Vm-dev

vm-dev@lists.squeakfoundation.org

11 participants
17005 discussions

VM Maker: VMMaker.oscog-eem.3368.mcz
by commits＠source.squeak.org 19 Apr '24

19 Apr '24

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.3368.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.3368 Author: eem Time: 19 April 2024, 11:41:04.395712 am UUID: 9f9dc531-0e56-45f7-8c71-b567cfe447f0 Ancestors: VMMaker.oscog-eem.3367 Fix a silly bug in primitiveShowDisplayRect, which checked for argument marcshalling failure *after* using the parameters!! MT VM: Apply ioEventThreadAffinity guard in primitiveForceDisplayUpdate & primitiveShowDisplayRect. Slang: allow breakpointing of creation of TMethods by adding a class var to CCodeGenerator (cuz the plumbing in VMMaker isn't conducive to getting at its code generator). =============== Diff against VMMaker.oscog-eem.3367 =============== Item was changed: Object subclass: #CCodeGenerator instanceVariableNames: 'vmClass structClasses translationDict asArgumentTranslationDict inlineList constants variables variableDeclarations scopeStack methods macros apiMethods apiVariables kernelReturnTypes currentMethod headerFiles globalVariableUsage useSymbolicConstants generateDeadCode requiredSelectors previousCommentMarksInlining previousCommenter logger suppressAsmLabels asmLabelCounts pools selectorTranslations staticallyResolvedPolymorphicReceivers optionsDictionary breakSrcInlineSelectors breakDestInlineSelectors breakOnInline vmMaker accessorDepthCache beganInlining' + classVariableNames: 'BreakSrcSelectors NoRegParmsInAssertVMs' - classVariableNames: 'NoRegParmsInAssertVMs' poolDictionaries: 'VMBasicConstants' category: 'VMMaker-Translation to C'! !CCodeGenerator commentStamp: 'tpr 5/2/2003 14:30' prior: 0! This class oversees the translation of a subset of Smalltalk to C, allowing the comforts of Smalltalk during development and the efficiency and portability of C for the resulting interpreter. See VMMaker for more useful info! Item was changed: ----- Method: CCodeGenerator>>compileToTMethodSelector:in: (in category 'utilities') ----- compileToTMethodSelector: selector in: aClass "Compile a method to a TMethod" (breakSrcInlineSelectors notNil and: [(breakSrcInlineSelectors includes: selector) and: [breakOnInline ~~ true]]) ifTrue: + [self halt: 'Compiling ', aClass name, '>>', selector]. - [self halt]. ^(aClass >> selector) asTranslationMethodOfClass: self translationMethodClass "was: | implementingClass | implementingClass := aClass. ^(Compiler new parse: ([aClass sourceCodeAt: selector] on: KeyNotFound do: [:ex| ""Quick hack for simulating Pharo images..."" (PharoVM and: [aClass == String class and: [selector == #findSubstringViaPrimitive:in:startingAt:matchTable:]]) ifFalse: [ex pass]. (implementingClass := ByteString) sourceCodeAt: #findSubstring:in:startingAt:matchTable:]) in: implementingClass notifying: nil) asTranslationMethodOfClass: self translationMethodClass"! Item was changed: ----- Method: CCodeGenerator>>initialize (in category 'initialize-release') ----- initialize translationDict := Dictionary new. inlineList := Array new. constants := Dictionary new: 100. variables := Set new: 100. variableDeclarations := Dictionary new: 100. methods := Dictionary new: 500. kernelReturnTypes := self computeKernelReturnTypes. macros := Dictionary new. self initializeCTranslationDictionary. headerFiles := OrderedCollection new. globalVariableUsage := Dictionary new. useSymbolicConstants := true. generateDeadCode := true. scopeStack := OrderedCollection new. self getLogger. pools := IdentitySet new. selectorTranslations := IdentityDictionary new. accessorDepthCache := IdentityDictionary new. beganInlining := false. suppressAsmLabels := false. previousCommentMarksInlining := false. previousCommenter := nil. breakSrcInlineSelectors := IdentitySet new. + breakDestInlineSelectors := IdentitySet new. + BreakSrcSelectors ifNotNil: [:them| breakSrcInlineSelectors addAll: them]! - breakDestInlineSelectors := IdentitySet new! Item was changed: ----- Method: CoInterpreterMT>>ioEventThreadAffinity (in category 'threading support') ----- ioEventThreadAffinity "Default" + <doNotGenerate> ^-1! Item was added: + ----- Method: CoInterpreterMT>>primitiveForceDisplayUpdate (in category 'I/O primitives') ----- + primitiveForceDisplayUpdate + "On some platforms, this primitive forces enqueued display updates to be processed immediately. On others, it does nothing. + Override so that if the platform requires GUI activity to be restricted to a given thread, we ony force update if on the right thread." + + (self ioEventThreadAffinity > 0 + and: [self ioEventThreadAffinity ~= cogThreadManager getVMOwner]) ifFalse: + [self ioForceDisplayUpdate]! Item was added: + ----- Method: CoInterpreterMT>>primitiveShowDisplayRect (in category 'I/O primitives') ----- + primitiveShowDisplayRect + "Force the given rectangular section of the Display to be copied to the screen. + On some platforms, this primitive forces enqueued display updates to be processed immediately. On others, it does nothing. + Override so that if the platform requires GUI activity to be restricted to a given thread, we ony force update if on the right thread." + | bottom top right left | + bottom := self stackIntegerValue: 0. + top := self stackIntegerValue: 1. + right := self stackIntegerValue: 2. + left := self stackIntegerValue: 3. + self successful ifTrue: + [self updateDisplayLeft: left Top: top Right: right Bottom: bottom. + (self ioEventThreadAffinity > 0 + and: [self ioEventThreadAffinity ~= cogThreadManager getVMOwner]) ifFalse: + [self ioForceDisplayUpdate]. + self methodReturnReceiver]! Item was changed: ----- Method: InterpreterPrimitives>>primitiveShowDisplayRect (in category 'I/O primitives') ----- primitiveShowDisplayRect "Force the given rectangular section of the Display to be copied to the screen." | bottom top right left | bottom := self stackIntegerValue: 0. top := self stackIntegerValue: 1. right := self stackIntegerValue: 2. left := self stackIntegerValue: 3. - self updateDisplayLeft: left Top: top Right: right Bottom: bottom. self successful ifTrue: + [self updateDisplayLeft: left Top: top Right: right Bottom: bottom. + self ioForceDisplayUpdate. + self methodReturnReceiver]! - [self ioForceDisplayUpdate. - self pop: 4]!

