Vm-dev May 2020

vm-dev@lists.squeakfoundation.org

39 participants
130 discussions

Support an updated image format number for Sista
by David T. Lewis 20 Jul '20

20 Jul '20

Hi Eliot, This is a belated follow up on earlier private discussion during a Squeak board meeting a few months ago, in which we agreed that identifying "Sista in use" in the image format number would be a useful thing to do. I previously added the documentation for this in ImageFormat-dtl.39.mcz (as well as the updated ckformat.c program to match). Attached here is a change set that can be loaded into a VMMaker image to add VM support for setting and reading the updated image format numbers. Here I used a primitive to initially set or clear the "sista bit". There are other ways this could be done, but this at least gets us started with a working implementation. By way of explanation, I am copying below both the change set preamble and the earlier ImageFormat commit notice: === Change Set: VMM-Sista-bit-dtl Date: 12 January 2020 Author: David T. Lewis Let the image inform the interpreter that alternate bytecodes either are or are not in use, and remember that setting when writing or reading the image format number in a shapshot file header. Provide primitiveMultipleBytecodesActive to allow the setting to be updated from the image. Expected usage is that the image can call this primitive after recompiling with Systa bytecodes, or after reverting back. Note that an alternative implemenation would be to use a VM parameter. The primitive was chosen here because it can perform parameter checking and does not require coordinated update the the VM parameter usage. Expected usage is that this feature will remain unused for some grace period (perhaps a few months) during which an updated VM can be widely circulated. After that period, images using Sista bytecodes may be safely saved with the updated image format number indicating that multiple bytecodes are in use. This change set includes helper methods in SmalltalkImage for testing or setting the multipleBytecodeSetsActive setting in the VM. Note the change set postscript which is necessary for initializing a class variable prior to code generation. === Name: ImageFormat-dtl.37 Author: dtl Time: 20 July 2019, 5:23:59.004 pm UUID: 52b464d5-0bf7-4326-bde3-4f18ad70b239 Ancestors: ImageFormat-dtl.36 Identify extended bytecodes for Sista in the image format number. Bit 10 identifies an image that contains extended bytecodes. Thus a 32 bit Spur image is 6521, with Sista it is 7033, and a 64 bit Spur image is 68021, with Sista it is 68533. It is expected that additional bytecode sets can be identified by an additional field in the image header, probably 32 bits containing two 16 bit fields, if both zero then Sista. Per discussion with Eliot and Bert in a Squeak oversight board meeting 2019-05-15. Dave

4 10

[OpenSmalltalk/opensmalltalk-vm] OpenSSL 1.0.2 end of support from 1 Jan 2020 (#455)
by Pierce Ng 22 Jun '20

22 Jun '20

As per https://www.openssl.org/policies/releasestrat.html. My fork has an [update3p branch](https://github.com/PierceNg/opensmalltalk-vm/tree/pierce_update3p) that builds with OpenSSL 1.1.1d and also related minor fixes for one or some of the Pharo mvm. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/455

3 7

[OpenSmalltalk/opensmalltalk-vm] Removes support for NSPlugin and browser mode on win32 (#501)
by Marcel Taeumel 13 Jun '20

13 Jun '20

The command line arguments "browserWindow" and "browserWindow:" are no longer recognized. For a discussion, see http://forum.world.st/Supporting-fBrowserMode-on-Windows-td5116489.html You can view, comment on, or merge this pull request online at: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/501 -- Commit Summary -- * Removes support for NSPlugin and browser mode on win32 -- File Changes -- M platforms/win32/plugins/DropPlugin/sqWin32Drop.c (4) M platforms/win32/vm/sqWin32.h (8) M platforms/win32/vm/sqWin32Main.c (25) M platforms/win32/vm/sqWin32PluginSupport.c (510) M platforms/win32/vm/sqWin32Window.c (128) -- Patch Links -- https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/501.patch https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/501.diff -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/501

2 2

[OpenSmalltalk/opensmalltalk-vm] Removes USE_DIB_SECTIONS as part of Windows CE removal (#503)
by Marcel Taeumel 13 Jun '20

