Vm-dev January 2017

vm-dev@lists.squeakfoundation.org

30 participants
136 discussions

[OpenSmalltalk/opensmalltalk-vm] 98fb4e: oops
by GitHub 31 Jan '17

31 Jan '17

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: 98fb4e4973db117828305e35d157ad2cb707d434 https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/98fb4e4973db117828… Author: Tobias Pape <tobias(a)netshed.de> Date: 2017-01-31 (Tue, 31 Jan 2017) Changed paths: M image/updatespurimage.sh Log Message: ----------- oops

1 0

Can we make prim 118 support mirror logic?
by Denis Kudriashov 31 Jan '17

31 Jan '17

Hi. Right now prim 118 is probably only reflective call which not support mirror logic. It named #tryPrimitive:withArguments:. And it fails when it is called with receiver as first argument. Can we add support for it? The reason why I ask is because now any proxy should define this method to support debugger. Best regards, Denis

3 3

[OpenSmalltalk/opensmalltalk-vm] bf1046: Update way of accessing files on files.squeak.org
by GitHub 31 Jan '17

31 Jan '17

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: bf1046fe0446972f3bd6669a1b0c93c73e579ad7 https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/bf1046fe0446972f3b… Author: Tobias Pape <tobias(a)netshed.de> Date: 2017-01-31 (Tue, 31 Jan 2017) Changed paths: M image/ensureSqueakV50sources.sh M image/getsqueak50.sh M image/updatespurimage.sh Log Message: ----------- Update way of accessing files on files.squeak.org

2 1

VM Maker: VMMaker.oscog-cb.2117.mcz
by commits＠source.squeak.org 31 Jan '17

31 Jan '17

ClementBera uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-cb.2117.mcz ==================== Summary ==================== Name: VMMaker.oscog-cb.2117 Author: cb Time: 31 January 2017, 2:41:26.404463 pm UUID: b56c34a6-88ae-4f17-930a-fc625b7ab100 Ancestors: VMMaker.oscog-eem.2116 Fixed a bug in read-only object where the attemptToAssign:withIndex: call-back would be called with a 1-based index by the interpreter and 0-based index by the JIT. Everybody now calls it with 1-based index. =============== Diff against VMMaker.oscog-eem.2116 =============== Item was changed: ----- Method: CoInterpreter>>ceCannotAssignTo:withIndex:valueToAssign: (in category 'trampolines') ----- ceCannotAssignTo: immutableObject withIndex: index valueToAssign: valueToAssign + "index is unboxed and 0-based. The call-back expects 1-based value (to perform the operation with instVarAt:put:" - "index is unboxed." <api> <option: #IMMUTABILITY> instructionPointer := self popStack. self push: immutableObject. self push: valueToAssign. + self push: (objectMemory integerObjectOf: index + 1). - self push: (objectMemory integerObjectOf: index). self push: instructionPointer. ^ self ceSendAbort: (objectMemory splObj: SelectorAttemptToAssign) to: immutableObject numArgs: 2!