1 0

Issue on generating VM
by Nicolas Sarfati 19 Apr '24

19 Apr '24

Hi everyone! My name is Nicolas Sarfati and I am working on my University Final work. For it, I am needing to generate an Image of the VM, but I am not getting lucky. On last step, when it converts the C code to the proper image, it's failing with the next error message: *(Attached you can get the full log)* clang -Wall -g -O2 -DNDEBUG -DDEBUGVM=0 -msse2 -DCOGMTVM=0 -pthread -DLSB_FIRST=1 -m64 -Wno-missing-braces -Wno-unknown-pragmas -Wno-unused-value -Wno-unused-label -Wno-unused-function -Wno-unused-variable -Wno-unused-but-set-variable -DHAVE_CONFIG_H -DSQUEAK_BUILTIN_PLUGIN -I/home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/building/linux64x64/squeak.cog.spur/build -I/home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/building/linux64x64/squeak.cog.spur/build -I/home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/unix/vm -I/home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/Cross/vm -I/home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/src/spur64.cog -I/usr/local/include -I/home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/Cross/vm -I/home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/unix/vm -I/home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/src/spur64.cog -I/home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/Cross/plugins/FilePlugin -I/home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/unix/plugins/B3DAcceleratorPlugin -m64 -Wno-missing-braces -Wno-unknown-pragmas -Wno-unused-value -Wno-unused-label -Wno-unused-function -Wno-unused-variable -Wno-unused-but-set-variable -c -o sqUnixMain.o /home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/unix/vm/sqUnixMain.c /home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/unix/vm/sqUnixMain.c:1832:10: warning: format string is not a string literal (potentially insecure) [-Wformat-security] printf(optionStrings[count]); ^~~~~~~~~~~~~~~~~~~~ /home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/unix/vm/sqUnixMain.c:1832:10: note: treat the string as an argument to avoid this printf(optionStrings[count]); ^ "%s", /home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/unix/vm/sqUnixMain.c:2324:38: error: incompatible function pointer types assigning to 'void (*)(int, siginfo_t *, void *)' from 'void (int, siginfo_t *, ucontext_t *)' (aka 'void (int, siginfo_t *, struct ucontext_t *)') [-Wincompatible-function-pointer-types] sigsegv_handler_action.sa_sigaction = sigsegv; ^ ~~~~~~~ /home/nicolas/fcen/tesis/04-04-24/opensmalltalk-vm/platforms/unix/vm/sqUnixMain.c:2331:38: error: incompatible function pointer types assigning to 'void (*)(int, siginfo_t *, void *)' from 'void (int, siginfo_t *, ucontext_t *)' (aka 'void (int, siginfo_t *, struct ucontext_t *)') [-Wincompatible-function-pointer-types] sigusr1_handler_action.sa_sigaction = sigusr1; ^ ~~~~~~~ 1 warning and 2 errors generated. make[1]: *** [Makefile:299: sqUnixMain.o] Error 1 make: *** [Makefile:413: vm/vm.a] Error 2 *Environment: Linux x64 Manjaro* *Kernel: 6.6.19-1-MANJARO* *These are the steps that I am doing:* 1. git clone git@github.com:OpenSmalltalk/opensmalltalk-vm.git 2. I am running the next command: *./scripts/updateSCCSVersions* 3. cd image 4. *./buildspurtrunkvmmaker64image.sh* || also tried with an specific version like these: *./buildspurtrunkvmmaker64image.sh -vm Squeak6.1alpha-23007-64bit* 5. *./buildspurtrunkreader64image.sh* 6. I am running the Squeak with the VMMaker image: *sqcogspur64linuxht/squeak Spur64VMMaker.image* 7. Then inside Squeak, I run some of the commands in the worksapce: 1. [VMMaker generateAllConfigurationsUnderVersionControl] valueSupplyingAnswer: false. 2. [VMMaker generateSqueakSpurCog64VM] valueSupplyingAnswer: false. 3. VMMaker generateVMPlugins. 8. After it finish, I go to the directory: cd building/linux64x64/squeak.cog.spur/build 9. An run the command* ./mvm* 1. Is asks if I want to make a clean build, which I answer Y 2. Then the configure commands outputs all right 3. But then the build breaks with the message which I paste before. 10. Attached you can get the full log. Any input will be really appreciated, since I am really stuck!! Thanks! Nicolas.

