Vm-dev August 2022

vm-dev@lists.squeakfoundation.org

11 participants
75 discussions

[OpenSmalltalk/opensmalltalk-vm] e9ec18: Nuke unwanted file
by Eliot Miranda 19 Aug '22

19 Aug '22

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: e9ec18457c587e4e7102337a38ec7d0221d84570 https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/e9ec18457c587e4e71… Author: Eliot Miranda <eliot.miranda(a)gmail.com> Date: 2022-08-19 (Fri, 19 Aug 2022) Changed paths: R building/linux64riscv/RiscV64ForVMMaker.tgz Log Message: ----------- Nuke unwanted file

1 0

Please remove RiscV64ForVMMaker.tgz
by ken.dickey＠whidbey.com 15 Aug '22

15 Aug '22

opensmalltalk-vm/building/linux64riscv/RiscV64ForVMMaker.tgz is no longer needed as code was submitted to AlienInbox. Please remove this file. It is no longer supported. Thanks, -KenD

1 0

Question about trampoline usage
by Angel Yan 15 Aug '22

15 Aug '22

Hi I'm doing some work on the cog vm for which I need to instrument cogged bytecodes to collect runtime information and my plan is to do this via trampolines. The idea is to update the Cogit such that calls to the relevant trampolines are generated before or after the usual machine code that is generated for these bytecodes. As a first test, I added a dummy trampoline that just prints a message to stdout and updated the machine code translation of a bytecode to generate a call to said trampoline at the beginning. For this I: 1. Updated StackToRegisterMappingCogit class>>#numTrampolines. 2. Added a ceDummyTrampoline inst var to StackToRegisterMappingCogit. 3. Updated StackToRegisterMappingCogit to generate the dummy trampoline. StackToRegisterMappingCogit>>#generateRunTimeTrampolines super generateRunTimeTrampolines. ceDummyTrampoline := self genTrampolineFor: #dummyTrampoline: called: 'ceDummyTrampoline' regsToSave: CallerSavedRegisterMask. 4. Added the dummy trampoline method in the CoInterpreter. CoInterpreter>>#dummyTrampoline <api> self cCode: 'printf("dummy trampoline\n");'. 5. Instrumented a bytecode. StackToRegisterMappingCogit>>#genStoreAndPopReceiverVariableBytecode self Call: ceDummyTrampoline. ^super genStoreAndPopReceiverVariableBytecode. 6. Generated and compiled a new vm with VMMaker generateSqueakSpurCog64VM. With the newly compiled vm I'm able to open an image and I can see the message from the trampoline being printed to stdout, however, it quickly crashes when I start doing things in the UI. These are some of the behaviors I've observed: * The vm does not seem to crash as long as I do simple things like moving around the cursor, opening context menus and dragging collapsed windows, however, it quickly -but not always immediately- crashes when I start opening windows. * The chances of a window crashing the vm increases with time. e.g. I'm able to open a browser if I do it as soon as I open the image, but not if I wait 5 seconds. * Some windows are more likely to crash the vm than others. e.g. I'm able to open multiple workspaces and transcripts before the vm crashes, but at most one browser. * The vm outputs a crash dump sometimes, but not always. I've attached to this message some of the crash dumps I've gotten. Any help will be greatly appreciated. I've tried debugging using the simulator and gdb but was not able to get very far as I don't know how to best troubleshoot these types of errors. Suggestions in this regard are welcome as well. Thanks! Regards Angel

2 2

Re: [Vm-dev] [squeak-dev] Problem creating MVCProject in Mac
by Eliot Miranda 15 Aug '22

