计算机组成与结构：第二部分-程序和数据的机器级表达 05-程序的机器级表示：基本

第3章程序的机器级表示

——

基本

计算机组成与结构

2016年4月主讲教师Today:MachineProgrammingI:BasicsHistoryofIntelprocessorsandarchitecturesC,assembly,machinecodeAssemblyBasics:Registers,operands,moveaddressingmode,addresscomputation*ArithmeticoperationsIntrotox86-64Intelx86Evolution:MilestonesName Date Transistors MHz8086 1978 29K 5-10First16-bitprocessor.BasisforIBMPC&DOS1MBaddressspace386 1985 275K 16-33First32bitprocessor,referredtoasIA32Added“flataddressing”,CapableofrunningUnixPentium4F 2004 125M 2800-3800First64-bitprocessor,referredtoasx86-64Core22006291M1060-3500Firstmulti-coreIntelprocessorCorei7 2008 731M 1700-3900Fourcores(oursharkmachines)Haswell20131.4B1900-3700On-chipGPUOurCoverageIA32Thetraditionalx86x86-64/EM64TTheemergingstandardPresentationBookpresentsIA32inSections3.1—3.12Coversx86-64in3.13WewillcoverbothsimultaneouslySomelabswillbebasedonx86-64,othersonIA32Today:MachineProgrammingI:BasicsHistoryofIntelprocessorsandarchitecturesC,assembly,machinecodeAssemblyBasics:Registers,operands,moveaddressingmode,addresscomputationArithmeticoperationsIntrotox86-64CPUAssemblyProgrammer’sViewProgrammer-VisibleStatePC:ProgramcounterAddressofnextinstructionCalled“EIP”(IA32)or“RIP”(x86-64)RegisterfileHeavilyusedprogramdataConditioncodesStorestatusinformationaboutmostrecentarithmeticoperationUsedforconditionalbranchingPCRegistersMemoryCodeDataStackAddressesDataInstructionsConditionCodesMemoryByteaddressablearrayCodeanduserdataStacktosupportprocedurestexttextbinarybinaryCompiler(gcc-S)Assembler(gccoras)Linker(gccorld)Cprogram(p1.cp2.c)Asmprogram(p1.sp2.s)Objectprogram(p1.op2.o)Executableprogram(p)Staticlibraries(.a)TurningCintoObjectCodeCodeinfilesp1.cp2.cCompilewithcommand:gcc–O1p1.cp2.c-opUsebasicoptimizations(-O1)PutresultingbinaryinfilepGCC的基本用法gcc[options][filenames]其中options就是编译器所需要的参数，filenames给出相关的文件名称-c，只编译，不链接为可执行。-o，确定输出文件的名称。如果不给出这个选项，gcc给出预设的可执行文件a.out。-s，产生汇编代码。-O1，对程序进行第一级优化编译、链接。-O2，比-O1更好的优化编译、链接，当然整个编译、链接过程会更慢。等等参考：百度百科GCC阅读：课程网站推荐阅读资料“GCC”专题CompilingIntoAssemblyCCodeintsum(intx,inty){intt=x+y;returnt;}GeneratedIA32Assemblysum: pushl%ebp movl%esp,%ebp movl12(%ebp),%eax addl8(%ebp),%eax popl%ebp retObtainwithcommandgcc–O1-Scode.cProducesfilecode.sSomecompilersuseinstruction“leave”MachineInstructionExampleCCodeAddtwosignedintegersAssemblyAdd24-byteintegers“Long”wordsinGCCparlanceSameinstructionwhethersignedorunsignedOperands:x: Register %eaxy: Memory M[%ebp+8]t: Register %eaxReturnfunctionvaluein%eaxObjectCode3-byteinstructionStoredataddress0x80483caintt=x+y;Addl8(%ebp),%eax0x80483ca:034508Similartoexpression:x+=yMoreprecisely:inteax;int*ebp;eax+=ebp[2]Codeforsum0x401040<sum>:0x550x890xe50x8b0x450x0c0x030x450x080x5d0xc3ObjectCodeAssemblerTranslates.sinto.oBinaryencodingofeachinstructionNearly-completeimageofexecutablecodeMissinglinkagesbetweencodeindifferentfilesLinkerResolvesreferencesbetweenfilesCombineswithstaticrun-timelibrariesE.g.,codeformalloc,printfSomelibrariesaredynamicallylinkedLinkingoccurswhenprogrambeginsexecutionTotalof11bytesEachinstruction1,2,or3bytesStartsataddress0x0401040Obtainwithcommandgcc–O1-ccode.cProducesfilecode.o目标代码code.oDisassembledDisassemblingObjectCodeDisassemblerobjdump–dcode.oUsefultoolforexaminingobjectcodeAnalyzesbitpatternofseriesofinstructionsProducesapproximaterenditionofassemblycodeCanberunoneithera.out(completeexecutable)or.ofile0x401040<sum>:0x401040:55push%ebp0x401041:89e5mov%esp,%ebp0x401043:8b450cmov0xc(%ebp),%eax0x401046:034508add0x8(%ebp),%eax0x401049:5dpop%ebp0x40104a:c3retDisassembledDumpofassemblercodeforfunctionsum:0x080483c4<sum+0>:push%ebp0x080483c5<sum+1>:mov%esp,%ebp0x080483c7<sum+3>:mov0xc(%ebp),%eax0x080483ca<sum+6>:add0x8(%ebp),%eax0x080483cd<sum+9>:pop%ebp0x080483ce<sum+10>:retAlternateDisassemblyWithingdbDebuggergdbPdisassemblesumDisassembleprocedurex/11xbsumExaminethe11bytesstartingatsumObject0x401040:0x550x890xe50x8b0x450x0c0x030x450x080x5d0xc3GCC使用举例将C源程序文件test.c，用GCC直接生成最终的可执行文件testgcc-O1test1.ctest2.c-otest选项-O1表示一级优化，-O2为二级优化，选项-o指出输出文件名目标文件可用“objdump-dtest.o”反汇编为汇编语言程序add:pushl %ebpmovl %esp,%ebpsubl $16,%espmovl 12(%ebp),%eaxmovl 8(%ebp),%edxleal (%edx,%eax),%eaxmovl %eax,-4(%ebp)movl -4(%ebp),%eaxleaveret00000000<add>:0:55 push%ebp1:89e5 mov%esp,%ebp3:83ec10sub$0x10,%esp6:8b450cmov0xc(%ebp),%eax9:8b5508mov0x8(%ebp),%edxc:8d0402lea(%edx,%eax,1),%eaxf:8945fcmov%eax,-0x4(%ebp)12:8b45fcmov-0x4(%ebp),%eax15:c9leave16:c3retgcc-Etest.c-otest.igcc-Stest.i-otest.s