2 4

VM Maker: VMMaker.oscog-eem.3367.mcz
by commits＠source.squeak.org 19 Apr '24

19 Apr '24

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.3367.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.3367 Author: eem Time: 18 April 2024, 7:51:44.62085 pm UUID: a7ed57b3-cc68-438c-833d-9b33287ad4ab Ancestors: VMMaker.oscog-eem.3366 Spur Cogit: Improve decoration of the rClass load when selector indices are used. Decode a linked send's class index to its class. =============== Diff against VMMaker.oscog-eem.3366 =============== Item was added: + ----- Method: CogAbstractInstruction>>cacheTagLoadInstructionBytes (in category 'accessing') ----- + cacheTagLoadInstructionBytes + "This is a default, overridden in a few choice cases." + ^self literalLoadInstructionBytes! Item was added: + ----- Method: CogObjectRepresentation>>lookupMaybeInlineCacheTag: (in category 'disassembly') ----- + lookupMaybeInlineCacheTag: anInteger + <doNotGenerate> + ^cogit lookupAddress: anInteger! Item was added: + ----- Method: CogObjectRepresentationForSpur>>lookupMaybeInlineCacheTag: (in category 'disassembly') ----- + lookupMaybeInlineCacheTag: anInteger + <doNotGenerate> + | classOrNil | + ^((objectMemory isValidClassIndex: anInteger) + and: [(classOrNil := objectMemory classOrNilAtIndex: anInteger) ~= objectMemory nilObject + and: [coInterpreter objCouldBeClassObj: classOrNil]]) + ifTrue: [coInterpreter nameOfClass: classOrNil] + ifFalse: [cogit lookupAddress: anInteger]! Item was added: + ----- Method: CogX64Compiler>>cacheTagLoadInstructionBytes (in category 'accessing') ----- + cacheTagLoadInstructionBytes + "The inline cache always uses a 32-bit load, which is always 7 bytes" + ^7! Item was changed: ----- Method: Cogit>>lookupAddress: (in category 'disassembly') ----- lookupAddress: address <doNotGenerate> + address isInteger ifFalse: + [^nil]. address < methodZone freeStart ifTrue: [^address >= methodZoneBase ifTrue: [(methodZone methodFor: address) ifNotNil: [:cogMethod| (self methodSelectorStringFor: cogMethod), '@', ((address - cogMethod asInteger) hex allButFirst: 3)]] ifFalse: [(self trampolineRangeFor: address) ifNotNil: [:range| (self codeEntryNameFor: range first) ifNotNil: [:name| name, (address = range first ifTrue: [''] ifFalse: [' + ', (address - range first) hex])]]]]. (simulatedTrampolines includesKey: address) ifTrue: [^self labelForSimulationAccessor: (simulatedTrampolines at: address)]. (simulatedVariableGetters includesKey: address) ifTrue: ["In methods varBaseAddress is typically stackLimitFromMachineCode, but in trampolines it is typically varBaseAddress..." (varBaseAddress isInteger and: [address = varBaseAddress and: [disassemblingMethod isNil]]) ifTrue: [^'VarBaseAddress']. ^self labelForSimulationAccessor: (simulatedVariableGetters at: address)]. ^(coInterpreter lookupAddress: address) ifNil: [address = self cStackPointerAddress ifTrue: [#CStackPointer] ifFalse: [address = self cFramePointerAddress ifTrue: [#CFramePointer]]]! Item was added: + ----- Method: Cogit>>lookupAddress:at:couldBeRClassLoad: (in category 'disassembly') ----- + lookupAddress: address at: instructionAddress couldBeRClassLoad: couldBeRClassLoad + <doNotGenerate> + "This is to allow us to informatively decorate selector literal index/class index loads in inline caches in Spur. + If the instruction is follwed by a call instruction then we can tell if it is linked by the address or not." + | offset | + offset := backEnd cacheTagLoadInstructionBytes + backEnd callInstructionByteSize. + (couldBeRClassLoad + and: [instructionAddress between: methodZoneBase and: methodZone freeStart - 1 - offset]) ifTrue: + [(methodZone methodFor: instructionAddress) ifNotNil: + [:cogMethod| | target selector | + (cogMethod cmType >= CMBlock + and: [cogMethod stackCheckOffset > 0 + and: [instructionAddress >= (cogMethod asInteger + cogMethod stackCheckOffset) + and: [backEnd isCallPrecedingReturnPC: instructionAddress + offset]]]) ifTrue: + [address isString ifTrue: + [| addressValue signExtendedAddress | + self assert: (address beginsWith: '0x'). + addressValue := Number readFrom: (address allButFirst: 2) base: 16. + signExtendedAddress := addressValue > (1 << (8 * objectRepresentation wordSize - 1) - 1) + ifTrue: [addressValue - (1 << (8 * objectRepresentation wordSize))] + ifFalse: [addressValue]. + ^self + lookupAddress: signExtendedAddress + at: instructionAddress + couldBeRClassLoad: couldBeRClassLoad]. + target := backEnd callTargetFromReturnAddress: instructionAddress + offset. + target > methodZoneBase ifTrue: "i.e. a linked send" + [^objectRepresentation lookupMaybeInlineCacheTag: address]. + self inlineCacheTagsAreIndexes ifTrue: + [selector := address < 0 + ifTrue: [(coInterpreter specialSelector: -1 - address)] + ifFalse: [coInterpreter literal: address ofMethod: cogMethod methodObject]. + ^self lookupAddress: selector]]]]. + ^self lookupAddress: address! Item was changed: ----- Method: Cogit>>lookupCHexString: (in category 'disassembly') ----- lookupCHexString: aCHexString <doNotGenerate> + | shortened address | - | pastLastZero shortened address | (aCHexString beginsWith: '0x') ifFalse: [^aCHexString]. + shortened := self shortenCHexString: aCHexString. - pastLastZero := aCHexString findFirst: [:c| c ~= $0 and: [c ~= $x]]. - shortened := pastLastZero = 0 - ifTrue: ['0x0'] - ifFalse: - [(aCHexString size >= 16 and: [pastLastZero >= 4]) - ifTrue: [aCHexString copyReplaceFrom: 3 to: pastLastZero - 1 with: ''] - ifFalse: [aCHexString]]. address := Number readFrom: (ReadStream on: shortened from: 3 to: shortened size) base: 16. (disassemblingMethod notNil and: [address > disassemblingMethod and: [address < (disassemblingMethod asInteger + disassemblingMethod blockSize)]]) ifTrue: [shortened := '.+', (address - disassemblingMethod asInteger printStringBase: 16 length: 4 padded: true)]. ^(self lookupAddress: (Number readFrom: (ReadStream on: shortened from: 3 to: shortened size) base: 16)) ifNotNil: [:string| shortened, '=', string] ifNil: [shortened]! Item was added: + ----- Method: Cogit>>shortenCHexString: (in category 'disassembly') ----- + shortenCHexString: aCHexString + <doNotGenerate> + | pastLastZero | + (aCHexString beginsWith: '0x') ifFalse: + [^aCHexString]. + pastLastZero := aCHexString findFirst: [:c| c ~= $0 and: [c ~= $x]]. + ^pastLastZero = 0 + ifTrue: ['0x0'] + ifFalse: + [(aCHexString size >= 16 and: [pastLastZero >= 4]) + ifTrue: [aCHexString copyReplaceFrom: 3 to: pastLastZero - 1 with: ''] + ifFalse: [aCHexString]]! Item was added: + ----- Method: CurrentImageCoInterpreterFacadeForSpurObjectRepresentation>>isValidClassIndex: (in category 'disassembly') ----- + isValidClassIndex: anInteger + (anInteger between: 1 and: (2 raisedTo: 22) - 1) ifTrue: + [Smalltalk allClassesDo: + [:c| + ((c hasIdentityHash and: [c identityHash = anInteger]) + or: [c class identityHash = anInteger]) ifTrue: + [^true]]]. + ^false!