1 0

VM Maker: VMMaker.oscogSPC-eem.2118.mcz
by commits＠source.squeak.org 30 Jan '17

30 Jan '17

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscogSPC-eem.2118.mcz ==================== Summary ==================== Name: VMMaker.oscogSPC-eem.2118 Author: eem Time: 30 January 2017, 12:36:15.524891 pm UUID: 240e5693-afd5-4976-9230-d1fc4bf3c4b0 Ancestors: VMMaker.oscogSPC-eem.2117 SpurPlanningCompactor: freeFrom:upTo:previousPin: must resect the limit it is given and not free beyond it. Doing so causes double freeing. updateSavedFirstFieldsSpaceIfNecessary must not reset the free space carefully built up by previous compaction passes. noMobileObjectsFrom: doesn't need to be inlined and is more helpful for debugging if not. SpurMemoryManager: resetFreeLists should reset the freeListsMask! (old bug) Have printFreeChunk:printAsTreeNode: print the address after for tree chunks. Put a limit on valid class indices in the assert in classAtIndex: Cogit: Rename offsetCacheTagAndCouldBeObjectAt:annotation:into: to the more accurate entryCacheTagAndCouldBeObjectAt:annotation:into: and extract entryPointTagIsSelector: for use below. Fix the check for valid selectors for 64-bits (unlinked cacheTags are literal/specialSelector indices) in checkIfValidOopRefAndTarget:pc:cogMethod:. Simulator: Print all those NOT FOUNDs to the right transcript. =============== Diff against VMMaker.oscogSPC-eem.2117 =============== Item was changed: ----- Method: CogVMSimulator>>primitiveDirectoryEntry (in category 'file primitives') ----- primitiveDirectoryEntry | name pathName arrayNilOrSymbol result | name := self stringOf: self stackTop. pathName := self stringOf: (self stackValue: 1). "temporary work-around to make it work in Pharo..." self cppIf: PharoVM ifTrue: [ pathName := Smalltalk imagePath ]. self successful ifFalse: [^self primitiveFail]. arrayNilOrSymbol := FileDirectory default primLookupEntryIn: pathName name: name. arrayNilOrSymbol ifNil: [self pop: 3 thenPush: objectMemory nilObject. ^self]. arrayNilOrSymbol isArray ifFalse: ["arrayNilOrSymbol ~~ #primFailed ifTrue: [self halt]. " + self transcript show: name, ' NOT FOUND'. - Transcript show: name, ' NOT FOUND'. ^self primitiveFail]. result := PharoVM ifTrue: [self makeDirEntryName: (arrayNilOrSymbol at: 1) size: (arrayNilOrSymbol at: 1) size createDate: (arrayNilOrSymbol at: 2) modDate: (arrayNilOrSymbol at: 3) isDir: (arrayNilOrSymbol at: 4) fileSize: (arrayNilOrSymbol at: 5) posixPermissions: (arrayNilOrSymbol at: 6) isSymlink: (arrayNilOrSymbol at: 7) ] ifFalse: [self makeDirEntryName: (arrayNilOrSymbol at: 1) size: (arrayNilOrSymbol at: 1) size createDate: (arrayNilOrSymbol at: 2) modDate: (arrayNilOrSymbol at: 3) isDir: (arrayNilOrSymbol at: 4) fileSize: (arrayNilOrSymbol at: 5) ]. self pop: 3 thenPush: result! Item was changed: ----- Method: Cogit>>checkIfValidOopRefAndTarget:pc:cogMethod: (in category 'garbage collection') ----- checkIfValidOopRefAndTarget: annotation pc: mcpc cogMethod: cogMethod "Check for a valid object reference, if any, at a map entry. Answer a code unique to each error for debugging." <var: #mcpc type: #'char *'> <var: #nsSendCache type: #'NSSendCache *'> | literal entryPoint | annotation = IsObjectReference ifTrue: [literal := literalsManager fetchLiteralAtAnnotatedAddress: mcpc asUnsignedInteger using: backEnd. (self asserta: (objectRepresentation checkValidOopReference: literal)) ifFalse: [^1]. ((objectRepresentation couldBeObject: literal) and: [objectMemory isReallyYoungObject: literal]) ifTrue: [(self asserta: (self cCoerceSimple: cogMethod to: #'CogMethod *') cmRefersToYoung) ifFalse: [^2]]]. NewspeakVM ifTrue: [annotation = IsNSSendCall ifTrue: [| nsSendCache classTag enclosingObject nsTargetMethod | nsSendCache := self nsSendCacheFromReturnAddress: mcpc asInteger. (self asserta: (objectRepresentation checkValidOopReference: nsSendCache selector)) ifFalse: [^9]. classTag := nsSendCache classTag. (self asserta: (classTag = 0 or: [objectRepresentation validInlineCacheTag: classTag])) ifFalse: [^10]. enclosingObject := nsSendCache enclosingObject. (self asserta: (enclosingObject = 0 or: [objectRepresentation checkValidOopReference: enclosingObject])) ifFalse: [^11]. entryPoint := nsSendCache target. entryPoint ~= 0 ifTrue: [ nsTargetMethod := self cCoerceSimple: entryPoint - cmNoCheckEntryOffset to: #'CogMethod *'. (self asserta: (nsTargetMethod cmType = CMMethod)) ifFalse: [^12]]]]. (self isPureSendAnnotation: annotation) ifTrue: [(self asserta: (self cCoerceSimple: cogMethod to: #'CogMethod *') cmType = CMMethod) ifFalse: [^3]. + self entryCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: + [:entryPt :cacheTag :tagCouldBeObject| + entryPoint := entryPt. - self offsetCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: - [:offset :cacheTag :tagCouldBeObject| tagCouldBeObject ifTrue: [(objectRepresentation couldBeObject: cacheTag) ifTrue: [(self asserta: (objectRepresentation checkValidOopReference: cacheTag)) ifFalse: [^4]] ifFalse: [(self asserta: (objectRepresentation validInlineCacheTag: cacheTag)) ifFalse: [^5]]. ((objectRepresentation couldBeObject: cacheTag) and: [objectMemory isReallyYoungObject: cacheTag]) ifTrue: [(self asserta: (self cCoerceSimple: cogMethod to: #'CogMethod *') cmRefersToYoung) ifFalse: [^6]]] ifFalse: + [(self inlineCacheTagsAreIndexes + and: [self self entryPointTagIsSelector: entryPoint]) + ifTrue: + [cacheTag signedIntFromLong < 0 + ifTrue: + [cacheTag signedIntFromLong negated > NumSpecialSelectors ifTrue: + [^7]] + ifFalse: + [cacheTag >= (objectMemory literalCountOf: enumeratingCogMethod methodObject) ifTrue: + [^8]]] + ifFalse: + [(self asserta: (objectRepresentation validInlineCacheTag: cacheTag)) ifFalse: + [^9]]]]. - [(self asserta: (objectRepresentation validInlineCacheTag: cacheTag)) ifFalse: - [^7]]]. - entryPoint := backEnd callTargetFromReturnAddress: mcpc asInteger. entryPoint > methodZoneBase ifTrue: ["It's a linked send; find which kind." self targetMethodAndSendTableFor: entryPoint annotation: annotation into: [:targetMethod :sendTable| (self asserta: (targetMethod cmType = CMMethod or: [targetMethod cmType = CMClosedPIC or: [targetMethod cmType = CMOpenPIC]])) ifFalse: + [^10]]]]. - [^8]]]]. ^0 "keep scanning"! Item was added: + ----- Method: Cogit>>entryCacheTagAndCouldBeObjectAt:annotation:into: (in category 'in-line cacheing') ----- + entryCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: trinaryBlock + "Evaluate trinaryBlock with the entryPoint, inline cache tag and whether the cache + tag could be an object, for the send at mcpc with annotation annotation." + <inline: true> + | cacheTag entryPoint tagCouldBeObj | + cacheTag := backEnd inlineCacheTagAt: mcpc asInteger. + entryPoint := backEnd callTargetFromReturnAddress: mcpc asInteger. + "in-line cache tags are the selectors of sends if sends are unlinked, + the selectors of super sends (entry offset = cmNoCheckEntryOffset), + the selectors of open PIC sends (entry offset = cmEntryOffset, target is an Open PIC) + or in-line cache tags (classes, class indices, immediate bit patterns, etc). + Note that selectors can be immediate so there is no guarantee that they + are markable/remappable objects." + tagCouldBeObj := self inlineCacheTagsAreIndexes not + and: [objectRepresentation inlineCacheTagsMayBeObjects + or: [self entryPointTagIsSelector: entryPoint]]. + trinaryBlock + value: entryPoint + value: cacheTag + value: tagCouldBeObj! Item was added: + ----- Method: Cogit>>entryPointTagIsSelector: (in category 'in-line cacheing') ----- + entryPointTagIsSelector: entryPoint + "Answer if the entryPoint's tag is expected to be a selector reference, as opposed to a class tag." + ^entryPoint < methodZoneBase + or: [(entryPoint bitAnd: entryPointMask) = uncheckedEntryAlignment + or: [(entryPoint bitAnd: entryPointMask) = checkedEntryAlignment + and: [(self cCoerceSimple: entryPoint - cmEntryOffset to: #'CogMethod *') cmType = CMOpenPIC]]]! Item was changed: ----- Method: Cogit>>markLiterals:pc:method: (in category 'garbage collection') ----- markLiterals: annotation pc: mcpc method: cogMethod "Mark and trace literals. Additionally in Newspeak, void push implicits that have unmarked classes." <var: #mcpc type: #'char *'> <var: #nsSendCache type: #'NSSendCache *'> | literal | annotation = IsObjectReference ifTrue: [literal := literalsManager fetchLiteralAtAnnotatedAddress: mcpc asUnsignedInteger using: backEnd. (objectRepresentation markAndTraceLiteral: literal in: (self cCoerceSimple: cogMethod to: #'CogMethod *') atpc: mcpc asUnsignedInteger) ifTrue: [codeModified := true]]. NewspeakVM ifTrue: [annotation = IsNSSendCall ifTrue: [| nsSendCache sel eo | nsSendCache := self nsSendCacheFromReturnAddress: mcpc asInteger. sel := nsSendCache selector. (objectMemory isForwarded: sel) ifFalse: [objectMemory markAndTrace: sel] ifTrue: [sel := objectMemory followForwarded: literal. nsSendCache selector: sel. self markAndTraceUpdatedLiteral: sel in: (self cCoerceSimple: cogMethod to: #'CogMethod *')]. eo := nsSendCache enclosingObject. eo ~= 0 ifTrue: [(objectMemory isForwarded: eo) ifFalse: [objectMemory markAndTrace: eo] ifTrue: [eo := objectMemory followForwarded: literal. nsSendCache enclosingObject: eo. self markAndTraceUpdatedLiteral: eo in: (self cCoerceSimple: cogMethod to: #'CogMethod *')]]]]. (self isPureSendAnnotation: annotation) ifTrue: + [self entryCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: - [self offsetCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: [:entryPoint :cacheTag :tagCouldBeObj | tagCouldBeObj ifTrue: [(objectRepresentation markAndTraceCacheTagLiteral: cacheTag in: (self cCoerceSimple: cogMethod to: #'CogMethod *') atpc: mcpc asUnsignedInteger) ifTrue: ["cacheTag is selector" codeModified := true]]]]. ^0 "keep scanning"! Item was changed: ----- Method: Cogit>>markLiteralsAndUnlinkIfUnmarkedSend:pc:method: (in category 'garbage collection') ----- markLiteralsAndUnlinkIfUnmarkedSend: annotation pc: mcpc method: cogMethod "Mark and trace literals. Unlink sends that have unmarked cache tags or targets." <var: #mcpc type: #'char *'> <var: #nsSendCache type: #'NSSendCache *'> | literal | annotation = IsObjectReference ifTrue: [literal := literalsManager fetchLiteralAtAnnotatedAddress: mcpc asUnsignedInteger using: backEnd. (objectRepresentation markAndTraceLiteral: literal in: (self cCoerceSimple: cogMethod to: #'CogMethod *') atpc: mcpc asUnsignedInteger) ifTrue: [codeModified := true]]. NewspeakVM ifTrue: [annotation = IsNSSendCall ifTrue: [| nsSendCache entryPoint targetMethod sel eo | nsSendCache := self nsSendCacheFromReturnAddress: mcpc asInteger. entryPoint := nsSendCache target. entryPoint ~= 0 ifTrue: "Send is linked" [targetMethod := self cCoerceSimple: entryPoint - cmNoCheckEntryOffset to: #'CogMethod *'. (self markAndTraceOrFreeCogMethod: targetMethod firstVisit: targetMethod asUnsignedInteger > mcpc asUnsignedInteger) ifTrue: [self voidNSSendCache: nsSendCache]]. sel := nsSendCache selector. (objectMemory isForwarded: sel) ifFalse: [objectMemory markAndTrace: sel] ifTrue: [sel := objectMemory followForwarded: literal. nsSendCache selector: sel. self markAndTraceUpdatedLiteral: sel in: (self cCoerceSimple: cogMethod to: #'CogMethod *')]. eo := nsSendCache enclosingObject. eo ~= 0 ifTrue: [(objectMemory isForwarded: eo) ifFalse: [objectMemory markAndTrace: eo] ifTrue: [eo := objectMemory followForwarded: literal. nsSendCache enclosingObject: eo. self markAndTraceUpdatedLiteral: eo in: (self cCoerceSimple: cogMethod to: #'CogMethod *')]]]]. (self isPureSendAnnotation: annotation) ifTrue: + [self entryCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: - [self offsetCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: [:entryPoint :cacheTag :tagCouldBeObj | | cacheTagMarked | cacheTagMarked := tagCouldBeObj and: [objectRepresentation cacheTagIsMarked: cacheTag]. entryPoint > methodZoneBase ifTrue: "It's a linked send." [self targetMethodAndSendTableFor: entryPoint annotation: annotation into: [:targetMethod :sendTable| (cacheTagMarked not or: [self markAndTraceOrFreeCogMethod: targetMethod firstVisit: targetMethod asUnsignedInteger > mcpc asUnsignedInteger]) ifTrue: ["Either the cacheTag is unmarked (e.g. new class) or the target has been freed (because it is unmarked), so unlink the send." self unlinkSendAt: mcpc targetMethod: targetMethod sendTable: sendTable. objectRepresentation markAndTraceLiteral: targetMethod selector in: targetMethod at: (self addressOf: targetMethod selector put: [:val| targetMethod selector: val])]]] ifFalse: "cacheTag is selector" [(objectRepresentation markAndTraceCacheTagLiteral: cacheTag in: (self cCoerceSimple: cogMethod to: #'CogMethod *') atpc: mcpc asUnsignedInteger) ifTrue: [codeModified := true]]]]. ^0 "keep scanning"! Item was changed: ----- Method: Cogit>>markYoungObjects:pc:method: (in category 'garbage collection') ----- markYoungObjects: annotation pc: mcpc method: cogMethod "Mark and trace young literals." <var: #mcpc type: #'char *'> <var: #nsSendCache type: #'NSSendCache *'> | literal | annotation = IsObjectReference ifTrue: [literal := literalsManager fetchLiteralAtAnnotatedAddress: mcpc asUnsignedInteger using: backEnd. objectRepresentation markAndTraceLiteralIfYoung: literal]. NewspeakVM ifTrue: [annotation = IsNSSendCall ifTrue: [| nsSendCache | nsSendCache := self nsSendCacheFromReturnAddress: mcpc asInteger. objectRepresentation markAndTraceLiteralIfYoung: nsSendCache selector. nsSendCache enclosingObject ~= 0 ifTrue: [objectRepresentation markAndTraceLiteralIfYoung: nsSendCache enclosingObject]]]. (self isPureSendAnnotation: annotation) ifTrue: + [self entryCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: - [self offsetCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: [:entryPoint :cacheTag :tagCouldBeObj | tagCouldBeObj ifTrue: [objectRepresentation markAndTraceLiteralIfYoung: cacheTag]]]. ^0 "keep scanning"! Item was removed: - ----- Method: Cogit>>offsetCacheTagAndCouldBeObjectAt:annotation:into: (in category 'in-line cacheing') ----- - offsetCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: trinaryBlock - "Evaluate trinaryBlock with the entry, inline cache tag and whether the cache - tag could be an object, for the send at mcpc with annotation annotation." - <inline: true> - | cacheTag entryPoint tagCouldBeObj | - cacheTag := backEnd inlineCacheTagAt: mcpc asInteger. - entryPoint := backEnd callTargetFromReturnAddress: mcpc asInteger. - "in-line cache tags are the selectors of sends if sends are unlinked, - the selectors of super sends (entry offset = cmNoCheckEntryOffset), - the selectors of open PIC sends (entry offset = cmEntryOffset, target is an Open PIC) - or in-line cache tags (classes, class indices, immediate bit patterns, etc). - Note that selectors can be immediate so there is no guarantee that they - are markable/remappable objects." - tagCouldBeObj := self inlineCacheTagsAreIndexes not - and: [objectRepresentation inlineCacheTagsMayBeObjects - or: [entryPoint < methodZoneBase - or: [(entryPoint bitAnd: entryPointMask) = uncheckedEntryAlignment - or: [(entryPoint bitAnd: entryPointMask) = checkedEntryAlignment - and: [(self cCoerceSimple: entryPoint - cmEntryOffset to: #'CogMethod *') cmType = CMOpenPIC]]]]]. - trinaryBlock - value: entryPoint - value: cacheTag - value: tagCouldBeObj! Item was changed: ----- Method: Cogit>>remapIfObjectRef:pc:hasYoung: (in category 'garbage collection') ----- remapIfObjectRef: annotation pc: mcpc hasYoung: hasYoungPtr <var: #mcpc type: #'char *'> <var: #targetMethod type: #'CogMethod *'> <var: #nsSendCache type: #'NSSendCache *'> annotation = IsObjectReference ifTrue: [| literal mappedLiteral | literal := literalsManager fetchLiteralAtAnnotatedAddress: mcpc asUnsignedInteger using: backEnd. (objectRepresentation couldBeObject: literal) ifTrue: [mappedLiteral := objectRepresentation remapObject: literal. literal ~= mappedLiteral ifTrue: [literalsManager storeLiteral: mappedLiteral atAnnotatedAddress: mcpc asUnsignedInteger using: backEnd. codeModified := true]. (hasYoungPtr ~= 0 and: [objectMemory isYoung: mappedLiteral]) ifTrue: [(self cCoerceSimple: hasYoungPtr to: #'sqInt *') at: 0 put: true]]]. NewspeakVM ifTrue: [annotation = IsNSSendCall ifTrue: [| nsSendCache oop mappedOop | nsSendCache := self nsSendCacheFromReturnAddress: mcpc. oop := nsSendCache selector. mappedOop := objectRepresentation remapObject: oop. oop ~= mappedOop ifTrue: [nsSendCache selector: mappedOop. (hasYoungPtr ~= 0 and: [objectMemory isYoung: mappedOop]) ifTrue: [(self cCoerceSimple: hasYoungPtr to: #'sqInt *') at: 0 put: true]]. oop := nsSendCache enclosingObject. oop ~= 0 ifTrue: [ mappedOop := objectRepresentation remapObject: oop. oop ~= mappedOop ifTrue: [nsSendCache enclosingObject: mappedOop. (hasYoungPtr ~= 0 and: [objectMemory isYoung: mappedOop]) ifTrue: [(self cCoerceSimple: hasYoungPtr to: #'sqInt *') at: 0 put: true]]]. ^0 "keep scanning"]]. (self isPureSendAnnotation: annotation) ifTrue: + [self entryCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: - [self offsetCacheTagAndCouldBeObjectAt: mcpc annotation: annotation into: [:entryPoint :cacheTag :tagCouldBeObj | | mappedCacheTag | (tagCouldBeObj and: [objectRepresentation couldBeObject: cacheTag]) ifTrue: [mappedCacheTag := objectRepresentation remapObject: cacheTag. cacheTag ~= mappedCacheTag ifTrue: [backEnd rewriteInlineCacheTag: mappedCacheTag at: mcpc asUnsignedInteger. codeModified := true]. (hasYoungPtr ~= 0 and: [objectMemory isYoung: mappedCacheTag]) ifTrue: [(self cCoerceSimple: hasYoungPtr to: #'sqInt *') at: 0 put: true]]. hasYoungPtr ~= 0 ifTrue: ["Since the unlinking routines may rewrite the cacheTag to the send's selector, and since they don't have the cogMethod to hand and can't add it to youngReferrers, the method must remain in youngReferrers if the targetMethod's selector is young." entryPoint > methodZoneBase ifTrue: "It's a linked send." [self targetMethodAndSendTableFor: entryPoint annotation: annotation into: [:targetMethod :ignored| (objectMemory