gcc–Stest.c–otest.s

test.s位移量机器指令汇编指令编译得到的与反汇编得到的汇编指令形式稍有差异

两种目标文件“objdump-dtest”结果00000000<add>:0:55 push%ebp1:89e5 mov%esp,%ebp3:83ec10sub$0x10,%esp6:8b450cmov0xc(%ebp),%eax9:8b5508mov0x8(%ebp),%edxc:8d0402lea(%edx,%eax,1),%eaxf:8945fcmov%eax,-0x4(%ebp)12:8b45fcmov-0x4(%ebp),%eax15:c9leave16:c3rettest.o中的代码从地址0开始，test中的代码从80483d4开始！080483d4<add>:80483d4:55push...80483d5:89e5…80483d7:83ec10…80483da:8b450c…80483dd:8b5508…80483e0:8d0402…80483e3:8945fc…80483e6:8b45fc…80483e9:c9…80483ea:c3

ret

“objdump-dtest.o”结果test.o：可重定位目标文件test：可执行目标文件Today:MachineProgrammingI:BasicsHistoryofIntelprocessorsandarchitecturesC,assembly,machinecodeAssemblyBasics:Registers,operands,moveaddressingmode,addresscomputation*ArithmeticoperationsIntrotox86-64AssemblyCharacteristics:DataTypes“Integer”dataof1,2,or4bytesDatavaluesAddresses(untypedpointers)Floatingpointdataof4,8,or10bytesNoaggregatetypessuchasarraysorstructuresJustcontiguouslyallocatedbytesinmemoryAssemblyCharacteristics:OperationsPerformarithmeticfunctiononregisterormemorydataTransferdatabetweenmemoryandregisterLoaddatafrommemoryintoregisterStoreregisterdataintomemoryTransfercontrolUnconditionaljumpsto/fromproceduresConditionalbranchesIntegerRegisters(IA32)%eax%ecx%edx%ebx%esi%edi%esp%ebp%ax%cx%dx%bx%si%di%sp%bp%ah%ch%dh%bh%al%cl%dl%bl16-bitvirtualregisters(backwardscompatibility)generalpurposeaccumulatecounterdatabasesourceindexdestinationindexstackpointerbasepointerOrigin(mostlyobsolete)MovingData:IA32MovingDatamovlSource,Dest:OperandTypesImmediate:

ConstantintegerdataExample:$0x400,$-533LikeCconstant,butprefixedwith‘$’Encodedwith1,2,or4bytesRegister:Oneof8integerregistersExample:%eax,%edxBut%espand%ebpreservedforspecialuseOthershavespecialusesforparticularinstructionsMemory:

4consecutivebytesofmemoryataddressgivenbyregisterSimplestexample:(%eax)Variousother“addressmodes”%eax%ecx%edx%ebx%esi%edi%esp%ebpmovlOperandCombinationsCannotdomemory-memorytransferwithasingleinstructionmovlImmRegMemRegMemRegMemRegSourceDestCAnalogmovl$0x4,%eaxtemp=0x4;movl$-147,(%eax)*p=-147;movl%eax,%edxtemp2=temp1;movl%eax,(%edx)*p=temp;movl(%eax),%edxtemp=*p;Src,DestToday:MachineProgrammingI:BasicsHistoryofIntelprocessorsandarchitecturesC,assembly,machinecodeAssemblyBasics:Registers,operands,moveaddressingmode,addresscomputation*ArithmeticoperationsIntrotox86-64SimpleMemoryAddressingModesNormal (R) Mem[Reg[R]]RegisterRspecifiesmemoryaddress

movl(%ecx),%eaxDisplacement D(R) Mem[Reg[R]+D]RegisterRspecifiesstartofmemoryregionConstantdisplacementDspecifiesoffset

