Vm-dev March 2020

vm-dev@lists.squeakfoundation.org

31 participants
136 discussions

[OpenSmalltalk/opensmalltalk-vm] Update OpenSSL download to v1.1.1d as v1.0.2q EOLed 1 Jan 2020. (#474)
by Pierce Ng 22 Apr '20

22 Apr '20

Fixes #455 You can view, comment on, or merge this pull request online at: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/474 -- Commit Summary -- * Update OpenSSL download to v1.1.1d as v1.0.2q EOLed 1 Jan 2020. -- File Changes -- M third-party/openssl.spec (14) -- Patch Links -- https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/474.patch https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/474.diff -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/474

3 3

pthread_setschedparam failed: Not owner
by stes 01 Apr '20

01 Apr '20

On Solaris when running the bash-4.4$ bin/squeak Squeak5.3-19431-32bit.image *pthread_setschedparam failed: Not owner* (the Squeak image starts up despite the warning/error, it seems to warn that "for best operation, this thread should run at a higher priority, however the VM was unable to change the priority.") I'm not really sure what the right Solaris way to fix this is, but by default the shell has for me bash-4.4$ priocntl -d $$ INTERACTIVE CLASS PROCESSES: PID[/LWP] IAUPRILIM IAUPRI IAMODE 3246 0 0 1 priority limit 0 and priority set to 0 I can raise this with "priocntl -s -m 10 -p 1 $$" to set the limit to 10 and the priority of the shell to 1 (not sure whether that is good idea) in any case in that case bash-4.4$ priocntl -d $$ INTERACTIVE CLASS PROCESSES: PID[/LWP] IAUPRILIM IAUPRI IAMODE 3246 10 1 1 the limit is then 10 when starting from that shell the squeak VM it does *not print the warning*, so perhaps it can succesfully issue the pthread_setschedparam() call. the LWP (light weight process) with "ps" (report process status) shows a difference indeed # ps -eLo class,pri,lwp,pid,comm | grep squeak IA 40 1 3256 bin/../lib/squeak/5.0-202003240214-solaris/squeak IA 51 2 3256 bin/../lib/squeak/5.0-202003240214-solaris/squeak while if the error/warning is printed the "pri' (priority) is the same for the 2 LWP (threads?) # ps -eLo class,pri,lwp,pid,comm | grep squeak IA 49 1 3306 bin/../lib/squeak/5.0-202003240214-solaris/squeak IA 49 2 3306 bin/../lib/squeak/5.0-202003240214-solaris/squeak -- Sent from: http://forum.world.st/Squeak-VM-f104410.html

3 15

Coming out
by Bert Freudenberg 31 Mar '20

31 Mar '20

Hi all, Today is Transgender Day of Visibility, and I am coming out publicly as a transgender woman. My name is Vanessa, and my pronouns are she / her. Many of you have known me for a long time as Bert Freudenberg. I’m still me, just more so. I will resubscribe to this list with my new email address, codefrau(a)gmail.com. Cheers! - Vanessa - -- -- Dr. Bert Freudenberg 7275 Franklin Avenue #210 Los Angeles CA 90046 +1 (818) 482-3991

11 10

squeak vm version numbering
by stes 31 Mar '20

31 Mar '20

There seems to have been in the past a "change" in version numbering. That's quite common of course, version numbering schemes always change As far as I currently understand there is the 4.x series of Squeak images and vm's. # squeak -version 4.16.7-3775 #1 XShm Tue Mar 31 10:36:25 CEST 2020 cc SunOS gecko 5.11 11.3 i86pc i386 i86pc plugin path: /usr/lib/squeak/4.16.7-3775 [default: /usr/lib/squeak/4.16.7-3775/] Note that this is X.Y.Z version 3 numbers : 4.16.7 On the other hand with the "Spur" recent 64 bit images, this is now replaced by GIT commit This is in the V5 Squeak with 5.x Squeak images in 64bit. # squeak -version 5.0-202003291930-solaris Mon Mar 30 17:47:52 CEST 2020 cc [Production Spur 64-bit VM] CoInterpreter VMMaker.oscog-eem.2731 uuid: 3c8dda9e-4706-4c3d-8c58-a284c47f5705 Mar 30 2020 StackToRegisterMappingCogit VMMaker.oscog-eem.2731 uuid: 3c8dda9e-4706-4c3d-8c58-a284c47f5705 Mar 30 2020 Date: Sun Mar 29 21:30:18 2020 CommitHash: 22060579c SunOS gecko 5.11 11.3 i86pc i386 i86pc plugin path: /usr/bin/../lib/squeak/5.0-202003291930-solaris [default: /usr/lib/squeak/5.0-202003291930-solaris/] Note that the plugin path and version here is X.Y 2 numbers 5.0. This is understandable as GIT has more concepts like "CommitHash" and so on , rather than 'old style' version numbers. However it could still be interesting I think to keep the tradition of the 3 number version, this is with the updateSCCSversion script, I believe. Is there a plan to keep the VM version number to keep matching the Squeak image versions ? For the squeak image it is 6.0 alpha or 5.3 I see. But they also have 5.3 rc1 , rc2 etc. That could correspond to 5.3.1 5.3.2 etc. -- Sent from: http://forum.world.st/Squeak-VM-f104410.html

1 1

Re: [Vm-dev] [pharo-project/pharo] Windows image has system time delay of 40 seconds (#2792)
by Eliot Miranda 31 Mar '20

31 Mar '20

Hi Sabine, I proposed a solution, and included an example implementation and asked for review here in February of 2016. Re: Time millisecondClockValue (was: The Trunk: Morphic-mt.1080.mcz) <http://forum.world.st/Time-millisecondClockValue-was-The-Trunk-Morphic-mt-1…> But I received no feedback whatsoever. I'm not particularly excited by doing all the heavy lifting while the unethical pass off my work as theirs, and contribute nothing back. But that's where we are. On Wed, Aug 28, 2019 at 11:55 PM Sabine <notifications(a)github.com> wrote: > I want to bring this up again. A summary of the history of this thread is > that on windows, the time of the Image (Time now) gets a gap to NTP Time > which grows by several seconds each day. > > This has already be discussed here > > http://forum.world.st/Time-millisecondClockValue-was-The-Trunk-Morphic-mt-1… > and here > http://forum.world.st/Windows-image-has-system-time-delay-td5096643.html#a5… > > I start my production server images new each week since 6 months now > because of this issue. > > And after a discussion at ESUG yesterday in the evening here again: yes, I > could find ways for workarounds. But I think this is a serious issue which > should be fixed because if you can not rely on the time in the image you > can get various problems. > > I have put a link to this issue here pharo-project/opensmalltalk-vm#12 > <https://github.com/pharo-project/opensmalltalk-vm/issues/12> (did not > know that it is a vm issue when reporting it 6 months ago). > > I really would appreciate having a good solution for this. > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <https://github.com/pharo-project/pharo/issues/2792?email_source=notificatio…>, > or mute the thread > <https://github.com/notifications/unsubscribe-auth/ADY5VUGI4LUMPBSNSN62NGTQG…> > . > -- _,,,^..^,,,_ best, Eliot