15 Aug '22

Hi Javier, > On Aug 13, 2022, at 3:41 PM, Javier Diaz-Reinoso via Squeak-dev <squeak-dev(a)lists.squeakfoundation.org> wrote: > > But -core-graphics works!, -metal have this message: > > 2022-08-13 17:35:32.628 Squeak[10943:21373230] -[sqSqueakOSXHeadlessView sqScreenSize]: unrecognized selector sent to instance 0x7ff77dd45c00 > > and stop, so probably the default was -opengl. > Thanks all for your help. That’s great! Now we have to make this happen automatically. I have an old laptop so I could probably get this to work. > >> The MBP is in 10.4 but with a patched OS, you can go up to Big Sur if you are very patient (it use Metal acceleration for the Finder for the first time). >> This mac don't support Metal, but the problem is only with MVC, Morphic (and the rest) works very well (so far). >> Testing with a lot of clicks I can get the browser (probably the menu open and the next click select the browser) and the browser works OK, so I think the problem is with (some) mouse events?. >> The -opengl flag don't help, same result. > >> On 13 Aug 2022, at 14:07, Eliot Miranda <eliot.miranda(a)gmail.com> wrote: >> >> https://support.apple.com/en-us/HT205073 >> >> _,,,^..^,,,_ (phone) >> >>>> On Aug 13, 2022, at 12:03 PM, Eliot Miranda <eliot.miranda(a)gmail.com> wrote: >>>> >>> Hi Javier, >>> >>> I expect it’s because the Mac doesn’t support Metal. If it is the same issue as we have with Virtend, the graphics subsystem required Metal support. There used to be a -opengl flag that would force the vm to use the opengl back end. This may still work. >>> >>> If it doesn’t we could possibly work together on vm-dev (we being the high dpi developers, you, and I’ll try and give some time; and I do have an old laptop that has 10.13 and does not support Metal on which to test) to try and get it going again. But if the code is missing from current MacOS, not merely deprecated, then it’s going to be difficult. >>> >>> _,,,^..^,,,_ (phone) >>> >>>>> On Aug 13, 2022, at 10:54 AM, Bruce O'Neel <bruce.oneel(a)pckswarms.ch> wrote: >>>>> >>>> >>>> So is the OS 10.13 or 10.14 then? >>>> >>>> If I would have to guess I would guess that some bit of our new HDPI code is falling over either on 10.13 or 10.14 on that type of display. >>>> >>>> It shouldn't, but, it might be very hard for someone to fix.... >>>> >>>> Thanks. >>>> >>>> cheers >>>> >>>> bruce >>>> >>>> On 2022-08-13T19:27:11.000+02:00, Javier Diaz-Reinoso <javier_diaz_r(a)mac.com> wrote: >>>> Its an old 2011 who is only Apple supported to 10.13: >>>> Hardware Overview: >>>> >>>> Model Name: MacBook Pro >>>> Model Identifier: MacBookPro8,1 >>>> Processor Name: Intel Core i7 >>>> Processor Speed: 2.7 GHz >>>> Number of Processors: 1 >>>> Total Number of Cores: 2 >>>> L2 Cache (per Core): 256 KB >>>> L3 Cache: 4 MB >>>> Hyper-Threading Technology: Enabled >>>> Memory: 8 GB >>>> Boot ROM Version: 87.0.0.0.0 >>>> SMC Version (system): 1.68f99 >>>> Serial Number (system): C02G669EDRJ9 >>>> Hardware UUID: DC85D6C6-A8B9-506A-8753-3AA32ACDA180 >>>> Sudden Motion Sensor: >>>> State: Enabled >>>> >>>> Graphics/Display: >>>> Intel HD Graphics 3000: >>>> >>>> Chipset Model: Intel HD Graphics 3000 >>>> Type: GPU >>>> Bus: Built-In >>>> VRAM (Dynamic, Max): 512 MB >>>> Vendor: Intel >>>> Device ID: 0x0126 >>>> Revision ID: 0x0009 >>>> Displays: >>>> Color LCD: >>>> Display Type: LCD >>>> Resolution: 1280 x 800 >>>> UI Looks like: 1280 x 800 >>>> Framebuffer Depth: 24-Bit Color (ARGB8888) >>>> Main Display: Yes >>>> Mirror: Off >>>> Online: Yes >>>> Rotation: Supported >>>> Automatically Adjust Brightness: No >>>> >>>> >>>> Really I don't use the MVC very much, so is not very important to me, I also have an Pi 3B+ who can use. >>>> >>>> >>>>> On 13 Aug 2022, at 11:29, Bruce O'Neel <bruce.oneel(a)pckswarms.ch> wrote: >>>>> >>>>> Hi, >>>>> >>>>> So which Mac Power Book? >>>>> >>>>> I'm trying to track down info about the display. >>>>> >>>>> It might help to run System Report, or what 10.14 called it to see the info about the display. >>>>> >>>>> cheers >>>>> >>>>> bruce >>>>> >>>>> >>>>> 2b456032378f72c0c1fc96275c00c479(a)infomaniak.com"><Screenshot 2022-08-13 at 18.24.53.png> >>>>> >>>>>> On 2022-08-13T17:33:42.000+02:00, Javier Diaz-Reinoso <javier_diaz_r(a)mac.com> wrote: >>>>>> >>>>>> >>>>>>> On 13 Aug 2022, at 02:21, Bruce O'Neel <bruce.oneel(a)pckswarms.ch> wrote: >>>>>>> >>>>>>> Hi, >>>>>>> >>>>>>> So it does work ok on 10.15. I guess you stay on 10.14 because 10.15 doesn't work on your mac, right? >>>>>> Yes >>>>>> >>>>>> I did see these sorts of problems when we were working towards a release and it might be related to the HDPI code that was put in. Is it possible to try changing the resolution of the screen and see if that fixes it? >>>>>> >>>>> No, this is a MBP 1280x800 I can only change to 1152x720 or 1024x840 neither works. >>>>> Testing in 5.3 I see the Quit entry in the mac menu also don't work to exit the image (why?) so that is not relevant. >>>>>> Also, what is the display? >>>>> Only the gray background: >>>>> >>>>> <Screen Shot 2022-08-13 at 10.17.12.png> >>>>>> >>>>>> Thanks. >>>>>> >>>>>> bruce >>>>>> >>>>>> fbe1c8cdc44f10e9e2867310e1395ab2(a)infomaniak.com"><Screenshot 2022-08-13 at 09.17.54.png> >>>>>> >>>>>> On 2022-08-12T21:02:21.000+02:00, Javier Diaz-Reinoso via Squeak-dev <squeak-dev(a)lists.squeakfoundation.org> wrote: >>>>>> I download Squeak6.0-22104-64bit.dmg, extract app, run, skip configuration, menu Create MVCProject, enter, click but the main menu don't open, I need to Force Quit to exit the image. >>>>>> >>>>>> Tested Squeak6.0-22104-64bit-202206021410-Windows-x64 in VirtualBox Win10 and that works OK, so is probably a VM bug. >>>>>> >>>>>> Mac is in Mojave 10.14.6. >>>>>> fbe1c8cdc44f10e9e2867310e1395ab2(a)infomaniak.com"><Screenshot 2022-08-13 at 09.17.54.png> >>>>> 2b456032378f72c0c1fc96275c00c479(a)infomaniak.com"><Screenshot 2022-08-13 at 18.24.53.png> >>>> >> > >

1 0

MPLR 2022: Call for Participation
by Stefan Marr 12 Aug '22

12 Aug '22

======================================================================== Call for Participation MPLR 2022 - 19th International Conference on Managed Programming Languages & Runtimes September 14-15, 2022 in Brussels, Belgium https://soft.vub.ac.be/mplr22/ Follow us @MPLR_Conf ======================================================================== The 19th International Conference on Managed Programming Languages & Runtimes (MPLR, formerly ManLang) is a premier forum for presenting and discussing novel results in all aspects of managed programming languages and runtime systems, which serve as building blocks for some of the most important computing systems in use, ranging from small-scale (embedded and real-time systems) to large-scale (cloud-computing and big-data platforms) and anything in between (desktop, mobile, IoT, and wearable applications). This year, MPLR will be held at the campus of the Vrije Universiteit Brussel(VUB) in Brussels, Belgium. Keynotes -------- Performance Optimizations in the .NET GC Dr. Maoni Stephens, Microsoft, USA Static Compilation of JavaScript Dr. Manuel Serrano, Inria Sophia-Antipolis, France Accepted Papers --------------- Automatic Array Transformation to Columnar Storage at Run Time Lukas Makor et al. Compressed Forwarding Tables Reconsidered Jonas Norlinder et al. Event-Based Out-of-place Debugging Tom Lauwaerts et al. Machine-Learning-Based Self-Optimizing Compiler Heuristics Raphael Mosaner et al. SecSharp: Towards Efficient Trusted Execution in Managed Languages Gilang Mentari Hamidy et al. A Model Checking Framework for a New Collector Framework Bochen Xu et al. Porting a JIT compiler to RISC-V: Challenges and Opportunities Quentin Ducasse et al. Analyzing and Predicting Energy Consumption of Garbage Collectors in OpenJDK Marina Shimchenko et al. Better Understanding the Costs and Benefits of Automatic Memory Management Kunal Sareen et al. Dynamic Taint Analysis with Label-Defined Semantics Jacob Kreindl et al. Attendance ---------- Thanks to our generous sponsors, we were able to significantly reduce this year’s registration fees. Early Registration Fees: Students: 175 Euro Regular: 250 Euro In addition to the conference participation, the registration includes: - 2-day transportation ticket in the Brussels area - welcome reception on Wednesday, September 14 - guided tour to the city hall and conference banquet on Thursday, September 15 More details, and link to registration system: https://soft.vub.ac.be/mplr22/registration/ Important Dates --------------- Early Registration: 23 August 2022 Conference Dates: 14-15 September 2022

1 0

VM Maker: VMMaker.oscog.seperateMarking-WoC.3254.mcz
by commits＠source.squeak.org 11 Aug '22

