给pc,系统进入sys_open函数,处理实质的open动作

注:用到的一些函数数据所在文件,如下所示

arch/arm/kernel/calls.S声明了系统调用函数

include/asm-arm/unistd.h定义了系统调用的调用号规则vector_swi定义在arch/arm/kernel/entry-common.Svector_IRQ定义在arch/arm/kernel/entry-armv.Svector_FIQ定义在arch/arm/kernel/entry-armv.Sarch/arm/kernel/entry-common.S中对sys_call_table进行了定义:.type sys_call_table, #objectENTRY(sys_call_table)#include "calls.S" //将calls.S中的内容顺序链接到这里源程序:ENTRY(vector_swi)save_user_regszero_fpget_scno //将[lr,#-4]中的中断号转储到scno(r7)arm710_bug_check scno, ip#ifdef CONFIG_ALIGNMENT_TRAPldr ip, __cr_alignmentldr ip, [ip]mcr p15, 0, ip, c1, c0 @ update control register#endifenable_irq ipstr r4, [sp, #-S_OFF]! @ push fifth argget_current_task tskldr ip, [tsk, #TSK_PTRACE] @ check for syscall tracingbic scno, scno, #0xff000000 @ mask off SWI op-code//#define OS_NUMBER 9[entry-header.S]//所以对于上面示例中open系统调用号scno=0x900005//eor scno,scno,#0x900000//之后scno=0x05eor scno, scno, #OS_NUMBER << 20 @ check OS number//sys_call_table项为calls.S的内容adr tbl, sys_call_table @ load syscall table pointertst ip, #PT_TRACESYS @ are we tracing syscalls?bne __sys_traceadrsvc al, lr, ret_fast_syscall @ return addresscmp scno, #NR_syscalls @ check upper syscall limit//执行sys_open函数ldrcc pc, [tbl, scno, lsl #2] @ call sys_* routineadd r1, sp, #S_OFF2: mov why, #0 @ no longer a real syscallcmp scno, #ARMSWI_OFFSETeor r0, scno, #OS_NUMBER << 20 @ put OS number backbcs SYMBOL_NAME(arm_syscall) b SYMBOL_NAME(sys_ni_syscall) @ not private func/** This is the really slow path. We're going to be doing* context switches, and waiting for our parent to respond.*/__sys_trace:add r1, sp, #S_OFFmov r0, #0 @ trace entry [IP = 0]bl SYMBOL_NAME(syscall_trace)/*//2007-07-01 gliethttp [entry-header.S]//Like adr, but force SVC mode (if required).macro adrsvc, cond, reg, labeladr\cond \reg, \label.endm//对应反汇编://add lr, pc, #16 ; lr = __sys_trace_return*/adrsvc al, lr, __sys_trace_return @ return addressadd 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 r1, spmov r0, #1 @ trace exit [IP = 1]bl SYMBOL_NAME(syscall_trace)b ret_disable_irq.align 5#ifdef CONFIG_ALIGNMENT_TRAP.type __cr_alignment, #object__cr_alignment:.word SYMBOL_NAME(cr_alignment)#endif.type sys_call_table, #objectENTRY(sys_call_table)#include "calls.S"

相关资源:敏捷开发V1.0.pptx
转载请注明原文地址: https://www.6miu.com/read-5051495.html

最新回复(0)