1 0

__VERSION__ in squeakvm classic
by stes 30 Mar '20

30 Mar '20

In the subversion sources of "Squeakvm" classic platforms/unix/vm/config.cmake there is a line SET (VM_BUILD_STRING "\"Unix built on \"__DATE__ \" \"__TIME__\" Compiler: \"__VERSION__") Now on Solaris 11.4 or 11.3 with SunPRO C compiler there is an issue with __VERSION__ the following sample program: bash-4.4$ cat x.c #include <stdio.h> int main(int argc,char *argv[]) { printf("%s\n",__DATE__); printf("%s\n",__TIME__); printf("%s\n",__VERSION__); } when compiled with GCC 9.2 on Solaris I get: $ gcc x.c $ ./a.out Mar 29 2020 15:43:33 9.2.0 when compiling with SunPRO C 12.6 on Solaris the result is : $ cc x.c "x.c", line 8: undefined symbol: __VERSION__ cc: acomp failed for x.c I could be using GCC 9.2 to build the squeakvm of course but because I tried with the SunPRO C, this only defines __DATE__ and __TIME__ but not __VERSION__ (as far as I know). So perhaps the best level to fix this (minor and only) issue with the compile of the squeakvm with the SunPRO C would be at the platforms/unix/vm/config.cmake level -- Sent from: http://forum.world.st/Squeak-VM-f104410.html

2 8

build64 in squeakvm classic
by stes 30 Mar '20

30 Mar '20

By default the example Makefile from ./platforms/unix/cmake/Makefile.example seems to be building for the build64 also a -m32 binary on Solaris bash-4.4$ file ./lib/squeak/4.16.7-3775_64bit/squeakvm64 ./lib/squeak/4.16.7-3775_64bit/squeakvm64: ELF 32-bit LSB executable 80386 Version 1 [SSE2 SSE], dynamically linked, not stripped If I change the Makefile.example to use -m64 (to generate 64bit) it works: bash-4.4$ file ./lib/squeak/4.16.7-3775_64bit/squeakvm64 ./lib/squeak/4.16.7-3775_64bit/squeakvm64: ELF 64-bit LSB executable AMD64 Version 1 [SSE2 SSE], dynamically linked, not stripped there was one patch required for Index: platforms/unix/vm-sound-Sun/sqUnixSoundSun.c =================================================================== --- platforms/unix/vm-sound-Sun/sqUnixSoundSun.c (revision 3775) +++ platforms/unix/vm-sound-Sun/sqUnixSoundSun.c (working copy) @@ -72,7 +72,7 @@ #endif static sqInt sound_Stop(void); -static int sound_AvailableSpace(void); *+static sqInt sound_AvailableSpace(void);* static int auFd= -1; /* open on /dev/audio */ static int auCtlFd= -1; /* open on /dev/audioctl */ the type "int" should be changed to "sqInt" for the 64 bit compile now I'm not sure how to test that it opens a 64bit Squeak image ... -- Sent from: http://forum.world.st/Squeak-VM-f104410.html

2 3

VM Maker: VMMaker.oscog-eem.2735.mcz
by commits＠source.squeak.org 30 Mar '20

30 Mar '20

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2735.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2735 Author: eem Time: 29 March 2020, 8:15:45.667712 pm UUID: 001c245b-2f3a-4d8e-923f-b4acac100afe Ancestors: VMMaker.oscog-eem.2734 Allow volatile variables into the VM's variable structure. Make sure CFramePointer and CStackPointer are marked volatile. =============== Diff against VMMaker.oscog-eem.2734 =============== Item was changed: ----- Method: CCodeGeneratorGlobalStructure>>placeInStructure: (in category 'C code generator') ----- placeInStructure: var "See if we should put this array into a structure. The variables listed are hardcoded as C in the interpreter thus they don't get resolved via TVariableNode logic. Also let's ignore variables that have special definitions that require initialization, and the function def which has problems." | check | check := variableDeclarations at: var ifAbsent: ['']. (check includes: $=) ifTrue: [^false]. (check includes: $() ifTrue: [^false]. (check includesSubstring: 'static') ifTrue: [^false]. - (check includesSubstring: 'volatile') ifTrue: [^false]. ^(vmClass mustBeGlobal: var) not! Item was changed: ----- Method: CoInterpreter class>>declareCVarsIn: (in category 'translation') ----- declareCVarsIn: aCCodeGenerator "Override to avoid repeating StackInterpreter's declarations and add our own extensions" self class == thisContext methodClass ifFalse: [^self]. "Don't duplicate decls in subclasses" aCCodeGenerator addHeaderFile:'"sqCogStackAlignment.h"'; addHeaderFile:'"cogmethod.h"'. NewspeakVM ifTrue: [aCCodeGenerator addHeaderFile:'"nssendcache.h"']. aCCodeGenerator addHeaderFile: (aCCodeGenerator vmClass isThreadedVM ifTrue: ['"cointerpmt.h"'] ifFalse: ['"cointerp.h"']); addHeaderFile:'"cogit.h"'. aCCodeGenerator vmClass declareInterpreterVersionIn: aCCodeGenerator defaultName: aCCodeGenerator interpreterVersion. aCCodeGenerator var: #heapBase type: #usqInt; var: #statCodeCompactionUsecs type: #usqLong; var: #maxLiteralCountForCompile declareC: 'sqInt maxLiteralCountForCompile = MaxLiteralCountForCompile /* ', MaxLiteralCountForCompile printString, ' */'; var: #minBackwardJumpCountForCompile declareC: 'sqInt minBackwardJumpCountForCompile = MinBackwardJumpCountForCompile /* ', MinBackwardJumpCountForCompile printString, ' */'. aCCodeGenerator removeVariable: 'atCache'. "Way too much trouble than it's worth in the Cog VM" aCCodeGenerator var: #primTraceLogIndex type: #'unsigned char'; var: #primTraceLog declareC: 'sqInt primTraceLog[256]'; var: #traceLog declareC: 'sqInt traceLog[TraceBufferSize /* ', TraceBufferSize printString, ' */]'; var: #traceSources type: #'char *' array: TraceSources. aCCodeGenerator + var: #CFramePointer type: #'volatile usqIntptr_t'; + var: #CStackPointer type: #'volatile usqIntptr_t'! - var: #CFramePointer type: #'usqIntptr_t'; - var: #CStackPointer type: #'usqIntptr_t'!