11 Aug '22

Tom Braun uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog.seperateMarking-WoC.3254.mcz ==================== Summary ==================== Name: VMMaker.oscog.seperateMarking-WoC.3254 Author: WoC Time: 11 August 2022, 9:37:57.506803 pm UUID: c10e995c-40c9-4433-85b9-6c4556f50088 Ancestors: VMMaker.oscog.seperateMarking-WoC.3253 add missing equality check =============== Diff against VMMaker.oscog.seperateMarking-WoC.3253 =============== Item was changed: ----- Method: SpurMemoryManager>>indexOf:in: (in category 'debug support') ----- indexOf: anElement in: anObject <api> | fmt numSlots | fmt := self formatOf: anObject. fmt <= self lastPointerFormat ifTrue: [numSlots := self numSlotsOf: anObject. 0 to: numSlots do: [:i| anElement = (self fetchPointer: i ofMaybeForwardedObject: anObject) ifTrue: [^i]]. -1]. fmt >= self firstByteFormat ifTrue: [fmt >= self firstCompiledMethodFormat ifTrue: [^self primitiveFailFor: PrimErrUnsupported]. numSlots := self numBytesOfBytes: anObject. 0 to: numSlots do: + [:i| anElement = (self fetchByte: i ofObject: anObject) ifTrue: [^i]]. - [:i| (self fetchByte: i ofObject: anObject) ifTrue: [^i]]. -1]. fmt >= self firstShortFormat ifTrue: [numSlots := self num16BitUnitsOf: anObject. 0 to: numSlots do: [:i| anElement = (self fetchUnsignedShort16: i ofObject: anObject) ifTrue: [^i]]. -1]. fmt = self sixtyFourBitIndexableFormat ifTrue: [numSlots := self num64BitUnitsOf: anObject. 0 to: numSlots do: [:i| anElement = (self fetchLong64: i ofObject: anObject) ifTrue: [^i]]. -1]. fmt >= self firstLongFormat ifTrue: [numSlots := self num32BitUnitsOf: anObject. 0 to: numSlots do: [:i| anElement = (self fetchLong32: i ofObject: anObject) ifTrue: [^i]]. -1]. ^-1!

1 0

VM Maker: VMMaker.oscog.seperateMarking-WoC.3253.mcz
by commits＠source.squeak.org 11 Aug '22