13 Jun '20

You can view, comment on, or merge this pull request online at: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/503 -- Commit Summary -- * Removes USE_DIB_SECTIONS as part of Windows CE removal * Removes comment about "NT vs CE" bc. not meaningful anymore. -- File Changes -- M platforms/win32/misc/Makefile.mingw32 (1) M platforms/win32/vm/sqWin32.h (2) M platforms/win32/vm/sqWin32Window.c (98) -- Patch Links -- https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/503.patch https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/503.diff -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/503

2 1

64 bit FFI (was: porting Croquet to Squeak6.0 alpha...)
by Bert Freudenberg 09 Jun '20

09 Jun '20

I'd suggest to get OpenGL working outside of Croquet first: http://www.squeaksource.com/CroquetGL.html Step 1: Verify this works in 32 bits. (assuming you are doing this on Linux, you can run 32 bit Squeak side-by-side with the 64 bit one) Step 2: Make it work in 64 bits. The second step requires that you understand how FFI works, and how it handles e.g. pointers and integer sizes. I am assuming we do have a working 64 bit FFI, at least for x86_64 machines. E.g. the OGLUnix>>glExtGetProcAddress: method returns a pointer. On a 32 bit system, that fits into a 'ulong' which is 32 bits. On a 64 bit system, a pointer is 64 bits wide so it would not fit into a 32 bit word. Now I don't know how many bits 'ulong' has in our 64 bit FFI, but that declaration may have to change. Etc. pp. If you have questions about FFI then those are best directed at the vm-dev list since it is dealing with VM-level interfaces. CC'ing, please follow up there. - Bert - On Wed, Mar 11, 2020 at 2:54 PM gettimothy via Squeak-dev < squeak-dev(a)lists.squeakfoundation.org> wrote: > Okey dokey, > > Poking along, there is a stray glyph in OGLUnix openGLLibraryName after > > openGLLibraryName > ^Smalltalk osVersion = 'linux' > ifTrue: ['libGL.so.1'] > ifFalse: ['GL'] > > I removed it in my install and got past that error. > > Working exclusively with Croquet(Master)... > > > > My next error is in OGLUnixX11LE(OpenGL)>>glMatrixMode: > > glMatrixMode: mode > "This method was automatically generated." > "void glMatrixMode(GLenum mode);" > <apicall: void 'glMatrixMode' (ulong) module: '#openGLLibraryName'> > ^self externalCallFailed > > The <apicall:...> fails > > How to think about this? > > Is Croquet behind OpenGL latest? > Would teaching myself OpenGL programming be of use to the Croquet project? > > cheers, > > tty > > >

7 13

[OpenSmalltalk/opensmalltalk-vm] POWER (ppc64le) port. (#493)
by kgardas 03 Jun '20

03 Jun '20

It would be great if there is also ppc64le/power platform supported. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/493

9 27

Cog on ARM
by Boris Shingarov 02 Jun '20

02 Jun '20

To build the latest Cog VM with the ARM simulation plugins doesn't seem to work for me. I thought maybe I had dirty state, so I started from a clean clone (this is the current tip of the Cog branch, 60973ef): $ git clone git@github.com:OpenSmalltalk/opensmalltalk-vm.git $ cd opensmalltalk-vm $ cd build.linux64x64/gdbarm32 $ ./conf.COG checking build system type... x86_64-pc-linux-gnu checking host system type... x86_64-pc-linux-gnu checking target system type... arm-unknown-linux-gnu checking for gcc... gcc checking whether the C compiler works... yes checking for C compiler default output file name... a.out ... checking for working strncmp... yes configure: creating ./config.status config.status: creating Makefile config.status: error: cannot find input file: `testsuite/Makefile.in' $ _ Is this the supposed way? Has anyone built ARM recently? I build a fair amount of ARM code somewhat regularly on this machine, therefore I doubt this is some basic dysfunctionality of my dev environment. My first reflex was to go and just hack into the problem, but if it doesn't even go this far out of the box, then I'd expect I am somehow wrong and I thought I'd ask before digging myself too deep into the wrong direction. Also I've been having doubts, if these configure scripts were broken, people would be complaining? does anyone do ARM? (and by that I mean "develop the VM" for which you need the simulator) Why isn't CI complaining? does CI only cover Intel? does CI only cover production-packaging?