isYoung: targetMethod selector) ifTrue: [(self cCoerceSimple: hasYoungPtr to: #'sqInt *') at: 0 put: true]]]]]]. ^0 "keep scanning"! Item was changed: ----- Method: SpurMemoryManager>>classAtIndex: (in category 'class table') ----- classAtIndex: classIndex <api> | classTablePage | + self assert: (classIndex >= 0 and: [classIndex <= self tagMask or: [classIndex >= self arrayClassIndexPun and: [classIndex <= self classIndexMask]]]). - self assert: (classIndex <= self tagMask or: [classIndex >= self arrayClassIndexPun]). classTablePage := self fetchPointer: classIndex >> self classTableMajorIndexShift ofObject: hiddenRootsObj. classTablePage = nilObj ifTrue: [^nil]. ^self fetchPointer: (classIndex bitAnd: self classTableMinorIndexMask) ofObject: classTablePage! Item was changed: ----- Method: SpurMemoryManager>>printFreeChunk:printAsTreeNode: (in category 'debug printing') ----- printFreeChunk: freeChunk printAsTreeNode: printAsTreeNode | numBytes | numBytes := self bytesInObject: freeChunk. coInterpreter + print: 'freeChunk '; printHexPtrnp: freeChunk. + printAsTreeNode ifTrue: + [coInterpreter print: ' - '; printHexPtrnp:(self addressAfter: freeChunk)]. + coInterpreter - print: 'freeChunk '; printHexPtrnp: freeChunk; print: ' bytes '; printNum: numBytes; print: ' next '; printHexPtrnp: (self fetchPointer: self freeChunkNextIndex ofFreeChunk: freeChunk). (numBytes >= (self numFreeLists * self allocationUnit) and: [printAsTreeNode]) ifTrue: [coInterpreter print: ' ^ '; printHexPtrnp: (self fetchPointer: self freeChunkParentIndex ofFreeChunk: freeChunk); print: ' < '; printHexPtrnp: (self fetchPointer: self freeChunkSmallerIndex ofFreeChunk: freeChunk); print: ' > '; printHexPtrnp: (self fetchPointer: self freeChunkLargerIndex ofFreeChunk: freeChunk)]. coInterpreter cr! Item was changed: ----- Method: SpurMemoryManager>>resetFreeListHeads (in category 'free space') ----- resetFreeListHeads + freeListsMask := 0. 0 to: self numFreeLists - 1 do: [:i| freeLists at: i put: 0]! Item was changed: ----- Method: SpurPlanningCompactor>>freeFrom:upTo:previousPin: (in category 'private') ----- freeFrom: toFinger upTo: limit previousPin: previousPinOrNil "Free from toFinger up to limit, dealing with a possible intervening run of pinned objects starting at previousPinOrNil." <inline: false> <var: 'limit' type: #usqInt> <var: 'toFinger' type: #usqInt> <var: 'previousPinOrNil' type: #usqInt> | effectiveToFinger pin nextUnpinned start seg | <var: 'nextUnpinned' type: #usqInt> <var: #seg type: #'SpurSegmentInfo *'> self cCode: [] inSmalltalk: [coInterpreter cr; cr; print: 'freeing at '; printHexnp: toFinger; print: ' up to '; printHexnp: limit; print: ' pin '; printHexnp: previousPinOrNil; cr]. effectiveToFinger := toFinger. pin := previousPinOrNil. "If the range toFinger to limit spans segments but there is no pin (as when freeing to the end of memory) segment boundaries must still be observed. So in this case use the nearest bridge above toFinger as the pin." pin ifNil: [seg := manager segmentManager segmentContainingObj: toFinger. + self deny: seg isNil. seg segLimit < limit ifTrue: [pin := manager segmentManager bridgeFor: seg]]. + [pin notNil and: [pin < limit]] whileTrue: - [pin notNil] whileTrue: [(start := manager startOfObject: pin) > toFinger ifTrue: [manager addFreeChunkWithBytes: start - effectiveToFinger at: effectiveToFinger]. nextUnpinned := self unmarkPinnedObjectsAndFindFirstUnpinnedOrFreeEntityFollowing: pin. nextUnpinned >= limit ifTrue: [^self]. effectiveToFinger := manager startOfObject: nextUnpinned. pin := self findNextMarkedPinnedAfter: nextUnpinned]. limit > effectiveToFinger ifTrue: [manager addFreeChunkWithBytes: limit - effectiveToFinger at: effectiveToFinger]! Item was changed: ----- Method: SpurPlanningCompactor>>noMobileObjectsFrom: (in category 'private') ----- noMobileObjectsFrom: mobileObj + <inline: false> manager allOldSpaceEntitiesFrom: mobileObj do: [:o| ((manager isMarked: o) and: [(manager isPinned: o) not]) ifTrue: [^false]]. ^true! Item was changed: ----- Method: SpurPlanningCompactor>>updateSavedFirstFieldsSpaceIfNecessary (in category 'space management') ----- updateSavedFirstFieldsSpaceIfNecessary "If savedFirstFieldsSpace is a free chunk then it may need to be repositioned if there is more than one pass." <inline: true> self savedFirstFieldsSpaceInFreeChunk ifTrue: [self useFreeChunkForSavedFirstFieldsSpace: manager findLargestFreeChunk]. + savedFirstFieldsSpace top: savedFirstFieldsSpace start - manager bytesPerOop! - savedFirstFieldsSpace top: savedFirstFieldsSpace start - manager bytesPerOop. - manager resetFreeListHeads! Item was changed: ----- Method: StackInterpreterSimulator>>primitiveDirectoryEntry (in category 'file primitives') ----- primitiveDirectoryEntry | name pathName arrayNilOrSymbol result | name := self stringOf: self stackTop. pathName := self stringOf: (self stackValue: 1). "temporary work-around to make it work in Pharo..." self cppIf: PharoVM ifTrue: [ pathName := Smalltalk imagePath ]. self successful ifFalse: [^self primitiveFail]. arrayNilOrSymbol := FileDirectory default primLookupEntryIn: pathName name: name. arrayNilOrSymbol ifNil: [self pop: 3 thenPush: objectMemory nilObject. ^self]. arrayNilOrSymbol isArray ifFalse: ["arrayNilOrSymbol ~~ #primFailed ifTrue: [self halt]. " + self transcript show: name , ' NOT FOUND'. - Transcript show: name , ' NOT FOUND'. ^self primitiveFail]. result := PharoVM ifTrue: [self makeDirEntryName: (arrayNilOrSymbol at: 1) size: (arrayNilOrSymbol at: 1) size createDate: (arrayNilOrSymbol at: 2) modDate: (arrayNilOrSymbol at: 3) isDir: (arrayNilOrSymbol at: 4) fileSize: (arrayNilOrSymbol at: 5) posixPermissions: (arrayNilOrSymbol at: 6) isSymlink: (arrayNilOrSymbol at: 7) ] ifFalse: [self makeDirEntryName: (arrayNilOrSymbol at: 1) size: (arrayNilOrSymbol at: 1) size createDate: (arrayNilOrSymbol at: 2) modDate: (arrayNilOrSymbol at: 3) isDir: (arrayNilOrSymbol at: 4) fileSize: (arrayNilOrSymbol at: 5) ]. self pop: 3 thenPush: result!