11 Aug '22

Tom Braun uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog.seperateMarking-WoC.3253.mcz ==================== Summary ==================== Name: VMMaker.oscog.seperateMarking-WoC.3253 Author: WoC Time: 11 August 2022, 3:46:03.942968 pm UUID: b99e4bbb-6ada-4c8e-a156-ccc460434dd4 Ancestors: VMMaker.oscog.seperateMarking-WoC.3252 shutdown incremental marking and reset the state correctly for enumeration primitives added missing class tracing for complete marking =============== Diff against VMMaker.oscog.seperateMarking-WoC.3252 =============== Item was added: + ----- Method: SpurGarbageCollector>>markObjectsForEnumerationPrimitives (in category 'as yet unclassified') ----- + markObjectsForEnumerationPrimitives + + ^ self shouldBeImplemented! Item was changed: ----- Method: SpurIncrementalGarbageCollector>>doIncrementalCollect (in category 'as yet unclassified') ----- doIncrementalCollect phase = InMarkingPhase ifTrue: [ marker incrementalMarkObjects ifTrue: [ "manager allPastSpaceObjectsDo: [:obj | self assert: (manager isWhite: obj)]." + manager + setCheckForLeaks: GCCheckFreeSpace + GCModeFull; + runLeakCheckerFor: GCModeFull; + checkFreeSpace: GCModeFull. - manager runLeakCheckerFor: GCModeFull. "when sweeping the mutator needs to allocate new objects black as we do not have any information about them. We only know if they should get swept after the next marking -> keep them alive for this cycle" self allocatorShouldAllocateBlack: true. phase := InSweepingPhase. "marking is done and thus all forwarding references are resolved -> we can use the now free segments that were compacted during the last cycle" compactor freePastSegmentsAndSetSegmentToFill. ^ self] ifFalse: [manager runLeakCheckerFor: GCModeIncremental]]. phase = InSweepingPhase ifTrue: [ compactor incrementalSweep ifTrue: [ self allocatorShouldAllocateBlack: false. manager allOldSpaceObjectsDo: [:ea | self assert: (manager isWhite: ea) ]. "self assert: manager allObjectsUnmarked." phase := InCompactingPhase. ^ self]]. phase = InCompactingPhase ifTrue: [ compactor incrementalCompact ifTrue: [phase := InMarkingPhase. ^ self]]! Item was added: + ----- Method: SpurIncrementalGarbageCollector>>markObjectsForEnumerationPrimitives (in category 'as yet unclassified') ----- + markObjectsForEnumerationPrimitives + + "do we need marking complete (and more accurate than incremental marking provides). For now lets do this only during marking although + probaly not 100% correct" + self flag: #Todo. "rework later on" + ^ phase = InMarkingPhase! Item was changed: ----- Method: SpurIncrementalMarker>>completeMark (in category 'marking - global') ----- completeMark "finishes the current mark pass" | currentObj slotsLeft | "manager objStack: manager markStack do: [:index :page | Transcript showln: (manager fetchPointer: index ofObject: page)]. manager sizeOfObjStack: manager markStack" currentObj := manager popObjStack: manager markStack. currentObj ifNil: [^ true]. "there is nothing more on the stack and we are done" slotsLeft := SlotLimitPerPass. [ | slotNumber slotsToVisit startIndex | "after passing the limit we push the current index on the stack. Is the currentObj only an index? " (manager isImmediate: currentObj) ifTrue: [startIndex := currentObj. currentObj := manager popObjStack: manager markStack.] ifFalse: [startIndex := 0]. + self markAndTraceClassOf: currentObj. + slotNumber := manager numStrongSlotsOfInephemeral: currentObj. slotsToVisit := slotNumber - startIndex. "we can mark all" slotsLeft := slotsLeft - slotsToVisit. self markFrom: startIndex nSlots: slotsToVisit of: currentObj. "we finished everything there is to be done with to obj -> make it black" self blackenObject: currentObj. currentObj := manager popObjStack: manager markStack. "repeat while there still are objects" currentObj notNil] whileTrue. ^ true! Item was changed: ----- Method: SpurIncrementalMarker>>completeMarkObjects (in category 'marking - global') ----- completeMarkObjects "this method is meant to be run for a complete GC that is used for snapshots. It discards previous marking information, because this will probably include some objects that should be collected It makes me a bit sad but I cannot see how this could be avoided" <inline: #never> "for profiling" - manager runLeakCheckerFor: GCModeFull. - "reset and reinitialize all helper structures and do actions to be done at the start of marking" self resetMarkProgress. self initializeForNewMarkingPassIfNecessary. self pushAllRootsOnMarkStack. self completeMark. + self finishMarking. + manager runLeakCheckerFor: GCModeFull. - self finishMarking ! Item was changed: ----- Method: SpurIncrementalMarker>>finishMarking (in category 'as yet unclassified') ----- finishMarking "marks the structures needed during GC" <inline: #never> 1 to: manager numClassTablePages - 1 do: [:i| manager setIsMarkedOf: (manager fetchPointer: i ofObject: manager hiddenRootsObj) to: true]. self flag: #Todo. "handle ephemerons" "lets assume there are not too many for now" self markWeaklingsAndMarkAndFireEphemerons. manager expungeDuplicateAndUnmarkedClasses: true. manager nilUnmarkedWeaklingSlots. + + self assert: (manager isEmptyObjStack: manager markStack). isCurrentlyMarking := false. marking := false! Item was changed: ----- Method: SpurIncrementalMarker>>markObjects: (in category 'as yet unclassified') ----- markObjects: objectsShouldBeUnmarkedAndUnmarkedClassesShouldBeExpunged + "mark objects has to mark all available objects in on go (e.g. used in allInstances where you only want live objects so you do not keep objects that should + be collected alive, as they get collected into an array that then holds a reference to the object) + In the incremental case we need to throw away our marking progress until now :( (in incremental collection garbage can stay around until the next round + of collection when we encountered an object during earlier stages of marking when an object is still alive, but it gets unreferenced during one of the + following mutator runs)" + + self flag: #Todo. "we made a (forced) complete marking. Lets use the results" + self completeMarkObjects. + self assert: (manager isEmptyObjStack: manager markStack).! - self flag: #Todo. "is this sane?" - self incrementalMarkObjects! Item was changed: ----- Method: SpurIncrementalSweeper class>>initialize (in category 'as yet unclassified') ----- initialize + MaxObjectsToFree := 10000! - MaxObjectsToFree := 100000! Item was changed: ----- Method: SpurMemoryManager>>allInstancesOf: (in category 'primitive support') ----- allInstancesOf: aClass "Attempt to answer an array of all objects, excluding those that may be garbage collected as a side effect of allocating the result array. If no memory is available answer the number of instances as a SmallInteger. Since objects are at least 16 bytes big, and the largest SmallInteger covers 1/4 of the address space, the count can never overflow." | classIndex freeChunk ptr start limit count bytes | classIndex := self rawHashBitsOf: aClass. classIndex = 0 ifTrue: [freeChunk := self allocateSlots: 0 format: self arrayFormat classIndex: ClassArrayCompactIndex. ^freeChunk]. + gc markObjectsForEnumerationPrimitives ifTrue: - MarkObjectsForEnumerationPrimitives ifTrue: [marker markObjects: true]. "may not want to revive objects unnecessarily; but marking is sloooow." freeChunk := self allocateLargestFreeChunk. "N.B. Does /not/ update totalFreeOldSpace" start := freeChunk + self baseHeaderSize. limit := self addressAfter: freeChunk. (self isClassAtUniqueIndex: aClass) ifTrue: [self uniqueIndex: classIndex allInstancesInto: start limit: limit resultsInto: [:c :p| count := c. ptr := p]] ifFalse: [self ambiguousClass: aClass allInstancesInto: start limit: limit resultsInto: [:c :p| count := c. ptr := p]]. self assert: (self isEmptyObjStack: markStack). + gc markObjectsForEnumerationPrimitives - MarkObjectsForEnumerationPrimitives ifTrue: [self assert: self allObjectsUnmarked. self emptyObjStack: weaklingStack] ifFalse: [self assert: (self isEmptyObjStack: weaklingStack)]. (count > (ptr - start / self bytesPerOop) "not enough room" or: [limit ~= ptr and: [limit - ptr <= self allocationUnit]]) ifTrue: "can't split a single word" [self freeObject: freeChunk. ^self integerObjectOf: count]. count < self numSlotsMask ifTrue: [| smallObj | smallObj := self allocateSlots: count format: self arrayFormat classIndex: ClassArrayCompactIndex. 