2 2

VM Maker: VMMaker.oscog-eem.2759.mcz
by commits＠source.squeak.org 01 Jun '20

01 Jun '20

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2759.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2759 Author: eem Time: 31 May 2020, 4:30:04.25407 pm UUID: 830cf133-03ef-4a4e-abbb-856c29f8d4a0 Ancestors: VMMaker.oscog-eem.2758 The minimum alignment of a structure should be determined by a structure that contains an element with minimal alignment requirements, not maximal ones. =============== Diff against VMMaker.oscog-eem.2758 =============== Item was changed: ----- Method: AlignmentStructType class>>printTypedefOn: (in category 'code generation') ----- printTypedefOn: aStream + #('Byte' 'Short' 'Int' 'LongLong' 'Float' 'Double') + with: #(#char #short #int #'long long' #float #double) do: + [:structName :type| + aStream + nextPutAll: 'typedef struct { char pad_to_misalgnment; '; + nextPutAll: type; + nextPutAll: ' element; } struct'; + nextPutAll: structName; + nextPut: $;; cr; cr]. - | union | - union := String streamContents: - [:unionStream| - #('Byte' 'Short' 'Int' 'LongLong' 'Float' 'Double') - with: #(#char #short #int #'long long' #float #double) do: - [:structName :type| - aStream - nextPutAll: 'typedef struct { char pad_to_misalgnment; '; - nextPutAll: type; - nextPutAll: ' element; } struct'; - nextPutAll: structName; - nextPut: $;; cr; cr. - unionStream crtab: 3; nextPutAll: type; space; nextPutAll: structName; nextPut: $;]]. aStream + nextPutAll: 'typedef struct { char pad_to_misalgnment; struct {char c; } element; } structStruct;'; + cr; cr! - nextPutAll: 'typedef struct { char pad_to_misalgnment;'; - crtab: 2; - nextPutAll: 'union {'; - nextPutAll: union; - crtab: 2; - nextPutAll: '} element; } structStruct;'!

1 0

VM Maker: VMMaker.oscog-eem.2758.mcz
by commits＠source.squeak.org 01 Jun '20

01 Jun '20

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2758.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2758 Author: eem Time: 31 May 2020, 3:48:23.333686 pm UUID: 12eca51a-b989-486f-9a89-a66bba48ccc1 Ancestors: VMMaker.oscog-eem.2757 undo two unintended changes form the previous commit. =============== Diff against VMMaker.oscog-eem.2757 =============== Item was removed: - ----- Method: CogIA32Compiler>>genAlignCStackSavingRegisters:numArgs:wordAlignment: (in category 'abi') ----- - genAlignCStackSavingRegisters: regMask numArgs: numArgs wordAlignment: alignment - <inline: true> - | regMaskCopy numRegsPushed wordsPushedModAlignment delta | - <var: 'regMaskCopy' type: #usqInt> - self shouldBeImplemented. - self flag: 'test the assert & debug VMs on Windows and Mac OS (while you still can). The Windows build produces stack alignment assert fails, probably due to ceInvokeInterpret. The issue is the alignment calculation may not take into account the return address.'. - regMaskCopy := regMask asUnsignedInteger. - numRegsPushed := 0. - [regMaskCopy ~= 0] whileTrue: - [numRegsPushed := numRegsPushed + (regMaskCopy bitAnd: 1). - regMaskCopy := regMaskCopy bitShift: -1]. - (numRegsPushed = 0 - and: [self numIntRegArgs >= numArgs]) ifTrue: - [^0]. - wordsPushedModAlignment := numRegsPushed + (numArgs - self numIntRegArgs max: 0) \\ alignment. - wordsPushedModAlignment ~= 0 ifTrue: - [delta := alignment - wordsPushedModAlignment. - cogit SubCq: delta * objectMemory wordSize R: SPReg]. - ^0! Item was changed: ----- Method: Cogit>>genInvokeInterpretTrampoline (in category 'initialization') ----- genInvokeInterpretTrampoline "Switch to the C stack (do *not* save the Smalltalk stack pointers; this is the caller's responsibility), and invoke interpret PDQ." | startAddress | <inline: false> startAddress := methodZoneBase. self zeroOpcodeIndex. backEnd hasVarBaseRegister ifTrue: [self MoveCq: self varBaseAddress R: VarBaseReg]. "Must happen first; value may be used in genLoadStackPointers" cFramePointerInUse ifTrue: [backEnd genLoadCStackPointers] ifFalse: [backEnd genLoadCStackPointer]. - cStackAlignment > objectMemory wordSize ifTrue: - [backEnd - genAlignCStackSavingRegisters: self emptyRegisterMask - numArgs: 0 - wordAlignment: cStackAlignment / objectMemory wordSize]. "Sideways call interpret so that the stack looks correct, for exception handling etc" backEnd genMarshallNArgs: 0 arg: nil arg: nil arg: nil arg: nil. backEnd hasLinkRegister ifTrue: [self MoveAw: coInterpreter cReturnAddressAddress R: LinkReg] ifFalse: [self MoveAw: coInterpreter cReturnAddressAddress R: ABIResultReg. backEnd genSubstituteReturnAddressR: ABIResultReg]. self JumpFullRT: (self cCode: [#interpret asUnsignedInteger] inSmalltalk: [self simulatedTrampolineFor: #interpret]). self outputInstructionsForGeneratedRuntimeAt: startAddress. self recordGeneratedRunTime: 'ceInvokeInterpret' address: startAddress. ^startAddress!