movl8(%ebp),%edxUsingSimpleAddressingModesvoidswap(int*xp,int*yp){intt0=*xp;intt1=*yp;*xp=t1;*yp=t0;}BodySetUpFinishswap:pushl%ebpmovl%esp,%ebppushl%ebxmovl8(%ebp),%edxmovl12(%ebp),%ecxmovl(%edx),%ebxmovl(%ecx),%eaxmovl%eax,(%edx)movl%ebx,(%ecx)popl%ebxpopl%ebpretUsingSimpleAddressingModesvoidswap(int*xp,int*yp){intt0=*xp;intt1=*yp;*xp=t1;*yp=t0;}swap:pushl%ebpmovl%esp,%ebppushl%ebxmovl 8(%ebp),%edxmovl 12(%ebp),%ecxmovl (%edx),%ebxmovl (%ecx),%eaxmovl %eax,(%edx)movl %ebx,(%ecx)popl %ebxpopl %ebpretBodySetUpFinish插播：函数调用——IA32栈结构示例C程序：intsum(intx,inty){ intt=x+y; returnt;}intmain(){ returnsum(1,3);}UnderstandingSwapvoidswap(int*xp,int*yp){intt0=*xp;intt1=*yp;*xp=t1;*yp=t0;}Stack(inmemory)Register Value%edxxp%ecxyp%ebxt0%eaxt1ypxpRtnadrOld%ebp%ebp04812Offset•••Old%ebx-4%espmovl8(%ebp),%edx#edx=xpmovl12(%ebp),%ecx#ecx=ypmovl(%edx),%ebx#ebx=*xp(t0)movl(%ecx),%eax#eax=*yp(t1)movl%eax,(%edx)#*xp=t1movl%ebx,(%ecx)#*yp=t0UnderstandingSwap0x1200x124Rtnadr%ebp04812Offset-4123456Address0x1240x1200x11c0x1180x1140x1100x10c0x1080x1040x100ypxp%eax%edx%ecx%ebx%esi%edi%esp%ebp0x104movl8(%ebp),%edx #edx=xpmovl12(%ebp),%ecx #ecx=ypmovl(%edx),%ebx #ebx=*xp(t0)movl(%ecx),%eax #eax=*yp(t1)movl%eax,(%edx) #*xp=t1movl%ebx,(%ecx) #*yp=t0UnderstandingSwap0x1200x124Rtnadr%ebp04812Offset-4123456Address0x1240x1200x11c0x1180x1140x1100x10c0x1080x1040x100ypxp%eax%edx%ecx%ebx%esi%edi%esp%ebp0x1240x1040x120movl8(%ebp),%edx #edx=xpmovl12(%ebp),%ecx #ecx=ypmovl(%edx),%ebx #ebx=*xp(t0)movl(%ecx),%eax #eax=*yp(t1)movl%eax,(%edx) #*xp=t1movl%ebx,(%ecx) #*yp=t0UnderstandingSwap0x1200x124Rtnadr%ebp04812Offset-4123456Address0x1240x1200x11c0x1180x1140x1100x10c0x1080x1040x100ypxp%eax%edx%ecx%ebx%esi%edi%esp%ebp0x1200x1040x1240x124movl8(%ebp),%edx #edx=xpmovl12(%ebp),%ecx #ecx=ypmovl(%edx),%ebx #ebx=*xp(t0)movl(%ecx),%eax #eax=*yp(t1)movl%eax,(%edx) #*xp=t1movl%ebx,(%ecx) #*yp=t0456UnderstandingSwap0x1200x124Rtnadr%ebp04812Offset-4123456Address0x1240x1200x11c0x1180x1140x1100x10c0x1080x1040x100ypxp%eax%edx%ecx%ebx%esi%edi%esp%ebp0x1240x1201230x104movl8(%ebp),%edx#edx=xpmovl12(%ebp),%ecx #ecx=ypmovl(%edx),%ebx #ebx=*xp(t0)movl(%ecx),%eax #eax=*yp(t1)movl%eax,(%edx) #*xp=t1movl%ebx,(%ecx) #*yp=t0UnderstandingSwap0x1200x124Rtnadr%ebp04812Offset-4123456Address0x1240x1200x11c0x1180x1140x1100x10c0x1080x1040x100ypxp%eax%edx%ecx%ebx%esi%edi%esp%ebp4560x1240x1200x104123123movl8(%ebp),%edx #edx=xpmovl12(%ebp),%ecx #ecx=ypmovl(%edx),%ebx #ebx=*xp(t0)movl(%ecx),%eax #eax=*yp(t1)movl%eax,(%edx) #*xp=t1movl%ebx,(%ecx) #*yp=t0456456UnderstandingSwap0x1200x124Rtnadr%ebp04812Offset-4Address0x1240x1200x11c0x1180x1140x1100x10c0x1080x1040x100ypxp%eax%edx%ecx%ebx%esi%edi%esp%ebp4564560x1240x1201230x104123movl8(%ebp),%edx #edx=xpmovl12(%ebp),%ecx #ecx=ypmovl(%edx),%ebx #ebx=*xp(t0)movl(%ecx),%eax #eax=*yp(t1)movl%eax,(%edx) #*xp=t1movl%ebx,(%ecx) #*yp=t0UnderstandingSwap0x1200x124Rtnadr%ebp04812Offset-4456123Address0x1240x1200x11c0x1180x1140x1100x10c0x1080x1040x100ypxp%eax%edx%ecx%ebx%esi%edi%esp%ebp4560x1240x1200x104123movl8(%ebp),%edx #edx=xpmovl12(%ebp),%ecx #ecx=ypmovl(%edx),%ebx #ebx=*xp(t0)movl(%ecx),%eax #eax=*yp(t1)movl%eax,(%edx) #*xp=t1movl%ebx,(%ecx) #*yp=t0CompleteMemoryAddressingModesMostGeneralFormD(Rb,Ri,S) Mem[Reg[Rb]+S*Reg[Ri]+D]D: Constant“displacement”1,2,or4bytesRb:Baseregister:Anyof8integerregistersRi: Indexregister:Any,exceptfor%espUnlikelyyou’duse%ebp,eitherS: Scale:1,2,4,or8(whythesenumbers?)SpecialCases(Rb,Ri) Mem[Reg[Rb]+Reg[Ri]]D(Rb,Ri) Mem[Reg[Rb]+Reg[Ri]+D](Rb,Ri,S) Mem[Reg[Rb]+S*Reg[Ri]]AddressComputationExamplesExpressionAddressComputationAddress0x8(%edx)0xf000+0x80xf008(%edx,%ecx)0xf000+0x1000xf100(%edx,%ecx,4)0xf000+4*0x1000xf4000x80(,%edx,2)2*0xf000+0x800x1e080%edx0xf000%ecx0x0100ExpressionAddressComputationAddress0x8(%edx)0xf000+0x80xf008(%edx,%ecx)0xf000+0x1000xf100(%edx,%ecx,4)0xf000+4*0x1000xf4000x80(,%edx,2)2*0xf000+0x800x1e080AddressComputationInstructionLealSrc,DestSrcisaddressmodeexpressionSetDesttoaddressdenotedbyexpressionUsesComputingaddresseswithoutamemoryreferenceE.g.,translationofp=&x[i];Computingarithmeticexpressionsoftheformx+k*yk=1,2,4,or8Exampleintmul12(intx){returnx*12;}leal(%eax,%eax,2),%eax;t<-x+x*2sall$2,%eax;returnt<<2ConvertedtoASMbycompiler:Today:MachineProgrammingI:BasicsHistoryofIntelprocessorsandarchitecturesC,assembly,machinecodeAssemblyBasics:Registers,operands,moveaddressingmode,addresscomputation*ArithmeticoperationsIntrotox86-64SomeArithmeticOperationsTwoOperandInstructions:FormatComputationaddl