0 to: count - 1 do: [:i| self storePointerUnchecked: i ofObject: smallObj withValue: (self fetchPointer: i ofFreeChunk: freeChunk)]. self freeChunkWithBytes: (self bytesInBody: freeChunk) at: (self startOfObject: freeChunk). self beRootIfOld: smallObj. self checkFreeSpace: GCModeFull. ^smallObj]. bytes := self largeObjectBytesForSlots: count. start := self startOfObject: freeChunk. self freeChunkWithBytes: limit - start - bytes at: start + bytes. totalFreeOldSpace := totalFreeOldSpace - bytes. self rawOverflowSlotsOf: freeChunk put: count. self set: freeChunk classIndexTo: ClassArrayCompactIndex formatTo: self arrayFormat. self possibleRootStoreInto: freeChunk. self checkFreeSpace: GCModeFull. self runLeakCheckerFor: GCModeFull. ^freeChunk! Item was changed: ----- Method: SpurMemoryManager>>allObjects (in category 'primitive support') ----- allObjects "Attempt to answer an array of all objects, excluding those that may be garbage collected as a side effect of allocating the result array. If no memory is available answer the number of objects as a SmallInteger. Since objects are at least 16 bytes big, and the largest SmallInteger covers 1/4 of the address space, the count can never overflow." | freeChunk ptr start limit count bytes | + gc markObjectsForEnumerationPrimitives ifTrue: - MarkObjectsForEnumerationPrimitives ifTrue: [marker markObjects: true]. "may not want to revive objects unnecessarily; but marking is sloooow." freeChunk := self allocateLargestFreeChunk. "N.B. Does /not/ update totalFreeOldSpace" ptr := start := freeChunk + self baseHeaderSize. limit := self addressAfter: freeChunk. count := 0. self allHeapEntitiesDo: [:obj| "continue enumerating even if no room so as to unmark all objects." + (gc markObjectsForEnumerationPrimitives - (MarkObjectsForEnumerationPrimitives ifTrue: [self isMarked: obj] ifFalse: [true]) ifTrue: [(self isNormalObject: obj) ifTrue: + [gc markObjectsForEnumerationPrimitives ifTrue: - [MarkObjectsForEnumerationPrimitives ifTrue: [self setIsMarkedOf: obj to: false]. count := count + 1. ptr < limit ifTrue: [self longAt: ptr put: obj. ptr := ptr + self bytesPerOop]] ifFalse: + [gc markObjectsForEnumerationPrimitives ifTrue: - [MarkObjectsForEnumerationPrimitives ifTrue: [(self isSegmentBridge: obj) ifFalse: [self setIsMarkedOf: obj to: false]]]]]. self assert: (self isEmptyObjStack: markStack). + gc markObjectsForEnumerationPrimitives - MarkObjectsForEnumerationPrimitives ifTrue: [self assert: self allObjectsUnmarked. self emptyObjStack: weaklingStack] ifFalse: [self assert: (self isEmptyObjStack: weaklingStack)]. self assert: count >= self numSlotsMask. (count > (ptr - start / self bytesPerOop) "not enough room" or: [limit ~= ptr and: [limit - ptr <= self allocationUnit]]) ifTrue: "can't split a single word" [self freeChunkWithBytes: (self bytesInBody: freeChunk) at: (self startOfObject: freeChunk). self checkFreeSpace: GCModeFull. ^self integerObjectOf: count]. bytes := self largeObjectBytesForSlots: count. start := self startOfObject: freeChunk. self freeChunkWithBytes: limit - start - bytes at: start + bytes. totalFreeOldSpace := totalFreeOldSpace - bytes. self rawOverflowSlotsOf: freeChunk put: count. self set: freeChunk classIndexTo: ClassArrayCompactIndex formatTo: self arrayFormat. self possibleRootStoreInto: freeChunk. self checkFreeSpace: GCModeFull. self runLeakCheckerFor: GCModeFull. ^freeChunk! Item was added: + ----- Method: SpurMemoryManager>>allObjectsWhite (in category 'gc - global') ----- + allObjectsWhite + self allObjectsDo: + [:o| ((self isMarked: o) or: [self isGrey: o]) ifTrue: [bogon := o. ^false]]. + ^true! Item was changed: ----- Method: SpurMemoryManager>>ambiguousClass:allInstancesInto:limit:resultsInto: (in category 'primitive support') ----- ambiguousClass: aClass allInstancesInto: start limit: limit resultsInto: binaryBlock "Dea with ambiguity and normalize indices." <inline: true> | expectedIndex count ptr | count := 0. ptr := start. expectedIndex := self rawHashBitsOf: aClass. self allHeapEntitiesDo: [:obj| | actualIndex | "continue enumerating even if no room so as to unmark all objects and/or normalize class indices." + (gc markObjectsForEnumerationPrimitives - (MarkObjectsForEnumerationPrimitives ifTrue: [self isMarked: obj] ifFalse: [true]) ifTrue: [(self isNormalObject: obj) ifTrue: + [gc markObjectsForEnumerationPrimitives ifTrue: - [MarkObjectsForEnumerationPrimitives ifTrue: [self setIsMarkedOf: obj to: false]. actualIndex := self classIndexOf: obj. (self classOrNilAtIndex: actualIndex) = aClass ifTrue: [actualIndex ~= expectedIndex ifTrue: [self setClassIndexOf: obj to: expectedIndex]. count := count + 1. ptr < limit ifTrue: [self longAt: ptr put: obj. ptr := ptr + self bytesPerOop]]] ifFalse: + [gc markObjectsForEnumerationPrimitives ifTrue: - [MarkObjectsForEnumerationPrimitives ifTrue: [(self isSegmentBridge: obj) ifFalse: [self setIsMarkedOf: obj to: false]]]]]. self purgeDuplicateClassTableEntriesFor: aClass. binaryBlock value: count value: ptr ! Item was added: + ----- Method: SpurMemoryManager>>makeAllObjectsWhite (in category 'primitive support') ----- + makeAllObjectsWhite + self allHeapEntitiesDo: + [:obj| + (self isWhite: obj) not ifTrue: + [(self isNormalObject: obj) + ifTrue: + [self setIsMarkedOf: obj to: false. + self setIsGreyOf: obj to: false] + ifFalse: + [(self isSegmentBridge: obj) ifFalse: + [self setIsMarkedOf: obj to: false. + self setIsGreyOf: obj to: false]]]]. + ! Item was added: + ----- Method: SpurMemoryManager>>markObjectsForEnumerationPrimitives (in category 'gc - incremental') ----- + markObjectsForEnumerationPrimitives + + <doNotGenerate> + ^ MarkObjectsForEnumerationPrimitives! Item was changed: ----- Method: SpurMemoryManager>>shutDownGlobalIncrementalGC: (in category 'gc - incremental') ----- shutDownGlobalIncrementalGC: objectsShouldBeUnmarked "If the incremental collector is running mark bits may be set; stop it and clear them if necessary." self flag: 'need to implement the global inc GC first...'. objectsShouldBeUnmarked ifTrue: + [self makeAllObjectsWhite. + self assert: self allObjectsUnmarked]! - [self assert: self allObjectsUnmarked]! Item was changed: ----- Method: SpurMemoryManager>>uniqueIndex:allInstancesInto:limit:resultsInto: (in category 'primitive support') ----- uniqueIndex: classIndex allInstancesInto: start limit: limit resultsInto: binaryBlock <inline: true> | count ptr | count := 0. ptr := start. self allHeapEntitiesDo: [:obj| "continue enumerating even if no room so as to unmark all objects." + (gc markObjectsForEnumerationPrimitives - (MarkObjectsForEnumerationPrimitives ifTrue: [self isMarked: obj] ifFalse: [true]) ifTrue: [(self isNormalObject: obj) ifTrue: + [gc markObjectsForEnumerationPrimitives ifTrue: - [MarkObjectsForEnumerationPrimitives ifTrue: [self setIsMarkedOf: obj to: false]. (self classIndexOf: obj) = classIndex ifTrue: [count := count + 1. ptr < limit ifTrue: [self longAt: ptr put: obj. ptr := ptr + self bytesPerOop]]] ifFalse: + [gc markObjectsForEnumerationPrimitives ifTrue: - [MarkObjectsForEnumerationPrimitives ifTrue: [(self isSegmentBridge: obj) ifFalse: [self setIsMarkedOf: obj to: false]]]]]. binaryBlock value: count value: ptr ! Item was added: + ----- Method: SpurStopTheWorldGarbageCollector>>markObjectsForEnumerationPrimitives (in category 'as yet unclassified') ----- + markObjectsForEnumerationPrimitives + + ^ manager markObjectsForEnumerationPrimitives! Item was changed: ----- Method: StackInterpreter>>incrementalMarkAndTraceStackPage: (in category 'object memory support') ----- incrementalMarkAndTraceStackPage: thePage | theSP theFP frameRcvrOffset callerFP oop | <var: #thePage type: #'StackPage *'> <var: #theSP type: #'char *'> <var: #theFP type: #'char *'> <var: #frameRcvrOffset type: #'char *'> <var: #callerFP type: #'char *'> <inline: false> self assert: (stackPages isFree: thePage) not. self assert: (self ifCurrentStackPageHasValidHeadPointers: thePage). thePage trace: StackPageTraced. theSP := thePage headSP. theFP := thePage headFP. "Skip the instruction pointer on top of stack of inactive pages." thePage = stackPage ifFalse: [theSP := theSP + objectMemory wordSize]. [frameRcvrOffset := self frameReceiverLocation: theFP. [theSP <= frameRcvrOffset] whileTrue: [oop := stackPages longAt: theSP. (objectMemory isOopForwarded: oop) ifTrue: [oop := objectMemory followForwarded: oop. stackPages longAt: theSP put: oop]. (objectMemory isImmediate: oop) ifFalse: + [objectMemory marker markAndShouldScan: oop]. - [objectMemory marker pushOnMarkingStackAndMakeGreyIfNecessary: oop]. theSP := theSP + objectMemory wordSize]. (self frameHasContext: theFP) ifTrue: [self assert: (objectMemory isContext: (self frameContext: theFP)). + objectMemory marker markAndShouldScan: (self frameContext: theFP)]. + objectMemory marker markAndShouldScan: (self iframeMethod: theFP). - objectMemory marker pushOnMarkingStackAndMakeGreyIfNecessary: (self frameContext: theFP)]. - objectMemory marker pushOnMarkingStackAndMakeGreyIfNecessary: (self iframeMethod: theFP). (callerFP := self frameCallerFP: theFP) ~= 0] whileTrue: [theSP := theFP + FoxCallerSavedIP + objectMemory wordSize. theFP := callerFP]. theSP := theFP + FoxCallerSavedIP. "caller ip is frameCallerContext in a base frame" [theSP <= thePage baseAddress] whileTrue: [oop := stackPages longAt: theSP. (objectMemory isOopForwarded: oop) ifTrue: [oop := objectMemory followForwarded: oop. stackPages longAt: theSP put: oop]. (objectMemory isImmediate: oop) ifFalse: + [objectMemory marker markAndShouldScan: oop]. - [objectMemory marker pushOnMarkingStackAndMakeGreyIfNecessary: oop]. theSP := theSP + objectMemory wordSize]!