1 0

[OpenSmalltalk/opensmalltalk-vm] use --whole-archive option instead of -whole-archive for solaris ld (PR #673)
by dcstes 19 Apr '24

19 Apr '24

I believe that the GNU link editor supports both single hyphen and double hyphen options -whole-archive and --whole-archive However the SUN (Solaris and SunOS and Illumos) link editor uses different names like "-z weakextract" Three is a GNU compatibility layer but the SUN link editor insists for GNU compatibility on the double hyphen format -whole-archive and --no-whole-archive David Stes You can view, comment on, or merge this pull request online at: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/673 -- Commit Summary -- * use --whole-archive option instead of -whole-archive for solaris ld -- File Changes -- M platforms/unix/config/Makefile.in (2) -- Patch Links -- https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/673.patch https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/673.diff -- Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/673 You are receiving this because you are subscribed to this thread. Message ID: <OpenSmalltalk/opensmalltalk-vm/pull/673(a)github.com>

2 4

[OpenSmalltalk/opensmalltalk-vm] af5f4f: use --whole-archive option instead of -whole-archi...
by David T Lewis 19 Apr '24

19 Apr '24

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: af5f4f883b6aca4369b52dc451f4be977f3c6967 https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/af5f4f883b6aca4369… Author: David Stes <stes(a)telenet.be> Date: 2024-01-06 (Sat, 06 Jan 2024) Changed paths: M platforms/unix/config/Makefile.in Log Message: ----------- use --whole-archive option instead of -whole-archive for solaris ld Commit: 5473cde3bc0d3b97679633d6333fd9ce40a58d81 https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/5473cde3bc0d3b9767… Author: David T Lewis <dtlewis290(a)gmail.com> Date: 2024-04-18 (Thu, 18 Apr 2024) Changed paths: M platforms/unix/config/Makefile.in Log Message: ----------- Merge pull request #673 from dcstes/whole-archive use --whole-archive option instead of -whole-archive for solaris ld Compare: https://github.com/OpenSmalltalk/opensmalltalk-vm/compare/d926670ee265...54… To unsubscribe from these emails, change your notification settings at https://github.com/OpenSmalltalk/opensmalltalk-vm/settings/notifications