1 0

VM Maker: VMMaker.oscog-eem.2757.mcz
by commits＠source.squeak.org 01 Jun '20

01 Jun '20

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2757.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2757 Author: eem Time: 31 May 2020, 3:42:59.428516 pm UUID: c2aedcf5-7fdb-48d6-90c2-15d41942bc06 Ancestors: VMMaker.oscog-ul.2756 Spur: Make sure prim fail codes in primitiveConstantFillSpur are correct. ThreadedFFIPlugin: Add ThreadedFFIPlugin>>primitiveStructureElementAlignment. Fix some comment typos. Slang: Allow a VMStructType to control adding its types to the struct name cache, ehnce allowing AlignmentStruct to encode multiple struct types. =============== Diff against VMMaker.oscog-ul.2756 =============== Item was added: + VMStructType subclass: #AlignmentStructType + instanceVariableNames: 'element' + classVariableNames: '' + poolDictionaries: '' + category: 'VMMaker-Plugins-FFI'! + + !AlignmentStructType commentStamp: 'eem 5/30/2020 18:50' prior: 0! + AlignmentStruct is a hack to declare and generate all the structure field alignment types used in ThreadedFFIPlugin>>primitiveStructureElementAlignment. + + Instance Variables + element: <char,short,int,long long, etc> + + element + - a C type that is defined according to the relevant type name + ! Item was added: + ----- Method: AlignmentStructType class>>addStructTypeNamesTo: (in category 'code generation') ----- + addStructTypeNamesTo: aSet + aSet addAll: #('structByte' 'structShort' 'structInt' 'structLongLong' 'structFloat' 'structDouble' 'structStruct')! Item was added: + ----- Method: AlignmentStructType class>>printTypedefOn: (in category 'code generation') ----- + printTypedefOn: aStream + | union | + union := String streamContents: + [:unionStream| + #('Byte' 'Short' 'Int' 'LongLong' 'Float' 'Double') + with: #(#char #short #int #'long long' #float #double) do: + [:structName :type| + aStream + nextPutAll: 'typedef struct { char pad_to_misalgnment; '; + nextPutAll: type; + nextPutAll: ' element; } struct'; + nextPutAll: structName; + nextPut: $;; cr; cr. + unionStream crtab: 3; nextPutAll: type; space; nextPutAll: structName; nextPut: $;]]. + aStream + nextPutAll: 'typedef struct { char pad_to_misalgnment;'; + crtab: 2; + nextPutAll: 'union {'; + nextPutAll: union; + crtab: 2; + nextPutAll: '} element; } structStruct;'! Item was changed: ----- Method: CCodeGenerator>>generateBetweenAnd:on:indent: (in category 'C translation') ----- generateBetweenAnd: msgNode on: aStream indent: level "Generate the C code for the between:and: message onto the given stream." aStream nextPutAll: '(('. self emitCExpression: msgNode receiver on: aStream. aStream nextPutAll: ' >= '. self emitCExpression: msgNode args first on: aStream. aStream nextPutAll: ') && ('. + self emitCExpression: (msgNode receiver isAssignment + ifTrue: [msgNode receiver variable] + ifFalse: [msgNode receiver]) + on: aStream. - self emitCExpression: msgNode receiver on: aStream. aStream nextPutAll: ' <= '. self emitCExpression: msgNode args second on: aStream. aStream nextPutAll: '))'! Item was changed: ----- Method: CoInterpreter>>internalExecuteNewMethod (in category 'message sending') ----- internalExecuteNewMethod <inline: true> "For interpreter performance and to ease the objectAsMethod implementation eagerly evaluate the primitive, i.e. if the method is cogged and has a primitive /do not/ evaluate the machine code primitive, just evaluate primitiveFunctionPointer