1 0

VM Maker: VMMaker.oscog.seperateMarking-WoC.3252.mcz
by commits＠source.squeak.org 10 Aug '22

10 Aug '22

Tom Braun uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog.seperateMarking-WoC.3252.mcz ==================== Summary ==================== Name: VMMaker.oscog.seperateMarking-WoC.3252 Author: WoC Time: 10 August 2022, 12:57:43.436127 pm UUID: a36ddae8-e692-4d21-9c93-353093cdbc64 Ancestors: VMMaker.oscog.seperateMarking-WoC.3251 fixed off by one error in incremental marking =============== Diff against VMMaker.oscog.seperateMarking-WoC.3251 =============== Item was changed: ----- Method: SpurIncrementalMarker>>incrementalMark (in category 'marking - incremental') ----- incrementalMark "does one marking cycle. Breaks after a certain amount of slots is marked and the last object, that amount is crossed in, is completely scanned" | currentObj slotsLeft | "manager objStack: manager markStack do: [:index :page | Transcript showln: (manager fetchPointer: index ofObject: page)]. manager sizeOfObjStack: manager markStack" currentObj := manager popObjStack: manager markStack. "skip young objects. They get already scanned as they are part of the roots" [(currentObj notNil) and: [(manager isNonImmediate: currentObj) and: [manager isYoung: currentObj]]] whileTrue: [currentObj := manager popObjStack: manager markStack.]. currentObj ifNil: [^ true]. "there is nothing more on the stack and we are done" slotsLeft := SlotLimitPerPass. [ | slotNumber slotsToVisit startIndex | "after passing the limit we push the current index on the stack. Is the currentObj only an index? " (manager isImmediate: currentObj) ifTrue: [startIndex := manager integerValueOf: currentObj. currentObj := manager popObjStack: manager markStack.] ifFalse: [startIndex := 0. self markAndTraceClassOf: currentObj. "eager color the object black. Either it will get scanned completely and the color is correct or we have at least scanned some of the slots. In the second case the mutator could modify one of the slots of the object that already were scanned and we would could lose this object. Therefore color the object early to trigger the write barrier on writes. There will be some overhead (trigger the barrier always although only the already scanned slots are technically black) but it seems we need to do this for correctness" self blackenObject: currentObj]. slotNumber := manager numStrongSlotsOfInephemeral: currentObj. slotsToVisit := slotNumber - startIndex. slotsLeft - slotsToVisit < 0 ifTrue: [ self markFrom: startIndex nSlots: slotsLeft of: currentObj. "If we need to abort earlier we push the index and the currently scanned object on the marking stack. Otherwise it is not possible for immediates to be on the stack (they have no fields to be scanned) -> we can use the immediated to detect this pattern" (manager topOfObjStack: manager markStack) ~= currentObj ifTrue: [manager push: currentObj onObjStack: manager markStack]. + manager push: (manager integerObjectOf: slotsLeft) onObjStack: manager markStack. - manager push: (manager integerObjectOf: slotsLeft + 1) onObjStack: manager markStack. "we need to abort early to not run into some extreme corner cases (giant objects) that would explode our mark time assumptions" ^ false] ifFalse: ["we can mark all" slotsLeft := slotsLeft - slotsToVisit. self markFrom: startIndex nSlots: slotsToVisit of: currentObj]. currentObj := manager popObjStack: manager markStack. [(currentObj notNil) and: [(manager isNonImmediate: currentObj) and: [manager isYoung: currentObj]]] whileTrue: [currentObj := manager popObjStack: manager markStack.]. "repeat while there still are objects" currentObj notNil] whileTrue. ^ true!

1 0

VM Maker: VMMaker.oscog.seperateMarking-WoC.3251.mcz
by commits＠source.squeak.org 10 Aug '22