1 0

VM Maker: VMMaker.oscog-eem.3366.mcz
by commits＠source.squeak.org 18 Apr '24

18 Apr '24

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.3366.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.3366 Author: eem Time: 18 April 2024, 11:18:57.12491 am UUID: 791f24bd-c2cf-4601-b1a7-3fae30ee6c9b Ancestors: VMMaker.oscog-eem.3365 More concise implementation of FloatMathPlugin>>#primitiveLog10 (& nicer simulation). Simulation fixes (from simulating a Cuis 5.0 image). a) fix the simulation of setting primitiveDoMixedArithmetic & upscaleDisplayIfHighDPI, which can now be overridden on the command line. b) provide a default implementation of ioEventThreadAffinity for the threaded VM c) fix disassembly of jitted V3 methods d) fix simulation of the log: (log: 3) method; it forgot to return a result e) fix simulation of the JPEGReadWriter2Plugin, and the infinite recursion in its #sizeof: method (!!) =============== Diff against VMMaker.oscog-eem.3365 =============== Item was changed: ----- Method: CoInterpreter>>setImageHeaderFlagsFrom: (in category 'image save/restore') ----- setImageHeaderFlagsFrom: headerFlags "Set the flags that are contained in the 7th long of the image header. This is sent at image load time." imageHeaderFlags := headerFlags. "so as to preserve unrecognised flags." fullScreenFlag := headerFlags bitAnd: 1. imageFloatsBigEndian := (headerFlags noMask: 2) ifTrue: [1] ifFalse: [0]. "processHasThreadAffinity := headerFlags anyMask: 4. specific to CoInterpreterMT" flagInterpretedMethods := headerFlags anyMask: 8. preemptionYields := headerFlags noMask: 16. "noThreadingOfGUIThread := headerFlags anyMask: 32. specific to CoInterpreterMT" newFinalization := headerFlags anyMask: 64. sendWheelEvents := headerFlags anyMask: 128. + (self cCode: [primitiveDoMixedArithmetic < 0] + inSmalltalk: [primitiveDoMixedArithmetic isInteger and: [primitiveDoMixedArithmetic < 0]]) ifTrue: "i.e. has it not been set on the command line?" - primitiveDoMixedArithmetic < 0 ifTrue: "i.e. has it not been set on the command line?" [primitiveDoMixedArithmetic := headerFlags noMask: 256]. "N.B. flag mask 512 is responded to by the FilePlugin & FileAttributesPlugin" + (self cCode: [upscaleDisplayIfHighDPI < 0] + inSmalltalk: [upscaleDisplayIfHighDPI isInteger and: [upscaleDisplayIfHighDPI < 0]]) ifTrue: "i.e. has it not been set on the command line?" - upscaleDisplayIfHighDPI < 0 ifTrue: "i.e. has it not been set on the command line?" [upscaleDisplayIfHighDPI := headerFlags noMask: 1024]! Item was added: + ----- Method: CoInterpreterMT>>ioEventThreadAffinity (in category 'threading support') ----- + ioEventThreadAffinity + "Default" + ^-1! Item was changed: ----- Method: Cogit>>methodSelectorStringFor: (in category 'disassembly') ----- methodSelectorStringFor: cogMethod "Answer either the hex string or the selector for a method, if it has a valid selector. For full blocks, prepend ``[full address] '' to the selector." <doNotGenerate> + (cogMethod cmType > CMOpenPIC + and: [cogMethod cmIsFullBlock]) ifTrue: - cogMethod cmIsFullBlock ifTrue: [| rawString | rawString := String streamContents: [:s| s nextPutAll: '[full '. cogMethod asInteger printOn: s base: 16. "omits 16r prefix". s nextPut: $]]. (coInterpreter maybeSelectorOfMethod: cogMethod methodObject) ifNotNil: [:selector| (objectRepresentation couldBeObject: selector) ifTrue: [^rawString, ' ', (coInterpreter stringOf: selector)]]. ^rawString]. cogMethod selector = objectMemory nilObject ifTrue: [^cogMethod asInteger hex,'(nil sel)']. (objectRepresentation couldBeObject: cogMethod selector) ifTrue: [^coInterpreter stringOf: cogMethod selector]. ^cogMethod asInteger hex! Item was changed: ----- Method: FloatMathPlugin>>primitiveLog10 (in category 'float primitives') ----- primitiveLog10 "Computes log10(receiver)" <export: true> <primitiveMetadata: #(FastCPrimitiveFlag FastCPrimitiveAlignForFloatsFlag)> | rcvr result | <var: #rcvr type: #double> <var: #result type: #double> rcvr := interpreterProxy stackFloatValue: 0. + interpreterProxy failed ifFalse: + [rcvr >= 0.0 ifTrue: + [result := self __ieee754_log10: rcvr. + (self isnan: result) ifFalse: + [^interpreterProxy methodReturnFloat: result]]]. + ^interpreterProxy primitiveFail! - interpreterProxy failed ifTrue:[^nil]. - rcvr < 0.0 ifTrue:[^interpreterProxy primitiveFail]. - result := self cCode: '__ieee754_log10(rcvr)' inSmalltalk: [rcvr log: 10]. - (self isnan: result) ifTrue:[^interpreterProxy primitiveFail]. - ^interpreterProxy methodReturnFloat: result! Item was added: + ----- Method: FloatMathPluginSimulator>>__ieee754_log10: (in category 'float primitives') ----- + __ieee754_log10: aFloat + ^aFloat log: 10! Item was changed: ----- Method: Integer>>coerceTo:sim: (in category '*VMMaker-interpreter simulator') ----- coerceTo: cTypeString sim: interpreter | bits unitSize | cTypeString last == $* ifTrue: "C pointer" [unitSize := cTypeString caseOf: { [#'char *'] -> [1]. [#'short *'] -> [2]. [#'int *'] -> [4]. [#'long long *'] -> [8]. [#'float *'] -> [^CFloatArray basicNew interpreter: interpreter address: self unitSize: 4; yourself]. [#'double *'] -> [^CFloatArray basicNew interpreter: interpreter address: self unitSize: 8; yourself]. [#'unsigned *'] -> [4]. [#'unsigned int *'] -> [4]. [#'unsigned char *'] -> [1]. [#'signed char *'] -> [1]. [#'unsigned short *'] -> [2]. [#'unsigned long long *'] -> [8]. [#'oop *'] -> [interpreter objectMemory bytesPerOop]. } otherwise: [interpreter objectMemory wordSize]. ^CArray basicNew interpreter: interpreter address: self unitSize: unitSize; yourself]. cTypeString first == $u ifTrue: [bits := cTypeString caseOf: { [#usqInt] -> [interpreter objectMemory wordSize * 8]. [#usqLong] -> [64]. [#unsigned] -> [32]. [#'unsigned char'] -> [8]. [#'unsigned int'] -> [8]. [#'unsigned long'] -> [48]. "LLP64 on Windows :-(" [#'unsigned long long'] -> [64]. [#'unsigned short'] -> [16]. } otherwise: [self error: 'unknown unsigned integer type name']. ^self bitAnd: 1 << bits - 1]. bits := cTypeString caseOf: { [#sqInt] -> [interpreter objectMemory wordSize * 8]. [#'sqIntptr_t'] -> [interpreter objectMemory wordSize * 8]. [#sqLong] -> [64]. [#char] -> [^self bitAnd: 255]. "char may be signed, may be unsigned; interpret as unsigned by default" [#'signed char'] -> [8]. [#'short'] -> [16]. [#int] -> [32]. [#long] -> [48]. "LLP64 on Windows :-(" [#'long long'] -> [64]. [#'wint_t'] -> [24]. "unsigned short on Windows; int elsewhere" [#double] -> [^self asFloat]. } + otherwise: + ["exceptions for unusual cases..." + cTypeString == #'j_decompress_ptr' ifTrue: + [^self coerceTo: 'j_decompress *' sim: interpreter]. + "abort..." + self error: 'unknown signed integer type name']. - otherwise: [self error: 'unknown signed integer type name']. ^(self bitAnd: (1 bitShift: bits) - 1) - ((self bitAnd: (1 bitShift: bits - 1)) bitShift: 1)! Item was changed: ----- Method: InterpreterPrimitives>>log: (in category 'primitive support') ----- log: aFloat "Simulate the C library log:(3) function." <doNotGenerate> + ^aFloat = aFloat - aFloat = aFloat ifTrue: [aFloat ln] ifFalse: [Float nan]! Item was changed: ----- Method: JPEGReadWriter2Plugin>>sizeof: (in category 'simulation') ----- sizeof: aSymbol <doNotGenerate> ^aSymbol caseOf: { + [#'struct jpeg_compress_struct'] -> [JPEGReadWriter2 new primJPEGCompressStructSize]. + [#'struct jpeg_decompress_struct'] -> [JPEGReadWriter2 new primJPEGDecompressStructSize]. + [#'struct error_mgr2'] -> [JPEGReadWriter2 new primJPEGErrorMgr2StructSize] } - [self sizeof: #'struct jpeg_compress_struct'] -> [JPEGReadWriter2 new primJPEGCompressStructSize]. - [self sizeof: #'struct jpeg_decompress_struct'] -> [JPEGReadWriter2 new primJPEGDecompressStructSize]. - [self sizeof: #'struct error_mgr2'] -> [JPEGReadWriter2 new primJPEGErrorMgr2StructSize] } otherwise: [super sizeof: aSymbol]!