Src,Dest ;Dest=Dest+Srcsubl

Src,Dest ;Dest=DestSrcimull

Src,Dest ;Dest=Dest*Srcsall

Src,Dest ;Dest=Dest<<Src(Alsocalledshll)sarl

Src,Dest ;Dest=Dest>>Src(Arithmetic)shrl

Src,Dest ;Dest=Dest>>Src(Logical)xorl

Src,Dest ;Dest=Dest^Srcandl

Src,Dest ;Dest=Dest&Srcorl

Src,Dest ;Dest=Dest|SrcWatchoutforargumentorder!Nodistinctionbetweensignedandunsignedint(why?)SomeArithmeticOperationsOneOperandInstructionsincl

Dest Dest=Dest+1decl

Dest Dest=Dest1negl

Dest Dest=Destnotl

Dest Dest=~DestSeebookformoreinstructionsArithmeticExpressionExampleintarith(intx,inty,intz){intt1=x+y;intt2=z+t1;intt3=x+4;intt4=y*48;intt5=t3+t4;intrval=t2*t5;returnrval;}arith:pushl %ebpmovl %esp,%ebpmovl 8(%ebp),%ecxmovl 12(%ebp),%edxleal (%edx,%edx,2),%eaxsall $4,%eaxleal 4(%ecx,%eax),%eaxaddl %ecx,%edxaddl 16(%ebp),%edximull %edx,%eaxpopl %ebpretBodySetUpFinish•••16z12y8x4Rtn

Addr0Old%ebpUnderstandingarithmovl 8(%ebp),%ecxmovl 12(%ebp),%edxleal (%edx,%edx,2),%eaxsall $4,%eaxleal 4(%ecx,%eax),%eaxaddl %ecx,%edxaddl 16(%ebp),%edximull %edx,%eax%ebpOffsetintarith(intx,inty,intz){intt1=x+y;intt2=z+t1;intt3=x+4;intt4=y*48;intt5=t3+t4;intrval=t2*t5;returnrval;}•••16z12y8x4Rtn

Addr0Old%ebpUnderstandingarith%ebpOffsetStackintarith(intx,inty,intz){intt1=x+y;intt2=z+t1;intt3=x+4;intt4=y*48;intt5=t3+t4;intrval=t2*t5;returnrval;}movl 8(%ebp),%ecx #ecx=xmovl 12(%ebp),%edx #edx=yleal (%edx,%edx,2),%eax #eax=y*3sall $4,%eax #eax*=16(t4)leal 4(%ecx,%eax),%eax #eax=t4+x+4(t5)addl %ecx,%edx #edx=x+y(t1)addl 16(%ebp),%edx #edx+=z(t2)imull %edx,%eax #eax=t2*t5(rval)ObservationsaboutarithInstructionsindifferentorderfromCcodeSomeexpressionsrequiremultipleinstructionsSomeinstructionscovermultipleexpressionsGetexactsamecodewhencompile:(x+y+z)*(x+4+48*y)movl 8(%ebp),%ecx #ecx=xmovl 12(%ebp),%edx #edx=yleal (%edx,%edx,2),%eax #eax=y*3sall $4,%eax #eax*=16(t4)leal 4(%ecx,%eax),%eax #eax=t4+x+4(t5)addl %ecx,%edx #edx=x+y(t1)addl 16(%ebp),%edx #edx+=z(t2)imull %edx,%eax #eax=t2*t5(rval)intarith(intx,inty,intz){intt1=x+y;intt2=z+t1;intt3=x+4;intt4=y*48;intt5=t3+t4;intrval=t2*t5;returnrval;}SpecialArithmeticOperations指令效果描述imullS%edx:%eax←S×%eax有符号全64位乘法mullS%edx

:%eax←S×%eax无符号全64位乘法cltd%edx:%eax←SignExtend(%eax)转为四字idivlS%edx←%edx:%eaxmodS;%eax←%edx:%eax÷S有符号除法divlS%edx←%edx

:%eaxmodS;%eax←%edx:%eax÷S无符号除法Today:MachineProgrammingI:BasicsHistoryofIntelprocessorsandarchitecturesC,assembly,machinecodeAssemblyBasics:Regi