1 0

VM Maker: VMMaker.oscog-eem.2734.mcz
by commits＠source.squeak.org 30 Mar '20

30 Mar '20

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2734.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2734 Author: eem Time: 29 March 2020, 6:42:58.093616 pm UUID: ab7ba9fb-7121-4fb0-a741-c22da8b1fbd0 Ancestors: VMMaker.oscog-eem.2733 CoInterpeter: No longer rely on setjmp/longjmp to get back into the interpreter from arbitrary locatons. Instead, since CoInterpreter maintains the base of the C stack in CFramePointer & CStackPointer, it is straight-forward for us to simply call interpret after doing the switch to the C stack, avoiding issues such as stack unwind problems in longjmp. And of course the implementation is simpler thn setjmp/longjmp and so faster." This commit adds the new machinery. A subsequent commit will clean-up and delete the old machinery. P.S. this change exposes what looks like a bug in the Bochs simulators, so currently neither x86 nor x64 simulate. We shall have to fix this very soon. =============== Diff against VMMaker.oscog-eem.2733 =============== Item was changed: ----- Method: CoInterpreter>>ceBaseFrameReturn: (in category 'trampolines') ----- ceBaseFrameReturn: returnValue "Return across a page boundary. The context to return to (which may be married) is stored in the first word of the stack. We get here when a return instruction jumps to the ceBaseFrameReturn: address that is the return pc for base frames. A consequence of this is that the current frame is no longer valid since an interrupt may have overwritten its state as soon as the stack pointer has been cut-back beyond the return pc. So to have a context to send the cannotReturn: message to we also store the base frame's context in the second word of the stack page." <api> | contextToReturnTo contextToReturnFrom isAContext thePage newPage frameAbove | <var: #thePage type: #'StackPage *'> <var: #newPage type: #'StackPage *'> <var: #frameAbove type: #'char *'> self assert: (stackPages stackPageFor: stackPointer) = stackPage. self assert: stackPages mostRecentlyUsedPage = stackPage. cogit assertCStackWellAligned. self assert: framePointer = 0. self assert: stackPointer <= (stackPage baseAddress - objectMemory wordSize). self assert: stackPage baseFP + (2 * objectMemory wordSize) < stackPage baseAddress. "We would like to use the following assert but we can't since the stack pointer will be above the base frame pointer in the base frame return and hence the 0 a base frame pointer points at could be overwritten which will cause the isBaseFrame assert in frameCallerContext: to fail." "self assert: (self frameCallerContext: stackPage baseFP) = (stackPages longAt: stackPage baseAddress)." self assert: ((objectMemory addressCouldBeObj: (stackPages longAt: stackPage baseAddress - objectMemory wordSize)) and: [objectMemory isContext: (stackPages longAt: stackPage baseAddress - objectMemory wordSize)]). contextToReturnTo := stackPages longAt: stackPage baseAddress. self assert: (objectMemory addressCouldBeObj: contextToReturnTo). "The stack page is effectively free now, so free it. We must free it to be correct in determining if contextToReturnTo is still married, and in case makeBaseFrameFor: cogs a method, which may cause a code compaction, in which case the frame must be free to avoid the relocation machinery tracing the dead frame. Since freeing now temporarily violates the page-list ordering invariant, use the assert-free version." stackPages freeStackPageNoAssert: stackPage. isAContext := objectMemory isContext: contextToReturnTo. (isAContext and: [self isStillMarriedContext: contextToReturnTo]) ifTrue: [framePointer := self frameOfMarriedContext: contextToReturnTo. thePage := stackPages stackPageFor: framePointer. framePointer = thePage headFP ifTrue: [stackPointer := thePage headSP] ifFalse: ["Returning to some interior frame, presumably because of a sender assignment. Move the frames above to another page (they may be in use, e.g. via coroutining). Make the interior frame the top frame." frameAbove := self findFrameAbove: framePointer inPage: thePage. "Since we've just deallocated a page we know that newStackPage won't deallocate an existing one." newPage := stackPages newStackPage. self assert: newPage = stackPage. self moveFramesIn: thePage through: frameAbove toPage: newPage. stackPages markStackPageMostRecentlyUsed: newPage. self setStackPointersFromPage: thePage]] ifFalse: [(isAContext and: [objectMemory isIntegerObject: (objectMemory fetchPointer: InstructionPointerIndex ofObject: contextToReturnTo)]) ifFalse: [contextToReturnFrom := stackPages longAt: stackPage baseAddress - objectMemory wordSize. self tearDownAndRebuildFrameForCannotReturnBaseFrameReturnFrom: contextToReturnFrom to: contextToReturnTo returnValue: returnValue. ^self externalCannotReturn: returnValue from: contextToReturnFrom]. "void the instructionPointer to stop it being incorrectly updated in a code compaction in makeBaseFrameFor:." instructionPointer := 0. thePage := self makeBaseFrameFor: contextToReturnTo. self setStackPointersFromPage: thePage]. self setStackPageAndLimit: thePage. self assert: (stackPages stackPageFor: framePointer) = stackPage. (self isMachineCodeFrame: framePointer) ifTrue: [self push: returnValue. cogit ceEnterCogCodePopReceiverReg. "NOTREACHED"]. instructionPointer := self stackTop. instructionPointer = cogit ceReturnToInterpreterPC ifTrue: [instructionPointer := self iframeSavedIP: framePointer]. self setMethod: (self iframeMethod: framePointer). self stackTopPut: returnValue. "a.k.a. pop saved ip then push result" self assert: (self checkIsStillMarriedContext: contextToReturnTo currentFP: framePointer). + self invokeInterpreterFromMachineCode. - self siglong: reenterInterpreter jmp: ReturnToInterpreter. "NOTREACHED" ^nil! Item was added: + ----- Method: CoInterpreter>>ceEnterInterpreterOnReturnFromCogCode (in category 'trampolines') ----- + ceEnterInterpreterOnReturnFromCogCode + "Perform a return from a machine code frame to an interpreted frame. + The machine code has executed a return instruction when the return address + is set to ceReturnToInterpreterPC. Push the result and call interpret." + <api> + self assert: (objectMemory addressCouldBeOop: self stackTop). + self deny: (self isMachineCodeFrame: framePointer). + self setMethod: (self iframeMethod: framePointer). + instructionPointer := self iframeSavedIP: framePointer. + self assertValidExecutionPointe: instructionPointer r: framePointer s: stackPointer imbar: true line: #'__LINE__'. + self invokeInterpreterFromMachineCode. + "NOTREACHED" + ^nil! Item was removed: - ----- Method: CoInterpreter>>ceReturnToInterpreter: (in category 'trampolines') ----- - ceReturnToInterpreter: anOop - "Perform a return from a machine code frame to an interpreted frame. - The machine code has executed a return instruction when the return address - is set to ceReturnToInterpreterPC. Return the result and switch to the interpreter." - <api> - self assert: (objectMemory addressCouldBeOop: anOop). - self flag: 'are you really sure setStackPageAndLimit: is needed?'