10 Aug '22

Tom Braun uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog.seperateMarking-WoC.3251.mcz ==================== Summary ==================== Name: VMMaker.oscog.seperateMarking-WoC.3251 Author: WoC Time: 10 August 2022, 1:26:34.109382 am UUID: 6e093abb-8fd3-4df3-b8f2-877c400b96d0 Ancestors: VMMaker.oscog.seperateMarking-WoC.3250 mark ephemeronStack insert some additional checks use write barrier only when marking =============== Diff against VMMaker.oscog.seperateMarking-WoC.3250 =============== Item was changed: SystemOrganization addCategory: #'VMMaker-Building'! SystemOrganization addCategory: #'VMMaker-Interpreter'! SystemOrganization addCategory: #'VMMaker-InterpreterSimulation'! SystemOrganization addCategory: #'VMMaker-JIT'! SystemOrganization addCategory: #'VMMaker-JITSimulation'! SystemOrganization addCategory: #'VMMaker-Multithreading'! SystemOrganization addCategory: #'VMMaker-Plugins'! SystemOrganization addCategory: #'VMMaker-Plugins-FFI'! SystemOrganization addCategory: #'VMMaker-Plugins-IOS'! SystemOrganization addCategory: #'VMMaker-PostProcessing'! SystemOrganization addCategory: #'VMMaker-SmartSyntaxPlugins'! + SystemOrganization addCategory: #'VMMaker-SpurGarbageCollector'! + SystemOrganization addCategory: #'VMMaker-SpurGarbageCollectorSimulation'! SystemOrganization addCategory: #'VMMaker-SpurMemoryManager'! SystemOrganization addCategory: #'VMMaker-SpurMemoryManagerSimulation'! - SystemOrganization addCategory: #'VMMaker-V3MemoryManager'! SystemOrganization addCategory: #'VMMaker-Support'! SystemOrganization addCategory: #'VMMaker-Tests'! SystemOrganization addCategory: #'VMMaker-Translation to C'! SystemOrganization addCategory: #'VMMaker-Utilities'! + SystemOrganization addCategory: #'VMMaker-V3MemoryManager'! - SystemOrganization addCategory: #'VMMaker-SpurGarbageCollectorSimulation'! - SystemOrganization addCategory: #'VMMaker-SpurGarbageCollector'! Item was changed: ----- Method: Spur64BitMMLESimulator>>setIsGreyOf:to: (in category 'header access') ----- setIsGreyOf: objOop to: aBoolean "objOop = 16rB26020 ifTrue: [self halt]." "(#(16r1971D0 16r196EE0 16r197048 16r197148) includes: objOop) ifTrue: [self halt]." GCEventLog register: ((aBoolean ifTrue: [GCGreyEvent] ifFalse: [GCUngreyEvent]) address: objOop). - - objOop = 16r448 ifTrue: [self halt]. super setIsGreyOf: objOop to: aBoolean. "(aBoolean and: [(self isContextNonImm: objOop) and: [(coInterpreter checkIsStillMarriedContext: objOop currentFP: coInterpreter framePointer) and: [(coInterpreter stackPages stackPageFor: (coInterpreter frameOfMarriedContext: objOop)) trace = 0]]]) ifTrue: [self halt]"! Item was added: + ----- Method: SpurIncrementalGarbageCollector>>doIncrementalCollect (in category 'as yet unclassified') ----- + doIncrementalCollect + + phase = InMarkingPhase + ifTrue: [ + marker incrementalMarkObjects + ifTrue: [ + "manager allPastSpaceObjectsDo: [:obj | self assert: (manager isWhite: obj)]." + manager runLeakCheckerFor: GCModeFull. + + "when sweeping the mutator needs to allocate new objects black as we do not have any information about them. + We only know if they should get swept after the next marking -> keep them alive for this cycle" + self allocatorShouldAllocateBlack: true. + phase := InSweepingPhase. + + "marking is done and thus all forwarding references are resolved -> we can use the now free segments that were + compacted during the last cycle" + compactor freePastSegmentsAndSetSegmentToFill. + + ^ self] + ifFalse: [manager runLeakCheckerFor: GCModeIncremental]]. + + phase = InSweepingPhase + ifTrue: [ + compactor incrementalSweep + ifTrue: [ + self allocatorShouldAllocateBlack: false. + manager allOldSpaceObjectsDo: [:ea | self assert: (manager isWhite: ea) ]. + "self assert: manager allObjectsUnmarked." + phase := InCompactingPhase. + ^ self]]. + + phase = InCompactingPhase + ifTrue: [ + compactor incrementalCompact + ifTrue: [phase := InMarkingPhase. + ^ self]]! Item was changed: ----- Method: SpurIncrementalGarbageCollector>>incrementalCollect (in category 'as yet unclassified') ----- incrementalCollect + self doIncrementalCollect. + + self assert: manager validObjStacks.! - phase = InMarkingPhase - ifTrue: [ - marker incrementalMarkObjects - ifTrue: [ - "manager allPastSpaceObjectsDo: [:obj | self assert: (manager isWhite: obj)]." - - "when sweeping the mutator needs to allocate new objects black as we do not have any information about them. - We only know if they should get swept after the next marking -> keep them alive for this cycle" - self allocatorShouldAllocateBlack: true. - phase := InSweepingPhase. - - "marking is done and thus all forwarding references are resolved -> we can use the now free segments that were - compacted during the last cycle" - compactor freePastSegmentsAndSetSegmentToFill. - - ^ self]]. - - phase = InSweepingPhase - ifTrue: [ - compactor incrementalSweep - ifTrue: [ - self allocatorShouldAllocateBlack: false. - manager allOldSpaceObjectsDo: [:ea | self assert: (manager isWhite: ea) ]. - "self assert: manager allObjectsUnmarked." - phase := InCompactingPhase. - ^ self]]. - - phase = InCompactingPhase - ifTrue: [ - compactor incrementalCompact - ifTrue: [phase := InMarkingPhase. - ^ self]]! Item was changed: ----- Method: SpurIncrementalMarker>>incrementalMarkObjects (in category 'marking - incremental') ----- incrementalMarkObjects "this method is to be run directly after a scavenge -> we can assume there are ony objects in the now past survivor space" <inline: #never> "for profiling" "manager runLeakCheckerFor: GCModeIncremental." self initializeForNewMarkingPassIfNecessary. [ | continueMarking | (manager isEmptyObjStack: manager markStack) ifTrue: [self pushAllRootsOnMarkStack. " manager sizeOfObjStack: manager markStack. did we finish marking?" (manager isEmptyObjStack: manager markStack) ifTrue: [self finishMarking. ^ true]]. + "due to a slang limitations we have to assign the result into variable => do not remove!!" - "due to a slang limitations we have to assign the result into variable" continueMarking := self incrementalMark. continueMarking] whileTrue. ^ false ! Item was changed: ----- Method: SpurIncrementalMarker>>pushHiddenRootsReferencesOnMarkingStack (in category 'root-scanning') ----- pushHiddenRootsReferencesOnMarkingStack | classTablePageSize | self markAndTraceObjStack: manager markStack andContents: false. self markAndTraceObjStack: manager weaklingStack andContents: false. self markAndTraceObjStack: manager mournQueue andContents: true. + self markAndTraceObjStack: manager ephemeronStack andContents: true. classTablePageSize := manager numStrongSlotsOfInephemeral: manager classTableFirstPage. self markNSlots: classTablePageSize of: manager classTableFirstPage. self blackenObject: manager classTableFirstPage! Item was changed: ----- Method: SpurIncrementalMarker>>writeBarrierFor:at:with: (in category 'barrier') ----- writeBarrierFor: anObject at: index with: value "a dijkstra style write barrier with the addition of the generation check objects that are not able to contain pointers are ignored too, as the write barries should ensure we lose no references and this objects do not hold any of them" <inline: true> self flag: #Todo. "we probably want the oldObject check to be the first one as it is only a pointer comparison and no dereferencing is needed" + (self marking and: [(self isLeafInObjectGraph: anObject) not and: [(manager isOldObject: anObject) and: [manager isMarked: anObject]]]) - ((self isLeafInObjectGraph: anObject) not and: [(manager isOldObject: anObject) and: [manager isMarked: anObject]]) ifTrue: [self pushOnMarkingStackAndMakeGreyIfNecessary: value]! Item was added: + ----- Method: SpurIncrementalMarkerSimulation>>writeBarrierFor:at:with: (in category 'as yet unclassified') ----- + writeBarrierFor: anObject at: index with: value + + GCEventLog + inContext: #writeBarrier + do: [super writeBarrierFor: anObject at: index with: value] + ! Item was added: + ----- Method: SpurIncrementalSweeperSimulator>>canUseAsFreeSpace: (in category 'as yet unclassified') ----- + canUseAsFreeSpace: objOop + + "objOop = 16r25FDBD8 ifTrue: [self halt]." + ^ super canUseAsFreeSpace: objOop!