1 0

VM Maker: VMMaker.oscog-eem.3365.mcz
by commits＠source.squeak.org 18 Apr '24

18 Apr '24

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.3365.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.3365 Author: eem Time: 17 April 2024, 4:17:09.846667 pm UUID: d067e715-4de3-4931-8c6e-5884b5fbab91 Ancestors: VMMaker.oscog-eem.3364 First pass at getting threaded VM to work on Mac. Only call ioProcessEvents if teh platform doesn't care (doesn't define ioEventThreadAffinity, or defines it as 0) or the current thread id is the same as ioEventThreadAffinity. Nuke vmOwnerAddress & vmOwnerFromMachineCode which were used only by the pre-C-atomics access code to generate the machine code accessors for vmOwner. =============== Diff against VMMaker.oscog-eem.3364 =============== Item was changed: ----- Method: CoInterpreterMT>>checkForEventsMayContextSwitch: (in category 'process primitive support') ----- checkForEventsMayContextSwitch: mayContextSwitch "Check for possible interrupts and handle one if necessary. Answer if a context switch has occurred." + | switched sema now threadingThwartsIOProcessEventsInvocation | - | switched sema now | <inline: false> <var: #now type: #usqLong> self assertSaneThreadAndProcess. cogit assertCStackWellAligned. statCheckForEvents := statCheckForEvents + 1. "restore the stackLimit if it has been smashed." self restoreStackLimit. self externalWriteBackHeadFramePointers. self assert: stackPage = stackPages mostRecentlyUsedPage. "Allow the platform to do anything it needs to do synchronously." self ioSynchronousCheckForEvents. self checkCogCompiledCodeCompactionCalledFor. objectMemory needGCFlag ifTrue: ["sufficientSpaceAfterGC: runs the incremental GC and then, if not enough space is available, the fullGC." (objectMemory sufficientSpaceAfterGC: 0) ifFalse: [self setSignalLowSpaceFlagAndSaveProcess]]. mayContextSwitch ifFalse: [^false]. switched := false. self assert: deferThreadSwitch not. deferThreadSwitch := true. (profileProcess ~= objectMemory nilObject or: [nextProfileTick > 0 and:[self ioHighResClock >= nextProfileTick]]) ifTrue: [self zeroNextProfileTick. "Take a sample (if not already done so) for the profiler if it is active. This must be done before any of the synchronousSignals below or else we will attribute a pause in ioRelinquishProcessor to the newly activated process." profileProcess = objectMemory nilObject ifTrue: [profileProcess := self activeProcess. profileMethod := objectMemory nilObject]. "and signal the profiler semaphore if it is present" (profileSemaphore ~= objectMemory nilObject and: [self synchronousSignal: profileSemaphore]) ifTrue: [switched := true]]. self cppIf: #LRPCheck ifTrue: [self checkDeliveryOfLongRunningPrimitiveSignal ifTrue: [switched := true]]. objectMemory signalLowSpace ifTrue: [objectMemory signalLowSpace: false. "reset flag" sema := objectMemory splObj: TheLowSpaceSemaphore. (sema ~= objectMemory nilObject and: [self synchronousSignal: sema]) ifTrue: [switched := true]]. + threadingThwartsIOProcessEventsInvocation := self checkInvokeIOProcessEvents: (now := self ioUTCMicroseconds). - "inIOProcessEvents prevents reentrancy into ioProcessEvents and allows disabling - ioProcessEvents e.g. for native GUIs. We would like to manage that here but can't - since the platform code may choose to call ioProcessEvents itself in various places." - false - ifTrue: - [((now := self ioUTCMicroseconds) >= nextPollUsecs - and: [inIOProcessEvents = 0]) ifTrue: - [statIOProcessEvents := statIOProcessEvents + 1. - inIOProcessEvents := inIOProcessEvents + 1. - self ioProcessEvents. "sets interruptPending if interrupt key pressed; may callback" - inIOProcessEvents > 0 ifTrue: - [inIOProcessEvents := inIOProcessEvents - 1]. - nextPollUsecs := now + 20000 - "msecs to wait before next call to ioProcessEvents. Note that strictly - speaking we might need to update 'now' at this point since - ioProcessEvents could take a very long time on some platforms"]] - ifFalse: - [(now := self ioUTCMicroseconds) >= nextPollUsecs ifTrue: - [statIOProcessEvents := statIOProcessEvents + 1. - self ioProcessEvents. "sets interruptPending if interrupt key pressed; may callback" - nextPollUsecs := now + 20000 - "msecs to wait before next call to ioProcessEvents. Note that strictly - speaking we might need to update 'now' at this point since - ioProcessEvents could take a very long time on some platforms"]]. interruptPending ifTrue: [interruptPending := false. "reset interrupt flag" sema := objectMemory splObj: TheInterruptSemaphore. (sema ~= objectMemory nilObject and: [self synchronousSignal: sema]) ifTrue: [switched := true]]. nextWakeupUsecs ~= 0 ifTrue: [now >= nextWakeupUsecs ifTrue: [nextWakeupUsecs := 0. "set timer interrupt to 0 for 'no timer'" sema := objectMemory splObj: TheTimerSemaphore. (sema ~= objectMemory nilObject and: [self synchronousSignal: sema]) ifTrue: [switched := true]]]. "signal any pending finalizations" pendingFinalizationSignals > 0 ifTrue: [pendingFinalizationSignals := 0. sema := objectMemory splObj: TheFinalizationSemaphore. (sema ~= objectMemory nilObject and: [self synchronousSignal: sema]) ifTrue: [switched := true]]. "signal all semaphores in semaphoresToSignal" self signalExternalSemaphores ifTrue: [switched := true]. deferThreadSwitch := false. + (threadingThwartsIOProcessEventsInvocation + or: [checkThreadActivation]) ifTrue: - checkThreadActivation ifTrue: [checkThreadActivation := false. self cedeToHigherPriorityThreads]. "N.B. This may not return if we do switch." self threadSwitchIfNecessary: self activeProcess from: CSCheckEvents. ^switched! Item was added: + ----- Method: CoInterpreterMT>>checkInvokeIOProcessEvents: (in category 'process primitive support') ----- + checkInvokeIOProcessEvents: now + "Some platforms allow event resaponse only from a subset of threads (e.g. on Mac, the main application thread). + Override to check if the platform allows us to invoke or not and answer true if ioProcessEvents should have been + invoked but wasn't because of thread affinity." + <inline: #always> + (self ioEventThreadAffinity > 0 + and: [self ioEventThreadAffinity ~= cogThreadManager getVMOwner]) ifTrue: + [^now >= nextPollUsecs]. + super checkInvokeIOProcessEvents: now. + ^false! Item was removed: - ----- Method: CoInterpreterMT>>vmOwnerAddress (in category 'simulation') ----- - vmOwnerAddress - <doNotGenerate> - ^cogThreadManager vmOwnerAddress! Item was removed: - ----- Method: CogThreadManager>>vmOwnerAddress (in category 'Cogit lock implementation') ----- - vmOwnerAddress - <api> "NB. For