1 0

f2c/g77 and problem with FFI 64bits returning float
by Nicolas Cellier 30 Jan '17

30 Jan '17

Hi, I have strange result with 64bits FFI function returning single precision float. Here is an example: (LapackSGEMatrix rows: #((2.3))) absMax. This matrix has a single element, 2.3 rounded to single precision float (2.299999952316284 when printed as a double precision) absMax is supposed to take the maximum of absolute values in the matrix. It does so thru Lapack function slange: " * Purpose * ======= * SLANGE returns the value of the one norm, or the Frobenius norm, or * the infinity norm, or the element of largest absolute value of a * real matrix A. " <cdecl: float 'slange_'( char * long * long * float * long * float * long )> Unfortunately above snippet returns 3.6893488147419103e19 It correctly calls this: floatRet = dispatchFunctionPointerwithwithwithwithwithwith(((float (*)(sqIntptr_t, sqIntptr_t, sqIntptr_t, sqIntptr_t, sqIntptr_t, sqIntptr_t)) procAddr), ((calloutState->integerRegisters))[0], ((calloutState->integerRegisters))[1], ((calloutState->integerRegisters))[2], ((calloutState->integerRegisters))[3], ((calloutState->integerRegisters))[4], ((calloutState->integerRegisters))[5]); which translates into something like: 0x10833c537 <+2615>: movq -0x28(%rbp), %rax 0x10833c53b <+2619>: movq -0xe8(%rbp), %rcx 0x10833c542 <+2626>: movq 0xd8(%rcx), %rdi 0x10833c549 <+2633>: movq -0xe8(%rbp), %rcx 0x10833c550 <+2640>: movq 0xe0(%rcx), %rsi 0x10833c557 <+2647>: movq -0xe8(%rbp), %rcx 0x10833c55e <+2654>: movq 0xe8(%rcx), %rdx 0x10833c565 <+2661>: movq -0xe8(%rbp), %rcx 0x10833c56c <+2668>: movq 0xf0(%rcx), %rcx 0x10833c573 <+2675>: movq -0xe8(%rbp), %r8 0x10833c57a <+2682>: movq 0xf8(%r8), %r8 0x10833c581 <+2689>: movq -0xe8(%rbp), %r9 0x10833c588 <+2696>: movq 0x100(%r9), %r9 -> 0x10833c58f <+2703>: callq *%rax 0x10833c591 <+2705>: cvtss2sd %xmm0, %xmm0 0x10833c595 <+2709>: movsd %xmm0, -0x150(%rbp) If I print $xmm0 just after the callq, then (lldb) nexti (lldb) print $xmm0 (unsigned char __attribute__((ext_vector_type(16)))) $212 = (0x00, 0x00, 0x00, 0x60, 0x66, 0x66, 0x02, 0x40, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00) and just after the connversion to double precision: (lldb) nexti (lldb) print $xmm0 (unsigned char __attribute__((ext_vector_type(16)))) $213 = (0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x44, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00) Let's see: tmp := #[16r00 16r00 16r00 16r60 16r66 16r66 16r02 16r40 ]. {tmp doubleAt: 1. tmp floatAt: 1}. #(2.299999952316284 3.6893488147419103e19) Bingo! that means that the value returned in xmm0 was already in double precision. When we convert it back to single precision (it's like interpreting the 4 LSB of the double as a single precision), then we get the incorrect value... So why was slange result promoted to double? I can reproduce on macosx with pre-installed veclib, and in win64 compiling LAPACK 3.3.1 from sources (translated by f2c) with MSVC10. Ah, Ah, f2c! Dont you promote float return values to double? YES But why this does not happen with the 32bits VM ??? That's what drove me off the solution for a while... It's the IA32 ABI... return value is stored in ST0 (allways promoted to double). So converting it to a double again like we do is a no-op and just works in 32bits. That's going to be a problem for FORTRAN functions on 64bits. IF compiled thru g77 or f2c conventions, then float results are promoted to double! IF compiled thru gfortran, then float result just remain float results. It means a major source of incompatibility: how to guess how this binary was compiled? (for example vecLib...) And how to adapt my FFI source code? Last thing, f2c might also be non standard when returning a complex value Big ball of mud...