directly." | succeeded methodHeader | primitiveFunctionPointer ~= 0 ifTrue: [self isPrimitiveFunctionPointerAnIndex ifTrue: [^self internalQuickPrimitiveResponse]. "slowPrimitiveResponse may of course context-switch. If so we must reenter the new process appropriately, returning only if we've found an interpreter frame." self externalizeIPandSP. succeeded := self slowPrimitiveResponse. instructionPointer = cogit ceReturnToInterpreterPC ifTrue: [instructionPointer := self iframeSavedIP: framePointer]. self internalizeIPandSP. succeeded ifTrue: [self return: self popStack toExecutive: true. self browserPluginReturnIfNeeded. ^nil]]. methodHeader := self rawHeaderOf: newMethod. "if not primitive, or primitive failed, activate the method" (self isCogMethodReference: methodHeader) ifFalse: [^self internalActivateNewMethod]. self iframeSavedIP: localFP put: localIP asInteger. instructionPointer := cogit ceReturnToInterpreterPC. self externalizeFPandSP. + "This may context switch and hence return..." - "THis may cintext switch and hence return..." self activateNewCogMethod: (self cCoerceSimple: methodHeader to: #'CogMethod *') inInterpreter: true. "Hence this si reachable..." self internalizeIPandSP! Item was added: + ----- Method: CogIA32Compiler>>genAlignCStackSavingRegisters:numArgs:wordAlignment: (in category 'abi') ----- + genAlignCStackSavingRegisters: regMask numArgs: numArgs wordAlignment: alignment + <inline: true> + | regMaskCopy numRegsPushed wordsPushedModAlignment delta | + <var: 'regMaskCopy' type: #usqInt> + self shouldBeImplemented. + self flag: 'test the assert & debug VMs on Windows and Mac OS (while you still can). The Windows build produces stack alignment assert fails, probably due to ceInvokeInterpret. The issue is the alignment calculation may not take into account the return address.'. + regMaskCopy := regMask asUnsignedInteger. + numRegsPushed := 0. + [regMaskCopy ~= 0] whileTrue: + [numRegsPushed := numRegsPushed + (regMaskCopy bitAnd: 1). + regMaskCopy := regMaskCopy bitShift: -1]. + (numRegsPushed = 0 + and: [self numIntRegArgs >= numArgs]) ifTrue: + [^0]. + wordsPushedModAlignment := numRegsPushed + (numArgs - self numIntRegArgs max: 0) \\ alignment. + wordsPushedModAlignment ~= 0 ifTrue: + [delta := alignment - wordsPushedModAlignment. + cogit SubCq: delta * objectMemory wordSize R: SPReg]. + ^0! Item was changed: ----- Method: Cogit>>genInvokeInterpretTrampoline (in category 'initialization') ----- genInvokeInterpretTrampoline "Switch to the C stack (do *not* save the Smalltalk stack pointers; this is the caller's responsibility), and invoke interpret PDQ." | startAddress | <inline: false> startAddress := methodZoneBase. self zeroOpcodeIndex. backEnd hasVarBaseRegister ifTrue: [self MoveCq: self varBaseAddress R: VarBaseReg]. "Must happen first; value may be used in genLoadStackPointers" cFramePointerInUse ifTrue: [backEnd genLoadCStackPointers] ifFalse: [backEnd genLoadCStackPointer]. + cStackAlignment > objectMemory wordSize ifTrue: + [backEnd + genAlignCStackSavingRegisters: self emptyRegisterMask + numArgs: 0 + wordAlignment: cStackAlignment / objectMemory wordSize]. "Sideways call interpret so that the stack looks correct, for exception handling etc" backEnd genMarshallNArgs: 0 arg: nil arg: nil arg: nil arg: nil. backEnd hasLinkRegister ifTrue: [self MoveAw: coInterpreter cReturnAddressAddress R: LinkReg] ifFalse: [self MoveAw: coInterpreter cReturnAddressAddress