the JIT only, so it can generate the lock & unlock functions." - <returnTypeC: #usqInt> - self error: 'Deprecated!! Replaced by proper atomic functions'. - ^self - cCode: [(self addressOf: vmOwner) asUnsignedInteger] - inSmalltalk: [cogit simulatedReadWriteVariableAddress: #vmOwnerFromMachineCode in: self]! Item was removed: - ----- Method: CogThreadManager>>vmOwnerFromMachineCode (in category 'Cogit lock implementation') ----- - vmOwnerFromMachineCode - <doNotGenerate> - self error: 'Deprecated!! Replaced by proper atomic C functions'. - ^vmOwner! Item was removed: - ----- Method: CogThreadManager>>vmOwnerFromMachineCode: (in category 'Cogit lock implementation') ----- - vmOwnerFromMachineCode: aValue - <doNotGenerate> - self error: 'Deprecated!! Replaced by proper atomic C functions'. - self assert: (aValue between: 0 and: numThreads). - vmOwner := aValue! Item was changed: ----- Method: StackInterpreter>>checkForEventsMayContextSwitch: (in category 'process primitive support') ----- checkForEventsMayContextSwitch: mayContextSwitch "Check for possible interrupts and handle one if necessary. Answer if a context switch has occurred." | switched sema now | <inline: false> <var: #now type: #usqLong> statCheckForEvents := statCheckForEvents + 1. "restore the stackLimit if it has been smashed." self restoreStackLimit. self externalWriteBackHeadFramePointers. self assert: stackPage = stackPages mostRecentlyUsedPage. "Allow the platform to do anything it needs to do synchronously." self ioSynchronousCheckForEvents. self checkCogCompiledCodeCompactionCalledFor. objectMemory needGCFlag ifTrue: ["sufficientSpaceAfterGC: runs the incremental GC and then, if not enough space is available, the fullGC." (objectMemory sufficientSpaceAfterGC: 0) ifFalse: [self setSignalLowSpaceFlagAndSaveProcess]]. mayContextSwitch ifFalse: [^false]. switched := false. (profileProcess ~= objectMemory nilObject or: [nextProfileTick > 0 and:[self ioHighResClock >= nextProfileTick]]) ifTrue: [self zeroNextProfileTick. "Take a sample (if not already done so) for the profiler if it is active. This must be done before any of the synchronousSignals below or else we will attribute a pause in ioRelinquishProcessor to the newly activated process." profileProcess = objectMemory nilObject ifTrue: [profileProcess := self activeProcess. profileMethod := objectMemory nilObject]. "and signal the profiler semaphore if it is present" (profileSemaphore ~= objectMemory nilObject and: [self synchronousSignal: profileSemaphore]) ifTrue: [switched := true]]. self cppIf: #LRPCheck ifTrue: [self checkDeliveryOfLongRunningPrimitiveSignal ifTrue: [switched := true]]. objectMemory signalLowSpace ifTrue: [objectMemory signalLowSpace: false. "reset flag" sema := objectMemory splObj: TheLowSpaceSemaphore. (sema ~= objectMemory nilObject and: [self synchronousSignal: sema]) ifTrue: [switched := true]]. + self checkInvokeIOProcessEvents: (now := self ioUTCMicroseconds). - "inIOProcessEvents prevents reentrancy into ioProcessEvents and allows disabling - ioProcessEvents e.g. for native GUIs. We would like to manage that here but can't - since the platform code may choose to call ioProcessEvents itself in various places." - false - ifTrue: - [((now := self ioUTCMicroseconds) >= nextPollUsecs - and: [inIOProcessEvents = 0]) ifTrue: - [statIOProcessEvents := statIOProcessEvents + 1. - inIOProcessEvents := inIOProcessEvents + 1. - self ioProcessEvents. "sets interruptPending if interrupt key pressed; may callback" - inIOProcessEvents > 0 ifTrue: - [inIOProcessEvents := inIOProcessEvents - 1]. - nextPollUsecs := now + 20000 - "msecs to wait before next call to ioProcessEvents. Note that strictly - speaking we might need to update 'now' at this point since - ioProcessEvents could take a very long time on some platforms"]] - ifFalse: - [(now := self ioUTCMicroseconds) >= nextPollUsecs ifTrue: - [statIOProcessEvents := statIOProcessEvents + 1. - self ioProcessEvents. "sets interruptPending if interrupt key pressed; may callback" - nextPollUsecs := now + 20000 - "msecs to wait before next call to ioProcessEvents. Note that strictly - speaking we might need to update 'now' at this point since - ioProcessEvents could take a very long time on some platforms"]]. interruptPending ifTrue: [interruptPending := false. "reset interrupt flag" sema := objectMemory splObj: TheInterruptSemaphore. (sema ~= objectMemory nilObject and: [self synchronousSignal: sema]) ifTrue: [switched := true]]. nextWakeupUsecs ~= 0 ifTrue: [now >= nextWakeupUsecs ifTrue: [nextWakeupUsecs := 0. "set timer interrupt to 0 for 'no timer'" sema := objectMemory splObj: TheTimerSemaphore. (sema ~= objectMemory nilObject and: [self synchronousSignal: sema]) ifTrue: [switched := true]]]. "signal any pending finalizations" pendingFinalizationSignals > 0 ifTrue: [pendingFinalizationSignals := 0. sema := objectMemory splObj: TheFinalizationSemaphore. (sema ~= objectMemory nilObject and: [self synchronousSignal: sema]) ifTrue: [switched := true]]. "signal all semaphores in semaphoresToSignal" self signalExternalSemaphores ifTrue: [switched := true]. ^switched! Item was added: + ----- Method: StackInterpreter>>checkInvokeIOProcessEvents: (in category 'process primitive support') ----- + checkInvokeIOProcessEvents: now + "inIOProcessEvents prevents reentrancy into ioProcessEvents and allows disabling + ioProcessEvents e.g. for native GUIs. We would like to manage that here but can't + since the platform code may choose to call ioProcessEvents itself in various places." + <inline: #always> + false + ifTrue: + [(now >= nextPollUsecs + and: [inIOProcessEvents = 0]) ifTrue: + [statIOProcessEvents := statIOProcessEvents + 1. + inIOProcessEvents := inIOProcessEvents + 1. + self ioProcessEvents. "sets interruptPending if interrupt key pressed; may callback" + inIOProcessEvents > 0 ifTrue: + [inIOProcessEvents := inIOProcessEvents - 1]. + nextPollUsecs := now + 20000 + "msecs to wait before next call to ioProcessEvents. Note that strictly + speaking we might need to update 'now' at this point since + ioProcessEvents could take a very long time on some platforms"]] + ifFalse: + [now >= nextPollUsecs ifTrue: + [statIOProcessEvents := statIOProcessEvents + 1. + self ioProcessEvents. "sets interruptPending if interrupt key pressed; may callback" + nextPollUsecs := now + 20000 + "msecs to wait before next call to ioProcessEvents. Note that strictly + speaking we might need to update 'now' at this point since + ioProcessEvents could take a very long time on some platforms"]].!