. - "I think you're only doing this for the markStackPageMostRecentlyUsed: - and that's probably not needed either" - self setStackPageAndLimit: stackPage. - self assert: (self isMachineCodeFrame: framePointer) not. - self setMethod: (self iframeMethod: framePointer). - self assertValidExecutionPointe: (self iframeSavedIP: framePointer) - r: framePointer - s: stackPointer - imbar: true - line: #'__LINE__'. - instructionPointer := self iframeSavedIP: framePointer. - self push: anOop. - self siglong: reenterInterpreter jmp: ReturnToInterpreter. - "NOTREACHED" - ^nil! Item was changed: ----- Method: CoInterpreter>>ceSendMustBeBooleanTo:interpretingAtDelta: (in category 'trampolines') ----- ceSendMustBeBooleanTo: aNonBooleanObject interpretingAtDelta: jumpSize "For RegisterAllocatingCogit we want the pc following a conditional branch not to be reachable, so we don't have to generate code to reload registers. But notionally the pc following a conditional branch is reached when continuing from a mustBeBoolean error. Instead of supporting this in the JIT, simply convert to an interpreter frame, backup the pc to the branch, reenter the interpreter and hence retry the mustBeBoolean send therein. N.B. We could do this for immutability violations too, but immutability is used in actual applications and so should be performant, whereas mustBeBoolean errors are extremely rare and so we choose brevity over performance in this case." <api> | cogMethod methodObj methodHeader startBcpc | <var: 'cogMethod' type: #'CogBlockMethod *'> <var: 'p' type: #'char *'> self assert: (objectMemory addressCouldBeOop: aNonBooleanObject). cogMethod := self mframeCogMethod: framePointer. ((self mframeIsBlockActivation: framePointer) and: [cogMethod cmIsFullBlock not]) ifTrue: [methodHeader := (self cCoerceSimple: cogMethod cmHomeMethod to: #'CogMethod *') methodHeader. methodObj := (self cCoerceSimple: cogMethod cmHomeMethod to: #'CogMethod *') methodObject. startBcpc := cogMethod startpc] ifFalse: [methodHeader := (self cCoerceSimple: cogMethod to: #'CogMethod *') methodHeader. methodObj := (self cCoerceSimple: cogMethod to: #'CogMethod *') methodObject. startBcpc := self startPCOfMethod: methodObj]. "Map the machine code instructionPointer to the interpreter instructionPointer of the branch." instructionPointer := self popStack. instructionPointer := cogit bytecodePCFor: instructionPointer startBcpc: startBcpc in: cogMethod. instructionPointer := methodObj + objectMemory baseHeaderSize + instructionPointer - jumpSize - 1. "pre-decrement" "Make space for the two extra fields in an interpreter frame" stackPointer to: framePointer + FoxMFReceiver by: objectMemory wordSize do: [:p| | oop | oop := objectMemory longAt: p. objectMemory longAt: p - objectMemory wordSize - objectMemory wordSize put: (objectMemory longAt: p)]. stackPointer := stackPointer - objectMemory wordSize - objectMemory wordSize. self push: aNonBooleanObject. "Fill in the fields" objectMemory longAt: framePointer + FoxIFrameFlags put: (self encodeFrameFieldHasContext: (self mframeHasContext: framePointer) isBlock: (self mframeIsBlockActivation: framePointer) numArgs: cogMethod cmNumArgs); longAt: framePointer + FoxIFSavedIP put: 0; longAt: framePointer + FoxMethod put: methodObj. "and now reenter the interpreter..." self setMethod: methodObj methodHeader: methodHeader. + self invokeInterpreterFromMachineCode. + "NOTREACHED" + ^nil! - self siglong: reenterInterpreter jmp: ReturnToInterpreter.! Item was changed: ----- Method: CoInterpreter>>flushMethodsWithMachineCodePrimitivesAndContinueAnswering: (in category 'primitive support') ----- flushMethodsWithMachineCodePrimitivesAndContinueAnswering: result "Arrange that any and all cog methods with machine code primitives can be and are discarded. Hence scan contexts and map their PCs to bytecode PCs if required, and scan frames, divorcing the frames of activationsif required. The continue execution answering result. THIS MUST BE INVOKED IN THE CONTEXT OF A PRIMITIVE. It exists to support vmParameterAt:put:." | activeContext theFrame thePage | <var: #theFrame type: #'char *'> <var: #thePage type: #'StackPage *'> activeContext := self ensureFrameIsMarried: framePointer SP: stackPointer. self ensurePushedInstructionPointer. self externalWriteBackHeadFramePointers. self divorceMachineCodeFramesWithMachineCodePrimitiveMethod. self ensureAllContextsWithMethodMachineCodePrimitiveMethodHaveBytecodePCs. cogit unlinkSendsToMachineCodePrimitiveMethodsAndFreeIf: true. "If flushing led to divorce continue in the interpreter." (self isStillMarriedContext: activeContext) ifFalse: [self zeroStackPage. "to avoid assert in marryContextInNewStackPageAndInitializeInterpreterRegisters:" self marryContextInNewStackPageAndInitializeInterpreterRegisters: activeContext. "pop bogus machine-code instructionPointer, arguments and receiver" self pop: argumentCount + 2 thenPush: result. + cogit ceInvokeInterpret - self siglong: reenterInterpreter jmp: ReturnToInterpreter. "NOTREACHED"]. "If not, work out where we are and continue" theFrame := self frameOfMarriedContext: activeContext. thePage := stackPages stackPageFor: theFrame. self assert: thePage headFP = theFrame. self setStackPageAndLimit: thePage. self setStackPointersFromPage: thePage. instructionPointer := self popStack. self pop: argumentCount + 1 thenPush: result! Item was changed: ----- Method: CoInterpreter>>interpretMethodFromMachineCode (in category 'message sending') ----- interpretMethodFromMachineCode "Execute a method interpretively from machine code. We assume (require) that newMethod messageSelector, primitiveFunctionPointer and argumentCount have been set in the caller. Once evaluated either continue in the interpreter via a jongjmp or in machine code via an enilopmart (a form of longjmp - a stinking rose by any other name)." <inline: false> cogit assertCStackWellAligned. self assert: (self validInstructionPointer: instructionPointer inFrame: framePointer). primitiveFunctionPointer ~= 0 ifTrue: [primitiveFunctionPointer = #primitiveInvokeObjectAsMethod ifTrue: [self assert: (objectMemory isOopCompiledMethod: newMethod) not] ifFalse: [self assert: ((objectMemory isOopCompiledMethod: newMethod) and: [(self primitiveIndexOf: newMethod) ~= 0])]. "Invoke an interpreter primitive (because the method is to be interpreted or has not yet been compiled). This is very similar to invoking an interpreter primitive from a compiled primitive (see e.g. SimpleStackBasedCogit>>compileInterpreterPrimitive:). Cut back the stack pointer (done above) to skip the return address and invoke the function. On return if it has succeeded simply continue otherwise restore the stackPointer, collect the pc and interpret. Note that frame building primitives such as primitiveClosureValue, primitiveEvaluateMethod et al will not return but will instead jump into either machine code or longjmp back to the interpreter." "Assign stackPage headFP so we can tell if the primitive built a frame. We can't simply save the framePointer since e.g. assignment to contexts (via primitiveInstVarAt:put:) can change the framePointer. But context assignments will change both the framePointer and stackPage headFP." self assert: (framePointer < stackPage baseAddress and: [framePointer > (stackPage realStackLimit - (LargeContextSlots * objectMemory bytesPerOop / 2))]). stackPage headFP: framePointer. self isPrimitiveFunctionPointerAnIndex ifTrue: [self externalQuickPrimitiveResponse. primFailCode := 0] ifFalse: [self slowPrimitiveResponse]. self successful ifTrue: [self return: self popStack toExecutive: false "NOTREACHED"]] ifFalse: [self assert: ((objectMemory isOopCompiledMethod: newMethod) and: [(self primitiveIndexOf: newMethod) = 0 or: [(self functionPointerFor: (self primitiveIndexOf: newMethod) inClass: objectMemory nilObject) = 0 or: [self isNullExternalPrimitiveCall: newMethod]]])]. "if not primitive, or primitive failed, activate the method and reenter the interpreter" self activateNewMethod. + cogit ceInvokeInterpret. - self siglong: reenterInterpreter jmp: ReturnToInterpreter. "NOTREACHED" ^nil! Item was added: + ----- Method: CoInterpreter>>invokeInterpreterFromMachineCode (in category 'trampolines') ----- + invokeInterpreterFromMachineCode + "This is just a rename for a send of interpret, but provides + a simulation hook; see the CogVMSimulator subclass." + <inline: #always> + self interpret + + "NOTREACHED"! Item was changed: ----- Method: CoInterpreter>>return:toExecutive: (in category 'enilopmarts') ----- return: returnValue toExecutive: inInterpreter "We have made a context switch, either when interpreting or from machine code. Effectively return to the current frame, either by entering machine code, or longjmp-ing back to the interpreter or simply returning, depending on where we are." cogit assertCStackWellAligned. (self isMachineCodeFrame: framePointer) ifTrue: [self assertValidExecutionPointe: instructionPointer r: framePointer s: stackPointer imbar: false line: #'__LINE__'. self push: instructionPointer. self push: returnValue. cogit ceEnterCogCodePopReceiverReg "NOTREACHED"]. self push: returnValue. self setMethod: (self iframeMethod: framePointer). self assertValidExecutionPointe: instructionPointer r: framePointer s: stackPointer imbar: true line: #'__LINE__'. instructionPointer = cogit ceReturnToInterpreterPC ifTrue: [instructionPointer := self iframeSavedIP: framePointer]. inInterpreter ifTrue: [^nil]. + cogit ceInvokeInterpret. - self siglong: reenterInterpreter jmp: ReturnToInterpreter. "NOTREACHED" ^nil! Item was changed: ----- Method: CoInterpreter>>returnToExecutive:postContextSwitch: (in category 'enilopmarts') ----- returnToExecutive: inInterpreter postContextSwitch: switchedContext "Return to the current frame, either by entering machine code, or longjmp-ing back to the interpreter or simply returning, depending on where we are. To know whether to return or enter machine code we have to know from whence we came. We could have come from the interpreter, either directly or via a machine code primitive. We could have come from machine code. The instructionPointer tells us where from. If it is above startOfMemory we're in the interpreter. If it is below, then we are in machine-code unless it is ceReturnToInterpreterPC, in which case we're in a machine-code primitive called from the interpreter." <inline: false> | cogMethod retValue fullyInInterpreter | <var: #cogMethod type: #'CogBlockMethod *'> cogit assertCStackWellAligned. (self isMachineCodeFrame: framePointer) ifTrue: [self assertValidExecutionPointe: instructionPointer r: framePointer s: stackPointer imbar: false line: #'__LINE__'. "If returning after a context switch then a result may have to be popped from the stack. If the process is suspended at a send then the result of the primitive in which the process was suspended is still on the stack and must be popped into ReceiverResultReg. If not, nothing should be popped and ReceiverResultReg gets the receiver." switchedContext ifTrue: [cogMethod := self mframeCogMethod: framePointer. self assert: (instructionPointer > cogit minCogMethodAddress and: [instructionPointer < cogit maxCogMethodAddress]). (instructionPointer ~= (cogMethod asInteger + cogMethod stackCheckOffset) and: [cogit isSendReturnPC: instructionPointer]) ifTrue: [self assert: (objectMemory addressCouldBeOop: self stackTop). retValue := self popStack] ifFalse: [retValue := self mframeReceiver: framePointer]] ifFalse: [retValue := self mframeReceiver: framePointer]. self push: instructionPointer. self push: retValue. cogit ceEnterCogCodePopReceiverReg "NOTREACHED"]. self setMethod: (self iframeMethod: framePointer). fullyInInterpreter := inInterpreter. instructionPointer = cogit ceReturnToInterpreterPC ifTrue: [instructionPointer := (self iframeSavedIP: framePointer) asUnsignedInteger. fullyInInterpreter := false]. self assertValidExecutionPointe: instructionPointer r: framePointer s: stackPointer imbar: true line: #'__LINE__'. fullyInInterpreter ifFalse: + [cogit ceInvokeInterpret - [self siglong: reenterInterpreter jmp: ReturnToInterpreter. "NOTREACHED"]. ^nil! Item was changed: ----- Method: CoInterpreterPrimitives>>primitiveLongRunningPrimitiveSemaphore (in category 'process primitives') ----- primitiveLongRunningPrimitiveSemaphore "Primitive. Install the semaphore to be used for collecting long-running primitives, or nil if no semaphore should be used." | sema flushState activeContext | <export: true> self methodArgumentCount ~= 1 ifTrue: [^self primitiveFailFor: PrimErrBadNumArgs]. sema := self stackValue: 0. sema = objectMemory nilObject ifTrue: [flushState := longRunningPrimitiveCheckSemaphore notNil. longRunningPrimitiveCheckSemaphore := nil] ifFalse: [flushState := longRunningPrimitiveCheckSemaphore isNil. (objectMemory isSemaphoreOop: sema) ifFalse: [^self primitiveFailFor: PrimErrBadArgument]. longRunningPrimitiveCheckSemaphore := sema]. "If we've switched checking on or off we must void machine code (and machine code pcs in contexts) since we will start or stop setting newMethod in machine code primitive invocations, and so generate slightly different code from here on in." flushState ifTrue: [self push: instructionPointer. activeContext := self voidVMStateForSnapshotFlushingExternalPrimitivesIf: false. self marryContextInNewStackPageAndInitializeInterpreterRegisters: activeContext. self assert: (((self stackValue: 0) = objectMemory nilObject and: [longRunningPrimitiveCheckSemaphore isNil]) or: [(self stackValue: 0) = longRunningPrimitiveCheckSemaphore and: [objectMemory isSemaphoreOop: sema]])]. self voidLongRunningPrimitive: 'install'. self pop: 1. flushState ifTrue: + [cogit ceInvokeInterpret]! - [self siglong: reenterInterpreter jmp: ReturnToInterpreter]! Item was changed: ----- Method: CoInterpreterPrimitives>>primitiveProfileSemaphore (in category 'process primitives') ----- primitiveProfileSemaphore "Primitive. Install the semaphore to be used for profiling, or nil if no semaphore should be used. See also primitiveProfileStart." | sema flushState activeContext | <export: true> self methodArgumentCount ~= 1 ifTrue: [^self primitiveFailFor: PrimErrBadNumArgs]. sema := self stackValue: 0. sema = objectMemory nilObject ifTrue: [flushState := profileSemaphore ~= objectMemory nilObject] ifFalse: [flushState := profileSemaphore = objectMemory nilObject. (objectMemory isSemaphoreOop: sema) ifFalse: [^self primitiveFailFor: PrimErrBadArgument]]. profileSemaphore := sema. "If we've switched profiling on or off we must void machine code (and machine code pcs in contexts) since we will start or stop testing the profile clock in machine code primitive invocations, and so generate slightly different code from here on in." flushState ifTrue: [self push: instructionPointer. activeContext := self voidVMStateForSnapshotFlushingExternalPrimitivesIf: false. self marryContextInNewStackPageAndInitializeInterpreterRegisters: activeContext. self assert: (((self stackValue: 0) = objectMemory nilObject and: [profileSemaphore = objectMemory nilObject]) or: [(self stackValue: 0) = profileSemaphore and: [objectMemory isSemaphoreOop: sema]])]. profileProcess := profileMethod := objectMemory nilObject. self pop: 1. flushState ifTrue: + [cogit ceInvokeInterpret]! - [self siglong: reenterInterpreter jmp: ReturnToInterpreter]! Item was changed: ----- Method: CoInterpreterPrimitives>>primitiveSnapshot (in category 'system control primitives') ----- primitiveSnapshot "Save a normal snapshot under the same name as it was loaded unless it has been renamed by the last primitiveImageName. Override to jump to the interpreter because the machine code zone is now void." <inline: false> self snapshot: false. (self iframeMethod: framePointer) = newMethod ifTrue: ["snapshot: has reached the end and built a frame. In the JIT we need to back-up the pc before reentering the interpreter." instructionPointer := instructionPointer - 1]. + cogit ceInvokeInterpret - self siglong: reenterInterpreter jmp: ReturnToInterpreter "NOTREACHED"! Item was changed: ----- Method: CoInterpreterPrimitives>>primitiveSnapshotEmbedded (in category 'system control primitives') ----- primitiveSnapshotEmbedded "Save an embedded snapshot. Override to jump to the interpreter because the machine code zone is now void." <inline: false> self snapshot: true. (self iframeMethod: framePointer) = newMethod ifTrue: ["snapshot: has reached the end and built a frame. In the JIT we need to back-up the pc before reentering the interpreter." instructionPointer := instructionPointer - 1]. + cogit ceInvokeInterpret - self siglong: reenterInterpreter jmp: ReturnToInterpreter "NOTREACHED"! Item was changed: ----- Method: CoInterpreterPrimitives>>primitiveVoidVMState (in category 'system control primitives') ----- primitiveVoidVMState "Void all internal VM state in the stack and machine code zones Override to jump to the interpreter because the machine code zone is now void." | activeContext | self push: instructionPointer. activeContext := self voidVMStateForSnapshotFlushingExternalPrimitivesIf: false. self marryContextInNewStackPageAndInitializeInterpreterRegisters: activeContext. + cogit ceInvokeInterpret - self siglong: reenterInterpreter jmp: ReturnToInterpreter "NOTREACHED"! Item was changed: ----- Method: CoInterpreterPrimitives>>primitiveVoidVMStateForMethod (in category 'system control primitives') ----- primitiveVoidVMStateForMethod "The receiver (or first argument) must be a compiledMethod. The optional (or second) argument must be a boolean. Clear all VM state associated with the method, including any machine code, or machine code pcs in context objects. If the optional boolean argument is false do not scan the heap looking for contexts." | activeContext methodObj scanHeapForContexts hasCogMethod theFrame thePage | <var: #theFrame type: #'char *'> <var: #thePage type: #'StackPage *'> scanHeapForContexts := true. "See comment ''One might