R: ABIResultReg. backEnd genSubstituteReturnAddressR: ABIResultReg]. self JumpFullRT: (self cCode: [#interpret asUnsignedInteger] inSmalltalk: [self simulatedTrampolineFor: #interpret]). self outputInstructionsForGeneratedRuntimeAt: startAddress. self recordGeneratedRunTime: 'ceInvokeInterpret' address: startAddress. ^startAddress! Item was changed: ----- Method: InterpreterPrimitives>>primitiveConstantFillSpur (in category 'sound primitives') ----- primitiveConstantFillSpur "Fill the receiver, which must be an indexable non-pointer object, with the given integer value." <inline: true> | fillValue rcvr format end i oddBytes | <var: #fillValue type: #usqLong> <var: #end type: #usqInt> <var: #i type: #usqInt> argumentCount ~= 1 ifTrue: [^self primitiveFailFor: PrimErrBadNumArgs]. rcvr := self stackValue: 1. ((objectMemory isNonImmediate: rcvr) and: [(format := objectMemory formatOf: rcvr) >= objectMemory sixtyFourBitIndexableFormat]) ifFalse: + [^self primitiveFailFor: PrimErrBadReceiver]. - [^self primitiveFailFor: PrimErrBadReceiver]. (objectMemory isObjImmutable: rcvr) ifTrue: [^self primitiveFailFor: PrimErrNoModification]. fillValue := self positive64BitValueOf: self stackTop. self successful ifFalse: [^self primitiveFailFor: PrimErrBadArgument]. format >= objectMemory firstShortFormat ifTrue: [format >= objectMemory firstByteFormat ifTrue: [(fillValue > 16rFF or: [format >= objectMemory firstCompiledMethodFormat]) ifTrue: + [^self primitiveFailFor: (fillValue > 16rFF ifTrue: [PrimErrBadArgument] ifFalse: [PrimErrBadReceiver])]. - [^self primitiveFail]. fillValue := fillValue + (fillValue << 8) + (fillValue << 16) + (fillValue << 24). oddBytes := format bitAnd: 7] ifFalse: [fillValue > 16rFFFF ifTrue: + [^self primitiveFailFor: PrimErrBadArgument]. - [^self primitiveFail]. fillValue := fillValue + (fillValue << 16). oddBytes := (format bitAnd: 3) << 1]. fillValue := fillValue + (fillValue << 32)] ifFalse: [format = objectMemory sixtyFourBitIndexableFormat ifTrue: [oddBytes := 0] ifFalse: [fillValue > 16rFFFFFFFF ifTrue: + [^self primitiveFailFor: PrimErrBadArgument]. - [^self primitiveFail]. fillValue := fillValue + (fillValue << 32). oddBytes := (format bitAnd: 1) << 2]]. end := objectMemory addressAfter: rcvr. i := rcvr + objectMemory baseHeaderSize. [i < end] whileTrue: [objectMemory long64At: i put: fillValue. i := i + 8]. "now ensure trailing bytes are zero" oddBytes > 0 ifTrue: [self flag: #endianness. fillValue := fillValue >> (8 * oddBytes). objectMemory long64At: i - 8 put: fillValue]. self pop: 1! Item was changed: ----- Method: ThreadedFFIPlugin class>>ancilliaryClasses (in category 'translation') ----- ancilliaryClasses + ^{ self calloutStateClass. AlignmentStructType }! - ^{ self calloutStateClass }! Item was added: + ----- Method: ThreadedFFIPlugin class>>primitiveAlignmentOf: (in category 'simulation') ----- + primitiveAlignmentOf: atomicTypeCode + "Answer the platform C compiler's alignment in a structure of an element of the + C type corresponding to atomicTypeCode, where atomicTypeCode is in the range + FFITypeUnsignedByte (2) to FFITypeDoubleFloat (13), or is FFIFlagStructure (65536)." + <primitive: #primitiveStructureElementAlignment module: #SqueakFFIPrims> + self primitiveFailed + + "(FFITypeUnsignedByte to: FFITypeDoubleFloat), {FFIFlagStructure} collect: [:n| self primitiveAlignmentOf: n]"! Item was changed: ----- Method: ThreadedFFIPlugin>>checkAlignmentOfStructSpec:OfLength:StartingAt: (in category 'marshalling-struct') ----- checkAlignmentOfStructSpec: specs OfLength: specSize StartingAt: startIndex "Check the alignment of a structure and return true if correctly aligned. If computed size = declared size, then the struct is assumed correctly aligned." | index spec computedSize fieldAlignment fieldSize declaredSize maxAlignment | <var: #specs type: #'unsigned int*'> - <var: #indexPtr type: #'unsigned int*'> <inline: false> index := startIndex. spec := specs at: index. self assert: (spec bitAnd: FFIFlagPointer + FFIFlagAtomic + FFIFlagStructure) = FFIFlagStructure. (self isUnionSpec: specs OfLength: specSize StartingAt: index) ifTrue: [^self checkAlignmentOfUnionSpec: specs OfLength: specSize StartingAt: startIndex]. declaredSize := spec bitAnd: FFIStructSizeMask. computedSize := 0. maxAlignment := 1. [index := index + 1. index < specSize] whileTrue: [spec := specs at: index. spec = FFIFlagStructure ifTrue: [^(computedSize - 1 bitOr: maxAlignment - 1) + 1 = declaredSize]. (spec anyMask: FFIFlagPointer) ifTrue: [fieldSize := BytesPerWord. fieldAlignment := fieldSize] ifFalse: [fieldSize := spec bitAnd: FFIStructSizeMask. (spec anyMask: FFIFlagStructure) ifTrue: [(self checkAlignmentOfStructSpec: specs OfLength: specSize StartingAt: index) ifFalse: [^false]. fieldAlignment := self alignmentOfStructSpec: specs OfLength: specSize StartingAt: (self addressOf: index)] ifFalse: [fieldAlignment := fieldSize]]. "round to fieldAlignment" maxAlignment := maxAlignment max: fieldAlignment. computedSize := (computedSize - 1 bitOr: fieldAlignment - 1) + 1. computedSize := computedSize + fieldSize]. ^(computedSize - 1 bitOr: maxAlignment - 1) + 1 = declaredSize! Item was changed: ----- Method: ThreadedFFIPlugin>>checkAlignmentOfUnionSpec:OfLength:StartingAt: (in category 'marshalling-struct') ----- checkAlignmentOfUnionSpec: specs OfLength: specSize StartingAt: startIndex "Check the alignment of a union and return true if correctly aligned. Union are correctly aligned, but a sub-structure might not." | index spec | <var: #specs type: #'unsigned int*'> - <var: #indexPtr type: #'unsigned int*'> <inline: false> index := startIndex. spec := specs at: index. [index := index + 1. index < specSize] whileTrue: [spec := specs at: index. spec = FFIFlagStructure ifTrue: [^true]. (spec anyMask: FFIFlagPointer) ifFalse: [(spec anyMask: FFIFlagStructure) ifTrue: [(self checkAlignmentOfStructSpec: specs OfLength: specSize StartingAt: index) ifFalse: [^false]]]]. ^true! Item was added: + ----- Method: ThreadedFFIPlugin>>primitiveStructureElementAlignment (in category 'primitives') ----- + primitiveStructureElementAlignment + "Answer the alignment of an element of an atomic type, or a structure, within a structure on the current platform." + <export: true> + | typeCode alignment | + typeCode := interpreterProxy stackValue: 0. + ((interpreterProxy isIntegerObject: typeCode) + and: [((typeCode := interpreterProxy integerValueOf: typeCode) between: FFITypeUnsignedByte and: FFITypeDoubleFloat) + or: [typeCode = FFIFlagStructure]]) ifFalse: + [^self primitiveFailFor: PrimErrBadArgument]. + + alignment := typeCode + caseOf: { + [FFITypeUnsignedByte] -> [self structOffsetOf: 'structByte *' atomicTypeCode: FFITypeUnsignedByte]. + [FFITypeSignedByte] -> [self structOffsetOf: 'structByte *' atomicTypeCode: FFITypeUnsignedByte]. + [FFITypeUnsignedShort] -> [self structOffsetOf: 'structShort *' atomicTypeCode: FFITypeUnsignedShort]. + [FFITypeSignedShort] -> [self structOffsetOf: 'structShort *' atomicTypeCode: FFITypeUnsignedShort]. + [FFITypeUnsignedInt] -> [self structOffsetOf: 'structInt *' atomicTypeCode: FFITypeUnsignedInt]. + [FFITypeSignedInt] -> [self structOffsetOf: 'structInt *' atomicTypeCode: FFITypeUnsignedInt]. + [FFITypeUnsignedLongLong] -> [self structOffsetOf: 'structLongLong *' atomicTypeCode: FFITypeUnsignedLongLong]. + [FFITypeSignedLongLong] -> [self structOffsetOf: 'structLongLong *' atomicTypeCode: FFITypeUnsignedLongLong]. + [FFITypeSingleFloat] -> [self structOffsetOf: 'structFloat *' atomicTypeCode: FFITypeSingleFloat]. + [FFITypeDoubleFloat] -> [self structOffsetOf: 'structDouble *' atomicTypeCode: FFITypeDoubleFloat]. + } + otherwise: [self structOffsetOf: 'structStruct *' atomicTypeCode: FFIFlagStructure]. + ^interpreterProxy methodReturnInteger: alignment! Item was added: + ----- Method: ThreadedFFIPlugin>>structOffsetOf:atomicTypeCode: (in category 'primitive support') ----- + structOffsetOf: structPointerType atomicTypeCode: atomicTypeCode + <inline: #always> + ^self cCode: [self addressOf: (self cCoerceSimple: 0 to: structPointerType) element] + inSmalltalk: [self class primitiveAlignmentOf: atomicTypeCode]! Item was changed: ----- Method: ThreadedX64SysVFFIPlugin class>>ancilliaryClasses (in category 'translation') ----- ancilliaryClasses + ^super ancilliaryClasses, + { ThreadedFFIX64SixteenByteReturnDD. - ^{ self calloutStateClass. - ThreadedFFIX64SixteenByteReturnDD. ThreadedFFIX64SixteenByteReturnDI. ThreadedFFIX64SixteenByteReturnID. ThreadedFFIX64SixteenByteReturnII }! Item was added: + ----- Method: VMStructType class>>addStructTypeNamesTo: (in category 'code generation') ----- + addStructTypeNamesTo: aSet + aSet add: self name; add: self structTypeName! Item was added: + ----- Method: VMStructType class>>ensureStructTypeNameCache (in category 'translation') ----- + ensureStructTypeNameCache + ^StructTypeNameCache ifNil: + [StructTypeNameCache := Set new. + self allSubclassesDo: + [:sc| sc addStructTypeNamesTo: StructTypeNameCache]. + StructTypeNameCache]! Item was changed: ----- Method: VMStructType class>>isTypePointerToStruct: (in category 'translation') ----- isTypePointerToStruct: type | index | - StructTypeNameCache ifNil: - [StructTypeNameCache := Set new. - self allSubclassesDo: - [:sc| StructTypeNameCache add: sc name; add: sc structTypeName ]]. ^type notNil and: [(index := type indexOf: $*) > 0 + and: [self ensureStructTypeNameCache anySatisfy: - and: [StructTypeNameCache anySatisfy: [:structType| (type beginsWith: structType) and: [index > structType size]]]]! Item was changed: ----- Method: VMStructType class>>structTargetKindForDeclaration: (in category 'translation') ----- structTargetKindForDeclaration: decl - StructTypeNameCache ifNil: - [StructTypeNameCache := Set new. - self allSubclassesDo: - [:sc| StructTypeNameCache add: sc name; add: sc structTypeName ]]. ^(decl notNil + and: [(self ensureStructTypeNameCache includes: decl) - and: [(StructTypeNameCache includes: decl) or: [StructTypeNameCache anySatisfy: [:structType| (decl beginsWith: structType) and: [(decl indexOf: $* ifAbsent: [decl indexOf: Character space]) > structType size]]]]) ifTrue: [(decl indexOf: $*) > 0 ifTrue: [#pointer] ifFalse: [#struct]]!

1 0

← Newer
1
2
3
4
5
6
7
...
13
Older →

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev May 2020