1 0

VM Maker: VMMaker.oscog.seperateMarking-WoC.3250.mcz
by commits＠source.squeak.org 09 Aug '22

09 Aug '22

Tom Braun uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog.seperateMarking-WoC.3250.mcz ==================== Summary ==================== Name: VMMaker.oscog.seperateMarking-WoC.3250 Author: WoC Time: 9 August 2022, 11:43:29.170785 pm UUID: 7538f8e5-09f4-4386-bc3e-183325a03dfe Ancestors: VMMaker.oscog.seperateMarking-WoC.3249 being able to build a SpurStackVM =============== Diff against VMMaker.oscog.seperateMarking-WoC.3249 =============== Item was changed: ----- Method: SpurIncrementalCompactor class>>declareCVarsIn: (in category 'as yet unclassified') ----- declareCVarsIn: aCCodeGenerator - super declareCVarsIn: aCCodeGenerator. aCCodeGenerator var: #segmentToFill type: #'SpurSegmentInfo *'.! Item was changed: ----- Method: SpurIncrementalCompactor>>compact (in category 'api') ----- compact + <inline: #never> + self incrementalCompact! Item was added: + ----- Method: SpurIncrementalCompactor>>postSwizzleAction (in category 'api') ----- + postSwizzleAction + "Since the compact abuses the swizzle field of segment, it needs to be reset after start-up." + + | segInfo | + <var: 'segInfo' type: #'SpurSegmentInfo *'> + 0 to: manager numSegments - 1 do: + [:i| + segInfo := self addressOf: (manager segmentManager segments at: i). + segInfo swizzle: 0 ]! Item was removed: - ----- Method: SpurIncrementalCompactor>>restoreSwizzle (in category 'api') ----- - restoreSwizzle - "Since the compact abuses the swizzle field of segment, it needs to be reset after start-up." - - | segInfo | - <var: 'segInfo' type: #'SpurSegmentInfo *'> - 0 to: manager numSegments - 1 do: - [:i| - segInfo := self addressOf: (manager segmentManager segments at: i). - segInfo swizzle: 0 ]! Item was changed: ----- Method: SpurIncrementalGarbageCollector class>>classesForTranslation (in category 'as yet unclassified') ----- classesForTranslation + ^ { SpurGarbageCollector . self . SpurGenerationScavenger . SpurMarker . SpurIncrementalMarker . SpurCompactor . SpurIncrementalSweeper . SpurIncrementalCompactor . SpurIncrementalSweepAndCompact }! - ^ { SpurGarbageCollector . self . SpurGenerationScavenger . SpurMarker . SpurIncrementalMarker . SpurIncrementalSweeper . SpurIncrementalCompactor . SpurIncrementalSweepAndCompact }! Item was changed: ----- Method: SpurIncrementalGarbageCollector>>incrementalCollect (in category 'as yet unclassified') ----- incrementalCollect phase = InMarkingPhase ifTrue: [ marker incrementalMarkObjects ifTrue: [ + "manager allPastSpaceObjectsDo: [:obj | self assert: (manager isWhite: obj)]." - manager allPastSpaceObjectsDo: [:obj | self assert: (manager isWhite: obj)]. "when sweeping the mutator needs to allocate new objects black as we do not have any information about them. We only know if they should get swept after the next marking -> keep them alive for this cycle" self allocatorShouldAllocateBlack: true. phase := InSweepingPhase. "marking is done and thus all forwarding references are resolved -> we can use the now free segments that were compacted during the last cycle" compactor freePastSegmentsAndSetSegmentToFill. ^ self]]. phase = InSweepingPhase ifTrue: [ compactor incrementalSweep ifTrue: [ self allocatorShouldAllocateBlack: false. manager allOldSpaceObjectsDo: [:ea | self assert: (manager isWhite: ea) ]. "self assert: manager allObjectsUnmarked." phase := InCompactingPhase. ^ self]]. phase = InCompactingPhase ifTrue: [ compactor incrementalCompact ifTrue: [phase := InMarkingPhase. ^ self]]! Item was added: + ----- Method: SpurIncrementalMarker>>markObjects: (in category 'as yet unclassified') ----- + markObjects: objectsShouldBeUnmarkedAndUnmarkedClassesShouldBeExpunged + + self flag: #Todo. "is this sane?" + self incrementalMarkObjects! Item was changed: ----- Method: SpurIncrementalMarker>>pushOnMarkingStackAndMakeGreyIfNecessary: (in category 'marking-stack') ----- pushOnMarkingStackAndMakeGreyIfNecessary: objOop + (manager isImmediate: objOop) + ifTrue: [^ self]. + (manager isWhite: objOop) ifTrue: [self pushOnMarkingStackAndMakeGrey: objOop]! Item was added: + ----- Method: SpurIncrementalSweepAndCompact class>>declareCVarsIn: (in category 'as yet unclassified') ----- + declareCVarsIn: aCCodeGenerator! Item was added: + ----- Method: SpurIncrementalSweepAndCompact>>biasForSnapshot (in category 'as yet unclassified') ----- + biasForSnapshot! Item was added: + ----- Method: SpurIncrementalSweepAndCompact>>isMobile: (in category 'as yet unclassified') ----- + isMobile: obj + + <inline: true> + self flag: #Todo. "investigate this one here" + ^ (manager isPinned: obj) not! Item was added: + ----- Method: SpurIncrementalSweeper class>>declareCVarsIn: (in category 'as yet unclassified') ----- + declareCVarsIn: aCCodeGenerator! Item was removed: - ----- Method: SpurIncrementalSweeper>>biasForGC (in category 'api') ----- - biasForGC! Item was removed: - ----- Method: SpurIncrementalSweeper>>biasForSnapshot (in category 'api') ----- - biasForSnapshot! Item was changed: ----- Method: SpurIncrementalSweeperSimulator>>unmark: (in category 'api') ----- unmark: objOop + super unmark: objOop + "^ GCEventLog - ^ GCEventLog register: (GCWhitenEvent address: objOop) expecting: { [:evt | evt = (GCUnmarkEvent address: objOop)]. [:evt | evt = (GCUngreyEvent address: objOop)]} + doing: [super unmark: objOop]"! - doing: [super unmark: objOop]! Item was added: + ----- Method: SpurMarker>>markObjects: (in category 'as yet unclassified') ----- + markObjects: objectsShouldBeUnmarkedAndUnmarkedClassesShouldBeExpunged + + self shouldBeImplemented!

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev August 2022