think...'' below" "In Smalltalk allow both aMethod voidCogVMState and aMethod voidCogVMStateScanningContextsIf: aBoolean" argumentCount = 0 ifTrue: [methodObj := self stackTop] ifFalse: [methodObj := self stackValue: 1. scanHeapForContexts := self booleanValueOf: self stackTop. self failed ifTrue: [^self primitiveFailFor: PrimErrBadArgument]]. NewspeakVM ifFalse: [argumentCount > 1 ifTrue: [^self primitiveFailFor: PrimErrBadNumArgs]] ifTrue: "In the NewspeakVM we allow VMMirror voidStateFor: method scanningIf: aBoolean as well as the Smalltalk forms." [argumentCount >= 2 ifTrue: [argumentCount > 2 ifTrue: [^self primitiveFailFor: PrimErrBadNumArgs]. (objectMemory isOopCompiledMethod: methodObj) ifFalse: [^self primitiveFailFor: PrimErrBadArgument]]]. self flushMethodCacheForMethod: methodObj. activeContext := self ensureFrameIsMarried: framePointer SP: stackPointer. self ensurePushedInstructionPointer. self externalWriteBackHeadFramePointers. (hasCogMethod := self methodHasCogMethod: methodObj) ifTrue: [self divorceMachineCodeFramesWithMethod: methodObj]. "One might think (as this author did) that the heap scan is unnecessary if the method does not have a cog method. But it could be the case that the code zone has recently been reclaimed and so not having a cog method is no indication that it didn't have a cog method some time in the recent past, and that there are indeed still contexts with machine code pcs out there. The only steps that can be avoided are divorcing frames in the stack zone, and scanning to unlink and free if there isn't a cog method, unless we are told otherwise." scanHeapForContexts ifTrue: [self ensureAllContextsWithMethodHaveBytecodePCs: methodObj]. hasCogMethod ifTrue: [cogit unlinkSendsTo: methodObj andFreeIf: true]. "If flushing led to divorce continue in the interpreter." (self isStillMarriedContext: activeContext) ifFalse: [self zeroStackPage. "to avoid assert in marryContextInNewStackPageAndInitializeInterpreterRegisters:" self marryContextInNewStackPageAndInitializeInterpreterRegisters: activeContext. self popStack. "pop bogus machine-code instructionPointer" self assert: (methodObj = self stackTop or: [argumentCount > 0 and: [methodObj = (self stackValue: 1)]]). self pop: argumentCount. + cogit ceInvokeInterpret - self siglong: reenterInterpreter jmp: ReturnToInterpreter. "NOTREACHED"]. "If not, work out where we are and continue" theFrame := self frameOfMarriedContext: activeContext. thePage := stackPages stackPageFor: theFrame. self assert: thePage headFP = theFrame. self setStackPageAndLimit: thePage. self setStackPointersFromPage: thePage. instructionPointer := self popStack. self assert: (methodObj = self stackTop or: [argumentCount > 0 and: [methodObj = (self stackValue: 1)]]). self pop: argumentCount! Item was added: + ----- Method: CogVMSimulator>>invokeInterpreterFromMachineCode (in category 'trampolines') ----- + invokeInterpreterFromMachineCode + "In simulation we raise an exception to unwind the stack and get back to the current level of interpret execution." + reenterInterpreter + returnValue: ReturnToInterpreter; + signal + + "NOTREACHED"! Item was changed: CogClass subclass: #Cogit (excessive size, no diff calculated) Item was changed: ----- Method: Cogit class>>declareCVarsIn: (in category 'translation') ----- declareCVarsIn: aCCodeGenerator | backEnd | backEnd := CogCompilerClass basicNew. #( 'coInterpreter' 'objectMemory' 'methodZone' 'objectRepresentation' 'cogBlockMethodSurrogateClass' 'cogMethodSurrogateClass' 'nsSendCacheSurrogateClass' 'threadManager' 'processor' 'lastNInstructions' 'simulatedAddresses' 'simulatedTrampolines' 'simulatedVariableGetters' 'simulatedVariableSetters' 'processorFrameValid' 'printRegisters' 'printInstructions' 'clickConfirm' 'singleStep') do: [:simulationVariableNotNeededForRealVM| aCCodeGenerator removeVariable: simulationVariableNotNeededForRealVM]. NewspeakVM ifFalse: [#( 'selfSendTrampolines' 'dynamicSuperSendTrampolines' 'implicitReceiverSendTrampolines' 'outerSendTrampolines' 'ceEnclosingObjectTrampoline' 'numIRCs' 'indexOfIRC' 'theIRCs') do: [:variableNotNeededInNormalVM| aCCodeGenerator removeVariable: variableNotNeededInNormalVM]]. aCCodeGenerator removeConstant: #COGMTVM. "this should be defined at compile time" aCCodeGenerator addHeaderFile:'<stddef.h>'; "for e.g. offsetof" addHeaderFile:'"sqCogStackAlignment.h"'; addHeaderFile:'"dispdbg.h"'; "must precede cointerp.h & cogit.h otherwise NoDbgRegParms gets screwed up" addHeaderFile:'"cogmethod.h"'. NewspeakVM ifTrue: [aCCodeGenerator addHeaderFile:'"nssendcache.h"']. aCCodeGenerator addHeaderFile:'#if COGMTVM'; addHeaderFile:'"cointerpmt.h"'; addHeaderFile:'#else'; addHeaderFile:'"cointerp.h"'; addHeaderFile:'#endif'; addHeaderFile:'"cogit.h"'. aCCodeGenerator var: #ceGetFP declareC: 'usqIntptr_t (*ceGetFP)(void)'; var: #ceGetSP declareC: 'usqIntptr_t (*ceGetSP)(void)'; var: #ceCaptureCStackPointers declareC: 'void (*ceCaptureCStackPointers)(void)'; + var: #ceInvokeInterpret + declareC: 'void (*ceInvokeInterpret)(void)'; var: #ceEnterCogCodePopReceiverReg declareC: 'void (*ceEnterCogCodePopReceiverReg)(void)'; var: #realCEEnterCogCodePopReceiverReg declareC: 'void (*realCEEnterCogCodePopReceiverReg)(void)'; var: #ceCallCogCodePopReceiverReg declareC: 'void (*ceCallCogCodePopReceiverReg)(void)'; var: #realCECallCogCodePopReceiverReg declareC: 'void (*realCECallCogCodePopReceiverReg)(void)'; var: #ceCallCogCodePopReceiverAndClassRegs declareC: 'void (*ceCallCogCodePopReceiverAndClassRegs)(void)'; var: #realCECallCogCodePopReceiverAndClassRegs declareC: 'void (*realCECallCogCodePopReceiverAndClassRegs)(void)'; var: #postCompileHook declareC: 'void (*postCompileHook)(CogMethod *)'; var: #openPICList declareC: 'CogMethod *openPICList = 0'; var: #maxMethodBefore type: #'CogBlockMethod *'; var: 'enumeratingCogMethod' type: #'CogMethod *'. aCCodeGenerator var: #ceTryLockVMOwner declareC: '#if COGMTVM\usqIntptr_t (*ceTryLockVMOwner)(void)'; var: #ceUnlockVMOwner declareC: 'void (*ceUnlockVMOwner)(void)\#endif /* COGMTVM */'. backEnd numCheckLZCNTOpcodes > 0 ifTrue: [aCCodeGenerator var: #ceCheckLZCNTFunction declareC: 'static usqIntptr_t (*ceCheckLZCNTFunction)(void)']. backEnd numCheckFeaturesOpcodes > 0 ifTrue: [aCCodeGenerator var: #ceCheckFeaturesFunction declareC: 'static usqIntptr_t (*ceCheckFeaturesFunction)(void)']. backEnd numICacheFlushOpcodes > 0 ifTrue: [aCCodeGenerator var: #ceFlushICache declareC: 'static void (*ceFlushICache)(usqIntptr_t from, usqIntptr_t to)']. aCCodeGenerator var: #ceFlushDCache declareC: '#if DUAL_MAPPED_CODE_ZONE\static void (*ceFlushDCache)(usqIntptr_t from, usqIntptr_t to)\#endif'; var: #codeToDataDelta declareC: '#if DUAL_MAPPED_CODE_ZONE\static sqInt codeToDataDelta\#else\# define codeToDataDelta 0\#endif'. aCCodeGenerator declareVar: 'aMethodLabel' type: #'AbstractInstruction'; "Has to come lexicographically before backEnd & methodLabel" var: #backEnd declareC: 'AbstractInstruction * const backEnd = &aMethodLabel'; var: #methodLabel declareC: 'AbstractInstruction * const methodLabel = &aMethodLabel'. self declareC: #(abstractOpcodes stackCheckLabel blockEntryLabel blockEntryNoContextSwitch stackOverflowCall sendMiss entry noCheckEntry selfSendEntry dynSuperEntry fullBlockNoContextSwitchEntry fullBlockEntry picMNUAbort picInterpretAbort endCPICCase0 endCPICCase1 cPICEndOfCodeLabel) as: #'AbstractInstruction *' in: aCCodeGenerator. aCCodeGenerator declareVar: #cPICPrototype type: #'CogMethod *'; declareVar: #blockStarts type: #'BlockStart *'; declareVar: #fixups type: #'BytecodeFixup *'; declareVar: #methodZoneBase type: #usqInt. aCCodeGenerator var: #ordinarySendTrampolines declareC: 'sqInt ordinarySendTrampolines[NumSendTrampolines]'; var: #superSendTrampolines declareC: 'sqInt superSendTrampolines[NumSendTrampolines]'. BytecodeSetHasDirectedSuperSend ifTrue: [aCCodeGenerator var: #directedSuperSendTrampolines declareC: 'sqInt directedSuperSendTrampolines[NumSendTrampolines]'; var: #directedSuperBindingSendTrampolines declareC: 'sqInt directedSuperBindingSendTrampolines[NumSendTrampolines]']. NewspeakVM ifTrue: [aCCodeGenerator var: #selfSendTrampolines declareC: 'sqInt selfSendTrampolines[NumSendTrampolines]'; var: #dynamicSuperSendTrampolines declareC: 'sqInt dynamicSuperSendTrampolines[NumSendTrampolines]'; var: #implicitReceiverSendTrampolines