3 2

VM Maker: VMMaker.oscog-eem.3364.mcz
by commits＠source.squeak.org 17 Apr '24

17 Apr '24

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.3364.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.3364 Author: eem Time: 17 April 2024, 4:13:25.775126 pm UUID: c0aef66b-b635-4039-897e-d58fa07a050b Ancestors: VMMaker.oscog-eem.3363 A better fix to the non-inlining of isAlien: in assert: is not to mark isAlien: as <inline: #always> but <inline: true>. That was a mistake on my part. Apologies. =============== Diff against VMMaker.oscog-eem.3363 =============== Item was changed: ----- Method: InterpreterPlugin>>isAlien: (in category 'alien support') ----- isAlien: oop "Answer if oop is an Alien. We could ask if isWordsOrBytes: first, but that doesn't help. We still have to do the is:KindOf: walk. We're not interested in fast falsehood, but as fast as possible truth, and with the current API this is it." + <inline: true> - <inline: #always> ^interpreterProxy is: oop KindOfClass: interpreterProxy classAlien! Item was changed: ----- Method: ThreadedFFIPlugin>>isDirectAlien: (in category 'primitive support') ----- isDirectAlien: oop <inline: #always> + self assert: (self isAlien: oop). - "self assert: (self isAlien: oop). -- nested inlining is ignored for #isAlien:. Expand here." - self assert: (interpreterProxy is: oop KindOfClass: interpreterProxy classAlien). ^(self sizeField: oop) > 0!

1 0

VM Maker: VMMaker.oscog-eem.3363.mcz
by commits＠source.squeak.org 17 Apr '24

17 Apr '24

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.3363.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.3363 Author: eem Time: 17 April 2024, 4:08:09.099496 pm UUID: ef2ede81-25f0-41f4-ab60-14f7bb01111a Ancestors: VMMaker.oscog-mt.3362 A couple of methods should use objectMemory displayObject rather than access TheDisplay directly. =============== Diff against VMMaker.oscog-mt.3362 =============== Item was changed: ----- Method: StackInterpreter>>postGCUpdateDisplayBits (in category 'object memory support') ----- postGCUpdateDisplayBits "Update the displayBits after a GC may have moved it. Answer if the displayBits appear valid. The wrinkle here is that the displayBits could be a surface handle." <inline: false> | displayObj bitsOop bitsNow | + displayObj := objectMemory displayObject. - displayObj := objectMemory splObj: TheDisplay. ((objectMemory isPointers: displayObj) and: [(objectMemory lengthOf: displayObj) >= 4]) ifFalse: [^false]. bitsOop := objectMemory followOopField: 0 ofObject: displayObj. (bitsOop = objectMemory nilObject "it ain't yet set" or: [objectMemory isIntegerObject: bitsOop]) ifTrue: "It's a surface; our work here is done..." [^true]. self assert: ((objectMemory addressCouldBeObj: bitsOop) and: [objectMemory isWordsOrBytes: bitsOop]). (objectMemory hasSpurMemoryManagerAPI and: [objectMemory isPinned: bitsOop]) ifFalse: [(objectMemory hasSpurMemoryManagerAPI and: [stackPage notNil]) ifTrue: "If stackPage is nil we're snapshotting and now is not the time to pin." [objectMemory pinObject: bitsOop. bitsOop := objectMemory followOopField: 0 ofObject: displayObj]. bitsNow := self cCode: [objectMemory firstIndexableField: bitsOop] inSmalltalk: [(objectMemory firstIndexableField: bitsOop) asInteger]. displayBits ~= bitsNow ifTrue: [displayBits := bitsNow. self ioNoteDisplayChanged: displayBits width: displayWidth height: displayHeight depth: displayDepth]]. ^true! Item was changed: ----- Method: StackInterpreter>>showDisplayBits:Left:Top:Right:Bottom: (in category 'I/O primitive support') ----- showDisplayBits: aForm Left: l Top: t Right: r Bottom: b "Repaint the portion of the Smalltalk screen bounded by the affected rectangle. Used to synchronize the screen after a Bitblt to the Smalltalk Display object." (deferDisplayUpdates + or: [aForm ~= objectMemory displayObject]) ifTrue: [^nil]. - or: [aForm ~= (objectMemory splObj: TheDisplay)]) ifTrue: [^nil]. self updateDisplayLeft: l Top: t Right: r Bottom: b!

1 0

[OpenSmalltalk/opensmalltalk-vm] pthread_jit_write_with_callback_np() vs. pthread_jit_write_protect_np() (Issue #682)
by Marcel Taeumel 17 Apr '24

17 Apr '24

Apple suggests `pthread_jit_write_with_callback_np()` and a *JIT callback allow list* via the `com.apple.security.cs.jit-write-allowlist` entitlement as the way to port a JIT to Apple silicon. - https://developer.apple.com/documentation/apple-silicon/porting-just-in-tim… We use `pthread_jit_write_protect_np()` at this point. How much longer is this going to work? Maybe we want to refactor this to the callback approach? - https://keith.github.io/xcode-man-pages/pthread_jit_write_protect_np.3.html… This is just a heads-up. 😺 -- Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/682 You are receiving this because you are subscribed to this thread. Message ID: <OpenSmalltalk/opensmalltalk-vm/issues/682(a)github.com>

2 1

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