3 5

New VM needed ...
by tim Rowledge 30 Jan '17

30 Jan '17

for this - https://www.top500.org/news/cray-to-deliver-arm-powered-supercomputer-to-uk… :-) tim -- tim Rowledge; tim(a)rowledge.org; http://www.rowledge.org/tim Gargoyle (n.), olive-flavored mouthwash.

1 0

VM Maker: VMMaker.oscogSPC-eem.2117.mcz
by commits＠source.squeak.org 28 Jan '17

28 Jan '17

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscogSPC-eem.2117.mcz ==================== Summary ==================== Name: VMMaker.oscogSPC-eem.2117 Author: eem Time: 28 January 2017, 12:58:43.099647 pm UUID: 7130d9ac-7881-4086-8350-391ecc3d61c6 Ancestors: VMMaker.oscogSPC-eem.2116 Must guard against trying ot free zero bytes at the end of memory. =============== Diff against VMMaker.oscogSPC-eem.2116 =============== Item was changed: ----- Method: SpurPlanningCompactor>>freeFrom:upTo:previousPin: (in category 'private') ----- freeFrom: toFinger upTo: limit previousPin: previousPinOrNil "Free from toFinger up to limit, dealing with a possible intervening run of pinned objects starting at previousPinOrNil." <inline: false> <var: 'limit' type: #usqInt> <var: 'toFinger' type: #usqInt> <var: 'previousPinOrNil' type: #usqInt> | effectiveToFinger pin nextUnpinned start seg | <var: 'nextUnpinned' type: #usqInt> <var: #seg type: #'SpurSegmentInfo *'> self cCode: [] inSmalltalk: [coInterpreter cr; cr; print: 'freeing at '; printHexnp: toFinger; print: ' up to '; printHexnp: limit; print: ' pin '; printHexnp: previousPinOrNil; cr]. effectiveToFinger := toFinger. pin := previousPinOrNil. "If the range toFinger to limit spans segments but there is no pin (as when freeing to the end of memory) segment boundaries must still be observed. So in this case use the nearest bridge above toFinger as the pin." pin ifNil: [seg := manager segmentManager segmentContainingObj: toFinger. seg segLimit < limit ifTrue: [pin := manager segmentManager bridgeFor: seg]]. [pin notNil] whileTrue: [(start := manager startOfObject: pin) > toFinger ifTrue: [manager addFreeChunkWithBytes: start - effectiveToFinger at: effectiveToFinger]. nextUnpinned := self unmarkPinnedObjectsAndFindFirstUnpinnedOrFreeEntityFollowing: pin. nextUnpinned >= limit ifTrue: [^self]. effectiveToFinger := manager startOfObject: nextUnpinned. pin := self findNextMarkedPinnedAfter: nextUnpinned]. + limit > effectiveToFinger ifTrue: + [manager addFreeChunkWithBytes: limit - effectiveToFinger at: effectiveToFinger]! - manager addFreeChunkWithBytes: limit - effectiveToFinger at: effectiveToFinger!

1 0

[OpenSmalltalk/opensmalltalk-vm] 368974: CogVM source as per VMMaker.oscogSPC-eem.2116
by GitHub 28 Jan '17

28 Jan '17

Branch: refs/heads/SpurPlanningCompactor Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: 3689748e3dd657560df5d011c6a54fd6e68313ce https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/3689748e3dd657560d… Author: Eliot Miranda <eliot.miranda(a)gmail.com> Date: 2017-01-28 (Sat, 28 Jan 2017) Changed paths: M nsspur64src/vm/cogit.h M nsspur64src/vm/cogitX64.c M nsspur64src/vm/cointerp.c M nsspur64src/vm/cointerp.h M nsspur64src/vm/gcc3x-cointerp.c M nsspursrc/vm/cogit.h M nsspursrc/vm/cogitARMv5.c M nsspursrc/vm/cogitIA32.c M nsspursrc/vm/cogitMIPSEL.c M nsspursrc/vm/cointerp.c M nsspursrc/vm/cointerp.h M nsspursrc/vm/gcc3x-cointerp.c M nsspurstack64src/vm/gcc3x-interp.c M nsspurstack64src/vm/interp.c M nsspurstacksrc/vm/gcc3x-interp.c M nsspurstacksrc/vm/interp.c M spur64src/vm/cogit.h M spur64src/vm/cogitX64.c M spur64src/vm/cointerp.c M spur64src/vm/cointerp.h M spur64src/vm/gcc3x-cointerp.c M spursista64src/vm/cogit.h M spursista64src/vm/cogitX64.c M spursista64src/vm/cointerp.c M spursista64src/vm/cointerp.h M spursista64src/vm/gcc3x-cointerp.c M spursistasrc/vm/cogit.h M spursistasrc/vm/cogitARMv5.c M spursistasrc/vm/cogitIA32.c M spursistasrc/vm/cogitMIPSEL.c M spursistasrc/vm/cointerp.c M spursistasrc/vm/cointerp.h M spursistasrc/vm/gcc3x-cointerp.c M spursrc/vm/cogit.h M spursrc/vm/cogitARMv5.c M spursrc/vm/cogitIA32.c M spursrc/vm/cogitMIPSEL.c M spursrc/vm/cointerp.c M spursrc/vm/cointerp.h M spursrc/vm/gcc3x-cointerp.c M spurstack64src/vm/gcc3x-interp.c M spurstack64src/vm/interp.c M spurstacksrc/vm/gcc3x-interp.c M spurstacksrc/vm/interp.c M src/vm/cogit.h M src/vm/cogitARMv5.c M src/vm/cogitIA32.c M src/vm/cogitMIPSEL.c M src/vm/cointerp.c M src/vm/cointerp.h M src/vm/cointerpmt.c M src/vm/cointerpmt.h M src/vm/gcc3x-cointerp.c M src/vm/gcc3x-cointerpmt.c M stacksrc/vm/gcc3x-interp.c M stacksrc/vm/interp.c Log Message: ----------- CogVM source as per VMMaker.oscogSPC-eem.2116 Temporary branch for SpurPlanningCompactor as default compactor. Get the termination condition for copyAndUnmarkMobileObjects right. Only exit when finding a marked object beyond lastMobileObject. Deal with coalescing free space since this new variation may produce two adjacent free chunks when trying to tread carefully past any pins. Add and use oldSpaceObjectAfterUse in SpurPlanningCompactor to avoid the overhead in objectAfter:. Also see commit commit de43045c722a64b96e4503ad0942fce4cc89dca1 on the Cog branch.