declareC: 'sqInt implicitReceiverSendTrampolines[NumSendTrampolines]'; var: #outerSendTrampolines declareC: 'sqInt outerSendTrampolines[NumSendTrampolines]']. aCCodeGenerator var: #trampolineAddresses declareC: 'static char *trampolineAddresses[NumTrampolines*2]'; var: #objectReferencesInRuntime declareC: 'static usqInt objectReferencesInRuntime[NumObjRefsInRuntime+1]'; var: #labelCounter type: #int; var: #traceFlags declareC: 'int traceFlags = 8 /* prim trace log on by default */'; var: #cStackAlignment declareC: 'const int cStackAlignment = STACK_ALIGN_BYTES'. aCCodeGenerator declareVar: #minValidCallAddress type: #'usqIntptr_t'. aCCodeGenerator vmClass generatorTable ifNotNil: [:bytecodeGenTable| aCCodeGenerator var: #generatorTable declareC: 'static BytecodeDescriptor generatorTable[', bytecodeGenTable size printString, ']', (self tableInitializerFor: bytecodeGenTable in: aCCodeGenerator)]. "In C the abstract opcode names clash with the Smalltalk generator syntactic sugar. Most of the syntactic sugar is inlined, but alas some remains. Rename the syntactic sugar to avoid the clash." (self organization listAtCategoryNamed: #'abstract instructions') do: [:s| aCCodeGenerator addSelectorTranslation: s to: 'g', (aCCodeGenerator cFunctionNameFor: s)]. aCCodeGenerator addSelectorTranslation: #halt: to: 'haltmsg'. self declareFlagVarsAsByteIn: aCCodeGenerator! Item was changed: ----- Method: Cogit class>>mustBeGlobal: (in category 'translation') ----- mustBeGlobal: var "Answer if a variable must be global and exported. Used for inst vars that are accessed from VM support code." + ^#('ceBaseFrameReturnTrampoline' ceCaptureCStackPointers 'ceCheckForInterruptTrampoline' - ^#('ceBaseFrameReturnTrampoline' 'ceCaptureCStackPointers' 'ceCheckForInterruptTrampoline' ceEnterCogCodePopReceiverReg realCEEnterCogCodePopReceiverReg ceCallCogCodePopReceiverReg realCECallCogCodePopReceiverReg ceCallCogCodePopReceiverAndClassRegs realCECallCogCodePopReceiverAndClassRegs + ceInvokeInterpret 'ceReturnToInterpreterTrampoline' 'ceCannotResumeTrampoline' - 'ceReturnToInterpreterTrampoline' 'ceCannotResumeTrampoline' ceTryLockVMOwner ceUnlockVMOwner 'cmEntryOffset' 'cmNoCheckEntryOffset' 'cmDynSuperEntryOffset' 'cmSelfSendEntryOffset' 'missOffset' 'cbEntryOffset' 'cbNoSwitchEntryOffset' 'blockNoContextSwitchOffset' breakPC ceGetFP ceGetSP cFramePointerInUse traceFlags traceStores) includes: var! Item was changed: ----- Method: Cogit class>>numTrampolines (in category 'trampoline support') ----- numTrampolines + ^37 "29 + 4 each for self and super sends" + (LowcodeVM ifTrue: [3] ifFalse: [0]) + CogCompilerClass numTrampolines - ^38 "30 + 4 each for self and super sends" + (LowcodeVM ifTrue: [1] ifFalse: [0]) + CogCompilerClass numTrampolines "self withAllSubclasses collect: [:c| {c. (c instVarNames select: [:ea| ea beginsWith: 'ce']) size}]"! Item was added: + ----- Method: Cogit>>ceInvokeInterpret (in category 'simulation only') ----- + ceInvokeInterpret + <api: 'extern void (*ceInvokeInterpret)()'> + <doNotGenerate> + self simulateEnilopmart: ceInvokeInterpret numArgs: 1! Item was added: + ----- Method: Cogit>>genInvokeInterpretTrampoline (in category 'initialization') ----- + genInvokeInterpretTrampoline + "Switch to the C stack (do *not* save the Smalltalk stack pointers; + this is the caller's responsibility), and invoke interpret PDQ." + | startAddress | + <inline: false> + startAddress := methodZoneBase. + self zeroOpcodeIndex. + cFramePointerInUse + ifTrue: [backEnd genLoadCStackPointers] + ifFalse: [backEnd genLoadCStackPointer]. + self + compileCallFor: #interpret + numArgs: 0 arg: nil arg: nil arg: nil arg: nil + resultReg: NoReg + regsToSave: self emptyRegisterMask. + self outputInstructionsForGeneratedRuntimeAt: startAddress. + self recordGeneratedRunTime: 'ceInvokeInterpret' address: startAddress. + ^startAddress! Item was added: + ----- Method: Cogit>>genReturnToInterpreterTrampoline (in category 'initialization') ----- + genReturnToInterpreterTrampoline + self zeroOpcodeIndex. + "Set the instruction pointer to the interpreter frame's saved ip, set the method and the bytecode set offset, + then call interpret." + self PushR: ReceiverResultReg. "The result" + ^self genTrampolineFor: #ceEnterInterpreterOnReturnFromCogCode + called: 'ceEnterInterpreterOnReturnFromCogCode' + numArgs: 0 arg: nil arg: nil arg: nil arg: nil + regsToSave: self emptyRegisterMask + pushLinkReg: false + resultReg: NoReg + appendOpcodes: true! Item was changed: ----- Method: Cogit>>generateRunTimeTrampolines (in category 'initialization') ----- generateRunTimeTrampolines "Generate the run-time entries at the base of the native code zone and update the base." ceSendMustBeBooleanAddFalseTrampoline := self genMustBeBooleanTrampolineFor: objectMemory falseObject called: 'ceSendMustBeBooleanAddFalseTrampoline'. ceSendMustBeBooleanAddTrueTrampoline := self genMustBeBooleanTrampolineFor: objectMemory trueObject called: 'ceSendMustBeBooleanAddTrueTrampoline'. ceNonLocalReturnTrampoline := self genNonLocalReturnTrampoline. ceCheckForInterruptTrampoline := self genCheckForInterruptsTrampoline. "Neither of the context inst var access trampolines save registers. Their operation could cause arbitrary update of stack frames, so the assumption is that callers flush the stack before calling the context inst var access trampolines, and that everything except the result is dead afterwards." ceFetchContextInstVarTrampoline := self genTrampolineFor: #ceContext:instVar: called: 'ceFetchContextInstVarTrampoline' arg: ReceiverResultReg arg: SendNumArgsReg result: SendNumArgsReg. ceStoreContextInstVarTrampoline := self genTrampolineFor: #ceContext:instVar:value: called: 'ceStoreContextInstVarTrampoline' arg: ReceiverResultReg arg: SendNumArgsReg arg: ClassReg result: ReceiverResultReg. "to keep ReceiverResultReg live.". ceCannotResumeTrampoline := self genTrampolineFor: #ceCannotResume called: 'ceCannotResumeTrampoline'. + "ceInvokeInterpreter is an optimization and a work-around. Historically we used setjmp/longjmp to reenter the + interpreter at the current C stack base. The C stack base is set at start-up and on each callback enter and + callback return. The interpreter must be invoked whenever a non-machine-code method must be run. That might + be when invoking an interpreter method from one of the send linking routines (ceSend:...), or on continuing from + an evaluation primitive such as primitiveExecuteMethod. The problem here is that such primitives could have + been invoked by the interpreter or by machine code. So some form of non-local jump is required. But at least as + early as MSVC Community 2017, the Microshaft longjmp performs stack unwinding which gets hoplessly confused + (bless its little heart) by any stack switch between machine code and C stack, and raises a spurious + Stack cookie instrumentation code detected a stack-based buffer overrun + error from the bowels of gs_report.c _GSHandlerCheck. + Since the CoInterpreter maintains the base of the C stack in CFramePointer & CStackPointer, it is straight-forward + for us to simply call interpret after doing the switch to the C stack, avoiding the stack unwind issue altogether." + ceInvokeInterpret := self genInvokeInterpretTrampoline. "These two are unusual; they are reached by return instructions." + ceReturnToInterpreterTrampoline := self genReturnToInterpreterTrampoline. ceBaseFrameReturnTrampoline := self genReturnTrampolineFor: #ceBaseFrameReturn: called: 'ceBaseFrameReturnTrampoline' arg: ReceiverResultReg. + LowcodeVM ifTrue: + [ceFFICalloutTrampoline := self genFFICalloutTrampoline. + ceMallocTrampoline := self genTrampolineFor: #ceMalloc: + called: 'ceMallocTrampoline' + arg: ReceiverResultReg + result: TempReg. + ceFreeTrampoline := self genTrampolineFor: #ceFree: + called: 'ceFreeTrampoline' + arg: ReceiverResultReg]! - ceReturnToInterpreterTrampoline := self - genReturnTrampolineFor: #ceReturnToInterpreter: - called: 'ceReturnToInterpreterTrampoline' - arg: ReceiverResultReg. - ceMallocTrampoline := self genTrampolineFor: #ceMalloc: - called: 'ceMallocTrampoline' - arg: ReceiverResultReg - result: TempReg. - ceFreeTrampoline := self genTrampolineFor: #ceFree: - called: 'ceFreeTrampoline' - arg: ReceiverResultReg. - LowcodeVM ifTrue: [ - ceFFICalloutTrampoline := self genFFICalloutTrampoline. - ]!