1 0

[OpenSmalltalk/opensmalltalk-vm] de4304: CogVM source as per VMMaker.oscog-eem.2116
by GitHub 28 Jan '17

28 Jan '17

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: de43045c722a64b96e4503ad0942fce4cc89dca1 https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/de43045c722a64b96e… Author: Eliot Miranda <eliot.miranda(a)gmail.com> Date: 2017-01-28 (Sat, 28 Jan 2017) Changed paths: M nsspur64src/vm/cogit.h M nsspur64src/vm/cointerp.c M nsspur64src/vm/cointerp.h M nsspur64src/vm/gcc3x-cointerp.c M nsspursrc/vm/cogit.h M nsspursrc/vm/cointerp.c M nsspursrc/vm/cointerp.h M nsspursrc/vm/gcc3x-cointerp.c M nsspurstack64src/vm/gcc3x-interp.c M nsspurstack64src/vm/interp.c M nsspurstacksrc/vm/gcc3x-interp.c M nsspurstacksrc/vm/interp.c M spur64src/vm/cogit.h M spur64src/vm/cointerp.c M spur64src/vm/cointerp.h M spur64src/vm/gcc3x-cointerp.c M spursista64src/vm/cogit.h M spursista64src/vm/cointerp.c M spursista64src/vm/cointerp.h M spursista64src/vm/gcc3x-cointerp.c M spursistasrc/vm/cogit.h M spursistasrc/vm/cointerp.c M spursistasrc/vm/cointerp.h M spursistasrc/vm/gcc3x-cointerp.c M spursrc/vm/cogit.h M spursrc/vm/cointerp.c M spursrc/vm/cointerp.h M spursrc/vm/gcc3x-cointerp.c M spurstack64src/vm/gcc3x-interp.c M spurstack64src/vm/interp.c M spurstacksrc/vm/gcc3x-interp.c M spurstacksrc/vm/interp.c Log Message: ----------- CogVM source as per VMMaker.oscog-eem.2116 Interpreter: Fixed a bug where the CogVMFeatureFlags were converted twice to SmallInteger (in getCogVMFeatureFlags and in primitiveVMParameter), shifting 1 time too many the flags. Hence MULTIPLEBYTECODESETS flag was always true in the image while SUPPORTSWRITEBARRIER was true only if the VM supported multiple bytecode set. Make printFrame: lax enough to accept a married context and print its frame without complaint. Decode a method header when doing longPrintOop:. Add printOopsSuchThat: for debugging. Spur: Include rawNumSlots and eliminate the hex prefix in printEntity:'s size info. Change attemptToShrink to request the segmentManager to shrink by at least growHeadroom, since it won't otherwise. Have shrinkObjectMemory: answer if shrinkage occurred, and only increment statShrinkMemory if so. Cogit: Add leak checking of CogMethod counters and implicit receiver caches. StackToRegisterMappingCogit: Follow jumps to jumps and push: aBoolean; jump:if:s, eliminating dead code. Elimnate jumps to the immediately following instruction in StackToRegisterMappingCogit>>generateInstructionsAt:. Sista: Added support for BitShiftRight: and bitShiftLeft: unsafe operations Added support for inlined primitive #characterAsInteger. Fixed a bug where branchIfNotInstanceOf: span function misscomputed the distance when ExtB is negated to inverse the branch. Set deadCode to true in trap bytecode to avoid generating merge machine code that is never reached. Use Spur's single-bit tests for immediates in the CogObjectRepresentationFor??BitSpur's branchIf:has[Not]ImmediateTag:target: Fix the ssFlushTo: call in genExtJumpIfNotInstanceOfBehaviorsBytecode Fix coging optimized methods with lots of literals. Change the machine compilation of BranchIfInstanceOf if there is one Behavior and the behavior is UndefinedObject, True and False to compile to a direct comparison to the unique instances instead of class index check. Add a new trampoline to create new hashes. The trampoline is Sista only and it's used in the inlined primitive. ThreadedFFIPlugin: Fix FFI in 64bits. The moduleHandle (a pointer) was incorrectly truncated to an int (32bits). Lowcode: 64 bits version of Lowcode. All of the tests for WodenMath are passing. Workaround a GCC crash in Windows when building a Lowcode VM. Too much register allocation pressure for calling a builtin memcpy.

1 0

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 January 2017