1 0

VM Maker: FileAttributesPlugin.oscog-eem.54.mcz
by commits＠source.squeak.org 30 Mar '20

30 Mar '20

Eliot Miranda uploaded a new version of FileAttributesPlugin to project VM Maker: http://source.squeak.org/VMMaker/FileAttributesPlugin.oscog-eem.54.mcz ==================== Summary ==================== Name: FileAttributesPlugin.oscog-eem.54 Author: eem Time: 29 March 2020, 6:33:34.8893 pm UUID: 4aa6e35d-11f2-49de-9e53-ac5600cf4684 Ancestors: FileAttributesPlugin.oscog-akg.53 Get the plugin closer to compiling under MSVC (2017). =============== Diff against FileAttributesPlugin.oscog-akg.53 =============== Item was changed: ----- Method: FileAttributesPlugin class>>declareCVarsIn: (in category 'translation') ----- declareCVarsIn: cg self declareC: #('sCLPfn' 'sCOFfn') as: #'void *' in: cg. "Assume the security plugin can be loaded until proven otherwise" cg var: 'hasSecurityPlugin' declareC: 'int hasSecurityPlugin = 1'. cg addHeaderFile: '<errno.h>'. cg addHeaderFile: '<limits.h>'. + cg addHeaderFile: '<sys/types.h> - cg addHeaderFile: '<sys/types.h>'. - cg addHeaderFile: '<unistd.h>'. - cg addHeaderFile: '<dirent.h> #if _WIN32 || _WIN64 # include <windows.h> # include <winbase.h> + # if !!defined(_MSC_VER) /* i.e. cygwin or mingw */ + # include <unistd.h> + # include <dirent.h> + # endif # define FAIL() { return -1; } # include "sqWin32File.h" # if !!defined(PATH_MAX) /* work around bug in 64-bit cygwin; sigh :-( */ # define PATH_MAX 4096 # endif # define HAVE_CHMOD 0 # define HAVE_CHOWN 0 #else + # include <unistd.h> + # include <dirent.h> + # define HAVE_CHMOD 1 + # define HAVE_CHOWN 1 - #define HAVE_CHMOD 1 - #define HAVE_CHOWN 1 #endif'. cg addHeaderFile: '"faCommon.h"'. cg addHeaderFile: '<sys/stat.h> #if !!defined(HAVE_LSTAT) && !!defined(_WIN32) && !!defined(_WIN64) # define HAVE_LSTAT 1 + #endif + - #endif'. - cg addHeaderFile: '<unistd.h> /* AKG 2018 - ', self moduleName, '.c translated from class ', self name, ' */'!

1 0

← Newer
1
2
3
4
5
6
7
...
14
Older →

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev March 2020